Philosophical Provocations: 55 Short Essays

Philosophical Provocations

Philosophical Provocations 55 Short Essays

Colin McGinn

The MIT Press Cambridge, Massachusetts London, England

© 2017 Massachusetts Institute of Technology All rights reserved. No part of this book may be reproduced in any form by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher. This book was set in ITC Stone Sans Std and ITC Stone Serif Std by Toppan Best-set Premedia Limited. Printed and bound in the United States of America. Library of Congress Cataloging-in-Publication Data Names: McGinn, Colin, 1950- author. Title: Philosophical provocations : 55 short essays / Colin McGinn. Description: Cambridge, MA : MIT Press, 2017. | Includes bibliographical references and index. Identifiers: LCCN 2016046174 | ISBN 9780262036191 (hardcover : alk. paper) Subjects: LCSH: Philosophy. Classification: LCC BD41 .M34 2017 | DDC 192--dc23 LC record available at https://lccn.loc.gov/2016046174 10 9 8 7 6 5 4 3 2 1

Contents

Preface  vii I Mind  1 The Mystery of the Unconscious  3 Concepts of Mind  15 Knowledge and Emotion: An Untenable Dualism  19 The Second Mind  25 Awareness of Time  29 Mind–Brain Identity Theories  33 Are There Actions?  41 Actions and Reasons  45 Consciousness and Light  49 Have We Already Solved the Mind–Body Problem?  53 The Reality of the Inner  57 The Thought of Language  63 II Language  67 Meaning Monism  69 Against Language-Games  75 Meaning without Language  79 For Privacy  87 On the Impossibility of a (Wholly) Public Language  91 Deciding to Mean  99 Truth, Verification, and Meaning  105 Meaning and Argument  109 III Knowledge  113 Knowledge and Truth  115 Proof of an External World  129 The Simulation Game  131

viContents

The Riddle of Knowledge  133 Does Knowledge Imply Truth?  139 Everything Is Hidden  143 Light and Our Knowledge of Matter  145 Seeing the Light  149 IV Metaphysics  155 Knowing and Necessity  157 Antirealism Refuted  177 The Puzzle of Paradox  181 The Secret Cement  187 Analysis and Mystery  193 Explanation and Necessity  199 Against Possible Worlds  203 The Concept of a Person  207 The Question of Being  211 Science as Metaphysics  215 Logic without Propositions (or Sentences)  219 V Biology  227 Selfish Genes and Moral Parasites  229 The Evolution of Color  241 The Language of Evolution  245 Immaterial Darwinism  251 Trait Selection  253 VI Ethics  257 The Trouble with Consequentialism  259 Absurd Utilitarianism  267 Why Is It Good to Be Alive?  269 Physical Noncognitivism  275 Child Liberation  277 Modesty and Self-Knowledge  281 Is Romantic Self-Love Possible?  285 Against Laughter  289 VII Religion  293 A Deontological Theodicy  295 God and the Devil  301 A Religion of Hate  303 Index  307

Preface

The essays collected here follow a particular pattern and style. They aim to be pithy, with no padding or extraneous citation. They each address a specific philosophical issue and try to make progress with it as efficiently as possible. There is little discussion of particular authors or “the literature”; my aim is to get down to the issues immediately and state a position. I avoid excessive qualification or self-protection, leaving it to the reader to fill in gaps. Thus: short, sharp, and breezy. The style is intended to contrast with the way academic philosophy tends to be written these days: long, leaden, citation-heavy, and painful to read. This means that the book can be read by someone not expert in the fields covered, though I would not deny that some of the essays are quite demanding—but not I think unnecessarily so. There is room in academic writing for the style I avoid, but there is something to be gained by the direct and unencumbered style I adopt here (I have written my fair share of the other kind of stuff). A great many topics are covered in the book, and I make no attempt to link them.1 Each occurred to me independently of the others, though there are thematic continuities. They often challenge orthodoxy in ways that might seem shocking and put forward views that may be condemned as eccentric. There is nothing safe about these essays—though I do help myself to a dose of common sense when needed. I am self-consciously trying to find new ways to think about old problems. Maybe not all the suggestions will stand the test of further reflection, but I hope they stir people to think in fresh ways—or at least to feel provoked at my audacity. My ideal over the period of writing these essays (2014 to 2016) was to resolve a serious philosophical problem in no more than a page. Needless to say, this ideal was just that—not a realistic prospect. Still, it was a useful guiding principle, because it forced 1.  In fact the essays published here constitute a selection from a larger number of essays written over the same period (one hundred and thirty in all). I hope to publish the remaining essays in a later volume.

viiiPreface

me to extremes of economy in formulation and argument. I ruthlessly excluded all preamble and qualification, knowing that “sophisticated” readers will pounce on certain passages; I tried to get right to the point and state my view with maximum clarity. Admittedly, I did occasionally employ arguments, in the standard style, but I tried to keep them short and sweet. No doubt I’ve sometimes lapsed into prolixity and professionalism, old habits being what they are, but I regard that as a fault, not a virtue. If an essay does not raise professional eyebrows at some point, I count it a failure.   Colin McGinn

I Mind

The Mystery of the Unconscious

Introduction For some time now consciousness has been at the center of discussions of the mind– body problem. We might say that today the mind–body problem is the consciousness–brain problem. But it was not always so. Thomas Nagel felt it necessary to begin his classic 1974 paper “What Is It Like to Be a Bat?” with these words: “Consciousness is what makes the mind–body problem really intractable. Perhaps that is why current discussions of the problem give it little attention or get it obviously wrong.” I introduced my 1989 paper “Can We Solve the Mind–Body Problem?” as follows: “The specific problem I want to discuss concerns consciousness, the hard nut of the mind– body problem. How is it possible for conscious states to depend upon brain states? How can Technicolor phenomenology arise from soggy grey matter?” David Chalmers has spoken of consciousness as the “hard problem,” distinguishing it from (relatively) “easy” problems of the mind, and finding in it a uniquely difficult phenomenon to bring within a materialist perspective. And many others have felt that the problem of consciousness is particularly and distinctively deep or profound—as if without it the mind would be far less intractable, perhaps not intractable at all. Consciousness is now thought to be the great enigma of the mind. I think we have all been wrong. I say this not because I believe that consciousness is tractable after all; it is rather that I think it is not uniquely intractable. To be more specific, I think that much the same problems that afflict consciousness also afflict unconsciousness: unconscious mental states are as problematic as conscious mental states, and for essentially the same reasons. The emphasis on consciousness is therefore misplaced, even though the authors cited have been right to discern a deep problem about consciousness. Consciousness is indeed a mystery, but so is the unconscious (hence my title for this essay). The plan of the essay is as follows. First I will explain how the standard arguments against materialism for conscious states carry over to unconscious states. Next I will offer a diagnosis of why it is harder to see this in the case of the unconscious than it is for the conscious. Then I will describe the intimate relations between the conscious mind and the unconscious mind, so that we can see that it is extremely unlikely that the former should be deeply intractable while the latter is relatively plain sailing. It is the mind in its entirety that poses an intractably hard problem for materialism, not just the conscious part of it. Antimaterialist Arguments It is not difficult to see how the standard arguments transpose to the unconscious, once the question has been formulated. I will therefore go over this ground quickly,

4

Part I

assuming that the reader is familiar with the arguments in question. I am going to use what I hope is an uncontroversial example of an unconscious mental state—not the Freudian or Jungian unconscious or anything exotic like that. I came to the position to be outlined here in the course of defending innate ideas in the Cartesian sense: these ideas had to be unconscious, given that babies are not born consciously thinking about mathematics or sensible qualities or anything else Descartes took to be innate (rightly, in my view). But I don’t want to enter this controversial territory now; instead I shall discuss a straightforward case of experiential memory. Suppose that yesterday I saw a red bird in my garden and noted its fine plumage. Then I went back to work and didn’t think any more about it. Today I (consciously) remember seeing the bird and wonder if I will see it again: I recall seeing its brilliant red plumage, just as I experienced it yesterday. I have a visual memory of my perceptual experience. Between these two times, evidently, I stored the experience of the red bird in my memory, wherein it resided unconsciously. That is, I was not conscious of the memory between the two times; like most of my memories it existed in an unconscious form, though it was not difficult to bring it to consciousness. The unconscious memory was sandwiched between an act of conscious perception and an act of conscious remembering: but it was not itself conscious. It existed in the unconscious part of my mind. For brevity, call this memory “the memory of red.” Now compare the memory of red with a perceptual experience of red, and consider the standard arguments. The standard arguments against viewing the experience of red as a brain state (or a functional state) are as follows: Nagel’s subjectivity argument, Jackson’s knowledge argument, Kripke’s modal argument, Putnam’s multiple realization argument, Chalmers’s zombie argument, Brentano-inspired arguments from intentionality, Block’s inversion and absence arguments (against functionalism), and the brute primitive intuition that experiences cannot in their very nature be identified with brain states. Each of these arguments purports to show that the experience of red cannot be reduced to a physical state. Thus, Nagel contends, the experience is subjective while the brain is objective: the man born blind cannot know what a subjective experience of red is like, having never had such an experience, though he may grasp the objective nature of the brain states that correlate with that experience. We cannot know what it is subjectively like to be a bat, though we can know the objective physiology of a bat’s brain. Jackson’s Mary could know all about color vision and not know what it is to see something as red—as is shown by her learning something new when she emerges from her black-and-white room and sees something red for the first time. We can imagine a possible world, Kripke tells us, in which someone has an experience of red but not the usual brain correlate found in the actual world, and equally we can imagine the usual brain correlate associated with a different color experience or no experience at all. Hence there cannot be an identity between the two,

Mind5

since identity is a necessary relation. Putnam argues that the experience of red may have a different physical realization in Martian brains, so it cannot be identified with the brain state humans happen to use to implement it. Chalmers argues that there is no a priori entailment between physical descriptions of the subject and descriptions of her experiences, so that zombies are logically possible, thus refuting the logical sufficiency of brains for conscious minds. Brentano observes that the experience of red has an intentional directedness to the color red, which cannot be found in the physical world. Block conjectures that we can conceive of cases in which an organism has the same functional characterization as an organism experiencing red and yet that organism experiences green (“inverted qualia”) or experiences no color at all (“absent qualia”). And many people instinctively feel that it is just massively implausible to suppose that an experience of red is nothing but a brain state of neurons firing, given how different the two things seem. The question is whether the same arguments can be applied to the memory of red. Suppose a bat uses its sense of echolocation and has the corresponding perceptual experience. Then it stores that experience in memory, perhaps for later use. The memory exists in an unconscious form, but it is an experiential memory. Can we grasp the nature of that memory? Not if we can’t grasp the nature of a bat’s experience, since that is what it is a memory of. To grasp it we would need to occupy the bat’s subjective point of view, as Nagel puts it; but we don’t, so we can’t. We can know all about a bat’s brain, including the memory centers, but that will not tell us the nature of a bat’s memory, since that memory incorporates the bat’s alien sense of echolocation. Yet the memory is unconscious. Note that I did not say that we are unable to grasp what it is like to have that memory, since (being unconscious) there is nothing it is like. What we fail to grasp—because we have no conception of it—is the nature or essence of the memory. The case is just like a blind person trying to grasp what a memory of having a red experience is. He does not grasp that concept, any more than he grasps the concept of a conscious perceptual experience of red. Of course, a memory of a conscious perceptual experience is precisely that—a memory whose content refers to a state of consciousness. But it doesn’t follow from this that it is itself an instance of consciousness. So the memory is problematic despite itself being unconscious. Mary learns everything there is to know about the physical basis of memory locked in her black-and-white room. She herself has no memories of having seen a red thing, since she hasn’t. She emerges and has her first red experience, which she commits to memory (she found it very memorable). Now she has the concept of a memory of red, which she lacked before. So she has learned something—she now knows what a memory of red is. So there are more facts about memory than she knew back in her room. So the memory of red is not a physical fact. Yet the memory is unconscious.

6

Part I

Consider a memory of red and its neural correlate, say M-fibers firing. Surely I can conceive of this memory existing in the absence of M-fibers; maybe in some other possible world that memory is associated with N-fibers firing. But identity is a necessary relation, so the two cannot be identical. Equally, can I not imagine a world in which M-fibers are firing and yet a memory of green is associated with that neural correlate, or no memory at all? There is an “intuition of contingency” here that is incompatible with the identity theory. Nor can that intuition be explained away by means of epistemic counterparts and the like, since there is no sense in the idea of a “seeming memory” of red that is not an (apparent) memory of red (like seeming water that isn’t really water). Yet the memory of red is unconscious. Martians have brains made of silicon chips, not neurons. But they are functionally just like us, specifically in regard to “recall behavior.” They clearly have memories and perceive colors—so they have memories of colors perceived, including red. Thus the memory of red can be multiply realized, and the identity theory must be false (at least for memory types). Yet the memory of red is unconscious. There could be a zombie that behaves just like us but has no actual memories. Since the zombie has no perceptual experiences, it has no memories of such experiences; so it has no memory of (seeing) red. It has neither a conscious mind nor an unconscious mind, being a total zombie. Therefore, being a physical duplicate of a creature with memory is not sufficient for having memory; so the memory of red is not necessitated by physical properties of the organism. So materialism is false. Yet the memory of red is unconscious. The memory of red has intentional directedness—it is a memory of red. It is just as strongly intentional as the perceptual experience of red. But intentionality is not reducible to physical relations. So the memory of red is not a physical state. Yet it is unconscious. There can be functional equivalents of us that exhibit inverted and absent memory: where we remember red they remember green, or they simply have no memories at all. So the memory of red cannot be a functional property. So functionalism is false. Yet the memory of red is unconscious. Finally, it just seems massively implausible that a memory of an experience could be nothing but neurons firing—they seem like such different things. So the memory of red is not a brain state. Yet it is unconscious. My purpose in going over these arguments, and noting their extension to unconscious memories, is not to endorse the arguments (though I have a lot of sympathy for them); it is rather to point out that they take the same form for both the conscious and the unconscious. So the force of those arguments does not depend on their being directed to conscious mental states specifically. They apply to mental states more generally, particularly unconscious memories (or other unconscious mental states). It is mentality that is the issue, not conscious mentality. Consciousness is therefore not

Mind7

uniquely intractable or hard or deep. It is a special case of a wider intractability, which applies also to unconscious mentality.1 Conceiving the Unconscious Why has this rather obvious point not been generally recognized and pressed (including by my earlier self)? I think it is because we are not confronted by the unconscious. With consciousness, the problematic thing is staring us in the face: we know we have it by direct introspection, by immediate acquaintance. Its existence is self-evident and undeniable—the primary Cartesian certainty. The nature of conscious states is also evident to us. Thus we have a basic intuition (right or wrong) that this thing we know so well is not merely a state of our brain, which we do not know so well. We are accordingly open to arguments that trade on that intuition. But the unconscious is precisely not an object of introspection, not known by acquaintance, and not self-evident and undeniable. We postulate the unconscious; we are not confronted by its reality day in and day out. It lies behind the scenes, revealing its presence only indirectly. Thus we have trouble recognizing its psychological reality (compare the world of atoms, which likewise does not confront us and must be postulated). But this is a bad way to assess the contents of the mind: there clearly have to be memories if we are to make sense of our common experience, and they are obviously not present to consciousness at all times. And then the question must arise as to their relation to the brain, with the same basic considerations applying as in the conscious case. Actually, the problem of our grasp of the unconscious goes deeper. Put simply, I think we have no adequate concept of unconscious mental states, including memories. In the case of consciousness we can conceive of it under the mode of presentation afforded by introspection—we conceive of conscious experiences of red as we introspect them. We are acquainted with them and we conceptualize them in terms of that acquaintance. But in the case of unconscious memories we are not acquainted with them, so we cannot conceptualize them that way. They are mental, but we cannot conceive of them in the way we conceive of conscious mental states. How then can we 1.  Do not be misled by the fact that the origin of the memory of a red bird is itself a conscious experience: this does not imply that the problems for the unconscious memory are really derivative from the problems of consciousness. First, although the original experience is conscious, the memory of it is not; it exists completely unconsciously prior to the recall. Second, strictly speaking what is remembered is the red bird, not the experience of it: the bird is the intentional object of the memory, not the experience of that bird. Third, we can easily describe a case where there is no temptation to make the content of the memory a conscious experience: just consider ordinary memories of historical facts or geographical layouts or where you left your keys—these too pose problems for materialism analogous to those posed by conscious knowledge of such things. In the simplest case, unconscious beliefs are as difficult for materialism as conscious beliefs inasmuch as both are beliefs; at any rate, they pose many of the same problems.

8

Part I

conceive of them? A bad way would be to try to conceive of them as if they are conscious to someone else—as if we had a homunculus inside for whom our unconscious memories are conscious. That is just desperate mythology. So what do we do instead? We either try to reduce the memory to a brain state or to a disposition to produce conscious states—for both of which we have adequate concepts. Neither attempt at reduction works (as I explain later), but they seem to be forced on us because otherwise we are left conceptually blank—we don’t really know what it is we are talking about when we talk of the unconscious. It has no natural mode of presentation to us, being neither an object of perception nor introspection. It exists in a kind of conceptual no-man’s land. We posit the unconscious for explanatory purposes, but it eludes our cognitive grasp. We have merely demonstrative concepts for it—that thing that fulfills a certain role, whatever exactly it is. Thus a memory of red is thought of as just “the unconscious mental state that explains why I am now consciously remembering an experience of red.” We are a bit like the blind man with regard to color perception: he can refer to it demonstratively and make true statements about it, but he doesn’t really know what he is talking about. Why? Because he is not conscious of such states—he has no acquaintance with them. They are a blank to him. Likewise, our own memories are alien to our comprehension—for they are presented to us neither perceptually nor introspectively. We know they are there but we don’t really grasp their nature, their mode of being. They are conceived as just “the residue of experience and the cause of recollection.” Our mode of referring to them is not cognitively penetrating—not revelatory. Thus we find ourselves conceptually uncomfortable with the unconscious; we can’t quite believe in it, though we are convinced it must exist. We are therefore inclined to instrumentalism about the unconscious, or reductionism, or maybe even eliminativism—because we suffer from a conceptual lacuna in relation to the unconscious. We lack a “clear and distinct idea” of the unconscious mind. But this dismissive reaction is misguided: the unconscious is psychologically real, no less robustly so than the conscious. If so, it presents all the same problems for materialism that consciousness presents (or many of them: see below). However, it takes a mental effort to appreciate this because of its conceptual elusiveness. We are less convinced of its intractability “in our gut”—that is, by immediate intuition. We find it hard to summon the unconscious to our thought, and hence we find it hard to appreciate its irreducibility. But that is a mental limitation on our part, not a count against the full psychological reality of the unconscious. And it is really quite easy to deploy the standard antimaterialist arguments against materialism with respect to the unconscious. They are no less cogent in the one case than in the other. If we could somehow (per impossibile) introspect our memories in their unconscious state— be directly confronted by them—then we would not hesitate to extend the standard arguments to them. But then consciousness is not really the heart of the mind–body

Mind9

problem, is not uniquely mysterious, is not the last bulwark against an encroaching materialism, and is not the be-all and end-all of the mind’s recalcitrance to physical explanation. The unconscious is a mystery too, and apparently a mystery of much the same kind. If we are cognitively closed to consciousness, we are also cognitively closed to unconsciousness. Unconscious mental states are no less physically inexplicable than conscious mental states. Is there any problem that consciousness adds to unconsciousness? Granted that the unconscious mind is problematic too, as suggested by the standard arguments, is there any extra dimension of difficulty presented by consciousness? I think this is a difficult question, turning on precisely how we are to understand consciousness. If we view consciousness as consisting in some kind of self-scanning, then presumably there will be no further deep problem: for a mental state to be conscious will be just for it to be the subject matter of a higher-order thought, or to be perceived by an inner sense, or to be attended to in some way. If we can explain these faculties in naturalistic fashion, then we will have explained consciousness. The question will be whether higher-order thoughts or inward perceptions or acts of attention can be physically explained. But if we think of consciousness in some other way, then things might not be so straightforward. Here is where we encounter talk of “qualia” and the “inner glow” and “phenomenal consciousness.” I admit to having a vague sense that there is something to consciousness that goes beyond the problems already posed by the unconscious, that is, mentality in general, but I also admit to having no good way to articulate exactly what this something might be. It cannot be anything revealed by the standard arguments, because they apply equally to the unconscious; but it is hard to put one’s finger on what the extra dimension of difficulty might be. Not that this makes consciousness any less mysterious than it is commonly acknowledged to be; it just means that consciousness is not alone in its mysteriousness—the unconscious is mysterious too, even in the form of a simple memory of a red bird. Conscious and Unconscious Someone might try to resist the argument so far by insisting that the unconscious is not quite real: that is, one might adopt a grudging or reductive or minimalist or eliminative view of it. This would be analogous to similarly dismissive attitudes toward consciousness before its existence became fully acknowledged by blinkered theorists. Given the current state of things, the most likely suggestion would be that unconscious mental states should be analyzed in terms of conscious mental states—for example, the unconscious memory of red should be analyzed as a disposition to produce conscious recollections of red. Then it will be claimed that the problems attaching to the memory derive from those attaching to consciousness, since the memory is nothing other than a disposition to undergo conscious states. But this is really no more plausible than the thesis that perceptual experience is a disposition to produce memories. It is true that

10

Part I

perceptual experiences have dispositions to give rise to memories, but it is not true that they reduce to such dispositions. Rather, the experience is the categorical ground of the disposition to produce memories; and in just that way the memory is the categorical ground of the disposition to produce acts of conscious remembering. In general, properties don’t reduce to the dispositions with which they are correlated and which they ground, though there is obviously an intimate relationship between a property and its corresponding dispositions. We can see this from thought experiments in which we can pull the two things apart: the memory could exist and not have the disposition, because of external interference (so the disposition is not necessary for the memory); and the disposition could exist without the memory, because of outside help—God steps in to make us remember, but no memory trace exists before that time (so the disposition is not sufficient for the memory). These are familiar enough points, which I don’t intend to belabor. What they tell us is that memories and dispositions to recall, though connected, are not the same thing. Memories are real internal states that persist unconsciously in the mind between the original experience and later conscious recall. That is precisely how we naively think of them, not as bare propensities to produce events of conscious recall. We have no conceptual difficulty in the idea of a repressed memory that refuses to see the light of day and exists hidden throughout its possessor’s life: it has a disposition not to be recalled. Another line of resistance suggests that unconscious mental states are deeply different, intrinsically so, from conscious mental states—belonging to a completely separate mental system. Memories are psychologically real, to be sure, but they are so removed from conscious experience that it is a mistake to assimilate whatever mysteries they may possess to the mystery of consciousness. I can sympathize with that position, because the difference between being conscious and being unconscious is a nontrivial difference, but I don’t think it stands up to scrutiny. Memories and perceptual experiences are just too closely intertwined for such an insulating view to be plausible. The first and most powerful point to make is that what exists in my experiential memory is the very thing that was earlier in my conscious experience: what I experienced when I saw that red bird is itself stored in my memory and later recalled. The same experience that I had yesterday is before my mind today, having been retained overnight in my memory. Similarly, it is the same belief that I retain when I stop thinking about it—so the very same mental state of belief is conscious and unconscious at different times. The experience I had yesterday comes back to me today (that experience), and it was preserved in my memory between times. The same perceptual experience is conscious at one time and unconscious at another time—it is precisely my experience of that red bird that is stored unconsciously in my memory over time. Here is a second point: An experience oscillates between being conscious and being unconscious, just like a belief. Freud spoke of a memory as being preconscious, that is,

Mind11

as something existing before conscious recollection; but he could equally have spoken of memory as postconscious, that is, as something I form following a conscious experience, say a perception. We have the sequence: perception, memory, recollection—the perception lays down the memory, which is later recalled. The first phase is conscious, the second unconscious, and the third conscious. These phases are tightly connected to each other, not separate and parallel—as the experience goes in and out of consciousness. What we have here is a perception-memory-recall system, in which the conscious and the unconscious work systematically together. It is not as if we have another self in which our unconscious mind resides, sealed off from consciousness. This is how the mind works, by this kind of conscious–unconscious oscillation. These are not separate systems but one integrated functional system. There is no dualism of conscious and unconscious; rather, there is a unified functional whole. So, if one part of the overall mental system is problematic, it is likely the other part will be. It would be extremely odd if the unconscious were mysterious and the conscious not; and it should seem equally odd if the conscious were steeped in mystery but the unconscious clear as a bell. In particular, it would be odd if the unconscious were straightforwardly physical but the conscious not, or vice versa. Beliefs certainly do not shed their problematic features just by becoming unconscious—that doesn’t turn them into something clearly physical. Another way to see how inseparable the conscious and unconscious are is to consider perceptual priming and subliminal perception. As is well known, if stimuli are presented very briefly they may not be consciously perceived but can still affect later conscious perceptions: memories are registered even though there was no conscious awareness of the stimuli. Increasing the exposure time will eventually cross the threshold for conscious perception, and so memories will then be formed from conscious percepts. But this is a gradual process, and it is hard to believe that an enormous threshold of theoretical recalcitrance is crossed when the stimuli start to be consciously perceived. The unconscious forming of memories (and hence priming effects) cannot be all that discontinuous from the conscious forming of memories. The same point can be made with regard to vision and attention. In the focal part of my visual field I am maximally conscious of what I see, but I also have peripheral vision in which I am marginally aware of what lies in my visual field. We can thus distinguish between degrees of conscious awareness, and it is hard to believe that theoretical recalcitrance increases dramatically with degree of focal awareness. Things are not so black and white when it comes to consciousness and unconsciousness, so it would be difficult to claim that consciousness harbors insoluble mysteries while unconsciousness does not. Better to accept that all of our mental life poses deep problems of understanding. Are we to suppose that a pain to which I am no longer attending suddenly becomes merely a brain state, while moments before it resisted physical reduction? This kind of unconscious pain is no more a candidate for physical reduction than a conscious pain, so long as it

12

Part I

really is a pain. Sliding in and out of consciousness is not sliding from a “hard problem” to an “easy problem” and vice versa. In response to these kinds of points, someone might suggest that what we are calling unconscious is really a case of consciousness: there is something it is like to have a memory or a subliminal perception or an unnoticed pain—but something it is like for someone else. In a split-brain case, the subject is not aware of what is going on in one of her hemispheres, so she is unconscious of those mental states; but it doesn’t follow that those mental states are conscious to nobody. For there might be another self associated with that hemisphere, and it is conscious of what is going on there. Similarly, it may be said, my unconscious (to me) memories may be conscious to some other self that is lurking within me. In general, all unconscious states, so called, are just unconscious to subject X but conscious to subject Y. This means that what are called unconscious mental states are all really conscious—rather as your mental states are not conscious to me but they are to you. Now one has to admire the ingenuity of this line of argument, and it would certainly undercut the position defended here—since it would turn out that the so-called unconscious is really conscious after all! But it is surely not remotely credible: can it be seriously supposed that I have two conscious selves—the usual one that perceives, thinks, and so on, and an extra one that is always fully conscious of all my memories? There is no evidence for such a second conscious self in charge, so to speak, of my memories; and the mind boggles at its putative state of consciousness—everything I have ever learned and retained is constantly before its capaciously conscious mind! No, the unconscious is unconscious period, not just relative to a particular self. There is nothing it is like for anybody to have my unconscious memories. It is worth considering higher-order thought theories here. Such theories are typically motivated by a desire to rid consciousness of its mysterious aura or reputation. Consciousness is really just a matter of having a thought about a thought (or sensation, etc.). My experiencing a red object consciously is just my thinking that I have such an experience. The natural response to this position is that it simply shifts the problem to the experience itself—there is still the question of how that can be physical, even if we allow that higher-order thoughts are physically explicable. But according to such theories, those experiences in themselves are not conscious—not until they are thought about. This implies that it is the unconscious that is problematic, not consciousness itself—that is, the first-order experiences, not the presence of a second-order thought about them. The unconscious is a mystery, but consciousness is not! I do not say that I agree with such an analysis of consciousness, but it is worth noting how the tables are turned once we view consciousness in this way. It is the mental state itself that poses the problem, whether it is conscious or not. A final objection may be attempted: consciousness is active but memory is passive—so it doesn’t have the dynamic nature of consciousness. Nothing is going on

Mind13

in memory; it is completely static. First, it is not clear why this counts against the psychological reality of memory, or how it bears on issues of materialism and the mental. But second, the assumption of stasis is false, because memory is actually a highly active system: memories change and mutate all the time. Obviously they fade with time, or become reinforced, but they can also be transformed by later experience, emotion, bias, and so forth—notoriously so. By no means are memories passive imprints on the unconscious mind; their activity behind the scenes is well attested. So there are unconscious processes of memory as well as unconscious states of memory—and these processes are no easier to understand physically than the processes of consciousness. So there is no interesting asymmetry between conscious and unconscious to be found here. Granted the psychological reality of the unconscious, particularly memory, but also innate mental states, and maybe others, we can say that the vast majority of mental states exist outside of consciousness: there is far more in my mind now than I am conscious of, or have been conscious of. If these unconscious states are as puzzling as conscious states, then consciousness is only one small part of the mind that is intractable—most of what is intractable is unconscious. If I reflect on my mind now, I can appreciate that my consciousness poses serious explanatory problems; but I should also reflect that the vast hidden ocean of my unconscious mind poses problems of comparable magnitude. I could lose my consciousness altogether and still be a mystery to science and philosophy. When God created me in my physical nature he had not yet created my unconscious mind (nor my conscious mind), and when he did create it he introduced a new level of mystery into the world. Adding consciousness to unconsciousness made little or no difference to the world’s opacity (or none that anyone has managed to articulate). It is creating mind, conscious or unconscious, that is the fundamental problem, not creating the conscious mind as such. Or at least there is a challenge to explain what makes consciousness special in the league of mental intractability.

Concepts of Mind

In our descriptions and categorizations of the phenomena of what we are pleased to call our “mental life,” a certain terminology has grown up and taken root. Thus we speak of the conscious mind, the unconscious mind, and the preconscious mind. We might allow that there are several unconscious or preconscious minds, but we cleave to the idea that there is a sharp contrast, or gulf, between conscious and unconscious— between what we are conscious of and what we are not conscious of. That epistemological distinction is thought to reflect a deep ontological division—a fundamental separation between the conscious and the not conscious. But consider an ordinary visual experience: the perception occurs at a particular time, leaving a memory of what was seen, which can be recalled later or lie dormant. The experience may flit in and out of consciousness as it is recalled or momentarily forgotten, and it is the same experience that is now conscious and now not conscious. It is not that the experience goes out of existence between moments of conscious recall; it exists in an unconscious form—that very experience. It becomes conscious while still being itself. Memories like this exist in parallel with the stream of conscious experience, while sometimes swimming into that stream. They switch from one place in the mind to another, not losing their identity as they do so. Memory in general is a separate mental existence alongside what exists in consciousness, but the barrier between the two is highly permeable. Freud called these accessible memories “preconscious” because they precede the conscious experience of remembering, but they are also postconscious with respect to conscious remembering. What I want to emphasize is that they exist simultaneously with conscious experience and that they pass smoothly between the two levels. The two systems are complementary (the same is not true of the Freudian “dynamic unconscious”).1 I propose that we call the place of ordinary memory the paraconscious (“para,” “beside” or “adjacent to”): it exists alongside consciousness and replicates it in crucial respects.2 The only difference is that we are not conscious of it—hence it is not called 1.  Freud did not hold that repressed unconscious desires actually become conscious during psychoanalysis: no one goes from sincerely denying that he has sexual desires for his mother to actually having the conscious desire to have sex with his mother. The patient may be brought to believe that he has such unconscious desires, but they don’t actually emerge into consciousness as fully conscious desires. So this is not like ordinary memory, where the unconscious literally becomes conscious. The Freudian unconscious really is opposed to consciousness. The same is true, mutatis mutandis, for the kind of unconscious postulated in cognitive science, such as early processing in the visual system. 2.  My model here is the so-called parasympathetic nervous system (PSNS), which exists alongside and complements the sympathetic nervous system. The PSNS is responsible for such functions as

16

Part I

“conscious.” But it is a full mental reality, as real as what occurs within consciousness, and intimately bound up with consciousness. It is not merely that which precedes consciousness (that could be a brain state), and it is not cut off from consciousness; it is where conscious states go when they stop being conscious, possibly only for a few seconds. Since memory is ubiquitous in mental life, informing perception, reasoning, imagination, creativity, and language, the paraconscious is ubiquitous: most of what exists in the mind exists in the paraconscious—it has far more storage capacity than consciousness. Memory is vast; what is present to consciousness is tiny by comparison. The paraconscious is an important, extensive, active, and robustly real component of the mind. We should not try to downplay it, eliminate it, or reduce it to something else (say, dispositions to consciousness). And it is certainly not unconscious in the way brain states are, since it flits in and out of consciousness; it is not something opposed to consciousness. That is why I propose a new term that does more justice to mental reality: the paraconscious is the place in the mind where mental states loiter when we are not conscious of them. If we understand that last concept in terms of attention, then we can say that the paraconscious is the place where mental states exist when we are not currently attending to them. The contents of the paraconscious are not easy to conceive. It is not a physical reality, like neurons, so it cannot be conceived under sensory concepts—it is not what appears thus and so to the senses. Nor is it conscious, so we cannot conceive of it as a mental state of which we are conscious—as what appears in a certain way to introspection. We tend to conceive of it functionally, while recognizing that it is not functionally definable: it is what bears a certain relation to conscious states, but it is thought of as fully and intrinsically mental. For example, it might be a memory of a recent conscious experience that is just about to be recalled to consciousness. There is a temptation to think of it as merely a brain state or as a disposition to cause conscious experiences, but in fact it is a mental state in its own right, just one that is not currently in consciousness. Psychologists speak of “memory traces” or “engrams,” but these are just empty labels for a type of existent we find hard to conceive. This may be the reason that its ontological status tends to be found problematic: we find it difficult to get a handle on what we are talking about. However, granting that the paraconscious exists, it opens up some interesting options and theoretical possibilities. First, consider the idea of paraconscious panpsychism. We are familiar with the doctrine that all matter contains elements of consciousness, but might we also suppose that all matter contains elements of paraconsciousness?

digestion, defecation, salivation, lacrimation, and sexual arousal. No doubt there is a part of the nervous system dedicated to the paraconscious mental system existing alongside the part dedicated to the conscious mental system.

Mind17

Matter contains mental states that it is not conscious of—and hence which are not conscious. These mental states are like memories in their ontological character, but they won’t be literally memories—unless we determine that matter contains conscious states and paraconscious memories of those states! This might deliver a more credible form of panpsychism than the kind that finds actual consciousness everywhere. Also, we might suppose that simple organisms have a paraconscious but not full consciousness, or that the paraconscious precedes the conscious both in evolution and individual development. The paraconscious is the almost-but-not-quite-conscious precursor of consciousness—how the mind is before consciousness puts its unmistakable stamp on it. Maybe it is how the mind has to be in order for consciousness to arise—its essential precondition. The mind has to first go through a paraconscious phase before it can attain consciousness proper. This suggests an intriguing possibility: all consciousness is the becoming conscious of a paraconscious mental state. When a memory is consciously recalled it becomes conscious: it preexisted conscious recall and it then undergoes a transformation to consciousness. Might this always be true? For instance, could conscious perception consist in the becoming conscious of a prior paraconscious perception? Or could conscious intention consist in the becoming conscious of a preexisting paraconscious intention (Libet)? Is consciousness always preceded and conditioned by an underlying nonconscious mental state that becomes a conscious state? If so, consciousness is always a becoming conscious: first there is the paraconscious, and then it gets transformed into the conscious (by what means we don’t know). Something must be added, but the adding is always to something already mental—as conscious remembering always is. In the case of conscious thinking we must exploit concepts that already exist in nonconscious form (either as memories or as innate ideas), so the conscious thought is a becoming conscious of these preexisting mental entities: it is the paraconscious rising to the level of consciousness. Thus consciousness never arises directly from nonmental antecedents (as it might be, brain states) but always arises from a preexisting mental substrate (the paraconscious). The paraconscious therefore has a kind of primacy. A further consequence is that we really have two mind–body problems, not one. There is the problem of consciousness (which itself can break down into subproblems), and there is the problem of the paraconscious—and also the third problem of how the conscious and the paraconscious are related. Thus we want to know how brain states give rise to consciousness and we want to know how brain states give rise to paraconsciousness. Solving one problem will not automatically solve the other, though it may help. A natural way to think about this is that the brain first produces the paraconscious and only then builds on that to produce consciousness: the brain produces the latter by producing the former. It is a two-step process. The point I want to emphasize here is that the paraconscious poses its own mind–body problem, distinct from that of consciousness.

18

Part I

It would be fair to say that the picture I am working with follows the spotlight model of consciousness: there are all sorts of items dwelling in the mind of which we are not conscious, each as real as any denizen of consciousness, and consciousness consists in singling some out for attention and shining a light on them. It is not that the mind is made of consciousness through and through. The concept of the paraconscious is intended to do justice to this picture by regarding the mind as running on parallel adjacent lines. To call the nonconscious part the “unconscious” is to fail to do justice to its proximity to the conscious: it is not the negation of consciousness but its indispensable partner. Many things are unconscious in the weak sense that they are not conscious (most of what happens in the body), but some things that are not conscious are tightly intertwined with consciousness; and it is these I wish to label “paraconscious.”3 3.  I would say there are two levels of psychological vocabulary: words for specific functions and phenomena such as “see,” “smell,” “think,” “remember,” and so on; and more general words like “mind,” “psyche,” “soul,” “consciousness,” “the unconscious,” and so on. The latter category is much more fraught and disputable than the former, and of more recent vintage. Thus the general categories we deploy to talk about the mind (including the category mind) are far more open to revision than the specific concepts we use to talk about the mind. The way we have come to talk about consciousness and its absence needs to be scrutinized carefully; for it can lead to inadequate and mistaken ideas about how the mind (psyche, soul) is organized.

Knowledge and Emotion: An Untenable Dualism

Traditional philosophical psychology distinguishes between the so-called cognitive and affective faculties. The two faculties are conceived as belonging to different provinces of the mind. They may interact, but they comprise distinct kinds of mental entity, uneasily coexisting within the psyche. Cognition and emotion involve separate kinds of psychological reality, variously called Thought and Feeling, Reason and Passion, Knowledge and Affection. Cognition is often supposed to need protection from disruptive emotion, and emotion is regarded as essentially nonrational. Thus our psychological makeup has two levels or compartments, each insulated from the other, with no intermingling. We have the idea of a dualism of the cognitive and the affective. Once that dualism has been established, we can formulate philosophical theories that treat one or the other faculty as primary, or invoke one faculty rather than the other in explaining an area of philosophical interest. Thus it will be said that man is a rational animal (Aristotle), not an emotional one, or that reason is our essence while feeling is adventitious (Descartes). Rationality connects us to divinity. By contrast, it may be claimed that rationality is a capitalist or patriarchal myth, with our being subject to political or emotional forces, which we may prefer to ignore. Debates then rage about whether human beings are governed more by reason or more by passion. Psychologists like Freud tell us that we are less rational than we suppose, being influenced by all sorts of emotional factors that escape conscious awareness. Defenders of Reason point out that thought is subject to logical norms that free it from subjugation to arbitrary emotional promptings. Both sides agree on the cognitive–affective dualism; they just differ over which side runs the show. Likewise, we find such doctrines as emotivism in ethics and aesthetics: moral or aesthetic utterances are not expressions of knowledge (hence not “cognitive”) but rather expressions of our emotions (hence “noncognitive”). The emotivist doctrine can be extended into a more general “sentimentalism” that seeks to view other mental phenomena as noncognitive rather than cognitive—even to the point of claiming that knowledge itself is an affective matter (noncognitivism about cognition). Cognition turns out to be a species of Passion (as it might be, enthusiastic whole-hearted assent to a proposition). Contrariwise, the rationalistically inclined philosopher (or psychologist) might maintain that the so-called affective faculty is reducible to the cognitive faculty—as it might be, knowledge about the value of things. Love, for example, is belief in the virtues of the beloved, while anger is belief that one has been wronged. This doctrine would be the analogue of cognitivism about perception, mental imagery, and even bodily sensations—all these are just species of cognition (judgment). Passions and perceptions are really cognitions; thought is the universal bedrock of the mind. According to this type of view, the sensation of pain is the thought that one is in pain—the self-ascription of the concept of pain. We can even have a cognitivist view

20

Part I

of consciousness itself: to be conscious is to have thoughts about one’s own thoughts. In ethics it would be the theory that moral emotions are really moral cognitions—the cognitivist theory of the moral passions, in contrast to the emotivist theory of moral judgments. But how solid is the traditional exclusive distinction between knowledge and emotion? Are these really polar opposites? I think it is not solid at all, so this whole way of thinking is radically misguided, as are the doctrines that are based on it. What has happened is that we have reified abstractions: we have converted a linguistic dualism into an ontological dualism. We have committed what might be called “a concept– reality confusion” (like the use–mention confusion): that is, we have misconstrued the distinctness of our concepts for a distinction in the things to which they refer. True, “know” and “feel” are not synonyms, but it by no means follows that they denote separate things. The dogma of separation does not follow from the duality of our concepts. The correct view is that the cognitive and affective are inseparable from one another, joined like Siamese twins, aspects of the same basic reality. We might think of this as the “double aspect” theory of knowledge and emotion: knowledge has an emotional aspect and emotion has a cognitive aspect—both things fall under both sorts of description. Thus knowledge is not opposed to emotion and emotion is not opposed to knowledge: each involves the other inextricably. We have mutual infusion, not mutual exclusion. We have been blinded to this elementary truth by a venerable tradition, tied to religion, and backed by an act of linguistic reification. Why do I say that? I am by no means the first to suggest that emotions have a cognitive dimension—indeed, the suggestion is commonplace in theories of emotion. Even if anger is not reducible to the judgment that one has been wronged, it surely involves such a judgment. Even if love is not just favorable opinion, it surely presupposes it—one does not love a person one despises. Even sexual emotions involve cognitive elements, such as the experience of discovery or the revelation of secrets—hence we speak of “carnal knowledge.” The emotions involved in aesthetic experience are clearly strongly imbued with cognitive content. In ordinary mature humans, at any rate, the emotions are infused with cognitive materials—thoughts, ideas, and perceptions. Our emotions thus reach high levels of sophistication and subtlety, and we cultivate our emotions by means of our cognitive powers. It is not that our emotions seethe and surge in one isolated corner of the mind, insulated from all external influence: they are shaped in their very essence by our so-called cognitive faculties—they could not even exist in the absence of the thoughts that shape them. Indeed, emotions are so suffused with thought that it is difficult to understand theoretically how exactly they differ from thoughts—though they assuredly do. However, the counterpart thesis for thought and knowledge is far less widely accepted, though it too seems obvious on reflection—the thesis, that is, that knowledge is inextricably bound up with emotion. Let me invoke Aristotle’s authority on

Mind21

the point when he says: “All men by nature desire to know. Thus, the senses are loved not only for their usefulness but also for themselves. Sight is loved best of all, for, of all the senses, it is the one that brings the most knowledge” (Metaphysics, Book 1). If human beings desire to know, then knowledge is the satisfaction of a universal human desire—and as such will carry an appropriate range of affect. Humans also desire food and sex, and the satisfaction of these desires will produce the corresponding affect— contentment, happiness, tranquility, and the like. We may therefore say that knowledge is the satisfaction of a desire: the desire has knowledge as its object, and attaining that object produces satisfaction and its affective corollaries. In fact, Aristotle somewhat understates the point by speaking of “desire,” since not all desires are very pressing (I desire a walk around the block, but not very intensely). We can also say that humans actively seek knowledge, long for it, languish without it, need it, feel thwarted if they do not obtain it, and in some cases make it the central purpose of their lives (scientists, scholars). Aristotle hits the right note when he remarks that the senses are loved for their ability to bring knowledge—because we do indeed love knowledge. It may be the highest reaches of scientific or philosophical knowledge, or it may be just local gossip and sports trivia, but people really love to know things and will go out of their way to acquire knowledge. Why do people love books? Because of all the knowledge contained in them, of course! People delight in learning things, they savor the knowledge they have acquired, and they revel in the extent of their erudition. They enthusiastically undertake to learn about new subjects—often quite difficult subjects—and they experience joy when they succeed. If that were not so, it is hard to see how they would be as motivated as they are to acquire knowledge. If we just coldly sought knowledge for purely practical reasons, or because we had been instructed to, we would not be motivated in the urgent way we manifestly are. No, we actively yearn for knowledge and thrill to its acquisition. We feel pride in what we know and shame at not knowing. We particularly prize the state we call “understanding”—when we finally get to the bottom of something, or make sense of something that has puzzled us. We are also transported by the creation of new knowledge—by discovery. Knowledge is exciting, uplifting, fascinating. These are the “epistemic emotions”—those that cluster around cognitive activity. They are integral to the search for knowledge and to its achievement. They are part of the human psychology of knowledge. It may be objected that not all knowledge involves flights of epistemic ecstasy; some is quite mundane, for example, the knowledge that it is raining outside. That is true enough, but not all emotion is flaring and soaring either—some emotions are quite mundane, for example, a general feeling of tranquility or a vague background unease. Fear is never far away from human consciousness, but it may be peripheral and low-wattage—part of being alive and alert to danger. Mundane knowledge may therefore come with low-key mundane emotion—such as that the rain today is a bit depressing. Not all knowledge is welcome, and some may occasion mild discomfort,

22

Part I

just as routine perceptual knowledge may carry only a hint of pleasure or interest. Still, an affective charge always seems to be hovering somewhere, since hardly anything is a matter of complete indifference. Even a judicial disinterestedness carries some emotional baggage—the feeling of cool detachment or sense of fairness. In any case, emotion is characteristic of knowledge—part of its natural phenomenology. Cognition is not an emotion-free zone, pursued without passion or purpose. It sings its own emotional tune. We can therefore say that knowledge is, or involves, an affective state, just as emotion is, or involves, a cognitive state. What falls under the description “knowledge” also falls under the description “feeling”—the referent is a double-aspect thing. Here it is tempting to fall back on a kind of mental chemistry: that such complex mental states are really compounds or composites of more primitive elements. Thus knowledge is made of a cognitive part and an affective part, and similarly for the emotions. Atomic mental components combine to produce the mental molecules we experience—and these atoms are purely one thing or the other. We thus resurrect the dualism of mental faculties. But this is a bad way to think: knowledge and emotion, as we experience them (as they are), are not composite entities with one part purely cognitive and the other part purely affective. There are no such separable atoms or parts or dimensions: if you try to subtract one element, you are left not with the other—but with nothing. Anger without a judgment of wrongdoing is simply not anger (or any other emotion); knowledge that does not spring from a desire to know is not really knowledge—not the very state that we have when we know something (compare insight, discovery, illumination, revelation, enlightenment). The identity of the mental state depends on its place in the network of mental states and dispositions. Knowledge and emotion are blends, not mixtures; they cannot be resolved into separable parts. For example, when I come to understand something after a long struggle and experience a feeling of enlightenment, we cannot suppose that my mental state resolves into separable components of cognition and affection: my understanding simply is a feeling of (justified) enlightenment. Similarly, my anger at someone just is the feeling I have when I reflect on her unjust treatment of me. It is not that I have the feeling of anger and then add that feeling to my judgment of injustice, like two peas in a mental pod. Nor is the joy of knowing something an emotion you could feel absent that knowing—just a free-floating aura, as it were. Knowing joyfully is not knowing and feeling joyful, like wearing a hat and a tie. We feel knowingly and we know feelingly. It is therefore a mistake to think in terms of the fire of emotion and the ice of reason. Emotion is icy fire and reason is fiery ice. There is no sharp dichotomy, no rift or split. So we cannot use one side of the putative division to explain the other side. We cannot reduce reason to emotion or emotion to reason, since reason is already emotion-infused and emotion is already reason-infused. One cannot be an “emotivist” in ethics or a “sentimentalist” in epistemology if that involves claiming that emotions are ontologically

Mind23

separate from, and conceptually prior to, cognitions; and one cannot be a “rationalist” about the emotions if that involves claiming that knowledge is ontologically separate from, or conceptually prior to, emotions. Emotion and reason presuppose each other; they take in each other’s laundry; they symbiotically support one another. There is just no clear and clean distinction between the cognitive and the affective, at least at the level of ontology. Consider an animal’s knowledge that it is about to be attacked by a predator: it would be quite wrong to think that its psychological state could be broken down into a strictly cognitive component (“The lion is about to bite my neck”) and a strictly affective component (“My stomach is in knots”); rather, the animal is in a cognitive-cum-affective state (“I am doomed!”). Talk of “reason” and “emotion” is really artificial abstraction, not reflective of life on the ground. We can “abstract away” from the cognitive or affective aspects of a given mental state, but that doesn’t mean that the state itself breaks down into separable parts. There is no pure faculty of reason and pure faculty of emotion, existing separately, but sometimes colliding in the hurly-burly of psychological life. There are, of course, things like thoughts and emotions, but that does not imply that each has no role to play in shaping the other. When we argue with someone about the appropriateness of his or her emotion, we presuppose a rational cognitive dimension; and when we say “I am thinking of you” or “You should be more thoughtful” we are clearly alluding to emotional matters. It is better to conceive of knowledge and emotion, not as twin psychological streams flowing in proximity to each other, sometimes touching and sometimes miles apart, but as a single stream that can be viewed from different angles. There is therefore something misleading about the label “cognitive science” and the picture of the mind it promotes. For there can be no such thing as a science of the “purely cognitive”: that is not how the human mind works, or how its states are constituted. Cognitive science must also be affective science. Insofar as it is not, we need an “affective revolution” just as we once had a “cognitive revolution”; or better, we should abandon these labels and the abstractions they encode. We certainly cannot expect to complete our cognitive psychology and then move on to our affective psychology (or vice versa). The computer model no doubt abets the dogma of separation, since we don’t tend to think that computers have feelings, while finding it easier to attribute cognitive states to them (memory, computation, even belief). But cognition in the absence of emotion is not human (or animal) cognition. Any good psychology will need to integrate knowledge and emotion, not try to hive one off from the other. The model of abstract computations, proceeding without affective content, cannot explain normal human psychology, in which the cognitive and the affective are deeply intertwined and inseparable. At best such models give us one aspect of a typical mental process, while ignoring the rest. It’s rather like describing colors in purely abstract structural terms, while ignoring the colors themselves. Cognition is affectively colored right down to its roots. We might try to idealize mental processes in our theories, so

24

Part I

that the cognitive and affective are treated separately, but in reality it is inextricability all the way down. These points apply both to the conscious mind and to the unconscious mind. Within consciousness, knowledge and emotion cling tightly together, each suffusing the other: there is phenomenological fusion. But there is no reason to doubt that the same is true of the unconscious, once we accept that the unconscious can be both cognitive and affective. If the unconscious contains desires, whether repressed or otherwise, then it will contain emotions, and these will connect with the cognitive contents of the unconscious. It is certainly so with dreams, in which knowledge and emotion are joined at the hip: we might aptly think of dream content as affective conception—seeing the world through emotional lenses. The unconscious mind, like the conscious mind, combines feeling and thinking into seamless psychological wholes. It isn’t easy to free oneself from the dualistic picture of knowledge and emotion. There is a long and entrenched tradition supporting it, and our ordinary ways of speaking reinforce it; also, we are generally prone to dualisms of all sorts. But it is surely not true to our mental life as we experience it: the inextricability of thought and emotion is evident to us from within. Our thoughts are tinged with emotion, and our emotions are shot through with thought. As Aristotle says, we desire, seek, and love knowledge— and that fact configures the very nature of knowledge. Knowledge, to repeat, is desire satisfied—epistemic desire. And it is equally true that human emotion could not exist without the cognitive sophistication that we bring to it: emotions are not like raw sensations in the gut—they are constructed around thought and judgment. This is how things are de re—even if our habitual modes of description tend to suggest ontological separation. It may well be true that some knowledge is more emotionally charged than other knowledge, and that some emotions are more cognitively constituted than other emotions; but this kind of gradation should not lead us to suppose that we are dealing with separate faculties or facts. In the normal course of mental life, we don’t alternate between one faculty and the other—as if we switch off emotion while thinking and switch off thinking while emoting. Rather, the two work intimately together to produce cognitive-affective hybrids. To put it as baldly as possible, my state of knowing (say) that the theory of evolution is true is an emotional state, as well as a cognitive state; and my state of being angry with a person X is my being in a state of knowing, as well as being in an affective state. Knowledge and emotion are two sides of the same coin. Accordingly, the philosophy and psychology of each cannot realistically ignore the other.

The Second Mind

It has been known for some time that the human body contains a nervous system in the bowels. This is known as “the second brain.” It consists of hundreds of millions of nerve cells embedded in the intestines, which operate independently of instructions from the brain in the head. You can sever the nerve connections between bowel and brain and the bowel will still operate normally. There are receptor cells that respond to pressure from inside the intestines and there are effector cells that cause contractions in the bowel (peristalsis). The system is very old phylogenetically and is widespread among animals. It enables the gut to go about its business separately from the main brain, which has obvious advantages. Michael Gershon, a leader in the field, writes: “The enteric nervous system is thus an independent site of neural integration and processing. This is what makes it the second brain. The enteric nervous system may never compose syllogisms, write poetry, or engage in Socratic dialogue, but it is a brain nevertheless. It runs its organ, the gut; and if push comes to shove (as it does in the millions of people who have had their vagus nerves surgically interrupted), it can do that all by itself” (The Second Brain, 17).1 He also tells us: “The enteric nervous system can, when it chooses, process data its sensory receptors pick up all by themselves, and it can act on the basis of those data to activate a set of effectors that it alone controls. The enteric nervous system is thus not a slave of the brain but a contrarian, independent spirit in the nervous organization of the body” (17). This noncranial second brain employs neurotransmitters, including serotonin, just like its higher altitude brother, and its constituent cells have the same kind of anatomy and physiology as those of the cranial brain. The second brain is thus essentially the same kind of organ as the first brain: a receptor-effector information-processing system consisting of a great many neural units—with axons, dendrites, synapses, the works. It exists in the stomach area, not in the head, but it is still a bona fide brain. At one point, discussing the extent of new knowledge about the second brain, Gershon even refers to “the mind of the bowel” (20). This raises intriguing philosophical questions. We can envisage a philosophy of the second mind. Given the anatomical and functional similarities between the first brain and the second brain, it is reasonable to suppose that the second brain gives rise to some sort of mind—just as we suppose that other animals with brains have minds too. This is a complex and subtle nervous system, functioning autonomously, so it may well have a mind of its own. We can therefore ask: What is it like to be an intestine? Some sort of sentience presumably attends the second brain’s use of its intestinal pressure detectors—there is some way things seem to it. It feels the pressure and as a result acts 1.  Michael Gershon, The Second Brain: A Groundbreaking New Understanding of Nervous Disorders of the Stomach and Intestine (New York: HarperCollins, 1998).

26

Part I

appropriately—by initiating peristaltic action. We often feel as if our bowels “have a will of their own”—well, maybe they do, literally. They sense and act, according to their own needs and job description. Presumably there are intestinal qualia. These qualia may or may not be like those delivered by any sense seated in the first brain— they may differ from the sort of feeling of pressure we have when touching an object or the feeling of a full stomach. We may, in fact, never know what it is like to be an intestine, though there is something it is like. The intestine has alien subjectivity. Presumably the second mind, in addition to being sentient, also has intentionality. It contains states that are about something—the condition of the bowels. It must also have something corresponding to intentions, at least in the sense that it performs actions as a result of its perceptions. So the philosophy of action would apply to its operations. Is there any distinction between its intentional actions and its subintentional actions? Does it involve trying to move the bowels? Is it capable of weakness of the will? Does it act for reasons? Does it plan ahead? We should be wary of thinking of it as a mechanical reflexive system, since it must exercise discretion about its actions and make decisions about exactly when to contract. It is no use expelling material from the intestine until the nutrients have been properly absorbed, and the chemical contents of the blood and tissues may affect what nutrients it works to extract. One can certainly envisage a very complex and subtle process of perception and decision, calibrated to the general nutritional state of the body. It is no easy matter to decide what to digest and when, how long to retain food material, and the moment to let go. It may be a biological advantage to have a thoughtful gut—to be gut-smart. We used to think that language was a simple mechanical system of stimulus and response, until the cognitive revolution taught us otherwise; maybe the second brain has a similarly complex internal structure and does not operate by means of responses elicited by stimuli according to simple laws. Maybe, indeed, the enteric nervous system contains its own language of thought—a symbol-manipulating system all about the secrets of good digestion. It certainly gives every appearance of being an information-processing system. There will be a mind–body problem for the second brain (Cartesian dualism, anyone?), as well as the problem of meaning for its representational states. Is there also a free will problem? Does this agent think of itself as free? Is it capable of self-awareness? It is hard to answer these questions, but it would be dogmatic to rule such things out, given the evident braininess of the second brain. This is a very sophisticated system performing a very sophisticated task. No machine has been built that can digest food like the average mammalian digestive system. There is, as yet, no convincing AD (artificial digestion). The enteric nervous system has evolved over many millions of years, and it must act with flexibility and subtlety: it may be much more minded and intelligent than we suspect. The reason we tend to dismiss its claims to advanced mindedness is that we have no access to it mentally—we just observe its external operations.

Mind27

And this behavior is itself mainly invisible. Nor do we interact socially with our intestines, so we have no need to evolve ways of reading their mind from their behavior. There is thus a severe problem of other minds with regard to the second mind. It might be that we house a second mind of remarkable richness and inner complexity but that we are unable to appreciate its nature. What if the second brain had turned out to be as complex as the first brain? We would still have no way to access its inner psychological landscape and would no doubt suspect it to be minimally endowed. But that would betray psychological chauvinism, lack of imagination, and cerebral discrimination. The second mind exists within our body as an isolated center of psychological reality, to which we have no more access than to the mind of a bat (in fact less, because at least the bat behaves in ways we can interpret using standard folk psychology). There is a mind in your gut of which you are blissfully unaware, even though it is vital to your survival and well-being. Maybe we should be thanking it, not dismissing it. (It dies too.) Are we then made up of two selves? Are there two sets of beliefs and desires in us, as well as two sets of emotions, cognitive abilities, and so on? That is not to be ruled out. Then we will have questions of personal identity about the enteric self, and questions of epistemology about the beliefs of this self, and so on. The full range of philosophical issues about the mind will carry over to our second mind. And the same for psychology: Is there an enteric unconscious? Are there innate intestinal ideas? Is there any digestive learning? Does the second mind know that there is a first mind? Is there anything akin to child development? How does the enteric mind build up its picture of the world? Can it conceive of anything outside the gut? Does it sleep and dream? Is it neurotic? Is it prone to mental breakdown or illness? Does it have a personality? Does it enjoy its work? We can ask these kinds of questions about minds other than our own and even provide answers, and in principle we can do the same for the mind that lurks in our bowels. Maybe it is not a very intelligent mind, as we define “intelligence,” but there might still be an interesting psychology to explore down there. It seems unlikely that it will resemble the mind of any whole animal, even a worm or snake, but it might surprise us with its unique form and mode of operation. It used to be said that some dinosaurs had a second brain near their tail, though this is now regarded as mistaken; still, the notion is not intrinsically absurd and a second mind could come along for the ride. It turns out that the dual-brain anatomical setup is more common than we thought, and begins at home. Most animals in fact have two brains, cranial and intestinal, and presumably two minds to go with them: one dedicated mainly to the skeletal muscles, and the other dedicated to the muscles involved in digestion—with an accompanying sensory apparatus and central processing system. When we speak of human psychology, then, we must, strictly speaking, clarify which human psychology we are referring to. What I find fascinating is the idea that all along

28

Part I

I have had another psychological being living inside of me—not as a parasite but as a symbiotic partner (I help him too). My gut has been supervised and run by an intelligent being of sorts, with thoughts and feelings of its own. It is just that I will never get to know this being—not unless huge strides are made in enteric psychology. I suppose that brain scans might in principle be employed on the second brain, enabling us to learn more about this particular other mind (do its pleasure centers light up when digestion is going well?). As things are, I have to accept that I am a divided creature in possession of two minds, one intimately known to me, the other not known at all. Given Kant’s principle of respect for persons, I should respect my bowels more. Maybe my gut is not the reflective rational being that I am, but it may be more interesting than I ever gave it credit for. Maybe one day we could even be friends.

Awareness of Time

The other day I decided to find out more about time. I lay down, closed my eyes, and tried to focus on time. I concentrated on the present moment, hoping to learn something about the nature of time, keeping everything else from my mind. It was like trying to find out more about the color red or the nature of consciousness by intently focusing on it. I needed to attend closely to time, as I spontaneously perceive it. Given that I am aware of time, and can direct my attention to it, I should be able to discern something of its nature—as I can direct my attention to space and grasp something of its nature. I would use my basic acquaintance with time to discover things about it. After a few minutes I realized I was coming up with nothing. I wasn’t learning anything about time, and the object of my efforts seemed maddeningly elusive. My supposed basic acquaintance with time was yielding no insight into it whatever. I wondered why the effort was proving so futile (I invite my reader to perform the same experiment). It was just as if I was not succeeding in focusing on time at all. Then the thought struck me: maybe I am not aware of time. Maybe I do not take time as an intentional object—it is just not an object of my apprehension. I am aware of matter, space, my own consciousness, and maybe some other things, but I am not aware of time. Time exists, but I have no direct awareness of it. That is why my efforts to concentrate on it were leading nowhere: it was not the intentional object of my mental acts. This struck me as a radical thought—and not something I would have entertained before. So I must have had a strong tendency to believe that I am aware of time. Given that I am a fairly representative human being, this would suggest that human beings are not aware of time and yet they tend to believe they are. That would explain why I thought I could learn about time by focusing on it and yet could learn nothing. Why didn’t I just use my senses to try to learn about time? Why did I lie down in a quiet place, close my eyes, and concentrate with my mind? Because I didn’t think that I could perceive time with my senses: I didn’t think time was visible or audible or touchable. Space and matter, yes, but not time. So I (we) don’t think that time is an object of sensory awareness: for how could it impinge on the senses, stimulate the receptors? I therefore needed to focus on time with a special sort of faculty—my “temporal awareness faculty.” This I must have conceived as being like the introspective faculty—hence the need for sensory blocking. But surely that was a strange assumption on my part: why should I think that I possess a “temporal awareness faculty”? Do I really have a sixth sense dedicated to revealing truths about time? My introspective faculty might be viewed as a sixth sense dedicated to revealing truths about my own state of mind, but why think that I have a further faculty that is specifically geared to time? The idea of such a faculty seems like a myth—I simply have no such faculty. I have no sense organ that picks up information about the nature of time. But then, with what might I hope to be aware of time? If I am aware of time, then I must have a faculty that enables

30

Part I

me to be aware of it; but I have no such faculty, so I am not aware of it. I do not sense time—detect it, resonate to it. I have nothing to sense it with. So I can’t use my putative faculty of perceiving time in order to discover truths about time. That doesn’t mean that time doesn’t exist: it might exist noumenally, that is, as an object that I cannot apprehend. As Kant would say, I have no “intuition” of time. Nor does it mean that I cannot refer to time, or think about it. We can refer to and think about things that we are not immediately aware of. We can have these cognitive relations to things that we cannot directly sense. I take it that we can think about and refer to time: “the past,” “this moment,” “yesterday,” “the duration of the Second World War.” It is just that we have no acquaintance with the objects of such reference; as Russell would say, our knowledge of time is only “by description.” When I refer to the present moment as “now” my words mean something like “the time at which this is going on,” where I demonstratively refer to some event in time. I sense the event and then I refer to its time of occurrence by forming a description that embeds reference to that event. But I don’t refer to or think about the time directly, that is, without such referential mediation. I cannot look at time, or scrutinize it, or try to get a better view of it. Time isn’t something that appears to me; I have (to use Hume’s expression) no impression of time. Of course, I am aware of clocks and can sense their operations. This is how I know how much time has elapsed. But clocks and time are not the same, the former being merely a measure of the latter (clocks tick, but time doesn’t tick). I know quite a lot about time, of a structural and abstract nature, but I don’t have the kind of knowledge of time that arises from direct acquaintance. It is also correct to say that I am aware of the passage of time, but again that is not the same as saying that I am aware of time itself. I am aware of time’s passage because I am aware of events and change, but as to the medium in which events and change occur, of that I have no awareness. I do not perceive moments succeeding moments, but only events succeeding events. I have no sense data of time as such: no sensory experience presents time to me. There is nothing like a moment of time looking or sounding a certain way to me. Can we even conceive of a faculty of time perception? How would it work? How could time impinge on our temporal receptors? What kind of anatomy would the organ have? It isn’t as if some animal species have a temporal sense that we lack, like echolocation. No animal can sense time directly, though they are no doubt aware of the passage of time and can even read natural clocks (like the apparent motions of the sun). It is not that some remarkable deep-sea fish is privileged to have time appear to its specially evolved time sense. Sensory awareness of time is impossible, not merely contingently lacking (so it is not like the invisibility of atoms). Given this peculiar epistemology, it is not surprising that time should be so perplexing to us. We grasp certain properties of time, but we don’t have immediate knowledge of its intrinsic nature. Our knowledge of time is like a blind person’s knowledge of

Mind31

color: structural and relational, not intrinsic. Time therefore strikes us as systematically elusive—we know it is there, but we cannot grasp its inner being. It thus differs from space, which is everywhere evident to us, even if its objective character is other than we suppose. You can see space, even empty space. Time is built into everything we experience and yet it is puzzling: we might even say that it is mysterious (it seems “occult”). It is not as if science has asked what time is and then has gone on to give the answer (“Water is H2O,” “Time is XYZ”). Time is not like a natural kind whose real essence we have empirically discovered. We don’t have anything analogous to the atomic theory of matter for time. We cannot peer into time, or put it under a microscope, or dissect it. It is true that we divide time into instants, but that is more of a mathematical convenience than an empirical discovery: we didn’t discover it by means of close examination or by bombarding time with particles to reveal its internal structure. Hence time is a subject of philosophical perplexity, not a subject of routine empirical science. Maybe from a God’s eye perspective time would appear less mysterious, more “natural.” But from our human perspective it strikes us as inherently “queer” (as Wittgenstein would say)—a sort of impalpable ghostly presence. We keeping asking ourselves, “What is time?” and no answer is forthcoming. But if we are simply not aware of time, such perplexities are predictable: we see it through a glass darkly, or rather we don’t see it at all. Why did I start out thinking that I could discover the nature of time by attending closely to it in immediate experience? If I am simply not aware of time, shouldn’t this fact have been evident to me? The answer presumably is that time is so familiar to us, and so much part of our everyday conceptual framework, that we tend to think we know more about it than we do. The idea that epistemic limitation could lie so close to home, in our most basic categories, seems like an insult to us as knowing beings; for we are incurable epistemic optimists—as I was when I lay down to concentrate and discover the nature of time. But the truth appears to be that we are cut off from some of the most fundamental aspects of reality. In the case of time, the thing-in-itself transcends our faculties of awareness. A moment is not the kind of thing we can sense.

Mind–Brain Identity Theories

I will articulate some perennial dualist intuitions surrounding claims of mind–brain identity. This will require going over familiar ground, but I hope with some new formulations and dialectical twists. First, I will discuss type identity theory, then token identity theory. Type identity theory is usually explained by reference to standard examples of theoretical identification. Thus “Pain is identical to C-fiber firing” is said to be analogous to “Water is identical to H2O” or “Heat is identical to molecular motion.” The critic will object that the cases are not analogous, because we can pull apart pain and C-fiber firing, but we can’t pull apart water and H2O or heat and molecular motion. We have an intuition of distinctness or contingency or irreducibility in the pain case, but not in the other two cases. The identity theorist will reply that the critic is confusing pain and the appearance of pain: we can pull water (H2O) apart from the appearance of water and heat (molecular motion) apart from the appearance of heat, and similarly for pain (C-fiber firing) and the appearance of pain—so the cases are analogous after all. The critic will respond that there is no gap between pain and the appearance of pain, while there is a gap between water and the appearance of water or heat and the appearance of heat. The identity theorist will now have either to give up the analogy or dig in his heels and claim that there is such a gap for pain. But then, the critic will persist, what about the appearance of pain—isn’t that also supposed to be identical to a physical state? So the identity theorist will have to apply a type identity theory to the appearance of pain too. But the critic will protest that he intuits that these two things are also separate and distinct. At this point the theorist will have to admit the lack of analogy with the heat and water cases, but will insist that the critic’s intuitions are misguided, since the appearance of pain (and hence pain) just is C-fiber firing according to his theory. We now seem to have reached a stalemate, though the critic has the dialectical advantage, because of the admitted failure of analogy with standard cases of theoretical identification. What he will claim is that his intuition that pain (or its appearance) is not C-fiber firing is exactly the same as his intuition that the appearance of water is not identical with H2O (i.e., water). Given that the identity theorist is a dualist about H2O and its appearance, not an identity theorist, he must adopt the same view about C-fiber firing and pain, according to the critic, since the intuitions are exactly the same in the two cases. The onus is therefore on the identity theorist to defend the claim of identity in the one case and not in the other. The critic is saying that the world could contain C-fibers without pain and pain without C-fibers, so that the two cannot be identical; but that it could not contain H2O without water or water without H2O—though it could easily contain H2O (water) without the appearance of water, say, by not containing any sentient beings. The identity theorist is driven to accept that he has to deny that intuition, holding that necessarily

34

Part I

C-fibers and pain (and the appearance of pain) must go together, despite what intuition suggests. Since his view is that pain just is C-fiber firing, there simply cannot be one without the other. The critic meanwhile keeps insisting that that claim is implausible: he thinks that (an appearance of) pain cannot have an essence constituted by C-fiber firing, though he readily accepts that water can have an essence constituted by H2O. He thinks that pain has one kind of essence and C-fibers quite another, while his opponent thinks that there is only one essence here, namely, the essence of C-fiber firing, which (he holds) is identical to the essence of pain (or its appearance). We then seem to reach dialectical deadlock. Is there a principled way to budge the identity theorist from his intransigence? Consider a new type of identity theorist: she identifies the appearance of water, not with a sensory state of a sentient being, but with an external physical property, namely water itself. She is not an internal materialist about appearances, identifying them with states of the nervous system; she is an external materialist who identifies the appearances of things with the physical properties of those things. She thinks, in short, that the appearance of water is identical to water, which is H2O. Her view is like the view of those who take the appearance of color to coincide with color itself (a property of external objects), and then reduce colors to wavelengths: color appearances thus turn out to be identical with physical wavelengths. Appearances are not in the head, according to the external materialist, but in external objects. This position is externalism about appearances combined with materialism about the external world—not “central-state materialism” but “external-state materialism.” When appearances become objects of awareness, what happens is that an external physical property becomes the intentional object of a sentient being: but this object is entirely physical. Once we have a materialist account of the intentionality relation, we will have a fully materialist account of sensory appearance. So water is H2O and the appearance of water to a perceiver is just that property becoming the object of a sentient being’s awareness. Thus appearances are physical too: they are identical to the physical properties they are appearances of. The critic might insist that he intuits that the appearance is not the same as the external physical property, being irreducibly mental; but the external materialist type identity theorist will respond that this begs the question, since her view is precisely that there is an identity here. When it comes to things like pain her view will be that the experience of pain that you get from being stuck with a sharp nail is just the sharp nail itself—that is the intentional content of the sensation, and sharp nails are just physical objects. Pain is directedness to an external pain-causing object: the so-called qualitative feel of pain is just the painful stimulus itself. Pain is not identical to C-fiber firing, on this view, but to a property of the external stimulus (more exactly, its qualitative content is identical to a property of the external stimulus). On a variant view, pain (its qualitative content) is identical to a bodily wound, not to the wounding external stimulus, and the wound is something physical.

Mind35

But the critic will shrilly and vehemently insist that such a view is deeply implausible, since it is clearly possible to pull these things apart: we could have external properties without appearances and appearances without external properties—for example, water without the appearance of water and the appearance of water without any water. For the world might not have contained any sentient beings, or we might all be brains in a vat. That sounds entirely correct, but the external identity theorist might dig in her heels, even more shrilly insisting that these so-called intuitions are misguided and anyway beg the question—for her view is precisely that no such things are possible, since appearances are external properties. She thus bites down extra hard on the bullet. But surely this is one bite too many: she has embraced as an acceptable consequence what is in fact a reductio ad absurdum. The intuitions adduced by the critic are entirely persuasive, and they defeat the external identity theorist; only a stipulation about the word “appearance” could justify this kind of theory. Sensory appearances are manifestly not identical to external physical properties: my sightings of water are not reducible to water, and my pains are not reducible to nails. You cannot be a credible materialist by identifying appearances (“qualia”) with external physical states of affairs, because one can clearly be pulled apart from the other. But now isn’t the internal identity theorist in much the same weak dialectical position? We have a strong and stubborn intuition, not easily dismissed, that appearances and internal brain states can be pulled apart, so that they must have different essences. Water and H2O indeed share an essence, but pain and C-fiber firing do not, which is why we can so readily conceive them separately. Merely insisting that this intuition must be discarded seems no better than the corresponding move in the external materialist case. The basic intuition is the same in all these cases: namely, the appearance of water can come apart from water (H2O), and the appearance of pain (i.e., pain) can come apart from C-fiber firing—we must therefore have a dualism of essences in both cases. We may call this the “single essence problem” for type identity theories: there is only one essence to play with according to such theories, but it is unable to explain the manifest independence of subjective appearances and brain states. Since the brain states are supposed to constitute the essence of subjective mental appearances, we should not be able to pull the two apart; but we can pull them apart, so there cannot be just one essence here. If there were nothing more to the experience of pain than C-fibers firing, then we should not be able to drive a wedge between the two—but we can. So the case is really just like the relation between water or heat and their subjective appearances—two distinct things that are contingently connected. A type identity theory here would be clearly absurd. Suppose we were actually to introduce a term for C-fiber firing in the way that natural kind terms are introduced in the standard analogues. We are investigating the brain and observe a part of it that looks anatomically distinctive: we call this area of brain tissue “niap,” just to have a name for it. Then we set about investigating it with

36

Part I

a microscope and find that the area consists of delicate fibers of a distinctive histological type, which we end up describing as “C-fibers” (they are in fact the nerves that correlate with sensations of pain). Then we can truly say, “Niap activity is identical to C-fiber firing.” Clearly “niap” is not a synonym of “pain”—we might not even know that the tissue in question is involved in pain. This case would be just like the water and heat cases: we would not be able to have a world in which C-fibers fire but there is no niap activity, or a world in which there is niap activity but no C-fibers firing. For niap activity just is C-fiber firing. We have no intuition of contingency here at all, and no sense whatever that we are dealing with two distinct essences. But the standard type identity theory of pain treats “pain” as if it were just like “niap”—a term introduced to denote whatever the underlying brain state might be. But it was not introduced in that way, and might have been introduced without any knowledge of the brain at all, just on the basis of our inner experience. If we imagine someone introducing terms for underlying brain states explicitly—“Let ‘riban’ stand for whatever underlying brain state I have when I see something red”—then we will find exactly the same story we find with standard natural kind terms. We will have no trouble supposing that “riban” stands for a specific brain state, and no trouble forming true and unproblematic identity statements of the form “Riban is identical to R-fibers firing.” But this is not how we introduce our ordinary mental words, or how we understand them, and so these words cannot be assimilated to terms that denote brain states in the usual natural kind style. A type identity theory for terms introduced expressly as names for underlying brain states will not occasion any of the dualist intuitions that surround the usual kind of materialist identity theory. Those intuitions must therefore be taken seriously, and they appear to undermine the type identity theory. It may be thought that dualistic troubles with type identity are not so significant for materialism, because we can always fall back on token identity. Token identity does not entail type identity (though the converse holds), so it need not be disturbed by the duality of essences we noted above. Token identity claims merely that for any token mental event m there is an event e such that m is identical to e and e also has a physical property. For instance, a token of pain, occurring at a particular time, is identical to an event e that also has the property of being a C-fiber firing. In other words, there are events that have both mental and physical properties, where these properties are not themselves identical. A particular instance of pain is both an instance of pain and an instance of C-fiber firing. Every token mental event is thus identical to a token physical event, that is, falls under a physical description. Token identity theories are double-aspect theories in that they hold that there are events that have both a mental and a physical aspect—they have a dual nature. There are no mental events that fail to exhibit such a dual nature. Thus every mental event is a physical event of some kind, even though mental aspects are not identical to physical aspects.

Mind37

That sounds coherent enough, but is it really? It sounds coherent because we generally recognize that events can fall under different sorts of description: the same event can be described in several ways, that is, have several properties. For instance, an assassination can also be described as the beginning of a war—so a single event is both an assassination and a trigger to war. We can write a true identity statement as follows: “the assassination = the beginning of the war.” There is one event and two descriptions (and two nonidentical properties). We have an assassination event and then various properties consequential on that event. As we might say, the event (an action) was in its nature an assassination, and various other properties accrued to it as a result—such as leading to war. It also occurred at a particular time and place, was observed by certain people, and had disastrous results. Thus the event has a certain constitutive property (assassination) and also certain consequential properties (initiating a war). But that is not the way we should characterize token identity theory, because in its case the event has two natures, not one: it is both a pain and a C-fiber firing. Each property by itself would be sufficient to make it into an event with a specific nature, but it also possesses a second nature existing alongside the first. Each aspect contributes its own specific nature to the event. And the worry is that this is one nature too many, since particular events don’t have multiple natures. They have only one nature, coexisting with many extrinsic accidental properties. Let’s call this problem for token identity “the double nature problem.” Why is it a problem? Because it looks as if the token identity theorist has just jammed two different events together by stipulation, not by dint of natural identity. Suppose I declare that there is an event e such that e is both an assassination and a book signing: you would be justifiably puzzled, wondering how a single event could combine those two very different aspects. You might wonder whether I had confused co-occurrence with identity—was there a book signing going on near the assassination? According to one view of the individuation of events, an event is just the occurrence of a property by an object at a time, so that there are as many events as there are properties manifested at a given place and time. If mental and physical types are not identical, then they cannot occur in the same event, since events are individuated by the property in question; no one event can have both properties. That seems too finegrained when we consider cases like the assassination and its aftermath, but it seems exactly right when we are considering the constitutive nature of an event. Events can have only one constitutive nature. What if I claimed that a token pain is not only also a case of C-fiber firing but also a case of some further kind, say an economic kind? The event in question is both a pain and a C-fiber firing and an instance of inflation. You would be inclined to reply, “Well, which is it?” Once we multiply natures we multiply events—that is just what events are, ontologically speaking. That is how they are individuated. We can add various descriptions, such as time, place, and consequences, but we cannot add completely new second natures. But that is exactly what the token

38

Part I

identity theorist is trying to do in claiming that one event can be both mental and physical: he is therefore under suspicion of concocting one event out of two events. No one would think that a mental event could be both a sensation and a thought, or that a physical event could be both a C-fiber firing and a D-fiber firing: so why suppose that a single event can be both a pain and a C-fiber firing, where these properties are distinct and independent? If the properties are different, then the token events in which they occur must be different, on pain of violating the “single-nature condition” on event individuation. The challenge to the token identity theorist is thus to find a single instance of event identity, outside of the psychophysical case, that also violates the single-nature condition. Generally, when we have several descriptions of the same event, these descriptions record extrinsic features of the event, not multiple distinct natures. Are there any counterexamples to this? It might be thought that there are, because of the fact that the constitutive properties of an event can themselves have a nature: for example, an explosion event might also have the property of being a rapid expansion of molecules event. But that is not a case of dual natures; it is a case of a single nature that can be severally described. What we have here, in effect, is a type identity relation— explosions are rapid expansions of molecules. There is just one nature that can be described in different ways. But with token identity theory (of the non-type-identity kind) the idea is that the two properties are quite independent of each other in their intrinsic natures—they just happen to be stuck together in the same token event. And that is what is problematic: for how can a single event be the locus of totally distinct intrinsic natures? If two, then why not three; and if three, why not a hundred? The token dualist will say that there are simply two contemporaneous events here, tied to two distinct properties, namely being a pain and being a C-fiber firing; and that the token identity theorist has bunched them into one by arbitrary stipulation. Intuitively, there is, on the one hand, an event in my mind when I experience a pain at a particular time and, on the other hand, there is a distinct event in my brain that is correlated with the first event—these being manifestations of totally different properties. How could these be one event given that the properties concerned are agreed not to be identical? Referring to a token pain in terms of its consequences (“the event that caused me to take a pain killer”) is nothing like trying to refer to it by means of its neural correlate (“the mental event that is identical to a C-fiber firing at t”): the former description adds no further substantial nature, but the latter certainly does—and that seems like one nature too many. One event can either be a pain or a C-fiber firing, but it can’t be both—not unless pain (the type) is C-fiber firing (the type). It thus appears that token identity presupposes type identity, because only that will reduce the event’s nature to a single property. The problem arises when the two properties are distinct and independent, for then we are crediting a single event with two independent natures—as if one event could be both an assassination and a book

Mind39

signing. Token identity with type identity is perfectly coherent, but token identity without type identity looks like an ontological monster—a creature of metaphysical invention not natural unity. This is presumably why the notion of token mind–brain identity seems so forced and artificial (without the backing of type identity): we are just declaring an identity, not discovering one. That seems remotely feasible only because we are using models like assassination and the start of the war, but these are not enough to provide a decent precedent for what the token identity theorist is actually proposing. Just as the standard analogies of empirical theoretical identification failed at the crucial point for type identity theories, so the standard analogies of identity statements for particular events fail at the crucial point for token identity theories. The natural dualistic intuitions in both cases cannot be accommodated by the standard analogies, and so persist in the face of materialist identity theories. Thus “pain = C-fiber firing” is not like “water = H2O” and “this pain = that C-fiber firing” is not like “the assassination = the start of the war.” The cases of water and the assassination provide perfectly unproblematic identity statements, but the cases of type and token pain do not. In the case of pain we are left wondering how such statements can be true: type identity theory is vulnerable to stubborn intuitions of contingency and distinctness, while token identity theory tries to bundle what are by any reasonable standards two events into one event. There thus appear to be no true mind–brain identity statements. One might wonder whether supervenience, or something like it, could help out the token identity theorist. Earlier I posed a dilemma: either we presuppose type identity, in which case we run into the problems attending that view, or we reject type identity, in which case we multiply natures beyond the possibility of unity. But what if we suppose that the mental and physical aspects of a given event are not independent but dependent (though nonidentical)? What if we take the mental aspect to supervene on the physical aspect? Would that solve the double-nature problem? Suppose we have an economic token event, say an event of inflation, and suppose that the economic properties of this event supervene on its psychological and other properties—it is in virtue of those properties that the inflationary event occurred. Can we say that this event has two sorts of nature, economic and noneconomic, without compromising its status as a single event? Likewise, can we say that an event has both a mental and a physical nature, with the former supervening on the latter, without compromising its status as a single event? The trouble with this suggestion is that the claim is subject to the same dilemma as before: if we have supervenience underwritten by reduction (i.e., a type identity theory), then we don’t have a case of a single event with a dual nature; but if we reject such a reduction, then we still have an unacceptable duality at the heart of the alleged unitary event. The economic event will not be genuinely identical to the underlying event or events, even if it does supervene; and the mental event will not be identical to the underlying physical event, even if it supervenes. For the supervening properties are still distinct properties that confer a nature on the event to which

40

Part I

they contribute; and so there is no room for a further nature in that event. Intuitively, the mental event of pain cannot also have a second nature, despite supervenience: being the event it is does not allow it to host an additional physical nature. It would be different if the mental property were identical to the physical property on which it supervenes, since then there would be a single nature satisfying two descriptions. But if the mental property is really distinct from any physical property, then introducing a physical property into the event violates the single-nature condition. It is therefore better to say that we have two events here, one mental and one physical, even though the properties of the former supervene on the properties of the latter. When first confronted by token identity theory one’s first instinct is to ask, “But is it mental or physical?” We are then pityingly informed that the theory is that a single event e has both a mental and a physical aspect—is both a pain and a C-fiber firing— and that we should cease our doubting. The wonderful event e enjoys a double nature (a double life), each side of e being irreducible to the other. It magically straddles the mental–physical divide. We are impressed and captivated by this protean being, but it is hard to suppress our unease, because we want to be told which of its sides constitutes its basic nature—with the other side merely derivative. It is then sternly insisted that both sides are equally basic, equally real, and equally irreducible. But we are left wondering how we can be dealing with a single unified event. Analogies with other kinds of token identity statement don’t assuage our concerns, because they don’t involve the idea of dual intrinsic natures, just the idea of different descriptions attaching to a single intrinsic nature—what we think of as the constitutive nature of the event. Thus the token identity theorist does not succeed in persuading us that what he proposes is really metaphysically possible. It still seems as if he is conjuring identity from nowhere, in the face of clear indications of nonidentity. The foregoing objections to standard forms of mind–brain identity theory cannot be said to establish or prove dualism, though they certainly move us in that general direction, because there may be powerful arguments in favor of mind–brain identity that we have not disarmed. But they do show that things are not as easy for the identity theorist, of either stripe, as has often been supposed. I think it is fair to say that the fundamental difficulties have been papered over by the use of dubious analogies that do not defuse those difficulties.

Are There Actions?

Apparently there are actions, since I can sign my name, wave to a friend, or punch someone. Such actions can be counted: I signed my name three times yesterday and delivered five punches. There are signings, waves, and punches; and these are actions. It is thus tempting to say that actions exist and that we are related to them in a certain way. Just as there are thoughts and sensations that we can be said to have, so there are actions that we can be said to … But what do we put in the blank: say, “do,” “make,” “produce,” “cause,” “execute,” “perform”? The favorite philosopher’s term here is “perform”—thus we “perform actions.” There are things called “actions” and we stand to them in the relation of “performing.” We don’t “have actions,” but we stand in another relation to our actions, with various names. The resulting locutions are awkward and strained: we don’t normally say that we “do actions” or “make actions” or “cause actions”—still less that we “do punches” or “make signings.” Nor do we say that we “perform” these things. Part of the problem is that these verbs are themselves verbs of action, so that we are being said to “act an action.” I punched someone and I did a punching—two actions, it seems. But now we have a regress, because my doing is itself an action that I did. Performing is an action, so if I perform an action, I must have performed my performing—and so on ad infinitum. To say that I perform an action is to say that there is an action and a performing of it, where these are distinct things. We do naturally speak of “doing something,” but the something here is not typically an action—it is more like the effect of an action. The something I do is not an action of mine, though the doing is. Thus I may point to a hole in the ground and say, “I did that.” To speak of me doing the action of raising my arm sounds peculiar—I raised my arm, but I didn’t do the raising. We are apt to infer from the truth of “I punched him three times” to the truth of “I performed three actions of punching him.” But this can be highly misleading: first, because it suggests the double-action theory of action; and second, because “action” is not a proper count noun. Were not my punches made up of subactions, such as flexing my muscles? If you ask how many actions were performed in a given time period, you will not get a clear answer—just as asking how many objects there are in a room has no clear answer. We need a proper count noun like “punch” or “table.” The noun “action” is like “object”—a kind of dummy count noun, with no clear individuation conditions. Thus we may say that there are no objects as such, as there are no actions as such—there are just specific types of object or action. But still, what is my relation to my signings and punches? It is very odd to say that I “perform” signings and punches—unless, perhaps, I am an actor on the stage pretending to do these things (I gave a good performance of punching someone). I simply punch someone; I don’t perform the punching. The relation between the action and me seems simply to be that I am the one acting—it was my action (not yours). It isn’t

42

Part I

that I stand in the performing relation to the action; rather, it was my action. All that can and should be meant by my “performing an action” is that I acted: but then we have said nothing about any supposed relation between agent and action. I certainly don’t “produce” my actions in the way that I may produce loaves of bread—the loaves are not actions. The phrase “perform an action” is logically misleading, by suggesting as it does a relation between agent and action that just isn’t there—like the relation between an actor and a character. I cannot be said to “create an action” as I might create a painting: the action of painting is not caused by another action of creating the first action (did I create the creating too?). It is different with what are called “involuntary actions,” such as tics or tremors. Here I can be said to “have” a tic or tremor—though I do not “perform” such things, since that implies intention. But we can’t say that I “have a book signing” when I simply sign my name in the normal way. I may have the sneezes, but I don’t “have the punches”—unless this is an involuntary tic. And the difference is not that I “have” tics and “perform” actions, since the latter locution does not stand for a relation between my action and me; it is just a misleading way of saying that I act. It is perfectly true that I punch people, but it is not true that I perform acts of punching, in the sense in which philosophers are apt to use this phrase (in contrast to being an actor on the stage playing a pugilist). That locution is reifying and conceptually suspect, except as a longwinded way of saying that I punch people. These points apply to the notion of “speech acts.” I assert and command things, but does it follow that I “perform speech acts”? Again, what is this performing relation, and isn’t it an action itself? Can I be said to “perform assertions” (as a surgeon can “perform an appendectomy”)? Is my asserting something any different from my performing the act of asserting something? Also, is the noun “speech act” a proper count noun? In asserting that the sky is blue I perform other linguistic acts, such as referring to the sky—so how many speech acts did I perform? I made one assertion, but indefinitely many other linguistic acts went into that (phonemes, etc.). To ask “how many speech acts” occurred during a given period is to ask a question with no clear answer; it depends on what specific count noun we choose. So what should we say in answer to the question of whether there are actions? I think we should say: “Yes and no.” There are punches and signings and waves, but there is no class of entities called “actions” that is such that we stand to them in a relation of performing or doing or making or producing or creating. Actions are not things we manufacture or generate or bring into being—for these are all themselves types of action. The logical form of an action statement is not “X does a at t,” where a is an action; we don’t do actions (compare “Debbie does Dallas” or “Have you done Venice?”). Actions are a bit like postures—forms our body gets into, not entities to which we stand in a mysterious “adoption” relation. It seems clear to me that philosophers of action have tacitly accepted the model of “having a sensation” and applied it to the

Mind43

case of action, thus speaking of “performing an action.” But there are no actions in that sense—a punch, say, is not a special kind of entity that I “perform” or “do” or “bring about.” I don’t do my deeds; my deeds are my doings. We should not analyze punching in the way suggested by these peculiar locutions. There is indeed a relation between the puncher and the punched, but there is no similar relation between the puncher and the punch (except the trivial one of being his punch). The puncher does not “perform” the punch or otherwise “carry it out.” So the philosophy of action is not about “the performing of actions”—and it is not about entities called “actions” in the sense that goes with that phrase. There are no actions in that sense. Compare our talk of events: we say that the bomb exploded, and also that there was an explosion, but we don’t say that the bomb “performed an explosion” (or “had an explosion”). We don’t think that in addition to the explosion there is an event of performing the explosion, whatever exactly that might be. Actions, being events, are like that: people act, and there are actions, but there is no performing of actions. There is no triadic structure of agent-performance-action. We can say that explosions occur because of bombs, and we can say that punches occur because of agents: but in neither case should we say that bombs or agents instigate explosions or punches—as if there has to be another event or action preceding the explosion or punch. To put it differently, while I can perform a role, produce a loaf of bread, make a mess, create a work of art, cause a commotion, or instigate a rebellion, I cannot do any of these things to my actions—I cannot stand in these kinds of relations to my actions. Yet we have a tendency to talk as if we do stand in such relations to our actions, as if we “act our actions,” thus distorting our concept of action. We end up with an “ontology of actions” that is metaphysically misguided. Actions, in this sense, don’t exist.

Actions and Reasons

Received wisdom, going back to Aristotle, has it that there are two kinds of reasons— reasons for belief and reasons for action. Thus we are taken to have “theoretical reason” and “practical reason” rooted in two different rational faculties. The conclusion of theoretical reasoning is said to be a belief, while the conclusion of practical reasoning is said to be an action. There are also “practical syllogisms” as well as “theoretical syllogisms.” Our rational faculty splits into two parts, according as it concerns belief formation or action production. There is a fundamental dualism in reasoning and reasons—we might call it “ratiocinative dualism” or “Aristotelian dualism.” Reasoning about what to do and reasoning about what to believe are held to be distinct kinds of activity, governed by distinct rules, with different kinds of outcome (belief or action). On the face of it, this is an unattractive doctrine. It is implausible that “reason” is ambiguous, or that radically different faculties are involved in discovering facts and deciding on the best course of action, or that reasoning can have quite different sorts of conclusion. Nevertheless, the position is deeply entrenched and seldom questioned. It is felt that beliefs and actions are clearly ontologically quite distinct kinds of thing, and that reasoning about them must reflect that diversity: from a dualism about beliefs and actions we infer a dualism about reasons and reasoning. I think we should question this ancient picture. Can we unify both sorts reasoning? Can we maintain that reasons are all of the same logical type? Is Aristotelian dualism as misguided as Cartesian dualism? This certainly seems like a worthwhile project—redescribing matters so that a fundamental unity emerges. So: is ratiocinative monism defensible? One possible line of thought is as follows: Belief formation is a type of action, and theoretical reasoning is reasoning about what beliefs to form; so it is reasoning about a subclass of actions. Reasons for belief are reasons for forming beliefs, but forming a belief is an action like tying one’s shoelaces; so all reasons are practical reasons, with actions as their outcome. The conclusion of a piece of theoretical reasoning is the action of belief formation. Voilà!—we have unified practical and theoretical reason. The trouble with this short route is that it is false that belief formation is a type of action (though some brave souls have tried to maintain that it is): we don’t and can’t decide to believe. But the suggestion is helpful in providing an example of what a unifying theory would look like: we simply subsume one side of the divide under the other. So let us try moving in the opposite direction: can we use the concept of belief to redescribe the case of action? Here is a simple and obvious proposal: the conclusion of so-called practical reasoning is a belief about which action is the most desirable in the circumstances. Suppose I am wondering what to do with my afternoon—go over to Eddy’s house to play tennis or go to the Venetian pool for a swim. I weigh my preferences, consider the pros and cons, and arrive at the conclusion that going to Eddy’s

46

Part I

for tennis is the most desirable course of action. That is, I judge that going to Eddy’s is the most desirable thing to do—this is the conclusion of my reasoning. The conclusion is not my going to Eddy’s—the action itself—but the belief (judgment, knowledge) that going there is the most desirable (worthwhile, good, agreeable, etc.). My reasons are reasons for that belief—as I have reasons for belief in various facts about the world. Thus all reasons are reasons for belief; it is just that some beliefs concern agents and their actions while other beliefs concern nonagential matters of fact. I believe we should act to reduce global warming, and I believe in Darwin’s theory of evolution: these are beliefs about different subject matters, but they are both beliefs based on reasons. Why should we suppose that these reasons are of logically different types? They are both reasons for belief; there are not action-reasons and belief-reasons. We can reason about the world and we can reason about our actions, but in both cases we seek to arrive at well-founded beliefs—that is, we aim at knowledge. That seems like a coherent and attractive story, but can it be motivated more strongly? I contend that reasons must always be reasons for something with propositional content: there cannot be a reason R for an entity E where E is not propositional. Thus I cannot have a reason for sweating or having freckles, since these are not propositional entities. Of course, there are reasons why these things are the case—that is, they have causes—but I do not have reasons for them. By contrast, I can have reasons for believing things, because beliefs are propositional. But actions are not propositional, being merely movements of the body: they cannot be justified by reasons (though they may be caused by them) because they are not the kind of thing that can be justified. Movements of the body can no more be justified by reasons than movements of worms or trees. The event of my arm going up two feet is not capable of justification—as opposed to my belief that it is desirable to raise my arm two feet. It is a category mistake to suppose that bodily movements per se can be justified, that is, can have reasons in their favor. Muscular contractions and efferent nerve signals are not rational conclusions of practical arguments; they are what constitute bodily movements. How could a movement of the body be the conclusion of anything? Conclusions must be propositional, which beliefs are and actions are not. We can see why actions are ineligible to be conclusions of practical reasoning by noting that they don’t provide the kind of intensional context that would allow reasons to gain purchase on them. Suppose I decide to drive over to Eddy’s for a game of tennis but that unknown to me the drive will result in a car accident (given the initial conditions and laws of nature): that is, the action of driving over will involve an accident I can’t foresee. Let us suppose that the drive over is identical to being a drive that results in an accident. I have a reason to drive over (to play tennis) but I don’t have a reason to be involved in a car accident: the reason works only “under a description”—it doesn’t cover the action as an event that has many descriptions. So it can’t be that the action itself has a reason for occurring, since described differently I have

Mind47

no reason for performing it. The context “I have a reason R for performing action A” is intensional: we cannot substitute any true description of A and preserve truth. Thus actions themselves, construed as concrete individuals, don’t have reasons—they must be construed under an appropriate description. That means there must be something propositional in the picture—something like belief. We may suppose that intentions to act have reasons, so that we are adverting to the intention behind the action, which is propositional, but intentions will involve judgments of desirability—to intend to do something requires believing it to be desirable to do it. In any case, it is not the bodily movement as such that is justified or reasonable, because it doesn’t have the requisite intensionality; it is some sort of propositional attitude. Judgments of desirability lurk in the background, and they are precisely the kind of thing that can be justified—unlike shifts of the body, contractions of the muscles, or nerve signals. None of the latter can be genuine conclusions of reasoning. Here is another way to look at it. Suppose you are a brain in a vat and don’t know it. You perform stretches of practical reasoning and arrive at judgments of desirability— but you never perform any actual actions. We therefore cannot say that any of your actions are justified or that you have reasons for your actions—since you don’t act. Yet your practical reasoning may be unimpaired and impeccable: you reach rational conclusions about what to do, just like anybody else. You are as justified in your practical reasoning as a person with a body. You have reasons for your conclusions, but by hypothesis these conclusions are never actions. Shouldn’t a theory of practical reason treat this case just like that of a normal agent? Deleting the actions does not change the nature of the reasoning. In both cases the reasoning concludes with a judgment of overall desirability (possibly also with an intention). A theory of the nature of practical reason should be indifferent to whether the person reasoning is an embodied agent or not; the existence of actual actions is adventitious to the process. It is not actions as such that are justified by reasons; it is judgments about actions. What are called “reasons for action” are really reasons for believing that an action is desirable. It should not be forgotten that practical reasoning does not always occur in the first person: I can reason about what you should do, for example, what college you should attend, as well as what we should do. In the former case my conclusion will take the form “You should do A.” It certainly does not take the form of my performing an action, say going to Harvard. My conclusion is a belief about what action is desirable for you—it is up to you whether to perform the action in question. Do we really want to say that my conclusion is your action? No, I merely arrive at a judgment about your action—that is my conclusion, not your action. I have concluded my reasoning when I reach that judgment; I don’t have to wait till you perform the recommended action. Failing to perform the action is not failing to conclude the train of reasoning (as also with the brain in a vat case). The only sense in which there are reasons for action is that there are reasons for believing that certain actions are desirable: but there are no

48

Part I

reasons for the actions themselves—they are not proper objects of justification. You cannot say of a bodily movement that it is rational or irrational; it is always beliefs that are rational or irrational (or intentions or desires—propositional entities). I conclude that not only can we unify so-called practical and theoretical reason by invoking belief in both cases, we should take this line because the usual idea of reasons for action is conceptually confused. There are reasons to desire things, intend things, and believe things, because these all have propositional content; but there are no reasons for the things that are done—that is, motions of the body are not entities that admit of justification. We may say that I am justified in doing things, because I have justified practical beliefs, but we may not say that the doings themselves are justified— because doings (in the intended sense) are not propositional entities. We can never say “X is justified” where X has no propositional content. Just as things that justify have to be propositional, so things that are justified have to be propositional. Justification is always a relation between propositional entities. Rocks cannot justify other rocks or be justified by them. Nor can movements of the body justify or be justified. Beliefs (or other representational entities) comprise the field of justification—in practical contexts as well as theoretical. Thus all reasons are reasons for belief (or intention or desire). What do reasons look like in the case of practical beliefs? They involve the identification and weighing of desires: the agent considers what desires he or she has, their relative importance, and so on. For example, when deciding whether to go over to Eddy’s I recognize that I have a desire to play tennis, and I weigh this desire against my desires to do other things; and so on. So the reasoning takes the form of determining what course of action would be desirable in the light of my desires. If I am asked to defend my decision, I will cite these desires: they are my “evidence” for the practical belief I form—my justification for deciding as I do. In the same way, I could reason on someone else’s behalf by assessing his or her desires and prescribing a course of action: my justification for believing that another person should act in a certain way is that he or she has certain desires. Information about desires is what warrants a particular belief about what it is desirable to do. Thus the reasoning here is essentially the same as elsewhere: we gather evidence and come to a conclusion. Practical reasoning is evidential reasoning about desires. It is not different in kind from “theoretical” reasoning: both involve the marshalling of facts—either about desires or other sorts of fact. There is no deep logical division here. Therefore, ratiocinative dualism is mistaken: reasoning about actions and reasoning about the world are fundamentally identical. All reasons are reasons for belief, but beliefs can be about different things, some of them being about actions. Reasoning about different types of thing is not engaging in different types of reasoning.

Consciousness and Light

The theory to be developed here is that the nature of consciousness results from the nature of light, and the reason consciousness exists is that light exists. This is putting it as provocatively as possible, but not, I believe, inaccurately. We will need to build up slowly to this theory—the “light theory of consciousness.” Consider a visual experience—the kind you have when you look around a room, say. What are its distinguishing features? It has intentionality, qualitative content, and a conscious subject, among other features (color, depth, foreground and background, center and periphery). But there is another feature not so often remarked on, perhaps because it is so obvious: the experience contains an enormous amount of information about the environment, an almost incalculable amount. Just think of all the shading, colors, overlaps, angles, spatial relations, wholes and parts—the exquisite detail and complexity. The visual experience presents a chunk of the world in all its variety and multiplicity, and this provides information about how things are out there. I shall call this “massive informational simultaneous saturation”—MISS, for short. The experience is saturated by simultaneously present bits of information, available for use by the perceiver. So much is obvious, but the following claim is by no means obvious: MISS is necessary and sufficient for consciousness. That is, conscious sensory states are distinguished from unconscious sensory states by being capable of MISS. To put it colloquially, a conscious visual experience is one that contains a hell of a lot of information. What evidence is there for that claim? Let’s look at unconscious sensory states and see if they are likewise rich in information. First, when you enter a state of unconsciousness (sleep, coma) you stop being visually aware of the environment, except perhaps to a very limited degree (your brain will respond to a bright light but not much else, even with your eyes open). Thus, when you’re in a state of unconsciousness, your sensory intake is minimal. Second, subliminal perception does not contain the vast amounts of information possessed by a conscious visual experience; it contains a small subset of such information. Subliminal vision is quite crude compared to conscious vision. Third, blindsight does not have the informational richness of conscious vision, being quite limited in what it can respond to. It is natural to assume that as we add to the amount of sensory information processed by the brain we reach a threshold after which the response is a conscious visual experience. Thus we might suppose that the function of visual consciousness is the representation of massive amounts of visual information; conscious visual experience evolved so that MISS would be possible. Why exactly unconscious visual processing should have less “bandwidth” than conscious visual processing is hard to say, but that it does seems undeniable. Conscious vision can “take in” much more than unconscious vision, and consciousness itself appears to aid this enlarged capacity. Consciousness is a device for increasing the informational

50

Part I

bandwidth—that is its function and that is why it evolved. Creatures with visual consciousness can process more visual information than creatures without it (at least as conditions stand on planet Earth), so such consciousness is selected for. Those animals that process light without the benefit of visual consciousness will not exhibit MISS; the zombies of our planet are low in informational capacity. How does light come into the story? It does so because light itself contains an enormous amount of information about the world—far more than any other physical phenomenon. It is the pattern of light hitting the retina that carries all that information, waiting to be extracted by the visual system. The ambient light encodes colossal amounts of fine detail about the objects from which it is reflected—far more than sounds emanating from an object, or chemicals released by it, or the touch of it. Moreover, light is everywhere, so that anywhere an eye is it can pick up the information contained in the local light. You can’t hear or smell or taste or touch very distant objects, but you can see them—even if they are light years away. Light is fast, ubiquitous, and incredibly fine-grained. Given this richness of information content, it makes sense that animals would evolve to exploit it—to make use of all the information contained in light. Thus there was selection pressure to evolve an organ that could tap into the full potential of light as a source of information—to extract the information that is naturally contained in it. That is, an organ will likely evolve that can handle the massive simultaneous banks of information contained in light—to wit, an eye. But if it is sufficient for visual consciousness that MISS be present, at least as things stand on Earth, then light will lead to consciousness, not merely to unconscious visual processing. That is, light is the reason that a MISS organ would evolve, but a MISS organ is a conscious organ—so light is what led to consciousness. Light was the cause of the origin of consciousness in the animal kingdom, to put it simply. According to this theory, there would be no visual consciousness in a world that was not as irradiated by light as our world is. If the Earth were only dimly illuminated by a distant star, so that the tiny amounts of light falling on it contained very little information about objects, then there would be no point in evolving the kind of visual awareness that is massively informational; at best eyes would evolve that processed light unconsciously, that is, with low bandwidth. No doubt conscious perceptual processing consumes a lot of energy, more so than unconscious perceptual processing; so it would be unwise to evolve a conscious visual organ when an unconscious one would do as well and be less metabolically costly. It is only when light is plentiful and rich in information that it makes sense to respond to it with conscious visual experience, given the metabolic demands of consciousness. But when light is plentiful and rich in information, it is adaptive to develop an organ that can exploit its full potential—that is, a visual system that is capable of MISS. Again, it is not clear why visual consciousness and MISS go together, but it does seem to be a biological law, at least on Earth, that they do. Given that fact, the explanation for the origin of consciousness is that light exists

Mind51

and has the nature it has. And given that conscious vision is essentially MISS, its nature (or an aspect of it) results from the nature of light: the former follows from the latter. As a matter of natural law, visual consciousness, as it exists on Earth, is shaped by the nature of light—not a very startling claim. What is perhaps surprising is the contention that the evolutionary cause of visual consciousness is light: the physical nature of light is the reason that creatures see consciously. There is a natural objection to this theory: how does it explain other kinds of consciousness, such as auditory or tactile? The answer is, with difficulty: the problem is that the other senses don’t exhibit MISS, so they may as well proceed unconsciously— yet they don’t. The reason is that the stimuli for other senses don’t have the physical powers of light: there is just not that much information in an impinging sound wave, or a chemical entering the nose or mouth, or a touching object. According to the present theory, consciousness arises when the stimulus reaches a certain threshold level of complexity, but that appears not to hold for hearing, touch, taste, and smell. Why then don’t these sensory processes remain subliminal? Let me mention four possible responses to this difficulty. First, it might be replied that we are underestimating the complexity of the stimuli in the nonvisual cases, and hence the amount of MISS present in the corresponding sensations. My current auditory field, say, contains a lot of information about sound—key tapping, birds singing, traffic, and the like—and in some cases my ears are replete with myriad sounds (say, listening to a symphony orchestra). Tastes, smells, and touches can be complex too—consider the totality of tactile sensations you now enjoy. This is a possible answer to our difficulty, though it is hard to see how the other types of stimuli could rival light for information content, given the physical nature of light. A second reply is that consciousness originally arose in connection with vision, and vision alone, but that once evolved it was transferred to the other senses. The picture is that in primitive organisms all sensing was subliminal but that at some point in evolutionary history sight began to exploit the full resources of light, which pushed it into conscious territory—and then this visual mode of consciousness spilled over to the other senses. Light and vision made consciousness accessible, but once it was possessed it could be recruited by the other senses. Light caused visual consciousness initially, and then the nonvisual senses tapped into the cerebral machinery for that kind of consciousness, so creating their own kind. This is a nice-sounding story, but it seems rather ad hoc and deus ex machina. Third, it might be claimed that the nonvisual senses are not as conscious as vision: their operations are more subliminal than the operations of vision. Isn’t it true that we are mostly unaware of what we are hearing or touching at any given moment? We have to pay attention to these senses in order to become conscious of what we are hearing and touching (and the same might be said of taste and smell). By contrast, vision occupies the forefront of consciousness, with light pressing itself upon us: what we see

52

Part I

is what we are primarily conscious of at any given moment. Sensory consciousness is primarily visual consciousness. There is something to be said for this from a phenomenological point of view, but it is surely exaggerated: what about the blind—are they unconscious? And what about closing your eyes and listening intently to a piece of music? Consciousness is clearly not the privilege of the visual. Fourth, it might be maintained that we shouldn’t be talking about “consciousness” at all in the general way that people do; rather, we should be discussing particular senses and how they work. These general concepts of consciousness and experience are dubious philosopher’s inventions, it may be said, born of a craving for generality—by contrast with the specific and solid concepts of seeing or hearing. The light theory of “visual consciousness” is really just a theory of one sense: it explains how that sense arose in virtue of the properties of light, bringing in phenomenological facts about the visual. Seeing is MISS, and light explains why that is so: we can say that without presupposing some general attribute of consciousness that is shared by all the senses. Thus there is no claim to account for consciousness in general, or even just sensory consciousness; we eschew such notions and limit ourselves to explaining the characteristics of one sense. We know that ordinary seeing is MISS, and that subliminal seeing is not MISS, and we postulate that what caused seeing of the former kind is the nature of light; we venture nothing about the other senses or some supposed general notion of consciousness. This last response is quite radical given current opinion, though worth pondering: maybe it is wrong to try to solve “the problem of consciousness” as if it were monolithic; maybe we should adopt a more piecemeal and pluralistic approach. Putting this troublesome question aside, the light theory tells us that our mental life, as we experience it, owes its existence and nature (part of it anyway) to the fact that we are surrounded by information-rich light. We have a sense that is an adaptation to this remarkable physical phenomenon, originally evolved in our fish ancestors no doubt. It was light in all its physical glory that triggered the kind of phenomenology that we find in vision. Granted that the kind of complexity we find in visual experience is not possible without the thing we call consciousness, at least not on planet Earth with its laws and initial conditions; light is what explains visual consciousness. If the planet were mainly dark, or if light were less rich in information, visual consciousness would not have evolved; animals would have had simple eyes (photoreceptors) and subliminal vision, with a far simpler visual cortex. It was the Sun in its proximity and power that caused consciousness to evolve on this planet, by sending out so much light to Earth. When God said, “Let there be light!” he was paving the way for consciousness.

Have We Already Solved the Mind–Body Problem?

We don’t experience ourselves as a metaphysical problem. We don’t wake up each day and think, “Oh, what a hard problem I am!” It is only when we get to philosophy (or maybe science) that we start to puzzle over ourselves. That we are problematic comes as a surprise, an affront even. This isn’t so with all mysterious matters: it is often obvious from the start that something is difficult to know, possibly unknowable. Thus facts about distant galaxies or very remote times present themselves to common sense as inherently difficult to know. But we don’t present ourselves to ourselves as likewise impenetrable. That is, we don’t experience ourselves as suffering from a mind–body problem (unless we are weak-willed, paralyzed, or some such): we don’t find ourselves perplexed at ourselves from early on, as if we were walking paradoxes, as if we don’t make sense. We feel ourselves to be straightforward entities, unified and integrated— but neither all body nor all mind nor some peculiar amalgam of the two. We view ourselves as ontologically perfectly kosher, even a paradigm of clarity in a confusing world. We take ourselves for granted, as if at least we pose no problems. As Popeye sang: “I am what I am and that’s all that I am, I’m Popeye the sailor man!” Thus there is a mismatch between our intuitive view of ourselves and the theoretical view to which we aspire (and we can be sure that other animals are not tormented by philosophical self-questioning either—they don’t live their lives in the shadow of the mind–body problem). When we learn a bit of philosophy we think, “I had no idea that I was so much of a problem; now I don’t know what to think of myself.” And yet, outside of the study, we return to our ordinary intuitive understanding of our self. We take it easy about the hard problem. There is a well-known but tantalizing passage from Descartes’s Meditations that bears on this tension: Nature also teaches by means of the sensations of pain, hunger, thirst, etc., that I am not present to my body only in the way that a pilot is present to a ship, but that I am very closely joined to it and almost merged with it to such an extent that, together with it, I compose a single entity. Otherwise, when my body is injured I (who am nothing but a thinking thing) would not feel pain as a result; instead I would perceive such an injury as a pilot perceives by sight if some part of the ship is damaged. Likewise, when my body needs food or drink, I would understand this more clearly and would not have confused sensations of hunger and thirst. For these sensations of thirst, hunger, pain, etc., are undoubtedly confused ways of thinking that result from the union and, as it were, the thorough mixing together of mind and body.

Descartes is not saying here that we are under an illusion as to our unity, given the truth of his metaphysical dualism, but rather that we know ourselves to be a unity—for that is what “nature teaches.” We apprehend ourselves as seamlessly unified. But the teaching of nature is not theoretical teaching; it is more basic and instinctual than that. Let’s call

54

Part I

this knowledge “intuitive knowledge” (or “instinctual knowledge” if you prefer). Then we can say that, according to Descartes, nature gives us intuitive knowledge of our fundamental unity—even though we cannot articulate this knowledge in theoretical terms (hence the metaphors of “merging” and “mixing”). The picture is that we are, ontologically, a natural unity, and we experience ourselves as a unity, but we have no articulate way to express the kind of unity we are. Thus we have a theoretical mind– body problem, even though we don’t have an intuitive mind–body problem. We feel and sense ourselves to be unitary—a form of embodied consciousness, in effect—but we can’t give a theoretical account of our unity. Our commonsense intuitive knowledge does not convert to discursive theoretical knowledge. This distinction opens up an interesting possibility: that there is a sense in which we already know the answer to the mind–body problem. Therefore, we do not go around feeling puzzled about ourselves—for we have intuitive knowledge of our nature. We don’t feel ignorant of ourselves, so that philosophical questions can expose this ignorance; we feel knowledgeable about ourselves, just as Descartes supposes. It is just that we can’t convert one kind of knowledge into the other kind. In a sense, then, we have already solved the mind–body problem. It isn’t that we are objectively unified but have no sense of this unity, no knowledge of it, no insight into it; rather, we do have a sense of it—but an inarticulate sense. We have, to use Russell’s term, acquaintance with our nature as a unified thing, and hence knowledge by acquaintance; but we don’t have knowledge by description—discursive theoretical knowledge. Nature has given us intuitive knowledge of our unity, but it has not given us a philosophical or scientific theory of our unity. I am not, objectively, a metaphysical enigma, a hard problem, or an impossible contraption—and I am aware that this is so, because I am aware of myself. But I cannot express this knowledge in discursive theoretical terms. I have a kind of animal knowledge of it—the kind that nature teaches, but not schools. Nature does not teach me that I am an enigma or beyond comprehension, with philosophy then merely confirming that impression; rather, it teaches me that I am an intelligible unity—though not one that I can systematically articulate. I can sense what I cannot explain; I can feel what I cannot put into words. Clearly, this possibility depends on a distinction between two kinds of knowledge: intuitive versus discursive, natural versus schooled, implicit versus explicit. Such distinctions have frequently been made, though they are not easy to demonstrate. What interests me, particularly, is the bearing of all this on the question of cognitive closure: for there is now the possibility that we have cognitive closure with respect to one kind of knowledge but not with respect to another kind. In one sense we already know the answer to the mind–body problem, but in another sense we don’t, and maybe can’t. If we accept strong cognitive closure, then we might even be in the predicament that we have already solved a problem we will never solve! We already know what we cannot know. This statement is saved from contradiction by the fact that “know” is ambiguous

Mind55

between intuitive (implicit, instinctual, natural) knowledge and discursive (explicit, theoretical, articulate) knowledge. Descartes’s own position is that dualism is true, but that we are nevertheless one unified thing, and we know we are—it is just that we cannot form an articulate conception (“clear and distinct idea”) of what our unity consists in. We have an inchoate understanding of ourselves, which is quite robust and apparently universal, but it resists discursive formulation—we might even say that it is not propositional. Descartes is an ontological dualist, a mysterian about the theoretical nature of our unity, and an epistemic optimist about our basic self-knowledge. He thinks we know what we are; we just can’t put it into words. To put it differently, our unified nature communicates itself to us, by dint of nature’s teaching, but it doesn’t inform us of what this unity consists in. We know it to be a fact, but we don’t discern the inner structure of the fact. This is why we can become puzzled about what does not normally puzzle us. It is a bit like knowing how to walk but being unable to explain it. The case is quite unlike our knowledge of the stars: here we are puzzled from the start (“What are those little glinting things up there?”) and are quite ready to admit our ignorance. But in the case of the self we are not naturally puzzled, because we do know what a self is; it’s just that this knowledge falls short of the kind of articulate knowledge we seek as philosophers and scientists. There is no sense in which we already know the answers to the hard problems of astronomy or physics, but there is a sense in which we already know the answers to the hard problems of the self. We know (intuitively) that we are not a metaphysical problem, but we don’t know (discursively) why we are not a problem. Cognitive closure might then coexist with cognitive accessibility.1 1. This combination might also hold for other philosophical problems: we grasp the solution implicitly by nature’s teaching, but we cannot produce a discursive theory. Free will might be a good candidate, or the nature of meaning, or a priori knowledge. In these cases too we have that odd combination of epistemic complacency and epistemic defeatism. It could be argued that evolution (or God) has installed implicit knowledge of the nature of these things in us, so as not to mire us in perplexity, but has not seen fit to supply us with the means to acquire the corresponding explicit knowledge.

The Reality of the Inner

The twentieth century set its face against the inner. Psychology and philosophy rejected the idea that mental states are special inner occurrences, sealed off from outside observation, private, known only to their possessor. Thus we have behaviorism (reductive and eliminative, Ryle and Watson), functionalism, Wittgenstein’s “outer criteria,” Quine’s rejection of the “museum myth,” and materialism in its several varieties. These doctrines make the mind a public thing, not something hidden inside—the mind is not something accessible only to the person whose mind it is. Some theorists accept a diluted form of “first-person authority” or “privileged access,” allowing for some sort of epistemic asymmetry between subject and observer; others simply abandon such notions, holding that others can know my mind as well as I can. What is rejected is the idea that it is of the essence of mind to be inner and private. For if that were so, the study of mind would be radically different from other studies of nature: we could only study the mind, as it is in itself, from an introspective point of view; there could be no objective third-person study of mind. That would make psychology radically discontinuous with the rest of science, which deals with what is public and outer. We could only integrate the mind into our general conception of nature if we abandoned the notion of its essential innerness. The mind must be a public thing or be no thing at all. The idea of a thing whose existence and nature is purely inward would separate the mind from the rest of nature, rendering it sui generis and unapproachable (save from the inside). Despite the prevalence these days of talk of consciousness, this twentieth-century position has not changed. We speak freely of consciousness, characterizing it with such locutions as “what it is like,” “phenomenal content,” “qualia,” “subjectivity,” and so on, but we do not generally find it spoken of using the language of the “inner,” the “private,” the “hidden.” The notion of the “inner process” is still taboo. That is why we don’t see people defining consciousness as simply what is inner, as opposed to things that are outer. Bats have an inner life, but rocks don’t; our inner life is different from that of bats; our inner life gives rise to a “hard problem.” This is quite a natural way to talk—the ordinary man or woman will know what you mean—but it is generally shunned: we can deal with consciousness, but we are reluctant to think of it as a wholly inner phenomenon—something hidden away, inaccessible, and imperceptible. We prefer the idea that the mind is both subjective and yet public. And indeed if the mind were radically inner it would differ dramatically from the body and the physical world—it would not be susceptible to third-person study. If the mind were in principle inaccessible to others, then it would (a) differ from everything else in nature and (b) not be investigable by standard scientific methods. It would be a “something” that might as well be a “nothing.”

58

Part I

It seems to me that this way of conceiving the issue is fundamentally correct: the inner is inimical to scientific treatment—at any rate, of the usual kind. And if the mind is by its essential nature inner, then it is different in kind from everything else in nature. It is not easy to articulate what this innerness consists in, though it has a strong intuitive content: the best we can do is to summon the usual language of “privacy,” “privileged access,” “self-intimation,” accepting that these terms do not do justice to our sense of the inner. The feeling is that while I have direct unimpeachable knowledge of my own mental states you can at best guess at them, infer them from my behavior, merely speculate about them. My mental states exist in a kind of private space that I can survey immediately but which you are barred from entering. You would have to be me to know what I know. Moreover, it is of the essence of mental states—part of their ontology—that they should occur thus privately: their what-itis-to-be is to be inner, unobservable, as if behind an impenetrable wall. The being of the (conscious) mind is to be evident to me but elusive to others.1 If I imagine a pink elephant, say, this mental occurrence, while immediately presented to me, is concealed from everyone else—it could not be apprehended by another as it is apprehended by me. It belongs, we want to say, in another world—the private world that each person carries within her- or himself.2 This private world offers itself to my introspective gaze, in delicious proximity, deep and full; but it is wholly invisible to anyone but me, a blank in the fabric of nature. I know my private world intimately, but I don’t know your private world—except perhaps tenuously. Thus I can speculate that there might 1. When we say that the mind is intrinsically unobservable, we attribute an epistemological property to it, but this property is grounded in the metaphysics of the mind—in what it is for the mind to exist. The mind’s mode of being is to exist privately, as an inner thing; it does not just happen to be unobservable, as if this condition could be remedied. When I have a thought, say, this occurrence has no outward face, no public identity, no intersubjectivity; it exists entirely for me. Its being is to be-for-me. It is radically inward in its essence: it could not escape its privacy and remain what it is. This is what I mean by ontological innerness: the ontological ground of the asymmetric epistemology of the mind. (Here I am trying to formulate the innerness of the mind as dramatically as possible, not shying away from it; of course, there are ways in which the mind can be known to others, because it can be inferred. Inner ontology is quite compatible with the possibility of inferential knowledge.) 2.  Perhaps the best example of privacy is dreaming: there is very little, if any, behavioral hint of what is going on within the dreamer. You can gaze at the dreamer for hours, with maximum attention, and you will have no idea what she is dreaming; meanwhile her mind throbs with inner life, rich private experience. Dreams prove the intrinsic opacity of the mind: so fully present to the subject yet so invisible to the observer. In dreams experience goes on with full force, but there is no sign of this in the public world. The dream is the inner without regard for the outer; it is the inner declining to communicate itself. It is content in its own concealment.

Mind59

be private worlds in objects not generally thought to have minds at all—trees, rivers. Our concept of mind is the concept of an inner reality that may or may not disclose itself; indeed, for each of us it is the fundamental reality—the one that is closest to us and that matters the most. We believe unshakably in the reality of the inner. But if that belief is correct, the mind is not as other things, and is not a potential subject for objective science. The mind is not something that exists in the public objective space to which all observers have access. At most we can study it from a firstperson point of view, introspectively, possibly sharing our findings with others similarly situated. We thus face a stark choice: we either give up the idea of the inner or we accept that the mind cannot be studied as other things in nature are. All the theories popular in the twentieth century give up on the inner; they make the mind something public and outer. This is obvious for behaviorist theories (including functionalism), but it is also true of materialist theories: these theories too render the mind accessible from a third-person perspective—the observer of the brain. If pain is C-fiber firing, then looking at C-fiber firing is looking at pain—that is what you are seeing when you peer into a brain. Such theories do indeed prepare the mind for objective scientific study, by conceiving it in outer terms, and by downplaying or denying its essential innerness. But I think they do violence to the mind in so conceiving it: the mind really is ontologically and epistemologically an inner thing. This is evident to common sense, but it is perfectly true that acknowledging it results in the collapse of the scientific model (as commonly understood). If the mind is constitutionally an inner thing, then it cannot be studied as if it were an outer thing. Of course, we can allow that the inner mind has outer symptoms—effects, manifestations. Behavior and brain can signal the presence of an inner state. But the state itself is an essentially hidden reality, lurking behind those outer symptoms. It is not present in its symptoms. So we are not studying the thing itself when we study its symptoms. We may be studying the embodiment of the mind, but we are not studying the mind. The mind is inherently an inner thing, while its symptoms are inherently outer things. I therefore think that the twentieth century in its flight from the inner proceeded from the right premises, but it reached the wrong conclusion. The correct premise was: the innerness of the mind cannot be reconciled with a naturalistic, homogeneous, seamless world of publicly accessible entities. But the right conclusion is not that innerness must be denied; it is that we do not live in such a homogeneous seamless world. The introspectionists in psychology were basically correct: the mind must be studied by a special method appropriate to its special nature, namely introspection. That doesn’t mean we cannot include behavior and the brain in our general study of the mind, since the mind does have these associations; but it does mean that the essential innerness of the mind requires us to approach it from a first-person point of view, if we are to

60

Part I

catch it naked, so to speak. We must accept that the mind (especially consciousness) is inner and face up to the consequences of that admission. Thoughts, sensations, and emotions are in their nature inner processes—private entities to which the subject has privileged access. To possess such a mental state is to undergo a modification in one’s private space—that is, in the self, the subject of awareness. The metaphor of the theater is quite unapt to capture this fact, since theaters are precisely public objects in public space. The privacy of the mental is sui generis, not to be compared to private rooms or beetles in boxes or processes buried in the bowels—for these are all in precisely the opposite category. The sense in which the mind is inner is not the sense in which one object can be inside another object, or internal to it: these are spatial concepts. The innerness of the mind is a matter of the inherent nature of the mind, not its relation to something else. It would be quite wrong to say that the mind is “hidden within the body.” If it were, we could, in principle, dig it out and take a look at it. No, the mind is necessarily private—inner by nature. It is the part of nature that nature does not reveal to the rest of nature—except the part that is that mind itself. Creatures have evolved that instantiate states directly known only to themselves; others can only surmise, knowing that they may be quite wrong. The mind is such that certainty about it is possible only from its own perspective. How does this bear on the mind–body problem? As follows: no solution to that problem can dispense with the innerness of the mind. Whatever is the correct theory must do justice to that innerness. The standard theories all fail this simple requirement—notably behaviorism and materialism. Nor can computational theories do what is necessary, since they too render the mind less than inner. Computers don’t have inner states in the intended sense. So reductions that don’t preserve innerness will fail as theories of the mind. We must add innerness to the list of properties of consciousness that make it problematic. Perhaps, indeed, we should put it at the top of the list, because the innerness property is about the most robust property consciousness possesses. We can argue about what-it’s-like and qualia and the phenomenal, but it is surely a datum that consciousness is inner—though it is a datum that has been routinely denied. We have both an inner life and an outer life—a life of the mind and a life of the body. The question is how that is possible. How can an inner life arise from an outer life? How can the inner emerge from the non-inner? How does the public produce the private (neurons are public, thoughts are not)? If this problem looks daunting, that indicates that we are on the right track. The problem is diamond-hard. Not for nothing did the twentieth century set its face against the inner. I would like to see a concerted effort to articulate more clearly what the innerness of the mind consists in. There is an epistemological question and an ontological question: what kind of knowledge do we have of our inner states, and what is the ontology of these states? What is it to be inner, and what is involved in first-person knowledge of the ontologically inner? These questions have been avoided, I suspect, because they

Mind61

pose such a threat to our general scientific world-view.3 It really is not clear how the project of an integrated scientific worldview is possible once the essential innerness of the mental is accepted. We seem confronted by an inner–outer dualism. There is something irreducibly “queer” in things. The mind refuses to fall into line with the rest of nature, despite being part of nature. 3.  There is also the inhibiting effect of Wittgenstein: he made philosophers afraid to go near the concept of the inner. I think he denied the inner because he knew quite well what a problem it poses. In this essay I have consciously flaunted my flouting of Wittgenstein’s position, knowing quite well how provocative that may be found. For me the inner stands in need, not of an outward criterion, but of an inward contemplation.

The Thought of Language

We are familiar with the idea that thought is a species of speech, a type of linguistic performance: there is the idea that thought is “subvocal speech,” and there is the different idea that thought operates by means of a “language of thought.” But might language be a species of thought, a type of cognitive performance: might speaking be a type of thinking? Might a speech act itself be an act of thinking? Can we think “outside the head”? Can we think with the larynx? As we can project language inward to account for thought, can we project thought outward so as to inhabit speech? Are there “external” thoughts? Descartes would suppose not, since thought is an attribute of a mental substance while speech is an attribute of the body—the body can’t think, only the mind can. Thoughts may cause acts of speech but they cannot be acts of speech. But that is not how we ordinarily talk: for we speak of “thinking aloud.” This construction is rather like the construction “reading aloud”: we can think silently or think aloud, as we can read silently or read aloud. There is silent reading “in the head” as well as vocal reading, and we talk as if there can be silent thinking “in the head” as well as vocal thinking. Are these locutions reflective of the psychological facts? We don’t speak of seeing or looking aloud, or believing, knowing and remembering aloud; and describing these things as done “silently” sounds distinctly odd. Is thinking like these, or is it like reading? We should first note the various ways in which language can be employed by the mind. First, there is ordinary speaking—producing sounds with the mouth—whether for communication with another or for enabling thought in oneself. Second, there is subtly moving the larynx and tongue while making no sound, as one might do when proof reading. Third, there is not using the organs of speech at all but merely producing words “in the head”: words going through one’s consciousness but with no bodily manifestation. The third category includes voluntarily producing words in the mind as well as hearing (or seeming to hear) voices in the head. In this case, language engages with the mind entirely without the mediation of the body (not including the brain), and there is no reason to doubt that it is as real as audible speaking—the words can have a purely “mental” existence. This is where we are most inclined to speak of thinking: we are silently and motionlessly “thinking in words.” Indeed, we are inclined to regard this kind of case as the paradigm of thinking—what thinking really is in its essence. Thinking, we suppose, is essentially having a conversation with oneself—silently, privately, without moving a muscle (there is no muscle for thinking, as there is a muscle for speaking out loud). I have no wish to dispute that this is indeed an example of thinking, and I agree that such thinking is completely independent of the vocal apparatus (no subtle movements of the larynx and tongue). The question is whether the other two cases are also cases of thinking.

64

Part I

It is hard to see how the second kind of case could fail to be a case of thinking: if I start out thinking in the nonbodily style and then modulate to moving my larynx silently, I surely don’t stop thinking. My brain is what enables me to think without moving my vocal apparatus, but my vocal apparatus can play the same role—it can enable the thought process. I might switch to it precisely because it improves my thought process—in some strange way it enables me to articulate my thoughts better. The process is like reading while moving one’s lips: some people can read in no other way, and others do it in order to savor what they are reading (especially poetry). In such a case the person is “thinking with her mouth”—and perhaps she could not think as she does without thus employing her vocal apparatus. We can imagine a species that cannot think at all without vocal involvement—they are psychologically unable to think without recruiting their voice muscles. Thus, we should allow that these are examples of thinking—thinking with the voice. When you observe such a person subtly moving her lips with a pensive look in her eye you should think, “Ah, she is deep in thought.” Her thinking is her moving her lips in that way (enabled by the brain, of course). Just as her reading is her moving her lips (etc.), so her thinking is her moving her lips (etc.). Her thinking is not “in the head.” The apparatus of thinking here involves the body (part of it). Or rather: some thinking is purely an inner process, but other thinking is partly an outer process. Thoughts come in different categories that straddle the inner–outer divide, real as that divide is. But why should adding actual sounds alter the picture? If I start to make sounds as I employ my vocal apparatus that does not prevent me from thinking or negate the fact that I am thinking—I am now, as we say, “thinking aloud.” My making those noises is my thinking—so thought can be audible as well as silent. We must be careful to understand this claim correctly: it is not that the thought is identical to the noises themselves, as if anything making those noises would be thinking; it is that my thinking is my making those noises—a certain kind of intentional act. Some intentional acts of thinking are purely inner occurrences, but other intentional acts of thinking are outer occurrences—in the latter case, the thinking is the speaking (though not the sounds considered in themselves). Some of my thinking I do privately, but I also think publicly on occasion—as when I sit alone in my study and mouth sentences to myself. This kind of self-directed conversation is my thinking; and by thus “talking to myself” I expedite my thinking. If someone comes into my study and observes me talking to myself in this way, she will rightly comment, “Shush, he’s thinking!” A stubborn dualist might object: “The thought is not identical to the speech act but the cause of it; the thought lies behind the speech act and accompanies it—so all thought, properly so-called, is silent, private, and motionless.” But this is an implausible position because we don’t perform two acts here, one private and one public. When I start to voice my thought, I switch to another mode of thinking; I don’t stay with my previous mode and add an extra type of act to it. I abandon the previous mode in the

Mind65

hope that the new mode will help me to think more effectively—and it often does. In the same way, if I switch from reading silently to reading aloud, I don’t carry on reading silently and add to that my audible reading—I just adopt a different kind of reading. Likewise, when I start thinking using spoken words I do not carry on in parallel using words inwardly, with the latter constituting the real thought; I stop employing language in the inward style and start employing in the outward style. If I am thinking while speaking it is in virtue of that fact and not the putative fact that I am inwardly rehearsing words as I produce additional outer words. It is not that I keep on speaking to myself inwardly and then superimpose acts of outer speech; I just switch from one type of speaking to the other. The question, then, is why we should resist classifying thinking aloud as precisely that—thinking aloud. Why should we not extend the mind to the vocal apparatus? It can’t just be the insistence that the mind is necessarily interior, because that looks like a dualist dogma; we need a reason to deny that thinking can be conducted in a (partly) bodily medium. The most perspicuous way to put the point is that we really have two minds or two types of mental process: inner and outer. Granted the mind is centrally something interior and private, but it is also something exterior and public. As we have a conscious mind and an unconscious mind (or many such), so we have an inner mind and an outer mind, corresponding to the types of thought of which we are capable. To repeat: I can think purely inwardly, without moving a muscle, but I can also think corporeally, by exercising my vocal muscles. These muscular movements are the vehicle of thought (or part of it). I would be a lesser thinker than I am without this vehicle, given that thinking aloud enhances my thought process. And it is the same with writing: some people cannot think properly without writing it down, and the writing constitutes their thinking—here it is the hand, not the larynx, that is the bodily vehicle. As we can communicate with our hands, so we can think with our hands: the mind is therefore in the hands (i.e., the moving of them). The thinking faculty extends to the hands: thinking is (in some cases) moving the hands. For the deaf (and others) who use a nonvocal sign language the same conclusion is warranted: the hands are the vehicle of thought for someone using sign language as an aid to thinking, as they gesticulate away in their study. Where there is language, of whatever medium, there is thought. It might be accepted that one use of language—as an aid to solitary cogitation— counts as thinking, but what about other uses, notably in interpersonal communication? Are all uses of language to be counted as instances of thinking? Of course, all communication involves propositional attitudes, if we follow Grice; but I don’t think we should thereby conclude that all speaking is thinking. When I make an assertion or issue a command I have certain intentions and beliefs, but it would be wrong to say that I am therefore thinking about something—in fact, I am typically thinking about nothing. Only a specific employment of language is appropriate for thinking—the kind

66

Part I

that is typically solitary, quiet, and self-directed. So we shouldn’t say that all speech acts are examples of thinking; only a specific subclass is. Note that I am not defending a behaviorist view of thought: the claim is not that thought reduces to behavior. On the contrary, some thought involves no behavior at all—the purely inner variety. The point is rather that some types of behavior count as thinking—so I am characterizing some behavior in psychological terms. I am saying that some verbal behavior can be described as inherently cognitive—as constituting a thought. I am extending the mind to behavior, not reducing the mind to behavior. In my view the mind comes in two forms, inner and outer, neither of which is reducible to bodily movements. I am thus neither an “internalist” nor an “externalist” about thinking: I am a “mongrelist,” that is, I believe the mind to be a mixture of different breeds. The mind has an interior part and an exterior part—it is partly inner and partly outer. It is a mixed bag. Thinking is rather like singing: we can sing out loud, under our breath, or just in our head.1 All are bona fide cases of singing. If a species did all of their singing silently in the head, it would still be a singing species; and if one day it started to sing out loud that also would be singing. We shouldn’t take one type of singing to fix the definition of singing, thus excluding the other types. We tend to think of the audible kind as primary, but that doesn’t mean we can’t sing silently to ourselves. In the case of thinking we tend to take the inner kind as primary, but that should not prevent us from counting the outer kind as genuine thinking. Some mental concepts (but not all) span the inner–outer divide. Thoughts can occur inside the head, or they can occur outside of it. 1.  It is actually quite remarkable that singing, like thinking, permits these three varieties, because they differ dramatically among themselves. Try singing a song out loud and then carrying on purely in your head: that is an abrupt mental transition, but the brain accomplishes it with barely a pause.

II Language

Meaning Monism

This is Wittgenstein on the irreducible plurality of language-games: “But how many kinds of sentence are there? Say assertion, question, and command?—There are countless kinds: countless different kinds of use of what we call ‘symbols,’ ‘words,’ ‘sentences.’ And this multiplicity is not something fixed, given once for all; but new types of language, new language-games, as we may say, come into existence, and others become obsolete and get forgotten. (We can get a rough picture of this from the changes in mathematics)” (Philosophical Investigations, sec. 23). He then goes on to compile a long list of different language-games, including “singing catches,” “guessing riddles,” and “asking, thanking, cursing, greeting, praying,” concluding with the words: “It is interesting to compare the multiplicity of the tools in language and of the ways they are used, the multiplicity of kinds of word and sentence, with what logicians have said about the structure of language. (Including the author of the Tractatus LogicoPhilosophicus.)” The sentiments expressed here are quite representative of a very influential trend in philosophy: the notion that different kinds of sentence have different kinds of meaning—that sentence meaning is not a unitary affair. The recognition of this fact has been held to be relevant to questions not expressly about language, but also about what we use language to talk about. We must be brought to see that our sentences may function in different ways and that their subject matter reflects these ways—for instance, ethical sentences should not be regarded as “fact-stating” but as “expressive.” We might call this view “meaning pluralism”—the view that there are many types of sentence meaning, many ways that sentences can mean (and that people can mean things by them). In contrast, the rejected view (attributed to the Wittgenstein of the Tractatus) can be called “meaning monism”—the doctrine that sentences are all of one type, that meanings are always uniformly the same, that there is something deeply common to all meaningful utterances. Even the differences between assertions, questions, and commands are superficial; underneath there is a unitary layer of meaning. Meaning is One, not Many. In this essay, I argue that meaning monism is true and meaning pluralism is false— the Wittgenstein of the Tractatus was right and the Wittgenstein of the Investigations was wrong. The argument for this conclusion is surprisingly simple, stemming from two well-received premises; but implementing the meaning monism that results is less straightforward. The two premises are as follows: (1) Words mean the same thing no matter what kind of sentence they occur in; and (2) the meaning of a word is its contribution to sentence meaning. For ease of reference, I will call these the “constancy principle” and the “context principle.” Since they are quite familiar I won’t say much in their defense, except to make a few remarks by way of clarification. As to constancy: the words “shut” and “door” mean the same thing in “The door is shut” and “Shut the door!” There is no ambiguity in these words, according as they

70

Part II

occur in one kind of sentence or the other. There is no reference shift or variation in sense (such as Frege detected in indirect discourse), and the same lexical mastery suffices for both types of occurrence. What differs semantically between the two sentences is just the grammatical mood, not the meaning of individual words. As to the context principle: the meaning of a word is its contribution to the meaning of whole sentences, since it is the function of words to compose sentences, not to stand nakedly alone. Only sentences can be used to perform a significant speech act; words are subordinate to that purpose. Thus the meaning of “door” and “shut” in “The door is shut” consists in their contribution to the meaning of the sentence. If we think of this meaning as possessing truth conditions, then the meaning of the words is their contribution to truth conditions. When we grasp the meaning of these words in that context, we grasp how they fix a particular truth condition, to the effect that “The door is shut” is true if and only if the door is shut. The meaning of a word in the context of an indicative sentence is constituted by the way it determines a specific aspect of the truth conditions of the sentence in question—generally, the objects and properties on whose disposition the truth conditions hinge. (This is to assume a truth-conditional theory of meaning, which I do so now for expository reasons: but see below.) Why do these two premises together entail meaning monism? Suppose that meaning pluralism were true: then one or other of the two principles would have to go. If different kinds of sentences had different kinds of meaning, then the context principle would entail that words vary in their meaning between sentence types, since their meaning would consist in contributing to different whole-sentence semantic properties. But if words have a constant meaning, they could not have their meaning fixed by contributing to different kinds of sentence meaning. If words have a constant meaning across sentence types, and if their meaning is their contribution to sentence meaning, then sentence meaning must be unitary. In other words, granted the context principle, meaning pluralism at the level of sentences implies meaning pluralism at the level of words—words can’t mean the same thing in different types of sentence. For example, if “door” and “shut” mean the same thing in both indicative and imperative sentences, then their meaning cannot consist in their contribution to fixing the different kinds of meaning possessed by those sentences. That is, their meaning cannot consist in determining either truth conditions or obedience conditions—for that would make the words systematically ambiguous. We give up word constancy if we accept that “door” and “shut,” depending on the sentential context, can mean either a part of a truth condition or a part of an obedience condition. We have here used two different semantic concepts at the level of whole sentences (truth and obedience), and then claimed that words can be defined using both of them: the same words either fix truth conditions or obedience conditions. In general, if we characterize sentence meaning pluralistically, we will end up denying word constancy, granted the context principle. Words will then mean indefinitely

Language71

many things, given the indefinitely many kinds of “language-games” (meaning types) they can contribute toward. If the meaning of a word is its “semantic role” in whole sentences, then the meaning will be multiple, given that sentence types are multiple, because the word will have many semantic roles depending on its context. We can save word constancy while accepting sentence pluralism by rejecting the context principle, because then word meaning will not be fixed by contribution to sentence meaning but by something else entirely; but the context principle is strong and well motivated. So meaning pluralism requires us to reject either the constancy principle or the context principle or both. By contrast, if meaning monism is true, we can keep both principles, because then there will be just one concept used to characterize the meaning of sentences, and hence words will have a fixed meaning, no matter the kind of sentence in which they appear. For instance, if all sentence meaning is a matter of truth conditions, then word meaning will just be contribution to truth conditions: semantic role will just be impact on truth conditions. Wittgenstein, in the passage quoted, speaks of different types of “sentence,” different types of “language game,” and different types of “use” of sentences. He shows no respect for the customary distinction between semantics and pragmatics. But it is entirely consistent to allow that speech acts may be of irreducibly many kinds while sentences are of a single kind—pragmatic multiplicity combined with semantic unity. Our intentions in speaking can indeed vary widely, as can the consequences of our speech; but it does not follow that sentences are of many different semantic types. To use Wittgenstein’s own analogy: the same tool can be used for many jobs, as when we use a screwdriver to insert screws or as a paperweight or as a weapon (consider all the things you can do with a stone). Speech act pluralism does not entail meaning pluralism. This is why you can issue a command without using an imperative sentence—as in uttering, “Would you like to shut the door?” or “I see the door isn’t shut,” as a way to get someone to shut the door. So we should not move too swiftly from the multiplicity of language-games (pragmatics) to the multiplicity of sentence types (semantics). I may have all sorts of motives for kicking you, and produce all sorts of effects by so doing, but it doesn’t follow that kicking is not a single type of action. There are no doubt many kinds of “kicking-games,” but kicking is constant among them. Still, aren’t there many grammatical types of sentence? Indicative sentences, imperative sentences, and interrogative sentences are all grammatically different (as marked in the syntax). So how can meaning monism be true? But if meaning monism were not true, then one or both of the constancy principle or the context principle would have to be mistaken. We therefore appear to be in an awkward theoretical position. One popular way out is to assimilate all grammatical types to a single type—with the indicative form chosen as the basic type. Thus we translate imperatives and interrogatives into indicatives, and apply a uniformly truth-theoretic semantics: “Shut the door!” means “I order you to shut the door” and “Is the door shut?” means “I want you to tell me

72

Part II

whether the door is shut”—or something along these lines. This is theoretically attractive, because now we can easily preserve word constancy and context dependence: all sentences have truth conditions as their meaning, and words always contribute to truth conditions. However, for various reasons, such a translational and reductionist approach is implausible and procrustean: it would be better if we could avoid having to go that way. At any rate, I am interested in whether we can preserve meaning monism without assuming such a program of paraphrase. That is the theoretical challenge raised by our discussion so far. We know that meaning monism has to be true, given our two premises, but it is unclear what kind of whole-sentence semantics to prefer. What we need is some way to unify indicatives, imperatives, and interrogatives without claiming that the latter two are just misleading forms of the indicative. We have a proof that some single concept unites the different sentence types, but we don’t yet know what that concept is. So what is it? Let’s begin by noting some obvious semantic facts. We can say the following three things: “The door is shut” is true if and only if the door is shut; “Shut the door!” is obeyed if and only if the addressee makes it the case that the door is shut; and “Is the door shut?” is correctly answered affirmatively if and only if the door is shut. In each case “if and only if” is followed by the indicative sentence “The door is shut.” So this is something each sentence has in common, despite their differences. What differs is the semantic predicate applicable to the sentences in question: “is true,” “is obeyed,” and “is correctly answered affirmatively.” We can say, quite naturally, that each sentence “corresponds” to the possible state of affairs of the door being shut—that is, what is expressed by “The door is shut.” So we have a correspondence of three sentence types with a single state of affairs—where that state of affairs can make a statement true or make a command obeyed or make a question correctly answered affirmatively. Each sentence can be said to correspond to, or be correlated with, the (possible) state of affairs of the door being shut—either truth-wise or obedience-wise or answered-wise. This is the state of affairs “represented” by the sentence in question—the state of affairs you have to identify in order to understand the sentence. There are three different relations to that state of affairs, as signified by the semantic predicates employed, but the same state of affairs is common to each sentence type. We can’t capture the common core of meaning with the semantic predicates themselves, since they mean different things, but we can unite them under a more abstract concept—what I am calling correspondence. Thus “The door is shut” corresponds to the door being shut, “Shut the door!” corresponds to the door being shut, and “Is the door shut?” corresponds to the door being shut. None of these sentences corresponds to grass being green or snow being white or any other state of affairs. So now the meaning of a word is its contribution to fixing correspondence, where this unitary semantic relation can be specialized into three semantic types, corresponding to the three grammatical moods.

Language73

Word constancy is thereby preserved and the context principle respected. We have the meaning monism we sought. It turns out to be strictly incorrect to say that word meaning is contribution to truth conditions, because that would exclude imperatives and interrogatives, but it is fine to say that word meaning is contribution to something that has truth conditions as a special case—a determinable of which truth conditions are a determinate. Word meaning is just a bit more abstract than we thought, since it consists in fixing the determinable correspondence. There is nothing in this general conception to preclude a language made up wholly of imperatives, where no sentence is ever a candidate for truth; and clearly such a language would contain meaningful sentences. What the language faculty basically contains are rules of semantic correspondence that are prior to rules that fix truth conditions or obedience conditions; these more abstract rules are interpreted with respect to particular grammatical types. The correspondence rules project sentence types onto possible states of affairs, without yet specifying what the grammatical mood is. Thus we can represent sentence processing as first mapping an input sentence onto a state of affairs and then determining what grammatical type is involved. My linguistic competence tells me I am dealing with a sentence that corresponds to the state of affairs of the door being shut and also that it corresponds in the indicative way or the imperative way or the interrogative way. Grasping the meaning of a word is strictly a matter of the first stage of this process, because that is where the unity of meaning lies, not in the consequent recognition that the sentence is either indicative or imperative or interrogative. Word meaning is pregrammatical in that sense: it doesn’t have grammatical mood built into it. Yet it is a function of sentence meaning. Sentence meaning has two layers, in effect: the upper layer containing grammatical mood, and the lower layer that is neutral with respect to mood, dealing only in abstract correspondence. Such a conception is quite revisionary, and doubtless rather obscure, but so far as I can see it is the best way to respect meaning monism, which is compulsory given our argument. First-level sentence meaning does not have grammatical mood built into it, so it is not like the classical proposition, which is inherently statement-like (truth-evaluable). So the present theory is not the same as “sentence radical” treatments of nonindicative sentences, since sentence radicals are conceived as themselves indicative sentences (as in “Make it the case that the door is shut!”). Rather, sentence meanings are constructed from a mood-neutral core (“correspondence”) and an overlay of grammatical mood. One level of meaning maps the sentence onto a possible state of affairs, without commitment to a specific mood, while a second level introduces the mode of that relation by supplying a mood (by means of syntax). We could represent the first stage as operating on the string, “door, shut,” and giving as output the state of affairs the door being shut, while the second stage inserts the mood by suitable symbols: “!,” “?,” and whatever we choose to stand for the indicative mood (say, Frege’s assertion sign). Thus a perspicuous representation of the semantic structure of, for example, a

74

Part II

complete imperative sentence would be, “!, door, shut.” In other terminology, the brain first computes the correspondence (or correlation) relations for “door, shut” and then computes the semantic mode corresponding to the mood signifier “!.” The meanings of “door” and “shut” belong to the first operation and do not include the output of the second operation. They are mode independent, yet sentence dependent. Thus we reconcile word-constancy with the context principle—by excluding grammatical mood from sentential context. What bearing does all this have on issues revolving around meaning pluralism? It tells us that all linguistic representation is fundamentally the same. Words mean the same no matter what their linguistic environment may be, and sentences share a core of meaning no matter how they may differ as to grammatical type. Words and sentences can have very different sorts of subject matter, but they do not represent their subject matter in varying ways. Just as we cannot read semantics off pragmatics, so we cannot read semantics off ontology. For example, numbers and material objects differ ontologically, no doubt, but sentences about numbers and material objects have the same kind of meaning. Similarly for sentences about facts and values: different subject matter, yes, but same semantic form. Wittgenstein tried to discourage us from ontological assimilations by pointing out the varieties of language-games (sentence types), but there are no such varieties at the basic semantic level; so any ontological discouraging needs to be justified in its own terms, not by attention to alleged linguistic multiplicity. Certainly, meaning monism does not entail ontological monism. Semantics and ontology are separate enterprises—one concerning representations, the other concerning what is represented. It is a use–mention confusion to conflate the two. In the Tractatus Wittgenstein thought that all meaningful sentences reduce to assertion. In the Investigations he thought that assertions are just one type of sentence in a heterogeneous collection, united merely by family resemblance. It turns out the correct view lies between these two extremes: sentences do come in irreducibly different types (indicative, imperative, interrogative, and perhaps others), but that does not thwart meaning monism, which is the only way to do justice to the constant manner in which words mean. All meaning is variation on a single theme—“meaning” is univocal. Wittgenstein later saw language as consisting of loose resemblances among members of a family; in reality it is more like the fixed four-letter code of a family’s DNA. Language has a deep unitary essence.

Against Language-Games

Wittgenstein introduces the concept of a language-game in section 7 of the Philosophical Investigations, using it repeatedly as the book progresses. It is clearly an important concept for him, and it has entered the philosophical lexicon. He writes: “We can also think of the whole process of using words in (2) [the section about the builders and the words ‘block,’ ‘pillar,’ ‘slab,’ and ‘beam’] as one of those games by means of which children learn their native language. I will call these games ‘language-games’ and will sometimes speak of a primitive language as a language-game. And the processes of naming the stones and of repeating words after someone might also be called language-games. Think of much of the use of words in games like ring-a-ring-a-roses. I shall also call the whole, consisting of language and the actions into which it is woven, the ‘language-game.’” We do well to scrutinize this passage carefully. Wittgenstein uses the word “game” in two distinct ways here: he uses it in its ordinary colloquial literal sense to refer to the games that children play; and he also uses it in a novel and technical sense to refer to uses of language that we would not normally describe as games in the previous sense. Ring-a-ring-a-roses is clearly a game in the ordinary sense, while using language to describe (say) a crime scene is not a game in any literal sense. Wittgenstein appears to believe that children acquire their native language (partially or wholly) by playing language-games in the literal sense, and he thinks that calling out names of types of stone and fetching stones in response is a type of game (he doesn’t tell us what this game is called). So he thinks we acquire mastery of “language-games” (technical sense) by playing “language-games” (literal sense). It will be helpful to have more examples of literal language-games so that we can gain a better sense of the concept involved (Wittgenstein mentions only ring-a-ringa-roses). I take it the following all qualify: I Spy, Scrabble, crossword puzzles, spelling bees, and rhyming games. These are all games we play with language, and they meet Bernard Suits’s definition of a game in terms of adopting inefficient means to reach a goal.1 So Wittgenstein (a) believes that such games feature prominently in language learning, and (b) holds that they form a good model for the other uses of language (those that are not literally language-games). Claim (a) is an empirical thesis about how children actually learn to speak, and it strikes me as implausible and unsubstantiated: such uses of language are too advanced to be basic to language learning, and surely it would be possible for children to learn language without playing any games with 1. Bernard Suits, The Grasshopper: Games, Life, and Utopia (Calgary, Ontario: Broadview Press, 2005).

76

Part II

language. However, I am more interested in claim (b), which is a conceptual claim: is the use of language in (real) language-games an apt model for the use of language outside of language-games? It is not, and for a very simple reason: the other uses of language are not games. If I use language to describe a crime scene or to write a physics paper or in a legal document, I am not playing a game—I am engaging in what Suits calls a “technical activity,” that is, an activity intended to achieve a certain goal as efficiently as possible. And, intuitively speaking, these technical uses of language are not remotely game-like—any more than fighting to the death on the battlefield is a kind of game (as opposed, say, to the game of Cowboys and Indians). Not all human activities are games, and not all uses of language are linguistic games; indeed these two points are connected, since we don’t use language in a game-like way when we are engaged in activities that are manifestly not games. There are deadly serious activities that contrast sharply to games, and similarly for uses of language. Not all “language-games” are language-games. So it is highly misleading to describe all uses of language as language-games. If this is intended merely as a metaphor, then it is a bad metaphor; if it is intended as literally true, then it simply isn’t. Wittgenstein never gives any argument or reason for describing all uses of language as language-games; he simply announces that he will speak that way. But we are entitled to ask what is to be gained by so speaking, given that it is literally false. Someone might say that it is just a loose metaphor intended to highlight the active nature of language—we shouldn’t take it too seriously. I would make three points about that line. First, Wittgenstein clearly takes it very seriously indeed, assuming it to reveal a deep truth about the nature of language. Second, what would we say if a philosopher writing about knowledge announced that he was going to refer to the ways we acquire knowledge as “knowledge-games”? He might point out that children often acquire knowledge by means of literal knowledge-games, such as quizzes and riddles, and then claim that it is illuminating to model all knowledge acquisition on such games. We would surely be right to protest: “But there is nothing game-like about acquiring historical or scientific or mathematical knowledge!” That is just an irresponsible and misleading use of the concept of a game. Or suppose a philosopher of science proposed to speak of “science-games” on analogy with real games: isn’t that just a strange and erroneous way to talk about the sciences? Science isn’t a game! To call biology and chemistry “games” is to misrepresent their very nature. It is to suggest that science is somehow unserious, frivolous, a form of entertainment. Third, why does Wittgenstein pick on the concept of a game to highlight the active and heterogeneous nature of language? It is true that he invokes the alleged family resemblance character of the concept of a game to model his family resemblance conception of language, but there are plenty of other (putative) family resemblance

Language77

concepts he could have chosen instead. If he merely wanted to point out that there are many heterogeneous types of language use, with no common underlying thread, he could have compared language to any of the following: faces, numbers, the arts, the sciences, dances, musical compositions, hairstyles, sports, furniture, and no doubt many others. All these show great variety in their instances, and it is arguable that they share no common essence (at least Wittgenstein would argue this way). So why doesn’t he compare language to those things? Why the particular family resemblance concept game? Why not speak of “language-sports” or “language-dances” or “language-arts”? He must presumably have thought that games in particular afford the best analogy for the many uses of language, but we search his text in vain to find any rationale for that decision. And it is evident that much language use is not at all game-like, if we take that concept literally. It is no more game-like than chopping wood. Can we fill the gap in Wittgenstein’s reasoning? Well, there is generally no winning and losing in language-games, and no stipulation of unnecessary obstacles. Nor is there typically any game equipment, like bats and balls. There are no fouls or penalty kicks either. It is true that both are rule-governed activities, but then so much else is as well (legal proceedings, royal behavior, table manners, religious observance, etc.); and the rules of grammar are not like the rules of typical games like chess—they are not rules that prescribe what counts as winning. There seems to be no interesting respect in which using sentences is like playing a game. Language is no more a game than science or art or industry: that is why we don’t speak of “science-games” or “art-games” or “industry-games”. So why should we accept Wittgenstein’s invitation to speak of “language-games”? Why don’t we bluntly retort: “But language isn’t a game!” We can accept that language use is very heterogeneous, as games are, without supposing that it is a type of game. We can also accept that we sometimes play games with language, without accepting that all language use is game-like (if you started playing around with language in a court of law, the judge would likely warn you, “This is not a game”). So why does Wittgenstein talk in that way, and why do his readers tend to go along with him? He never really tells us and I have no good idea, but it may have something to do, subliminally, with the feeling that language use is not practical in the way nonlinguistic behavior is apt to be—that is, words in themselves don’t have practical effects. Language consists of “mere words,” not consequential deeds (“Sticks and stones may break my bones, but words will never hurt me”). Using language is like playing a game in that it is removed from serious practical life. If that is the basis for the attraction of the game metaphor, then it strikes me as feeble and wrongheaded, as well as philosophically shallow. Absent any other rationale, talk of language-games is pointless and misleading. It should be dropped. We can speak of different language types, or

78

Part II

different language forms, or different language practices, but we should not employ the concept of a language-game to characterize the general nature of language. There are no language-games, except those that are properly so-called.2 2.  Characterizing the entirety of language with the concept of a game is like characterizing the entirety of culture with that concept—all the customs, institutions, and social structures of a society. No doubt games make up a part of culture, but it would be enormously misleading (short of some startling new theory) to speak as if all of culture consisted of games. We might then wonder how the subclass of “games” typically so called differs from all the other activities that have been brought under that concept. An anthropologist who simply announced that he was going to call every aspect of culture a “culture-game” would invite puzzlement and protest.

Meaning without Language

Suppose there is a planet somewhere in the galaxy on which the following peculiar conditions obtain. There are rational agents living there who form Gricean intentions to communicate: they intend to produce the belief that p in some audience by means of the audience’s recognition of that intention. However, contrary to what they believe, they lack the power of speech—they possess no functioning vocal apparatus. When they form an intention to make a particular utterance, they make no sound at all, though it seems to them that they do. They intend to produce a belief by uttering a sentence, but all that happens is that they have the illusion that they are speaking. Let us suppose, though, that a benevolent alien, knowing their vocal plight, arranges things so that a corresponding illusion occurs in the audience’s mind—it is as if they hear an utterance of the sentence in question. The audience accordingly forms the intended belief. It thus appears that on this planet (call it “Speechless Twin Earth”), the inhabitants mean things and manage to communicate, despite the absence of speech. There is no actual linguistic behavior, just the appearance of it. There are no utterances in a public language. But does language exist on Speechless Twin Earth? That is difficult to say. We might say there is an illusion of language, but no language. Merely having the impression of language is not sufficient for language, it might be thought. If so, then there is meaning and communication on this planet, but no language. But it might also be thought that there is language, because verbal strings go through the minds of “speakers” and “hearers”—and each understands these strings. There is no linguistic behavior of the kind found on Earth, but still it might be said that language exists on Speechless Twin Earth—in the heads of the inhabitants. But we can change the example slightly to sidestep this issue: let’s suppose that though the speaker (who does not speak) has an illusion of speech the hearer has no illusion of hearing—he just forms the belief by suddenly coming to know what the speaker intended to communicate (this is just the way the helpful aliens have arranged things). We would not say that these people communicate by means of language, since nothing about the speaker’s illusion of speech plays a role in causing the right belief in the audience. There is no shared language, and the speaker merely hears sounds in his head when he forms communicative intentions—no actual speech act is performed. In no sense is one person talking to another by means of language. I would say that in these conditions there is no communicative language on this planet. Here is another case: people form Gricean intentions, but when they speak only garbled words come out. We can call this “Aphasic Twin Earth.” However, the aliens arrange things so that the aphasia makes no difference to communication: they ensure that the target audience forms the belief that the speaker intended (by coming to know the speaker’s intentions). Since there is no grammatical structure to the utterances

80

Part II

on Aphasic Twin Earth, it would be wrong to say that those utterances constitute a language—it’s all just random noise. Yet there is “speaker” meaning and communication. There is not even the illusion of coherent speech. The speaker’s intentions simply bypass the defective vocal organs, thus allowing meaning and communication without the mediation of language. But once we have accepted that, we can take the thought experiment one step further: we cut out any attempt at speech at all. Suppose the inhabitants learn of their problems with speech, so that they no longer even attempt to utter coherent sentences; but they also learn of the helpful aliens and reason that they can simply form the right Gricean intentions and let the aliens take care of the rest—the appropriate beliefs will be instilled in the audience in the intended way (by recognition of the intention to produce belief). In fact, we can cut the aliens out of the story and assume a naturally evolved mechanism for belief transmission. Suppose the brain states correlated with Gricean intentions send out electromagnetic waves that can be detected by a suitable sense organ. Then what happens is that would-be communicators form the right Gricean intentions, and then rely on the sensory mechanism to alert audiences to their attempts to communicate. The communicator has the intention to produce the belief that p in audience a by means of a’s recognition of that intention, and the recipient detects the presence of that intention (by means of her sensory mechanism) and forms the requisite belief for the reason intended. In this story, no mention of language is made—whether illusory or garbled. No linguistic act is performed, internally or externally. On this planet (call it “Telepathic Twin Earth”) there is no language of any kind—yet there is meaning (“speaker meaning”) and communication. People mean things and they get their intentions across—but there is no language. All we have are Gricean intentions and brain scans. Thus there can be meaning and communication without language—and not just “natural meaning” but what Grice called “non-natural meaning.” Another way of putting the point is that it is possible to intend to produce in a the belief that p by means of a’s recognition of that intention without making an utterance u—and to succeed in that intention. We can cut the utterance out and keep the rest. Grice spoke of having certain intentions in making an utterance u, which confer meaning on u; but in my Twin Earth cases there is no u that is uttered, and yet there is still meaning and communication. The existence of u is accidental to the existence of meaning. What lessons can we draw from these thought experiments? The first is that the study of meaning and communication is not eo ipso the study of language: for meaning has no essential connection to language. Granted that the constitutive elements of meaning are contained in Grice’s theory, they can exist without benefit of language. And if we accept that speaker meaning is the basis of linguistic meaning, we don’t need to say anything specifically about language in order to account for linguistic meaning—beyond some account of how speaker meaning gets attached to linguistic symbols (say, by convention). The core concept of meaning applies even without language in

Language81

the picture. There is nothing inherently or essentially linguistic about meaning and communication as such. Just as there can be thought without language, so there can be meaning without language. So the theory of meaning, properly so called, is not necessarily concerned with language at all—except derivatively. The theory of meaning is about psychology, not speech acts, linguistic behavior, or language itself. Meaning is, in one sense, in the head—not in public sounds, marks, and gestures. Meaning consists in having a certain complex intention, and communicating consists in others becoming aware of that intention, by one means or another (by means of linguistic utterances or in some other way). The second point is that on the speechless planets there can be no “linguistic turn”— since there is nothing linguistic to turn to. The philosophers there cannot suggest that philosophy is best pursued through the study of language, there being none—though they can suggest that philosophy is best pursued by means of the theory of meaning. There is nothing to stop them from formulating different theories of meaning—truth conditions theory, verification conditions theory, imagistic theory, direct reference theory, or what have you. It is just that these theories will concern psychological contents (propositions) not sentences. It is hard to see how they could contemplate a use theory, though, because there are no words to be used (so linguistic use is not constitutive of meaning as such). We will no doubt want a theory of Gricean intentions that reveals them as complex and compositional, finitely based, and acquirable. So many of the same concerns of typical philosophy of language will carry over to nonlinguistic meaning, but they will be differently formulated and pursued. The theory of meaning, in its pure form, is no more essentially about language than it is about vocal language: just as language can be vocal or gestural or even olfactory and still be meaningful, so it need not exist at all for meaning to exist. The connection between meaning and language is contingent. There could have been meaning and communication in the world without language ever existing, as there could have been thought in a language-less world. (I am not speaking of a supposed language of thought here; I mean that meaning can exist without an ordinary public language.) Then what is philosophy of language up to? Does it rest on a mistake—the mistake of thinking that meaning and language are inextricable? No, that would be too strong; but we do need to rethink the connection between meaning and language. In what sense exactly do words and sentences have meaning? Here we need to tread carefully and critically, because we have potent sources of error to contend with. The simplest thing to say—and it would not be completely wrong—is that words and sentences don’t have meaning at all: only people can mean things. Just as only people can refer or state or insist or express, so only people can mean: words and sentences cannot do any of these things—they are essentially the acts of people. But that seems too revisionary, because it is surely not false to say that “snow is white” means that snow is white (in English). It may not be false but it can certainly be misleading, and has misled. The

82

Part II

natural thing to say is that words mean things derivatively—while people mean things nonderivatively. We confer meaning on words by using our original personal meaning; we impose meaning on words, projecting from the inside out. Words therefore don’t have meaning nonderivatively (“intrinsically”), but they do have meaning—they are not meaningless! That is fine, so far as it goes, but it is not yet quite strong enough to capture the precise way in which words “have meaning.” What does this mean exactly—or what should it mean? Suppose I paint a wall red. I have “conferred” the color on the wall, “imposed” it, brought it to the wall—in contrast to the red naturally found in a red rose, which is “nonderivative.” But still the red wall has that color intrinsically once painted, just like the rose. The same could be true of the shape or weight of something. But meaning isn’t like that—it doesn’t become a property of a word comparable to its other properties. It is not that the word “snow” has a certain sound or shape and is made of four letters and means snow—as if the last property were as intrinsic as the others. You can’t impose a meaning on a word as you can impose a color on it (with colored ink). The meaning remains extrinsic to the word, merely tacked on. The contrast I have in mind is similar to the contrast between the function of an artifact and other properties it has. A table has the function of supporting plates of food, say, as well as having four legs or being made of wood or being brown; but it would be wrong to suppose that the function and these other properties were on a par. The function is not intrinsic to the table, as its color and shape are (even if these have been “imposed” from outside). Hence we have an inclination to say that the table doesn’t have that function—we use it to perform a function. Left to its own devices, the table has no function at all: its functionality is completely relative to its users. This is not like the function of the heart: that is not imposed by us, or relative to our uses of the heart, but is a property it would have even if we never even knew of its existence. But words don’t have meaning in that way; they have meaning in the way an artifact has a function—as a result of intentional human use. If the use ends, the meaning and the function disappear. Hence it is natural to say that a word means nothing—it is only we who mean things by words. If we didn’t mean things by words, they would be entirely meaningless. So we need to add to the point about derivativeness the extra point that words don’t have their meanings inherently—as part of their nature as marks or sounds. When we say that words have their meaning “in virtue of” Gricean intentions, we don’t (or shouldn’t) mean that in the sense in which we say that a sculpture has its form “in virtue of” the actions of the sculptor. Words have meaning in roughly the way puppets do: we can say that a puppet performs speech acts, but we would be badly confused if we thought that it does so independently of the ventriloquist. It is an improvement to say that a puppet only speaks “derivatively,” but again we would be mistaken if we thought that this was like the way it has a certain color or size. Rather, a puppet can be

Language83

said to speak only in the sense that someone is speaking through the puppet. Words are like puppets that we speak through: we mean things by words, but words don’t mean anything themselves, not really. We would probably do well to avoid the phrase “have meaning” in application to words, given its similarity to “have shape,” and instead always use “we mean things by words.” To say a word has a meaning is just to say that we mean something by it (nothing comparable holds of “has shape”). In an ideal language, no one would ever say that words (tokens or types) have meaning or refer or state things; we would always say that people mean things or refer or state things (by the use of words). This point is really quite obvious, but it is amazing how often and easily it is forgotten. It may well be true that it is as if words have their meaning inherently—that is the way they strike us. We see or hear them as meaningful, not as empty signs that need meaning brought to them. Words seem pregnant with meaning, suffused by meaning. But this is really an illusion, borne of excessive familiarity—like the illusion that knives and forks have their function inscribed on them. We understand words so quickly and easily that we think their meaning is as present in them as their sound or shape, but actually it is a matter of learned association—of long-term memory (when we forget what a word means the illusion of immediacy disappears). We certainly don’t see or hear meanings, as opposed to seeing or hearing that a word means something. Nor do we literally “put our meaning into words,” as we might put a certain accent into their pronunciation. We are thus prone to “delusions of inherency” where meaning is concerned, and so we think that meaning and language are more closely intertwined than they are. Really, words are just conventional vehicles of meaning; they are not units of meaning themselves. Meanings are composed of intentions, according to the Gricean analysis, and language merely goes proxy for those intentions. It is rather like the relationship between mathematics and mathematical notation: we should not confuse numbers with numerals (though we are prone to), thus conflating mathematical entities with mathematical notation. Language is to meaning as mathematical notation is to mathematics: a dispensable contingent accompaniment. Or again, meaning and language are like money and currency: it is not that bits of paper and coins have economic value in themselves (though people are prone to think they do), but rather that they are dispensable vehicles of economic transactions. The study of money is not the study of paper and metal; and there are possible planets on which economic transactions occur without the benefit of tangible currency (even taking place by means of direct brain connections). All this is fairly obvious, even banal, but it is not always fully absorbed; what is not obvious is the consequential point that the subject called the “philosophy of language” is often not really about language at all. Consider, for example, Russell’s theory of descriptions: this is not really about bits of language, conceived as marks or sounds (or

84

Part II

even abstract syntactic strings), but about thoughts (better, the content of thoughts, i.e., propositions); and similarly for other issues of logical form and semantic analysis (names, demonstratives, quantifiers, etc.). The reason theorists speak of language is that language is more concrete and palpable, as well as because of a lingering behaviorism; but the object of interest is not the vehicle of meaning but meaning itself, which is detachable from language. We are really interested in sense, to put it in Frege’s terms, not signs of sense. The philosophy of meaning is therefore not the philosophy of language: pulling apart meaning and language, as I did with my Twin Earth thought experiments, merely drives home this conceptual distinction. The phrase “philosophy of language” is really a misnomer, analogous to calling the philosophy of mind “the philosophy of behavior.” This is even more obvious if we explicitly conceive of language as consisting in linguistic behavior—whether speech acts, the use of words, or dispositions to assent. None of these has anything essential to do with meaning, as the thought experiments demonstrate (assuming Grice to be on the right track about meaning in general). Meaning is a matter of the mind, not the larynx. The study of meaning is not the study of symbols. People mean by having suitable intentions; language is a contingent accompaniment. What is linguistics about? I would say it is mainly about the vehicle of meaning not meaning as such—phonetics, grammar, pragmatics, the pairing of sound and meaning. It is not about what meaning is—that is, communicative intentions, following Grice. Intentions are combinations of a psychological type and a proposition—intentions to get others to form beliefs are intentions embedding another psychological type (belief). So the study of meaning is fundamentally the study of intentions, beliefs, and propositions; where these may or may not issue in verbal utterances (i.e., language). That is not the concern of a linguist, strictly so called, since they are not part of language—they belong to psychology (empirical and philosophical) and to logic (the analysis of propositions). A study of meaning and communication as it exists on Telepathic Twin Earth will have nothing to say about language as such, and so will fall outside the purview of linguistics. The philosophy of language can certainly concern itself with language—as, say, in the study of speech acts—but the theory of meaning is not per se a theory of language. Words and sentences “have meaning” in the derivative sense specified earlier, but meaning itself is not essentially a linguistic matter. Meaning is something more general than language, and separable from it. In principle, meaning and communication can exist in the complete absence of language. Grice’s point was that sentence meaning is not separate from speaker meaning: rather, sentence meaning is constituted by speaker meaning, plus something like convention. So there is no more to sentence meaning than speaker meaning (plus convention), which is a matter of having suitable intentions. But if I am right, speaker meaning does not logically presuppose language—it can occur in the total absence

Language85

of anything in the nature of utterance. The essential ingredients of speaker meaning don’t even presuppose that the “speaker” speaks or has any mode of symbolic expression. It is possible to mean and communicate without having any vehicle of expression, even of the simplest kind. Thus we can have meaning without language, communication without utterance, information transmission without symbols. The study of signs is the wrong way to study meaning: it conflates the essential with the incidental. It is no better than studying thought by studying sentences, or studying pain by studying grimaces. The linguistic turn is as wrong for meaning as it is for other aspects of the mind.

For Privacy

In section 258 of the Philosophical Investigations Wittgenstein considers the case of someone wanting to keep a diary about the recurrence of a certain sensation. He wonders how this person might bring it about that his word refers to that type of sensation and to that alone. He writes: “A definition surely serves to establish the meaning of a sign.—Well, that is done precisely by the concentrating of my attention; for in this way I impress on myself the connexion between the sign and the sensation.—But ‘I impress it on myself’ can only mean: this process brings it about that I remember the connexion right in the future. But in the present case I have no criterion of correctness. One would like to say: whatever is going to seem right to me is right. And that only means we can’t talk about ‘right.’” The idea is that the diary keeper focuses on the sensation S and makes an effort to remember that this is the sensation named by “S.” He then relies on his memory to decide what to call the sensation he is now having: if it is the same as the sensation he had yesterday when he called it “S,” he should call it “S” today. He simply has to recall what type of sensation he yesterday called “S.” But Wittgenstein thinks there is a problem here, because the person must remember correctly what he meant by “S” yesterday—and he has “no criterion of correctness.” Presumably Wittgenstein must mean he has no other criterion except that of his apparent memory: he can (attempt to) justify his assertion that he meant S by “S” yesterday by saying that he has an apparent memory of doing so. But if someone challenges this memory he has nothing to fall back on—he can only repeat that it sure seems to him that he has a veridical memory that that is what he meant yesterday by “S.” And he has to admit that memory is fallible, so that he might have meant sensation S′ by “S” (as it might be a sensation of green, not a sensation of red). That all seems true enough: he has only his memory to rely on, and memory is fallible. But then Wittgenstein goes on to report what “one would like to say”—namely that “whatever is going to seem right to me is right.” Whether one would “like” to say that or not, it is not a good thing to say—in fact, it is totally misguided. It is clearly false to assert that whatever I (seem to) remember to be correct is correct: that would be to deny that memory is fallible (cf. “whatever I believe to be correct is correct”). Wittgenstein concludes that we can’t “talk about ‘right’” in such circumstances, since there is now no distinction between seeming right and being right. If so, there cannot be a “private language” for sensations: the only way to make room for the notion of correctness here is to allow that the sensation has a behavioral expression that enables other people to correct one’s apparent memories. They will then be able justifiably to say, “You are misremembering what you meant by ‘S’ yesterday” because they can know your sensations by observing how you behave. Hence, Wittgenstein thinks, there can only be a language for sensations if the sensations have public expressions.

88

Part II

That is the gist of the so-called private language argument. The standard objection to it is that it is verificationist in form: it claims that there cannot be a fact of the matter about what I meant yesterday by “S” if I cannot verify my memory of what I meant (and no one else can either); or again, that there cannot be a fact of the matter about whether my memory is correct if I cannot independently verify that my memory is correct. But why should we accept this kind of verificationism here any more than elsewhere? There are lots of things I can’t verify, conclusively or inconclusively, but that doesn’t mean there is no fact of the matter. It is a fact that I meant S by “S” yesterday whether I remember that fact correctly or not, and it is a fact that I mean the same thing by “S” today as yesterday whether I can verify this fact or not. In general, it is a fact what sensations I have on particular days whether I can verify it or not. I might have a terrible memory for sensations, but that has no bearing on what sensations I have. So there will always be a distinction between using a word correctly over time for a sensation and merely appearing to oneself to be so doing. The notion of misremembering the meaning of a word for a sensation is perfectly meaningful, even if there is no one to point out that one is misremembering. This is the standard response to Wittgenstein’s argument, and I am fully persuaded by it. Insofar as there is an argument presented in section 258, this is the right thing to say about it. But there is a deeper and less obvious problem with the argument, namely that it proves too much: for, were it sound, it would prove that there cannot be “private memory” or “private belief” or “private knowledge” or “private perception.” We should first observe that Wittgenstein is not trying to demonstrate that there cannot be private sensations, that is, sensations without behavioral expressions; he is trying to demonstrate that there cannot be a language for such private sensations. The sensations are (logically) possible, but there cannot be meaningful words for them: meaning conceptually requires public criteria even though sensations themselves don’t. A creature could have sensations undetectable to anyone else, but that creature could not talk about them: the private language argument concerns the possibility of semantic facts, not the possibility of psychological facts. Could someone have such private sensations and also have memories of them? On the face of it that seems logically possible, but Wittgenstein could ask whether there is any criterion of correctness for such memories (apart from their seeming accuracy); if not, he will say that the distinction between seeming right and being right is lost. So memories of private sensations are impossible according to this argument. And the same thing is true for knowledge or belief about private sensations: it now seems to me that I am having sensation S, but what is the criterion of correctness for such an attribution? It can’t be just that I feel flooded with the sensation S—someone else has to have an independent criterion for whether I am having S. But then I can’t know that I am having S, according to the “private knowledge argument.” The same argument applies to introspection construed as a perceptual capacity: I can’t be said to

Language89

inwardly perceive S because there has to be a distinction between seeming to perceive S and really perceiving S, and that distinction relies (so the argument goes) on an external check by another person. It is not just a private language that is impossible according to the argument; it is any form of mental directedness to a private sensation—whether by memory, knowledge, belief, or perception. We can have the sensations but we can’t think about them, let alone refer to them linguistically. The answer to this strengthened “private mental representation argument” is that it is (a) hugely implausible and (b) open to the same antiverificationist objection we rehearsed above. There is always a conceptual distinction between being so and appearing to be so, even for private sensations, and that is enough to ensure that the notion of correctness doesn’t disintegrate: there is always the fact, on the one hand, and the mental representation of the fact, on the other. Let us take stock of what the strengthened antiprivacy argument is contending: here I am experiencing a range of sensations with no behavioral expression and I am trying to think about these sensations; the argument says that I cannot succeed in this effort, even if I can think about public things like tables and tigers. I suffer from complete cognitive and semantic impotence vis-à-vis my sensations, even though I can refer to, remember, and know about other things. Just because no one else can know my sensations, I cannot even think about them! What is to stop me forming concepts of the sensations I have? Suppose I go from not having public expressions of my sensations to having them—as it might be, I recover from paralysis: do I then suddenly acquire the ability to refer to, remember, and know about my sensations? And if I lose the public expressions again, do I lose those semantic and cognitive abilities? If I have the concept red, applicable to public objects, and I have sensations of red, do I not thereby have enough to have the concept sensation of red? Why should I need to have other people to confirm (or disconfirm) my impressions that I know, remember, and refer to my private sensations—any more than I need them to confirm my having such sensations? What goes for sensations goes for other mental states—beliefs, desires, intentions, and so on. There is no sound argument that shows that these things must have behavioral expressions, either considered in themselves or as a condition for referring to them (or remembering and knowing them). There might indeed have to be certain connections between mental states and symbols in order for there to be reference— perhaps causal connections—but that is not to say that outer behavior is logically required, such as groaning in the case of pain or sorting behavior in the case of color sensations. No one else need be able to tell what you are feeling, thinking, or intending in order for you actually to feel, think, or intend—or in order for you to refer to or conceptualize what it is that you feel, think, or intend. Full mental privacy is conceptually possible. True, our sensations do typically have behavioral effects that signal to others what we are feeling, but there is no conceptual incoherence in the idea of removing those effects and leaving the sensations and language about them intact.

90

Part II

And there is one more point: Wittgenstein appears to suppose that there can be a private language only if I can mean the same thing today as I meant in the past by a certain word; but what about a language-of-the-moment? Can’t I now mean S by “S” even if I cannot succeed in meaning the same thing by my words over time? What if I have a terrible memory and can’t keep track of what I used to mean by my words: what is to stop me from starting from scratch each day? Maybe in ten minutes I will forget what I meant by “S,” but at the time I knew it perfectly well, and my word had the meaning I bestowed on it. The possibility of this momentary language will not be susceptible to worries about the fallibility of memory (even though those worries turn out to be groundless upon examination). I can thus have a private language-of-themoment in which words refer to sensations and other mental states that are undetectable to others. A private language is therefore possible, contrary to Wittgenstein’s argument (or the argument often extracted from his rather obscure text). But it is a further question whether such a language actually exists (how knowable by others are our sensations?), or whether in fact such a language is necessary for the existence of any kind of language. Can there be an argument showing that words and sentences about public objects presuppose the existence of words and sentences about private objects? That is an interesting question, which I won’t pursue now, but I think the answer is that such an argument can be constructed.1 1. The question is discussed in “On the Impossibility of a (Wholly) Public Language,” this volume.

On the Impossibility of a (Wholly) Public Language

At one extreme lies the view that a public language is impossible: no language can refer to anything except private experiences (those whose existence and nature can be known only to the subject of the experience). Such a view would follow from a radical form of verificationism combined with the idea that experiences are necessarily private. If every statement had its meaning given in terms of verifying experiences, and such experiences could not be known to anyone except their subject, then all language would be private, that is, understood by just one person. If every statement is really about private experience, then the meaning of any statement could be known only to the person having the experience: for only that person could know what his words refer to. At the other extreme lies the view that a private language is impossible: there cannot be a language whose words refer to private experiences (this is Wittgenstein’s view). There can only be a public language—one that refers to publicly accessible things (tables, trees, and sensations that have a recognizable behavioral expression). My position is that neither extreme view is correct, though the former view is closer to the truth. The truth is that both types of language are possible but a public language is possible only because a private language is possible. That is, a public language presupposes the existence of a private language—it is a necessary condition of referring to public things that a language also refers to private things. Thus a purely public language is not possible—though a purely private language is possible. There could not be a language referring only to public things. Our own language lends itself to a mixed position: some of it is clearly public, while some of it appears private. We refer to public things and also to private things. States of mind are private because no one but the subject can really know what they are—others cannot know at all, or their evidence is too shaky to ground properly justified belief in what the relevant words mean. What you mean by “experience of green” might be very different from what I mean, and I can never find out if it is or not. The possibility of a private language thus follows from standard skepticism about other minds. But this kind of mixed position does not hold that public language is dependent on private language; it simply holds that the two coexist. My position is therefore much stronger: public language is inextricably bound up with private language, so that the latter is fundamental. However, this is not because of any kind of verificationism; it simply reflects the way meaning works, even for meanings that involve public objects and unverifiable propositions. One cannot know the meaning of public words without knowing the meaning of private words: the very possibility of public meaning depends on the existence of private meaning. This may seem like a surprising claim, but actually it follows from some elementary and widely accepted assumptions—indeed truisms.

92

Part II

The quickest way to see the point is to consider Gricean accounts of meaning. Let me put it very simply: meaning consists in having conscious intentions, but conscious intentions are private, so meaning in general presupposes words for private intentions, that is, a private language. When I utter a meaningful sentence I have Gricean intentions, and I know what I mean by knowing what these intentions are—so I must be able to refer to my Gricean intentions. But these intentions could occur in the absence of any recognizable behavioral expression (say, if I were a brain in a vat), so my reference to them is essentially private—I know what intentions I have, but no one else does or can. If I say, “The cat is on the mat,” I have an intention to cause the belief that the cat is on the mat in the mind of an audience (real or imagined), but that intention might be known only to me, and my words for it can therefore be understood only by me (no one else can know what mental state I am referring to). Now Wittgenstein thought he had an argument against the possibility of a private language in this sense, but I think (along with many others) that his argument rests on verificationist principles that should be rejected. I don’t propose to discuss Wittgenstein’s argument here;1 my point is that, given the possibility of a private language, all language depends on a private language. In order to understand sentences about cats and mats, we have to understand sentences about intentions to communicate, but these sentences refer to private mental states—things that the subject can know even if others cannot. The same kind of conclusion follows from other constitutive connections between meaning and mind. Suppose I refer to a particular cat with the words “that cat” while I am looking at a cat. My reference with the demonstrative depends in complex ways on my perceptual state and its more basic intentionality—simply put, I am aware of the cat (if I were not, why would I refer to it?). But then to understand my reference one has to grasp this relation of referential dependence, which means grasping what state of awareness I am in. So one must be able to refer to such awareness—as in, “He is visually aware of a black cat in front of him.” But states of perceptual awareness are essentially private: the subject knows just what the state is, while anyone else is at best guessing on the basis of behavior—and anyway we can abrogate any behavioral evidence by imagining a case in which it is absent (paralysis, brain in a vat, disembodiment). At best a mental state like this is contingently public: its essential privacy can be revealed by removing any public criteria from the picture. When a person understands a sentence containing a demonstrative she must grasp the mental states that underlie its use, but such states are essentially private; so there cannot be public reference without private reference. Reference to public objects depends on reference to private objects—basically, states of consciousness. We can’t refer to public objects unless we can already refer to private objects, because reference to public objects proceeds against 1.  I discuss it in “For Privacy,” this volume.

Language93

a background of reference to conscious states, this being what meaning ultimately rests upon. In short, we need to be able to identify a speaker’s mental states if we are to identify his public reference. There are two parts to this argument that need further comment. First, as things stand our language does not obviously include a private fragment, since our conscious states do typically have behavioral expressions. So it appears possible to have a public language that has no actually private fragment. The issue depends on the view we take of our knowledge of other minds: do we now know the mental states of others? Maybe the alleged behavioral signs are inadequate to deliver actual knowledge, so only the subject can really know his mental states—and hence understand his words for those mental states. But we can sidestep this issue by claiming that our present language is only contingently public, because we can imagine situations in which the behavioral signs have been removed without removing the mental states or words for them. Since the mental states could be radically private, without detriment to the existence of a language for them (pace Wittgenstein), we can say that our words for inner states are essentially private—that is their natural condition, a reflection of what mental states are intrinsically. A speaker could have a language for his inner states that was completely private, with no behavioral expression at all, and the words in it would mean what our words mean: this private language would then be the foundation for the speaker’s language for public objects. To put it differently, the private part of our language consists of those words that can be converted into a private language by deleting any behavioral signs that may exist. The thesis will then be that any public language rests on a language that is convertible into a straightforwardly private language. Thus our actual language is not a counterexample to the thesis I am defending. The second point concerns reference to the substratum of mental reality presupposed in meaningful language. It might be objected that the substratum needs to exist, it being what constitutes meaning, but that it does not need to be referred to in the language whose meaning it constitutes. Thus, I must indeed have Gricean intentions in order to mean anything, but it doesn’t follow from this that I have to refer to such intentions in order to mean anything—in which case I may need private mental entities to have a meaningful language, but not a private language for them. I need a mind to mean, and that can be agreed to be private, but why suppose that I need words for that mind in order to mean? This is a perfectly reasonable point, and it brings up an interesting issue: whether reference to meaning is necessary for meaning to exist. After all, a brain is necessary for meaning and reference too, but surely not reference to a brain. The answer to this question is that we need knowledge of meaning in order to mean: it is not enough for me to have Gricean intentions—I must also know that I have them. The reason is that in meaning something by my words I must know what

94

Part II

I have said, and what I have said is constituted by my intentions (as well as the prevailing conventions). One cannot say something and not know what one has said—as if one were unconscious of what one is saying. When we mean, we know what we mean. There are intentions and knowledge of those intentions (this is generally true of intentions). And that requires that I possess a means of mentally representing such intentions, that is, reference to them: they are objects of knowledge. Still, it may be replied, do I need words for my communicative intentions, or just concepts of those intentions? Here we can make two kinds of response. One is that the issue is really about the possibility of public language and private concepts, not words as such: we could reframe the whole issue as being about whether there can be a public language without any private concepts. And if the argument is sound, there cannot, because we need conceptual representations of private mental states, such as intentions. Knowledge of intentions requires concepts of intentions, even if it doesn’t require words for intentions. The other response is to weaken the thesis as follows: every language with words for public objects needs an implicit private language, in the sense that it must at least allow for the potential of a private fragment. Once we introduce private concepts we satisfy this condition, because all we need now are words for these antecedently existing concepts. The potential for an explicit private language is there, because we can easily extend the language by introducing words to express the concepts of mental states that are agreed to be necessary for knowledge of meaning to exist, and hence for meaning to exist. Thus a public language needs an implicit or explicit private fragment—incorporating words for the private substratum. Speakers need to be able to talk about mind in order to talk about body, that is, public objects: for all talk about anything depends upon a grasp of the mental facts that constitute meaning. If we grasp those facts conceptually, then we are only inches from referring to them linguistically, given that we are already speakers of a language. Thus linguistic reference to meaning-constituting mental states is built into any language, at least implicitly. We can therefore conclude that reference to public objects presupposes reference to private objects. We can refer to tables and trees only because we can refer to mental states about tables and trees. It follows that ground-level reference depends on a kind of second-level reference: to refer to public objects we must be able to refer to reference to those objects, that is, mental reference to them (in the form of Gricean intentions or perceptions of particulars or suitable beliefs). Reference to a public object requires reference to a private object that refers to that public object. For example, to refer to a cat I need to be able to refer to my mental representation of a cat, since I must grasp what makes my reference to a cat possible, which is being in a certain mental state. To express it from the audience’s point of view, the hearer will understand my utterance to be a reference to a cat only if she grasps that I have

Language95

certain communicative intentions with regard to cats; but that requires her to grasp a mental state that is about a cat, which is to say she must be able to refer to this mental state of mine (“He is intending to get me to form a belief about a cat”). She can understand my reference to a cat only if she understands her own reference to my intention—and this involves reference to a private object. So ground-level reference to public objects presupposes higher-level reference to private mental states about public objects, because grasp of reference requires grasp of speaker meaning, and that brings in the psychological basis of meaning, that is, Gricean intentions (and whatever other psychological factors you think are constitutive of meaning, such as perceptual modes of presentation). To summarize all this, let us imagine a would-be speaker who is completely paralyzed, so that he has no behavioral expression for his inner states. He perceives various physical objects around him and proposes to introduce words for them. For instance, he introduces the words “table” and “tree” for tables and trees. Other people can observe the things to which he is referring because they see the public objects that prompt his utterances (let’s suppose that he can make utterances despite his paralysis). So this part of his language is seemingly public. However, for that to be possible he must have certain intentions in relation to his words, and he must know what these intentions are, which requires reference to such intentions. Suppose he now introduces words for his intentions, as well as words for other conscious states that accompany his use of words for public objects: these words will not be intelligible to his audience, since by hypothesis his audience has no knowledge of his inner states—they are known only to him. According to my argument, his public words are meaningful only because of the private goings-on in his mind, where these are expressed in a private language that only he understands. He could not mean anything by his public words unless he had an additional private language referring to private mental entities. So the audience cannot really understand even his words for public things, since they cannot know his inner meaning-conferring mental states. Yet these words may nevertheless have a perfectly determinate meaning, generated by his private inner states. This paralyzed speaker dramatizes our own position with respect to meaning: we do give behavioral signs of our inner states, even if inadequate ones, but our essential predicament is that we make reference to public objects in virtue of being able to make reference to private objects—ones that need no public expression in order to be referred to. The fundamental nature of language is that reference to a public world is possible only because of reference to a private world, that is, reference to states of mind that at best have contingent behavioral expression. Thus, if in fact I am paralyzed and don’t know it, erroneously supposing that my states of mind can be known by others, this does not imply that I fail to refer to my mental states, or fail to refer to public objects. I can successfully refer to things even though no one else can know what I am referring to (again, pace

96

Part II

Wittgenstein). The unknowability of my mind is not a disqualification from being able to speak a meaningful language. This is what is not possible: I possess a language while being a complete zombie. Even if my external behavior matches that of a normal speaker, I do not mean anything by my words if I have no intentions and no other mental states. It may look as if I could refer to public objects in such a mindless condition but not to my private objects (since I have none), but that is wrong: I cannot mean anything by any words unless I possess the right psychological substratum, and I lack it in the zombie condition. The best that can be said is that I behave like someone who has a meaningful language—as I behave like someone who is in pain, without actually being in pain. In this kind of case there is no public language because there is no private language—no language for inner goings-on, there being none. Speaking a language requires meaning, and meaning requires a mind to mean—so zombies can’t mean. To say that a certain kind of language is possible is not to say that others can understand that language. Indeed, a private language for sensations just is one that is possible but not graspable by others. In the case of the paralyzed speaker, other people cannot understand his language (unless they rely on other methods to discover his mental states, such as examining his brain): knowledge of another’s meaning requires knowledge of his mind. This is true not just of the fragment of his language devoted to his own mental states, but also to the part ostensibly about public objects. For despite the fact that his utterances are observably prompted by public objects of certain kinds, it may be that his inner intentions are quite different from those assumed by observers. The observers may just guess wrong about his intentions, and these are constitutive of what he means by his words. And the same basic point applies to us nonparalyzed speakers: we may not understand each other, despite superficial appearances. Maybe you have an inverted spectrum relative to me, in which case I don’t understand your color language (I think you mean red by “red” but in fact you mean green). Maybe you are a complete zombie and mean nothing by anything. This is just a special case of the problem of other minds. I have said nothing to try to answer such skepticism; my point has concerned only the necessary logical conditions for public language to be possible. I have claimed (a) that private language is possible, (b) that public language is possible, and (c) that public language is possible only because private language is possible (but not vice versa). We can speak of what is outside us only because we can speak of what is inside us—reference to the outer requires reference to the inner. This is not because the outer is a version of the inner (as with idealism) but because meaning itself is inner, and we must refer to meaning in order to refer to anything. To understand a sentence is to know its meaning, but meaning consists in the inner states that accompany uttering the sentence—things like Gricean intentions, perceptions, and beliefs. So we must know and refer to inner states if we are to understand a sentence. And these inner states are

Language97

precisely things that (a) are known in a privileged way by their subject, (b) can exist without the benefit of any behavioral expression, and (c) need not be knowable by any outside observer. It is thus a condition of the possibility of public meaning that there be private meaning. No one could speak and understand a purely public language. This is fundamentally because the possibility of meaning depends upon the possibility of thought about meaning. Meaning is not something that could occur outside of the boundaries of awareness: to mean something we have to be conscious of what we mean. Meanings have to be known.

Deciding to Mean

Imagine a catalog of all the mental events, states, and attributes of which human beings are capable. Some of these will be things that we can decide to exemplify and some will not. We cannot decide to believe, know, understand, dream, or see (hear, etc.): these things are not “subject to the will.” But we can decide to imagine, think about, dwell upon, recall, calculate, or suppose—here the will is effective. Where does meaning fall with respect to this division? Can we decide to mean? In some senses we clearly can. I can decide to perform an act of speaker meaning, uttering the sentence “It’s close today” to mean that it is humid today. I decided to say something meaning that it’s humid and I went ahead and did it. I can also decide to mean someone by the nickname “Wily,” as I can decide to name my daughter “Bertha.” So meaning is subject to the will in some respects—“deciding to mean M by W” is perfectly meaningful. When this happens, a word comes to mean what I decide that it shall mean. I create a meaning—I bring a meaning into existence. This is a free choice on my part: words come to mean what I choose they shall mean. Nothing from the outside imposes the meaning on me: it is up to me what I mean. Words can have the meaning they have as a consequence of my free decision. But how far does this freedom reach? Can I decide that any two names I use for the same object shall have the same sense? That is, can I decide that Frege’s theory of sense and reference shall not apply to my use of names? I might regard the distinct senses of “Hesperus” and “Phosphorus” as a logical defect, so I choose to eliminate it from my logically perfect language: all the names I use will mean the same if they refer to the same object. This would be like choosing not to have modal words in my language because I don’t like modality on metaphysical grounds, or getting rid of vague expressions. Can I decide that my definite descriptions shall all conform to Russell’s theory? I simply stipulate that they do. It appears that I can indeed decide these things—my semantic will can create semantic facts. I can decide what my words and sentences mean, since this is a matter of stipulation. Meaning is malleable and voluntary, not imposed from without. Just as I can decide what to imagine, so I can decide what to mean. Meaning is ultimately a matter of human decision. But can I decide the nature of my meaning? Can I decide what theory will be true of my meaning? Here things become murkier. Consider truth conditions theories of meaning versus verification conditions theories: can I decide which of these theories shall be true of my meanings? Suppose I am attracted to metaphysical realism and relish the concept of truth; I therefore decide that my sentences will have their meaning given by truth conditions. It is not that I think they already conform to a truth conditions theory—I may be agnostic on the question or even skeptical (what with all the misplaced verificationism around these days). No, I stipulate that truth conditions constitute my meaning. I know what truth conditions are and I want my sentences to

100

Part II

express truth conditions; so I decide to mean truth conditions by my sentences, not verification conditions. What is to stop me from doing that? I freely assign truth conditions to my sentences as their meanings. It’s a free country, semantically speaking. I can mean what I choose to mean—and I choose to mean truth conditions. I might even announce outright: “The meaning of my utterances is to be understood in terms of truth conditions, not verification conditions.” But couldn’t someone of positivist or antirealist sympathies make the contrary decision? This person suspects that the sentences she has inherited from her elders are tainted with metaphysics, and she regards the concept of truth with suspicion; she wants her meaning to be determined entirely by verification conditions. She thus stipulates that her sentences are to be understood in terms of verification conditions, not truth conditions. When she says, “John is in pain” she means that the assertion conditions for that sentence are satisfied (John is groaning, writhing, etc.). She insists that there is no inaccessible private something lurking behind the behavioral evidence—no mysterious “truth condition”; there are just the criteria we use for making assertions of this type. She accordingly stipulates that her meanings shall conform to the verification conditions theory. This does not seem logically impossible: there could be a language conforming to the verificationist conception, given appropriate beliefs and intentions on the part of speakers. The traditional dispute has been over whether our actual language is subject to a truth conditions or a verification conditions theory, not over whether each theory is logically coherent. The truth conditions theorist thinks that objective conditions in the world are what our meanings express; the verification conditions theorist thinks that humanly accessible evidence is what our meanings express: both seem like logical possibilities, and the question is which of them is realized in our actual language. But even if one theory applies to actual language and not the other, that doesn’t rule out a possible language conforming to the other theory. Compare semantic tone: our actual language has semantic tone (“cur” versus “dog,” etc.), but there is no logical necessity that all languages must have semantic tone. We might choose to stipulate it away. Likewise, we could add tone to a hitherto toneless language, if we thought it a good idea. The idea of tone is not logically incoherent just because a given language lacks it. So why aren’t there possible languages that exemplify a truth-based semantics or an evidence-based semantics? Maybe there could be a transcendental argument that would rule out one of these as inherently incoherent, but it is hard to see what it could be. And even if there were such an argument, what is to stop someone from ignoring it and simply legislating what he is going to mean—even if it is somehow incoherent? Couldn’t a sufficiently determined speaker create a language that rested on concepts that are deeply problematic to the trained logical eye? What is to stop someone from indulging in an ultimately incoherent semantics—a logically imperfect language? After all, we can write incoherent sentences and believe

Language101

incoherent theories. Aren’t our own meanings arguably incoherent in places, what with the semantic paradoxes and all? And is it really incoherent to stipulate that the string “John is in pain” is to mean a set of suitable assertion conditions relating to behavior? Note that this is not the claim that that is what the English sentence “John is in pain” actually means; it is a purely stipulative claim about what a string of signs might mean. If speakers can in principle decide to mean what they want to mean, how did we come to mean what we do? Let’s suppose that our meanings are constituted by truth conditions, not verification conditions: why is that? I think the answer must be that we decided to make it that way. We believed in objective truth, facts, bivalence, realism, the unknowable, and so on, and we decided that these beliefs would fix our meanings: we let our metaphysics shape our semantics. We wanted to convey information about an objective world that exists independently of human minds and human evidence, so we fashioned our meanings accordingly. We could have had other predilections of a more positivist or idealist stripe, in which case we would have fashioned our meanings differently; but as it happens, we decided on the realist truth conditions option. That is what we decided to mean, given our beliefs, and nothing could prevent us making it semantic reality. So if a truth conditions theory is correct for our language, it is so because of our earlier semantic stipulations, now long entrenched. That theory is, in effect, a theory of what we decided meaning should be. If we had decided differently, then a different theory would apply. Maybe Martians did decide differently, being a resolutely positivist bunch, in which case a truth conditions theory would not be right for their language—for that, we would need a verification conditions theory. Maybe no such systematic theory would be possible for their language (there is no Martian Tarski of justification, defining “s is justified if and only if p”); but that is not a point about their meaning, merely about its susceptibility to systematic theoretical treatment. The correct theory of meaning for a language is the one that fits the speakers’ linguistic decisions (customs, predilections, prejudices, practices, etc.). Meaning does not descend on a group of people against their will, giving them no choice about what they mean; it results from their own decisions and wishes. Our actual language has a truth-based semantics (if it does) because that is how we set it up—if not by explicit conscious stipulation, then by implicit assumption and ingrained habit. We could have set it up differently. In light of these points, the usual way philosophers of language talk and think is misguided. They talk as if they are providing a theory of meaning that applies to meaning as such, that is, to all meaning, not to the semantic decisions of a particular linguistic group. Thus we hear of the “truth conditions theory of meaning,” the “verification conditions theory of meaning, the “use theory of meaning,” “possible worlds semantics,” “situation semantics,” and so on. These may (or may not) be good theories of the meaning of normal human languages as they now exist, but they are not theories

102

Part II

that can apply to all conceivable cases of meaning—since conceivable speakers may fashion their language with other agendas in mind. For example, consider possible worlds semantics: speakers may or may not be attracted to the underlying metaphysics of such a theory. One group may simply love the concept of possible worlds (they are natural-born David Lewis enthusiasts): they believe in them wholeheartedly, as objective constituents of reality, and they want to be construed as making constant reference to them in their modal talk (it may even be a kind of religion with them). They consciously decide that possible worlds are what they want to mean by their modal words, or maybe they have just always found it natural to think this way. Then possible worlds semantics will fit their language nicely—they will welcome such a semantic theory when explicitly formulated (“Yes, that’s what we always meant!”). But another group may absolutely hate the idea (they are natural Quine followers) and will have nothing to do with it. If you tell them that their modal words refer to possible worlds, they will spit in your face and assure you that that is certainly not what they mean. How can you persuade them that they really mean possible worlds when they violently reject the whole idea? They will insist that they can mean what they decide to mean, and they have decided not to mean that (they mean primitive modal operators or predicate modifiers or some such). Speakers cannot be forced to mean what they choose not to mean; they mean precisely what they mean to mean. The upshot is that theories of meaning are inherently language relative—or better, population relative. They are theories of what groups of people intend or stipulate or choose semantically, not theories of some supposed transcendent semantic reality that foists itself on speakers, whether they like it or not. Speakers in effect choose what theory of meaning will apply to their language by deciding what their words mean, according to their beliefs and preferences, explicit and implicit. There is no such thing as the “theory of meaning” tout court; there is only “theory of meaning for P,” where P is a population. And our language may not be a good model for all possible languages, given our parochial semantic preferences. Maybe whales and dolphins have other ideas about the world and about the purposes of communication, shaping their meanings to fit their specific viewpoint. Tarski-type realist truth theories might be quite alien to the meanings they have contrived and bestowed on their signals. Maybe their language is “feeling-theoretic”: the meanings of their utterances are given in terms of the emotions they produce—which are rich and varied in their case. They are across-the-board semantic emotivists (neither truth conditions nor verification conditions), imbuing their language with emotions that correspond to particular utterances. All they do is express emotions to one another, never claiming that some fact obtains or that they have evidence that such-and-such. A theory of meaning for them will be nothing like a truth-based or evidence-based semantics, but will correspond to their specific semantic decisions—to their linguistic “form of life.”

Language103

There is no one “theory of meaning”; there are many theories, corresponding to the many semantic decisions speakers might choose to make, human or other. To suppose otherwise is like thinking that a “theory of culture” need only take account of one culture, not the full range of actual and possible cultures. Maybe universals will be found, in cultures as well as meanings, but it may also be that there are fundamental differences in how meaning is constituted in different cases—from truth conditions to verification conditions to emotion conditions to other conditions undreamt of (one hears of some very exotic folk in certain remote galaxies). There is evidently enormous freedom in what speakers can in principle decide to mean, and theories of meaning must respect that versatility.

Truth, Verification, and Meaning

It is often supposed that there is such a thing as a verification conditions theory of meaning as opposed to a truth conditions theory of meaning for natural human languages. The former is viewed as an alternative to the latter, so that it makes sense to reject a truth conditions theory in favor of a verification conditions theory. I will contest this. Take a simple sentence like “The table has four legs”: the usual assumption is that the sentence has a truth condition (that the table has four legs) and also a set of verification conditions (its visually seeming to me that the table has four legs, someone telling me this, the general likelihood of four legs, and so on). One theory says the truth condition is the meaning; the other theory says the meaning is constituted by the verification conditions. Let’s focus on the sentence “It looks to me that the table has four legs”; and let’s say that this alone constitutes the meaning of the original sentence. The question then is what to make of the usual truth condition—what is it up to if not constituting the meaning of the sentence? What role does it play? There seem to be three possibilities: (1) the sentence has no truth condition; (2) it has a truth condition but this is not part of the sentence’s meaning; (3) the truth condition exists and is identical to the verification condition. Possibility (1) is very hard to accept: surely the sentence can be true or false, and surely there is such a state of affairs as there being a four-legged table; and what is evidence for, except the truth of sentences? The sentence must have a truth condition and evidence must be evidence for the obtaining of that condition—on pain of denying that we use language to speak of reality. The verification conditions theorist does not wish to deny this, only to deny that truth conditions are what fix meaning; the point is not some sort of nihilism about truth, to the effect that there is no such thing. So the natural view will be that (2) is the intended position: the truth condition exists but is outside of meaning (lots of facts about sentences are not part of their meaning). The trouble with this view is that it is very hard to see how truth conditions could fail to be part of meaning: for surely they are what is represented by the sentence— what it stands for. The words in the sentence refer, and what they refer to determines the truth condition. This is why a speaker will always know what the truth conditions are—his understanding of the sentence will yield this knowledge. But how could that be if truth conditions were extraneous to meaning? Could we really learn the meaning of a sentence and only subsequently discover the conditions that make it true? Could I know the meaning of “The table has four legs” and be quite unaware that this sentence is true if and only if the table has four legs? That seems massively implausible: if a speaker didn’t know this we would assume that he didn’t know what the sentence means. It’s not like failing to know what the sentence sounds like in a particular dialect, or the etymology of “leg,” or even the correct syntactic analysis of the sentence.

106

Part II

Indeed, if the speaker knows the verification conditions of the sentence, doesn’t he thereby know its truth conditions, since he knows what the evidence is evidence for? Looking to me as if the table has four legs is precisely evidence for it being true that the table has four legs; these two things cannot be separated. It thus appears that option (3) must be the intended position. We then hear murky talk about adopting an “epistemic notion of truth,” where truth somehow consists in verification. The truth of “The table has four legs” amounts to nothing more than it looking to me that the table has four legs. That is undoubtedly a very peculiar view, being quite blatantly idealist—what it is for a table to have four legs is identical to a sense experience of a certain sort! Moreover, why choose this verification condition rather than others as what the truth of the sentence is held to consist in? But these various conditions are not identical to each other, so how can any of them be identical to the table’s having four legs? Is the sentence multiply ambiguous? Intuitively, the truth condition stands apart from any of its many and various verification conditions; it does not reduce to them. The view that it does so reduce is radically revisionary, implying a metaphysics that was surely not part of the original intent. This is why the other two options are genuine options—because they allow a verification conditions theory of meaning not to collapse into total idealism. The problem is that those two options don’t work either. Only option (2) stands a chance of being acceptable, but it faces the decisive objection that truth conditions cannot plausibly be located outside of a sentence’s meaning. And there is a further problem: what should we say about the meaning of the evidential statements themselves? The sentence “It looks me that the table has four legs” has a meaning: is that meaning its truth condition or its verification condition? If it is the truth condition, then we have accepted a truth conditions theory for a subclass of sentences—and then there is the question of why we should abandon this theory for other sentences. But if it is a verification condition, we need to know (a) what this condition is and (b) what in turn is the correct account of its meaning. Is the answer to (a), “It seems to me (introspectively) to seem to me (visually) that the table has four legs”? If so, we can ask what the evidence might be for that—and so we generate an infinite regress. At some point the evidence has no further evidence to back it up. So the meaning of the ultimate evidential statement must be given in terms of itself. But isn’t it just straightforwardly true that it looks to me that the table has four legs? The evidential statement is itself true, and hence has truth conditions; so we must give a truth conditions theory of meaning for it. The answer to (b) must therefore be that a truth conditions theory is right for the basic evidential sentences. The basic sentences have truth conditions, and they provide evidence for the nonbasic sentences—whose meaning is constituted by verification conditions, not truth conditions. So we have a dual semantic theory—which again raises the question of why we don’t just adopt a truth conditions theory across the board. The only way out of this is to insist that the

Language107

verification conditions theory applies everywhere, including to basic sense-datum-like sentences. But that is very implausible, since those sentences just record truths about sensory seeming—there is no separate verification condition for them. Put simply, we have true evidential statements, and they provide evidence for other statements— that’s what verification is. We can’t have a verification conditions theory of the basic evidential statements: the meaning of “It looks to me that the table has four legs” is not given by some distinct statement that provides evidence for that statement. The meaning is contained in the fact that the sentence is true if and only if it looks to me that the table has four legs. Thus a verification conditions theory of meaning inevitably depends on a truth conditions theory at the level of evidential sentences. The meaning of an evidential sentence is not supplied by some sentence that is evidence for this evidence; it is supplied by the state of affairs that makes the sentence true. To this someone might reply that the basic sentence is evidence for itself. This is a funny way to talk, but it doesn’t get around the problem: the sentence is still true and hence has a truth condition— and, as argued above, this truth condition cannot be extraneous to meaning. The simple fact is that statements that record evidence are true if and only that evidence exists—they have truth conditions. The meaning of such a sentence must include its truth condition—for example, the condition that it looks to me as if the table has four legs. Thus verification conditions theories presuppose truth conditions theories. To put it differently, if we propose a theory of meaning in terms of criteria, we must reckon with the statements that record the criteria; but these statements don’t themselves have criterion-based meaning—they are just true statements about matters of fact. In the end, we will have statements about the basic criteria, and these will be true or false depending on the facts—that is, their meaning will consist in truth conditions. For example, if behavior provides the criteria for third-person psychological ascriptions, then reports of behavior will have a meaning given by truth conditions—unless we identify further criteria for behavioral statements, and then the same point will apply to them. And, of course, it is not generally supposed that statements about someone’s behavior are subject to a criterion-based theory of meaning; such statements are supposed to contrast with third-person psychological statements. This is really what verificationists have assumed all along: a two-tier system consisting of a basic level of true statements recording evidence and a superimposed level of statements verified by the first level. The meaning of the second level is given in terms of the first level, but the meaning of the first level is not given by an extra evidential level—it is given by the conditions of truth for the first level. No one has ever been a verification conditions theorist across the board; it is the “theoretical” sentences that have been subject to verificationist treatment. At most they have assumed that at the basic level, verification conditions and truth conditions coincide: but that is to accept that the meaning of such sentences consists in their truth conditions. All of this is just

108

Part II

to say that I have evidence for “The table has four legs” if the sentence “It looks to me that the table has four legs” is true. So a verification conditions theory presupposes a truth conditions theory for some sentences. It is not therefore a self-sufficient theory of meaning. Truth conditions theories, by contrast, have no such dependency: they are not committed to a verification conditions theory of some sentences. But then, why be a verification conditions theorist at all? Such a theorist is always a closet truth conditions theorist at the level of evidence. But if that is so, you may as well adopt a uniformly truth-conditional account of meaning—especially given the argument of the first part of this essay.

Meaning and Argument

In the course of a conversation, we often construct arguments designed to convince our interlocutor to form a particular belief. We are engaging in persuasion, and arguments have the power to persuade. If the audience is to be persuaded to adopt a certain belief, they will need to be offered reasons for that belief; otherwise no belief will be formed. Thus sentences will function as parts of arguments, as words and phrases function as parts of sentences: sentences are elements in larger linguistic structures—sequences of sentences in argumentative order. This suggests a thesis: the meaning of a sentence is its role in an argument. As a word can occur in many sentences, making its distinctive contribution, so a sentence can occur in many arguments, also making its distinctive contribution. We might even say that the function of a sentence is to feature in arguments, since that is what we do with sentences—we use them to persuade. If I am trying to convince you that Socrates is mortal, I may produce the sentences “All men are mortal” and “Socrates is a man,” thus deriving the sentence “Socrates is mortal”; and the purpose of those two premise sentences is precisely to get you to accept the conclusion sentence. The meaning of the premises is such as to make the conclusion true; the sentences are locked together in logical relations that reflect their meaning. Sentences are constituents of arguments, and that is an important fact about them; it is not merely incidental. The arguments can be deductive, inductive, or abductive, but they all contain sentences as their working parts—rather as sentences contain words as their working parts. We might dignify this observation with the phrase “argument-theoretic semantics” because it takes the argument as the primary unit of language use. People like to say that the utterance of a sentence counts as a “move in the language-game,” not the utterance of a mere word; but the argument is the overarching “game” in which that move is made. Persuasion is the point, and argument is the means. It isn’t truth or verification that is primary in semantics but validity, because validity is what arguments have (or lack). Philosophers of language have located the primary unit of meaning in the wrong place: not the sentence in isolation but the sentence in its argumentative role—as a part of something larger. As chess is the game within which chess moves make sense and gain point, so chains of argument are the “game” in which utterances of sentences make sense and gain point. Two objections may be made to this position. The first is that not all utterances are aimed at getting people to form beliefs: assertion may be, but what about command? Here there are two replies: (a) even command can be construed as inducing the belief in the audience that they should do such-and-such; and (b) we could allow for arguments that have actions as conclusions, so we can accommodate commands that way. The second objection is that not all utterances of sentences occur as parts of arguments, though admittedly many do—so what about the solitary assertion? Suppose I

110

Part II

just assert, “It’s raining”—have I not said something significant, even though there is no argument in the offing of which that utterance is a part? Similarly for an isolated utterance of “Shut the door!” Thus not all utterances of sentences take place in the context of an argument: yet the speaker is still trying to induce a belief, doing so by making a solitary assertion, not by constructing an argument. Many people indeed make a habit of unsupported assertion, so their sentences surely don’t get their meaning and point from a larger linguistic whole. Aren’t we overintellectualizing ordinary speech? Here is where things get interesting. For the natural question is how beliefs can be induced in an audience without producing anything in the way of an argument. Why doesn’t the hearer simply respond, “Why should I believe that p?” The hearer needs a reason to believe that p, so the mere unsupported assertion that p is unlikely to secure belief. If we think of it Grice’s way, the speaker has the intention to produce in the audience the belief that p; but by what means can he produce that belief? He can’t simply command, “Believe that p!” He needs to persuade the audience to adopt the belief in question; and for that he needs a suitable argument. What is that argument? The answer, I suggest, is that an assertion is an argument in enthymematic form: there are premises and a conclusion once the full import of the speech act is spelled out. The premises concern the making of an assertive utterance, while the conclusion concerns the belief the speaker intends to induce. Thus: “I have just asserted that p” (premise); therefore “p” (conclusion). The reason this argument works is that most speakers are reliable in their assertions, so their assertions give good grounds for believing what they assert. We can register this by adding a second premise: “The speaker is generally reliable in his assertions.” So the assertion is interpreted by the hearer as an argument, who goes on to form the belief that forms the conclusion of the argument. It is not a deductively valid argument, to be sure, but it does provide genuine reasons for belief, since testimony is a rational way to form beliefs. The point for current purposes is that the act of assertion is an implicit argument for a conclusion: it justifies that conclusion by basing it on the fact of assertion itself (plus general reliability). Accordingly, even the isolated assertion has the force of an argument, with the sentence asserted figuring as the conclusion of the argument. It is the same with imperatives. If I say, “Shut the door!” my speech act is an argument of the following general shape: “Given that I have commanded you to shut the door, you are well advised to shut it.” The listener comes to the conclusion that she should shut the door because I have just commanded it and I am in a position of authority. If these premises were not true, she would have no reason to shut the door— there would be no argument for shutting the door. Of course, she might shut it out of the goodness of her heart, but then the argument will record that fact as a premise—he has asked me to shut the door and I can see that he is too tired to do it himself; therefore I will shut the door.

Language111

On this conception, speech acts are enthymematic arguments, not different in kind from arguments consisting of explicit sequences of sentences. They are attempts to persuade by argumentative means. They are not merely hopes that the audience will form the belief in question; they give reasons for forming that belief. The hearer is therefore involved in reasoning when responding to acts of speech: she doesn’t just “grasp truth conditions” but reconstructs the argument offered by the speaker, evaluating it for cogency. The hearer is a logician (in the broad sense), employing norms of reasoning, assessing arguments. Thus, speaking and hearing are rational activities in the strong sense that they involve constructing, presenting, and evaluating arguments. Speaking is arguing, and hearing is deciding whether the argument is good. This is true even when the speaking involves a single simple sentence, as in asserting “It’s raining”: for even in this case the hearer’s reason for believing that it’s raining must advert to the performance of a speech act and the reliability of the speaker—what he is saying must be true because otherwise he wouldn’t say it, and he is a generally reliable source; therefore it’s raining. There is thus a close connection between meaning and logic: grasp of meaning embeds grasp of logical relations (deductive, inductive, and abductive). It is not just that sentences have logical form; they must also be seen in the context of logical arguments of which they are parts. To understand sentences we must have a mastery of inferences—of how one sentence can support another. Whenever we form a belief as a result of a speech act, we have performed an inference concerning the speaker. Also, there is something self-referential about speech acts: when I assert a sentence intending to induce a belief I make implicit reference to my asserting it—since I offer the act of assertion as a premise in a piece of reasoning. I am in effect saying that the reason the hearer should believe that p is that I am now asserting that p. I am intending that you take my act of assertion as your reason for believing that p. My assertion is functioning as your premise. Are there any counterexamples to this theory? Are there any speech acts that cannot be viewed as arguments? This will involve some complicated questions about interrogatives and other grammatical moods, which I won’t go into here,1 but I think the most natural candidates are very simple utterances like “Boo,” “Hurrah,” and “Ouch.” My suggestion is that these are intended as arguments for holding beliefs about the speaker’s beliefs and other attitudes: the speaker wants the hearer to infer, on the basis of his speech act, what his state of mind is. For example, by uttering “Boo” he wants the audience to believe that he disapproves of what he is seeing or hearing—and the audience will reason correctly that he does, from the premise that he wouldn’t say “Boo” unless he disapproved of something in the vicinity. They could of course reject the argument 1.  I mean questions like whether questions are really commands—“Tell me whether p!” Or are they assertions of the speaker’s ignorance? How should we view the optative mood?

112

Part II

if they thought the speaker was being insincere—which shows that they are reasoning to a certain conclusion. In the case of “Ouch” the intention is to get the audience to believe one is in pain, by means of their inference that one wouldn’t normally say that unless one were in pain. So the speaker is offering the audience an argument that has a reference to uttering “Ouch” in a premise, the conclusion of which is that he is in pain. That is, he is trying to get the audience to form a belief about himself by arguing from the occurrence of a certain speech act. Communicating is arguing and sentences are the cogs of arguments. A sentence is a move in a “game” of argument. Persuasion is the purpose and argument is the means. How else could we persuade? This means that speaker and hearer function always as rational inferential beings. To be master of a language is to be adept in arguing with that language.

III Knowledge

Knowledge and Truth

A Priori Knowledge and A Priori Truth Tradition insists that knowledge falls into two broad classes: a priori knowledge and a posteriori knowledge. These categories are conceived as exclusive and exhaustive: no piece of knowledge is both a priori and a posteriori, and any piece of knowledge is one or the other. Without going into detail, we can characterize a posteriori knowledge as knowledge acquired by means of the senses (“by experience”) and a priori knowledge as knowledge not so acquired, but rather acquired “by reason alone” or “intuitively.” Then the traditional doctrine is that all knowledge is either acquired by means of the senses or acquired by pure reason, but not both. There is an exclusive and exhaustive partition of knowledge into these broad categories, with knowledge of logic, mathematics, analytic truths, philosophy, and ethics in one category and knowledge of physics, history, geography, psychology, and anthropology in the other category. But in addition to this epistemic partition, the tradition also recognizes a partition into two great classes of truths or propositions or facts: the a posteriori truths and the a priori truths. Thus we regularly find it said, for example, that the proposition that Hesperus is Hesperus is an a priori truth while the proposition that Hesperus is Phosphorus is an a posteriori truth, or that “2 + 2 = 4” expresses an a priori truth while “Water arrived on Earth from meteors” expresses an a posteriori truth. That is, we find both pieces of knowledge and the truths known classified by means of the terms in question. On the face of it, this is peculiar. Why do we apply the terms so promiscuously, to both knowledge and what is known? The answer, of course, is that we take there to be a connection between the two uses—in fact, a definitional connection. The simplest way to state the connection is as follows: an a priori truth is one that is known a priori, and an a posteriori truth is one that is known a posteriori. We define the partition of truths by means of the partition of knowledge: each type of truth has its characteristic epistemology—the way it comes to be known. Mathematical truths are a priori because they are known a priori, while truths of physics are a posteriori because they are known a posteriori. In other terminology, truths of reason (“intuitive truths”) are those known by the exercise of reason alone, while empirical truths are those that are known empirically, that is, by means of the senses. Thus we define a type of truth by reference to a type of knowledge. Of course, we must make allowance for the fact that not all truths are known: we can’t say that an a priori truth is one that is known a priori, or else undecided (but true) mathematical propositions would fail to count as a priori truths. So instead we should say: an a priori truth is one that is known a priori when it is known, and similarly for a posteriori truth. It may not even be knowable (say, an undecidable mathematical proposition), but were it to be known it would be known a priori. It is of a kind that is

116

Part III

known a priori, when known at all. In this way we can allow for a priori truths before the existence of people capable of knowing them. We do not define a priori knowledge as knowledge of an a priori truth, and then try to say what an a priori truth is; we say what a priori knowledge is first, and then define a priori truth in those terms. The primary notion is epistemological, but we extend it to propositions in the natural way just sketched. The simple definition needs to be amended further: we are not just saying that a priori truths happen to be known a priori (similarly for a posteriori truths) but that they must be so known. We do not want to allow that at some point in the future we might start to know a priori truths a posteriori or that intelligent aliens know all a priori truths a posteriori; the idea is rather that a priori truths are necessarily known a priori. So the definition should be formulated as follows: an a priori truth is one that must be known a priori (when it is known at all) and an a posteriori truth is one that must be known a posteriori (when it is known at all). With this we have a bipartite division of the class of truths, which is exclusive and exhaustive. It will be useful to compare this case with two others: introspective knowledge versus perceptual knowledge and innate knowledge versus acquired knowledge. These are epistemological distinctions, but we can extend them into distinctions between truths: introspective truths versus perceptual truths and innate truths versus acquired truths. Thus, an introspective truth is one that must be known introspectively (similarly for perceptual truths) and an innate truth is one that must be known innately. For example, the truth that I am in pain is an introspective truth in virtue of being known by me introspectively, while the truth that everything is identical to itself is an innate truth in virtue of being known by me innately (assuming I have innate knowledge of the law of identity). We impose a classification on truths by linking those truths with types of knowledge. Truths are classified in terms of how they are characteristically or essentially known. This seems like a perfectly coherent and defensible procedure, one that has tradition behind it. Counterexamples The foregoing might appear straightforward enough, scarcely worth articulating: but in fact it is not. I will begin my discussion of putative counterexamples by recalling some remarks of Saul Kripke: I will say that some philosophers somehow change the modality in this characterization from can to must [he has just been considering the definition of an a priori truth as one that “can be known independently of experience”]. They think that if something belongs to the realm of a priori knowledge, it couldn’t possibly be known empirically. This is just a mistake. Something may belong in the realm of such statements that can be known a priori but still may be known by particular people on the basis of experience. To give a really common sense example: anyone

Knowledge117

who has worked with a computing machine knows that the computing machine may give an answer to whether such and such a number is prime. No one has calculated or proved that the number is prime; but the machine has given the answer: this number is prime. We, then, if we believe that the number is prime, believe it on the basis of our knowledge of the laws of physics, the construction of the machine, and so on. We therefore do not believe this on the basis of purely a priori evidence. We believe it (if anything is a posteriori at all) on the basis of a posteriori evidence. Nevertheless, maybe this could be known a priori by someone who made the requisite calculations. So ‘can be known a priori’ doesn’t mean ‘must be known a priori’. (35)1

Kripke’s counterexample is clear enough: it shows that it is too strong to define an a priori truth as one that must be known a priori, since such a truth can be known both a priori and a posteriori. It is possible to know mathematical truths on the basis of empirical evidence. The counterexample generalizes to all cases involving knowledge by testimony (of which Kripke’s example is a special case). The computer is a reliable source of mathematical information, but so is a qualified mathematician. If a mathematician had told me that the number in question is prime, I would have been justified in believing him, and could acquire the corresponding knowledge. My evidence is just the empirical evidence of general reliability and listening to what the mathematical authority tells me in this case—that is what I would cite as my warrant. Clearly, I could acquire a great many beliefs this way, about all manner of mathematical and logical matters, and hence have a posteriori knowledge of a priori truths. But then we can’t define an a posteriori truth as one that is known a posteriori, or else mathematical truths will come out a posteriori (relative to a particular believer) as well as a priori (relative to another believer)—we lose exclusivity. It is apparently not correct (“just a mistake”) to say that an a priori truth is one that must be known a priori. Another kind of example concerns knowledge of analytic truths. Suppose we have a computer with access to a dictionary of synonyms. When I ask it whether some pairing of words produces an analytic sentence, it can respond more quickly than the human brain. I might acquire knowledge of synonymies by consulting this machine, thus knowing a posteriori what I could have figured out a priori if I hadn’t been so slow and lazy. I might follow the same procedure with a human lexicographer. Granted that analytic truths are a priori, I would thus acquire a posteriori knowledge of a priori analytic truths. We might now be tempted to conclude that anything a priori could in some circumstances be known by someone a posteriori—just by replacing direct reflection by testimony. If ethical truths are a priori, for example, we can imagine someone coming to know them by reliance on moral authority, and hence a posteriori. 1.  All quotes in this essay are from Saul Kripke, Naming and Necessity (Cambridge, MA: Harvard University Press, 1980).

118

Part III

Does this ruin the partition of truths into a priori and a posteriori? Not quite yet, because we can define a posteriori truths in a different way: instead of saying an a posteriori truth is one that can be known a posteriori (which would ruin the partition), we can stipulate that an a posteriori truth is one that must be known a posteriori. Then we can maintain exclusiveness, since no proposition that is a posteriori will be knowable a priori. Admittedly, there is an odd modal asymmetry in the definitions, but we don’t have to accept that some propositions can be both a priori and a posteriori. But is it plausible that typical a posteriori propositions must be known a posteriori? Kripke gives the beginnings of a negative answer in a footnote: “Quite a different question [from the question of metaphysical possibility] is the epistemological question, how any particular person knows that I gave these lectures today. I suppose that in that case he does know this a posteriori. But, if someone were born with an innate belief that I was to give these lectures today, who knows? Right now, anyway, let’s suppose that people know this a posteriori” (40). Now Kripke does not come right out and say it, but he is clearly thinking that the a posteriori truth that he gave the lectures today might be known by someone innately, and hence a priori, that is, not by means of the senses. Indeed, any proposition that is currently known a posteriori might be known innately, and hence a priori. So it is not necessary for a posteriori truths to be known a posteriori. It is logically possible to know such truths independently of sense experience. Again, this kind of point can be expanded and generalized: what if God implants beliefs in us that he thinks will be helpful, and so spares us the trouble of empirical investigation? He could do the implanting at birth or at some point during adult life when the piece of knowledge would come in handy. Would we not then know things about the material world—empirical facts—prior to, and independently of, sense experience? We would know them by divine intervention, not by means of our senses. And what about God’s own knowledge of material reality? It is surely not arrived at by sensory observation; so God’s knowledge of the universe is presumably a priori. That does not appear conceptually incoherent; so it is conceivable that what we would regard as a posteriori truths should be known a priori. Pushing this point to its logical limit, can we not imagine wholesale epistemic inversion? That is, can we not imagine a creature that inverts the human ways of knowing the truths that are there to be known? This creature has all its logical, mathematical, and moral knowledge based on testimony (and hence a posteriori), and also all its physical and historical knowledge based on innate beliefs or divine intervention (and hence a priori). Both ways of knowing truths are logically contingent, so we can pull them apart from particular propositions, thus producing complete epistemic inversion. And this means that there is no chance of defining a particular class of truths by reference to the two ways of knowing: the concepts of an a priori truth or an a posteriori truth appear to crumble before us. We still have the notions of a priori and a posteriori knowledge, but there is no hope of

Knowledge119

converting that distinction into a determinate partition of truths. Every truth can in principle be known in either way. The case looks analogous to the distinction between innate and acquired: there will be far too many propositions that can in principle be both, so the prospects for defining a class of innate truths and a class of acquired truths look dim (except relative to a creature). Indeed, in this case the very idea of exclusive classes of truths appears to involve a category mistake: truths cannot be intrinsically innate or acquired, only knowledge can be. We cannot project the attributes of being innate or acquired onto the truths known, since there is no one way they must be known—you can be born knowing them or you can acquire knowledge of them later. The identity of the proposition does not entail one way of knowing rather than the other. You cannot peer into the proposition and read off whether it will be innately known or known by learning. So talk of innate truth or learned truth serves no useful purpose and yields no exclusive and exhaustive classification of truths. It looks, in fact, like an oxymoron (compare “innate facts” and “acquired facts”). By contrast, the distinction between analytic and synthetic truths is not vulnerable to the kind of counterexamples we have seen for the other two epistemic distinctions. For anyone who believes in the analytic–synthetic distinction at all, these two concepts partition the class of truths exclusively and exhaustively; and they ascribe intrinsic properties to the propositions to which they apply. It is not that the same proposition can be both analytic and synthetic, or that a proposition is analytic if it can be known by analysis. We are not trying to generate a distinction between truths from a distinction between types of knowledge (“analytic knowledge” and “synthetic knowledge”). If there is any epistemic distinction here, it arises from the nature of the truths themselves, not the other way around. But the distinction between a priori truths and a posteriori truths looks, in the light of the counterexamples, like the distinction between innate truths and acquired truths—a confused way of speaking that fails to identify a significant partition between two kinds of truth. To be sure, there are two kinds of knowledge, but they don’t correlate with two kinds of truths—any more than knowledge acquired in the kitchen as opposed to the living room generates a distinct class of truths. You can come to know exactly the same truth in these two ways, say by reading a book in the two rooms—as, apparently, the same truths can be known both in the a priori way and in the a posteriori way. So, at least, it appears from the considerations adduced so far. Possible Responses If we take the counterexamples at face value, they appear to indicate a profound skepticism about the ideas of a priori and a posteriori truth. The legitimate distinction between a priori and a posteriori knowledge has been illicitly projected onto truths, producing a category mistake. The distinction is an epistemic distinction, not a metaphysical

120

Part III

distinction between types of truth or fact—or a semantic distinction between types of sentence or proposition. It is like calling a proposition justified, when all that can be meant is that someone’s belief in it may be justified: the same proposition can be justified for one person but not for another—the proposition is not somehow justified in itself (though it may be true or false). Justification is a property of beliefs not propositions or facts; there are not two kinds of truths, the justified ones and the unjustified ones. Similarly, all that can be meant by calling a truth a priori or a posteriori is a purely relational statement, to the effect “p is a priori for x,” that is, “x knows p a priori.” This allows that p may well not be a priori for y (and similarly for a posteriori). Thus, in the interests of clarity, we should stop speaking of truths in this way and restrict ourselves to speaking only of a priori and a posteriori knowledge (or evidence or reasons or justifications). We should not suppose that the world divides up into an a priori part and an a posteriori part. If the world is the totality of facts (or true propositions), then there is no partition of these facts into the a priori ones and the a posteriori ones. That is a confusion of metaphysics with epistemology. This is not, of course, to deny that there is a distinction between, say, mathematical truths and truths of physics; it is just to say that we cannot hope to capture that distinction by speaking of mathematics as a priori and physics as a posteriori. For some possible species, that classification could be inverted, because it depends on the contingent way the truths in question get to be known. So we should abolish all talk that presupposes that truths or facts themselves fall into one category rather than the other. Thus we can no longer speak of “empirical facts,” unless we mean simply “fact that is empirical for knower x.” There are just facts—that may be physical or historical or mathematical—but there is no general class of empirical facts (as there are no “rational facts,” i.e., truths of reason). When we say, loosely, that “Hesperus = Hesperus” and “Hesperus = Phosphorous” express a priori and a posteriori truths, respectively, we are strictly speaking mistaken (or at least misleading), since the former proposition could be known a posteriori (by testimony) and the latter proposition could be known a priori (innately or by divine intervention). So we need to reform our casual philosophical talk about these matters quite dramatically: we need to stop treating relative traits as if they were absolute. It is as if we had got into the habit of speaking of “visual facts,” forgetting that what is sensed visually by one person may not be sensed visually by another. The concept of vision applies properly to a particular sense, not to the objects of that sense—as if being square, say, were somehow intrinsically visual. The world does not divide up into visual objects and nonvisual objects: we just happen to sense objects in one of these ways rather than another. In the land of the blind, there are no visual objects, though the objects are the same as those seen by the sighted. We must not project our subjective categories onto nature. It is simply anthropocentric to suppose that physical facts are inherently empirical and mathematical facts are inherently

Knowledge121

rational. So the counterexamples appear to suggest, backed by a general separation of epistemology and metaphysics. The position just sketched, which deserves to be called radical, raises the question of whether the class of all truths falls into two great categories that correlate with the old classification into a priori and a posteriori truths. Were we perhaps mistaking a genuine distinction for the misguided one currently being rejected? Here two positions are possible: (a) there is no bipartite division corresponding to the old discarded one, or (b) there is such a division but it is entirely nonepistemic. The first position acknowledges that truths divide into groups—mathematical, logical, physical, historical, and so on—but denies that there is any meaningful division into a pair of classes. There are many kinds of truths, which all belong to one overarching class, but there is no way to divide them into two classes that line up with the old division into a priori and a posteriori. There is no interesting metaphysical dualism of truths. The duality of ways of knowing finds no counterpart at the level of metaphysics. The second position is less radical, suggesting that we can find a metaphysical distinction to replace the old epistemic distinction. The general thought here is that some bodies of truths deal with abstract, timeless, unchanging realities, while some deal with material, temporal, and changing realities. Plato’s ontological dualism of universals and particulars is the model for such a view: truths about the former differ from truths about the latter, in view of their distinctive types of subject matter. This (it may be said) is the metaphysical distinction that lies behind the misguided epistemological distinction of a priori and a posteriori truths. For instance, “2 + 2 = 4” refers to abstract entities that exist outside space and time and never change, while “Water came to Earth on meteors” deals with spatiotemporal particulars involving events and transitions. Here we have said nothing about ways of knowing, but the division maps onto the old epistemic division in obvious ways. So perhaps we can generalize and find a nonepistemic distinction that corresponds extensionally with the old epistemic distinction—for brevity, let’s call it the “abstract– concrete” distinction. Then reality does divide neatly into two big classes, which the tradition wrongly characterized in epistemic terms. Our instincts were on the right track, but we misconstrued an ontological distinction as an epistemological distinction. Instead of using the epistemic locutions “it is an a priori truth that” and “it is an a posteriori truth that” we should use locutions like “it is abstractly true that” and “it is concretely true that”; that way we capture the underlying metaphysical distinction accurately. We may go on to say that abstract truths can in principle be known either a priori or a posteriori, and concrete truths likewise. It is strange that the philosophical tradition was so fundamentally confused about all this, conflating metaphysics and epistemology so crassly; but now that we have clarified matters we can proceed with the requisite conceptual clarity. When speaking of knowledge we can continue to use

122

Part III

the concepts of the a priori and a posteriori, but when we speak of truths we must use the concepts of the abstract and the concrete (or some such). A Better View Perhaps the reader will sense that I am not entirely happy with the position just outlined. Granted, the position is clear enough, and apparently well motivated, but it seems just a bit too radical, a bit too dismissive. Was the tradition really so misguided? Is there really nothing to the idea of a distinction between a priori and a posteriori truth, taken to mark an inherent difference in the types of truth? But weren’t the counterexamples clear and convincing? Or did they perhaps move a little too quickly? Let us reexamine them to see if they might be faulty in some way. Kripke’s computer case invites an obvious response: isn’t this rather an indirect way to arrive at knowledge of a mathematical truth? Consider testimony knowledge generally: I can learn something from an expert, but that person did not learn it from someone else—at some point, someone has to know it not by testimony but “directly.” Someone in the chain of testimony has to know the truth in question basically—not by deriving his or her knowledge from someone else. Thus we need to distinguish basic knowledge from derived (by testimony) knowledge. Then what is to prevent us from stipulating that an a priori truth is one that must be known a priori in the basic sense? That is, basic or direct knowledge of the truth in question must always be a priori—whatever might be said of derivative or testimony knowledge. The a priori way of knowing a mathematical truth has priority, since it must ground any chain of testimony, and it is clearly correlated with the distinctive subject matter of the proposition. So it is perfectly well-motivated to insist that it is the type of knowledge suited to defining a priori truth: an a priori truth is one that must be known a priori, in the sense proper to the truth in question—the one that reflects the subject matter directly. And it is notable that there is no way of knowing a mathematical truth a posteriori that is not by testimony (or no uncontroversial way). So we are within our rights to exclude knowledge by testimony in our definition of a priori truth. Let us say, summarizing, that an a priori truth is one that is canonically known a priori—meaning originally known, basically known, properly known. Then we can say that an a priori truth is one the canonical knowledge of which must be a priori. That this is not an ad hoc maneuver can be seen from the fact that the same distinction applies to a posteriori truth: we can and should distinguish between canonical knowledge of a simple observational fact and noncanonical knowledge of that fact—that is, observing the fact by means of the senses and hearing about it from someone else. The latter depends essentially on the former, which is the basic or direct way of knowing the fact. It would not be plausible to claim that an object’s being red is not an empirical fact because it can be known by testimony, and testimony provides a priori knowledge (where this latter claim is justified by arguing that testimony involves access to abstract

Knowledge123

propositions and so is not really based on the evidence of the senses). We need this kind of distinction across the board, and we can invoke it to rule out the kinds of counterexamples suggested by Kripke’s computer case. We have an inclination to think that knowledge by testimony is an inferior mode of knowledge, if it is really knowledge at all. It is very easy to be a skeptic about testimony knowledge (the expert might be trying to mislead you or be incompetent, and so on). You are taking it “on trust,” putting your “faith” in your informant, “taking his word for it.” This is a rather shabby second-rate way to know, compared to the way the expert himself knows. Consider coming to know that necessarily 2 + 2 = 4: you might arrive at knowledge of this modal proposition by testimony, but surely that is inferior to seeing the truth of it for yourself. Just conceivably, you might be so lazy and inept that you cannot figure out for yourself that it is analytic that bachelors are unmarried males, relying instead on a linguistic expert to assure you that this is so; but your knowledge in that case is defective and second-rate, compared to someone who takes the trouble to think it through for herself. We are surely entitled to frame our definition of a priori truth to rule this kind of case out: the truth that bachelors are unmarried men is canonically known by a priori reflection on the concepts, not by asking someone else to tell you whether the sentence in question is true or not. Also, are there not a priori truths that cannot be known a posteriori, contrary to the claim of total epistemic inversion? Is it really possible to know that Hesperus is Hesperus purely a posteriori? What would it be to implore someone to inform you whether Hesperus is identical to Hesperus? Isn’t this something you know just by understanding the name “Hesperus”? Can you come to know by testimony alone that nothing can be both red and not red? How could you be genuinely informed that you exist? These are all things we know prior to sense experience: could we really know them in any other way? If someone were to claim, as with Kripke’s computer case, that it is possible to know such truths by testimony, you would be inclined to ask whether such a person really understood the question. How could I know what “I exist” means and yet have no idea whether it is true? Upon being told that I do exist, can I really respond: “Thank you for telling me; I had my doubts for a moment”? These kinds of propositions are presupposed in our encounters with the empirical world, not derived from such encounters; and it is hard to see how they could have any other epistemic status. If so, not everything now known a priori could be known a posteriori. But what should we say about a posteriori truths and epistemic inversion? Is it not plausible that anything known by means of the senses could have been known innately? Can’t beliefs arrived at by experience also be arrived at by genetic engineering? Can’t God decide to implant any belief in me that he likes, including those beliefs that I usually form by use of my senses? It certainly seems correct to say that whatever is believed empirically could also be believed innately or by divine intervention, so that beliefs do not logically require causation by sense experience. But it does not follow

124

Part III

that knowledge obeys that principle—because knowledge requires justification. I could clearly have an unjustified innate belief, but it would not be knowledge. If I have an empirically justified belief that it’s raining outside, which counts as knowledge, we do not preserve its epistemic status by supposing the same belief to be innate and unjustified. A posteriori knowledge requires a posteriori justification, as with citing a sense experience, but if we remove the justification we remove the knowledge. Thus innate beliefs are not eo ipso innate knowledge, and so we do not yet have a case in which an a posteriori truth is known but not by experience. Nor is it at all plausible to suppose that someone born with such a belief could contrive some sort of a priori justification for it—since no such justification exists or is possible. There is no such thing as an a priori justification for the proposition that water came to Earth on meteors. So someone born with that belief has no justification for it and accordingly does not know. The case is completely different from being born with an innate belief in a mathematical truth: here it is possible to provide a justification without venturing into the world of the senses. The point is that a posteriori propositions demand a posteriori justifications, that is, the evidence of the senses. They cannot then be known in the absence of sense experience, merely by being installed at birth. An objection may be raised: couldn’t the person with inborn empirical beliefs provide a justification of another kind, namely that inborn beliefs are likely to be true— either because of evolutionary fitness or divine guarantee? Thus, I could know that meteors brought water to Earth by reasoning as follows: “I wouldn’t be born believing this unless it were true, because evolution or God would not implant it in me if it were false; so this justifies my confidence in its truth, and hence I know.” Now whatever may be thought about this kind of argument, as a source of epistemic warrant, the reply to it is that this is really just another sort of a posteriori reasoning, and so does not provide an example of an empirical truth known a priori. My ground for belief in the proposition is just that I believe on empirical grounds that my innate beliefs are likely to be true. What we need is a case where I have an innate empirical belief that counts as knowledge but which has no empirical justification. No such case has yet been provided, as opposed to describing a case of someone just finding himself with certain beliefs, either innately or divinely caused. Indeed, there is something troubling even about cases so described: for what am I to think of these beliefs that I find in myself? As they are my beliefs, I take them to be true—but what is my reason for taking them to be true? I have none, by stipulation—I just have the bare belief. I have neither an a priori nor an a posteriori justification, and yet I am supposed to hold a belief—am I not then completely irrational? But I am not irrational and will want to cast around for a justification for the belief I find implanted in me; and finding none, I will promptly jettison the belief in question. The belief will not be stable in the circumstances described. We have not provided a case in which someone is born with stable a priori knowledge of an empirical truth, only a

Knowledge125

case of unstable unwarranted quasi belief. Thus we are within our rights to insist that an a posteriori truth is one that must be known a posteriori, that is, by means of sense experience, since we have not been given a case of a priori knowledge of an a posteriori truth. No counterexample to that definition has yet been produced, despite initial impressions. For all that has been said, therefore, we have no reason to revise our initial strong formulation: an a priori truth is one that must be known (canonically) a priori, and an a posteriori truth is one that must be known (properly known) a posteriori. We can have noncanonical a posteriori knowledge of an a priori truth and we can have unjustified nonsensory belief in an a posteriori truth; but neither of these points counts against the definitions when correctly understood. In consequence, we can cleave to the traditional insistence on two broad classes of truth, a priori and a posteriori. We need not accept the radical revisionism contemplated earlier. What is the picture suggested by these reflections? How do truths and knowledge link up? The short answer is: subtly. An a priori truth imposes constraints on how it can be known, and similarly for a posteriori truths. The proposition dictates the appropriate way of knowing it. There are de jure ways of knowing, not just de facto ways. An a priori proposition can be known de facto by testimony (hence empirically), but the basic and proper way of knowing it—the de jure way—is by a priori means. That way reflects the inner constitution of the proposition—the nature of its subject matter. To put it briefly: the knower sees that the a priori proposition could not be false, and hence must be true; no recourse to sense experience is then required to certify it as true. And similarly for a posteriori truth: the proposition itself lays down rules for how it is to be known—it must be known on the basis of experience, if it is to be properly known. The knower can see that the proposition is not self-certifying, and therefore warrant for believing it must come from outside—from the senses. To put it crudely, the right way to know truths that concern the “abstract” is a priori, and the right way to know truths about the “concrete” is a posteriori. The metaphysics constrains the epistemology: the epistemology must follow the grain of the metaphysics. The two are not independent and free floating but intertwined. It is in the essence of the proposition that it must be (canonically) known in one way rather than the other; this is not something superimposed from the outside. We could say: an a priori truth is one such that in its nature it requires to be known a priori (similarly for a posteriori truth). We might go against that nature, coming to know it empirically, but it has that nature anyway. If we could come to know an a posteriori truth a priori (which we have seen to be doubtful), that too would go against its intrinsic nature—how it is meant to be known. We might say that the proposition specifies how it is ideally known—the kind of knowing it prefers. I would even venture to assert that the facts of the world have this kind of epistemological preference built into them: the fact that 2 + 2 = 4 itself contains the proper mode of knowing that fact (by

126

Part III

reason); the fact that water came to Earth on meteors itself contains the proper mode of knowing that fact (by sensory observation). That is precisely why the tradition insisted as it did—because the notion of an a priori truth (or fact) was recognized to be well defined and essential (and the same for a posteriori truth). Some facts are intrinsically a priori and some are intrinsically a posteriori; and that is why they are characteristically known in one way rather than the other. They are apt for one way of knowing or the other. It is not that we just happen to know them in one way rather than the other, as I just happen to know there is a cat next to me visually rather than by means of touch or hearing. The metaphysics and the epistemology interlock. Thus, our cognitive faculties fit the world in its broadest outlines. The world is the totality of two kinds of facts, and we have two kinds of cognitive faculty geared to these kinds of facts. That is fortunate, or else we would miss out on vast tracts of reality. The realist would say that this metaphysical division of facts predates the existence of our faculties, and that our faculties exist to provide insight into these mind-independent facts. The idealist would say that the faculties came first and that the metaphysical division is a reflection (a projection) of the operations of the faculties. But both agree that facts and faculties line up: the a priori facts are known through the a priori faculty, and the a posteriori facts are known through the a posteriori faculty. There is no fundamental crosscutting, as epistemic inversion suggested—as if reality were strictly neutral about how it should be known. The ontological dualism feeds into the epistemological dualism. As a realist, it strikes me as remarkable that mind and world line up in this orderly way: the very idea of a priori or a posteriori truth, which to a realist looks like an oxymoron, turns out to be the sober truth of the matter. Truths have ways of knowing inscribed in them! From the point of view of divine epistemology, it is hard to make sense of the distinction between a priori and a posteriori knowledge, and therefore hard to make sense of a distinction between a priori and a posteriori truths or facts. But from the human point of view, the structure of our minds seems written into reality (or the structure of reality seems written into our minds). The world divides into two compartments and our mind divides into two compartments, and the compartments march in parallel: the a priori compartment of the world maps onto the a priori compartment of our mind, and the a posteriori compartment of the world maps onto the a posteriori compartment of our mind. In other words, a priori truths are known a priori and a posteriori truths are known a posteriori—and ne’er the twain shall meet (except in noncanonical corners of the universe). Perhaps I can shed light on why this seems remarkable to me by reverting to the two analogies I mentioned earlier: innate beliefs versus acquired beliefs and introspective knowledge versus perceptual knowledge. In those cases there is no traditional insistence on a parallelism between truths and knowledge (or belief): we don’t speak freely of innate and acquired truths or introspective and perceptual truths. Here we recognize that the epistemic classifications have no deep ontological significance. The same

Knowledge127

belief can in principle be innate or acquired (at least in some cases), so there is no sense in the notion of an intrinsically innate truth versus an intrinsically acquired truth— truths are truths, whether belief in them is innate or acquired. The world does not objectively divide up into truths that are believed innately and truths that are believed by acquisition; these are completely subject-relative distinctions. Similarly, the same proposition that can be known introspectively by one person can be known perceptually by another—as when I know I am in pain by introspection and you know I am in pain by looking at me. It would be strange to proclaim a special class of “introspective truths,” as if the first-person epistemology were written uniquely into the fact—as if such truths could be known in no other way, say by observing behavior or examining the brain. Here the same fact can be known in two different ways, period. But in the case of the a priori and a posteriori the link between metaphysics and epistemology is much closer, much more intimate: here we are entitled to elevate the epistemological into the metaphysical, by speaking outright of a priori truths and a posteriori truths. Or perhaps it would be better to say that we read the metaphysical into the epistemological: a priori knowledge counts as such precisely because it is knowledge of an a priori truth—this being just what we call knowledge that has that kind of subject matter. That would make the metaphysics conceptually prior to the epistemology, not vice versa. At any rate, the two are enmeshed and inextricable: the truths and our knowledge of them are inseparable—despite some suggestions to the contrary. In the book of the world there are two kinds of truths written, a priori truths and a posteriori truths. We know these two kinds of truths in two very different ways, a priori and a posteriori. The way we read the book thus coincides with the way it is written. We view the world through the lens provided by our epistemic faculties, but the world invites us to view it that way. Ways of being and ways of knowing dovetail. Truth and knowledge of truth converge.

Proof of an External World

Suppose that I believe in the existence of butterflies even though I have never seen or otherwise experienced one; nor have I ever had their existence corroborated by an authority. I believe in them merely on the basis of unsubstantiated rumor—I have no evidence of their existence. You can call me irrational and I won’t put up much of a fight. Then one day a pair of butterflies fly into my garden, large as life. I capture them and arrange to show them to you, a butterfly doubter. As I do so, I say, “See, I told you butterflies exist!” Have I not proved that butterflies exist? I have done so by presenting you with an actual example of butterflies—what more could you possibly want? I used to have no evidence to back up my belief, but now I have irrefutable evidence—I have a pair of butterflies right in front of me. Suppose now that I believe in an external world of material objects even though I have never seen or otherwise experienced one; nor have I ever had their existence corroborated by an authority. I don’t even have a body! All I ever have in the way of sensory experience is hallucinations that don’t even seem like veridical perceptions— weird patterns, fuzzy and unreal seeming. I don’t believe in an external world because of these. I believe in an external world because of unsubstantiated rumors—I have no evidence for that belief at all. You can call me irrational and I won’t put up much of a fight. Then one day I magically acquire a human body that I can see and feel. I ask you, an external world doubter, to come over to see for yourself. As you witness my demonstration I say, “Here is one hand, and here is another,” holding out my new hands; “See, I told you material objects exist!” Have I not proved the existence of an external world? I have done so by presenting you with actual examples of material objects—what more could you possibly want? I used to have no evidence to back up my belief but now I have irrefutable evidence—I have a pair of material objects right in front of me.

The Simulation Game

The following document recently fell into my hands: Report to the Commissioner of Games: We recently met to discuss, plan and implement a new game, to be called the Simulation Game (hereafter SG). For this purpose we have created a small group (about 6 billion) of individual centers of consciousness, each with finite and quite restricted intelligence. These are the pieces in the game and they are currently stored in warehouse 7,000,042, suitably hooked up to the simulation machine SM 5000. The system is now fully operational, with each individual experiencing a fully simulated world. Our technicians have verified that there are no glitches. Each individual believes that he or she is living in a world that really exists. The point of SG is to provide clues to the pieces that this is not so and then wait to see when they realize they are in a simulation. We considered inserting some obvious clues into their stream of consciousness, such as skywriting that says “This is all a simulation—you are being fooled,” but that was deemed too obvious, even taking into account the limited intelligence of the pieces. To make the game more interesting, and to net the greatest gambling revenues, we decided to make the clues subtler, though of course any of our species would recognize them immediately. We have therefore arranged it so that the world they experience is incoherent and unintelligible—quite literally impossible. This is not so clear on the surface, but in the game it is gradually revealed, as the pieces apply their limited intelligence to the appearances. The bets are on who will get there first, if anyone. Without going into unnecessary detail, we have built into the simulation a few telltale incoherencies—such as the impression that consciousness depends upon the brain, some logical paradoxes, and the measurement problem surrounding quantum physics. In SG the pieces are allowed to discover and reflect upon their “world” and to ask themselves whether it really makes any sense. Once they realize it doesn’t, the question is when they will hit on the correct explanation of their predicament: that they are pieces in a simulation game. So far the vast majority are clueless but a few have begun to suspect that all is not well—they are starting to feel that they live in an impossible world (or “world”). At later stages of the game the point of interest will be whether they can persuade others of the truth. SG promises to be quite fun and completely harmless (unlike that game Galaxy Busters dreamt up recently by some irresponsible gamers). We ask merely that you, as commissioner, list the new game in your records and grant us the appropriate patent. Thanking you for your kind attention, we are the Society for Responsible Gaming, Section 345, Plasma System 68,000,333.

What should we make of this peculiar document? First, if it is genuine, then these super-gamers are by no means infallible, since the document gives the game away completely. But perhaps they are just being clever, since the existence of such a document by means proves the truth of what it contains. So let us put that aside. The story seems perfectly intelligible: it is logically possible to create a simulated world that contains hidden incoherencies—as with many works of fiction or even dreams. Thus there can be internal evidence that a narrative is a form of fiction, not fact. A

132

Part III

simulation might undermine itself in this way, either by design or through incompetence. We might think that the authors’ breezy assumption of incoherence in the simulated world they have created is questionable. Admittedly, the puzzles of mind and body, of the logical paradoxes, and of the quantum realm are serious and hitherto resistant to intelligible solution; but maybe this is a just a matter of temporary perplexity, or even permanent cognitive limitations on our part. Why should we think that a world in which these problems arise is impossible? Well, that depends on how seriously unintelligible you think the world is—whether such a world would be genuinely impossible. Of course, a world cannot be impossible if it exists to be lived in; but it may be that our “world” is unintelligible simply because it doesn’t exist. That is, there are no bodies and brains and objects of the kind that we suppose, governed by the laws we think we have discovered. It is an impossible fictional world—a perfectly intelligible notion. This is the view of the designers of the Simulation Game: such a world is literally impossible, and the question is whether we will come to realize that and draw the obvious (to them) conclusion, namely that we are pieces in a game of simulation. The form of the argument is straightforward: unintelligible worlds cannot exist; our (apparent) world is unintelligible; therefore, our world does not exist. Given that we experience an apparent world, the best explanation is that we are living in a simulation contrived by superior aliens. Does this story raise the probability that we are subjects in a simulation game? Suppose that we are: do we now have clues that this is our situation, if only we interpret them rightly? If our true situation is revealed tomorrow, will we be right to say, “Yes, it was clear all along, if only we had heeded the signs.” I just wish the document had contained information about what would happen if we arrive at the correct view of our condition. What happens to us at the end of the game? Will it be a case of “game over” with all the pieces disposed of? Maybe we should avoid drawing the “obvious” conclusion for as long as possible.

The Riddle of Knowledge

It is hard to pin down exactly what knowledge is: defining it has never been easy. Part of the problem is that knowledge is pervasive and protean: we know about all sorts of things, in all sorts of ways, and knowledge is found across the animal kingdom. We have sensory knowledge of our immediate environment, knowledge of the past and future, self-knowledge, knowledge of other minds, knowledge of logic and mathematics, knowledge of right and wrong, philosophical knowledge, scientific knowledge, knowledge of how to cook an egg, knowledge of etiquette, knowledge of the position of our own body, innate knowledge, knowledge of people and places. Does anything unite these various types of knowledge? And knowledge is not limited to adult humans: children are also steeped in knowledge, as are many species of animals (from apes to bats, rats, and cats). Knowledge is a basic biological fact—a trait of evolved organisms. It is not a cultural frippery or a leisured indulgence. Knowledge is pretty much coterminous with consciousness: where there is awareness there is also knowledge. Wherever there is something it is like there is something that is known, in one way or another (as the bat knows where its prey is by using its sonar sense). So knowledge is well-nigh universal in animals, and it comes in many forms, but when we try to say what it is exactly we are brought up short. It presents itself as a riddle—so general and yet so elusive, so omnipresent and yet so hard to articulate. We feel we should be able to say what it is straight off, but it leaves us speechless. We shake our heads and ask, “What is knowledge?” What about knowledge as true justified belief? That may work reasonably well for certain kinds of knowledge, despite some well-known problems, but it hardly captures the full range. It is how Plato distinguished knowledge from the mere “opinion” provided by the Sophists, and hence belongs in the context of disputes between discursive rational beings. But how does it work for simple sensory knowledge, or the knowledge of children and animals? Not all knowledge is grounded in belief, in any natural sense, and providing justifications for what one believes is not typical of knowledge generally. How does the definition work for innate knowledge, or proprioceptive knowledge, or knowledge of the way the wind is blowing? Socratic knowledge (as we may call it) is just one kind of knowledge, and not very typical—the reflective school-educated argumentative kind. It is knowledge you could survive without. Compare the concept of consciousness: if we want to know what consciousness is in general, it will not do to define it as “thought” in the manner of Descartes. That is just one kind of consciousness, and it is not typical of all the rest (what about sensations, emotions, and pangs of desire?). The Platonic definition of knowledge is similar: too narrow, too intellectualist, and too geared to reflective adult humans. We need something more general and more basic, but then we find ourselves floundering in unhelpful synonyms (“cognize,” “apprehend,” “be aware of”).

134

Part III

Historically, the response has been to offer metaphors. The most popular is visual: knowing is a kind of seeing. We do know things by seeing, so maybe all knowledge is seeing in some form (with the “mind’s eye”). But knowledge has also been compared to touching and grasping—the idea that knowing is like holding an object in one’s hand (knowing as “prehending”). Then there is the metaphor of acquaintance: we understand what it is to be acquainted with a person or place—well, knowing is like that. Or knowledge is like ability: the ability to act effectively (pragmatism), or the ability to sift evidence, or some such. Knowledge has even been compared to ingestion—it is the taking in, the absorption, of information (“I found your position difficult to digest—I need to chew on it a little longer”). None of these metaphors really resolves the riddle, though they may have some resonance and point. It is difficult even to decide what form an account of knowledge should take—how should we set about giving a philosophical theory of knowledge? Should we just declare knowledge primitive and inexplicable? Should we search for better metaphors? Should we opt for a squishy family resemblance story? Should we leave it to science? When in doubt, consult the dictionary. The Oxford English Dictionary has this for “knowledge”: “information and skills acquired through experience or education.” That doesn’t give us much to go on, but it does provide a significant clue, namely that knowledge is information that has a source—“experience or education.” Let me focus first on education, and let me paraphrase the dictionary definition as follows: knowledge (of one kind) is information gained by testimony, that is, by speech acts (and possibly other kinds of acts). We learn things by being told them. People speak to us and we thereby acquire knowledge—we become informed. The general structure here is this: “x is informed of y by z.” For example, John was informed of the date of the Battle of Hastings by his history teacher: thus it is that John came to know the date of the Battle of Hastings (1066). We could then say: x knows (by testimony) that p if and only if x has been informed that p by some suitable source s. I want to note two things about this formulation: (a) it has a triadic structure, with essential reference made to a source of information (in this case a speaker); and (b) the subject of knowledge is conceived as passive in relation to the active source of the information—he or she has been informed. The agent actively tells the student something and as a result the student passively acquires information. The student may actively (intentionally) listen, but the absorption itself is passive—it is an effect of the source’s agency. Thus, knowledge by testimony is (a) relational and (b) passive. That sounds all well and good, but how does it help to characterize knowledge in general, given that much knowledge (the majority) is not arrived at by testimony? The dictionary definition suggests a way forward in its alternation “experience or education”: maybe experience plays a role very like education in the production of knowledge. After all, we are often told, “experience is the best education.” Can’t we say that x was informed of y by z, where z is an “experience”? Let me replace the

Knowledge135

talk of experience with something else: the world. Then we can say that subjects are informed of things by the world. Just to enliven intuition, indulge me in saying that the world speaks to us—and we thereby become informed, that is, acquire knowledge. The world is a source of information, as our teachers are—indeed, they ultimately derived their information from the world. The world is the ultimate source of all information. It is as if the world is speaking to us—the metaphor is apt. That rustling tree is informing me of its presence—metaphorically speaking. If it could speak, it would proclaim its existence. As things are, it conveys information to observers: it tells us things. Not intentionally, of course, but as a matter of natural fact—information is transmitted from tree to mind, and we end up informed. Knowledge then is being informed of something by something: it has an object (what is known) and a source (the origin of the knowledge). Knowledge by testimony is the model, but the general structure applies across the board. This general structure is what is characteristic of knowledge in all of its variety. Thus, we have the instructional theory of knowledge: knowledge is obtained by being instructed—by the world or by other people. In coming to possess knowledge, we are passively instructed (informed, enlightened, educated) by an active source. Knowledge comes to us from elsewhere—from “experience or education.” That may sound on the right track for perceptual knowledge, but does it work for all cases of knowledge? It is supposed to, so the question is germane. Here is where things get interesting, because it is not at all obvious that the model works for all cases. It seems to work smoothly enough for self-knowledge and even knowledge of other minds, because we can regard mental states as sources of information that is transmitted to a receiver—we can even adopt a causal model of the process. But what should we say about knowledge of right and wrong or knowledge of mathematical truth? Can these areas be regarded as active sources of (passive) information reception? A causal model appears ruled out, for familiar reasons, but it doesn’t yet follow that there is no source for the information in question. And it seems to me true to say that, as a matter of phenomenology, it is as if we are being spoken to in these areas (as people sometimes feel that God is speaking to them): for we experience the “dictates” of morality, we hear the “voice of conscience,” we bow down before the “moral law” (conceived as a set of commands). It is as if morality were speaking directly to us, informing us of what is required of us, issuing prescriptions—instilling moral knowledge. We listen and absorb its instructions. Life teaches us moral “lessons.” Morality imposes its “demands” on us, and so we come to know its content—at least that is how it strikes us. In the case of logic and mathematics, it is more that the symbols seem to be acting as a source of information, of instruction (this may explain some of the attraction of formalism). The symbols convey mathematical truth to us, and they are bits of language. And doesn’t Plato talk as if his geometric universals are conversing with him? He personifies them, treating them as active beings: the circle is especially eloquent. It is

136

Part III

because of mathematical truth that we accept what it offers to us—the truth seems to be announcing itself. Obviously, this is very obscure territory, with much controversy surrounding it, but I don’t think the instructional theory is completely ruled out in these areas; it just needs careful and sensitive formulation. Mathematical intuition might be aptly conceived as being spoken to by the mathematical facts. The case of analytic truth is instructive: here the source of knowledge is meaning itself, and meaning is literally the kind of thing that acts as a medium of communication. Meaning is telling us quite authoritatively that “bachelors are unmarried males” is true. Meaning speaks to us. Thus we are informed by meanings, by mathematical truths, and by moral norms. Facts in general are eager to announce themselves to anyone who will listen—they can be loud and stubborn, insistent and irrefutable. Facts act as sources of information; in that sense they are not passive. When facts become manifest, we have knowledge—we are informed by a source. There are many kinds of source, and many ways of being informed by one: but all knowledge consists in being informed by a source in some way. All knowledge results from instruction, personal or natural. Here is a story about how the concept of knowledge evolved. First, we applied the concept exclusively to knowledge obtained by testimony, so that the idea of an active intentional source was presupposed. The initial concept of knowledge was that of being told something by somebody (say, a tribal elder or wise man). Then we noticed an analogy with the way we acquired other sorts of information, specifically of the perceived environment—it was as if it spoke to us. Objects also conveyed information to us (perhaps we had rather animistic views about objects). So we extended our concept of knowledge to these cases too, taking testimony as our paradigm. We started speaking as if we had testimony knowledge in the full range of cases, exploiting the analogy. We conceived of nature as a font of information, not just people. I have no idea whether this story is true, though it seems perfectly conceivable; but if it is true, we have just traced the concept of knowledge back to its origins. Perhaps we first viewed the world in an animistic way, so that we really thought it was speaking to us—the clouds were actually saying that it is about to rain. So we described ourselves as having knowledge in such cases, because they belonged with testimony cases. Then, when we gave up the animism, we kept the old conceptual practice, converting the literal story into a new metaphorical story. It is as if the clouds speak to us, thus conveying information; and they really do convey information, even if not by actual speech acts.1 Knowledge 1. In mathematical information theory, it is common to speak of natural events as conveying information, not just human agents. In this style people often talk about the physical world as containing signals. Brains are accordingly described as processing informative signals from the external world. This is all grist to my mill: we tend to think of the world as a source of information and hence knowledge derived from the world is a response to that information. Reality acts as an informant.

Knowledge137

by testimony was the original paradigm of knowledge, which was slowly modified to form the broad concept of knowledge we have today. Thus a scientist might even now in a moment of frustration exclaim: “Oh, mother nature, speak to me—tell me your secrets!” The concept of knowledge is the concept of information stemming from a source, whether the source is a full intentional agent or a mindless quasi agent. How does the theory apply to innate knowledge? Who or what is doing the informing? It depends on how the knowledge got there: if it came from God, then God is the source; if it came from genetic inheritance, then the genes are the source (or whatever installed those particular genes). Either way, something is providing information—conveying a message. So it is actually wrong of the dictionary editors to define knowledge by reference to “experience or education,” since not all knowledge has that kind of source: some is innate (or could be, which is all that is required to make the conceptual point). However, there is a source that generates the knowledge even in the case of innate knowledge, so it fits the schema I have outlined. Knowledge comes either from experience or education or inheritance. The source of innate knowledge is our ancestors, given that we inherited it from them. They are to be thanked for supplying us with useful information at birth. The metaphor of the Book of Nature comports nicely with the picture suggested here. When we read that enormous book, it speaks to us—it provides us with knowledge by testimony. Knowledge is precisely that which arises by such means; so if we have knowledge of nature, it must result from something of that general kind. Nature must be like a book, given that we know about it, since knowledge requires an information source. Similarly, when St. John’s gospel announces, “In the beginning was the word,” it is natural to interpret this as an imputation to nature of something language-like—as if reality testifies on its own behalf. We gain information from nature because it sends information to us. The world educates us in its ways. It may even on occasion deceive us, as in perceptual illusions—it misrepresents itself to us. It may also set us riddles to solve (such as the riddle of knowledge). It acts as if it is trying to communicate with us—at any rate, we tend to interpret it that way. The concept of knowledge is structured around these ways of thinking—to know is to be told. Someone may object that the theory is circular, because of the use of the notion of information. Doesn’t “information” just mean “knowledge”? When I am informed of some fact, don’t I simply know it? There are two replies to this objection. First, it is really not clear that we have synonymy here—the word “information” has a different grammar from “knowledge.” And there are notions of information that are clearly not just variants on the concept of knowledge, as in mathematical information theory. But second, the point is not to produce a noncircular analysis in the classic style; it is to articulate a common theme running through all knowledge attributions. That theme is the notion of being receptive to a source of information, the central case of which is knowledge by testimony. All knowledge is like knowledge by testimony (as some say all

138

Part III

knowledge is like seeing). Even in the case of speechless animals, knowledge is information received from a source that acts like a speaker. There is a similar difficulty about characterizing the concept of consciousness. It is hard to define directly, and it is easy to fall into overly narrow characterizations. As a heuristic substitute we find ourselves resorting to the catchy phrase “there is something it is like”: there is something it is like to be an octopus, but not something it is like to be a rock. In the same spirit, we could define the notion of a cognitive being as follows: a cognitive being is a being who is informed about something by a source, either person or thing; and a noncognitive being is a being that is not informed of anything by any source (e.g., a rock). This is not a nice classic definition of the concept, employing noncircular necessary and sufficient conditions; but it serves to tie it down to some degree—just as the “something-it-is-like” definition works to delimit the notion of consciousness. We probably should not expect more, given how basic both concepts are. Knowledge is not definable, but we can at least say something about its general character. What I have suggested here is that the concept of knowledge is best understood by taking knowledge by testimony as primary and other types of knowledge as secondary, instead of the usual procedure of regarding testimony knowledge as a conceptual extension of nontestimony knowledge. Knowing is receiving information from a reliable source, whether person or thing.2 2.  One can imagine a form of theism in which nature is a divine agent speaking truth to mortals. Then all knowledge would be literally a form of education.

Does Knowledge Imply Truth?

The traditional analysis of knowledge as true justified belief has been questioned in a number of ways. The sufficiency of the conditions has been questioned (Gettier cases), and the necessity of justification and belief has been questioned (basic unjustified knowledge, knowledge unaccompanied by belief). But no one denies that knowledge implies truth: truth is taken to be an indisputably necessary condition for knowledge. You cannot know that snow is black or that Paris is the capital of England! On further reflection, however, the truth requirement seems rather too strict and inflexible—it fails to match the actual scope of what we call knowledge. Knowledge is something broader than assent to truth; we can know propositions that are not strictly true—or so I shall argue in what follows. Suppose I set an examination in Golden Mountain Studies, which contains such questions as the following: “Is the Golden Mountain golden?,” “Is the Golden Mountain worth a lot of money?,” “Is the Golden Mountain made of cream cheese?,” and “Is the Golden Mountain hard to climb?” I allow as possible answers: yes, no, and don’t know. I mark yes as correct for the first two questions, no for the third, and don’t know for the fourth. I could give a similar examination in King of France Studies: “Is the king of France rich?,” “Does the king of France speak French?,” “Is the king of France heterosexual?” Answers: yes, yes, and don’t know. I could do the same with a name of a nonexistent but well-known individual (“Zeus,” “Vulcan”). I think that my students would exhibit knowledge by answering in the way I deem correct: they can demonstrate that they know, for example, that the Golden Mountain is golden (not puce). I am testing their knowledge. And yet the propositions they are demonstrating their knowledge of are not true: they are either false (Russell) or neither true nor false (Strawson)—because the Golden Mountain doesn’t exist (like the king of France and Zeus). Suppose now that I am testing my students’ knowledge of ethics and I devise an examination with questions like these: “Is genocide wrong?,” “Is it evil to be charitable?,” “Is pain good?,” “Is it right to keep your promises?” (this is a very elementary examination for the morally ignorant). I grade the examination in the obvious way. I am testing my students’ ethical knowledge. Yet I might also be an ethical emotivist who denies that ethical utterances have truth-value at all. Am I being inconsistent? It doesn’t appear that I am: so again we have knowledge of propositions or sentences that are neither true nor false. Clearly we could do the same with any sentences that fail of truth-value, such as (on some views) vague sentences. If we subscribed to the idea of degrees of truth, we might ascribe knowledge even in cases where the degree of truth is quite low—where the sentence is “roughly true” (but also “significantly false”). If these observations are correct, then knowledge does not require truth—certainly not clear unblemished truth. So what should we say about the traditional truth requirement? Should we say that it is possible to know falsehoods? Is it then possible to know,

140

Part III

say, that snow is black? No, that would be an exaggerated response; what we need to do is expand the concept of truth into something more capacious. We need to see truth as a special case of something more general. And here is where ordinary language proves instructive, because we have many expressions for truth-like concepts; and these fit the kind of examples I have described. Philosophers use “valid” to refer only to logically correct arguments, but in ordinary speech the word is commonly used in the place of “true,” especially where “true” is felt to be too strong (“You are making a valid point, but is it strictly speaking true?”). In this vernacular sense it might be said that yes is a “valid answer” to the question, “Is the Golden Mountain worth a lot of money?” Then too, we have words like “accurate,” “right,” “correct,” “veracious,” and “apt.” On the negative side we have “bogus,” “phony,” “fishy,” “dodgy,” “rubbish,” “tripe,” and “wrong.” In addition, we have many colloquial phrases that express much the same thing: “along the right lines,” “spot on,” “dead right,” “holds good,” “right on,” “on target,” “damn straight,” “on the money,” “rock solid”—with corresponding negative phrases. J. L. Austin spoke disparagingly of the “true/false fetish”—the idea that every utterance can be neatly classified as either true or false (he promised to “play old Harry” with that hallowed dichotomy). He preferred the generic notion of the “felicitous,” which can include more than simply truth. We have a plurality of concepts that resemble truth without being truth—which are truth-like or “truthy.” We may not wish to say that the sentence “The Golden Mountain is worth a lot of money” is true, but we have no hesitation in saying such things as that it is “along the right lines”—as compared to “The Golden Mountain is as light as a feather.” And we can know these things: we can believe them, justify them, assert them—we just can’t say that they are strictly speaking true. But they are (and must be) positive in some way: you can’t be said to know what is “complete rubbish” or “dead wrong.” In the case of ethics and aesthetics (or anywhere else where an expressivist theory seems attractive) people have felt that truth is the wrong concept to apply, preferring “apt” or “appropriate”; but then they have supposed that they must reject “cognitivism” in ethics and elsewhere—ethical principles are not objects of knowledge. But if I am right, we can break these things apart: knowledge does not require truth, so we can have ethical knowledge in the absence of ethical truth. If we are inclined to a very demanding conception of truth, this will be an attractive option—we can be cognitivists but not “truth-ists.” You can know a lot about ethics, even if none of it is actually true—just as you can know a lot about the Golden Mountain, even though there are no truths about it (only truth-value gaps). One approach in these matters is to adopt a very free and loose notion of truth, which allows every declarative sentence to be either true of false (and even nondeclarative sentences, if we get very loose). Then we can continue to take truth to be a necessary condition for knowledge, given that we want universal cognitivism to hold. That tends to lead to divisions within the broad concept of truth, so that all we have is family

Knowledge141

resemblance or even ambiguity; and then the point of the broad notion is brought into question. The option I have explored here allows us to keep a fixed and univocal notion of truth, which is quite restrictive, but without then denying the possibility of knowledge in areas where the notion of truth does not apply. Whether this is the right approach in ethics is another question (I doubt it myself since I am happy with the notion of ethical truth), but it opens up the logical space a bit. What I would insist is that we don’t want our notion of knowledge to be hostage to the notion of truth. It is possible to know something without knowing it to be true—it suffices to know it to be quasi-true (“on the money,” “spot on,” “the opposite of rubbish”). That is, it is possible to have a restrictive concept of truth and an expansive concept of knowledge. Suppose we came to the conclusion that the concept of truth is deeply confused and bankrupt: it can’t be “naturalized”; it is bound up with an indefensible metaphysical realism; it is mired in intractable semantic paradox. We decide to junk it completely— we just stop using the word. We hold that there is no such thing as the property of being true. Would that mean that we have no knowledge, since knowledge presupposes truth? That would surely be a case of throwing the baby out with the bathwater. We could still know things, just not know them as true. We would need some replacement notion that escapes the troubles of truth, so that we could distinguish the “good” propositions from the “bad” ones. Call that replacement notion “acceptability”: then we will be able to say that knowledge requires acceptability. This seems perfectly coherent, and it shows that the concept of knowledge is not logically committed to the concept of truth. We can eliminate truth without eliminating knowledge. We need something like propositions for there to be propositional knowledge, but whether propositions have the property of truth is a separate question. If we took coherence to be our replacement for truth, then knowledge would be of propositions that cohere with other propositions; truth (in a correspondence sense) would not come into the picture. Knowledge requires acceptability of some sort, but it does not require truth in any nontrivial sense (such as “correspondence to fact” or “denoting objective reality”).

Everything Is Hidden

The hidden is rampant in nature. Other minds are the standard case: what you are thinking and feeling is hidden from me. I can guess at it, make speculative inferences, but the way you are inwardly is deeply concealed to me, as if cordoned behind a permanently opaque barrier. Wittgenstein bravely maintained that other minds are not so hidden, but he knew he was denying the self-evident (hence the excitement of his denial).1 My mind is indeed not hidden to me (though see below), but it is certainly not presented to others with the transparency with which it is presented to me. No amount of gazing and prying will crack open the nut that is someone else’s inner consciousness (scrutinizing the brain is just perceiving more of the shell). So large tracts of nature are hidden to any individual knower—the minds of all humans and animals apart from that individual himself. The hidden exists wherever another consciousness exists—and not just hidden temporarily and contingently, but permanently and necessarily. But there is another layer of the hidden within each conscious subject, and this too is systemic and irrevocable. In addition to the conscious mind there is the unconscious mind—or the many unconscious minds, since the unconscious is not unified. By definition I don’t know what is going on in my unconscious mind (or minds): it is hidden from my conscious knowledge. I can guess at it, make speculative inferences, but I can’t know it as I know my conscious mind; it is as cut off from my awareness as other minds. It is so hidden that it is quite coherent to deny its existence, or to be a skeptic about its content. The unconscious is a region of ignorance, speculation, and darkness. There is no cogito for the unconscious (“I have unconscious desires, therefore I exist”). It is a concealed mental reality. So this is a second area in which nature hides its contents from us, where an epistemic barrier has been erected. The beetle sits invisibly inside its opaque box.2 One might suppose that this is the limit of the naturally hidden: the individual consciousness is hidden from others, and the unconscious is hidden even from its own 1.  The phrase “nothing is hidden” comes from Wittgenstein’s Investigations, section 435 (“How do sentences do it?—Don’t you know? For nothing is hidden”). I am interpreting it far more broadly than just the question of how sentences manage to represent. The concept of the hidden seems to me important in epistemology (it doesn’t coincide with the concept of the unknown, though the two concepts are connected). 2. It might be claimed, following Freud, that the contents of the (or one) unconscious can be become conscious, and so are not necessarily hidden. But that does not gainsay the point that we can never be aware of our unconscious as unconscious—so the unconscious is hidden from us in its condition of unconsciousness. And even if someone comes to know what is in his unconscious by means of inference, this is still a case of the unconscious remaining hidden, since there is no immediate awareness of it—as a matter of principle.

144

Part III

bearer. But the physical world is surely not hidden! It can be seen and heard, examined and surveyed. But can it? According to a well-known view, we observe the symptoms of the physical world and describe it mathematically, but we don’t really know its essential nature. We don’t know the intrinsic reality of matter, motion, energy, space, and time, only their structural relations: they are “know-not-what’s” that we merely name and then bring under mathematical laws. The underlying reality of the physical world is thus hidden from us. Some theorists have supposed this covert aspect of nature to be itself mental, with the part in our brain known directly, so that our epistemic relation to the physical world is like our epistemic relation to other minds. The physical world is one vast Other Mind brimming with hidden mental reality, but not revealed to us. And even if it is not mental in nature, but something quite unfamiliar to us—perhaps beyond our conceptual scheme altogether—it is still hidden, still inscrutable. We know it only under the description “the cause of so-and-so,” where “so-and-so” refers to something observable, perhaps states of our own consciousness (sense data and the like). If this is right, then the physical world as a whole is deeply hidden, despite the success of physics (compare behavioral psychology). Everything is hidden, at least in part. To be cloaked and veiled is endemic to nature. The only area of the nonhidden is one’s consciousness of one’s own consciousness.3 But one’s consciousness is hidden from every other consciousness, and everyone’s unconscious is hidden from everyone. Moreover, the physical world (in its ultimate nature) is hidden from everyone. Nor are these limits remediable: they are built into the structure of reality and our means of knowing about it. It is as if the world has been designed to thwart epistemic access, to insist on secrecy. When people speak of “nature’s secrets,” they hit on an important truth: nature is filled with hidden realities, tantalizingly out of reach. Our thirst for epistemic intimacy is gratified only in the case of our own individual consciousness. Our so-called knowledge exists only within a vast ocean of the unrevealed and ulterior. Nature is constitutionally reticent about itself. 3.  Even in the case of my own consciousness, it may be that there exist hidden aspects; so we should more cautiously say that the surface of my consciousness is not hidden from me.

Light and Our Knowledge of Matter

It has been said that we don’t know the intrinsic nature of matter, only its abstract mathematical structure and how it appears to our mode of sensibility. Sometimes this view is defended on the grounds that we only know our own inner experiences intrinsically. That defense presupposes a questionable empiricist epistemology and a highly restrictive theory of knowledge. I propose to defend it differently: on the basis of the physics of light. How do we know about matter? Primarily through vision: we see objects or their traces. This holds for the immediate environment as well as for the furthest reaches of the universe: I see the trees through my window, and astronomers see the red shift of distant receding galaxies (or if they don’t actually see the red shift, they infer it from seeing what their instruments tell them). In either case the knowledge is mediated by light; and it has been remarked that if it were not for light astronomy would be impossible. We cannot touch, hear, or smell distant stars and galaxies, but we can see the light that emanates from them, and this is the basis of our astronomical knowledge. But the same basic point applies even to nearby objects: vision is the primary source of our knowledge. It is true that we often can hear, touch, and smell nearby objects, though most such objects are outside the range of these senses; but consider how impoverished such data are compared to vision. The case of hearing is instructive: strictly speaking we don’t hear material objects; we hear the sounds they produce—from that we infer the nature of the remote object. If we heard only sounds from objects, we would be naturally receptive to the idea that our knowledge of matter is limited. Even in the case of touch, it is evident that our knowledge is limited to how matter interacts with the human body (sensations of resistance and hot and cold)—touch doesn’t reach into the heart of matter in its objective being. If we removed the other senses from the picture, our knowledge of matter—both common sense and scientific—would be much the same, but if we removed sight our knowledge would be greatly diminished. That is, if we removed the sense that depends on the properties of light, our knowledge of matter would be seriously impoverished. It is light above all that reveals the material world to us. The question is how much of matter does light reveal. For the power of matter to emit or reflect light is just one aspect of its nature. Light begins its journey from the Sun and arrives at planet Earth in short order—a voluminous and frantic army of tiny particles known as photons. It encounters a material object, which stops it dead in its tracks (despite that famous velocity); or rather, some regiments of particles penetrate the object while some rebound from it. The rebounding regiments strike out in new directions and some of them enter the eye of a beholder. As a result of intricate operations, the object is, as we say, seen. That darting light is the bearer of visual information, and as a result of it the beholder forms his or her picture of the material world, including the astronomical world. But how much of the intrinsic nature of the material object is conveyed in the light that rebounds from it? The answer is: only so much of matter

146

Part III

as is required by the physics of light. It is in virtue of certain properties of matter that light is reflected from it (or emitted by it), but these properties do not exhaust the full nature of matter; they concern only the power of matter to absorb and reflect a certain type of particle, viz. the photon. Light reveals matter only in one of its aspects. Compare sound and matter: when sounds emanate from an object certain of its properties are active, but not all. We don’t suppose that everything about matter is somehow contained in the sounds that it makes; this is just a very narrow band of its full range of properties and powers. But light is the same: emitting light is just one thing matter can do—it has many other tricks up its sleeve (occupying space, mass, gravity, motion, magnetism, etc.). From a physical point of view, light transactions are just one among many properties of matter; but from an epistemological point of view, light is all we have to go on (plus limited data from the other senses). So there is a mismatch between the ontology of matter and its epistemology. If there were no light, or very little of it, we would know next to nothing of matter; but even with floods of light, only a single aspect of matter becomes manifest to us. Matter-in-relation-to-light is just part of its nature; and it is physically contingent, since light does not exist necessarily—there could be matter in a world without (visible) light. Our knowledge of matter is essentially light relative, and hence partial. In the land of the blind, where only sounds provide information about matter, knowledge of matter would be sound relative, and hence partial (and the same goes for smell, taste, and even touch). Thus it is that we fail to know the intrinsic nature of matter, knowing only one of its relational properties—how it configures light (or sound, etc.). The point can be made vivid by recalling Russell’s view of matter and our knowledge of it, namely that only in introspection is the intrinsic nature of matter revealed to us. Matter actually has a mental nature, but this is closed off to our ordinary perception of it; only in introspection is its inner nature revealed. Matter does not look mental! But, given that Russellian view, we can see that the ability of matter to emit light has nothing to do with its real intrinsic nature: for the mental nature of matter is not part of its power to interact with light (pending some ingenious theory that connects the two). The underlying nature of matter is mental, but our ordinary knowledge of matter, as mediated by light, tells us nothing about that nature, being confined to the light-interactive properties of matter. Putting aside the Russellian view, we can say that if the essential intrinsic nature of matter consists in possessing property N, the light-interactive powers of matter are unlikely to exhaust, or even touch, the property N. The full nature of matter is simply not revealed in its interactions with light, but our knowledge of matter depends crucially on these interactions (slightly supplemented by the other senses). We try to infer the further properties of matter from this slender basis, but there is absolutely no guarantee that we can reach much beyond the partial view we are presented with. Just as it is hard to infer the real objective nature of matter from our subjective experiences of it, so it is hard to infer the real nature of matter from the patterns of light it projects in

Knowledge147

our direction. What we get is not the kind of intimate knowledge we have of our own consciousness but an abstract and structural conception of matter—a mere skeleton or diagram of matter inferred from the light it sends our way. Matter in its intrinsic being remains inscrutable. Note that I am not trying to prove the ignorance thesis with respect to matter from these considerations about light; I am offering these considerations as the explanation (or part of the explanation) for why we know so little about matter (apart from the kind of abstract mathematical knowledge we find in physics). Nothing like this holds in the case of our knowledge of our own minds: we don’t know the nature of our conscious experiences via the mediation of some sort of energy that reflects only an aspect of their nature—as it were, the patterns of light or sound waves emitted by experiences. So we should not expect an ignorance thesis about our knowledge of consciousness. But given the photon-based way we know about matter, we should expect just such a thesis about matter—purely in virtue of the physics of matter and light. If our knowledge of matter were confined to only a superficial property of matter that happened to be relevant to our biological survival, say its power to induce certain kinds of headache, then we would not suppose that we knew much about matter. But it is not essentially different with light and vision: light is important to us because vision is important to us, but it may not be all that important to matter. To be sure, light allows us to know all sorts of things about matter we would not know without it (such as the recession of the galaxies), but it would be an unfounded dogma to insist that light can tell us everything about matter. Light (visible light) offers just the outer skin of matter, not its bones and innards. To change the metaphor, what we see of objects is just the image they project in the form of a package of light; what lies behind that image in the projection room is entirely another matter—quite literally. Light affords us merely the face that matter chooses to project to the world.1 1. Dark matter, dark energy, and black holes are interesting in this connection. These are all forms of matter that do not reflect or emit light, so we know very little about them. But if physicists somehow find a way to coax them into interacting with light in such a way that we can see this light, we should not suppose that we thereby achieve deep knowledge of their inner being. Evidently, their being does not depend on their power to interact with light in the manner of visible objects, so it can hardly be that their material nature is a light-involving nature. Dark matter is matter at its purest and most inaccessible, beyond the reach of photonic detection. Dark matter may remain forever dark, though fully real and internally structured. Here we will need no persuading of the unknown nature of matter. Yet we may attain some structural mathematical knowledge of this elusive substance. Maybe there are universes in which all the matter and energy is dark, with nothing but black holes; such a universe is as crammed with matter as our own. In the case of our actual universe we tend to think the darkness is limited, with most matter revealed to us; but according to the ignorance thesis all matter is somewhat dark. The darkness is obscured by the light, giving us a vivid image of matter, but light can only penetrate so far into matter’s hidden recesses.

Seeing the Light

Philosophy of perception has been dominated by two theories: the sense datum theory says that mental items are the immediate objects of perception, while naïve realism insists that material objects themselves (or possibly their parts or facets) are what is immediately perceived. The sense that is generally considered is vision, but the traditional two options would not appear exhaustive if hearing were the focus. In the case of hearing there is the possibility that sounds are the proper objects of perception: not the objects that make these sounds, and not the sensations of sound (or some such), but simply the sounds—the physical sound waves that emanate from objects. These are objective physical particulars, but they are not the objects that produce them, and they are not internal to the perceiver; they are perturbations in the atmosphere (or whatever the medium may be). Similarly, in the case of smell we do not smell the remote object itself but rather the chemicals that reach our nose—they are what is “immediately perceived” (the remote object may be said to be perceived “indirectly”). What I shall argue here is that we should adopt this kind of theory for vision: we don’t see the remote object (except derivatively perhaps) but rather the packet of light emanating from that object. All we see is light, never material objects, and never visual sense data. So it is not literally true that I see the cup in front of me; what I see is the light reflected off it into my eyes. We may say loosely that I see the cup, but that can only mean that the cup is the object from which the light emanates that I see. Logically, it is just like the naïve realist position that I never strictly see the whole object but only a part or facet of it; we can speak loosely of seeing the whole object, but underlying that locution is the fact that what I immediately see is a part or facet of the object. The difference is that for the theory advocated here it is the light that I strictly see, not a part of facet of the object—but we may continue to speak loosely of seeing the object, because of the relations that exist between what we strictly see and the object. I will drop this qualification from now on and speak simply of what we see (or “really see”); so the thesis is that what we see is light, and that is all that we see. In a very real sense, matter is cut off from the realm of the seen: the visual world is composed wholly of light. I mean to speak of light in the physicist’s sense: streams of photons, electromagnetic waves—not any kind of subjective experience of light (sense data of light). So what we see are streams of photons or electromagnetic waves, which obey the laws of physics (including quantum theory). It is not that we see light as such streams or waves; rather, these things are the de re objects of vision. When I see a patch of light on a lake, what I am seeing is the thing described by physics. I see the light on the lake, and that light just is a stream of photons (I will omit mention of electromagnetic waves from now on). So the thesis is that all I ever see are streams of photons, never the objects that reflect or emit them, still less my own sensations. I therefore see what lies between

150

Part III

external objects and my mind. It is a question whether we should call light “material,” but if it is not then I never see material objects—nor do I see mental objects. What I see is a type of radiation: that phenomenon that travels at 186,000 miles per second, breaks up into the color spectrum, and obeys the peculiar rules of quantum theory. I don’t see cabbages and kings or chairs and tables (though I do refer to them); I see only the light with which these things interact. Compare: I don’t hear trains and violins or people and birds but only the sounds they produce—what I hear is strictly a world of sounds, not objects. The only sense in which I can be said to see or hear objects is that I do so by hearing sounds and seeing light (not processing sound and light but actually hearing sounds and seeing light). If we removed material objects from the world, leaving only sound and light, we would not reduce the number of things (directly) heard and seen; a world of only sound and light, with no material objects, is still a perceptual world—in fact, it is the same perceptual world that we actually live in, since sound and light are all that we really sense. How might we argue for this radical-sounding position? First, we need to note that we can see light: light is visible. Thus we see moonlight on water, light bulbs, dappled sunlight, and the radiant Sun itself; also gleams, flashes, glares, blazes, glimmers, glints, glows, and sparkles. In fact, whenever it is light, we see light: the light permeates space and our eyes respond to it. We see the light from the Sun all day long. So it is not that light itself is invisible while rendering objects visible; it is itself visible. We literally see photons (de re). The question is whether when (as we say) we see an object, we in the process see its light: do we see the light reflected from an object when it is observed in ordinary daylight? More, is that all that we see (strictly)? It is true that we wouldn’t ordinarily say that we are seeing the light from an object, but for familiar reasons of conversational implicature that has no relevance to the question of whether it is true that we see an object’s light. When I see a gleam of light on an object I am seeing light, but am I also seeing light when I see its matte-colored surface? Here is an argument leading to that conclusion, which I shall call “the argument from material-object subtraction,” or “the argument from subtraction,” for short. It is analogous to the argument from illusion, which is better called “the argument from hallucination,” in that both arguments involve subtracting the material object present in ordinary cases of perception. So suppose we subtract the actual cup from my perception of the cup leaving only the packet of light that emanates from the cup (a “photonic duplicate”): the photons are still there, but the cup is gone. What do I see if anything? Not the cup, to be sure, and not my sense datum of the cup, but the light that remains—the very same packet of light that is normally created by the cup. This is the cause of my experience, and it causes the experience in exactly the same way light normally causes experiences of objects (no deviant causal chains). It is like a gleam: a visual object in its own right. It is literally true to say, “He sees a packet of light that normally comes from a cup.” The subject certainly sees something—it is not that he

Knowledge151

sees nothing at all (as is arguably the case in complete hallucination). He may think he is seeing a cup, or he may not if he is apprised of the situation; in either case, what he actually sees is not a cup but a cup-like packet of light. But then, if he is seeing light in the subtraction case, is he not seeing the same thing in the original case? If we put the cup back, he hardly stops seeing the light he saw a moment earlier; so isn’t it right to say that in both cases he is seeing a packet of light? It doesn’t yet follow that he is not seeing the cup in the veridical case, but only its light; but it is surely clear that in both cases the relevant packet of light is seen (a “disjunctive” analysis seems highly implausible in this kind of case). The light is there to be seen in both cases, and it is seen in the subtraction case, so it must also be seen in the veridical case. We don’t render light invisible by backing it up with lumps of matter: the cup does not form some kind of wall between the perceiver and the light projected by the cup! Now consider reflections. If you see something in a mirror, light from the object falls onto the mirror’s surface, in virtue of which (as we say) you see the reflected object. Do you see this light or not? It is hard to deny that you do, since it is right there in front of you stimulating your eyes. If the object were to disappear suddenly, leaving only its light, would you hesitate to say that you see the residual patch of light? You see the reflection as much as you see the reflected. In fact, you directly see the reflection and only indirectly see the object reflected: for it is in virtue of seeing the former that you see the latter—you see the object by seeing its reflection. Suppose you only saw objects by way of reflections (you live in a mirror world): wouldn’t you have zero qualms about saying that reflections are what you primarily and directly see? And reflections just are packets of light, that is, swarms of photons. When you see your face in the mirror, what you are really seeing is the swarm of photons currently emanating from your face as it strikes the mirror’s surface (which you see as a face). The difference between the mirror case and a nonmirror case is just that the photons take a more circuitous route in the mirror case: but in both cases it is the photons that are seen. It is much the same story with seeing objects through a refractive medium, like distorting glass or rippling water: you see the light that travels through the medium, which is why illusions arise in this way (e.g., the bent stick in water). There is an old puzzle about distant galaxies, so far away that their light reaches the human eye long after they have ceased to exist: do we see such long-gone galaxies? That seems a strange thing to say because we normally suppose that objects exist at the time of seeing them: to say “x sees y at t but y does not exist at t” sounds contradictory. But it is uncontroversial that the light that emanated from the extinct galaxy still exists—it is right now hitting your eye. So shouldn’t we say simply that you are seeing the light from the nonexistent galaxy but not the galaxy itself (except in some derivative way)? But it is also true that any object we are said to see might have gone out of existence just before its light hits the viewer’s eye. Suppose you lived in a world where this always happened—ordinary nearby objects are like remote extinct galaxies:

152

Part III

wouldn’t it be natural to say that light is what is seen, not objects (save derivatively), since objects no longer exist at the moment of perception? But then in our world there is always a time lag as light makes its journey from object to eye. If light were to slow down to a snail’s space, so that objects were forever going out of existence or moving away before the eye was alerted to their presence, wouldn’t it be natural say that it is light that is directly seen? Finally, consider the case of a pure light world, analogous to a pure sound world: there are no material objects in this world, just free-floating light, darting and coalescing. This is a world made entirely of gleams and glimmerings, of radiance and incandescence. Here there is nothing to be seen except light—so nothing competes with light for the title of what is seen. All the inhabitants ever see is light, pure and unadulterated. If some of this light forms into packets that resemble those deriving from objects in other worlds, the inhabitants still see this light. And if objects eventually come along to populate the light world, adding their ponderous presence, we shall still say that light is what vision is all about in this world. If visual concepts had developed in such a world, the inhabitants might find themselves reluctant to extend them to these newly arrived chunks of matter (so slow, so heavy!), reserving the word “see” for good old-fashioned splashes of pure light. That is what is really seen! Once you have come to think of vision in this light-centered way, it starts to seem strange to suppose that we ever really see material objects. For vision is par excellence an affair of light: vision occurs only because of light, and the visual system is a system for processing patterns of light. Objects merely exist on the other side of light, so to speak. Objects crouch behind the light speeding from their surface. What we perceive of objects is wholly a matter of how they interact with light—with the light that they produce. We see their properties only as they are reflected in light: how we “see” objects is a function of the light they deliver (as well as our own constitution). The color and shading of an object are effects of its light, with the actual object lurking in the shadows. It is the same with sound: hearing is an affair of sound, not of the objects that produce the sound. It is strange to suppose that we hear objects themselves; we hear the sound of objects. Similarly, we see the light of objects. We can learn about objects by seeing their light, as we can learn about objects by hearing their sounds; but when it comes to sense perception itself sound and light are what it is all about. Hearing is the sense responsive to sound and vision is the sense responsive to light: hence it is natural to acknowledge that it is proper to say that sound and light are the objects of these senses. We just need to rethink our customary ways of speaking; and when we do, the idea of seeing objects starts to seem strained (really seeing them). This is not so much conceptual revision as conceptual recognition. We used to think of vision as some sort of prehensile reaching out to objects, as if it were not mediated by influxes of energy; we now recognize that the properties of light are crucial to vision. The next step is to recognize that light itself is what is primarily seen. In the case of stars, we used to think

Knowledge153

that we saw the actual celestial body; now we recognize that we are really seeing the light it gives off, the star being too distant to see (its light radius is much larger than its physical radius). We need to make the same kind of revision in the way we think of terrestrial vision. As a final nudge in that direction, consider seeing in low light. If the light is very low it reveals little of the objects it is reflected from—we might be seeing, say, a cat or a stool in that corner. Yet we still see the dim light that falls on these hard-to-identify objects—that is perfectly clear. We can see the light well enough, but our vision of objects is drastically reduced. Here we are inclined to say that we can’t see the objects around us but only the weak light playing on their surface. Well that is really our predicament with respect to objects and strong light: we see the light playing on their surface, but not the objects themselves.1 And if the light falling on objects became much stronger than it is now, so much so that we were half-blinded by ordinary objects (like looking at the Sun), wouldn’t we then be ready to accept that we are not seeing the objects, but only their blinding light? If objects become too bright, they overwhelm our visual system, so that we can’t make out detail and form; then we shall want to say that we can’t see the object but only its light. It is hard to find a suitable name for the theory described here to be set beside “naïve realism” and “the sense datum theory.” We could try “the light packet theory” or “photon realism” or “the proximal stimulus theory,” but none of these really resounds. Maybe we could just call it “the gleam theory”: at least that is unpretentiously brief and sounds nice. 1.  The experience we have is as of a material object with certain material properties, even though what we see is a congeries of light. We experience the light as representing an object, and the object may well have the very properties it is represented as having. Thus the experience is veridical. However, the direct de re object of our experience is the light, not the material object. Nothing in this implies that there is any actual error in ordinary visual perception: objects can be as they seem to be, even if they are not directly seen.

IV Metaphysics

Knowing and Necessity

I hope that some people see no connection between the two topics in the title.1 In any case, the absence of any connection will be developed in the course of the essay—a complete and clean separation. The way I think about these matters is, in some ways, quite different from what people take for granted these days, and is certainly very different from the orthodox position during most of the twentieth century. Some of my views may strike people at first sight as obviously mistaken, indeed as scarcely intelligible. Among my more surprising claims, to be defended subsequently, are the following: All necessity is uniformly de re; there is simply no such thing as de dicto necessity. The customary distinction between de re and de dicto necessity is an untenable dualism. There is, however, no such thing as empirical essence or “a posteriori necessity.” But nor is there such a thing as a priori essence. Neither are there any empirical facts, still less a priori facts. In addition, there is no separate category of epistemic modality—no epistemic necessities or contingencies. There is no intrinsic or conceptual relation between types of knowledge and types of modality at any point. What is called “conceptual necessity” is not to be understood in terms of knowledge of concepts or a priori truth. In fact, every truth is both empirical and a priori. So my views are somewhat surprising: but I intend to establish all these strange views in what follows. Where We Stand Today In the glory days of positivism, all necessity was understood as uniformly the same: a necessary truth was always an a priori truth, while contingent truths were always a posteriori. The attribute of necessity was taken to be reducible to the attribute of apriority. The attribute of apriority was in turn taken to be reducible to the attribute of 1.  In this paragraph I playfully parody the opening of Saul Kripke’s Naming and Necessity (Cambridge, MA: Harvard University Press, 1980), which begins: “I hope that some people see some connection between the two topics in the title.” In fact, I think this is a misleading statement on Kripke’s part, because the ensuing text never does establish any close connection between the two topics, and the spirit of the book is actually opposed to such an idea. For, first, Kripke’s notion of metaphysical necessity has nothing essentially to do with naming, being inherently nonlinguistic; certainly, such necessity does not arise from names. Metaphysical necessities would exist even if names did not. Second, necessary identity statements can readily be formed using rigid descriptions, as in “the successor of 2 is the predecessor of 4,” just as they can by using names, yet I doubt Kripke would want to say that there a special connection between describing and necessity. The same is true of demonstratives, which are also rigid designators; so there is nothing distinctive about names here. It is also odd to use the word “naming” in the title instead of “names,” since the text is hardly about the act of naming at all, though it contains a lot about names.

158

Part IV

analyticity, conceived in terms of something like Kantian conceptual containment. And then, finally, analyticity was held to be reducible to linguistic conventions or rules or stipulations. Thus, necessity was reducible to linguistic convention: all necessity was accordingly deemed de dicto, that is, a property grounded in meanings or propositions or sentences. Necessity did not reside in the world outside of linguistic or conceptual representations but had its origin and being within the world of symbols: it was essentially a reflection of the conceiving mind, as expressed in language. Thus every instance of necessity was deemed a priori and analytic, unlike the attributes dealt with by science and other reality-directed enterprises. Modality was all “in the head”—a product of thinking and symbolism, not objective reality. It may seem to us that some necessities belong in the world, but that is an illusion; strictly speaking, necessity reduces in all cases to an epistemic or cognitive attribute. To be necessary is just to be known a priori.2 In Naming and Necessity Saul Kripke challenged the reductionist positivist picture, root and branch. He begins by remarking, correctly, that many philosophers of the time (ca. 1972) use “necessary” and “a priori” interchangeably, taking the two terms to be virtual synonyms. But, he notes, the concept of the a priori is clearly a concept of epistemology, signifying something like “knowable independently of experience.” He goes on: The second concept which is in question is that of necessity. Sometimes this is used in an epistemological way and might then just mean a priori. And of course, sometimes it is used in a physical way when people distinguish between physical and logical necessity. But what I am concerned with here is a notion which is not a notion of epistemology but of metaphysics, in some (I hope) non-pejorative sense. We ask whether something might have been true, or might have been false. Well, if something is false, it’s obviously not necessarily true. If it is true, might it have been otherwise? Is it possible that, in this respect, the world should have been different from the way it is? If the answer is ‘no’, then this fact about the world is a necessary one. If the answer is ‘yes’, then this fact about the world is a contingent one. This in and of itself has nothing to do with anyone’s knowledge of anything. It’s certainly a philosophical thesis, and not a matter of obvious definitional equivalence, either that everything a priori is necessary or that everything necessary is a priori. Both concepts may be vague. That may be another problem. But at any rate they are dealing with two different domains, two different areas, the epistemological and the metaphysical. (35–36) 2.  A simple example would be explaining the necessity of the number two being even by saying that we have decided to use “two” and “even” in such a way that if the former applies so does the latter. Since we know such linguistic stipulations a priori, we cannot envisage a case in which they fail once made. Two is necessarily even because the meaning of “two” includes the meaning of “even” by stipulation, so that the necessity reflects a more basic linguistic fact and cannot hold without it. Hence the necessity is declared de dicto.

Metaphysics159

Along with many others, I have long found Kripke’s insistence on the distinction between the concepts of the a priori and the necessary salutary and undeniable; and I applaud his formulation of necessity in terms of the how the world could be, with his reference to facts as the proper bearers of necessity and contingency (not sentences or propositions or beliefs). He is absolutely right to keep sharply distinct the metaphysical and the epistemological, assigning necessity to the former category and apriority to the latter. Having made these general observations, he then proceeds (famously) to offer examples in which the two concepts come apart, thus showing that they are not even coextensive, let alone synonymous or identifiable. He proposes cases of empirically known necessary truths and contingent truths that are known a priori (“water is H2O,” “this rod is one meter,” respectively). A firm wedge has now been driven between the epistemological concepts and the metaphysical concepts: we must not confuse these concepts, and the positivists were wrong to try to reduce one to the other. Undoubtedly, the most influential point in all this has been Kripke’s contention that not all necessary truths are known a priori: some truths are both necessary and a posteriori. As examples, he offers necessities of identity, constitution, and origin. This flatly refutes the positivist thesis that all necessary truths are known a priori, and hence the thesis that necessity resides in language or the knowing mind. Thus, philosophers took to saying that Kripke had established the credentials of de re necessity: not all necessities are de dicto—some are de re. Now it is a notable fact about Naming and Necessity that it contains hardly any discussion of analytic truth or conceptual necessity; so we don’t know how Kripke thinks this category of necessary truth fits into his overall framework. But a common interpretation of his contribution, and one that is entirely consistent with everything in the text, is that he established that in addition to the class of analytic de dicto necessities there is also a class of synthetic de re necessities. He didn’t show, nor did he attempt to, that the positivists were completely wrong in identifying a class of de dicto necessities; he showed rather that these necessities are not the only kind of necessity. There thus grew up the habit of talking of two opposing types of necessity: de re necessities versus de dicto necessities, empirical necessities versus conceptual necessities, synthetic necessities versus analytic necessities, a posteriori necessities versus a priori necessities. Nor did Kripke’s exposition discourage the promulgation of these dichotomies; indeed, it appeared to provide solid ground for them. And the talk of very different types of necessity naturally leads to the idea that “necessary” is ambiguous—sometimes meaning necessity in the de re sense and sometimes meaning necessity in the de dicto sense. At any rate, we are often sternly warned not to conflate or assimilate the two, in the style of the bad old positivists. The underlying thought is: the positivists can have their traditional idea of conceptual necessity,

160

Part IV

but we also must recognize a quite distinct class of necessities. The attribute of necessity cannot be identified with the attribute of apriority for the new class of empirical necessities, but there might still be such an identification of attributes for the conceptual necessities. The positivists were not fundamentally wrong about those; their mistake was just to think there is no other kind of necessity. The reduction does not work for all necessities, but it still works for a subclass of necessities. At any rate, this divided position is quite consistent with the arguments of Naming and Necessity. I venture to suggest, indeed, that it is the conventional wisdom in the philosophical community today: some necessity is empirical and de re and some necessity is a priori and de dicto—where this latter kind of necessity has a quite different ground from the former kind. Crudely, one kind of necessity springs from the world, while the other kind springs from language or concepts. In the latter case, then, it is hardly worth distinguishing between being necessary and being a priori; after all, there are no empirical analytic propositions—they are all uniformly a priori. Consequently, post-Kripke, we warn our students to beware of the ambiguity of “necessary” as between the de re and the de dicto, pointing out that some necessary truths owe their existence to the world—to the objective essences of things. What is important is that Kripke taught us to expand the category of necessary truth beyond the traditional class of the de dicto analytic a priori necessary truths. Necessity Homogenized That is pretty much where we stand today, and where Kripke left the matter: it might be called the “the doctrine of modal dualism.” But there has to be something wrong with the received dualistic picture. First, the idea of an ambiguity in “necessary,” as between empirically known cases and cases known a priori, is distinctly unappealing—as is the idea that necessity itself really divides into two sui generis types. This seems like a philosophical thesis, not a record of what the word “necessary” actually means in the mouth of the naïve speaker. Don’t we use the word “necessary” in the same sense when we apply it to empirical and a priori propositions? It is nothing like using “bank” to refer to riverbanks and financial banks. Second, when Kripke rightly insists on the distinction between the a priori as an epistemological concept and the necessary as a metaphysical concept, he does not qualify his words to make room for the analytically necessary: he simply speaks of necessity as relating to how the world could be and what kinds of facts it could contain—he doesn’t create a separate category of the necessary to deal with the analytic cases. The concept of necessity, he assumes, is always defined nonepistemically, in terms of how the world could be—it is not identifiable with the a priori for a certain subclass of cases. So, instead of supposing that all necessity is de dicto, like the positivists (and others), what about exploring the idea that it is all de re? That would restore the univocal sense of “necessary” and also vindicate Kripke’s universal assignment of necessity to the class of

Metaphysics161

metaphysical concepts. Can we then defensibly maintain the view that all necessity is de re?3 It turns out that this is no great stretch; indeed, I am tempted to describe it as trivially true. Note immediately that the claim is not that all necessity is known empirically, with the notion of the a priori expunged from the picture; nor is it a rejection of the analytic–synthetic distinction. I am in fact totally against such draconian ideas.4 The claim is rather that the ground of all necessity is the objective world—not concepts or words or conventions. This, perhaps surprisingly, is readily established once certain habitual prejudices are put aside. Why is the proposition that water is H2O necessarily true? Because water, the objective stuff, has, as its essence, the chemical composition H2O: what it is to be water includes having that chemical composition. Being X consists in being Y. In this answer we refer only to water and its properties—not to the concept water, nor to the word “water.” Our talk is entirely de re. Water is a certain chemical way objectively, and that way constitutes its essence. There is nothing de dicto about it: water would be necessarily H2O even if there were no language or thought or even propositions. And that is now the commonly accepted view about water and other natural kinds. Now consider the proposition that a husband is a married man: that proposition is necessarily true. Why is it necessarily true? Because being a husband is constituted by being a man and being married. The property of being a husband comprises the property of being a man and the property of being married (as the property of being water includes the properties of being oxygen and hydrogen). Therefore, the property of being a husband cannot be instantiated unless the other two properties are instantiated, since they fix what a husband is. A conjunctive property cannot be instantiated unless its constituent conjuncts are. That is the reason why the proposition that a husband is a married man is a necessary truth. As it were, this is the very “composition” of 3.  I hope it is clear that I am discussing types of necessity and their ontological standing; I am not discussing the possibility of different scope readings of modal expressions in sentences. I do not wish to deny that “necessarily” can take different scopes with respect to quantifiers and descriptions, thus generating propositions with varying truth-value; and these scopes are commonly referred to as de re and de dicto readings of the modal expression. The terminology can be confusing, if it gives rise to the expectation that necessities themselves can be located either in the world or in language—which I am in the business of denying. It would probably be best simply to refer to wide and narrow scope readings and drop the talk of de re and de dicto readings. (Similar confusions surround the use of “de re” and “de dicto” in connection with belief: we must not confuse the possibility of different scope readings with the idea that beliefs themselves come in two ontological types, which I also doubt.) 4.  See my Truth by Analysis: Games, Names, and Philosophy (New York: Oxford University Press, 2011), especially chapter 6.

162

Part IV

the property of being a husband. In giving this straightforward answer we refer only to certain properties and to the instantiation and inclusion of those properties—there is no reference to concepts or words or conventions. Husbands would be necessarily male and married even if there were no concepts or words or even propositions. The answer is entirely de re. Being X consists in being Y. The point obviously generalizes: knowledge, say, necessarily includes belief and truth because of what knowledge is, independently of any concept we may have of it or any word we use for it—that is simply its de re essence. Being a game (to use a less familiar example) necessarily involves the use of inefficient means to achieve a goal, because of what games are in themselves, not because of how we describe them or conceptualize them.5 These kinds of analyses are simply and straightforwardly de re: they tell us the essence of a certain kind of thing—knowledge or games. They are not less de re than the usual examples of water, lightning, tables, and people. How we know the respective statements of essence to be true is altogether another matter: unlike the water and H2O type of case, they are known to be true a priori, by the operation we call “conceptual analysis.” But when it comes to the ground of the necessity itself, construed as a metaphysical notion—that resides in the nature of the kind or attribute itself, that is, what it is to be knowledge or a game or a husband (the things, not the concepts of the things). Analytic truths are necessary in virtue of the essences of the things to which they refer—and they refer to kinds, properties, and attributes. Water has a de re essence, but so does the attribute of being a bachelor: to be a bachelor is essentially to be an unmarried man—de re. Or again, it is a priori that the number 2 is even, and also necessary: in virtue of what is it necessary? It is necessary because 2, the number itself, has evenness in its essence—de re. It is not necessary because it is known a priori—to say that would be grossly to conflate the metaphysical with the epistemological. The necessity is de re just in the sense that the ground of it lies in the number 2 itself—just as the ground of the necessity of water being H2O lies in the nature of water. The propositions differ dramatically in their epistemic status, but the ontology of necessity remains uniform throughout: it is always de re. There could not be a world in which it is a fact that water is not H2O; and there could not be a world in which it is a fact that husbands are unmarried, or 2 is odd, or knowledge is false belief, or games use maximally efficient means to achieve a goal. All necessity is accordingly de re. Someone might protest: what about necessities of language and concepts? Surely they must be de dicto, not de re! The protest is confused. Let us by all means agree that words and concepts can be the bearers of necessity: the word “man” (in English) 5.  The definition of games I am citing is developed by Bernard Suits in The Grasshopper: Games, Life and Utopia (Peterborough, Ontario: Broadview Press, 2005; orig. published 1978). I discuss this definition in Truth by Analysis (see n. 4), in which I defend the idea of de re analysis.

Metaphysics163

is necessarily a noun and is necessarily shorter than “hippopotamus”; and the concept red is necessarily combinable in thought with shape concepts and is necessarily not possessed by blind people (pick other necessary truths about concepts if you don’t like these). But these are de re necessities too, where now the “res” is that part of the world comprising words and concepts. What makes these necessities hold is the nature of the entities in question—words or concepts—not our words or concepts about those entities. It is not propositions that refer to words and concepts that confer on these entities their essences; it is the objects referred to, considered in themselves. Language and thought can have de re essence as much as anything else. We don’t have a new type of necessity here, different in kind from that applicable to things like water or people, but simply a new subject matter for the usual kind of necessity—ordinary de re necessity. Kripke defines the concept of a rigid designator in modal terms: it is an expression designating the same object in all possible worlds. Given the equivalence of universal quantification over worlds and necessity, the concept of rigidity must then be explicable in terms of necessity. It is easy enough to see how this goes (though Kripke himself never explains the concept of rigidity in terms of a primitive modal operator, preferring the possible worlds formulation): “a” is a rigid designator of x if and only if necessarily “a” designates x, that is, it is not possible for “a” to designate anything other than x (in a semantically fixed language). Correspondingly, a term is a nonrigid designator if and only if it does not necessarily designate what it actually designates. The question is then: what kind of necessity is this? Kripke doesn’t address himself to the question, but his general stance would suggest that it is straightforward metaphysical necessity (it is certainly not epistemic necessity). Thus: it is metaphysically necessary that a name designate its actual bearer. In other words, it is part of the de re essence of the name (meaning what it does) that it should designate only what it actually designates. For example, it is part of the essence of (the meaning of) “Aristotle” that “Aristotle” could denote nothing but Aristotle—the name essentially denotes that particular individual. Any expression possibly denoting someone other than Aristotle, for example “the teacher of Alexander,” could not be the name “Aristotle” (as it now exists in English). Thus we can subsume Kripke’s notion of rigidity under his general conception of de re metaphysical necessity: this is a case of linguistic individual essence, in effect. (Any idea that we might define essence in terms of rigidity would therefore run in a circle, since rigidity simply is a type of essence—of language or meaning.) If all necessity is de re, from the a posteriori to the a priori, being uniform and unambiguous, then our knowledge of necessity will also be uniform in its objects. Here I am not speaking of our knowledge of the truth of a necessary proposition; I am speaking of our knowledge that such a proposition is necessary. That is, I am speaking of modal

164

Part IV

knowledge as such.6 Some necessary propositions are a priori and some are a posteriori, so we know their truth in these different ways; but our knowledge of their necessity might still be all of the same basic type. And I think it is: all knowledge of modality itself is a priori. This is as true for empirical necessities as it is for a priori necessities: for instance, we know that “Hesperus is Phosphorus” is necessarily true because of our general a priori grasp of the nature of identity and our a priori grasp of the rigidity of names. This comports well with the thesis that all necessity is de re, since knowledge of necessity will invariably take the same type of object—a de re necessity. All such knowledge involves insight into de re essence—seeing that something is an essential property of the thing in question (water, husbands, words, and so on). I do not say that such knowledge is unproblematic philosophically—it is not—but I do say that it is all of the same general type, namely, a priori knowledge of de re essence. Empirical Necessity and Conceptual Necessity If all necessity is de re, what becomes of the familiar distinction between empirical necessity and conceptual or a priori necessity? The answer should now be obvious: that is an epistemological distinction, not a metaphysical distinction (or a semantic distinction). There are not two distinct species of necessity, one residing in the world and the other conceptually based; there are just two ways of knowing the same kind of metaphysical fact. This follows immediately from a sharp distinction between epistemology and metaphysics: two ways of knowing do not entail two kinds of thing known. But it is also demonstrable from a consideration of specific cases. Consider (1) “Hesperus is Hesperus” and (2) “Hesperus is Phosphorus”: (1) is known a priori, while (2) is known a posteriori. Are there then two types of necessity here? No: the necessity of both rests on the same thing, namely the nature of identity and the semantics of names as rigid designators. Our modal knowledge of the necessity of both therefore derives from the same type of reasoning (and hence is a priori). But we know the truth of (1) and (2) by different means: in (1) by simple reflection on the meanings of the terms, in (2) by empirical investigation of the movements of the planets. It would simply be unwarranted reification to project the epistemological distinction onto metaphysical reality—supposing that two kinds of necessity are at work. That would be like supposing that the thing I touch has to be different from the thing I see because touch and sight are different epistemic faculties. It is the necessity of identity—a de re essence 6.  I discuss the distinction between knowing a necessary truth to be true and knowing it to be necessary in “A Priori and A Posteriori Knowledge,” reprinted in my Knowledge and Reality: Selected Essays (Oxford: Clarendon Press, 1999; orig. published 1976). Knowledge of necessity as such is always a priori, I contend, even when the necessary truth is known a posteriori. Of course, there are major philosophical issues surrounding the nature of modal knowledge, which I will not go into here.

Metaphysics165

of a particular relation—that constitutes the necessity of both (1) and (2); but the two propositions are contents of different types of knowledge. The modal fact is the same in both cases (the necessary identity of a certain planet with itself), but the two propositions represent it in different ways (under different “modes of presentation”). Both propositions are necessary in virtue of the same fact—the identity of the referents—but their epistemic status depends on something other than this common fact: it depends on how that reference is picked out. Reference fixes modal status; sense fixes epistemic status. We can even tease apart the epistemic and modal features of the two statements, showing how independent of each other they are. Consider (3) “It is necessary and a posteriori that Hesperus is Phosphorus”: something about the proposition that Hesperus is Phosphorus is responsible for its necessity and something is also responsible for its a posteriori status—but are these the same thing? No: it is the identity relation that accounts for the necessity and the different modes of presentation that account for the a posteriori status. What makes it necessary is not what makes it empirical. If we consider (4) “It is necessary and a priori that Hesperus is Hesperus,” we find the same modal component, but the epistemic component has changed. In fact, if we analyze the names as descriptions and then apply Russell’s theory of descriptions, thus generating scope ambiguities, we find that the modal operator applies to the same open sentence for both (3) and (4), while the epistemic operator attaches to a quite different part of the complex conjunctive proposition. We thus get for (3) something like this: “There is some object that is uniquely F and there is some object that is uniquely G, and it is a posteriori that the concept F refers to the same object as the concept G, and it is necessary that the former object is the latter.” In logical symbols, the last conjunct would consist of free variables within the scope of the modal operator, bound by the initial quantifiers. But we would have exactly the same final conjunct in a like Russellian paraphrase of (4), only now with the epistemic conjunct as a priori instead of a posteriori. In these compositional analyses, no single proposition is the object of both epistemic and modal operators. So, strictly speaking, no proposition is both necessary and a posteriori: once the Russellian analysis is carried out, with scopes duly assigned, different conjuncts bear necessity and a posteriori status. In logical form, nothing is both necessary and a posteriori. Thus the modal part of the truth conditions for (3) and (4) is identical and unaffected by the different epistemic status of the original unanalyzed proposition. This is just what we would expect, given that both (1) and (2) correspond to the same modal fact, that is, the self-identity of a particular planet. There are not two kinds of modality at work here, de re and de dicto; there are just two ways of knowing the same modal fact. In particular, it is not the case that in the a priori propositions the necessity switches its location from the world to meaning, so that the likes of (1) derive their necessity from something internal to the proposition, instead of from the de re

166

Part IV

essence of the identity relation. That would be, again, to confuse the modal status of the proposition with its epistemic status. Are There Essentially Empirical Essences? It may be felt that the relationship between the modal and the epistemic is stronger than I am allowing. And here, I think, we reach the rub: aren’t some essences essentially empirical, and some essentially a priori? Isn’t the necessity of water being H2O an inherently empirical necessity, as the necessity of husbands being married men is an inherently a priori necessity? It is, it may be said, not just contingent that we know the essence of water in an a posteriori fashion: it is of the essence of this essence that it be empirically known (and of the essence of the essence of husbands that this essence be known a priori). I think this view is widely held, if only implicitly: Kripke showed us that some essences are intrinsically empirical, not just accidentally empirical—so that the empirical status of these essences is embedded in them. The metaphysical thus bears ineluctably the imprint of the epistemological. So it is tempting to suppose, considering the actual status of our knowledge of essence. But it seems to me that on reflection this is quite wrong—for any empirical essence could have been known a priori. Consider a species of speaker with microscopequality eyes and an innate knowledge of chemistry: when introduced as children to the word “water” they know immediately and spontaneously that the liquid designated by “water” is H2O, just by looking. Thus linguistic mastery of “water” for them is accompanied by knowledge of what Locke would call real essence, as our introduction to “water” is accompanied by knowledge of nominal essence (the superficial properties of water). Then surely “Water is H2O” would be known a priori by these speakers, since the description “H2O” captures the sense of “water” for them (it is what they would say if you asked them what they mean by “water”). The mode of presentation that water has for them is simply the inner chemical composition of water, so this gets incorporated into the sense of the term. Maybe they are simultaneously ignorant of the nominal essence that we typically know, because their senses are only geared to fine-grained molecular properties, not phenomenal properties like transparency or absence of taste. Their commonsense or “folk” conception of water includes its molecular composition, so that predicating “H2O” of water is merely analytic for these super-perceptive speakers. The mode of presentation of water for them includes its molecular composition, so that the corresponding statement comes out analytic (yet the distribution of water on their planet may be an a posteriori matter, as may its phenomenal qualities as revealed to us). The lesson is that epistemic status is relative to epistemic faculties, not absolute and inherent. The truth of “Water is H2O” is therefore not essentially empirically known; it could be known in the way we know the truth of “Water is colorless and transparent”—that is, just by grasping the mode of presentation typically associated with

Metaphysics167

“water.” There are really no empirical essences in the absolute sense: the epistemology is not imprinted on the metaphysics, but only contingently tied to it. I could have made the same point more simply by reference to God: God’s knowledge of the essence of water is not like ours—an arduous a posteriori discovery of real essence, preceded by a commonsense apprehension of nominal essence. God sees directly into the inner nature of water, so “Water is H2O” is analytic and a priori for him.7 I didn’t make the point in this way only because wheeling in God’s knowledge is apt to excite philosophical objection—but it does serve to make the point vivid. The underlying principle is that the epistemic status of a truth as a priori or a posteriori depends on the epistemic powers of the knower in question, but its modal status does not, being entirely nonrelative. Perhaps even more surprisingly, the same kind of relativity point can be made with respect to necessities that are a priori for us. Take Bernard Suits’s definition of a game as a rule-governed activity in which the rules proscribe the use of maximally efficient means for securing a goal (like putting golf balls into the holes by hand). Suppose some alien radical interpreters are trying to translate our language into theirs, but have no games in their society and no prior grasp of the concept—game is a concept quite foreign to them. They are working on interpreting our word “game” and are starting to grasp a bit about its extension, noting that games are rule-governed intentional activities—but they don’t grasp the crucial point about inefficient means yet. They bandy the word “game” about (or their purported translation of it), making various true statements about games. Then, further down the line, some bright alien interpreter finally figures out that games are a special subclass of rule-governed activity and adds the crucial clause about restricted means to the definition-in-progress. They have made what is for them an empirical discovery about games, relative to their initial grasp of the concept—finally coming to know the real essence of games. In effect, “game” has been functioning as a natural kind term for them, with some known superficial marks of the kind, but an undiscovered real essence. What for us is a priori 7.  Kripke comes close to appreciating this point in footnote 11 of Naming and Necessity, in which he asks whether someone’s knowledge that he is giving these lectures must be a posteriori. He remarks: “But, if someone were born with an innate belief that I was going to give these lectures today, who knows?” (40). Here he seems to be envisaging the possibility that the fact in question might be knowable a priori: we just need to make suitable alterations in the epistemic powers of the knowing subject—the fact itself does not dictate the manner of knowing it. Another problem in all this, of course, is the relative obscurity of the very notion of the a priori—what exactly counts as a priori knowledge. What is clear is that someone could in principle know any ordinary fact without using his senses to do so—the knowledge could just be directly implanted in his brain. The senses are one route to knowledge; they do not determine the very nature of the fact known, or indeed the knowing of it. (Here we see a major flaw in traditional empiricism: total across-the-board nativism cannot be ruled out as logically impossible.)

168

Part IV

is for them a posteriori. (Similarly, what is a priori for my super-perceptive speakers with respect to “water” is a posteriori for us: we don’t immediately grasp what they mean by “water” if we don’t already know that water is H2O.) Obviously, the same kind of story could be told about “husband,” “bachelor,” and so on. Kant thought that “bodies are extended” is a priori and analytic, but can’t we imagine beings that are unaware of this property of matter, having no extensionrevealing senses and poor geometrical intelligence? They could introduce the term “body” in a natural kind spirit, hazily and ignorantly, and then later discover that bodies are extended. The fact that bodies are extended is not in itself an a priori fact; it is a priori only relative to the knowledge of the speakers who express the fact in language and thought—to their particular conceptions of things. In Locke’s terminology, real essences could become nominal essences and nominal essences could become real essences, simply by altering epistemic powers. No essence is inherently real or nominal, empirical or a priori. Metaphysically speaking, there are no empirical essences or a priori essences. There are just de re essences that are contingently of one epistemic status or the other. All we have are de re essences taking on different epistemic values relative to one endowment of cognitive powers or another. And this is entirely what we would expect if we had taken fully to heart the strict separation of epistemology and metaphysics with respect to modality upon which Kripke insisted. There Are No Empirical Facts We can now take the separation thesis one stage further, to assert that no facts of any kind are inherently empirical or a priori. For any fact expressed by an empirical proposition, there is an a priori proposition expressing the same fact; and the same can be said, mutatis mutandis, for any a priori proposition expressing a fact. Again, this claim may seem surprising, so deep is the tendency to conflate epistemology and metaphysics, but actually it is not so difficult to establish. Kripke offers a quick way to see the point in his notion of reference-fixing: “London is the capital of England” can be a posteriori for one person, but a priori for another, depending on how they understand the name. If you introduce the name by means of that description, then the resulting sentence will be known by you to be true a priori; but if the name is not so introduced, it can be an empirical discovery that London is the capital of England. Basically, it is just a matter of describing the subject by means of the predicate, thus producing a kind of analytic truth. Is “the earth orbits the sun” analytic or synthetic? You may reply that it is obviously synthetic, and I agree: but what if we define “the earth” as “the blue planet that orbits the sun”? Then the sentence will come out analytic. It seems that we can perform this trick for almost any empirically known sentence— for instance, we rephrase “Gravity obeys an inverse square law” as “The attractive force that obeys an inverse square law obeys an inverse square law.” The language-

Metaphysics169

independent fact remains the same, but the proposition varies in epistemic value. If we know enough about the world, we can generate these kinds of paraphrases at will. And remember: every fact that is empirical for us is a priori for an omniscient being like God, or a being that knows everything innately. We can also contrive the same kind of epistemic status switch for facts that we typically regard as a priori, say mathematical facts. Take “2 + 2 = 4,” which is as a priori as one could wish. Now redescribe the numbers referred to as follows: “The number of my cats plus the number of my cars equals the number of my kayaks” (I have two cats, two cars, and four kayaks). Then, the proposition goes from a priori to a posteriori (and from necessary to contingent). You just pick your descriptions at will and shift the epistemic value. The fact about numbers (the mathematical objects and their relations) is not an inherently a priori fact; what makes the sentence a priori is the particular proposition it expresses—sense not reference, mode of presentation not object presented. The very concept of an a priori fact (or an a posteriori fact) is a confusion of epistemology and metaphysics. If I am right, virtually any fact can in principle be known either by a priori means or by empirical means (if not by us then by some possible knower), depending on the proposition we choose to represent that fact. But even if that were not so, it would still be confused to suppose that the world divides into two sorts of fact, as an ontological matter; rather, we have two cognitive faculties that make us know facts in two different ways. Again, God helps us understand: do you think that God has any time for a distinction between a priori facts and a posteriori facts? He may well have time for a distinction between mathematical facts and physical facts, or abstract facts and concrete facts; but he regards talk of a priori and empirical facts, interpreted ontologically, as mere projective reification. We can talk of facts being known a priori or known a posteriori, so long as we are aware of the relativity of such knowledge; but it is a mistake to foist these different modes of epistemic access onto the nature of the fact itself. When giving an inventory of the world, we supply “just the facts”; we don’t divide them up according to how they are contingently and extrinsically known by us—as if the objective world respected our ways of knowing about it. Facts, considered in and of themselves, are epistemically neutral. Of course, there is nothing wrong with using the phrase “empirical fact” to mean “fact that happens contingently to be known empirically by beings of a certain type,” so that the self-same fact can also be said to be an a priori fact: but I think it is clear that the way philosophers have used the former phrase embodies a far greater commitment to ontological significance than such a relativistic paraphrase suggests.8 8.  In an unpublished essay entitled “Ontology without Epistemology” I ask whether it is possible to fashion an objective correlate for the traditional distinction between a priori and a posteriori truths—something ontological that corresponds to the epistemological distinction. I suggest that this is possible, using the notions of event and change. (Let me also remark that there is some

170

Part IV

Epistemic Necessity Someone who has followed me so far might concede that all metaphysical necessity is uniformly de re, and accept the strict separation of epistemology and ontology upon which I am insisting, but protest that it must be wrong, grotesquely so, to assimilate epistemic necessity to de re metaphysical necessity. Isn’t the distinction between epistemic and metaphysical necessity one of the principal and irrevocable achievements of Naming and Necessity? And isn’t it just a plain fact of the English language that “might” is ambiguous between an ontological and an epistemic reading? Surely, here, we find two genuinely distinct kinds of necessity! For how else are we to explain the fact that “Water is H2O” is both metaphysically necessary and epistemically contingent? If modal words had no ambiguity, this would be a simple contradiction; but evidently it is not. Now I thoroughly sympathize with the thought behind these sentiments, but I reject the suggestion that what is true here requires the multiplication of kinds of necessity or any kind of ambiguity claim. Once more, there is nothing afoot except ordinary univocal de re metaphysical necessity. Again, this may seem like an outrageous claim, but actually it is easy to establish, and is already implicit in Kripke’s own text. It is true that Kripke appears to commit himself to an ambiguity thesis in such remarks as this: “Obviously, the ‘might’ here is purely ‘epistemic’—it merely expresses our present state of ignorance, or uncertainty” (103). But when he goes on to explain what is meant by this kind of “might,” the picture changes. He says: And so it’s true that given the evidence that someone has antecedent to his empirical investigation, he can be placed in a sense in exactly the same situation, that is a qualitatively identical epistemic situation, and call two heavenly bodies ‘Hesperus’ and ‘Phosphorus’, without their being identical … being put in a situation where we have exactly the same evidence, qualitatively speaking, it could have turned out that Hesperus was not Phosphorus; that is, in a counterfactual world in which ‘Hesperus’ and ‘Phosphorus’ were not used in the way that we use them, as names of this planet, but as names of some other objects, one could have had [my italics] qualitatively identical evidence and concluded that ‘Hesperus’ and ‘Phosphorus’ name two different objects. … We could have [again, my italics] evidence qualitatively indistinguishable from the evidence we have and determine the reference of the two names by the position of two planets in the sky, without the planets being the same. (103–104)

Kripke is here analyzing what it is for something to be epistemically possible, and he employs the apparatus of a qualitatively identical evidential situation combined with

tension between the position advanced in this essay and the position taken in “Knowledge and Truth” in this book. The latter essay was written some time after the present one and reconsiders the question of how a priori truth and a priori knowledge are connected. The question is indeed a difficult one, and both sides can offer persuasive considerations.)

Metaphysics171

a modal condition: one could be in an identical evidential situation and it not be true that “Hesperus” and “Phosphorus” co-refer. I italicized the modal expressions in the quotation from Kripke to highlight the appeal to a modality in analyzing epistemic possibility. What kind of modality is this? It had better not be epistemic modality or else the analysis is completely circular, and anyway it clearly is not. It is regular good old metaphysical modality. Not that Kripke says as much explicitly, any more than he says explicitly that the modality used to define rigid designation is metaphysical modality. But the entire thrust of the analysis makes it plain that this must be what he intends: that is, spelling it out, there is a metaphysically possible world in which the same evidence we have in the actual world at a given time is coupled with “Hesperus” and “Phosphorus” not having the same reference in that world. In other words, the evidence we have does not necessitate the truth of the identity statement—it is possible for the evidence to obtain but for the corresponding statement not to be true. Kripke is, in effect, analyzing so-called epistemic modality in terms of the concept of evidential identity and metaphysical modality. It is metaphysically possible to have such and such evidence and a “qualitatively identical” statement not to be true. Accordingly, this is simply another case of de re metaphysical modality doing its routine work: it is not part of the de re essence of our evidence that it guarantee the truth of our identity claims. Our evidence is only contingently coupled with the truth of the identity statement we infer from it. Epistemic contingency is the same thing as the metaphysically contingent relation between evidence and conclusion.9 Similarly, in the case of epistemic necessity, the analysis is that it is in the essence of our premises that they guarantee the truth of what we assert on their basis: the premises could not hold without the conclusion being true—they necessitate the conclusion. There is no metaphysically possible world in which the premises are true and the conclusion is not true—this is a de re necessity concerning the premises. To illustrate the point, consider Descartes’s cogito. Descartes claims that I can be certain that I exist—the proposition that I exist is an epistemic necessity. (It is clearly not a metaphysical necessity, since it is not a metaphysically necessary truth that I exist.) Why is my existence epistemically necessary, according to Kripke’s analysis of that notion? Because the holding of the premises for that belief is inconsistent with its falsity (and the premises themselves are certain): I can be certain that I am thinking, and my existence follows (for Descartes) from my thinking. The crucial point is that the conclusion logically follows from the premises (and the premises are certain), so that in all worlds if the premises hold so does the conclusion (this is a deduction). But this notion of entailment 9.  I am only discussing Kripke’s conception of epistemic necessity and possibility here, not others that might be proposed. We could use these phrases to denote simply certainty and its lack, with certainty defined without invoking metaphysical modality—assuming this to be possible. But then there are not two kinds of necessity in any interesting sense, just necessity and certainty.

172

Part IV

simply is a kind of metaphysical necessity, expressing an essential property of a particular premise (just as it is an essential property of a conjunction that its conjuncts are true if it is). Thus the notion of epistemic necessity is analyzed by using metaphysical necessity. Given my evidence, I couldn’t now be in a world in which I don’t exist (though I could be in a world in which there is no external world). So it is not that we need to recognize two sorts of modal operators, which are logically independent of each other; rather, one sort suffices to define the other sort (with some supplementary non-modal apparatus). No ambiguity need be posited. I can put the point succinctly as follows: while Hesperus could not be anything other than identical to Phosphorus, it could have a counterpart planet that was not identical to Phosphorus—where the word “could” here is used in exactly the same sense at both occurrences. Hesperus itself must be identical to Phosphorus, but it is not the case that epistemic counterparts to Hesperus must be identical to Phosphorus— where again the modal words are used purely in the metaphysical sense throughout (there being no other). In sum, so-called epistemic modality reduces to metaphysical modality, and metaphysical modality (as we have seen) is uniformly de re modality. The only (irreducible) modalities that exist are de re metaphysical modalities. There is no such thing as epistemic modality—only metaphysical modality combining with notions of evidence and entailment. There are not two ways of something being necessary, metaphysical and epistemic; talk of epistemic necessity is just a way to register certain epistemological facts—the modal component derives entirely from a particular application of metaphysical modality. It would aid the cause of clarity to ban the language of “epistemic necessity” and “epistemic possibility” as encouraging misleading ideas, and speak instead of certainty and fallibility, which are frankly epistemological notions. Ordinary talk involving an epistemic use of “necessarily,” as in “What you are saying is not necessarily true,” intended to convey the message that the speaker’s evidence does not entitle him to his conclusion, can be readily parsed in terms of the uniform notion of metaphysical necessity—as in “your premises don’t necessitate your conclusion.” Theoretical parsimony suggests such a course, as does Kripke’s plausible analysis of the import of claims of (so-called) epistemic possibility; and it is the natural way to interpret all uses of modal terms to convey an epistemic point. Other Modalities Is monism about modality able to handle absolutely all modalities? Here I will briefly consider nomological modality and logical modality. According to some views, tentatively endorsed by Kripke, nomological necessities are metaphysical necessities, at least in certain cases; then the assimilation is automatic. Maybe nothing could be gravity (that force) unless it obeyed an inverse square law. But without committing ourselves to that strong thesis, we can still declare that nomological necessity is a species of de re necessity: objects, properties, and forces have “nomological essences”—ways

Metaphysics173

they have to be or behave given the laws of nature. These essences are not de dicto by anyone’s standards, since laws of nature are not a priori or analytic. The thesis that all nomological necessity is de re, not de dicto, is not likely to provoke serious dissent. Nomological necessity may not be identical to metaphysical necessity, but it is certainly not less de re than the latter—not less a matter of how the world is independently of thought and language. Indeed, if we had begun our investigation of modality with nomological necessity, we would have been primed (prejudices apart) to accept that all metaphysical modality is de re. That paradigm would have suggested a more general lesson: the idea of some distinctive type of de dicto necessity would have seemed freakish and anomalous—as would the conflation of necessity with apriority. Nomological necessity is, we might say, aggressively de re—not shy about its objective nonlinguistic status.10 Logical necessity presents a more troubling case. By “logical necessity” here I mean the narrow notion of a class of necessary truths made necessary by certain expressions labeled “logical constants.” Let us restrict the logical constants to the standard quantifiers “all” and “some” and the connectives “and” and “not.” There are certainly many formulas (infinitely many) that are necessarily true by virtue of just these expressions, as with the standard tautologies of predicate logic. We can say, following tradition, that such formulas are necessarily true “in virtue of their logical form.” Then what is the res on which this kind of necessity depends? In the case of “Water is H2O” the res is water, and in the case of “games use inefficient means” the res is games; but what is the res that corresponds to the logical constants? They don’t, apparently, denote anything that could function as the res: the necessities they generate seem to arise from their meaning not their reference. So are these narrowly logical necessities, at least, genuinely de dicto? If so, the general picture of necessity that I am advocating would be spoiled by this one awkward exception. Such is life, one might say. However, it is not compulsory to despair of finding theoretical unity, because we can appeal to de re locutions in characterizing even this kind of necessity. Thus we can speak as follows: It is in the nature of conjunction that if two propositions are combined by conjunction into a truth, then each of the conjoined propositions is true. It is in the nature of negation that applying negation twice to a proposition is equivalent to the proposition operated on. It is in the nature of universality that if everything is F, then some particular thing is F. It is in the nature of individuality that if an individual is F, then something is F. That is, we speak of the essential properties of such things as conjunction, negation, universality, and individuality. To fix ideas, these things can be 10. I am not worried about different strengths of necessity, with the nomological kind weaker than the metaphysical/logical kind, only about the idea that necessity can be grounded in propositions rather than facts (to put it simply). What must be so can be more or less stringent, but it always has its roots in what we talk about, not in how we talk about it.

174

Part IV

construed as something like Fregean functions, so that logical constants (the expressions) have both sense and reference, like every other expression. Then we are asserting that these functions have certain de re essences concerning their impact on the truth of propositions. We simply wheel in a custom-made ontology to play the role of the res we need. I am actually quite receptive to this move, being a bit of an ontological liberal, but I know that many people will regard the move as unappealing and ad hoc. For the ontologically fainthearted, then, I offer an alternative that still secures the broad result we are aiming for: regard logical necessities as essentially metalinguistic, making explicit reference to words or meanings, and then take those entities to be the res we seek. This is really what the positivists suggested quite generally: all necessary propositions are to be analyzed as being about meanings or linguistic entities. For instance, for conjunction elimination we say: the meaning or sense of “and” is necessarily such that if “p and q” is true, then “p” is true and “q” is true—that is, the de re essence of the meaning or sense of “and” is such as to generate those entailments. Semantically speaking, this is like analyzing the sentence “five necessarily has four letters” as “the word ‘five’ necessarily has four letters,” where we straighten out a use-mention error and redirect the necessity to its proper place, that is, in the region of words for numbers, not numbers. What we must not allow, to prevent any resurgence of the discredited notion of de dicto necessity, is that logically necessarily formulas can be necessary in virtue of something other than what they are about—their expressed sense as distinct from their explicit reference. And metalinguistic statements are precisely about words, or the meanings of words.11 Thus we can always find a suitable entity, even for logical necessities, to be the bearer of the essence we need. Necessity accordingly always arises from the de re essence of what is referred to. The Metaphysical Picture In conclusion, let us step back from the details and focus on the broader picture. I view it as Aristotelian in tendency.12 The world consists of a great variety of things 11.  If the positivists were to take this line in general, they would be assimilating the essence of references to the essence of words or meanings/concepts. That would be not to eliminate de re necessity but to acknowledge it at another level. But what is gained philosophically by denying that objects have essences while accepting that concepts do? And, of course, it is massively implausible to try to reduce essences of objects to essences of concepts—as if the necessity of my cat Lucy being a cat is the same as the necessity of my concept of Lucy including the concept cat. Cats are necessarily cats even if there are no people with concepts of them. 12.  In recent work the author who seems to me to express this metaphysical picture most clearly is Kit Fine: see his “Essence and Modality,” Philosophical Perspectives 8 (1994): 1–16. Our views do not coincide at every point, however.

Metaphysics175

with specific essences: particles, people, animal species, chemical kinds, human activities, psychological states, institutions, numbers, meanings, words, and whatnot. These things have their essence—their “what-it-is-to-be”—independently of our representations of them (unless they are themselves representational entities). Propositions or statements can express or describe these essences—from the outside, so to speak. These propositions can be known in different ways, notably empirically or a priori. But the knowledge is not intrinsic to the essence itself; it comes after the essence, not as a precondition of it. Essences are not propositional creatures: they do not themselves incorporate any mode of presentation or sense—they belong to the realm of reference or fact. (This is true even when the entity in question is itself a sense or meaning or word.) To fully specify an essence you never need to select a description under which the essence holds. Essences are not proposition relative. But knowledge is always proposition relative, always under a description, never irreducibly de re. Even when an ascription of knowledge takes a de re form, with a term outside the scope of the epistemic operator, there always exists a full propositional dictum that captures the subject’s state of knowledge. You cannot fully specify someone’s state of knowledge without specifying a proposition. Thus necessity and knowledge belong at different ontological levels: essence is purely de re, but knowledge is inherently de dicto. It is therefore a kind of category mistake to try to reduce necessity to knowledge. The dualism of knowledge and necessity is complete, sharp, and unwavering. A fact is what is necessary; that fact may be expressed by a proposition; that proposition may become known: but the knowledge is strictly posterior to the fact. Divisions between types of knowledge are irrelevant to the way necessities are distributed in the world. Necessities are ontologically uniform, being everywhere de re, even though there is pronounced epistemic variety in our knowledge of them. The world is replete with necessities, but none of them is constituted by human cognition or representation (unless they are necessities of cognitions and representations). In no sense, then, is necessity a product of the conceiving mind— even when the necessity is “conceptual” (recall my treatment of analytic truth above). Knowledge is no more internally connected to modal facts than it is to any other kind of fact. When God had created all the necessities in the world, he still had not got started on creating mind and language. In fact, at this stage, the epistemological distinction between a priori and a posteriori knowledge had not even crossed God’s mind. Necessity concerns the way things have to be, not how humans happen to conceptualize the world or represent it in thought or language.

Antirealism Refuted

How would we talk if antirealism were true and we knew it to be true? We would talk as if it were true, presumably. For example, we would utter hypothetical statements about sense experience, not categorical statements about material objects, assuming we knew phenomenalism to be true. We would speak of dispositions to behavior, not internal mental states, if we accepted behaviorism. We would refrain from reference to elementary particles, if we thought that there were no such particles to refer to. We would speak only of words and other symbols if we thought, as convinced nominalists, that no abstract entities exist (numbers, universals). We would restrict ourselves to overt expressions of emotion if we rejected the idea of moral values as objective entities, saying, “I approve of generosity,” not “Generosity is good.” That is, if we were genuine antirealists, as a matter of unreflective common sense, and had been forever, we would talk in the indicated manner. We would not talk misleadingly, as if realism were true, but accurately, reflecting our antirealist convictions. We would talk in the way we now talk about things that we are antirealist about—witches, ghosts, the ether, and the gods. But that is not the way we actually talk. We talk as if realism were true: as if material objects were independent of experiences, as if mental states lie behind and cause behavior, as if elementary particles were tiny invisible bits of matter, as if numbers were different from numerals, as if moral values exist independently of human emotions. That is why antirealism is always understood as a revisionary doctrine, not a purely descriptive one. We are natural realists—naïve realists, in the usual phrase. The antirealist suggests that our normal and spontaneous realism is mistaken—so that we must change our views, and even our language. The antirealist therefore sees himself as a critic of our ordinary ways of thinking, as they are expressed in our ordinary language. Thus we would (and should) speak and think differently, once we embrace the antirealist’s position. This means that antirealism is always an error theory: there is some sort of mistake or distortion or sloppiness embedded in our usual discourse. The antirealist about witches finds error in the discourse of those who speak uncritically of witches, and the antirealist about material objects finds error in the notion that objects are distinct from sense experiences. Hence antirealism is felt as surprising and disturbing. It would not be felt in that way if we were habitual antirealists from birth till death. There would be no need to urge antirealism on us if we already accepted it: in that situation it is realism that would be perceived as revisionary. But if antirealism is always an error theory, then it must account for the error. Why we do we make mistakes about ontological matters? Human error can arise in a number of different ways: perceptual illusion, indoctrination, prejudice, carelessness, random

178

Part IV

interference, and so on. Thus we can explain errors in astronomy by perceptual illusions, errors in politics by prejudice, errors in morality by indoctrination. There are no inexplicable errors—errors that come from nowhere, for no reason, even if it is just random neural firings that are responsible. Much human error is temporary and quickly corrected, as with simple errors of fact, for example, errors about the time of day, though some may take decades or centuries to be rectified. In all cases the error has some kind of intelligible explanation. But what is the antirealist’s explanation of the errors that she detects? On the face of it, none—she has no explanation. She supposes that human beings have made enormous metaphysical errors, persisting over millennia, which have not been corrected in the usual ways: but nothing much is said about how such errors might have arisen. And the usual kinds of explanation for errors don’t seem to apply: no perceptual illusions or indoctrination or prejudice or hastiness. Many people have no doubt been browbeaten into accepting certain erroneous moral attitudes—at school, in church, and in the home—but surely no one has ever indoctrinated a child into being a moral realist or a perceptual realist or a Platonic realist (or if they have, it would be very rare). We don’t accept these realist positions because we have been coerced into them at an early age, still less because we are subject to perceptual illusions that suggest them; we just find ourselves holding realist opinions. We are not victims of relentless realist propaganda or a misfiring of the senses, being pushed toward a realist position we would naturally reject. So why do we commit the errors attributed to us by the antirealist? Some have suggested that ordinary language is to blame. Our perception of language is misleading as to its true nature—or some such. It is as if we gaze languidly at language and it actively produces metaphysical illusion in us—the illusion that realism is true. Thus it might be said that moral words look a lot like words for material objects, so we transfer realism from the latter to the former. But that would assume realism for material objects—so how do we explain the error that antirealism detects in that area? Also, this kind of error theory is surely massively implausible: how could we be so easily bamboozled by the surface forms of our language? Why did no one point out the illusion centuries ago? Isn’t it just silly to suppose that the subject matter of a piece of discourse should mirror the syntax of the discourse itself? Is it really remotely plausible to suppose that our habitual realism is the result of committing bizarre non sequiturs from language to reality? And why is language so defective to begin with, given the truth of antirealism? Would it be reasonable to claim that people believed in witches because of the way the word “witch” looks? Ordinary language, as we normally experience it, just doesn’t have the power to generate the kind of large-scale metaphysical error that the antirealist alleges. So it appears that (a) antirealism is an error theory and (b) it has no workable theory of error. Realism, by contrast, is not an error theory, and can simply claim that our

Metaphysics179

commitment to it reflects the truth. If antirealism has no explanation of the error it imputes, and if no such theory can be plausibly produced, then it must be itself erroneous. We thus have good reason to reject it. More strongly, antirealism, insofar as it is an error theory, is a false theory—there is no such error in our ordinary thought and talk. Hence we should accept realism; and not just realism in this area or that, but realism across the board, since the problem with antirealism is general and systematic. At any rate, the antirealist must meet the challenge to account for the error she imputes, which is none too easy a thing to do.

The Puzzle of Paradox

Paradoxes are puzzling, but it is also puzzling why there are any paradoxes to begin with. Certain concepts lead to paradox: the concept of truth (the semantic paradoxes), the concept of a set (Russell’s paradox), any vague concept (the Sorites paradox). These concepts do not strike us as paradoxical on their face—indeed, they seem as solid as any concept we have—but they generate paradoxical results. Nor has it been easy to resolve such paradoxes; they are deep and stubborn, not the result of some simple slip of reasoning. Why do we have such paradoxical concepts? What is the source of paradox? The case is not like that of mystery: we have mysteries too, but it is usually easy to see how they arise—from limits on human knowledge. But paradoxes are not merely gaps in our knowledge; they are intellectual catastrophes, contradictions in how we think. A paradox is an affront to the intellect, a slap in the face: there shouldn’t be any paradox. We accordingly feel the need to get rid of paradoxes. So how do they arise—what causes them? If we think of them as like a disease, then where does the disease originate? One possible theory would be that they come from the world: the world is a paradoxical place, home to intellectual monsters. Thus truth, sets, and vague properties are objective mind-independent features of things, and they are inherently paradoxical. Similarly, it might be said that reality contains other paradoxical or absurd things: free will, quantum events, action-at-a-distance, moral values, and so on. All of these raise difficult problems of understanding and possibility; they are all “queer.” Reality, according to this view, is inherently bizarre, even incomprehensible—like a box of magic tricks. Paradoxes are just one more absurdity in an absurd world: reality is paradoxical de re. It’s a mad, mad, mad world! This view is hard to swallow: reality surely cannot be paradoxical or “queer” in itself—it cannot be the origin of our conceptual troubles. How could reality objectively contain impossibilities? Impossibilities cannot be! There couldn’t be a set that was both a member of itself and not a member of itself, or a proposition that was both true and not true, or a man who was bald and not bald. Far more attractive is the idea that paradoxes have their source in our concepts. It is the way we think of reality that causes paradox. Reality itself exists nonparadoxically, but we bring paradox to it by representing it in various ways. The fault lies in us, not in the world. The disease is self-generated. Our concepts are defective, crooked in some way, in need of reform. Once we reconceptualize the world, the paradoxes will melt away. Perhaps it is our language that leads us astray—in which case we just need to fix it. Our language is sloppy and undisciplined, and it gives rise to paradox—as it gives rise to other metaphysical monsters (as with Meinong’s ontology). Thus we will be advised: no self-reference, no hierarchy violations, no vague predicates! But this endogenous theory of the source of paradox encounters its own objections, for it seems both shallow and overly optimistic. It is shallow because it is hard to believe that the paradoxes

182

Part IV

go no deeper than some sort of conceptual muddle or deficit: rather, they seem to grow from the very nature of truth, sets, and vagueness. Any expression of truth, sets, or vague properties will generate paradox. And it is overly optimistic because it should be possible to rid ourselves of the paradoxes by the appropriate conceptual revision: we just have to think differently about truth, sets, and vagueness. But that is precisely what is so difficult: the things in question resist being rethought into coherence—they remain stubbornly paradoxical. It’s not like simply applying Russell’s theory of descriptions to get rid of Meinong’s ontology, or realizing that “no one” is not a singular term. And why exactly should our conceptual scheme harbor such monsters? Is it some gratuitous human peculiarity? Would Martians with a superior conceptual scheme encounter no paradoxes concerning truth, sets, and vagueness? Why don’t we already have a nonparadoxical conceptual scheme? Why do we make things so difficult for ourselves? No, it is hard to avoid the conclusion that the paradoxes do not spring merely from our contingent concepts, considered as psychological attributes, but from what these concepts are about: truth, sets, and vagueness are inherently paradoxical—paradox is in their nature. But then we are back with the mad mad world theory, with paradox stalking objective reality itself. Notice how the issue as so far articulated presupposes a deeply entrenched dualism of mind and world. Either paradox exists in the world or it exists in the mind: the disease is either “out there” or “in here.” The origin of the problem is either inner or outer—these being the only two possibilities. But such a dualism is not unquestionable, and indeed has some obvious weaknesses. What do we mean by “mind” such that it can produce a single notion of mind independence? And what is “the world”? Is there really a single unitary thing denoted by that phrase? Maybe we do better to recognize that there are many things we call “mental” and many things we assign to “the world.” We should be pluralists, not dualists. But how would this help with the puzzle of the paradoxes? Well, consider the idea of “the fictional world”—the world of made-up stories. We don’t want to say this is a “mental world,” so that Sherlock Holmes is a mental entity: he is rather the object of human thought. Nor do we want to say that the fictional world is part of “the real world.” So it seems not wrong and unmotivated to speak of two worlds here: the fictional world and the real world. And it is noteworthy that the fictional world is tolerant of contradiction: fictional characters can have contradictory attributes, if they are so described. The fictional world is not as constrained as the real world. In fact, there can be paradox in the fictional world—in literature (e.g., time travel) and in painting (e.g., Escher drawings). Fiction is hospitable to paradox, whereas fact is not. How does this help? It helps because it gives us the possibility of a third option: paradox belongs to a third realm, arising neither from our concepts nor from objective reality. The obvious model here is Kant’s distinction between the phenomenal world and the noumenal world. The former world is supposed to correspond to the world

Metaphysics183

we experience, and to possess objectivity; it is not merely the reified contents of our mind. The latter world is conceived as reality as it exists independently of human (or any other) subjectivity. Then this (taking a deep breath) is what I want to suggest, or at least play with: in the noumenal world there is no truth, there are no sets, and there is no vagueness; but in the phenomenal world there are such things. In a certain sense, truth, sets, and vagueness do not exist objectively, but they do belong to the familiar world revealed in our experience. So objective reality, in the purest sense, is not inherently paradoxical, but the human world (or that of any other intelligent species) is inherently paradoxical. Yes, I know it sounds pretty wacky, but let’s see whether it can be defended. It certainly seems nicely designed to resolve our problem, but is the cost too high? David Lewis multiplied existent worlds in order to solve philosophical problems (inviting those “incredulous stares”); I want to explore the possibility of a solution to our problem that adds one extra world—this one being actual as well as existent. Kant’s dualism of worlds will be my inspiration. The thesis, then, is that the noumenal world—the most austerely objective world— contains neither truth nor sets nor vagueness. We need not regard this world as totally unknowable, though our knowledge of it may be sketchy and limited. It contains nothing geared toward human minds or any other form of sensibility: any natural kinds it contains are completely independent of human interests or the human sense of similarity. I submit that no one is bald in that world, nor is anyone possessed of a fine head of hair. These are human classifications, reflecting our interests and perceptual saliencies. There may be hairs on heads in specific numbers (though even these categories reflect human interests), but there is no such property as baldness (or its opposite). We could think of this world, for concreteness, as “strings-in-the-void,” conceived purely mathematically. In the case of truth this is a concept that belongs with human belief and propositions; it is not part of reality considered independently of rational agents. There was no such thing as truth until creatures with beliefs came along; there were just facts—things with properties. Truth is part of the human Umwelt. As to sets, they enter our thought by way of the operation of collection—as when we collect several objects into one group. The concept of set is a Gestalt concept. In the stripped-down noumenal world, I surmise, there is no collecting and no collections: there are no sets there, just assemblages or distributions. So there is nothing potentially paradoxical in the noumenal world, the world as it exists independently of any form of perceptual or cognitive sensibility (a world we find it difficult to conceive). None of its intrinsic properties may even be imaginable by us—we may have only the vaguest and most abstract idea of what it is like. This is the furthest reach of objectivity, possibly beyond human conception. But it is fundamental reality—and it contains nothing paradoxical at all. But there is another parallel world—the one that we experience and conceive. This world contains many properties not found in the other world, such as colors, Euclidian

184

Part IV

shapes, values, social properties, and many others. Among these we find the property of being a set, the property of being true, and the property of being bald. These are real properties, not figments, because the phenomenal world is real—it is not a dream or a hallucination. It is not a piece of fiction, but a realm of fact. It is not produced or constituted by our minds. How exactly it is related to our minds is a difficult question (with which Kant grappled), but it is to be viewed as somehow commensurate with our minds. It is the world as experienced. Then the suggestion is that it is in this world that paradox lurks: the paradoxes arise from genuine properties instantiated in the phenomenal world. It is not that this world actually contains contradictions—it is not like the fictional world in that respect—but rather that it contains properties that naturally generate paradox. It is not that it is merely our contingent human concepts that produce paradox, so that we could remedy the situation by suitable conceptual change. The properties themselves produce paradox. Being bald, say, is a genuinely paradoxical property, as are truth and set-hood. These properties inevitably lead to paradox; it is not just the way we happen to conceive of them that does (or some muddle we have got ourselves into). But that doesn’t mean that the objective fundamental noumenal world is paradoxical, because these properties don’t exist in that world. So there is a paradoxical world and a nonparadoxical world. The source of paradox is one of these worlds (i.e., the properties instantiated there), not human concepts (or the concepts of other rational beings). We could then say that the paradoxes are objectively based without being ontologically fundamental. None of this is intended to make the paradoxes go away or render them any more comfortable. The idea is just to find a way between two unpalatable extremes: on the one hand, that paradox characterizes the very nature of ultimate reality; on the other hand, that paradox arises purely from within our conceptual scheme. Paradox arises, instead, at an intermediate level, analogous to Kant’s phenomenal world—which is neither ultimate nor merely subjective. We might compare the paradoxical properties to colors conceived in a certain way. Colors are not to be regarded as constituents of ultimate mind-independent reality, but neither are they merely fictions generated by the mind: they are primitive nonmental properties that objects instantiate—though not objects as they exist in full objectivity. It is as if the world of universals can be tapped into by the mind to represent a world of things, but underlying that world we have a reality that exists without instantiating those universals. Some of these universals are intrinsically puzzling and even paradoxical, but they don’t belong in the noumenal world. They occupy a middle ground. The phenomenal world is a world of paradoxes, while the noumenal world is a world free of paradox. We thus do justice both to the robustness of the paradoxes (they are not merely “conceptual”) and also to the intuition that the real world cannot be ultimately paradoxical (paradox cannot be written into the world as it exists independently of human sensibility). We remove the puzzle of paradox by locating it in a world that can tolerate

Metaphysics185

it. This is the world that also contains colors, values, numbers, propositions, and the like—the world according to conscious beings. The noumenal world, which is in some sense the basis of the phenomenal world, by contrast eludes our grasp except in the most tenuous of ways—it is the world as viewed from nowhere (to use Thomas Nagel’s phrase). The point, put in these terms, is that paradoxical properties belong in the world as viewed from somewhere (which is not the same thing as the viewing of it—so this is not idealism); they do not belong in the world considered as transcending any conceivable viewpoint—the world considered from the null viewpoint. These are difficult and obscure ideas, but I do not know of any other way to remove the puzzle of paradox. Even solving the paradoxes to everyone’s satisfaction would not remove the puzzle, because there would still be the question of how they arose in the first place: was it reality itself that suggested them, or were they merely the by-product of some unfortunate conceptual screw-up? The paradoxes go deep, according to the view suggested here, but not so deep as to afflict the very foundation of reality. If we think of the phenomenal world as supervening on the noumenal world (surely a plausible assumption), then we can say that the paradoxical properties belong at the supervening level, not at the underlying level. The supervening properties are distinct from the properties on which they supervene, so that their paradoxical nature does not penetrate to the underlying level.1 1.  This is the kind of essay one feels like apologizing for, because it is so speculative and obscure (not to say wild); but actually I think such essays can be important in philosophy. Some philosophical perplexity is real but inherently incapable of sharp rigorous treatment. We need to keep a sense of inarticulate puzzlement and not be afraid of wacky-sounding theories. So I won’t apologize, after all.

The Secret Cement

Our line is too short to fathom such immense abysses. —David Hume, An Enquiry Concerning Human Understanding, Part I

Hume begins his classic discussion of causation in the Enquiry with the cautionary words:1 “There are no ideas, which occur in metaphysics, more obscure and uncertain, than those of power, force, energy, or necessary connexion, of which it is every moment necessary for us to treat in all our disquisitions” (45). The first point he wishes to impress upon his readers is that these ideas are not derivable from the senses: “When we look about us towards external objects, and consider the operation of causes, we are never able, in a single instance, to discover any power or necessary connexion; any quality which binds the effect to the cause, and renders the one an infallible consequence of the other” (46). That is, the property denoted by these concepts is not perceptually available to us. This is why “from the first appearance of an object, we never can conjecture what effect will result from it. But were the power or energy of any cause discoverable by the mind, we could foresee the effect, even without experience; and might, at first, pronounce with certainty concerning it, by the mere dint of thought and reasoning” (46). We resort to “experience” to discover cause-effect relations because we lack perceptual access to the underlying property on which causation depends; if we could “discover” the power or energy inherent in a cause, we could see directly what effects it would bring about. None of the observable properties of matter supply us with the idea of causal power: “Solidity, extension, motion; these qualities are all complete in themselves, and never point out any other event which may result from them. The scenes of the universe are continually shifting, and one object follows another in an uninterrupted succession; but the power or force, which actuates the whole machine, is entirely concealed from us, and never discovers itself in the sensible qualities of body” (46–47). Nor can the fugitive powers be derived from inward reflection. Speaking of the influence of will upon body, Hume writes: “Of this we are every moment conscious. But the means, by which this is effected; the energy, by which the will performs so extraordinary an operation; of this we are so far from being immediately conscious, that it must forever escape our most diligent enquiry” (47). Thus he concludes: “Experience only teaches us, how one event constantly follows another; without instructing us in the secret connexion, which binds them together, and renders them inseparable” (48). Accordingly, he says, the power to move our body at will is “mysterious and unintelligible” and “wholly beyond our comprehension” (49). This unintelligibility is not confined to mental causation: “even in the most familiar events, the energy of the cause 1. These quotes are taken from David Hume, An Enquiry Concerning Human Understanding, ed. Peter Millican (Oxford: Oxford University Press, 2008).

188

Part IV

is as unintelligible as in the most unusual” (51). Billiard ball causation is as inscrutable as volitional causation. This reflects “the weakness of human reason, and the narrow limits, to which it is confined in all its operations” (53). Failing to acknowledge these limits, we stray into “fairy land,” he warns, following this up with the wonderful line I quote at the beginning. The simple fact is that “We are ignorant … of the manner in which bodies operate on each other. Their force or energy is entirely incomprehensible” (53). We can sum up Hume’s position as follows: the nature of causation as it exists in objects is conceptually inaccessible to us. That is, there is a property in objects that constitutes necessary connection, but our reason is too feeble and narrow to grasp it. If we could grasp it, we would not need to rely on “experience” to discover causal relations, but could simply infer the effect from knowledge of the property in question. Powers exist in objects, but they are mysterious to us, not being represented by clear and distinct ideas. The reality of causation thus transcends our cognitive capacities. In sum, Hume takes causation (involving necessary connection) to be real but mysterious. I want to ask whether Hume is right about this. In modern jargon, is it right to be a realistic mysterian with respect to causation? And exactly what form should such realistic mysterianism take? One drawback of his way of arguing the point is that he presupposes that any intelligible property must be a perceptible property (empiricism, in one version); but it may be objected that a property does not need to be perceptible in order to be cognitively accessible. However, Hume’s underlying point does not need to invoke such a strong empiricist principle: he can say instead that no conceivable quality of objects (physical or mental) could constitute power or force or necessary connection. This is as true for nonperceptual properties of atoms as it is for perceptible properties of ordinary objects. Any such properties would be merely actual properties and could not include the possible effects of the object—they could not “point out any other event that may result from them.” No categorical property of an object could constitute its power to bring about a certain kind of effect, since it would always be just an actual feature “complete in itself.” The gap between actual categorical properties and possibility-laden powers is logical or metaphysical. Perceived extension could not constitute the power to bring about certain effects, but neither could imperceptible extension, or extension in more than three dimensions, or some kind of weird nonEuclidian extension. Powers are potentials, not actually instantiated qualities; they have modality written into them. This is why it sounds wide of the mark to suggest that Hume is too hasty in declaring powers not to be cognitively accessible to us—as if we could just try a bit harder perceptually or wait for science to reveal more qualities of objects. For no such qualities could add up to power as we understand it—as inclusive of its effects, and as pure potential. That is why this family of ideas strikes us as so obscure and problematic—because we cannot make sense of them in terms of the usual properties of objects (primary or secondary qualities). Yet Hume believes that there is nothing chimerical about powers—they undoubtedly do exist in objects. It is just that

Metaphysics189

we have no clear conception (“just notion,” “accurate precise idea” [123]) of what they consist in, merely a “vulgar, inaccurate idea.” He thinks we are permanently precluded from understanding what these properties are as they exist in objects, though they no doubt have their special and determinate mode of being. If Hume is right about this, we can expect to see the characteristic pattern of philosophical responses to the mystery: what I call the DIME shape.2 Some philosophers have attempted to domesticate causation, by reducing it to something less mysterious: constant conjunction, counterfactual dependence, or energy transfer. Others have declared it irreducible and inexplicable: objects simply have a basic property of possessing causal power, and nothing further can or need be said—it is a simple self-standing quality. Then there are those who invoke the magical and miraculous: causation must be founded in the divine and omnipotent will (occasionalists), where God is the cause of every effect. Finally, we have the outright eliminativists who simply eject causation from their worldview: there is simply no such thing as causation, just one thing following another without any means of production. The D option is interpreted by its critics as covertly E, while the I option comes under suspicion of M implications. A fifth option (transcendental naturalism: TN) holds that none of the DIME options is plausible and that we have here an instance of ineradicable philosophical mystery: there is such a thing as necessitating causal power objectively existing in objects, and we can even refer to it in language and thought, but it systematically eludes our comprehension—we have no articulate knowledge of its nature. That is essentially Hume’s position: instead of being restricted to the unsatisfactory DIME options, which all presuppose our cognitive adequacy, we accept that there is something in reality that exceeds our grasp—our line cannot fathom this abyss. Our ideas of causation, power, and the like are crude and opaque labels for we know not what—they fail to reveal the nature of that which they denote. One advantage of adopting the TN position is that we are no longer driven to accept any of the unattractive DIME options, which come to seem like so many desperate attempts to avoid having to admit deep ignorance. We experience the outward manifestations of causal power, but we fail to comprehend its inner nature. Causation is, to use another resonant phrase of Hume’s, a “mystery of nature.” Given this kind of position, how should we understand our ordinary notions of causation? I suggest that we think of the relevant terms as functioning much like natural kind terms: we introduce them by means of reference-fixing descriptions, without being able to characterize what the objective nature of their reference is. Thus “power” is to denote whatever it is in objects that is responsible for their effects, whether knowable or not. As to the internal structure of our concept of power, it amounts to little more than “whatever it is that ‘power’ denotes”—a description that contains the word itself. That is, the semantic analysis of “power” is metalinguistic: if asked what power is, 2.  See my Problems in Philosophy: The Limits of Enquiry (Oxford: Blackwell, 1993).

190

Part IV

all we can say is that power is that which “power” denotes. This is not the correct kind of analysis of “red” or “square”: here we know quite well what we denote and are not forced to go metalinguistic. We might also include a demonstrative component in our concepts of power: power is that phenomenon, whatever exactly it may be. What we don’t have is a descriptively rich, revelatory concept of power. We stand at some cognitive distance from the thing itself, merely casting a line in its direction and hoping to hook something. We can succeed in our referential intentions, but that does not mean that we comprehend the thing to which we have referred. Someone might suggest that this Humean mysterianism is plausible for remote causes like gravitational attraction but not for proximate contact causation. The former is indeed mysterious and apparently “occult,” it may be admitted, but the latter is transparent and “mechanical.” So maybe some causation is subject to TN but some is quite readily domesticated. Isn’t that Newton’s own position, contrasting his peculiar action at a distance with the mechanical causation of the Cartesians? Granted we don’t understand the power of gravity (we can merely state its empirical laws), but don’t we understand quite well how billiard balls impart motion to each other? That is evidently not Hume’s position, and I think with good reason: for what is it about moving billiard balls that enables them to cause motion in other billiard balls? Is it their extension or their mass or their shape or their path through space? Hume will insist that none of the observable properties of billiard balls can add up to a causal power, since the power is defined relative to its effects, and these properties are not (being “complete in themselves”). Contact causation is therefore also mysterious, because we still don’t grasp the nature of causal power. Also, as we now know, so-called contact causation is not as proximate as we once thought, since objects give off forces of electrical resistance that prevent their atoms from actually touching—so all causation is action at a distance, though sometimes a very short distance. Nor is mental causation intelligible to us, since it too depends on an underlying causal power that is not presented to us in reflection. We experience the passing show, inwardly or outwardly, but the machine behind the scenes remains cloaked in mystery. Even high-powered microscopes fail to reveal it. The cement of the universe is invisible cement. Causation is hidden. In A Treatise of Human Nature Hume writes: “I am, indeed, ready to allow, that there may be several qualities both in material and immaterial objects, with which we are utterly unacquainted; and if we please to call these power or efficacy, ’twill be of little consequences to the world. But when, instead of meaning these unknown qualities, we make the terms power and efficacy signify something, of which we have a clear idea, and which is incompatible with those objects, to which we apply it, obscurity and error then begin to take place, and we are led astray by a false philosophy” (218).3 He 3.  David Hume, A Treatise of Human Nature, ed. David Fate Norton and Mary J. Norton (Oxford: Oxford University Press, 2011).

Metaphysics191

is urging here that we must use “power” and “efficacy” to refer to unknown qualities of objects, not known qualities (as in the DIME options). But what he does not tell us, here or elsewhere, is why we are so ignorant of the unknown qualities on which causation depends; he treats our ignorance as a kind of primitive fact about us in relation to the world. He has no theory of our causal ignorance. This is not a criticism of Hume, just to note a lacuna it would be nice to fill. I have no very good ideas about how to fill it, though it does seem as if the mystery is deep and principled. We can make the routine point that humans and other animals have no biological need to grasp the ultimate essence of causation, so long as we can make useful causal predictions; adding an ability to extend our cognitive line deeper into the abyss might be metabolically costly and adaptively pointless. But we have quite a lot of knowledge of doubtful evolutionary value, so it is a question why we so decisively draw a blank here. Furthermore, the cognitive gap appears to reflect a conceptual difficulty: what would it even be to grasp the principle of causation? What would it look like? Of course, our bafflement here might reflect the very ignorance Hume is pointing out— we can’t understand why we can’t understand because we can’t understand. We seem to be swimming in a sea of obscurity, in which even phrases like “unknown quality” might be inapposite—the basic principle of causal power might not be a quality at all. At any rate, we cannot conceive of it on analogy with qualities with which we are familiar (primary and secondary qualities). We are reduced to speaking obscurely of causal oomph. We gain firmer ground when we consider a supervenience claim, namely that causal powers supervene on categorical properties of objects. If two objects agree on all their observable and unobservable categorical properties, then they must share their causal powers. The case is very like mind and body, in which mystery and supervenience also puzzlingly coexist. There is clearly more to causal powers than categorical properties, but powers are strongly constrained by such properties. Hume never considers the question, but I don’t see why he could not accept such supervenience: then the unknown would supervene on the known—as the mental supervenes on the physical, or the evaluative on the descriptive. For some reason we can grasp the categorical properties but not the supervening powers—though we can refer to these powers (albeit obscurely and indirectly). The supervenience relation might itself be opaque, as well as that which supervenes; but still, such supervenience appears to be the case. Causal powers are tied to categorical properties in some inextricable (if inexplicable) way. What we don’t grasp is the origin of propensity, power, force, or energy—the oomph. Thus we don’t really understand what it is that makes things change. We merely chart superficial correlations on the basis of experience, without exposing underlying necessities. Yet necessities there are, somewhere in the secret bowels of the universe. There is clearly something puzzling about possibility—potential, power, disposition, capacity. Wittgenstein comments on this in two sections of the Investigations about

192

Part IV

machines and their possible movements: “As if it were not enough for the movements in question to be empirically determined in advance, but they had to be really—in a mysterious sense—already present” (193).4 Then he asks: “When does one have the thought: the possible movements of a machine are already there in it in some mysterious way? Well, when one is doing philosophy” (194). We can substitute “object” for “machine” and ask the same question: In what sense is the potential to do various things present in an object? In what way does a cause already contain its possible effects? As Wittgenstein says, a possible movement is not a movement, and it is not the mere physical conditions for a movement either; it is (we imagine) more “like a shadow of the movement itself.” But it can’t literally be a shadow (or a ghost or a shimmering reflection): it is something both present and absent, both concrete and abstract, both real and unreal. It taxes our cognitive powers—our methods of understanding. Even ascriptions of possible properties are puzzling: I am possibly a bricklayer—so is this a property I really possess in some way (in some possible world, maybe)? It hovers around me or beside me, like a shadow. It seems real but not in reality. No wonder philosophers have sought to banish such properties from the world, by rejecting modality altogether or reducing it to something deemed less mysterious (analyticity, human convention). Causal necessity and possibility seem both inherent in objects and yet curiously removed from them. We say a seed has the potential to grow into an oak—the tree is an effect of the seed—but how is an oak present in a seed? Don’t reply, “Because of the DNA,” because that just invites the question of how DNA can contain oaks: How do genes have causal powers? How does anything cause anything? That was really Hume’s basic intuition: how can any mere object produce anything? In virtue of what does an object bring about effects? Shouldn’t it all just be one damn thing after another? And yet causation is real, and potential and power are too. There is what Hume calls a “tie” between things (Enquiry, 64). Bafflement as to how causation works thus coexists with the conviction that it is as real as anything in nature. Realistic mysterianism is the view that best captures and explains these intuitions and perplexities. Of course, it has the consequence that the world in general is very much less intelligible to us than we tend uncritically to suppose. But it is a lot to expect that the world should be intelligible to us. The world simply is, and we catch whatever glimpses of it that we can. 4.  These quotes are taken from Wittgenstein, Philosophical Investigations, ed. and trans. G. E. M. Anscombe (Oxford: Blackwell, 1958).

Analysis and Mystery

The metaphilosophy I favor contains two elements: (a) philosophy is conceptual analysis, and (b) philosophical problems are often insoluble. According to (a), we arrive at philosophical knowledge by a priori investigation of concepts; according to (b), many philosophical questions cannot be answered by us, since they concern mysteries that exceed the power of the human intellect. It might be felt that these two positions are incompatible: if philosophical knowledge takes the form of knowledge of analytic truths concerning concepts, then how is it possible for such knowledge to be in principle unobtainable? To acquire philosophical knowledge, we need merely to reflect on our concepts—so how can it be so difficult to obtain such knowledge? How can it be a deep mystery what is involved in concepts? Isn’t philosophical knowledge already contained in our minds in an implicit form? There is thus an apparent tension between these two components of my metaphilosophical position. We might seek to relieve the tension by adopting one or other of the following two lines. First, it might be said that conceptual analysis is not as easy as it sounds. It is not as if the internal structure of our concepts is laid bare to the introspective eye in all its glory; it can be remarkably difficult to excavate the inner workings of a concept. This is why even successful conceptual analyses can prove elusive. Russell didn’t arrive at the theory of descriptions just by casting a casual glance at his ordinary understanding of “the”: there were many false turns and wrong starts. This point may be conceded: there are indeed hard problems of conceptual analysis. But it is difficult to see how hard problems of conceptual analysis might amount to total intractable permanent mystery. Surely questions of conceptual analysis are solvable in principle—yet according to the mystery thesis some philosophical problems are in principle unsolvable by the human mind. So the admitted difficulty of conceptual analysis does not go deep enough, if the mystery thesis is correct. A second response is to hold that some problems that seem philosophical are not really so; they are really scientific problems. Thus the insolubility of the problem does not arise from an inaccessible conceptual truth but from whatever it is that gives rise to scientific mysteries. Scientific mysteries presumably arise from such factors as limitations of available data or the difficulty of empirical theory construction. The tension is removed by recognizing that the problems that produce mysteries are not really problems of conceptual analysis; they are problems of scientific verification and theory construction. For instance, it might be said that the mind–body problem is inherently a scientific problem, not a problem of a priori conceptual analysis: we don’t have enough data and we have yet to produce the Einstein of consciousness and the brain. Again, this option should not be ruled out completely: it may be that a problem can appear to be philosophical but turn out to be empirical. Impossible science can look a lot like difficult philosophy (quantum mechanics?). But this diagnosis does not

194

Part IV

cover the full range of philosophical mystery, because some problems in philosophy do seem to stem from conceptual inadequacies and not merely lack of data or scientific genius. In the case of free will, say, it is hard to deny that our concept of free will is producing the problem, not merely a paucity of empirical data about free actions. It is not as if free actions are found only in a part of the universe that is remote from our means of observation. So we would be advised to seek some other explanation for why some philosophical problems generate intractable mystery. The problem is that it is hard to reconcile the conceptual character of philosophical problems with the admission that it is excruciatingly difficult to solve such problems. How can it be impossible to analyze a concept? The reconciliation I favor is that philosophical mysteries arise primarily from conceptual lack. It is not that we possess the concept but cannot gain insight into it; it is that we don’t possess the concept at all. The answer to the philosophical problem does indeed lie in a piece of conceptual analysis, but the concept in question is not one that we possess—though it may be possessed by other possible beings. It is not that the right concept is possessed by us but is maddeningly opaque; nor is it that the question is really empirical in nature: the answer lies in the analysis of a concept that lies outside of our cognitive range. This conceptual transcendence could in principle come in different strengths, according to how inaccessible to us the concept in question may be. It might be simply not possessed by us within our current conceptual scheme but could easily be added to that scheme—so the insolubility is relative to a particular moment in conceptual history. In that case, the problem is only temporarily insoluble and requires no major upgrading of our conceptual resources. On the other hand, the concept may be inaccessible without a substantial increase in brain size or a fundamental alteration of neural processing—so that further evolution or genetic engineering is needed to put the concept within our cognitive reach. More strongly still, the concept may be permanently unavailable to creatures sharing our basic modes of awareness of the world, or having brains made of organic tissue, or even to all finite beings. Clearly, the philosophical problem will be more or less intractable according to how inaccessible the needed concepts are. To see how this type of explanation might work in particular cases, consider again consciousness and free will. We currently have concepts that refer to these things, expressed by the words “consciousness” and “free will.” Employing these concepts, we encounter intractable philosophical problems regarding the references of the words. But there may be possible beings that conceptualize the same references in radically different ways—and with respect to those beings, an analysis of their expanded or revised concepts might resolve such problems. Or again, we possess certain concepts of space and time that appear unable to resolve ontological questions about the entities designated; but there may be superior beings possessing richer or more penetrating concepts for which the ontology is clear upon determined analysis of these concepts (superior

Metaphysics195

from a philosophical point of view). Clearly, two concepts can have the same reference and yet receive different conceptual analyses, one more informative about the reference than the other (senses can differ in cognitive content). To use old-fashioned terminology, one concept may be adequate to capture the essence of the referent, while another may be inadequate to capture it. One way in which a concept can be inadequate, in the intended sense, is that it is merely name-like, with no serious descriptive content—while an adequate concept will be one that is rich in descriptive content and thus capable of yielding a range of analytic truths. If our human concepts for a certain phenomenon are inadequate, then we will not be able to resolve philosophical problems about the phenomenon by analyzing these concepts. To solve the problem we would need conceptual augmentation. Because of this theoretical possibility, then, the following two propositions are entirely consistent: (a) all philosophical knowledge consists of conceptual analysis (and hence is of analytic truths), and (b) some philosophical problems are necessarily insoluble by human beings, that is, permanent mysteries. The key to resolving the apparent tension is simply to recognize that human concepts may be condemned to being a subset of all possible concepts. Yes, philosophical truth consists of conceptual analysis; but no, not all concepts are accessible to humans. The mysteries for us might then be straightforward conceptual truths for other types of intelligent being. Certainly, if there were such conceptual lacks, we would see the kinds of mysteries postulated by the metaphilosophy I favor. The root of philosophical mystery would then differ from that responsible for scientific mysteries, in which a lack of data is typically the prime problem (e.g., how life originated on Earth). Philosophical mysteries result from conceptual lack, while scientific mysteries result from data lack, roughly speaking. (It isn’t that biologists have an inadequate concept of life and hence can’t explain how life began; they just don’t have enough information about conditions on Earth four billion years ago). According to the present diagnosis, philosophical mysteries stem from a gap in the human conceptual scheme—a conceptual blind spot. They stem from conceptual poverty or paucity or primitiveness. It follows that philosophical mystification is a relative predicament: we may be mystified by a phenomenon X, but a species of cognitively superior beings may not be mystified by X (though they may be mystified by some Y that does not mystify us). Let me illustrate how this relativity might work in particular cases. Many animals possess such psychological traits as perception, belief, knowledge, intention, and even meaning; but they do not possess concepts of these traits. There are philosophical questions that arise about the traits they possess—notably constitutive questions about what the traits consist in. But since the animals don’t have concepts corresponding to the traits, they are not in a position to answer such questions by means of conceptual analysis. There are philosophical questions about these animals that they are not equipped to answer, through conceptual lack. Suppose the conceptual apparatus of sense and reference is

196

Part IV

what is needed to resolve some of these questions; then they will be constitutionally unable to answer such questions, owing to a lack of conceptual sophistication. We might be in an analogous position with respect to a species cognitively superior to us: we have traits (consciousness, free will) that raise philosophical problems, but we lack the conceptual sophistication necessary to resolve these problems—we just don’t have rich enough concepts on which to perform the requisite analysis. Only a cognitively superior species can resolve the problems that are raised by our nature and are recognized by us. There is a difference between animals and humans here: we can appreciate and raise the problems, but they cannot. They have traits that raise the problems, but they don’t have the sophistication to appreciate the problems, let alone resolve them. We are in an intermediate position with respect to such animals and the philosophically superior beings: we can raise the problems, but we can’t answer them (if they are really mysteries for us). How is that possible? Consider our early ancestors and their gradually dawning cognitive sophistication (say, Neanderthals). At some point philosophical problems will have begun to occur to them, even if inchoately, as they became more self-aware. They were developing concepts of their own given traits and beginning to reflect on them. But they had not yet developed concepts that contain the resources to answer their questions—they had inadequate concepts of X. Thus they can raise the questions but not produce the answers. The simplest way to model their conceptual predicament is that they have developed name-like or demonstrative concepts for the traits they possess (consciousness, free will), but that such concepts are too crude and inarticulate to yield any serious conceptual dividends with respect to what they merely label. Suppose they have the trait of knowledge, which consists (to simplify) of true justified belief, but that the concept they have that refers to that trait is merely name-like and contains no analysis of what knowledge is. Then they can ask, “What is knowledge?” using their name-like concept, but they cannot analyze the content of that concept in order to generate the correct account of what knowledge (the trait) is. We might be in an analogous predicament with respect to our traits of consciousness and free will and the concepts we now possess of these traits—our concepts are descriptively inadequate, too thin and unrevealing. We can raise conceptual questions by using these concepts, but we cannot answer the questions by analyzing the concepts used (much the same might be said of our cognitive relation to space and time). So we have three basic cases: (i) a species that has traits that raise philosophical questions but cannot itself consider these questions, (ii) a species that can consider the questions but whose concepts do not contain the resources to answer them (there is a class of mysteries for this species), and (iii) a species that can consider the questions and answer them by means of analysis of its comparatively rich set of concepts. Thus: beasts, humanity, and God (or Martians)—to put it simply.

Metaphysics197

Further insight into the human propensity for philosophical perplexity would require deeper investigation of the human conceptual system and its powers of advancement. Here we will offer only some sketchy general remarks. The idea of inherent limitations to human understanding is strongly suggested by both the nativist and empiricist traditions. The nativist presumes innate constraints on the range of concepts available to the human mind—as is the case with the concepts possessed by other species—because concepts are inherited from a fixed stock of ancestral concepts that have evolved for the usual biological reasons. The empiricist, holding concepts to derive from the senses, also presupposes limitations, given the obvious limits on the range and acuity of the senses. On both views there may be questions that exceed contingent human cognitive powers—and so philosophical mysteries will naturally arise for us. Conceptual lack will be part of the human condition (as it is for all other evolved species). We should expect a certain degree of conceptual fixity and hence stasis, not limitless plasticity or outright omniscience. And it is notable that human conceptual advance during recorded history seems minimal when measured by interspecies comparisons. Different species differ among themselves quite dramatically, cognitively speaking, as they do anatomically; but there seems little basic cognitive difference between recorded early human thought and more recent thought. Such progress as there has been looks like conjunction and combination, not radical conceptual innovation or cognitive restructuring (as in the evolution of species). And that is not surprising, given the short evolutionary timescale. The human brain has been much the same for many thousands of years, and the human mind has been pretty unvarying too. There is little reason to expect any truly radical cognitive advances in the near evolutionary future (unless by direct genetic engineering). The area in which conceptual innovation seems the most impressive (though such judgments are very fallible) is mathematics: here humans do seem to have made genuine conceptual advances in understanding infinity, probability, sets, zero, and the like. For the rest we just seem to be recycling old concepts in new settings (e.g., the concepts of force and field or unconscious and computation). Major conceptual advances in human history look few and far between—as compared to the large cognitive gaps that exist between different evolved species (e.g., mice and monkeys). So if our long-standing philosophical mysteries stem from conceptual absences, it is unlikely these absences are going to be filled any time soon. We may indeed never be able to perform the conceptual analyses that would resolve some of our deepest mysteries, even though the analyses exist (if only in Platonic heaven). We are simply precluded from possessing the necessary concepts. Thus we may say that, on the one hand, philosophy is quite easy methodologically, while, on the other hand, it is impossibly difficult. It depends on who you are conceptually. Conceptual analysis is not all that difficult in itself, but if you lack the concepts to analyze, constitutionally so, then you are out of luck. Fugitive concepts are the root of philosophical intractability.

Explanation and Necessity

What is the difference between necessary truths and contingent truths? Some say that necessary truths are analytic while contingent truths are synthetic; some say that necessary truths are a priori while contingent truths are a posteriori; some say that necessary truths hold in all possible worlds while contingent truths hold only in some possible worlds. The first two suggestions face familiar counterexamples; the third suggestion is not very illuminating, since it stays within the modal realm. We might as well say that the difference is simply that necessary truths are necessary while contingent truths are contingent—that is, take the notions as primitive. Can we venture anything more illuminating? I propose the following: contingent truths can always be explained, but necessary truths can never be explained. If you ask why there are three books on my table, I can explain that fact—because I put three books there. Similarly for any other contingent truth, going all the way back to the beginning of the universe (if there was one). What are called “empirical facts” have explanations or causes—something brought them about. There is a reason for their existence, as opposed to some other state of affairs. But with necessary truths, we don’t have this kind of explanation: there is no causal explanation of 2 + 2 = 4, or the identity of Hesperus and Phosphorus, or water being H2O, or your having had a particular parental origin, or red being a color, or pain being felt a certain way, or nothing being both square and not square. These necessary facts don’t have causal explanations—reasons they are the way they are rather than some other way. They must be as they are. There is no antecedent state of affairs from which they derive. For example, there is an explanation for why Hesperus is in a certain position (gravity, its mass, its earlier position), but there is no explanation for why it is identical to Phosphorus (as opposed to why it is also called “Phosphorus”). It just is. Similarly, we can’t explain why I have the parents I have, though we can explain why I have certain traits by reference to my parents. We can explain why there are four rabbits in my garden, but not why two rabbits plus two rabbits equals four rabbits. Necessary facts are not brought about by anything—they have no history, no genesis. They simply are. Suppose we ask why this table is made of this specific piece of wood. The question seems peculiar. We should not confuse this question with the question of why its maker used wood and not metal, which does have an explanation, or why I choose to use a wooden table. The question is why this particular table is made of this particular piece of wood. That question has no answer, though we can explain why the table is chipped or varnished. Nothing explains the fact that this table is composed

200

Part IV

of this piece of wood—any more than anything explains the fact that the table is self-identical. The things that constitute an object’s essence don’t have causal explanations. But the contingent properties of an object do have such explanations. Someone might say that there is an explanation of necessary truths, namely that they are necessary: their being necessary explains why they are true. Why is 2 + 2 = 4 true? Because it has to be true—it could not be otherwise. But this kind of explanation is nothing like the kind of historical-causal explanation that applies to contingent truths. We certainly don’t explain the fact that my table has three books on it by saying that this is a contingent fact! We could, if we like, say that the only explanation that necessary truths have is that they are necessary, instead of saying they have no explanation at all; either way they stand apart from contingent truths in respect of explanation. Necessary truths are not the upshot of prior conditions of the world, but contingent truths are. If this is right, the necessary–contingent distinction tracks another distinction, namely that between what can be explained and what cannot be explained (in the causal-historical style). So there is something special and distinctive about necessary truths (apart from their being necessary): but it is not that they are analytic or a priori or true in all worlds—rather, they are inexplicable. They simply are. Granted that I am necessarily human, the fact that I am human has no explanation; but the fact that I am a philosopher or a tennis player does have an explanation. The fact that a human was born in the time and place I was born has an explanation (it was the fact that my parents are human); but there is no explanation of the fact that I am human. For what would such an explanation look like? It is not that I was caused to be human, whatever that might mean. I was already human! Nothing caused me to be identical with myself, either—just by existing I am identical with myself. Nothing caused the table to be made of that chunk of wood—just by existing it (that table) was that chunk and no other. True, something caused the chunk to be made into a table, namely the labor of the carpenter. But it is only a contingent truth that this chunk is a table—it might have been made into a cabinet. What is not contingent is that this table is made of that chunk—and this has no explanation. Nor did anything make the number 2 even—it was even from the start. Necessary facts are not caused to obtain, even if the objects involved are caused to exist. This point about explanation in relation to necessity allows us to get at necessity from outside the modal circle, and it doesn’t have the counterexamples that the other traditional accounts have (a priori contingencies, a posteriori synthetic necessities). It is also rather surprising, since there seems no obvious reason for things to line up as they apparently do: for what has necessity got to do with explanation, or its lack? What is

Metaphysics201

true in all worlds turns out to coincide with what has no explanation. What must be so cannot be explained to be so. Interesting, is it not?1 1. The position suggested here would not be surprising to an old-fashioned positivist: if all necessity is verbal or conceptual necessity, necessary truths will not have empirical explanations. There is no causal explanation of the fact that bachelors are unmarried males, this being an analytic truth. On the other hand, the post-Kripke orthodoxy is that statements can be exactly alike except in respect of modal status: necessary truths, like contingent truths, can be synthetic, a posteriori, de re, and “factual.” Thus my position may be surprising to a modern Kripkean, schooled to accept that necessity is irreducible to anything else. Given that positivism is mistaken in holding that all necessary truths are analytic, we need to recognize that necessary truths are not distinguished from contingent truths by being analytic; but yet they are not simply homogeneous with contingent truths, as the Kripkean picture appears to suggest. We need an analogue of analyticity to distinguish necessary truths from contingent truths—viz. the kind of statement whose truth is not explicable in the causal-historical manner. That is, we need a position strategically between positivism and Kripkeanism. We need, I am suggesting, the notion of the “synthetic inexplicable” to mark out necessary truths. This may require some rethinking of traditional categories.

Against Possible Worlds

The possible worlds theory of modality postulates a substantive ontology of existing entities. Modal propositions are true or false depending on the properties of these entities (concrete particulars, in some versions). Without the existence of these worlds, modal propositions could not be true or false: it is possible that p only if there exists a world in which p, and it is necessary that p only if in every such existing world p. Modal truth allegedly depends on the existence of this domain of objects. But does modal truth really depend on the existence of such entities? Surely it is possible to destroy the actual world. We can envisage parts of it being destroyed, so why not all of it? If the big bang created the actual world, couldn’t there be a big crunch that destroys it? If God created it, couldn’t he remove it from the face of reality? Our universe may not always exist; it may devour itself or be devoured. Just as we can destroy any concrete particular within the actual world, so we can destroy the concrete particular that is the whole actual world. It exists contingently, and can be done away with. The actual world is destructible. But then so are the other possible worlds, which don’t differ intrinsically from the actual world, themselves being big concrete particulars. Cataclysms may occur in these worlds too, the end result of which is the disappearance of that world. Couldn’t God wipe out the whole lot in a single stroke, thus reversing his initial act of creation? The domain of possible worlds is a domain of entities that do not exist necessarily, and which may in principle be destroyed. Suppose that day comes—all the possible worlds are summarily destroyed. What happens to the truth-values of modal propositions? What, say, is the truth-value of “Necessarily 2 + 2 = 4”? This sentence is not true in each of the worlds that heretofore existed, since they no longer exist—so must we say it is false?1 It used to be true at all those worlds, but now it is true at none of them—so now must we say that “It is impossible that 2 + 2 = 4” is true? If impossibility is defined as truth at no world, then that is the situation for every proposition after the destruction occurs. Similarly, what about “Possibly there are electrons” after the destruction? This implies, in the standard conception, that there exists a world in which there are electrons: but there is now no such world. So is the sentence rendered false? That sounds absurd. Mathematical truths will not cease to be necessary if the domain of worlds is extinguished, and possibilities will not cease to be possibilities just because there are no existing worlds at which they hold. Thus the condition of the supposed 1.  We might try saying that it is trivially true, since there is no world at which it is false—like “All dodos fly.” That is, we could suppose that “all worlds” has no existential import. But then every proposition will turn out to be a necessary truth when the worlds are all destroyed, since even contingent propositions will not be false at any world, there being none.

204

Part IV

domain of possible worlds is irrelevant to modal truth. Modal truths do not depend on the existence of a totality of possible worlds, conceived in the standard way—that is, as real concrete entities like the actual world. That is just the wrong way to think about their truth conditions. The case is quite different with temporal and spatial locutions, where a quantificational analysis looks very attractive. If we were to destroy space and time, this would have a huge impact on the truth-values of the sentences involved, turning them from true to false immediately. Thus, “Always p” or “Everywhere p” could not be true if there were no times and places to quantify over, and similarly for “Sometime p” and “Somewhere p.” Here the ontology is essential to the truth-value of the sentences in question. But in the case of modal sentences, the absence of the worlds does not change a sentence like “Necessarily 2 + 2 = 4” from true to false, because the postulated worlds are not required for truth-values as we have them. Hence destroying the worlds does not alter truth-values in their case. The postulated worlds are really irrelevant to the truth-values of modal propositions. Suppose now that a group of speakers accepts the above argument against possible worlds as a theory of the meaning of modal expressions (or some other argument to the same effect). They thus accept that modal concepts are not definable in terms of quantification over possible worlds; indeed, we can suppose them to be vigorously against the whole ontology of possible worlds—they reject it outright. They are also good Griceans about meaning, supposing it to be determined by speakers’ intentions. They wish to introduce a vocabulary of modal expressions, to express the modal thoughts that have lately been crowding in on them as a result of philosophical reflection. Thus they begin uttering sentences of the form “Necessarily p” and “Possibly p.” What do they mean by these utterances? Surely not that in all possible worlds p or there exists a possible world in which p: for they flatly reject any such ontology. They intend to produce modal beliefs in their audience by means of its recognition of their intention, but not beliefs about possible worlds: they intend, say, to produce the belief that necessarily 2 + 2 = 4 by uttering “Necessarily 2 + 2 = 4,” but they reject the idea that necessity has anything to do with possible worlds. Given that what their words mean is fixed by their intentions in uttering them, we cannot analyze their sentences in terms of an ontology of possible worlds: that cannot be the correct semantics for their language. But surely we also can repudiate the ontology of possible worlds as we speak: if most of us reject the possible worlds analysis, how can it be what we really mean? How can our modal language diverge in its meaning from what we mean by it? The possible worlds semantics cannot be forced on us nonbelievers as what we must mean. We are free to protest: “But I don’t mean that!” The basic problem is that possible worlds semantics imports an alien ontology into the meaning of modal expressions; so the two can be pulled tidily apart. There can be necessary truths even though there are no possible worlds (understood in the standard way). We might compare the case to a similar conception of negative propositions,

Metaphysics205

namely the view that a negative proposition is true if and only if it is true in the “negative world,” that is, the world that contains all the states of affairs that don’t hold in the world of positive facts. The negative world consists of all the negative facts, for example, the fact that snow is not black. But surely this is one world too many: we have no need to postulate such a world in order to make sense of “not.” If that world were destroyed, by some perverse act of God, we could still have true negative propositions; and someone could use the word “not” meaningfully while rejecting the ontology of the negative world. Possible worlds are like that: misguided metaphysical fictions, superfluous to requirements. Even if there were such entities, they could be destroyed and modal truth would not be affected. They are like Meinongian entities: definite descriptions can be meaningful without there being any such entities, and even if they did exist they could be destroyed without detriment to the meaningfulness of definite descriptions. We could put all the subsistent entities on a bonfire and reduce them to ashes—descriptions would not thereby lose their meaning. Descriptions don’t depend on them for meaning (as Russell’s theory shows); likewise, modal expressions don’t depend on possible worlds to have meaning and truth-value, even if they exist. The proof is that we can destroy them and leave modal sentences as meaningful and true as ever.2 2.  Insofar as destructible worlds are used in the analysis of counterfactuals or causation or anything else, there is the same problem: if the destruction of the worlds leaves counterfactuals or causation intact, they cannot be constitutive of these things—they cannot be what counterfactual and causal truth consist in. And it certainly seems as if they would be unaffected by the elimination of all the possible worlds (except perhaps the actual world). Smoking won’t cease to cause cancer at the moment God annihilates all the possible worlds he had earlier created. So how could causation be determined by what happens in those worlds?

The Concept of a Person

I have come to the unsettling conclusion that the concept of a person, as philosophers employ that concept, is a bad concept. It leads to the formulation of bad questions that have no clear answers. The concept does not pick out any natural kind and is quite misleading. It should be abandoned as an important concept in philosophy, except in a very restricted setting. What counts as a person? We typically apply the word to ordinary adult humans of sound mind, assuming a certain set of mental characteristics—intelligence, consciousness, self-reflection, self-governance, memory, and so on. But what about children: When does a human child become a person? Is it at the age of sexual maturity, or when the child starts to walk and talk, or at birth, or in the third trimester, or at conception? Opinions differ radically. According to the standard Lockean definition, in terms of conscious self-reflection, persons must have advanced cognitive skills, so that personhood begins only when the mind reaches a certain level of sophistication— possibly around puberty or later, depending on the individual. So many human children are deemed nonpersons—though they have human bodies, minds, language, and will. What about those suffering from various forms of mental deficit—are they persons? Is an autistic adult a person? Does Alzheimer’s destroy personhood? Does coma eliminate the person? Are you a person while asleep, or just before you die in your sleep? And is there a science of persons? Is this concept useful in biology, or psychology? Why do we have the concept? What does it do for us? We are apt to restrict the concept of a person to the human species—only humans are said to be persons. Our pets are not deemed persons, nor are our closest biological relatives. Would we call other hominids persons if they still existed—Neanderthals, Homo erectus, et al.? Didn’t some people once deny that individuals of other races are persons? But are we just wrong to impose these restrictions—might we discover that gorillas, say, really are persons after all? What would such a discovery involve? Might their DNA make us accept that gorillas are persons, as it might make us accept that they belong to the same family that includes lemurs? Could their personhood be a scientific discovery? And if they are persons, what about other primates, other mammals, or even reptiles—might they too be persons for all we know? Is it that we know empirically that crocodiles are not persons, as we know they are not warm-blooded? Is that a scientific fact? Is it conceivable that turtles might turn out to be persons—but not sharks or octopuses? And when did the biological kind persons evolve? Might we stop calling ourselves persons if our mental faculties drop below a certain level (“We used to be persons but now we don’t measure up”)? There is a philosophical subject called “personal identity” in which we strive to find what constitutes the continued existence of a person. The subject involves many ingenious thought experiments, and it is difficult to come up with a satisfactory theory.

208

Part IV

Presumably the question is not supposed to include nonpersons: we are not seeking the conditions of nonpersonal identity—the question is supposed to be exclusively about persons as such. So we are not officially interested in young children and members of other species, since they don’t count as persons. But the same thought experiments, and the same theories, can be applied to these nonpersons too. A human child, say, three-year-old Jill, persists through time, and we can ask what her persistence consists in—what makes this child Jill. Is it her body or brain or memories or consciousness or personality—or none of the above? We can envisage swapping her brain for Jack’s brain, or dividing her brain in two, or erasing her memories—the usual philosophical moves. Yet none of this is about personal identity, Jill not being a person (yet), as we may suppose. Or if you think human children do count as persons, even going as far back as the fetus, what about cats and dogs—what does their identity through time consist in? What makes Fido, Fido? We can swap Fido’s brain, zap his memories, tinker with his personality, and subject him to teletransportation—the philosophical works. Yet none of this concerns a question of personal identity—just canine identity: “Is it the same dog?” not “Is it the same person?” But surely these questions about nonpersons are really the same as questions about the identity of persons—we have not got two philosophical problems here, one about persons and the other about nonpersons. So the question of personal identity, as it is normally pursued, is not really a question about personal identity as such. That is a misnomer. The concept of a person is not the concept we need to pursue these kinds of questions: it is too restrictive. And quite possibly it makes the questions needlessly intractable, because the concept itself is so vague, messy, and unnatural. We can ask what constitutes identity of body, identity of mind, and identity of animal (dog or gorilla, say), but asking what constitutes the identity of a “person” is not a very well-defined question, pending some clearer idea of what a person is. What question is left over when we have answered those other questions—in particular, when we have answered the question of what canine identity or human identity is? If we have a theory of human identity over time, don’t we have all we need? In other words, why not focus on species concepts and formulate the question that way? These are sortals in good standing, unlike the putative sortal “person,” which admits of so much indeterminacy. If we find we can settle questions of animal identity—dogs, turtles, humans—why bother with the supposed further question of personal identity? Maybe this just generates pseudo-questions that simply have no answer. We can also meaningfully inquire about the identity through time of minds—what makes me have the same mind today that I had yesterday or a year ago? The answer will specify my mental capacities, as well as certain kinds of psychological continuity; and the question posed may have a clear answer. But this will not satisfy the seeker after the secret of personal identity, which is construed as a question about another kind of entity entirely—the person. For why—that seeker will ask—couldn’t the same person have a different mind at different times, and why

Metaphysics209

couldn’t different persons share the same mind? Such questions appear fanciful, but they are easily generated from the assumption that there is a substantial further issue about personal identity. However, if we simply stop asking that question—that is, stop going on about the supposed category of persons—we can still cover all the ground that really matters, namely identity of body, animal, and mind. There is really no additional question in this neighborhood worth asking—so at least the skeptic about “personal identity” will contend. There is clearly a challenge here to explain what well-defined question remains once those other questions have been dealt with. The concept of a person is really quite a recent addition to our conceptual repertoire (replacing “soul” or “spirit”), but surely there were questions about our identity through time before it made its entrance. It is suspicious that we don’t have a term corresponding to “person” for other species. Some well-meaning people suggest that we should extend the concept to other species, because of their psychological similarities to us; but that seems rather forced and stipulative. What is odd is that we don’t have a more general concept of which person is a special case, given that we recognize that animals have minds as well as bodies. We think Fido is the same dog from day to day, as we think Bill is the same human from day to day; but we don’t have a term corresponding to “person” to add to our description in the case of Fido. There is a natural psychological kind here that subsumes both Bill and Fido, and which resembles the concept of a person, but we don’t actually have a word that does this job—hence we have to say bluntly that Fido is a nonperson. We might try using “psychological subject” or “ego” or “self,” but these don’t capture the notion of a nonhuman but person-like being (a “quasi person”?). What this suggests to me is that the concept of a person is not really a natural kind concept at all—it is not intended to capture significant natural traits of things. It has a completely different function. That is why we don’t have a more capacious notion of a person, despite recognizing similarities between ourselves and other species, and indeed between adult humans and juvenile humans (as well as others). The job of the word “person” is not to capture the nature of a certain kind of thing; rather, it is to enforce a certain kind of division—to stipulate a certain kind of exclusion. It is intentionally invidious. Locke remarked that “person” is a forensic term, that is, a term of the law. Let me rather say that it is a political and legal term, as is the concept expressed. To classify an individual as a person is to grant him or her certain rights—legal, political, moral. A person is precisely someone who possesses, or is deemed to possess, these rights—a right-holder. It is like calling someone a “gentleman” as opposed to a “commoner”: the point is to indicate how such a one is to be treated, not to get at some natural essence. We don’t refuse to call children and animals persons because we think they differ fundamentally from us in their objective nature; we do it because we are marking them out as beyond the normative sphere to which normal human adults belong—the sphere of

210

Part IV

responsibility, legal obligation, ownership, and so on. True, there are real differences that underlie this kind of forensic distinction, but the term “person” is employed to abstract away from these and focus on matters of law and politics. We declare a young human a person upon the attainment of a certain position in society, as we might stipulate a gorilla to be a person if gorillas come to be accorded legal rights comparable to those applicable to adult humans. It is not that we discover these creatures to be persons by observation or analysis—though we may discover relevant facts about their minds or bodies. The term “person” is a kind of honorific or status term, intended to signify belonging—it connotes legal and political standing. It is like “citizen” or “aristocrat” or “star” or “lady.” It is not the concept of a certain kind of natural entity. If this is right, we can see what is going wrong with the philosopher’s use of the concept of a person. It is not a concept designed for, or useful in, metaphysical or scientific contexts, but in political or legal contexts. There is no such question therefore as the “nature of persons” or “personal identity through time”—though there are real questions about the nature of animals and their minds and about the identity through time of animals and their minds. We can certainly ask about minds of different levels of complexity, up to and including the Lockean conception of a self-reflective conscious being that can “consider itself as itself.” But this should not be interpreted as a division into “persons” and “nonpersons”: there are just too many grades of animal (and human) mindedness for that dichotomy to be realistic. There is no such ontological subject as persons—at least as that concept is normally understood by philosophers. The kind “normal adult human with legal rights and obligations” is not a metaphysical kind, as philosophers have attempted to make it. Philosophers have extracted the concept of a person from its natural forensic context and tried to press it into metaphysical service, by asking questions about a supposed ontological category. The failure to make much progress with these questions is an indication that this appropriation was misconceived. Let us then drop the concept of a person from metaphysics and return it to its proper place in law and politics. We can still discuss the nature of animals, humans included, and ask about the identity through time of these entities—recognizing that they are essentially embodied minds—but we will not do so under the rubric “persons” or “personal identity.” There are no persons, as philosophers have employed the concept, primitive or nonprimitive.

The Question of Being

There is a huge gap at the heart of philosophy: the nature of existence. In fact, we might see the history of (Western) philosophy as a systematic avoidance of this problem. We have not confronted the question of being, not head on anyway. Even the formulation of the question has eluded us. What I mean can be explained by contrasting the question with two other questions, which have been amply discussed. One question concerns the analysis of “exists”: is it a first-level predicate or a second-level predicate? Is existence a property of things or does it reduce to a concept having instances? The other question concerns which things exist: do material objects exist, or other minds or numbers or universals or moral values? These are both good questions, but they are not the question I am interested in, namely what it is for something to have being. What does existence itself consist in—what is its nature? When something exists, what exactly is true of it? What kind of condition is existence? How does an existent thing differ from a nonexistent thing? The trouble with this question is that no answer suggests itself, so we get various sorts of nonanswer. One nonanswer is that existence is primitive and unanalyzable— all we can say is that an existent thing exists. Here we reach conceptual bedrock and nothing further can be said. But even if the concept is primitive, it should be possible to say something illuminating about it—not provide a classical noncircular analysis, perhaps, but at least offer some elucidatory remarks. If not, we must admit that we can say nothing informative about arguably the most basic and general concept in our conceptual scheme. This is what I mean by the strategy of avoidance: we avoid trying to say the unsayable—we admit defeat and call it indefinabilty. Or we just don’t go near the question, sensing its difficulty. Another way of dealing with the question is to answer it by some sort of human-centered theory—verificationist, pragmatist, or idealist. Thus existence is explained as a property that affects us in certain ways: it is what we must take account of, what we see and touch, what matters to us. If a thing doesn’t exist, we need not be afraid of it; but if it does, it must be reckoned with. The trouble with this kind of approach is that it fails to grasp the mind independence of existence: there could be a universe of existing things and no one in it to be affected by those things. Things would have being in that world, contrasting sharply with things that lack being, but this could not consist in how intelligent beings relate to those things. Being is an attribute that things have considered in themselves. A rock has being whether or not there is anyone around to bump into it—it exists in virtue of something inherent to it. In one sense, of course, we know just what existence is; we are quite able to distinguish existence from nonexistence. But when we ask ourselves what exactly it is that we know when we know that something exists we are brought up short—all we can do is sputter out unhelpful synonyms, like “being,” “reality,” or “thingness.” Existence is

212

Part IV

what is; nonexistence is what is not. Existence is what something possesses when it has made it into the world; nonexistence is the condition of banishment or exclusion from reality. Or again, existence is what God has created, while nonexistence is what God has passed over. Existence is a kind of fullness, we feel, while nonexistence is something thin and etiolated. But these comments, natural as they are, inescapable even, do little to produce any illumination on the question of being. We can say more about almost any other general concept we have (identity, necessity, goodness, and so on), but this concept leaves us completely stumped. Why? We can dilate on color, shape, mass, animal kinds, numbers, meaning, inference, and so on and on—but when it comes to the most basic aspect of things (and is even this the right way to talk?) we have nothing interesting to report. There is all the difference in the world between being and nonbeing, but for the life of us we can’t say what it is. When Descartes proved that he exists he achieved something substantial—there is a profound difference between a self that exists and one that does not—but he had nothing to say about what this prized existence is. It seems a vaguely good thing to have (witness the ontological argument), but what precisely is it? When the theist affirms that God exists, what is the attribute that seems so important to him? What difference does it make if God lacks this attribute? We think he would be lesser, ontologically impoverished, but we find it hard to say what the diminution amounts to. Existence adds no further specification of nature to a thing—no enrichment of properties—and yet no difference could be greater than that between being and nonbeing. It is like the difference between life and death, except that we can say what that difference amounts to. Existence presents itself as an ineffable something (certainly not a “nothing”). When I think of my nonexistence I feel a terrible vertigo, but I can’t spell out what this feeling is based on. Thus we are mute in the face of being, tongue-tied and perplexed. Now I am not about to announce an answer to the problem; my point has been that it is a genuine question that has been ignored, evaded perhaps. We can imagine early man noticing the difference between fictions, dreams, and illusions, on the one hand, and rocks, trees, and people, on the other. Just to give the difference a label he starts to speak of “being” and “nonbeing,” not giving too much thought to what this difference consists in.1 Then an early philosopher of the African savannah starts wondering 1.  It is tempting to imagine some along-ago ancestor of ours confronting the world one bright morning and thinking, with a tingle in her spine, “Existence!” The thought was so inarticulate as to be inscrutable. Thus was born a concept that is arguably our most basic, primordial, and baffling. It was our primitive reaction to the beautiful and terrible plenitude of things. Now it lives deep in our mental DNA, part of our inherited conceptual makeup. What is surprising is how little progress we have made in clarifying what it was exactly that our stunned and tingling ancestor had in mind. (We may also imagine, millennia later, another one of our ancestors waking up one morning and thinking, with a comparable tingle, “My existence!” This individual

Metaphysics213

what this idea of existence really comes to, and no answer suggests itself; so she moves uneasily on to less taxing topics (consciousness, free will). When philosophers started to organize and educate, a few thousand years ago, they would be tacitly aware of the problem but had nothing useful to communicate, so they left it alone, kept it off the syllabus, and discouraged students from raising it. And today we still have no idea what to say, beyond the two questions I mentioned earlier. All we can do is reiterate the word. Certainly science can do nothing to help us—there cannot be an empirical science of existence-as-such. If the question of being is the most fundamental question of philosophy, it is also the least well developed.

too was visited by a thought she found hard to articulate—her own existence had struck her.) The question of being is the question of what trait of reality such a basic thought records. We might attempt to regain our ancestor’s tingling state of mind by paying attention to the experience of waking up from a deep sleep: for in this experience we can become conscious again of the pressing reality of existence, as if encountering it for the first time. Existential thoughts tend to occur at such a moment—as in, “Existence, I remember you, nice to see you again” or “Bloody existence!” Normally the concept of existence hovers in the background of our thoughts, but sometimes it strides to the center. I like to gaze at trees and let their bulky existence sink into me—if only I knew what it was that was thus sinking in!

Science as Metaphysics

Nomenclature matters. It shapes the way we think. The word “metaphysics” continues to trouble and flummox. As every student of the history of philosophy knows, the work of Aristotle’s that is now titled Metaphysics was not so called by him; he did not even use that word to describe his field of study. It was an early editor of Aristotle’s works who coined the name and titled a collection of fragments by Aristotle with it. The word “metaphysics” simply means “after the Physics,” another work of Aristotle’s. This can mean either “the work Aristotle wrote after writing the Physics” or “the work to be studied after studying the Physics.” It doesn’t mean anything essentially connected to physics by definition—such as “that which transcends physics.” Nor does the word “meta” mean the same thing as in “metalanguage” or “metapsychology.” There is nothing in it to imply that metaphysics is somehow “above” physics—still less that anything mystical or religious is meant. If Aristotle had written a work of biology after writing the Physics, then that work would have been called “metaphysics”—the book that comes after the Physics. And the name might have stuck, with biology now known as “metaphysics.” The description “the book that follows the Physics” is a nonrigid designator: in other possible worlds it denotes (books about) biology, chemistry, history, logic, ethics, or animal husbandry. In fact, it could even denote physics, since Aristotle might have written another book on physics after writing the Physics. The subject matter of the book actually so named has nothing to do with the title, which is entirely arbitrary; yet to many people its descriptive content still carries weight. In philosophy we still refer to that subject matter with the term that the intrusive editor foisted on Aristotle’s work; no doubt Aristotle would have protested at this prosaic and prissy title for his treatise. Imagine if he had written the Physics after writing the Nicomachean Ethics: then that early literal-minded editor might have called his book on physics Metaethics! Despite the survival of this piece of linguistic legerdemain, we do have another term for the same area of study: “ontology.” Both terms gain their meaning by contrast with another term: “epistemology.” We need a term to refer to an area of study that is not epistemology, so that we can distinguish studying X from studying knowledge of X. This is because in philosophy we study both things, and it is easy to confuse the two sorts of question. It is not so in other areas of study: physics, chemistry, biology, psychology, history, and so on. These areas don’t study both (say) the physical world and our knowledge of it—the latter is left to philosophy of science or perhaps psychology. If physics did study both things, it might be necessary to introduce terminology to mark the distinction: say, “entity physics” and “epistemic physics.” But in philosophy we do study both things, so we need terms to contrast the two branches of study—and “metaphysics” and “epistemology” are the labels that have stuck, despite the odd origins of the former term. We also have “ontology” to fall back

216

Part IV

on in drawing the needed distinction; so we routinely tell our students that “ontology” refers to the study of being/reality/existence, while “epistemology” refers to the study of knowledge of being/reality/existence. Aristotle’s book could have aptly been titled Ontology, which is both descriptively accurate and not likely to mislead. As it is, we must contend with “metaphysics” as the name of the area of study in question, despite its shortcomings. But now a new question is raised, and a very reasonable one: if “ontology” just means “the study of what exists” or “the study of things,” as opposed to the study of knowledge, don’t the sciences qualify for that label? Doesn’t the physicist study the existing things of the physical world? And similarly for all the other sciences: don’t they all study a certain class of existing things—biology, astronomy, psychology, and so on? There are various entities in reality and the various sciences study the nature of those entities—planets, organisms, subjects of consciousness, and so on. Isn’t a scientist by definition an ontologist? The answer must surely be yes: the scientist studies the order of being, or a certain category of beings. He or she wants to know what kinds of being exist, how they should be classified, how they work, what laws or principles govern them. Science is therefore a kind of ontology—a systematic study of what exists in the world. It certainly isn’t epistemology, or axiology. It is the study of what is, why it is, and what it is. Science is the study of being (not the study of nonbeing). But then, granted the synonymy of “ontology” and “metaphysics” (as that term is now understood), science is also metaphysics. There is no contrast between science and metaphysics; science is a special case of metaphysics. The physicist is a metaphysician (= ontologist), quite literally, even when his concerns are thoroughly of this world. Theories of motion, say, are metaphysical theories—because they are ontological theories (not epistemological theories). Darwin had a metaphysical theory of life on Earth. There are metaphysical facts, like the rotation of the Earth or the boiling point of water. Philosophers also do metaphysics, of course, but they do so in the company of scientists: we are all practicing metaphysicians, for we all study being. We all do what Aristotle was doing in the book he wrote after writing the Physics. We study objective reality in a rigorous and systematic way, aiming to produce a general picture of things, seeking to keep bias and human idiosyncrasy out of it.1 1.  Scientists until recently were called “philosophers”—lovers of wisdom. Their kinship with the people narrowly called philosophers today was recognized. Latterly, the designation attracted the qualifier “natural,” so that we had “natural philosophers” as well as their colleagues, the plain old “philosophers.” There is really nothing wrong with calling contemporary scientists “philosophers,” despite obvious differences between the fields. I am suggesting that it is the same with “ontologist” and “metaphysician”: these terms also apply quite broadly, semantically speaking, and there is good reason to allow them their full scope. There will be plenty of time later to note divisions and distinctions within the broad category. We all can be said to pursue wisdom

Metaphysics217

This is not to deny any distinction between the kind of metaphysics (ontology) that philosophers do and the kind that scientists do. There are all sorts of distinctions between the kinds of metaphysics the various students of the world engage in— physicists or biologists, chemists or philosophers. No doubt every field differs from all the others in some way. There are many ways to be an ontologist, i.e. metaphysician, though that is what we all are. It is a matter of controversy what constitutes the philosophical kind of ontologist—especially what kind of methodology he or she adopts. Some see themselves as continuous with the scientific ontologists, perhaps arranging their several results into a big perspicuous ontological map. Some rely on the method of conceptual analysis to further their ontological goals. Others appeal to a special faculty of ontological intuition (they tend to be frowned upon by their tougher-minded laboratory-centered ontological colleagues). Aristotle understands his enterprise as differing from that of other ontologists merely in respect of generality. Where the physicist investigates substances of one kind—physical substances—the philosophical ontologist investigates the general category of substance. Where the chemist looks for the cause of particular chemical reactions, the philosopher looks at the nature of causation in general. These restricted ontologists want to know the nature of particular physical and chemical substances and causes; the philosophical ontologist wants to know the nature of substances and causation in general. They are both studying the same thing—being, reality—but they study it at different levels of generality. Thus philosophical metaphysics is fundamentally the same kind of enterprise as scientific metaphysics—though, of course, there are differences of method and scope. All are correctly classified as metaphysics (not epistemology or axiology). That is the right descriptive nomenclature to adopt. I therefore invite my colleagues in the sciences to share the label “metaphysician” with the philosophers, as well as the safer-sounding “ontologist.” The label simply serves to classify them more generally than their field-specific labels, and also than the term “scientist” (itself a recent invention). We are all metaphysicians (including mathematicians, who are interested in mathematical being—numbers, sets, geometrical forms). For we are all students of what is. I hope the scientists welcome the label,

(“philosophy”) about being (“ontology”) in the manner of Aristotle (“metaphysics”). Some of us do it “scientifically,” others do it “philosophically”; but we are all concerned with what is. (I also hold that philosophy can be described as a science, so that ontology and metaphysics count as science. Thus philosophers are scientists and scientists are philosophers, i.e., ontologists and metaphysicians. This way of dividing disciplines up strikes me as much healthier and more revealing than the usual exclusive divisions. To mark distinctions we can speak of “empirical philosophy/ontology/metaphysics” and “conceptual philosophy/ontology/metaphysics”—or any terminology that fits your view about the nature of the disciplines so labeled.)

218

Part IV

with all its resonance and impressiveness (yes, humble botanists, you too are metaphysicians!).2 Perhaps, too, this taxonomic unification will bridge certain gaps, break down certain barriers, and foster mutual respect. If I were running a university, I would have a Faculty of Metaphysics that included all the sciences as well as philosophy (but not sports or medicine or law or English literature). Faculty and students would be required to know the origins of the word “metaphysics,” as well as its contemporary philosophical sense. They would be encouraged to converse with one another using the term. We would all be members of one big happy intellectual family. World peace would assuredly soon follow. 2. I also hope that bookshop managers take my strictures to heart: the “metaphysics” section would then have a quite different content, including science, but excluding mysticism, new ageism, astrology, and so on. This latter section might be relabeled “antiphysics” or “alternative physics.”

Logic without Propositions (or Sentences)

Logic has been presented for a hundred years or so as a theory of the logical relations between propositions. Propositions have entailments and figure as the premises and conclusions of arguments. Not much is said about the nature of propositions in the standard textbook explanations of logic, but we are to assume that they correspond to the meanings of sentences—declarative sentences. So logic deals with representational entities—things that stand for states of affairs in the world. It does not deal with states of affairs themselves—with objects and properties. Sometimes talk of propositions is “eschewed” (Quine) and sentences are made the subject matter of logic, construed as marks and sounds, or some such. Then we hear what is called “propositional logic” described as “sentential logic.” If we wanted to go one stage further in the direction of concreteness, we could redescribe propositional logic as “statement logic” or “utterance logic,” where these are conceived as actual speech acts. Thus we would investigate the logical relations between speech acts. It is the same for what is called “predicate logic”: logic investigates the logical relations between predicates, especially as they interact with quantifier expressions. We are still investigating sentences, but we analyze them into predicates and quantifiers. If we don’t like the talk of predicates (bits of language), we could rename this branch of logic “concept logic”: then proposition logic and concept logic would both deal with what is expressed by language, while sentential logic and predicate logic address themselves to linguistic expressions. No matter how we formulate it, logic is conceived to operate at the level of representational entities, with logical relations defined over these entities. Logic is essentially concerned with the discursive. So conceived, modern logic is “the logic of (discursive) representations.” Consequence, consistency, and contradiction are all regarded as relations between sentence-like discursive entities (propositions, sentences, statements, assertions, utterances). Premises and conclusions of arguments are precisely such entities, and the laws of logic are the laws of the logical relations between these entities. But two points about logical laws call this representational conception into question. The first is that we presumably want logical laws to apply to worlds in which there are no representations. Suppose that no representational beings had ever evolved in the universe, so that there are neither sentences nor propositions (I will ignore Frege-style Platonism about propositions)—there is no language and no thought. Then logical relations defined over representations will not exist in that universe; there will be no logical laws of this kind. But will there be no logical laws of any kind? Surely not: the universe will still be governed by the laws of logic, as they are traditionally conceived. Contradictions will still be impossible, by the laws of logic: but they will not be defined over anything propositional. Logical laws like this are no more language dependent than natural laws, such as the law of gravity. We can state logical and natural laws by means of propositions, but the laws themselves don’t concern propositions. The laws

220

Part IV

can exist without the existence of any statement of them. So logical laws are not inherently propositional: they can hold in a world in which there are no propositions (a fortiori for sentences and speech acts). The universe would be subject to the laws of logic even if no thinking beings ever came into existence. The second point is that the traditional way of formulating logical laws does not make them about propositions or sentences. Thus: “Everything is identical to itself”; “Nothing can both have a property and lack it”; “Everything either has a given property or lacks it.” In stating these logical laws, no mention is made of propositions or sentences; the subject matter consists entirely of objects and their properties. There is thus no need to invoke propositions when stating logical laws; and such laws can clearly hold in a world without representations—you just need objects and properties, with logical relations defined with respect to them. So are there two kinds of logical law—laws of propositions and laws of objects and properties? That seems unappealing: one would like a uniform account of what a logical law is. The same goes for nonstandard logics, like modal or deontic logic: they also hold in worlds that contain no propositions (or sentences). If we add to the traditional three laws of logic so as to include further logical truths, such as Leibniz’s law of identity or the logical relations between colors, then again we have logical laws that are not defined over propositions—they concern the logical nature of the identity relation and of colors themselves. They deal with logical necessities that are not formulated by reference to propositional entailments: it is a logical truth about identity, say, that (in addition to being reflexive, transitive, and symmetrical) if a is identical to b, then a and b have all properties in common. Again, we talk here only of objects and properties (or relations), not of propositions about them. These are de re necessities, not de dicto necessities. Identity itself entails that identical objects are indiscernible, not propositions about identity—just as having a particular property itself entails not having the negation of that property, not propositions about the property. Logical facts, like other facts, obtain independently of discursive entities like propositions or sentences. Rather than accepting that there are two kinds of logical law, it would be better to demonstrate some kind of relationship of dependence between them. It seems too much simply to deny that propositions enter into logical relations, since that would be to condemn standard logic as completely misguided, based on an outright falsehood. Instead, we could try to see its entailments as derivative from deeper logical laws that are not inherently propositional: thus propositions have “derived logicality.” But how do we set about doing that? I propose that we reconceptualize the matter along the following lines. Suppose we accept an ontology consisting of particulars and universals (objects and properties); then we can distinguish the following three areas of investigation: (i) which particulars instantiate which universals, (ii) what the nomological relationships are between universals, and (iii) what the logical laws governing universals are. That is, there are three sorts of fact about universals: first, which objects fall

Metaphysics221

under them, how many, and so on; second, what laws of nature apply to universals (e.g., the laws of motion); third, what logical characteristics universals have. Each of these questions is about universals themselves, not about propositions or concepts. We are interested here in the third question, but it is worth observing how it relates to the other two questions, which are clearly not at all concerned with propositions or sentences. And the answer we would give will reflect the nature of the question: we will refer only to universals and their inherent logical relations (though of course we will be using propositions or sentences to do so). These relations, I suggest, will be of four basic kinds: identity, exclusion, consequence, and combination. Logic is then fundamentally about these four basic logical relations—with proposition-centered logic depending on the more basic logical facts. The logical relation of identity is captured in the law of identity for universals (also objects): every universal is identical to itself, and to no other universal. Then we will detail the logical properties of identity, noting also that identity is a necessary relation. None of this concerns propositions or sentences about identity, though there will be consequences for identity statements of familiar kinds. By exclusion I mean the way one universal excludes others from being instantiated in the same object—any of which is incompatible with the first. Thus being square will exclude being not square, say, by being triangular or circular. Every universal necessarily (logically) excludes other universals—that is a logical law. This is a de re necessity, not a truth about concepts or predicates: it could obtain without there being any concepts or predicates. By consequence I mean the way one universal can be sufficient for another: it is sufficient for being an animal that something is a cat, sufficient for being a man that someone is a bachelor, sufficient for having a successor that something is a number. One universal necessitates another, and perhaps another in turn. Logic (in a broad sense) traces out these consequence relations. By combination I mean logical properties of collections of universals: for example, if an object x instantiates a collection of universals U, then x instantiates each member of U; and if an object x instantiates a given universal F, then x instantiates F or any other collection of universals (these laws are intended to correspond to the standard rules of conjunction elimination and disjunction introduction). The idea here is that we can move from facts about collections of universals to facts about specific universals, and from facts about specific universals to facts about collections of them. Intuitively: if x instantiates F and G, then x instantiates F; and if x instantiates F, then x instantiates F or G. Here we logically link objects with universals considered as members of collections. Objects can be in the intersection of two universals (F and G) and be in the union of two universals (F or G). All these logical laws are stated over objects and properties. The claim then is that this is the metaphysical basis of logical laws as they are stated over propositions. It is in virtue of the former laws that the latter laws hold. It is fairly obvious how this goes: we just need to make a step of semantic ascent. Thus: if being F necessitates being G,

222

Part IV

then “x is F” entails “x is G”—and similarly for exclusion. The logical laws of “and” and “or” fall out of logical laws concerning objects and properties, as just outlined. The law of existential generalization is based on the fact that if a particular object instantiates a universal, then something does; and the law of universal instantiation is based on the fact that if everything instantiates a given universal, then any particular thing does. In the case of “not,” used as a sentence operator, we can take negation as applicable to universals themselves, so that not-F is itself a universal. Then we can interpret “not-not F” as meaning “the negative of the negative of F,” that is, F. Alternatively, we can construe negation as equivalent to “belongs to the complement of F” (e.g., “x is not red”). What we are doing is simply taking negation to apply to properties, not concepts or words; and similarly for conjunction and disjunction. An object can have the property of being F and G, or the property of being F or G, or the property of being not-F. All the standard so-called sentence operators have a more fundamental interpretation as operations on universals, forming complex universals from simpler ones. There are then logical relations between these universals, and hence logical laws. This allows such laws to obtain in worlds that lack language or anything representational. It makes them de re, not de dicto—about reality, not our description of it. We could express all this by speaking of states of affairs, but I think we get the basic ontology right by sticking to talk of objects and properties (particulars and universals)— these being what states of affairs are all about. Objects and properties have logical laws governing them, on this conception, as they have natural laws governing them, and as they form particular facts about the distribution of properties in the universe. None of these facts depends on propositions or concepts or words. Of course, we can formulate propositions about these laws and facts, but they are not themselves constituted by anything internal to propositions. A logical principle stated at the level of propositions is thus derivative from the more basic level of the logic of universals. Predicates entail other predicates because the universals they denote or express themselves necessitate other universals—this being an entirely nonlinguistic matter. So-called predicate logic is really property logic, seen through the prism of language. Strictly speaking, predicates don’t have logical relations, except derivatively on properties. If there were no properties obeying logical laws, then there would be no predicate logic. If there were no universals that inherently exclude each other, then there would be no law of noncontradiction at the level of propositions or sentences. Words cannot inherently exclude one another, and neither can concepts, construed independently of properties (as, say, dispositions to assent, or bits of syntax in the language of thought). The things that stand in logical relations at the most fundamental level are objects and properties; any other logical relations are transmitted upward from that basis. It is meaning that transmits logic from its original home in the world to language. If we try to view meaning as cut off from objects and properties, then we lose logic defined at the discursive level. Objects and properties are “logic-makers” as well as “truth-makers”—they are

Metaphysics223

ultimately where logical truth and truth in general come from. Nothing is true but reality makes it so, as Quine observed—even logical truth. Frege described logic as “the laws of truth,” thus locating it at the level of truth-bearers (“thoughts,” in his terminology). But this very formulation points to a different conception, since truth turns on the condition of the world beyond representation— and likewise for logical relations. Just as a proposition is true in virtue of the way the world is, so its entailments hold in virtue of the way the world is—specifically, the logical relations between universals. The truth-makers are also the logic-makers. The laws of logic are not fundamentally laws of truth but laws of what make truths true—that is, the logical matrix in which universals are embedded: identity, exclusion, consequence, and combination. Logic does not exist independently of the world, as if confined to the level of propositions—as if it reflected the structure of human thought—but rather is immanent in the world, part of what constitutes it. It is not that we impose logic on the world, having first found it in thought; rather, logic imposes itself on thought, having its origin in the world beyond thought. The propositional calculus and the predicate calculus, as they exist today, are really encodings of a mind-independent logical reality, which exists outside of sentences and propositions; they are not the primary bearers of logical relations (the same goes for modal logic, etc.). This way of looking at things clearly depends on a robust ontology of properties or universals—they cannot be identified with predicates or even concepts in the mind, or else the contrast I am insisting on would collapse. The logic of universals would simply be the logic of predicates or concepts. Perhaps this kind of nominalism or psychologism about universals is part of the motivation for the view of logic I am rejecting; but I take it that such views should not be accepted uncritically, and indeed are very implausible—for how then could objects have properties in a world lacking words or human concepts? Once we accept the reality of universals, fully and unapologetically, the approach I am defending begins to look attractive, indeed unavoidable. This incidentally implies that the usual separation between first-order logic and second-order logic is philosophically misguided (though technically correct): we are essentially concerned with properties and their relations even at the level of first-order logic, because we need to interpret the predicates as denoting universals that form the basis of logical laws. Particulars and universals are the foundation of the whole logical edifice, even when we are not quantifying explicitly over the latter. Universals are ontologically basic and enter into all our thought: they are the original ground of logical laws, even when dealing with first-order logic. Frege opposed psychologism about logic—the idea that logical laws have to do with the mind (apart from being apprehended by the mind). To this end he fashioned his ontology of objective “thoughts”—a clear oxymoron. These thoughts were taken to exist independently of the mind and to precede the existence of the mind. I won’t argue against this position here, merely noting its extravagance; but I will say that I

224

Part IV

agree with the motivation behind it—we don’t want logic to depend on human constructions, whether psychological or linguistic. When logic is conceived as the deductive science of propositions there is a distinct danger of psychologism, but the way to avoid it is not to objectify propositions; rather, we should locate logic at a deeper level—in the world beyond thought. There is nothing at all psychological about universals, for a realist about universals—they exist quite independently of minds. They are the building blocks of reality, since there is no particular that precedes universals— there are no property-free objects. Thus psychologism is avoided by locating logical laws in the nonpsychological world of objects and universals, not (pace Frege) in a supposed realm of objective transcendent “thoughts.” It is a consequence of the position advocated here that some knowledge of extramental reality is a priori: for we know the laws of logic a priori, and yet these laws characterize the world beyond the mind. Again, this consequence may be part of the motivation for a propositional view of logical laws, because then we can confine a priori knowledge to the contents of the mind (with language reckoned to the mind). If we think of logical necessity as analytic, and construe analyticity as arising from language and concepts, then we will be inclined to suppose that logical laws arise from the inner nature of mental representations or words. But again, such views must not be accepted uncritically or assumed without acknowledgment—and upon examination, they are very problematic. I won’t undertake a criticism here, merely noting that we need to take seriously the possibility that some a priori knowledge just is knowledge of the structure of extramental and extralinguistic reality. We know from our grasp of the nature of universals that they have certain kinds of exclusion and consequence relations—however jarring that may sound to certain kinds of empiricist or positivist assumptions. We have a priori knowledge of logical laws, and these laws characterize the objective nature of universals—so we have a priori knowledge of the general structure of mind-independent reality. This is just the way things are, like it or not. Finally, does the notion of logical form rest on a mistake? Philosophers and logicians have been apt to speak of the logical form of propositions or sentences, but an astute follower of the logical realism defended here might protest that this is a category mistake, since logical form properly belongs to states of affairs, not to sentences or propositions about states of affairs. I agree with the spirit of this protest and admire its extremism, but I think it goes a bit too far. We can agree that universals themselves exhibit logical form, in the sense that they are arranged in a logically determined totality, as defined by identity, exclusion, consequence, and combination. But there is nothing to stop us from supposing that this form is reflected in the structure of propositions themselves. The subject-predicate form, say, is a reflection of the object-property form: two complementary elements in a relation of mutual entanglement (predication and instantiation, respectively). Nor is there any objection to selecting a class of expressions designated as logical constants, and then defining a notion of logical form on that basis

Metaphysics225

(though this may be more arbitrary than has been recognized). What is mistaken is the composite idea that logical relations depend on logical form and that logical form is an intrinsic feature of propositions, considered independently of reality. That is just the dogma of logical representationalism (to give it a name) stated another way. Logical relations, to repeat, cannot be defined purely over representations, as a matter of their very nature: so they cannot result from the logical form of representations. Any logical form that propositions have must be derivative from a more basic logical reality—the logical form inherent in the underlying universals. If my position in this essay is correct, we should stop talking of propositional and predicate logic (though we may still speak of the propositional and predicate calculus— this being a type of notation). For that gives the metaphysically misleading impression that logic is grounded in propositions or predicates, not in the logical order of the world itself. We have different symbolic systems for representing (a fragment of) natural languages, but logical reality itself has nothing essentially to do with these systems. Logical reality is external to such systems, being essentially not a matter of symbols at all (so “symbolic logic” is misleading too). Logical laws are grounded in the world outside of all representation, and it is the job of our systems of representation to reflect their nature as best we can. They may do so without claiming to be constitutive of logical laws. The laws of logic stand outside of any notation for representing them, though they can be stated in a system of notation, more or less perspicuously.

V Biology

Selfish Genes and Moral Parasites

How is altruism possible in a world of selfish genes? If the genes produce only organisms that favor their own survival, then it will not be possible for there to be organisms that act so as to benefit other organisms. If an altruistic organism evolved by chance, because of some random mutation, it would be at a reproductive disadvantage compared to a selfish organism—so it would be less likely to reproduce than its selfish competitor. There seems to be no room for altruism in the biological world. To be sure, this stark picture needs to be qualified somewhat, because of the existence of reciprocal altruism and kin altruism; but these exceptions do not alleviate the basic problem, since altruism of the kind we observe in humans is not confined to these very limited kinds of altruism. Nor does it seem plausible that genuine human altruism arises as a necessary by-product of some highly advantageous adaptation—as it might be, rational thought—since it is hard to see why something so disadvantageous would not be selected against, or why rational thought would necessarily give rise to altruism as a by-product. In any case, I am going to assume in this essay that these avenues for explaining human altruism are closed, so that we need some other account of what we observe (I also rule out group selection for the standard reasons). I will also take it for granted that genuine altruism does exist in the human species: that is, humans do sometimes act against their best interests or the interests of their genes in order to help others. I want to know how and why this trait evolved, consistently with selfish gene theory. If our genes are invariably selfish, why are we not (not counting reciprocal and kin altruism)? Why do we sometimes act so as to maximize the reproductive prospects of other people’s genes? Why, that is, are we moral? I begin with a well-known paper by Richard Dawkins, “Arms Races and Manipulation.”1 In this paper Dawkins makes the point that individual animals sometimes act so as to benefit the genes of other animals because they are manipulated into doing so. For example, the reed warbler is manipulated by the cuckoo into rearing its nestlings, at the expense of its own genetic legacy. The cuckoo’s eggs mimic those of the warbler, and the cuckoo nestlings do a good imitation of warbler nestlings. Thus the brain circuits of the warbler are pirated by the cuckoo and made to serve the interests of the cuckoo. The case is similar to the anglerfish that presents a lure to its prey, causing it to stray to within the angler’s strike range. Dawkins compares these strategies to advertising, propaganda, hypnosis, and drugs in humans: devices of mental manipulation that tap into susceptibilities in the object of manipulation. Instead of using physical force, as in standard predation, the animal attacks the other animal’s mind-brain, causing it to act in the interests of the attacker animal. It is not that the manipulated animal has a gene for the behavior elicited, so that there is some genetic 1.  In Richard Dawkins, The Extended Phenotype (Oxford: Oxford University Press, 1982).

230

Part V

basis within it for the behavior that serves the interest of the other animal; rather, it is made to act against its own genetic interest by the manipulating genes of the other animal. A very naïve observer might suppose that warblers are acting altruistically toward cuckoos, generously donating their precious resources to raise orphans, and then wonder why such an unselfish gene in the warbler is not selected against. Similarly, a naïve observer might admire the angler’s prey that so altruistically strays near its predator’s mouth and provides it with an easy meal. But in neither case is this really what is going on: the seemingly altruistic animal is simply being fooled and manipulated. It has no altruistic genes, yet it behaves in a way that superficially resembles an animal that has such genes. The case is not different in principle from ordinary predation: the prey has no altruistic tendencies toward its predator, as if offering itself up in self-sacrifice; rather, it is physically forced to become food. Dawkins’s point is that physical force and mental manipulation have the same genetic underpinning: genes that influence the behavior of others in ways that serve their own interests, either by proximal contact or distal manipulation. Some predatory genes cause grabbing behavior, while others cause the subtler strategy of mental manipulation. Neither should be interpreted as altruism on the part of the grabbed or manipulated. In principle, we can imagine extreme examples of such manipulation. Suppose that Martians come to Earth and wish to enlist our services in their own interests, irrespective of our own. They simply implant a microchip in every human brain and send in signals that recruit parts of the brain in such a way as to produce behavior beneficial to Martians. Overnight we change into Martian providers, sacrificing our own interests at every turn. A naïve observer might suppose we have become extremely altruistic toward Martians, maybe harboring a gene for such altruism that has just been expressed. But that would be quite false: rather, we are being manipulated, mentally coerced. A drug could do the same, as could mass hypnosis or very effective propaganda. This raises the question of why such manipulation is not more widespread in the animal kingdom. It is clearly a highly effective strategy: exploit the energy resources of other animals in the project of perpetuating one’s own genes, thus saving time and effort—a clever manipulator could have enormous numbers of offspring this way. The reason it is not more widespread and more ruthless is, as Dawkins explains, that there is an arms race going on between the exploiting and the exploited. The warbler evolves ever more sophisticated ways to perceptually discriminate its own eggs and nestlings from those of cuckoos, but the cuckoo gets correspondingly better at mimicking the warbler’s eggs and nestlings. The same is true of the anglerfish’s lure. There comes a kind of equilibrium point at which the target animal accepts its exploitation because to resist any further would be too costly relative to its losses. The warbler only has to raise the odd cuckoo and it would be an intolerable diversion of resources to improve its ability to detect impostors. It would be different if the cuckoo were squeezing out the

Biology231

warblers’ young entirely, but as things stand the costs are low enough to allow some degree of exploitation. Dawkins slots this point about manipulation into his theory of the extended phenotype. In effect, the cuckoo’s phenotype includes the warbler’s body, because its genes control this body by means of manipulation. It is no different from the snail and its shell, or the beaver and its dam: the environment is exploited to serve the interests of a particular collection of genes. Hence the “long reach” of the gene: the genes in one animal are causing behavior in another animal that serves their selfish purposes. No unselfish genes need to be postulated in any of this: the genes cause behavior that is entirely selfish relative to their interests—except that the behavior might be that of another animal. That other animal has no unselfish genes either, despite the fact that it acts so as to benefit the genes of another animal; it is simply being manipulated by genes sitting in another animal’s body. An appearance of altruism thus dissolves into selfish genes, manipulation, and the extended phenotype. There is therefore no problem of reconciling the warbler’s “generous” behavior with selfish gene theory; we can readily see how the warbler’s apparent altruism stems from rigidly selfish principles at the level of genes. Of course, the warbler is not aware that it is being manipulated, and may indeed experience a surge of parental pride in (unknowingly) bringing cuckoo offspring to adulthood; but it is being manipulated nonetheless. It feels no resentment at its manipulation, so successful is that manipulation. What we have here is a parasite species, the cuckoo, cleverly exploiting the resources of a host species, the warbler. The parasite gets its way by fooling the host. We can envisage other ways in which such parasitism might be brought about: the parasite might insert chemicals into the brain of the host, as with certain species of ant discussed by Dawkins, or it might engage in some kind of brainwashing, or it might use hypnosis. Any strategy will be favored by evolution so long as it subverts the host species into helping the parasite species. We might expect strong selection pressure in favor of these kinds of manipulation. Parasitism is in general a remarkably effective evolutionary strategy, which is why there are so many pesky parasites. But it is not necessary actually to occupy the body of your host—you can be a parasite-at-a-distance. An exceptionally naïve observer might marvel at the altruistic behavior of animals that kindly allow parasites to occupy their bodies, sharing their precious resources; but we all know that this arrangement is really the result of an arms race between parasite and host. The host animal certainly has no gene for “altruistically” tolerating its body-snatching parasites. And the same is true for those manipulative parasites-at-a-distance. My suggestion can now be stated, somewhat brutally, as follows: human altruistic behavior is the result of parasitic manipulation. I mean this both intraspecifically and interspecifically. We are moral parasites with respect to each other, and members of other species can be moral parasites with respect to us. We have no gene for altruism and such a gene would be strongly selected against, yet we act altruistically. Why?

232

Part V

It’s because we are manipulated into doing so, ultimately by the genes of others. The great merit of this theory is that we can avoid the problem of postulating an unselfish gene—there is no such thing, yet there are altruistic acts, motives, thoughts, and so on. The reason altruism exists in me is that manipulation evolved in you, and the reason altruism exists in you is that manipulation evolved in me. Manipulative genes were favored by natural selection, for obvious reasons; but altruism in the “host” was never favored by natural selection—indeed, it is disfavored by natural selection. It exists only because of an equilibrium reached in an arms race. The mechanism is the same as the warbler and the cuckoo. Thus generalized altruism of the kind we find in humans has a completely different explanation from altruism toward family members and the quid pro quo kind. There are genes for kin altruism and for reciprocal altruism, since these kinds of altruism benefit the genes of the agent; but general altruism (“stranger altruism”) does not benefit the genes of the agent—it benefits the genes of another agent. It arises only via the mechanism of manipulative parasitism. We thus resolve the puzzle of the existence of altruism in a world of selfish genes. To be more exact, we resolve the problem of reconciling altruistic-appearing behavior with a world of selfish genes. We can state this theory in terms of the extended phenotype. Just as the genes of an animal can reach out and shape its environment, as with beavers and their dams and cuckoos and their warbler hosts, so the genes of one human can shape the behavior of another human. When this happens, the shaped environment becomes part of the phenotype of the animal whose genes are doing the reaching out. Thus, in effect, the body of one human becomes part of the phenotype of another human, because the behavior of that body serves to further the interests of the genes that sit in another body. It is as if the other body acts as a protective shell for the animal that controls it. By means of mental manipulation, I can control your body into acting as a protective shell for my body, by getting you to act altruistically toward me. The (apparent) altruism of others toward me is just my genes extending their phenotype in the direction of other people’s behavior. Instead of physically using another body as a shield (not a very practical proposition), I manipulate that body into acting in a way that serves my interests. Shells and dams are obviously not inherently altruistic, caring about the animal they protect; they are just part of the extended phenotype of a bunch of cunning and selfish genes. According to the present hypothesis, people are not altruistic “out of the goodness of their genes”; they are manipulated into being so by selfish genes in another animal that has extended its phenotype beyond the boundaries of its own body. Thus every phenotype, confined or extended, serves the interests of selfish genes—despite the fact that some behavior is altruistic in the sense that it does not serve the interests of the animal whose behavior it is. For the behavior of one animal can be part of the (extended) phenotype of another. Your altruistic behavior toward me is my extended phenotype catering to the dictates of my genes. Genes are successful

Biology233

according to whether they produce phenotypes that enable them to reproduce, and extending their phenotype to include the behavior of others that benefits them is just another device for reproduction. Creating altruism in others, by manipulative methods, is an ingenious strategy for successful gene reproduction, extending one’s phenotype into the very will of another. How might we set about testing this bracing (shocking?) hypothesis? The first thing to note is that we are as a species exceptionally susceptible to manipulation. We are gullible, credulous, superstitious, prone to magical thinking, easily hypnotized, prey to advertisers, suckers for jingles and slogans, vulnerable to cults, and fond of drugs. It sometimes seems as if we are just aching to be manipulated. We long to hand mental control over to someone else. We can’t stand our own autonomy (decisions, decisions!). Why this should be is hard to say. Perhaps it partly stems from our extreme educability—in order to learn what’s new, we have to be very open to whatever comes our way. Perhaps it stems from some deep existential angst. Whatever it is, it is powerful and endemic. We are like exceptionally gullible reed warblers. Our buttons are eminently push-able. Today, with the advent of science and so on, we are perhaps more critical than we used to be; but surely for most of our evolutionary past we were incorrigible suckers and pushovers. Second, and surely connected, we are also highly manipulative creatures, always trying to gain influence over others. This just seems like an instinctive human tendency: try to get control of the minds around you. There is no point in doing this if people are not susceptible to manipulation, but if they are there is survival value in exploiting their weaknesses. Our social life depends on “mind control,” so we are skilled in the science and art of manipulation; and it is likely that this talent has a basis in our genes. Politics and religion, as we know them, surely owe much to human powers of manipulation and to our weakness for being manipulated. A third point is that for manipulation to succeed it must not present itself as manipulation. The warbler mother does not feel that she is being manipulated by the cuckoo; she is oblivious to that reality. She feels like a warm and caring mother, not a slave or puppet. If human altruism itself results from manipulation, it will not feel like manipulation. That would only produce resentment and reluctance. No, the best manipulator must create in his target the feeling that the actions in question are done freely, happily, with pride and self-esteem. The gene that manipulates the mind-brain of others the best is the one that makes the target enjoy what is in fact manipulation. Since pride is an emotion much favored by humans, an optimal strategy will be to instill a feeling of moral pride in the one being manipulated; the real nature of what is going on must be kept hidden (so it is with all manipulation). The cleverest gene for manipulation will make its victims relish their enslavement. In the case of altruism, we will value this trait very highly, despite the fact that it actually results from piratical parasitism. We are like the warbler mother who feels exceptional joy and pride in feeding that oversized

234

Part V

rather peculiar-looking member of her brood—the one that is ruthlessly manipulating her into neglecting her own family and its stock of precious genes. A further source of evidence concerns the equilibrium point in human altruism. We are only so altruistic. The warbler would not keep feeding cuckoos if their demands became too clamorous, because her genes would not reproduce in sufficient numbers; at a certain point her “altruism” would fade out. Similarly, if you were to demand enormous sacrifice on my part for a small benefit to you, which has a substantial impact on my ability to reproduce, then I would not survive to reproduce, and my tendency to extreme sacrifice would fade out. People will generally be altruistic only up to a point, or else the costs become too great. In fact, they will be pretty selfish most of the time. There may be a sucker born every minute, but nobody is a total sucker. So the theory predicts moderate degrees of altruism, firmly constrained by evolutionary demands on the altruist. But isn’t this pretty much what we find, statistically speaking? Yes, we can be manipulated into helping others, but after a certain point we resist the manipulation and assert our selfish nature. Our selfish genes will tolerate only so much unselfish behavior. They would prefer to tolerate no unselfish behavior, but their host organism is built in such a way that some degree of manipulation is unavoidable—it’s just too costly to eliminate all manipulation (that might undermine our ability to learn and be flexible, say). Some moralists have bemoaned the limits of human altruism, but (according to the present hypothesis) these limits result from a commendable desire to resist moral parasitism. We can take only so much manipulation before our selfish genes rebel. But is there any evidence of manipulative intent? Do we do things that can plausibly be construed as manipulative? I don’t mean actions that are consciously intended as manipulative, but rather actions that are functionally manipulative. I doubt that the cuckoo goes to the warbler’s nest with conscious manipulative intentions; rather, it has been designed by evolution to behave in ways that functionally manipulate the warbler. So, what behavior of ours might have such a manipulative function? A good place to start is the human baby. The baby craves attention, lots of it. It behaves in ways that secure that attention. It howls and cries, whimpers and moans. The parents come running and attend to its needs. The parent’s altruism here can be explained by kin selection, though there must also be an element of simply wanting those loud shrill noises to stop. There are circuits in the parent’s brain that are activated by the baby’s sounds and sights and which reflexively elicit altruistic behavior. Now suppose you wanted to manipulate someone into behaving altruistically toward you, but you are a baby no longer: what would you do? You might act like a baby—mimic the actions of a baby. You have had an accident, say, and are in pain, and you need help: how do you secure the requisite attention? A good strategy would be to tap into the brain circuits of other people that activate care of babies. Hence you commence to howl and cry, whimper and moan. Passersby, though equipped only with selfish genes, find their

Biology235

baby-caring brain circuits activated, and they rush to help, unaware that they are being manipulated, as you are unaware that you are manipulating (you aren’t faking the howls). Their selfish genes are screaming, “Don’t do it! We are the ones you should be caring about!”; but the vehicle of those genes is susceptible to manipulation, so it acts against its own genetic self-interest. It is the ploy of the cuckoo nestling with its gaping baby-warbler mouth. That is, it exploits a kind of mimicry. Why do we groan when we are in pain? It doesn’t much help with the pain, but it does let other people know we need help—and, according to the present hypothesis, it works because of manipulative mimicry. It is as if the baby-helping brain circuits overgeneralize, making them susceptible to more than distressed babies—to the baby stimuli themselves rather than the babies that produce these stimuli. The brain simplifies, and in so doing it becomes overly inclusive. Hence we are prone to manipulation by means of (unconscious) baby mimicry. Or think of advertising. Advertising, as everyone knows, is not just about informing the consumer of your wares; it is persuading him or her to buy them, by any means available. The advertiser uses any trick of human psychology to worm his way into your psyche, influencing your buying behavior. Sheer repetition is one good way to achieve this. It may not be in the interests of your genes to be so influenced, but your mind has so many blind spots, weaknesses, and irrational tendencies that your genes can be thwarted by clever advertising techniques, despite their ingrained selfishness. Similarly, “moral advertising” might exert control over your mind and behavior: repetition, use of music, subliminal cues, and emotional vulnerabilities. Just think of all the tactics that are employed when someone wishes to persuade you of a particular moral point of view: not all of these consist of above-board rational arguments (see below) but instead employ tricks of manipulation, emotional and other. Babies, notably, tend to feature heavily in advertising, both for goods and services and for moral positions. Charities can be very persuasive when it comes to eliciting moral emotions from people, and not all of this persuasion renounces manipulation. Language plays a pivotal role in mental manipulation and raises many interesting questions in its own right in the present context. It is very natural to suppose that language is used to achieve moral manipulation. It may also be used in moral reasoning (again see below), but it is surely part of the arsenal of the moral manipulator: rhetoric, intonation, repetition, poetry, sheer volume—all these can exert a hold on listeners. There are those smooth talkers, spellbinding orators, and hypnotic raconteurs. Perhaps language evolved, at least in part, as a tool for mental manipulation: we wanted to secure the altruistic cooperation of others, so we developed language as a means to bring this about. If cuckoos could talk, they might be hypnotic speakers to the ears of reed warblers. Dawkins suggests that all animal communication is manipulative and drug-like, especially mating calls. Clearly, the function of much human speech is persuasion, getting the listener to do as you want—and this may be achieved by any

236

Part V

means that the language centers of the brain allow. Hence we have such speech acts as complaining, moaning, wheedling, haranguing, exhorting, pleading, demanding, insinuating, subtly suggesting, brow-beating, and shouting. We want the listener on our side, well disposed toward us, primed for altruism. A tremendous amount of ordinary human communication is about moral persuasion, by hook or by crook. Language brings moral parasitism to a new level of sophistication. The speaker is dispensing verbal intoxicants to his listener, hoping these will serve his manipulative purposes (not always consciously). To put it differently, the genes for language are also in the business of psychopharmacology: they want to drug the listener into serving their selfish ends. And the listener is quite an addict. Language is a mind-altering substance. I am speaking metaphorically, of course (or am I?). I am speaking of the functional properties of language with respect to persuasion and manipulation. I am suggesting that these play a role in eliciting altruism from others. But someone might object that this picture of language is far too cynical: don’t we also, and primarily, altruistically share information with other people? Isn’t that what assertion is? This raises an interesting theoretical question, which I shall call the “puzzle of linguistic altruism.” How could linguistic altruism evolve and persist over multiple generations? Like all altruism, it involves expenditures of energy that might be spent on more selfish tasks, like growing larger gonads (Dawkins’s favorite example). Why don’t I just keep my knowledge to myself instead of strenuously broadcasting it? Why do I answer other people’s questions at all? Reciprocal altruism and kin selection might explain some of my assertive behavior, but surely not all. Maybe we can subsume linguistic altruism under our general theory of altruism, regarding it as elicited by manipulation—so that other people are effectively parasites with respect to my stock of knowledge. My motives are pure as I verbally share my knowledge, even if my listeners are manipulating me into wasting my energies informing them about things. But there are other possible explanations that strike me as more credible. One is that my aim is less to enlighten others than to influence them: I want them to share my way of seeing the world so that their actions will conform better to my desires. What if telling them the truth will seriously harm me? My genes will program me to make only such assertions as will enable me to stay ahead of the game, one way or another. If they did not, they would be wiped out in succeeding generations. My genes will ensure that I have a tendency to shape my speech acts so as to respect their interests. Alternatively, we might, even more cynically, regard truthful assertions as a necessary prelude to more sinister linguistic intentions: I tell you the unvarnished truth a few times, thus securing your trust, and then I feel free to tell lies when it serves my selfish purposes to do so. My sharing of information is not as altruistic as it might appear, being merely a means to a selfish end. At any rate, there are clear problems in the idea that speech consists of altruistic acts of sharing—for the reasons that count against any claim of

Biology237

unqualified altruism. Energy spent on one thing is energy not spent on another thing; so persistent and energetic altruism will always be selected against, verbal or nonverbal. Language is just one kind of human behavior with respect to which this inexorable biological law applies. We need a conception of language (in its pragmatic dimension) that makes it biologically possible. We need to know, that is, how conversation can exist in a world of selfish genes. It can, but only if we rightly understand how linguistic altruism arises. The old “good of the species” answer is clearly not going to cut it, since no animal ever does anything for the good of the species. Things must be more convoluted and less goodie-goodie. I cannot resist making a point about God, which could scarcely sound more cynical. How does the idea of God enter human consciousness? Well, here is a theory deriving from the position defended in this paper: the idea of God arises as a tactic invented by selfish genes to further their manipulative purposes. If you want to manipulate others into altruistic acts, it would be helpful to give them the idea that their acts will be rewarded or punished by an omnipotent deity. So a really clever gene will not only manipulate others in the way described, as well as make them feel good about being so manipulated; this super-clever gene will also give them the idea of a God that cares about their moral status and rewards their altruistic acts. That would confer a certain authority on the impulses felt by the individual being manipulated. That is, this superclever manipulative gene will produce brain states that express the idea of God, so that the altruism it seeks will be reinforced at the highest level, so to speak. Imagine if cuckoo genes managed to produce in warbler brains, not only care of their offspring, accompanied by parental pride in so doing, but also the idea of a warbler God that particularly favors feeding the big awkward nestlings in the nest. That would really seal the deal. I do not claim to know that this is how the idea of God arose in human brains in the course of evolutionary history; I merely assert that it is a logically possible way for that idea to have arisen—and a way that makes a certain amount of theoretical sense. Certainly, the hypothesis that religion arose because of the machinations of selfish genes, hell bent on parasitic manipulation, has a certain perverse appeal. Even if the idea of God (or gods) arose in some other way, say in the explanation of natural phenomena, the genes could still recruit the idea in the service of their manipulative agenda, thus joining God (or gods) to morality. Referring to God functions as a means of moral manipulation. The view I have defended (or at least mooted) about the place of altruism in the natural biological world may strike the reader as extremely cynical and indeed nihilistic. Am I not saying that morality is a huge confidence trick, a drug-induced convulsion, a complete and utter delusion? Should we then try our best to resist the manipulations of those devious selfish genes and abandon altruism altogether? You may be relieved to hear that I think not, because I believe that what those genes manipulate us into

238

Part V

doing is in fact morally right. Suppose that the warbler mother comes to realize what the true situation is—that she is being tricked into taking care of another bird’s offspring. Should she then elbow it out of the nest and let it starve? No! Because that baby cuckoo needs help and its biological mother is nowhere to be found. The right thing to do is feed it. The case is no different from taking care of an orphan that has been smuggled into your house and left there. In general, it is a good thing to be altruistic. All I have maintained so far in this essay is that we need to be able to reconcile the existence of altruism with selfish gene theory and that the parasite-manipulation theory seems the best option. I have said nothing about the status of morality itself. Given, however, that what we are manipulated to do is (at least sometimes) the right thing to do, we can say that we have here a happy coincidence. When the genes of others amorally manipulate you into serving their interests, they are causing you to do what is in fact morally right. Moreover, you can see that it is right: you judge it to be right, quite rationally. This moral judgment can itself motivate. So there is a kind of double motivation at work, or a convergence of causal-explanatory factors: on the one hand, you judge your altruistic action to be right, and this can motivate you to perform it; on the other hand, you are being distally manipulated by a moral parasite to perform the same action. Thus the action is right, you judge it to be right, and you are thereby motivated to perform it, and you are (unconsciously) manipulated into performing the action. In the case of the cuckoo and the warbler, presumably only the second sort of cause is operative, since birds don’t make moral judgments. But in the human case the etiology is complex. There is a substructure of genes that determine the basic interpersonal dynamics and then a superstructure of conscious moral judgments. What is important, given the aims of this essay, is that altruism is not inconsistent with biology—we can see how altruism might evolve and persist, quite robustly. Structurally, altruism has the logic of parasitism at-a-distance. As far as I can see, this is quite compatible even with a fully Kantian approach to moral reasoning and moral truth. There are just two levels of moral psychology at work in determining moral action. But the compatibility of the present theory with moral realism does not imply that everything typically included in morality deserves a place there. If the theory is right, we might expect some areas of traditional morality to be infected with ideas deriving from its ultimate basis. Excessive emphasis on notions of meekness, humility, obedience, self-sacrifice, authority, and the like should be regarded with suspicion, because these are redolent of manipulation and subservience. The manipulation model makes of each person a mere means for the ends of others (a “doormat”), but this should be resisted as morally unacceptable. We don’t want our official morality to include bits of detritus surfacing from the biological substructure. If the cuckoo regards the warbler as a mere means to its ends, demanding from it meek surrender and self-effacing humility, even characterizing these as deep virtues, then I think the cuckoo is morally out of line.

Biology239

It would do better to think of the warbler as an autonomous agent that the cuckoo’s genes have designed it to manipulate. Some regret should enter into its feelings for the surrogate mother of its children. The cuckoo should respect the animal it manipulates. In the case of humans, of course, the manipulation goes both ways, since I manipulate you and you manipulate me. Still, it would be wrong to elevate the fact of manipulation into some sort of moral virtue, on either side. There is no virtue in being manipulated and no virtue in manipulating. A theory of how altruism is biologically possible cannot be converted into a theory of what morality should contain.

The Evolution of Color

According to projectivist views about color, color properties do not precede color vision. It is in virtue of color experiences in perceivers that things come to be colored. Color is mind dependent. I will assume this view here. My concern is with the consequences of this view for the theory of evolution. Sensations evolve: they result from genetic mutations acted upon by natural selection. They are adaptive traits, just like bodily organs; they are there for a reason. Thus the feeling of pain has biological utility (as a warning sign and motivator), and even the phenomenological details of the sensation will have been subject to fine-tuned natural selection. The feeling of orgasm must be similar: it is the way it is for strictly biological reasons—as the best method the genes have for securing their survival. These traits are internal—sensations occur “inside” animals. They are part of what we might call the restricted phenotype: they belong with eyes and stomachs and other organs of the body. Their design follows the general rules of trait selection: they are solutions to evolutionary problems, more or less efficient, constrained by the past, and handed down through the generations. Tastes and smells are similar, only now it is external objects that have tastes and smells—whereas it is organisms that have sensations. Things taste and smell as they do for strictly biological reasons. As organisms evolve, tastes and smells come into being, though they are tastes and smells of external things. They are projected, not inherent; relational, not intrinsic. And it is the same for colors: objects are colored but they are so only in virtue of the existence of evolved organisms that see them that way. There were no tastes, smells, and colors before sentient organisms evolved. Thus we can say that colors evolve—like sensations of color. Red objects, say, came into existence (qua red) by means of genetic mutation and natural selection. If red were not an adaptive color for organisms to see, then it would not have evolved: red is a biologically useful color to project. In general, the colors organisms see must have been specifically selected for their adaptive value—presumably because of their ability to provide sharp contrasts (among other things). In other words, natural selection operates on colors—even though it is external objects that are colored (in virtue of color vision). If the genes for color vision were to mutate so as to produce a completely different set of perceived colors, and these new colors were more adaptive than the ones we now see, then we would find a selective pressure in favor of these mutated genes. The world contains the colors it does because of the selective pressures operating on organisms. That may sound odd, because colors, unlike sensations, are not properties of organisms—they are properties of external objects (though projected there). They are not part of the organism’s restricted phenotype, that is, existing within its individual boundaries. But here we must remember the notion of the extended phenotype: it is

242

Part V

not just the individual body type that is selected, but also what that body produces environmentally. Thus the beaver’s dam and the bird’s nest are part of these animals’ extended phenotype: natural selection works on the combination of body and external product, so that good dams and nests are favored, along with good limbs and brains. Body plans and behavioral capacities evolve, but so do the adaptive products of those things—she who builds the better dam or nest is most likely to pass on her genes. The unit of natural selection is the extended phenotype, not merely the restricted phenotype. And now the point I want to make is that colors are part of an organism’s extended phenotype. They are products of minds and brains, but they exist outside the boundaries of the organism—hence they belong to the extended phenotype. They evolve by the same rules as bodies, but they are not parts of bodies. As dams and nests evolve, so colors evolve (and sounds, tastes, and smells). Colors are created by genes, ultimately, and the better the color the more chance it has of surviving. The colors we see now in the world have stood the test of evolutionary time. The colors we project are the colors that have passed selective muster. Red, for example, has proved itself a highly adaptive color, along with the usual color spectrum that we see. Wishy-washy or indistinct colors might not do as well in the fight to survive—just as feeble or painful orgasms would not be apt to survive, in contrast to more pleasurable ones. In the case of colors, we might say that they belong to the projected phenotype—which is a subclass of the extended phenotype. The organism builds its physical environment (sometimes), but it also constructs its perceptual environment. It constructs a phenomenal world. This world consists of colored objects (among other things); so colored objects evolve (though not the matter they are made from). Evolution thus operates selectively on phenomenal worlds, as it operates selectively on limbs and brains. Whole species of phenomenal world come into existence by mutation and natural selection, and they can also go out of existence with extinctions. Some properties of objects do not evolve—those that precede and are independent of organisms—but some do. Projected properties do, because they reflect the perceptual receptivity of organisms. When I say that colored objects evolve, I obviously don’t mean that the objects themselves have evolved by natural selection; I mean that their having the colors they have is a result of natural selection—there are no colored objects on the planet without natural selection. It is because colors are both properties of objects and projected by the mind that they belong to the extended phenotype of the organism. If they were psychological properties of organisms, they would be part of the restricted phenotype; and if they were inherent properties of external things, not projected properties, they would simply be part of the nonevolving environment. The point is not that experiences of color evolve—that follows simply from the fact that sensations evolve. It is rather that the colors that are the objects of such experience evolve—as dams and nests evolve. Dams and nests are adaptive traits for beavers and

Biology243

birds, and colors are adaptive traits for visual organisms—though these traits all belong in the extended phenotype. As adaptive traits, they are subject to evolution by natural selection. The colors that exist in our world are those that have survived the rigors of natural selection. It may help in understanding the point if I make the ontology of colors clear. The projectivist view of color is naturally associated with a dispositional theory of color ascriptions: objects have the colors they are disposed to produce experiences of—an object is red, say, in virtue of being disposed to produce experiences of red. This need not imply that colors are identical to such dispositions: we can hold that colors supervene on these types of dispositions, without being identical to them. Then we can say that colors are simple qualities of objects, not in themselves mental, but that they are instantiated in objects in virtue of dispositions to produce color experiences. Color experiences clearly evolve, and objects have dispositions to produce such experiences only in virtue of the existence of evolved organisms; but it is a further claim that colors themselves evolve—conceived as simply qualities of external things. We thus have the nontrivial thesis that objects come to have simple color properties in virtue of evolution by natural selection. A mutation caused some object to look red, and hence (by projection) to be red; then natural selection favored that way of seeing, and hence what is seen. Colors came into the world by the same mechanism as hearts and kidneys. To put it paradoxically: it is adaptive for us to be surrounded by colored objects. Does this view of color generalize? It certainly generalizes to other secondary qualities, but it may also generalize to qualities traditionally regarded as primary qualities, such as shape and motion. For it may well be that such qualities do not rightly belong in the austerely objective world described by physics; rather, they reflect our evolved sensibility—how we have been programmed to experience the world. If so, shape and motion—the perceptible qualities we experience—are also evolved (and evolving) entities. Objects have shape and motion, as we perceive them (if not in the austere world of physics), but they do so only as a result of our evolved sensibility; so they too are subject to evolution by natural selection. The whole world of colored objects with shapes and in motion is caught up in the evolutionary process: it originated in that way and its survival depends on the usual evolutionary pressures. In other words, the world we experience is an evolutionary product—like limbs and brains, dams and nests. The empirical world is really part of our extended phenotype.1 So the extended phenotype extends quite far into reality (though not all the way). When did 1.  But not the whole of reality, since there is an objective world out there that owes nothing to our evolved modes of experience. It is the world as it appears to us that belongs to the projected extended phenotype—but this world contains real (though projected) properties of things.

244

Part V

this naturally evolved empirical world come to exist? It probably had its early origins in the projective mind of a sentient fish a billion or so years ago. Then it was that the world began to be clothed in color (and maybe shape and motion, as we conceive them in common sense). It has been evolving ever since, becoming ever more complex and subtle. It will cease to exist when there are no more sentient organisms projecting properties onto the world. Colors will eventually become extinct, joining the dinosaurs.

The Language of Evolution

Despite the scientific solidity of the theory of evolution, it is a terminological mess. The usual descriptive vocabulary is inaccurate, metaphorical, and misleading. It leads to conceptual confusion and unwarranted suspicion. The discipline needs terminological reform. However, this is not as easy as it sounds, which is no doubt why the mess persists. I try to make some modest suggestions here. There is nothing amiss with the word “evolution” or “evolve”: organisms can be said to evolve from earlier organisms. This notion has two elements: (a) later organisms derive from earlier organisms, by procreation; (b) organisms change down the generations. There is a causal dependence with variation. When we speak of the evolution of organisms, we mean to imply that one gives rise to another (by reproduction) and that there is development or alteration, possibly improvement. We mean to rule out the idea that organisms are created ab initio and that they remain static down the generations. The word captures the import of the theory perfectly: dependence, succession, variation, and improvement. So when we speak of Darwin’s “theory of evolution” we speak accurately (though the word “theory” can mislead, since we are dealing with a scientific fact not a speculation). The problems arise, rather, with the word “selection,” as Darwin uses that word and as current biology follows his example. Darwin explained his theory by first discussing human breeding practices, for example with dogs. All current breeds of dog are derived from wolves by means of selective breeding—by restricting canine reproduction according to human preference. Darwin called this “artificial selection.” The word “select” is appropriate, but why “artificial”? What is the intended contrast? Presumably the natural behavior of dogs (earlier, wolves): that is, the contrast is with the reproductive practices favored by dogs themselves. Darwin could have just said the evolution of dogs from wolves was brought about by human selection, not canine selection; but there is nothing particularly “artificial” about this—it’s a natural fact. He used this word because he wanted a contrast with his use of “natural selection”—hence “artificial” versus “natural.” But the contrast that works in the case of breeding is not the contrast Darwin intends in his use of “natural selection”: natural selection is selection by nature, not by dogs. We can easily see the difference between human selection of dog mates and canine selection of dog mates— both are agents that can meaningfully be said to “select”—but when we move to the Darwinian notion of “natural selection” we are not contrasting one agent of selection with another. We are contrasting selection by an agent with selection by something that is not an agent. And that notion of selection is not clearly meaningful, if taken literally. For what can it mean for X to select something if X is not an agent? Surely, in the ordinary meaning of the verb “to select” only agents can select, as only agents can choose or prefer—but that is not what Darwin intends. The point I am making now is that Darwin does not ground a workable notion of nonagent natural selection in the

246

Part V

contrast between the “artificial” selection of dogs by humans and the “natural” selection of dogs by dogs, since the latter is a case of agent selection. The correct way to think about the human breeding of dogs is that it is an intervention by humans in the choices dogs would make if left to their own devices. So we can say that dogs have evolved over time as a result of human intervention. It is not helpful to refer to this as “artificial selection,” as if it were somehow false or not real selection (as in “artificial silk”): it is real selection according to human (not canine) intention. We could risk pleonasm by speaking of “intentional selection,” so that dogs have evolved by means of intentional selection (by humans). Our species has intentionally intervened to change subsequent generations of dogs from what they would have been without such intervention. I would call this “evolution by human intervention,” not “artificial selection.” If we want to keep the word “selection,” we could speak of “canine evolution by human (intentional) selection”: the changes in dogs down the generations are caused by human intentional action, not by dog preferences. This terminological change helps with the description of other kinds of interspecies evolutionary dependency. Take bees and flowers: bees’ preference for flowers of certain kinds determines pollination, so bee behavior shapes the evolution of flowers: the flowers that now exist have been selected by bees. This is just like the case of human preference for certain breeds of dog—one species intervening in the evolution of another. We may be reluctant to speak of bee intentions, but clearly bees are biological agents whose preferences shape the reproduction of flowers—there is bee selection with respect to flowers. But should this be called “artificial selection”? Hardly: it is not artificial at all but entirely natural—part of the order of nature (just like human selection of dog breeds—humans being part of nature). Selection by agents of types of organism is natural selection; there is nothing “artificial” about it. Would Darwin want to deny that this kind of “artificial selection” is also “natural selection”? That would be absurd, accepting the ordinary meaning of these terms. This brings us to the heart of the matter: the phrase “natural selection.” What does it mean? It presumably means “selection by nature,” as opposed to selection by agents, whether human or nonhuman (e.g., bees). Nature includes the weather, facts of geography and geology, the laws of chemistry and physics, anything “natural”—and these are not agents. So the idea must be that nonagents can be said to “select” organisms— determining which shall reproduce and which shall not. Nature therefore acts like an agent in that it can be said to select; it is analogous to a selecting agent. Consider this passage from the first edition of Darwin’s On the Origin of Species: “It may be said that natural selection is daily and hourly scrutinizing, throughout the world, every variation, even the slightest; rejecting that which is bad, preserving and adding up all that is good; silently and insensibly working, whenever and wherever opportunity offers, at the improvement of each organic being in relation to its organic and inorganic conditions of life.” In later editions Darwin inserted “metaphorically” after “It may,” no

Biology247

doubt aware that it would be unfortunate were he to be taken literally. Of course, such language, though suggestive, even illuminating, is merely colorful, and Darwin intends no anthropomorphic view of nature; anyone who understands the theory can see that. And it is perfectly possible to state the core of the theory without use of such metaphorical language. Still, the fact remains that Darwin’s central way of describing the process in question—“natural selection”—lends itself to anthropomorphic interpretation, and is at best a mere metaphor. In the literal sense, nature does not select anything, naturally or artificially. The Oxford English Dictionary defines “select” as “carefully choose as being the best or most suitable,” and nature certainly does nothing of the sort. As Wallace urged on Darwin, the very term “natural selection” invites misunderstanding and would be better purged (he suggested replacing it with Spencer’s “survival of the fittest”). Don’t we want a canonical formulation of the theory that is straightforwardly and literally true? But it is difficult to come up with an equivalent phrase, because all the synonyms for “select” carry the same agential connotation—“choose,” “cull,” “decide,” “elect.” The verb “select” is actually the least agential-sounding of the group. So there is no easy remedy for Wallace’s linguistic problem. The word “natural” derives whatever meaning it has in this context from Darwin’s use of “artificial,” but that is itself objectionable; and “selection by nature” doesn’t help (aren’t bees part of nature?). So both parts of the ubiquitous phrase “natural selection” are ill chosen, despite the ease with which they trip off the contemporary tongue. But it is the use of “select” in this context that really cries out for replacement—yet nothing comes to mind. It is no use resorting to words like “causal” or “mechanical” to produce the phrases “evolution by natural causation” or “evolution by natural mechanical means,” because that would be to assume that agential selection is not causal or mechanical, among other problems. There seems no simple way to fix the problem, and that can produce worries about the theory: how can Darwin’s theory be (literally) true if it is not possible to formulate it without using a suspect metaphor? Why can a skeptic not protest that evolution does not occur by natural selection, since there can be no such thing as nonagential selection? Isn’t the very idea of natural selection an oxymoron—selection without a selector? Only if we assume an anthropomorphic view of nature—say, a form of pantheism—can we rightly describe nature as in the business of selecting things. Nature may be said to cause things to happen, but it doesn’t select what it causes. It doesn’t, to use Darwin’s words, scrutinize, reject, preserve, or work at the improvement of.1 1.  Darwin also used the phrase “sexual selection,” but here there is genuine selection by an agent— one animal chooses another to mate with. The peahen chooses the peacock to mate with. Thus we can say that the peahen scrutinizes, rejects, favors, prefers, or plumps for a particular peacock. This intentional activity belongs with human “artificial selection”: but both contrast sharply with so-called natural selection, in which no agent selects anything. Using all three phrases without marking this difference leads to confusion in the unwary, and is strictly speaking false.

248

Part V

I think that the phrase “natural selection” is unsalvageable, except as a suggestive metaphor, and also that there is no way to fix it by a suitable replacement for “selection.” I suspect it came to Darwin because of his desire to compare evolution with selective breeding: he wanted to say it is like selective breeding but also unlike it. Hence the contrived phrases “artificial selection” and “natural selection,” neither of which is satisfactory. But this approach, mainly prompted by expository and polemical needs, should be abandoned in these more enlightened times in favor of a far more literal description of the basic evolutionary process. I would thus call the dog and bee cases “evolution by interspecific intervention”—and would allow the use of “select” to describe the nature of the intervention. What Darwin calls “natural selection” I would prefer to call “evolution by intraspecific competition,” or less stiffly “evolution by reproductive rivalry.” This is the actual content of the theory, after all, and it makes no reference to any notion of selection by nature. The idea of the theory is that individuals compete with each other to reproduce, by living long enough to do so and by finding an appropriate mate; in that way, the “winners” pass on their genes, thus making them more frequent in the gene pool. These individuals may indeed be said to select certain courses of action, as well as mates, and this is all perfectly “natural”; but we say nothing about an impersonal nature doing any selecting. Animals compete to reproduce against a background of nature containing scarce resources, but it is odd to speak of these scarce resources as in any way selecting who is to reproduce and how effectively. Do we really want to say that bodies of water, say, do any selecting? The sober truth is that reproductive rivalry involves natural resources that are not equally available to all, so some do better than others in the competition. The ones that do better pass on their genes and hence their phenotype. That is all—there is nothing here worth calling “natural selection.” At best this is a misleading redundancy (or possibly a useful introductory heuristic, to be jettisoned once understanding has been achieved). Darwin’s theory thus maintains that evolution (succession with variation) occurs in virtue of reproductive rivalry in a world of scarce resources—evolution by differential propagation, we might say. It is not that nature selects; rather, some organisms outperform others. This is essentially the idea contained in the phrase “survival of the fittest,” but that phrase focuses on survival, not reproduction, and wrongly suggests that fitness is what matters, that is, strength or endurance, not sheer reproductive prowess. It is not so much survival of the fittest as proliferation of the most seductive. Here is an analogy. Some inorganic things persist and some are destroyed, according to the forces of nature. Thus mountains and lakes may come and go; glaciers melt; rocks disintegrate. Some things survive the gauntlet and some don’t. They don’t survive to reproduce, but they do survive. Some types of things may be better at surviving the forces of nature than other types, say, harder rocks or more compact ice. We may even speak of “survival of the fittest,” where fitness is some intrinsic property conducive to warding off the forces of nature, such as wind or temperature. But would it be

Biology249

right to describe these facts as “evolution by natural selection”? We can justify the use of “evolution” because there is change over time (physicists speak of the evolution of the universe, prior to organic evolution), but do we really want to say that the physical objects that now exist—the ones that have survived the forces of nature—exist in virtue of “natural selection”? Isn’t this just a misleading way of saying that they have withstood the forces of nature? Nature didn’t do any “selecting,” except in an entirely metaphorical sense—still less any “scrutinizing” or “rejecting.” We might want to contrast this kind of survival (continued existence) with the man-made kind—the “artificial selection” whereby we keep certain things in existence, not letting nature destroy them. We may then say that some things evolve by artificial human selection and some things evolve by natural selection: we repair our houses and artworks, but we leave mountains and glaciers to nature. One sees the point of this way of talking, but the terminology is far from satisfactory. We could even add an element of competition to the story by supposing that raw materials can shift from one thing to another. Consider a desert of sand dunes being eroded by wind: the grains of sand may swirl around forming different dunes at different times—one dune goes out of existence and the grains form another dune a mile away. Here we could say (metaphorically) that sand dunes compete with each other for scarce resources. But is there any point in saying that there is “natural selection” favoring dunes that can acquire more sand? Some dunes may be better at acquiring sand than others, given their location and the nature of the prevailing winds; but should we say that they survive better than their rivals by natural selection? We could say that it is natural for them to survive better than other dunes, but why suppose that there is any “selecting” going on? Does the Earth select its mountain ranges and oceans—as opposed to simply failing to destroy them? An earthquake may destroy a mountain, but does it thereby select some other mountain that it could have destroyed? Nature may destroy an organism, say by fire or flood, but should we say that it thereby selected another organism for survival? Is every existing organism here by dint of nature’s positive selections? This seems no more reasonable a thing to say than that existing mountains are here because of nature’s positive selections. The idea of natural selection seems to have been born in analogy with human intentional selection, but it has little point outside of that analogy. And the analogy is crucially flawed because it is not literally true that nature selects—it is merely as if it selects. Once we purge the theory of these metaphors, we find that there is no notion of selection that unites agents and nature; instead, agents (humans, bees) purposefully select and organisms reproductively compete—and that’s all there is to it. We should therefore, in our sober clear-eyed moments, stop speaking of Darwin’s “theory of evolution by natural selection”—though I expect the practice to persist because of the power of tradition. In an ideal language of biology, no such phrase would occur. Perhaps we can live with the continued use of the phrase if we bear clearly in mind the kind of conceptual point insisted on here—though I would prefer complete linguistic reform, myself.

Immaterial Darwinism

Consider the following imaginary world (I say “imaginary” and not “possible” because I doubt this world is really metaphysically possible). There is a range of disembodied minds divided into different kinds in this world, analogous to animal species, numbering in the millions. There are also differences among the individuals belonging to each kind. These individuals can and do reproduce—they have children. They can also die, sometimes before reproducing. Our imaginary beings are completely immaterial and hence have no molecular parts. About these beings we can ask an origin question: how did they come to exist? One possible answer would be that they were created by another disembodied being, vastly superior to them, some 6,000 years ago. There was never any transformation of one species of mind into another; each species was created separately by an all-powerful God. But there is another possible origin theory: the theory that the disembodied minds evolved by natural selection from relatively primitive origins. This theory postulates that once, billions of years ago, there were very simple disembodied minds, and these minds evolved by natural selection into the minds that we see today. How did this happen? When the minds reproduce, making copies of themselves, in the form of offspring, errors can be made—the copying isn’t always perfect. When an error occurs the offspring differs slightly from the parent, since the error is not corrected. The error produces a variation in the properties of the mind that is produced—say, we get a mind with a slightly higher IQ or a reduction of affect. Natural selection then operates to favor or disfavor the change, which then gets passed to the next generation, or fails to. These selected changes accumulate over long time periods, producing varieties of mind. Competition for reproductive mates gives further bite to natural selection, so that traits are favored that increase the probability of mating, and hence producing copies. In other words, we have random variation, self-replication, and natural selection operating together to generate the immaterial beings that exist in our imaginary world. There are no genes, no bodies, and no physical processes of any kind—but there is evolution by natural selection. The lesson of this little thought experiment is that the basic explanatory scheme of Darwinian explanation is not essentially materialist. As things exist in our world, animals have material bodies, material genes, and material behavior: the mechanism of random variation, reproduction, and natural selection applies to material entities. But the mechanism itself is topic neutral: it is sufficiently abstract to apply even to immaterial beings—so long as the basic conditions of variation, copying, and natural selection apply. Just as it is possible to run an evolutionary program on a computer, producing more complex patterns from simpler ones by random variation and natural selection, so it is possible to conceive a world that runs by Darwinian principles but is quite immaterial. Spirits could evolve by random mutation and natural selection, so far

252

Part V

as the theory is concerned. Nothing in the theory itself entails that it applies only to material entities. Even gods could be subject to Darwinian evolution. How the abstract principles are implemented in different kinds of being differs from case to case, but the principles themselves are ontologically neutral. Thus it is logically conceivable for a dualist like Descartes to be a Darwinian. On the one hand, the animal body evolves by material natural selection in the standard way, involving DNA. On the other hand, the immaterial mind itself evolves on a parallel but separate path: it is subject to internal changes (“mutations”) that can be passed on to the next mind, assuming that there is a parallel mechanism to the genetic one; and these changes can be selected for or against. As the body creates copies of itself using DNA, so the mind creates copies of itself using whatever immaterial resources it possesses. We have two-track Darwinian evolution to match the dualist ontology. No doubt no such thing happens in the actual world, but we can imagine a world in which body and mind, conceived as separate substances, evolve in parallel, both subject to Darwinian principles. So you can consistently be a Darwinian anticreationist while also accepting Cartesian dualism, or even Berkeleyan idealism. The logic of Darwinian explanation is neutral between metaphysical systems.

Trait Selection

What does nature select when it naturally selects (if we may speak this way for expository convenience)? Evolution is the survival of the fittest—but the fittest what? The usual candidates are: genes, individuals, groups, and species. Each of these things belongs to what we might call biological ontology, but the question is which of them is the object of natural selection. The orthodox view today is that animals act so as to benefit their genes, not the group or species or even themselves (because of kin altruism). Genes build bodies that ensure the propagation of themselves by equipping bodies with reproductive advantage. Thus the gene is said to be “the unit of natural selection”—the genes that survive being the ones that build the best bodies. When the individual acts to secure its own survival it thereby ensures the survival of its genes, which get passed on to the next generation; but it can also act to secure the survival of other individuals that share its genes. A gene for helping others who share your genes has obvious survival value. To put it differently, when an individual produces copies of itself by reproduction it also produces copies of its genes—but it can also produce copies of its genes by helping others who share those genes. However, this helping does not extend to the group or species, because outside of narrow kin relations genetic overlap tails off. Let us accept this broad picture of how natural selection works. I want to suggest that there is an additional candidate for the unit of natural selection that has not been considered, and quite a natural one: traits. Just as there are copies of individuals and copies of genes, so there are copies of traits—organs, body parts, phenotypic features. These too get reproduced, passed on—hence the similarity of parents and offspring. They are passed on because the genes are passed on: the genes make the traits (with some outside assistance). So what about the idea that the trait is the unit of natural selection? That is, natural selection primarily selects traits as the things that survive— not genes or individuals (or groups or species). Not the selfish gene, but the selfish trait. Traits function so as to ensure their own survival: they have causal powers that lead to their persistence down the generations. The genes are their servants—the job of genes is to program adaptive traits and allow them to recur. Suppose a mutation produces a new trait, generally a modification of a preexisting trait: that trait will do well or ill according to its construction and competition from the traits of other animals. For example, the trait might be a subtle change in the eye, which can confer an adaptive advantage or disadvantage on the possessing organism. If the new trait is adaptive it will have a greater chance of being passed on than if it is not adaptive. So the trait is subject to selective pressure: will it persist or be eliminated? Later generations might see a proliferation of the trait. No doubt this is how eyes evolved to begin with. The trait arises by genetic mutation and then is selected for or against, which determines its frequency in subsequent generations. There are now a

254

Part V

great many pairs of eyes in the world where once there were none, and similarly for all other traits of organisms. The trait has persisted and multiplied. The point here is that genes are useless without accompanying traits: there would be no genes if there were no traits that they program. Genes are not self-sufficient entities. Nor would there be individual organisms without traits: an individual is a collection of traits, and the organism’s genes are merely devices for passing traits on to individuals. Traits are where all the selective action is—they are the things that primarily determine the survival prospects of both genes and individuals. Genes get transmitted if (and only if) they produce good traits, and organisms can survive to reproduce only if they comprise the right suite of traits. Evolution selects good eyes, good wings, good claws, good digestive systems, and so on; everything else that survives (or not) is derivative on such traits. The traits that survive best are those that compete best with other traits—of members of the same or different species. Traits are the things that are in competition— that must prove their worth in the battle for survival. They are like weapons of war, except that they constitute the entire nature of an individual animal. Genes, however, are like the blueprint for weapons, not the weapons themselves. The traits constitute the biological frontline. They are what are primarily selected to survive and be copied into the next generation. We can adopt a “trait’s-eye view” of evolution, using the expository device of personification. The trait is trying to adapt itself to the prevailing conditions, to be the best trait it can be. If it were a conscious agent it would seek to cooperate with other traits and to perfect its own inner nature. It would selfishly try to get itself into the next generation, collaborating with other traits to make that happen—just as genes are said to do these things (when personified). In fact, the gene selection view and the trait selection view are fully compatible with each other, since genes produce traits and traits rely on genes to be transmitted. So we should be able to translate gene selection theory into trait selection theory—nothing vital will be lost in translation. If genes are units of natural selection, then so are traits: both determine the viability of the organism, and both are potentially immortal (unlike individuals). If an animal helps its genetic kin, it also enables the traits of that animal to be passed on, via the genes. Thus we can have a trait-centered picture of the evolutionary process. This picture is superior to the gene-centered picture because traits are the actual things that selection acts on—the things that have direct biological utility. Genes only have biological utility because they program the body to contain suitable traits; they have none in themselves. Also, if we think of a whole organism as just a collection of traits, then we won’t want to say that the individual as such plays any distinctive role; rather, collections of traits compete with other collections of traits and succeed in producing similar collections of traits. Indeed, the concept of the biological individual is redundant from a theoretical

Biology255

point of view. Consider caterpillars and butterflies: are these the same individual or not? It really doesn’t matter from the point of view of natural selection; what matters is that each organism is characterized by a suite of adaptive traits. We could eliminate the notion of an individual animal from the science of biology and replace it with talk of trait complexes, thereby dispensing with the ontology of individual organisms and speaking only of bundles of traits (compare replacing the ordinary ontology of material objects with “bundles of qualities”). The basic fact is that eyes produce eyes: that is, an organism with eyes will tend to produce (via the genes) another organism with eyes, granted that natural selection favors eyes. Compare selective breeding: the breeder doesn’t select genes directly, but only indirectly via traits. She likes the look of certain canine traits, or values them in some other way (e.g., greyhounds), so she ensures that these traits are propagated: the traits are the objects of the breeder’s intentional selection—it just happens that genes come along with them. Similarly, natural selection favors certain traits over others, because of their survival power, and genes come along for the ride. Evolution selects for traits as breeders select for traits. If reproduction could occur without genes, possibly in some alien life form on a remote planet, evolution would still operate on traits, just like selective breeding. Genes are conceptually and theoretically parochial (DNA certainly is), given that it is logically possible to have evolution by natural selection without even the existence of genes. Maybe on that remote planet reproduction works by something akin to a photographic process—the traits of the offspring are caused by a photograph of the traits of the parent. What if breeders found a way to produce copies of dogs, and mixtures of dogs, by some nongenetic means—say, by mapping their bodies and then creating a modified version of the original by means of a computer simulation? The selection would still be on traits, with genes playing no role (the animals might even be rendered sterile). And the same could be true of natural evolution. The point is that, from a logical point of view, trait selection theory has broader application than gene selection theory, since genes are local and traits universal. We can talk in different ways about the mechanism of evolution and thereby gain different kinds of illumination. The gene’s-eye view has its virtues (it tells us what is happening behind the scenes), but so also does the trait’s-eye view (it gives us a view from the front row). I submit that the latter is simple and natural, as well as true to the facts; I wonder then why it has not been suggested before (as far as I know). Is it because it is harder to personify traits, compared to individual organisms and genes? We apparently need to think of evolution as if controlled by some person-like entity (God, the individual, the gene), but the trait selection view makes that difficult. On that view, biological ontology basically consists of traits bundled together—this is what natural selection operates on and favors or disfavors. Traits are not person-like entities, yet they are what natural selection “aims” to promote: they are the units of currency in

256

Part V

the biological economy.1 There could be evolution without anything of a person-like nature on the scene—just collections of self-reproducing traits. Evolution is essentially the survival of the fittest traits. Darwin should have called his book On the Origin of Biological Traits.2 1.  We could postulate the notion of a “trait pool” by analogy with a “gene pool”; this would be the totality of traits belonging to a species at a given time. Then the question will be which of the traits in this pool is currently favored by natural selection—which traits have a bright future and which don’t. 2.  We can imagine a biological world without clearly defined species, without genes and DNA, without even isolable individual organisms—an amorphous biomass reproducing without chromosomes, the double helix, etc. Yet this world will be subject to Darwinian natural selection, as some parts of the sprawling mass are favored over others to reproduce. This is a world in which biological traits are the only game in town—not genes, not individual organisms, not groups, not species. What survive are constellations of traits. That is the bare essence of evolution, with traits alone as the units of natural selection.

VI Ethics

The Trouble with Consequentialism

Philosophers have raised many problems with consequentialism over the years, usually claiming that it gives the wrong intuitive results in certain special cases, particularly concerning justice. But the problem I want to raise is more structural and cuts deeper: consequentialism is wrong as a matter of principle, not just as a matter of detail. It contains an inbuilt defect. The point is actually quite obvious once it is articulated, and I am surprised it has not been noticed before—but it took me a long time to notice it.1 If I am right, we have no moral obligations of the kind that consequentialists have alleged. We need a different kind of ethical theory altogether. The consequentialist claims that we have an obligation to promote good states of affairs (and prevent bad ones). A state of affairs is (intrinsically) good if it contains various basic goods, such as pleasure, happiness, knowledge, friendship, and aesthetic experience. The theorist makes a list of basic intrinsic goods and then claims that we have a duty to bring these goods about. If we agree that these things are good, we must agree that the world should contain them so far as is possible, and so we should do what we can to contribute to promoting the goods in question. All this seems self-evident, and so consequentialism seems undeniable. Questions may arise about how much good we should do and about what goods are basic and about how best to bring the goods about, but the logical structure of the position seems unassailable— we surely have a duty to make the world a better place (or a less bad place). Given a population P and a set of resources R, an agent A has an obligation to deploy R to produce the maximal good in P, as defined by the listing of basic goods. We must maximize the good. Now suppose that A has an exhaustive and accurate listing of goods and proposes to maximize them in P using R. By stipulation, then, A has the correct conception of the good. But suppose also that the members of P do not agree with A’s listing of goods. The members of P (think of them as extreme Spartans) prefer a quite different listing, rejecting pleasure, knowledge, and so on, but including the martial virtues of valor, fitness, discipline, and so on. They regard their listing as defining the primary goods for human beings, or at least for them. Let us suppose, for the sake of argument, that they are wrong to do so: despite their impassioned defense of their values, they have the wrong values, while A’s list includes the right values. Does A have a duty to impose those correct values on P? To do so would be to override the sincere and express wishes of P, even granting that it would produce an objectively better state of affairs than the one produced by the values actually held by P. I think the answer to our question is no, for a familiar reason: it would violate the personal autonomy of rational beings. 1.  I am not going to say that nothing like it has ever been noticed, but I am not aware of anything that puts it the way I do here.

260

Part VI

To override their wishes would be to fail to respect members of P as persons—rational agents with a right to pursue their own, freely chosen, values. The problem is that the consequentialist prescription is inherently paternalist: it treats people as if they must be protected from their own folly, ignorance, and immaturity—in certain conceivable cases. There is thus a clash between the Kantian ideal of respect for persons and the consequentialist principle that our duty is to maximize the good. Simply put, other people may not agree with our conception of the good, correct though it may be. The consequentialist rule in effect tells us to maximize the good irrespective of what our potential beneficiaries may themselves believe and desire. We are instructed to do what is “good for them” whether they agree or not. We must “save them from themselves.” There is no such problem with respect to animals and children. Here the idea of rational consent gets no purchase: we are entitled to overrule their protests, knowing that we are thereby producing what is good for them. We take our children and pets to the hospital even though they complain loudly. But for adults we cannot act in that way: they have a right to have their opinion respected, even when it is wrong—even obviously wrong. In the case I described, not only do we have no obligation to maximize the good for P; we have an obligation not to overrule their wishes, and hence an obligation to produce less good in P than we could. That is what I meant by calling the problem structural: consequentialism fails to recognize a fundamental moral principle that conflicts with its prescriptions. In conceivable cases it gives completely the wrong result, because it ignores personal autonomy—and it is in the very nature of the position to ignore that value. There is therefore no absolute unconditional obligation to maximize the good, even when you easily can and wish to do so. Indeed, there is an obligation in certain cases to limit the good you can easily and happily bring about. It can be right not to do the good you can. It can be wrong to bring about what is objectively good. It might be replied that we can reformulate consequentialism so as to get around this problem, by bringing in preference satisfaction. The good to be maximized is just having preferences satisfied, so that we can respect the wishes of others while maximizing good consequences. We give the Spartans what they want, not what we know to be intrinsically good—we maximize their preference satisfaction. The problem with this is that they subscribe to a quite different set of values from those of our beneficent agent A (rightly, as stipulated). By maximizing their preference satisfaction, A is thereby promoting states of affairs that he does not think morally good—indeed, states of affairs he knows to be bad (war, torture, killing, etc.). Why should he have an obligation to do that just because members of P prefer that way of life? Why should he support their bad choices? On the contrary, he has an obligation not to promote such values and conditions, since (by hypothesis) he knows them to be bad. Even if there is some good merely in satisfying preferences, however misguided or evil such preferences may be,

Ethics261

that cannot override the fact that the values of P are not those of A. I surely have no duty to promote what I rightly regard as bad! If the Spartans believe that beating their children daily to make them tough is a good thing, I have no duty to support them in that (by stipulation) misguided policy. If other people prefer the bad, that is no reason for me to enable them to pursue it. So moving to preference satisfaction does not help the consequentialist. Consequentialism thus faces a dilemma: either we impose our conception of the good on recalcitrant populations in a paternalistic spirit, or we respect the actual preferences of others and risk promoting objectively bad states of affairs. Neither of these alternatives is morally acceptable, and so consequentialism is false as a theory of moral obligation. It might now be thought that we can save consequentialism by combining both versions of it into a single doctrine, namely, A’s duty is to promote the good in P if and only if members of P agree with A’s conception of the good. That is, if P desires the good as A (rightly) conceives it, and so consents to A’s plans, then A is obliged to use R to bring about that agreed-on good. This is a conjunctive doctrine: right action maximizes the good and respects persons. Alternatively, the doctrine is conditional: maximize the good if (and only if) the people concerned want the good in question. So now we have two types of consequentialist doctrine depending on our choice of P: the simple nonconsent doctrine for animals and children, and the conjunctive consent doctrine for rational adults. And isn’t this closer to how we actually operate? We help people not just when they need help but when they want help. If a person in need does not want your help, you don’t give it; but if she does, you do. Nor do you deem it incumbent on you to help people pursue values you despise or otherwise deplore. We might call this “consent consequentialism.” Now I see nothing wrong with that as a useful moral rule, but it is not the theory of consequentialism as originally conceived. The rule does not prescribe maximizing good states of affairs as the sole guiding moral principle; it adds a quite different moral principle, namely, respect for persons. That is a deontological principle, not a consequentialist one. Someone might try to give the principle a consequentialist interpretation in the standard rule-utilitarian way, but for the usual reasons I think this is unsatisfactory. In addition, the same consent problem applies for this new type of good: what if members of P reject the idea of respect for persons? Suppose I am trying to promote this principle in P because I think it will lead to overall happiness in P, but members of P disagree with it. Then I will find myself having to override their wishes, which involves rejecting the very principle I am trying to promote. The point is that respect for persons is not a consequentialist principle, so that consent consequentialism is not real consequentialism, that is, a unifying ethical theory of what constitutes right action. We can see this just by considering the difference between this doctrine and genuine consequentialism in particular cases: when members of P reject A’s proposals,

262

Part VI

real consequentialism recommends overriding them in the name of the greater good, but consent consequentialism tells us to abjure that recommendation, instead calling for nonaction. It tells us what not to do, based on a nonconsequentialist principle, but it gives us no guidance as to what we should do. Should we just leave them to their own devices? Should we try to persuade them of our values? Should we put something in the drinking water to change their misguided attitudes? In the case of animals and children, we have a clear mandate in all cases—maximize the good—but once autonomous rational agents enter the picture we no longer have an automatic answer to the question of where our duty lies. We are left with the feeble advice not to impose our values on others, but to help others if they accept our offers (what if they are uncertain or hesitant or unfocused?). Where is the grand overarching ethical theory? Where is the deduction of a theory of right action from an account of the good? Perhaps a consequentialist will dig in her heels at this point and insist that we should follow her prescriptions in all cases, but exercise a bit of discretion in doing so. We should circumvent the objections of our potential beneficiaries, because they will thank us in the end, once they experience the good life we as define it. True, we shouldn’t coerce them into accepting our generosity by force of arms, but there are subtler ways of bringing the good life to them—say, by compulsory re-education or soothing propaganda or sly deception. It simply is our duty to maximize the good, and we may do this by any means necessary, so long as we produce no suffering in P. We are paternalist when it comes to seatbelts and crash helmets, so why not more generally? What is so wrong with the right-thinking chemical placed secretly in the water? I can see how this approach might seem attractive, but I think it is a very dangerous and unappealing way to go. There is not only the problem that we might be wrong in our conception of the good, but also it is morally objectionable to abrogate respect for persons in the way proposed. How would you like it if some beneficent outsider came along and ignored your sincere and deeply held commitments concerning the good life? Autonomy is a fundamental right of any rational self-determining being, as Kant maintained: if consequentialism conflicts with it, then so much the worse for consequentialism. I don’t think the classic consequentialists self-consciously envisaged rejecting it; they just assumed that people would agree about the good life once it was properly explained. For how could anyone in his right mind object to pleasure, knowledge, friendship, and aesthetic experience as types of good? Maybe no one in fact would, but the logical possibility of objecting to these things is enough to undermine consequentialism as an account of what right action or personal obligation consists in. It shows that we don’t have the correct analysis of obligation. There is a political side to this. If you have great political power, and great selfconfidence, you tend to assume that you have a right to impose your values on other people. Everyone else is thought of as, in effect, a child. The proponents of consequentialism were mainly citizens of an imperial power, Great Britain. They assumed a right

Ethics263

to rule others and to impose on them their own conception of the good. The idea of democratic decision making (an expression of the idea of personal autonomy) was strictly limited; it did not include rebellious and ill-educated foreigners. In this kind of political atmosphere it is very natural to suppose that it is permissible and obligatory to bring the good life (as you conceive it) to others, without much regard for their consent. The very fact that it is the good life is thought a sufficient justification for spreading it, in one way or another. But this is imperialist in the worst sense: it involves ignoring or overriding the considered wishes of others. You don’t have to ask them what they want; you just determine what is good and then bring it about. For instance, you might hold that democracy is a basic value, conducive to the highest good for human beings, so you regard it as your duty to ensure its spread. You might well be right so to regard it, but there is still a substantive question about how you bring it about. If we take on board the Kantian principle of respect for persons, then our duty is not to cause democracy to obtain as widely as possible, whether people welcome it or not; our duty is rather to offer it to others for their consideration as persuasively as we can. That is, given that democracy is a basic good, our obligation is to try to get others to agree with us by rational means—and if they don’t, not to impose it in any other way. Thus my duty is not to maximize the good tout court but to maximize my chances of persuading others to adopt my conception of the good (the power of persuasive speech thus assumes central moral importance). My duty is not to propagate the good but to promulgate the good—to persuade, not to enforce. I have this duty precisely because the consequences are likely to be good—so much is right in consequentialism—but the duty is importantly different from that envisaged by the classic consequentialist. A duty to produce good consequences is not the same as a duty to persuade others to allow me to produce good consequences. I have the former duty with respect to animals and children, but not with respect to rational adults; I have the latter duty with respect to adults, but not with respect to animals and children. The duty to make the world a better place is best thought of as a pedagogical duty, not an engineering duty—at least when it comes to getting others to share and adopt one’s conception of the good. If I encounter people resistant to the good life as I conceive it (rightly, we may suppose), my reaction should not be to ignore them or overrule them or circumvent them, in the laudable aim of improving their lives (even if I do improve them); it should be to try to persuade them that I am right and they are wrong. If I fail in that endeavor, I may regret the outcome, but I have not fallen short of my duty. I cannot blame myself if my improving words fall on deaf ears. This is the kind of attitude one would expect to see in a political environment of autonomous groups, not one of imperial power. If I am a member of a tribe surrounded by other tribes, each equal in power, and I have a conception of the good life that I want to spread, I cannot do so by fiat or force; I must resort to the persuasive arts. We would not expect to see consequentialist philosophy of the classic British variety in

264

Part VI

this kind of political environment. It is no use saying, “Maximize the good!” when surrounded by people who have the power to resist your attempts to bring them the good and who might not agree with your view of the good. The consequentialist maxim should instead read, “Maximize the good by trying to persuade others to share your view of the good!” If you succeed, you will have brought about a good state of affairs without infringing anyone’s autonomy. And if you fail, bringing about no good at all, perhaps even harm, you have not fallen short in your moral obligations, since they never included actually bringing about any good. You could live a morally praiseworthy life and never do anybody any good at all, despite having the power to do so, as long as you tried your best to persuade people to see it your way—they just refused to listen. That is not something a classic consequentialist could say, since the prime moral directive is actually to produce good states of affairs, by hook or by crook. It is notable that these kinds of problems do not apply to deontological theories. Rules about lying, promising, stealing, and so on are not hostage to other people’s consent. Even if someone objects to my keeping a promise to him, on the grounds that he doesn’t hold with all that irresponsible promising, that has no tendency to undermine my commitment to keeping my promise—I have not imposed my values on him. I have not forced him to adopt promising himself. I cannot be accused of infringing his personal autonomy if I follow such moral rules. This is why we are uneasy about seatbelt laws but do not worry that allowing people to keep their promises will infringe on the rights of others. It is a great merit of deontological theories that they run no risk of disrespecting persons; and most deontological rules precisely concern respecting other persons—by not lying to them, stealing from them, and so on. But consequentialist theories systematically and inherently entail that it is acceptable in principle to disrespect and disregard the beliefs and wishes of others, because of their inbuilt paternalism. By contrast, there is nothing paternalist about deontological prescriptions: we never have to treat the convictions of others as just a barrier to be overcome in the project of doing what is right. We treat others as ends in themselves, not merely as means to achieving good states of affairs in the world. We don’t think of them as mere passive receptacles of the good—opportunities for good production—but as rational agents whose views must be taken into account. Suppose that A is the only person in the world with the right conception of the good. Not one of the billions of others shares his enlightened perspective. Suppose too that A has unlimited resources and powers—he can promote his vision of the good and thereby produce a much better world. But he meets opposition from the targets of his philanthropy—reasoned opposition, though mistaken (we can also suppose that A is far more intelligent than everyone else). What are A’s moral obligations? For the consequentialist they are to produce by whatever means necessary the good life for all, as he (rightly) sees it. But that will involve overriding the autonomy of billions of people—everyone on the planet except him. Surely it is not his solemn moral duty to

Ethics265

go ahead and do that; his duty is not to do it. He might undertake a program of rational persuasion, but he cannot just ignore the wishes of everyone else. That would be an intolerable exercise of power and a complete abridgement of human rights. He should not impose his vision on the billions of reluctant others. If this is right, I simply do not have an absolute obligation to promote the good (or reduce the bad). All I have (at most) is an obligation to help others if they want my help. My attitude to others should not be that they are opportunities for me to spread the good life; rather, my attitude should be that they are autonomous agents whose wishes must be respected. I am not morally like a father or a pet owner, dispensing goods to grateful recipients (children or animals); I am more like a coworker who must respect the values and point of view of my equals, misguided though they may often be. The rightness of my actions is not to be calculated by summing up the good and bad consequences of my actions, but by a whole network of deontological principles, only one of which pertains to the future good I bring about (“help others who want your help”). I cannot be faulted if I help no one in my life and improve the good of the world by not one jot, even when I have the power to do these things; it all depends on whether others consent to my intervention. It might be frustrating to be thwarted in one’s altruistic aims, but it is not immoral to do nothing toward improving the common good—if nobody wants you to contribute. You have, quite properly, respected others as persons with the dignity appropriate to persons. You have an obligation to respect the wishes of others, but you have no obligation to promote the good among them, in the circumstances described. At most you have an obligation to inform them of your view of the good and hope that they respond positively. The final point I shall make is that this is not as abstract and theoretical as it may sound. When goods are described in abstract terms we don’t tend to find much disagreement about them, but when it comes to actual policies—individual and collective—we always need to get more specific. We say abstractly that knowledge is a good, but knowledge about what? What kind of knowledge should be maximized with the resources available—knowledge of baseball or knowledge of mathematics? Aren’t some kinds of knowledge better than others? Similarly for pleasure: there is the old distinction between higher and lower forms of pleasure, but the pleasures falling in each category can also be ranked and disagreed about—is the pleasure of pushpin better than the pleasure of cricket? People will differ about that kind of question. What kind of friendship is best—same-sex or different-sex, collegial or casual? What kind of aesthetic experience is best—opera or pop music, art or nature? In all these cases people are liable to disagree about the specific goods to be promoted; and even if one side is right and the other wrong, it is unacceptable to ignore the opinions of others and impose your views on them instead. That is, it is not morally permissible to override the autonomy of others in deciding what specific goods they will enjoy, even if you are right about the relative values of the goods in question. In practical life, then, there will be many

266

Part VI

occasions on which the consequentialist prescription gives the wrong results, precisely because of the structural defect that lies at its heart: it will inevitably result in violations of personal autonomy. This is not just an issue of conceptual analysis and fancy thought experiments. The underlying problem is that consequentialism views the moral agent as a kind of all-powerful benevolent dictator, doling out the good to grateful recipients. Then the objective good of things leads to a straightforward prescription for how the moral agent must conduct herself, namely, she should maximize the good by hook or by crook. But that is not the situation the moral agent is in; rather, she must negotiate with other autonomous beings, taking into account their views of what constitutes the good, without necessarily agreeing with them. She may find sharply different conceptions of the good, especially at the level of specifics, and even a general reluctance to accept any kind of help; and these will place strong limits on her sphere of moral action. The moral agent cannot just blunder in and dictatorially bring about the good, even if it is the good. This means there is always a kind of frustration in the moral life, since we cannot create a perfect world unless we can secure the consent of others. It can be immoral to create a utopia.

Absurd Utilitarianism

The basic principle of utilitarianism is that we are morally obliged to “maximize utility”: that is, given a fixed and finite resource R, we should act so as to produce the largest amount of utility (happiness, pleasure, preference satisfaction) in the world with R. We can call this the strong principle of beneficence (SPB); it tells us to use our resources in such a way that more utility is produced in that way than could be produced in any other way—and irrespective of other potentially competing demands. SPB tells us to create the best world we can, as defined by maximizing utility—the “greatest happiness of the greatest number.” It has been pointed out that this doctrine has the consequence that we should increase the number of people in the world even if the average level of happiness is lower than when there are fewer people, because this will maximize aggregate happiness. Given R, a distribution of resources among two million people may produce a greater total of happiness than among one million, even though each person is less happy under that distribution. Thus we are morally obliged by SPB to increase the population, given this empirical fact. The more people the more merriment, though each person may be less merry than under an alternative arrangement. (If we decided that we should increase average utility, not total utility, then we should reduce the population in order to increase the utility of each individual, despite the overall loss of utility.) This result has struck many people as counterintuitive, especially when pushed to the limit (say, ten times as many people, each a tenth as happy as the original million); but some have been willing to accept that this is the morally right result. I want to push the point even further in what follows, by way of reductio ad absurdum. Suppose we discover from neuroscience that we can repurpose parts of the human brain as pleasure centers—we can convert brain tissue for intelligence into brain tissue for pleasure (happiness, preference satisfaction). By a simple operation we can thereby increase each person’s level of utility considerably—though their intelligence would be correspondingly reduced. By SPB we should perform this operation on everybody as soon as possible, thus producing a greater amount of utility in the world. Or again, we can genetically engineer people to have more of their brain involved in producing pleasure and less in producing intelligence. We reduce the general IQ but we increase the general UQ (utility quotient). We have happy smiling people, but less intelligent people. That seems obviously wrong to me—we are not doing anyone any favors by this kind of brain modification. Suppose we take it to an extreme, dedicating large portions of the cortex to pleasure, while reducing intelligence dramatically—say, to an average IQ of 50. Someone may have to manage these people’s lives for them, because

268

Part VI

of their general lack of intelligence, but SPB has no objection of principle to that—so long as utility is maximized.1 Again, some people may be prepared to gnaw the pellet here too, so let’s consider another possibility. Suppose it turns out that the pleasure centers in the rat brain are more potent than those in the human brain, so that our resource R can produce a greater quantity of utility in the rat brain than in the human brain—we get more utilitarian bang for our buck by stimulating the rat’s pleasure centers. There will therefore be more pleasure in the world if there are more rats than humans, given R. Are we then morally obliged to reduce the number of humans and increase the number of rats? In the extreme, does SPB instruct us to phase out humans and produce a massive population of rats, each with a highly sensitive pleasure center? It appears that it does instruct us to do just that. Or suppose certain humans had pleasure centers that could be run more economically than those of other humans: should we seek to select those humans and phase out the others? These strike me as absurd (and indeed monstrous) consequences—so the unrestricted SPB is morally unacceptable. We are under no obligation to maximize utility—a world with more utility in it than another world is not necessarily a better world. In fact, we are under an obligation not to maximize utility, given the costs. Here is a final case: Suppose that it is cheaper to run brains in vats than in moving, perceiving bodies, so that R can be used more efficiently to produce utility in brains in vats than in functioning organisms. Are we then obliged to remove everyone’s brain and put it in a vat (assuming the technology is available), so that we can support more happy brains with R this way? We will certainly produce more utility overall. Combining this with my first case, are we obliged to repurpose brains in the direction of greater pleasure and then put them all in vats? Is it a better world to have a huge number of rather stupid brains in vats than fewer intelligent normal people walking around? Surely not—that would be absurd. Suppose we could produce the greatest quantity of utility in the world by repurposing rat’s brains to devote more tissue to their pleasure centers, eliminating all other species, and then putting all the rat brains in vats. That seems like a perfectly plausible scenario—we might well maximize total utility relative to R that way. But I take it no one is prepared to go that far: billions and billions of stupid rat brains floating in vats as the best world we can create. This refutes utilitarianism. 1.  You might object that their “higher pleasures” have been removed by lowering their intelligence: but suppose that this is compensated for by a sufficient increase in their “lower pleasures.”

Why Is It Good to Be Alive?

We tend to assume that it is good to be alive—much better than being dead. But why do we assume that? What makes being alive a good thing? This question is much more difficult to answer than you might suppose, and what I have to say here will provide no definitive answer. I really don’t know why it is good to be alive, though I am convinced that the standard suggestions don’t work. It is different with beauty and virtue. If we ask why it is good for there to be beauty and virtue in the world, then the answer is immediate: because beauty and virtue are good things. Goodness is written into them, as a matter of definition. But the property of being alive is not itself an evaluative property, so we need to find something associated with it that has value. But what is it? Mere existence cannot be the answer, though undoubtedly living things exist: there is nothing valuable in existence as such—or if there is, it is not the distinctive kind of value that we think attaches to being alive. If we agree with St. Anselm that it is better to exist than not to exist, then we are either thinking of living beings like ourselves or employing a very thin notion of “better than.” We should also rid ourselves of the idea that life exists because it is good that it should exist. The evolutionary process was not initiated and governed by any such normative principle; it isn’t that evolution recognized that life is good and therefore strove to bring it about. Evolution is blind and any goodness that life has is merely accidental. God may bring about life because he judges that it is good, but that’s not how evolution brings living things about. And even if God did bring about life because it is good, that does not explain why it is good. Maybe we will do better if we get more precise about what life is. Do we think the life of plants, bacteria, and lower animals is good? If we do, we don’t tend to think it is good in the way our life is—or the lives of other higher animals. It might be supposed that there is something valuable about self-organization and complexity, but that is not what someone is alluding to who declares, “It’s good to be alive!” It must be being alive in a particular way that confers goodness. Here it is natural to invoke consciousness: the reason being alive is good in our case (and that of other higher animals) is that we are conscious. Consciousness is inherently good, and it is what makes life good. But is consciousness inherently good? What if it is painful, miserable, deprived, and depraved? What if it is very primitive, as in snails and jellyfish perhaps? Again, we might suppose that any consciousness, no matter how primitive or long suffering, has some intrinsic value, but surely we mean something more when we speak of our conscious lives as having value. It is not consciousness as such that makes the difference but something special to our kind of consciousness. Also, it is not clear that consciousness is a necessary condition for life having value. If a person is permanently unconscious and dreaming, her life might still have value—her

270

Part VI

dreams might be creative, pleasurable, and deep. And what if the life we have is really just a sleeping dream—does it then lack all value? It might be replied that dreams are actually conscious states, even though the dreamer is “unconscious.” But now consider the unconscious mind: does its existence lack all value? Do we think that a person in a coma with an active unconscious is necessarily living a worthless life? It depends on how we think of the unconscious. Here is a thought experiment: Render every mental state of a normal conscious being unconscious, and then ask yourself whether there is now no value in that person’s life. What if we create a special kind of zombie, one with no consciousness at all but a fully functioning unconscious, suitably hooked up to a body? This zombie might behave just like a conscious being, and have the same mental states as a conscious being—it is just that none of these mental states is conscious. It sounds discriminatory to say that this kind of “disability” amounts to total life worthlessness. What if that is the way Martians are naturally—no conscious mind, but a rich and sophisticated unconscious mind? We don’t want to be guilty “consciousness-ism” (cf. “speciesism”). Even if one disagrees with the metaphysics of this case, finding the idea of such a being metaphysically impossible, it is still possible to judge that life could have value in such unusual circumstances. If we are attracted to a self-monitoring theory of what consciousness is, then we shall not regard the difference between conscious mental states and unconscious mental states as dramatic enough to draw such a sharp normative line. So what if the “zombie” fails to perceive or think about its first-order mental states—what matters is that it has those states. It is mind that matters, not the conscious mind alone. So it is not consciousness per se that makes being alive a good thing. Is it then mentality in the broader sense? But surely not all mentality is valuable— not misery and foul intentions. It must be something more specific and restricted. Now a natural thought suggests itself: being alive (having a mind) allows us to accomplish worthwhile things—art, science, social communities, and moral virtue. We need not limit these accomplishments to the conscious beings among us; we can, less chauvinistically, allow that unconscious (but minded!) beings could accomplish such worthwhile things—if there can be beings of this type. It is said that the unconscious plays a large role in creativity anyway, so we can allow creativity to flourish in the unconsciously alive denizens of the universe (or some possible universe). The advantage of this suggestion is that we are building goodness into the trait in question, that is, worthwhile things. That seems a promising line of thought, but unfortunately it runs into another problem of chauvinism (indeed speciesism). For what are we to say about children, animals, and the less productive humans? Are their lives then devoid of value—no better than being dead? And can’t it be worthwhile to be alive when one is simply idly enjoying the passing show—the trees, the flowers, the sky? The present suggestion is too high-minded and intellectualist. Life can be good, we tend to think, even when genius is lacking and nothing much is being accomplished.

Ethics271

Now an old idea offers itself: pleasure. Life is good because it is pleasurable. We can grant that pleasure is indeed a good thing, though we will want to distinguish higher from lower pleasures and subject the notion of pleasure to some scrutiny. But the problem is that being alive can be worthwhile in the absence of pleasure. If my pleasure centers are destroyed, is my life thereby rendered totally worthless? Don’t you go through large parts of the day feeling no pleasure at all? And yet wasn’t your life worthwhile during those hedonic lulls—better than temporary death anyway? What if our Martians were antihedonists who had abolished pleasure from their lives long ago—can they now be slaughtered without scruple? So it is not pleasure alone and as such that confers value on life. Might we make more progress if we acknowledged a continuum of value to lives? This is a perilous route to take, being open to all manner of discriminatory “isms”: why should a human life—any human life—be taken to be better than the life of a monkey or a dog or an elephant? And do we really want to say that some human lives are more valuable than others? Do some humans have more of a right to their life than others? Isn’t every life uniquely valuable to the being whose life it is? But maybe we can steer clear of these ethical shoals while accepting that what makes life good can come in degrees, even if everyone who meets the condition has a nonnegotiable right to life. What is it that every adult human, every human child, and every higher animal (as well as some “lower” ones: beware the chauvinism!) has in common that makes him or her have a worthwhile life? And let us not forget those unconscious zombies—they must be included too. They are all alive, to be sure, but what is it about their ontological condition that makes their aliveness amount to something good? As I said, I am really not sure, but here is a thought to play with: they all have knowledge. Knowledge is something that can come in different kinds and quantities in different creatures, and hence can confer different degrees of goodness. I here mean to include not just propositional knowledge or “book learning” but also acquaintance with objects and properties, immediate awareness of the environment, apprehension, perception, introspection, proprioception, and even dreaming—the full range of cognitive capacities. These are the basic capacities of any minded creature (though not total zombies that lack even an unconscious mind), so there is not much danger of chauvinism. The question is going to be whether, and in what sense, knowledge in this capacious sense is good. Certainly, it is traditional to regard knowledge as an intrinsic good, and intuition resonates to the suggestion; but we need to say more about what its goodness consists in. It sounds rather feeble to say that it consists in having true justified beliefs: why is that thought to be so life enhancing? What has knowledge in that sense done for you lately? One point that has been made is that knowledge is enriching: it adds to the form and quality of the self, it enlarges the self, it connects the self to what lies beyond its solitary confines. But it is not terribly clear why this is supposed to be such a good thing: eating

272

Part VI

a lot enlarges the self too, and connects the self to the world beyond, but obesity is not generally thought to be what makes being alive worthwhile. The enlargement has to be of something recognized to be good in itself. It might be suggested that knowledge increases understanding and empathy, but here again we run the risk of excluding the less exalted kinds of life, as well as raising questions about those putative values. Mere cognitive contact with things seems a slim basis on which to erect a theory of the value of life. We are in distinct danger of showing that there is nothing that makes being alive good, in the sense in which we seem to mean it. Maybe this is just a false belief that the genes have installed in us for their own selfish reasons—like the belief that there is something especially meritorious about our kin as opposed to strangers. There are worthwhile things that are effects or correlates of life—such as pleasure, beauty, virtue, and accomplishment—but being alive itself seems to lack any inherent value. We can agree that knowledge in the broad sense is characteristic of at least the “higher” forms of life, being built into all mentality, conscious or unconscious; but it is less clear what intrinsic value such knowledge has. Why is it so valuable that an owl can see things with great acuity (putting aside its utility in catching prey)? Why is my ability to be aware of the position of the sun so crucial to the value of my life? What is so good about knowing of terrible tragedies? Here is a suggestion: knowledge gives point, not to the living knower, but to reality outside of the living knower. It is not that human life would be meaningless without knowledge but that the universe would be meaningless without it. It is good to be alive because without the life of knowing beings the universe itself is pointless. What I mean can best be explained as follows. Consider the universe before life ever evolved: nothing was aware of it (I’m ruling out God), nothing cognized its being. It simply was, a brute mindless thing, unconsidered and unknown. Imagine if life had never evolved, so that the universe existed eternally in a state of complete unknownness— just vast stretches of matter, time, and space, with nothing aware of any of it. Doesn’t that seem bleak, desolate, and even tragic? To be so unwitnessed, so unappreciated, so neglected—it seems the epitome of pointlessness. It would be like a book that is never read by anyone ever. Nothing in its tumultuous history would be registered by any cognitive being. It would just churn mechanically away. That seems like a sad state of affairs. Imagine if the universe actually had a soul, a kind of inner psychic dimension: would it not feel an immense loneliness, as if it had been eternally ignored and shunned? Would it not long to create a knowing being that could observe and think about it? What if primitive life evolved in some sector of the universe, creating marvelous and intricate plants—and yet no cognitive being ever became aware of these plants? What if animals evolved somewhere but they completely lacked in

Ethics273

mind, conscious or unconscious—so that they too remained forever unperceived and unknown? What if planet Earth hosted evolved species but none of them had any cognitive capacity, so that nothing of the Earth ever became known to anyone? It sounds so utterly pointless, so futile, and so dead. The very existence of the universe would remain completely unknown, just a giant brute fact, signifying nothing and signified by nothing. Its being would be to be not perceived. Its reality would go utterly unheeded. It seems, then, that a universe that is known is better than a universe that is not known—the former is a better state of affairs than the latter (even if it is hard to see what knowledge does for the knower). So it is good to be alive for the universe, if not for the one who is alive. Our being alive and conscious allows the universe to escape the tragedy of being radically unknown. It is analogous to the universe not being useful if there are no living beings in it: what is the point of containing all that energy if no living thing ever makes use of it? Take the sun: with life it takes on a vital function, being the ground of all organic existence on earth; but if life had never evolved here, all that energy would have gone to waste—and what would the point be of the sun then? It would just be a massive fiery ball hanging up there for no reason, doing no good to anybody. At least if there is life for it to nourish it is performing a useful function. Not that it cares one way or the other; we care that it would exist for no reason. We don’t like to think of the universe as useless and unknown—it offends our sense of decorum and economy. Why create such an enormous and spectacular entity at all if it will never matter to anyone? Mere being is no reason to be. We don’t like to think of ourselves as useless and unknown, and we extend the same sentiment to the universe. A universe without cognizing life in it seems like an enormous waste of time and effort, a pointless joke, an ontological blunder. So when we consider the universe and our place in it, we can tell ourselves that we confer value on it simply by being aware of it—we save it from the fate of existing pointlessly. It is good to be alive because we give meaning to the “life” of the universe. Our life has value insofar as it gives value to something other than us, that is, mind-independent reality. Still, even if there is something to this line of thought—and it is certainly a stretch—we have not succeeded in finding a defense of the idea that it is good to be alive for us. We have found goods in life (pleasure, accomplishment, etc.) but we have found no good to life. That sounds like a rather bleak and depressing result, but it does have a silver lining: death now doesn’t seem like such a bad thing. If we cannot say that life is inherently good, then it is not inherently superior to death: the mere state of being alive is not a better state than the state of being dead. It is true that while alive we can instantiate goods that are not possible while dead (pleasure, creativity, virtue, etc.), but the state of being alive is not in itself superior to the state of being dead. There is nothing intrinsically good about being alive as such, even

274

Part VI

when not in pain and misery. The genes have brainwashed us into believing that life is a supreme good, in contrast to death, but rational investigation has not disclosed any solid basis for that instinctive belief. If life becomes unbearable for one reason or another, we are not violating any normative principle by choosing to enter the state of death. We cannot be accused of negating a basic good by ending a life that has no good things in it. Life may indeed be good in virtue of having good things in it, but not otherwise.

Physical Noncognitivism

Imagine a race of beings elsewhere in the universe that have a psychology rather different from ours. When they make judgments about the physical world around them they experience a set of distinctive emotions: in particular, their color and shape judgments are associated with particular emotions. Suppose they see just three colors—red, blue, and yellow—and that these give rise to emotions of affection, amusement, and ennui, respectively. Thus when they judge that an object is red, say, they experience a strong rush of affection for the object; and similarly for the other two colors. Their judgments about shape are also associated with emotions: circles make them feel joy, rectangles anger, and triangles confusion. This is just the way they are psychologically hooked up; we can think of it as a very pronounced form of synesthesia. Their thoughts and judgments about physical facts are thus saturated with emotion; there are no emotionally neutral physical judgments for them. Let us further suppose that they are noncognitivists about such judgments— specifically, they are emotivists. They do not believe in color and shape facts; they do not think statements about colors and shapes can be true or false; they hold that such statements (if they can be so classified) are expressions of emotion, possibly with prescriptive or hortatory force, not descriptions of anything. If asked what they make of their sensory experience of color and shape, they reply that the senses are misleading: color and shape are not real properties of things at all, though they can give that appearance. Upon further interrogation they explain that they are suspicious of the very idea of a material world (their leading philosopher is a distinguished gray-haired lady called Georgina Barkley), and they think that adopting a “realist” view of color and shape is buying into that suspect metaphysics. They patiently explain to us that speech acts can be meaningful without stating facts, so they are not committed to rejecting all talk of color and shape as meaningless; it is just that they adopt an emotivist view of what the relevant speech acts are all about—they serve to express the emotions that are so strongly associated with talk about so-called color and shape. Strictly, there are no such things as color and shape—we should refrain from “ontological commitment” in their regard; we just, lamentably, talk as if there were such things while experiencing certain emotions. Some of their philosophers prefer to put the truth in prescriptive terms: if someone says, “That thing is red,” she is really recommending that people should feel affection for it. What is important for them is the negative thesis that statements apparently “about” color and shape are about nothing real; these putative entities are just projections of noncognitive states that we express using certain sentences. They grant that sentences containing color and shape words are logically misleading, but they insist that these sentences can be paraphrased to avoid any impression of physical realism. The only facts in this vicinity are facts about emotions.

276

Part VI

Imagine in addition that these hyper-emotional beings are notably cool when it comes to moral judgments. They engage in moral reasoning and judgment as much as we do, but they are not psychologically organized in such a way as to feel anything in the way of moral emotion. When they declare something morally right they experience no emotional pro-attitude, and similarly for declarations of moral wrongness. They have a kind of extreme Kantian moral psychology. Moreover, they are inclined to Platonism philosophically, believing fervently in the Good as an objective universal— they are perfectly happy with the idea of moral facts, moral truth, and moral objectivity. They are accordingly cognitivists with respect to morality. It would be odd for them to take emotivism seriously, because they just don’t have moral emotions, despite their sincere moral judgments—except those that are common to all truth-seeking enterprises. They therefore draw a sharp contrast between ethical statements and physical statements: the former are deemed cognitive judgments in good standing, while the latter are to be interpreted noncognitively. Given their contingent psychology, they are natural ethical cognitivists and natural physical noncognitivists. On hearing about our own psychology and corresponding philosophical opinions, they express some bemusement but accept that our erroneous philosophical opinions are quite explicable, given our peculiar kind of psychological makeup. They earnestly advise us not to let our contingent human psychology dictate the correct philosophical treatment of the judgments in question. They return to hugging red objects and tut-tutting about rectangles while remaining calm about moral matters. The question is: is our type of noncognitivism more defensible than theirs?

Child Liberation

It is a disconcerting fact that even the most enlightened thinkers of the nineteenth century were quite blind to what we would regard today as obvious ethical and political truths. In particular, they were unabashed racists and sexists. They evidently could not see that these attitudes are factually and morally wrong, despite their generally progressive tendencies. This prompts the question: Are there any areas today where even progressive thinkers fail to recognize cases of clear injustice? It seems unlikely inductively that we are free from moral blind spots, even if we pride ourselves on complete enlightenment; at every period people have regarded themselves as morally omniscient, or at least not badly wrong. Why should we be the exception? And yet diligent reflection fails to come up with a blind spot comparable to racism and sexism: so have we finally got it right? Surely at some point in time it ought to be possible for moral thought to become fully enlightened—maybe that time is now. Still, it behooves us to try our best to see where we might be blinkered. One area in which it may be fairly claimed that we are morally myopic is that of animals. We exploit animals in all sorts of ways and we don’t find much of a problem with it. I don’t disagree with this judgment, and I think we are only at the beginning of understanding how wrong and unjust our treatment of animals really is. We are going to need a thorough moral and political overhaul where animals are concerned. However, in this case there is a recognition that something is amiss in our treatment of animals; many progressive thinkers are already onto the point and reform is in progress. This is not an area in which even the most advanced moralists are blind. Many still are, to be sure, but that is equally true of racism, sexism, and other issues. I want to try to identify an area where the blindness extends to the highest level. Consider a society in which adulthood is thought to occur much later than in our society. It is not that people actually mature later in this society: they mature at the same time as ordinary humans, but it is not thought that this is real maturity. Let’s suppose this imaginary society adds an extra ten years to what we regard as the threshold of adulthood. In our society particular ages are selected as marking the time at which certain rights are granted: sixteen to drive, eighteen to vote and give sexual consent, twenty-one to drink alcohol (I am speaking of American society, but other places have a similar set of ages). Suppose we add a decade to these ages, so that the rights in question begin to apply at twenty-six, twenty-eight, and thirty-one, respectively—yet the individuals in question are exactly like the people in our society at those ages. Our reaction to such a society is likely to be that this is far too late to be granted the rights in question. We would not choose to live in a society like that from behind the veil of ignorance, and it seems clearly to be excessively cautious and factually misguided. We can suppose that the young adults in this society are not happy about these laws, but they are powerless to do anything about them. We might even imagine a resistance

278

Part VI

developing, with political organization and protests. Their objection will be that they are competent and mature enough to be able to drive, vote, have responsible sex, and drink. They are, they insist, being deprived of their natural rights. But the real objection is not that the ages chosen are simply too advanced; it is that age itself is the wrong way to allot the rights in question. Calendar age is at best a mere indicator of maturity, not maturity itself. Let’s suppose that in our imaginary society a third of the people under the statutory ages actually are incompetent to exercise the relevant rights—they are just not psychologically mature enough. That fact might be offered in defense of the existing laws: surely we don’t want these childlike individuals driving, drinking, fornicating, marrying, and voting! The obvious reply is that age itself should not be used as the relevant criterion; competence and psychological maturity should. If necessary, people should be tested to determine whether they are sufficiently psychologically advanced. Using chronological age is just hopelessly undiscriminating: people should be judged by their actual physical and mental maturity not by their years on the planet. Some people never grow up—does that mean no one should have the privileges of adulthood? The question, then, is whether our current treatment of children is as unjust and wrongheaded as in my imaginary society. And the problem is that we are operating with a very similar “one size fits all” standard, which has the same injustices. Psychological and physical maturation follow a normal distribution, in the statistical sense, so that some people will be further along than others at specific ages: why should the more advanced individuals not be fast-tracked, as they are in school? Thus, the rights of adulthood should be distributed according to merit, not number of years alive; and people should be permitted to work toward these goals as early as they wish. If a child of twelve can demonstrate the requisite levels of competence and emotional maturity, then he or she should be allowed to do as adults do. Not to accept this principle is “ageism,” like sexism or racism or speciesism. Age, like sex or race or species, cannot be used to discriminate against individuals who meet the appropriate standards of competence and maturity, cognitive and emotional. To do so is arbitrary and morally unjustifiable. Age is just a number. It might be agreed that this is correct as a general principle but argued that we have the ages about right as things stand, at least as an average. It would be costly to test children for maturity, so we just use a rough guide, namely, number of years since birth. This is obviously an empirical and practical question, and jurisdictions vary in the ages they select, but it is notable that many children resent their social and legal status—and one can imagine more resenting it once their consciousness has been raised. What if a political party formed (the United Children’s Front) consisting of children and their advocates, campaigning for lower age restrictions? What if a substantial majority of children were found to be in favor of a general lowering? What if they marched, protested, and engaged in civil disobedience? And clearly many children today do feel that

Ethics279

they should have more freedom. They don’t like being told what to do. Is it possible that they are right and that we have been repressing them for too long? Consider mandatory education: many children don’t like school, especially as it now exists—they would like to get out as early as they can. Don’t they have a right to do so? Not all children go to college, so why should all have to go to high school? The answer will be that they are not intelligent enough to make a rational decision—so they have to be forced. But the same argument was used against women’s suffrage and in support of slavery: are we really so sure that children don’t know what’s good for them? What if children begin to mature earlier—shouldn’t their opinions count for more? We have to be very careful before we discount the autonomy of individual persons—are we being careful enough with children? Do the current arrangements reflect adult prejudice and unequal power? We should at least think seriously about this question. One area in which we definitely do infringe children’s rights is in the matter of indoctrination. We fill their heads with nonsense, religious and other, and they might well protest against the damaging effects of such indoctrination. They have a right not to be deceived and manipulated. We also require obedience from children that is more in our interest than theirs. Then there is corporal punishment and other forms of discipline. I can imagine that the future will go even further in liberating children from such treatment than we have already gone. We no longer think we can flog, imprison, and enslave children, but we still seem to assume that we can force them to attend school, go to bed early, wash their hands, and so on. Are we perhaps just in the middle of child liberation? Might we look back and be shocked at how benighted we were in the way we used to treat children? No doubt we are concerned for their welfare, and they often do stupid things, but the same is true of adults. Of course, we have to care for children in their early years or else they will perish, but it is conceivable that their later selves might deserve liberation from adult supervision and control. Other animals don’t seem to insist on controlling their offspring after the point at which they choose to fend for themselves. Are we adult humans, at this point of history, hanging on for too long, keeping our offspring subservient beyond the natural season of autonomy? We want to protect them from danger, but that is true no matter what age they are; and anyway protection is not the only value we need to respect—what about freedom? Are we guilty of excessive paternalism when it comes to children, particularly teenagers? I can imagine a future in which such thoughts gain a political foothold. If the world becomes economically flush, so that there is no material need, we might find children agitating for greater freedom to live where and as they wish. They might even outpace adults in technical skills, as well as innovative thinking. They might start to complain about “ageist” language, such as the use of “childlike,” “puerile,” and “infantile.” Child liberation! We might come to look back on the present period as a time of oppression and blindness where children are concerned—as we look back

280

Part VI

on earlier times in a similar way. Remember that children lack power—military, economic, and political—and where people lack power they are likely to be victims of oppression. Imagine what the world would be like if children suddenly acquired power over adults, so that they could become the oppressors, or at least gained equal power. That world might be closer to the world that ought to be. Could there ever be a tenyear-old president? Remember that a hundred years ago it was inconceivable that a woman or a black man could be president of the United States. Maybe we are still blind to the natural rights of children.

Modesty and Self-Knowledge

It is said of John Rawls that he was a modest man despite being the best moral philosopher of the twentieth century. We ought to be puzzled by this description: how could Rawls be so modest about his philosophical abilities and yet have those abilities to such an outstanding degree? Call this “the puzzle of modesty.” One way he could be thus modest is if he simply had no knowledge of his philosophical abilities or the importance of A Theory of Justice—he lacked self-knowledge about these things. Such a condition is no doubt possible: he vastly underestimated himself, thinking he wasn’t much better than a typical graduate student. People often overestimate their abilities (looks, talent, virtue, etc.), so why can’t they underestimate them? But that would be a bizarre form of ignorance: he could correctly estimate the abilities of other people but he somehow had a blind spot about himself. This just seems to indicate poor judgment, which is hardly a virtue. We don’t want to say that Rawls’s particular virtue was his lack of self-knowledge. Should we say that he fully recognized his intellectual distinction but refused to admit to it? Alone in his room he allowed himself to entertain thoughts of his greatness, smiling inwardly, fully aware of his importance as a thinker, but if anyone were to ask him about himself he would resolutely proclaim his mediocrity. Suppose his wife had no competence to judge his work and no contact with those who could, and she wondered whether all those hours he spent in his study not attending to his family were worth the effort. One day she asks him whether he is wasting his time and he replies, “I’m a philosophical mediocrity and it’s not worth it.” He doesn’t believe this for a second, and he understands why his wife might be concerned about time spent away from the family, but modesty requires him to lie. That too seems silly and hardly virtuous. He should be honest and sincere with his wife, not flaunt an image of humility he doesn’t feel. Is it then that he would reply honestly to questions about his ability but that he would never assert or allude to it unasked? If you ask him for his self-evaluation he cheerfully responds, “Oh yes, I am the foremost moral philosopher of the twentieth century,” but he never gives any hint of this opinion except when prompted. That too doesn’t ring true: the reply is not at all modest, and his generally not revealing his awareness of his brilliance looks like a form of pretense or deception. Shouldn’t he conduct himself outwardly in accordance with his self-knowledge? Also, he knows that others know how distinguished he is, so he is living with the fact that they know he must know it too, but that he refuses to admit to such knowledge. He is trying not to let them know that he knows what they know. Why is that good? Another option is that he knows his ability and admits to it but that he devalues what it is an ability to do. Sure, he is a brilliant moral philosopher, but moral

282

Part VI

philosophy is no big deal—it is like being a brilliant floor-sweeper or teeth-brusher. I take it this position is obviously absurd. The puzzle of modesty, then, is the puzzle of why self-ignorance should be thought virtuous. Wouldn’t Rawls be a better man if he simply acknowledged his own importance, to himself and others, and then took it from there—not flaunting it or using it to get his way or lording it over others? Wouldn’t he then have self-wisdom? But then, in what way is he modest? He will have proper self-pride, commensurate with the pride he feels in the achievements of his children—there is no virtue in modesty about them (“Oh, my daughter is a mediocre pianist,” said when she is an outstanding pianist). He might downplay his own part in achieving his success, emphasizing his perfect parents, excellent education, and pure luck; but if this veers into dishonesty, there is nothing virtuous about it. What we want from Rawls is a clear-eyed estimation of the merits of the author of A Theory of Justice, both internally and externally, not disingenuous self-criticism and coy obfuscation. But that is precisely what those who praise his modesty seem to be ruling out—they want him to go on as if he is unaware of his brilliance and importance (of course, I am just using Rawls as an example—my point is a general one about ascriptions of modesty). They want him to be blind to it. They want him to be like a Jimi Hendrix who can’t tell that he is a brilliant guitar player: it is a fact evident to all, but he just can’t see it. That is a bizarre state to be in. There is a further layer to the puzzle of modesty: Rawls is also supposed to be modest about his own moral quality, as well as his philosophical excellence. He is modest about his modesty: he has the virtue of modesty but he is modest about having it (and so on up the hierarchy of modesty—as in modesty about one’s modesty about one’s modesty). But again, what is this supposed to consist in? Is it that he simply doesn’t know he is modest? Is it that he knows it but won’t admit it? Is it that he knows it and admits it, but thinks it’s no big deal? None of these sounds right. Rather, Rawls should have self-knowledge, including knowledge of his virtues (if they are virtues). Indeed, we might even go so far as to say that he should be ashamed of his modesty if it results from a lack of self-knowledge: what if it arose from a general absence of self-scrutiny that permitted various vices to blossom? The Oracle exhorted, “Know thyself!” not “Avoid knowledge of thy good qualities!” Of course, preening, boasting, arrogance, and self-promotion are not admirable, but they are not entailed by a simple recognition of one’s actual place in the scheme of things. A sheepish admission of excellence strikes me as more estimable than a feigned ignorance of one’s achievements. What if someone disagrees with Rawls in his modest self-assessment, insisting that he is a great thinker—should he strenuously dispute the opinion of this right-thinking person? The disagreement is either sincere, in which case Rawls is sorely lacking in self-knowledge; or it is insincere, in which case one would hope for more honesty and less coyness. Perhaps what is called modesty in cases like this is a kind of self-deception brought on by misplaced puritanism. Perhaps Rawls was told as a boy that he should never be

Ethics283

boastful or self-important; so despite knowing he was exceptional he deceived himself into thinking he wasn’t. Is that what modesty is in general? If so, it hardly seems particularly virtuous, since it involves deceiving oneself about oneself. Surely the way of virtue recommends that one should form an accurate assessment of one’s strengths and weaknesses, as well as proper comportment with others. If that is incompatible with modesty, then we should be against modesty. I actually doubt that Rawls was modest in the sense people suppose; he just behaved well around others. He had what is called “quiet self-assurance,” with a bit of shyness thrown in. In any case, I would not object if he had a high opinion of himself.

Is Romantic Self-Love Possible?

Is it possible to fall in love with oneself? Can I become infatuated with myself in the way I can become infatuated with another person? If I can’t, why can’t I? Self-love certainly seems possible, indeed widespread, so why not romantic self-love? First, we need to distinguish self-love de re and self-love de dicto: can it be true of myself that I am in love with that person (de re), and can it be true that I love myself as myself (de dicto)? That is, can I love myself without realizing that it is myself that I love, and can I love myself in the full knowledge that the object of my love is none other than myself? In the case of ordinary self-love both things are possible, but what about romantic love? The de re kind of romantic self-love is surely perfectly possible, if unlikely to occur. I might suffer extreme amnesia and then learn all sorts of facts about myself that cause me to become romantically involved with myself, without realizing that it is myself that I am learning about. I might read my journals, see photographs of myself, and hear stories about myself, without realizing whom it is I am learning about. I believe I am getting to know someone else when it is really myself all the time. Someone might play an elaborate trick on me, whereby I am induced to develop romantic feelings for myself, perhaps using hypnosis to fool me into believing that it is a distinct person with whom I am falling in love. I am hypnotized into writing a love letter, which I am then promptly caused to forget, and then later I read the letter taking it to be from someone else to me. Film is taken of me, in which I am disguised from myself, showing me in attractive poses, alluringly attired. This might even be done in such a way as to conceal my sex, representing me as female (though of course there is romantic homosexual love). If the person is called “Nicola” by those around me (who are in on the trick), then I will suppose that I am in love with Nicola, so that the sentence “Colin is in love with Nicola” will be true. And that will be so even though the sentence “Colin is identical to Nicola” is also true. The case is no different, logically, from a case in which I believe of myself that I am a spy, even though I would reject the sentence “Colin is a spy.” I could come to hate myself in exactly the same way, not realizing that the object of my hate is just myself under a different mode of presentation. It is an interesting question what would happen to my feelings once I uncover the trick, but there seems no conceptual difficulty about such a case: I can be romantically involved with a person P without realizing that P is identical to myself. We can imagine a romantic comedy in which just such a theme is explored (“The Man Who Loved Himself”). But the case of de dicto romantic self-love is far more conceptually problematic. On the one hand, the very idea seems out of the question: no matter how narcissistic a person may be, it is impossible to be in love with oneself in the way one can be in love with another—how could one fall in love with oneself? On the other hand, a person can surely make certain judgments about herself that are associated with

286

Part VI

romantic love: she might judge that she is very attractive, amusing, charming, attentive, lovable, and worthy of anyone’s amorous passion. If someone judges herself to be romantically lovable, and rightly so, why can she not love herself romantically? We have a puzzle, a conundrum. Why can’t ordinary self-love shade into romantic self-love? Why couldn’t a case of de re romantic self-love turn into a case of de dicto romantic self-love, once the identity of the beloved is discovered? I am already in love with myself de re, and I judge myself to be eminently lovable by others, so why not go the extra step and become enamored of myself? What is to stop me from doing that? It would be desirable in many ways: I never have to be apart from myself, or worry about my fidelity, or be subject to doubts about my sincerity. It would give me a kind of romantic self-sufficiency. I could live alone and yet experience the joys of romance. I could wake up every morning glad that I am in bed with my beloved. I could say, “I love you” to myself in the mirror and always find my declaration reciprocated. Is it just a prejudice that we can’t romantically love ourselves—should we just try a bit harder, be more open to the possibility? Should I give myself a chance to find a place in my romantic affections? But here is the problem: the emotion of romantic love is bound up with other attitudes and emotions that require that the object of love be numerically distinct from oneself. Take jealousy: normally romantic love is accompanied by the possibility of jealousy, but that is not possible in the case of romantic self-love. I cannot be jealous that the object of my affections has become interested in someone else, since that would involve my being interested in someone else. I cannot be jealous that I am in love with another person! If, in addition to loving myself, I also love Wendy, that is not a reason for me to feel jealous about Wendy—Wendy can’t take me away from myself! Jealousy requires a triadic structure, not a dyadic one. So there is no possibility of jealousy in the case of romantic self-love. Nor is there any possibility of insecurity about the beloved’s affections: I cannot feel doubt about whether the one I love returns my feelings, because I am that person. Insecurity requires another mind whose contents may be doubted, but I know perfectly well my feelings about myself. Nor can I miss myself while I am away—I cannot pine for myself. In other-directed romantic love there is always the possibility of separation—the other can be out of contact. But I cannot be removed from myself; so I cannot yearn to see myself again. Also, in the case of ordinary romantic love there is the joy of discovery, as I get to know the quirks and essence of the person I love; but I cannot discover myself in that way. Lastly, I cannot have unrequited love in relation to myself: if I love myself, then myself loves me. The object of my affection cannot fall out of love with me unless I fall out of love with him. Nor can it happen that I love the beloved more than he loves me. These points are obvious, if never articulated. What is less obvious, though evidently true, is that romantic love essentially involves these other emotions and their ontological underpinning. You cannot love someone in that way without the possibility

Ethics287

of jealousy, insecurity, missing the person, discovery, and the chance of unrequited love. That is what romantic love is—what constitutes it. It is not just an isolated feeling—a kind of qualitative sensation—that can exist independently of the surroundings of romantic love; it is a complex of attitudes and beliefs that require the numerical distinctness of the lover and the beloved. If I had the belief that I alone exist, then I would not be able to feel romantic love, no matter how much I might yearn for it. I could feel ordinary self-love, but not the romantic kind. In fact, there is something odd about calling ordinary self-love an emotion at all, whereas romantic love is an emotion par excellence. Romantic love is a complex multifaceted emotion, which may explain some of its peculiar power.

Against Laughter

One of the jobs assigned to the philosopher is to suggest new and better ways of living. In that spirit I want to suggest that laughter is not a good thing and should probably be banned. I am aware that my suggestion will sound counterintuitive to some readers— indeed curmudgeonly and insane—but once a certain distinction is made it should emerge as anything but. The distinction is between amusement and laughter: I am all in favor of the former; it is the latter that I wish to outlaw. Laughter is a dispensable and crude expression of amusement, with much to be said against it, and nothing for it—a kind of vocal thigh slapping. Compare shrieking. When children gather, especially around water, they tend to shriek as they play. They are excited and enjoying themselves, and their natural expression of this is shrieking. It is high-pitched, loud, and unpleasant to listen to. Adults tell them not to do it, or to keep it down. As children mature, they grow out of shrieking, even when hit by an unexpected wave. This is good and admirable: an adult who insists on shrieking like a child when hit by a wave does not earn our approbation. Nor is it necessary to experience excitement and enjoyment that shrieking should be going on. Laughter is like that: it is an immature and annoying expression of amusement, which is in itself a fine and valuable thing. We should therefore strive to grow out of laughter. It should be consigned to our evolutionary past, like baying at the moon. Most people do not giggle and guffaw all the time, though they may be continuously amused. They are aware that that would be considered extreme, uncalled for, and disagreeable to others. Some people, often those with an acute sense of humor, rarely laugh. Older people tend to laugh less than younger people, though they are not less amused. Laughter may indeed get in the way of amusement, because it is so physically convulsive—it interferes with concentration (you don’t want to be laughing your head off all the time while reading, say, Max Beerbohm). Laughter involves loud noises, a moistening of the eyes, and an inability to speak. The face contorts and writhes. The inward amusement spills out in a grotesque pantomime—like a sort of reflexive shrieking. We sometimes can’t stop laughing, as our will is commandeered and stolen from us. We may do it in the most inappropriate of places—funerals, churches, lecture-halls, or the doctor’s office. It is anarchic and disruptive. Someone who can’t stop laughing is a figure of mild contempt, or perhaps sympathy. Laughing is like one of those involuntary motor disorders that prompt a visit to the neurologist—a kind of universal Tourette’s syndrome. This is why, if we tolerate laughter at all, we insist that it be kept within reasonable bounds. We clearly don’t think untrammeled laughter is a good thing. As laughter must be distinguished from amusement, so must it be distinguished from smiling. I am not opposed to smiling—it is laughing without the bad bits. It

290

Part VI

is quiet, seemly, and controlled—nothing like that wild shrieking. If laughing is a sort of shouting, then smiling is a whisper, a decorous exhalation. Smiling communicates amusement without screaming it from the rooftops. In the course of growing up, smiling generally replaces laughing—the civilized replacing the raucous. The Mona Lisa smiles demurely; she does not crease her angelic face with uncontrolled guffaws. People tend to laugh most when in groups, when they “get together”—say, when they gather in a theater to watch a comedian. Here they lose all sense of restraint and decorum: they are loud, backslapping, and generally savage. They try to outdo each other in the intensity of their howling. Being in among them is apt to thwart real inward amusement, so distracting is their ceaseless snorting and whinnying. They had a “good laugh”: but was their laughter good? Another count against laughter is that it is often allied with cruelty—by no means do people laugh only at what is funny. Laughter can express viciousness, a sense of superiority, delight in suffering. A man with absolutely no sense of humor could spend a lifetime laughing—at the misfortunes of others. Such a man may never even smile, though a rictus of laughter contorts his sneering lips. He has an “evil laugh.” Do we want to imitate him? Surely not: we want to keep as far away from him as possible. Now it may be asserted by the devotee of laughter that amusement is not really possible without laughter—the James–Lange theory of amusement: we don’t laugh because we are amused, we are amused because we laugh. But that is obviously a false theory, or else there could be no such thing as silent, calm amusement; and even when we are laughing, this does not constitute our amusement. We never need to laugh in order to be amused, even mightily amused. Mute paralysis is quite compatible with a richly humorous inner life. When a person “laughs till she cries” the tears are not a necessary condition of the inward amusement; they are merely a physiological accompaniment, and quite dispensable. We should reject behaviorism about amusement: amusement is essentially an inner state. So much for theory; what about practice? I would ban laughter from public places, like smoking—who wants to be subjected to other people’s second-hand laughter? No public cachinnation! People are not permitted to shout in a restaurant or a movie theater, so why should they be free to emit loud cackling noises? Reproving glances should suffice to discourage excesses of mirth, unless children are involved (we must be tolerant of their weakness). For those without the self-control to stifle a laugh, I recommend treating it like sneezing: don’t do it all over other people, but smother it in a handkerchief and apologize afterward. We should also all endeavor in our private moments to put a damper on the snickering and snorting, even when entirely alone. Let us by all means be amused, let us be convulsed inwardly, let us be full of the joy of humor, but

Ethics291

please let us not succumb to the urge to “crack up,” to “lose it,” to “roll in the aisles.” No more piercing uncouth guffaws; no more griefless tears; no more baring of the teeth and stretching of the face. A subdued smile is perfectly acceptable, with a sparkle of the eyes and an expression of appreciation—but no more falling about, no more gasping for air, no more simian screeching. Without such distractions and excesses, amusement can pursue its proper course, and perhaps reach even higher peaks of refinement and intensity. In the humorous life laughter should have no part.

VII Religion

A Deontological Theodicy

The classic problem of evil claims to find an inconsistency in the joint existence of God and evil. If we define God as a being that is all-knowing, all-powerful, and all-good, then God’s existence will be inconsistent with the existence of evil, since any perfectly good being will prevent evil given that he knows about it and has the power to do so. Thus the traditional concept of God is said to be paradoxical, given the existence of evil. Other familiar concepts also lead to paradox: truth (the semantic paradoxes), sets (Russell’s paradox), and any vague concept (the Sorites paradox). Yet we don’t normally suppose that these familiar paradoxes demonstrate the nonexistence of the things in question: we don’t conclude that nothing is true or that no sets exist or that no one is bald. That is because we think we have good antecedent reasons to believe that these things exist, and so we set about trying to resolve the paradoxes—not eliminating the things in question. We often find no convincing resolution. If we had strong reason to believe in the existence of God, then presumably we would adopt the same attitude toward the paradox presented by the concept of God: if we thought the ontological argument was sound, say, we would not reject the existence of God in the face of the problem of evil, but try to disarm the paradox. But since we don’t have such a convincing antecedent argument, we tend to suppose that the paradox of God and evil disproves God’s existence. Since we can’t deny the existence of evil, we end up denying the existence of God. I am interested in whether the problem of evil really demonstrates that the existence of God is inconsistent with the existence of evil. For if it does, the inconsistency is not as obvious as (say) the inconsistency of the concept of a round square, or the concept of a chair that is not a piece of furniture, or the concept of a married bachelor. My question, then, is whether it is possible to dissolve the appearance of paradox, thus rendering God’s existence at least consistent with evil (it is a further question whether God exists). Is the traditional concept of God as omniscient, omnipotent, and perfectly virtuous inherently and irremediably paradoxical, given the existence of evil? I shall argue that it is not. The essence of the problem concerns God’s apparent tolerance for suffering, especially extreme suffering on the part of innocent people (and animals). This seems to show that he is not perfectly good, given his power and knowledge; indeed, it is often taken to show he must himself be quite evil. Before I tackle this problem head on I want to consider a related problem, which I shall call “the problem of insufficient goodness.” There is a certain amount of goodness in the world, but there could be more goodness in it—people could be happier, more knowledgeable, more virtuous, more aesthetically sensitive, and whatever else you think constitutes goodness. Given that God is benevolent, why doesn’t he produce more goodness? Why doesn’t he produce the maximum amount of goodness? Not to do so looks like a moral failing on God’s part. But then

296

Part VII

God cannot be perfectly good, which by definition he is supposed to be—in which case he does not exist. God tolerates a lack of maximal goodness in the world, which shows that he is less than perfectly virtuous; so the traditional concept of God is inconsistent with the moral imperfection of the world, given that this is not the best of all possible worlds. That argument sounds fishy and unpersuasive—why? Because it presupposes a form of extreme utilitarianism: it assumes that one of God’s duties is the maximization of happiness (or other intrinsic goods). But that type of normative ethics has all sorts of well-known problems, construed as a theory of moral duty or obligation. I won’t rehearse these problems here, merely observing that we don’t generally think an agent is immoral just because she does not commit herself to generating as much happiness in the world as is humanly possible. We don’t think we are obliged to drop everything and travel abroad to help the poor, leaving family and responsibilities behind, even if we accept that doing so will maximize net utility. Of course, there are philosophers who think we are seriously immoral for not doing so—those who accept the extreme utilitarian theory. But many do not—those who adopt a more deontological approach to ethics. The former philosophers think we are the moral equivalent of murderers for not devoting ourselves to helping the poor, sick, and dying; but the latter philosophers protest that this is a highly revisionary account of moral obligation, which they see no reason to accept. They define moral duty by a set of specific rules, not by the imperative to maximize utility. Let us suppose that God is a deontologist, not a utilitarian (rightly so, in the opinion of many). He therefore does not believe that he has any duty to spread as much happiness as he possibly can; nor does he think that mortal beings have any such duty. He will accept that it is a good thing if people (and animals) are happy, but he does not believe it is morally binding on him strenuously to bring about such happiness—just as human deontologists do not believe it is morally binding on us to do everything we can to increase human (and animal) happiness to the maximum. As a deontologist, God thinks it is his duty to obey a list of moral rules—such as not lying, stealing, murdering, breaking promises, being unjust, committing adultery, coveting one’s neighbor’s ox, and so forth—but among these rules there is none that requires him to maximize happiness among all sentient beings (the Ten Commandments contain no such edict). So failing to live up to that extreme utilitarian principle does not, for God, count as a lapse from moral perfection—any more than it does for you and me. This means that if we live in a world in which happiness is not at the maximum possible, God cannot be accused of moral imperfection—so there would be no argument against his existence based on the absence of perfect happiness (whatever exactly that might mean). There is therefore no paradox involved in the idea of a less than perfect world that contains an all-knowing, all-powerful, all-good God. And the reason is that morality, correctly understood, simply does not demand the production of maximum general happiness.

Religion297

If God were to be found guilty of some other moral transgression—say, lying or breaking a promise—then indeed that would show him to be less than perfect, in which case he could not exist as traditionally conceived. But he cannot be faulted for failing to do what is not morally required of him. To use traditional language, he can be convicted of no sin. Can he be faulted for not removing all forms of suffering? Can an ordinary mortal be faulted for that? Again, the utilitarian says yes, but the deontologist says no. Suffering comes in many forms, from mild irritation or displeasure to agonizing pain. Is it anyone’s duty to set about alleviating all forms of suffering, no matter how trivial and mild? Are you required to devote yourself to making people feel a little more comfortable, by removing every hint of disagreeableness from their lives (so long as you do not suffer as a result of your generosity)? Hardly. Nor is God required to do any such thing. The world might be a better place without any suffering in it, but it is not anyone’s moral duty to bring such a world about, even if he or she has the ability to do so. Certainly, we do not go around blaming people for not embarking on comprehensive antisuffering expeditions—striving to remediate every slight pang of unpleasantness in as many lives as possible. We do not take ourselves to have a general duty to remove suffering wherever it is found. But what should we say about extreme suffering? Already we can see that this will not be a simple all-or-nothing matter, given that not all suffering must be alleviated. Is God at fault for tolerating extreme suffering? What about death, destruction, and maiming, caused by earthquakes, fires, and floods? Here is where our moral intuitions start to turn against God, because we think we would never allow such terrible things if we were in his position. But is that so obvious? Let us imagine that there are many universes, each containing innumerable life forms: let’s say a billion universes with a billion advanced life forms in each. Suppose God lives in one of them, but that he has power over all. Suppose too that he has other things on his mind, apart from mortal suffering and its alleviation (he has to think about creating further universes, about his plans for sending a savior to a given universe, about where to send the billions upon billions of dead, and so forth). At any given moment there is a lot of suffering in the totality of universes, which God could in principle prevent. Is he morally deficient for not doing so? Well, consider his emissary Jesus: he lived in one place all his life and did nothing for people outside of that place, despite his evident ability to perform miracles. Why didn’t he travel more widely and use that ability to help people in foreign lands? Why didn’t he devote himself to combating extreme suffering wherever it was found? He didn’t, but we don’t generally suppose him to be a moral monster. The reason is that we don’t take ourselves to be moral monsters just because we ignore extreme suffering in different parts of the world. That is, we don’t think we have a moral obligation to abandon home, family, friends, and work, in order to alleviate as much severe suffering as we possibly can. For we are not extreme utilitarians, but moderate deontologists: we think

298

Part VII

we have core local duties that are binding, and these do not include generalized impersonal suffering-alleviation. It is quite true that I know of distant suffering and also that I have the power to do something about it, but I don’t think I have a duty to do everything I can to stop it (such as contribute most of my income to charity). I don’t think I am morally obliged to tend to animals dying in remote forests, nor to help with the pain of strangers I know are in pain. If the suffering were right in front of me, I might think differently; but as things are I don’t think my duties extend as far as the extreme utilitarian suggests. In fact, I think the utilitarian philosophy leads to a bad ethical outlook, because of the various problems with it that have been pointed out (especially the way it conflicts with questions of justice). We do regard ourselves as failing in our duties if we lie or break promises or act unjustly, but we don’t regard ourselves as under an obligation to extinguish as much suffering in the world as is humanly possible, even when the suffering is extreme (and how extreme would it have to be?). Then why is God under an obligation to eliminate suffering? He may deplore suffering and feel compassion for the sufferer, but why must he do everything he can to prevent and alleviate it? Like the rest of us, he has other things to think about and doesn’t want the distraction; he may also have more local concerns (such as whether to send his only son back to Earth on a return mission). As a deontologist, he accepts that he must not cause harm and suffering, but he doesn’t accept that he has to remedy it whenever it occurs. He thus acts rightly according to his own morality, and according to the morality that most humans share. God is a good Kantian. So far, then, we have found nothing in God’s conduct that conflicts with this kind of morality, and hence nothing that could disqualify him from moral perfection (assuming such a morality to be correct). He is perfect, from a deontological point of view. He has never broken any sound moral rule. He has never sinned. He “tolerates suffering” in just the way the rest of us do: he allows a vast amount of it to happen when in principle he could prevent it. If God intentionally brought about extreme suffering, then he would have a lot to answer for, since there are moral rules against that; but it is a very different matter merely to allow it to happen when you could in principle do something about it. Only a simple-minded utilitarian could fail to appreciate that distinction, but a deontologist finds it vital. Two objections might be raised. The first is that it is clearly wrong to fail to alleviate preventable suffering when it is right in front of you. That may be true, but why suppose that God is present at the scene of suffering whenever it occurs? God is not present in the spatiotemporal world at all, as we are; he always transcends that world. We might think of him as always distant, as we are in relation to foreign lands, but existing in his own dimension—though, being omniscient, he knows exactly what is going on. The distinction between being there and not being there does not exist for him, so he is not ruled by the demands of spatial or temporal proximity. He is not literally standing right there next to the wounded man and coldly declining to help him.

Religion299

The second objection is that we have not yet explained how a perfect God could create an imperfect world—one that he knew would have lots of preventable suffering in it. But that is a separate question, not the question I am trying to answer here—which is why God tolerates suffering when it occurs. He might have had all sorts of reasons for creating the kind of world that he did (one might be that the notion of a perfect world is incoherent). Note that people who create things often know that evil will occur in those things, but they do it anyway—as when parents create a child that they know will suffer in the normal course of life, or an architect creates a building that is bound to have bad things happen in it. We do not generally regard people as to blame for these creative acts. In any case, my question is different, namely how to explain God’s tolerance for suffering. And the point is that tolerance for suffering, of the kind that God evidently exhibits, is not actually inconsistent with unimpeachable virtue—if we accept the deontologist’s view of morality. In effect, the critic of God is presupposing an extreme form of utilitarianism: but that is not a position we are required uncritically to accept. God is not claimed to be unjust by the traditional argument from evil: if he were, that would be straightforwardly incompatible with his complete virtue. But what should we say about the punishments of hell? They certainly seem disproportionate and cruel, not just at all. I agree: the existence of eternal damnation would be inconsistent with God’s virtue. But supervising hell is not part of the very definition of God, merely a contingent add-on: so the right thing to say is that there is no such place as hell. There cannot be a hell, given that God is just—which he by definition is. So that part of traditional theology has to go—and a good thing too. But earthly suffering is a fact, so we cannot just deny its existence. What we need to do is explain how the existence of suffering is consistent with the existence of God as morally unimpeachable—and the suggestion is that it is because of the correctness of deontological ethics, which does not condemn the mere tolerance of suffering. If we had a proof of God’s existence, we could use the problem of evil to prove the correctness of deontological ethics, given that it is the only way to remove the apparent inconsistency. What I am suggesting, reversing that, is that if we assume deontological ethics we can render the existence of God nonparadoxical. I think this is easy to see by considering the (alleged) problem of insufficient good, since we don’t have much of a tendency to think there is a moral rule requiring everyone to promote as much good as possible as widely as possible. But then it is a small step to accepting that neither do we have an obligation to alleviate all suffering, whether mild or extreme. Our moral duties are more limited and more local than that, as deontologists have long urged. Thus the resolution of the problem of evil turns on which ethical theory is correct; there is no knockdown argument that God must be failing in his moral duty unless he prevents or remedies every instance of suffering. Since I favor deontological theories, I am ready to give God a pass with respect to his moral behavior. Nothing in the traditional problem

300

Part VII

of evil shows that God is a sinner. The concept of God coexisting with evil is not paradoxical after all. It is because we are operating with a deontological conception of morality that we don’t immediately accept that the existence of evil shows that God must be less than perfectly good. It takes an injection of utilitarianism to make us wonder whether he might be flawed after all. Let me end with a parable designed to make the point vivid. Suppose a certain individual, call him Peter, has always had a strong desire to help people, especially medically. He studies medicine and becomes a doctor. He is an exemplary member of his profession, working long hours, often for no pay, always kind and attentive to his patients. Moreover, he is a sterling family man, a great husband and father. Peter’s code of personal ethics is very demanding: he has never told a lie, stolen anything, broken a promise, or committed adultery in his heart. In addition, he does a lot of work for charity, giving up what is left of his free time. Peter is widely regarded as a “saint”; no one has a word to say against him. He ends his life without a blot on his record and is sorely missed. There are many who describe him as morally perfect. But there is one person in Peter’s town who is a convinced utilitarian: he argues that Peter is actually not much better than a murderer, since he never left his town to work with dying people in Africa. He could have saved more lives if he had abandoned his practice and his family and moved abroad. Suppose this is true: by the utilitarian calculus Peter really could have maximized happiness by moving to Africa, thus saving more lives and alleviating more suffering. That is just the way the balance of utilities pans out. Instead of curing 10,000 people he could have cured 15,000 people. The question is: Who do you think is right, the utilitarian or the rest of the people in the town? Is Peter evil for not acting in the utilitarian way? He “tolerated evil” by not preventing as much of it as he could. Isn’t God in the same position as Peter?

God and the Devil

Suppose you live in a town with an exceptionally good mayor, call him Keith. Not only is Keith an excellent mayor, he is the best mayor in the world—none greater. He is admired by all, even worshiped. But there is also another man in town who is a terrible criminal, call him Mick. Mick is a very bad man indeed, constantly doing quite evil things, hated and feared by all. He spoils the experience of living in the town. He can occasionally be seen about the place, with his orange skin and flamboyant hairpiece, always up to no good. Keith also is regularly glimpsed, with his healthy glow and fine head of hair. They are never seen together, though—no doubt because of their very different lifestyles and tastes. Everyone just assumes that Keith wouldn’t be seen dead with Mick, yukking it up in some seedy strip bar. There is a puzzle, though. Keith is not only good; he is powerful—he has the police force under firm control (and there is no corruption there). Why doesn’t he just arrest Mick, put him out of action, thus removing evil from the town? Hell, why doesn’t he just have Mick quietly eliminated? You, as a concerned citizen, find yourself obsessed with this question—though other people seem unperturbed by it. Keith has the power, he is supremely good as a mayor (no kickbacks and such), and he is well aware that Mick causes a lot of suffering to the townsfolk (Keith has security cameras everywhere). Yet he allows Mick to persist and thrive, always up to his evil tricks, quite incorrigible. It is true that Keith sometimes publicly reprimands Mick and warns young people to stay away from him, but he never does anything—he lets Mick be. Why? Why? After some years of observing this curious coexistence and puzzling over it, you formulate a daring hypothesis: Keith is Mick! That is, Keith and Mick are one and the same man: it’s a classic Frege case of one individual having two modes of presentation, and hence being supposed nonidentical. This is why they are never seen together, and why Keith lets Mick live among them and get away with his evil deeds. You have also noticed a strong resemblance in the slope of the nose and other telltale signs. Why Keith should feel the need to live this double life is an additional puzzle—maybe he just enjoys the deception, or likes to slum it once in a while. And suppose you are quite right: the two men are indeed one. You have good reason to believe this, and it is true, and nothing else makes sense—so we can say that you know that Keith and Mick are numerically identical. This is despite the fact that one is supremely good and the other supremely bad—it just turns out that one person has both sets of characteristics. Odd, but human nature is odd. Do we not face the same puzzle with respect to God and the Devil? God and the Devil coexist—one supremely good and the other supremely bad. God is also endowed with tremendous power, being actually omnipotent. He could easily eliminate the Devil or curb his evil ways, and yet he does nothing—he lets the Devil be. They look and sound very different, to be sure, but that proves nothing. Also, we observe a suspicious

302

Part VII

similarity between the two (analogous to the nose): the Devil actively pursues evil, but God allows evil to occur when he could clearly prevent it. God also seems excessively punitive when it comes to his punishments in the afterlife, almost as if he relishes the eternal suffering of sinners. Some theologians try to explain these unfortunate traits away, but it is hard to deny that there is something worrying here. So we might reasonably consider the hypothesis that God is identical to the Devil: that would explain the fact that their apparent rivalry never turns into actual elimination, or at least confinement. And, of course, they have never been spotted in the same room together, so that their numerical distinctness can be witnessed. God, in fact, always seems curiously otherwise engaged when the Devil is going about his devilish business. We have here a classic Frege case at the level of supernatural beings: different modes of presentation of the same individual. The identity statement “God = the Devil” is a posteriori and synthetic, with the singular terms in it bearing different senses, though having the same reference. We have discovered it to be true (let us suppose) by reasoning about the case—it certainly wasn’t self-evident simply from the meaning of the terms. It would be quite wrong to reject the identity hypothesis on the grounds that our concepts of God and the Devil are very different—that would be a conflation of sense and reference.1 So the hypothesis is an epistemic possibility. It is true that the identity claim would cause a good deal of cognitive dissonance in us, as in the case of Keith and Mick; but from a logical point of view, it cannot be ruled out. 1.  Someone might argue that our concept of God includes the idea that God is perfectly good, which would not be the case if he were identical to the Devil. But this argument begs the question against the identity claim, since that is precisely what we should abandon if we accept identity. And there is also the point that God is already under some suspicion because of evil and hell. Why God might choose to live such a double life is a difficult question of divine psychology; but it is not to be supposed that the psychology of God is totally transparent to us. We do know that God must be capable of evil, because of his omnipotence; so he must have his inscrutable reasons for exercising that capacity in the guise of the Devil. This makes more sense than supposing that he allows the Devil to exist and ply his trade when he could so easily improve things by putting him out of business.

A Religion of Hate

Christianity exhorts everyone to love everyone. Instead of loving only yourself, your family and friends, your tribe or country, you are told to love even your enemies, as well as those who are complete strangers. The idea is that universal love will lead to universal justice—good ethical behavior will result from generalized love. Christianity says: Don’t hate anyone! The trouble with this prescription is that it is unrealistic: we cannot spread our love so widely, and some people really don’t deserve our love. The pretense that we can will lead only to emotional dilution and insincerity, as we force ourselves into a posture of love that we do not feel. Nor is such universal love necessary for justice: it is possible to treat someone justly—an enemy or a stranger—without feeling any affection for that person. Indeed, it is possible to treat a person justly that you hate (this is the best form of justice). But that leaves the question of what emotion you should feel for people in general: should there be no uniform emotion, just variations on love, hate, and indifference? That kind of mixture is what seems natural to humans, so advocates of universal love are opposing what comes naturally. But the problem with the mixed kind of emotional setup is that it leads to strife, injustice, and disharmony. That is precisely why Jesus and others advocated universal love. So are we condemned by our emotional nature to human discord? Some religions attempt to transcend human emotion—they recommend a kind of emotional distancing from others. No hate, but no love either—just detachment. But that too is psychologically unrealistic, and certainly not natural to humans. Then what is left? Universal love won’t work, a mixture of love and hate is the basic problem, and detachment is not feasible either. That leaves only one possibility: universal hate. This is the type of religion I wish to defend. Note that the hate must be universal: you must hate everyone (with the possible exception of yourself, but even then …). Then you won’t be guilty of treating some people better than others—you will respect the value of equality. If you find everyone hateful, your hate won’t prejudice you against any one individual or group. To recommend universal hate is not to condone unethical behavior: you must treat everyone fairly and justly, with due respect. You must be universally ethical, while hating the objects of your good actions. This kind of emotional stance seems to me psychologically realistic: it is quite easy for us to hate others, and people are objectively hateful. If we are all sinners, we should be regarded as such—hate the sin and hate the sinner. We are a fallen species, violent, vengeful, cruel, selfish, petty, envious, and spiteful—so let us acknowledge that. What is crucial is that we don’t acknowledge it only for certain people: hatefulness begins at home, in family and neighborhood.1 It is not difficult to get into a state of generalized hatred—you just 1. I read a story in a newspaper about a mother whose children asked her which of them she loved the most. She replied that she disliked each of them equally. This is the spirit of the religion of hate.

304

Part VII

need to cultivate your inner misanthrope. You know what people are like from your own personal experience, as well as from knowledge of history, so you can safely generalize a judgment of universal hatefulness. And it is certainly true that the people you happen to associate with are unlikely to be objectively particularly lovable—they will be as hateful as any other group of human beings. An objection may be raised: What about love? Are we forbidden from loving anyone in the religion of hate? No! For we can love and hate: love and hate are contraries, not contradictories. Our generalized emotion of hate can be overlaid by an emotion of love, thus producing a state of ambivalence, more or less pronounced. Hate is the default condition, but love may supervene. Every marriage is a love-hate relationship, because human nature is clearly revealed in marriage—and it is not always (ever?) pretty. Also, this love will be more valuable and meaningful against a background of universal hate; it will be experienced as an achievement, both of the lover and the beloved, not just a reflex feeling. You must love despite the hate. Both emotions must be kept in precarious balance, each leavening the other. That is mature emotion—informed, realistic, and clear-eyed. It is not the unthinking simple-minded love of an animal or child (they don’t know how hateful we humans are). We may even within the religion of hate form a casual liking for someone we meet—while never forgetting that we also hate him or her. We are not against love; but we are for hate. But now a further objection may arise: Doesn’t the acceptance of local love vitiate the general purpose of the religion of hate? We were trying to prevent the ills of mixed love and hate, by advocating universal hate, but now we have reintroduced love, thus restoring the problematic asymmetry between the loved and the hated. Aren’t we back where we started (with universal Christian love now threatening)? I take this objection very seriously, and we do indeed need to guard against the ill effects of local love. What saves the religion of hate from relapsing into the mixed position is that universal hate is preserved in the face of local love—we must still hate those we love. We cannot divide humanity exclusively into those we love and those we hate, because that leads to all the human wickedness we witness; but we can avoid this by insisting that everyone be accorded his or her fair share of hate. No one is beyond our hate, not even those we love most. Thus we preserve basic emotional equality. Some may feel that hatred is intrinsically a bad thing, while love is intrinsically a good thing. But that is completely wrong: there is nothing good about loving Hitler and Nazism, and there is nothing wrong with hating injustice and cruelty (as well as the people who perpetrate such things). It depends on what you love or hate and for what reasons. Hatred is not in itself bad, though many instances of it undoubtedly are. Nor is hate intrinsically unpleasant, especially when it is fully justified; it can be quite invigorating. Hatred is part of our natural emotional economy, and it can be both rational and useful. Of course, we should not aim to promote irrational hatred

Religion305

of humanity as a whole; but a rational and moderate hatred is perfectly justified. If we consider a hypothetical species, stipulated to be especially hateful, then a religion of hate would appear to be the only rational system, perhaps tempered with some local love. If this species insisted on dividing themselves into the objectively hateful and lovable, they would not only controvert the facts, they would also generate the same kind of strife we see in human populations. And preaching universal love to them would be a doomed project, since the evidence of their hatefulness would be palpable. True, they should always treat each other justly and ethically, but that is quite orthogonal to the question of love and hate. It seems to me that the story of Jesus Christ fits the religion of hate better than the religion of love. Judas, the Pharisees, Pontius Pilate—are any of these characters lovable? No, they are hateful, even those of Jesus’ own tribe, including his disciples. His story justifies extreme misanthropy, not love of humanity. We killed the Son of God! And why exactly should we contort ourselves into loving others, despite their manifest hatefulness? Don’t say it is because this will enable us to treat others morally: morality requires not love, but justice. As Kant says, we must be moral to those for whom we feel no affection—indeed, that is the highest form and essence of morality. Morality is about overcoming our hatred for others—our sense that others are cowardly, defective, malicious, and evil. The crucifixion of Jesus represents all that is hateful in humanity, so we should acknowledge that people deserve hatred (though they may also in some instances deserve love). The story of Jesus (or Socrates) illustrates perfectly the hatefulness of humankind; the question is how to incorporate recognition of that into a religion that minimizes conflict and harm. My suggestion is that generalized hatred is the best method, because it is (a) justified, (b) natural, and (c) egalitarian. It does not discriminate, by demonizing some and sanctifying others. Above all, it prevents preference for those close to us and animosity for those outside the circle of those we naturally love—at any rate, it helps to counteract unfair preference based on love. We should feel partially alienated from everyone, suspicious of everyone, and not entirely thrilled with anyone—so that no sharp division exists in our minds. Can we hate some people more than others? By all means—in fact, I strongly recommend it. Not everyone is equally hateful. Agreed, this is likely to generate divisions in our emotional attachments, but as long as an underlying note of hatred exists the worst kind of insider–outsider division can be avoided. Shades of gray in degrees of hatred are natural and healthy, and they are compatible with a basic emotional egalitarianism: first hate all equally (but not immoderately), but then hate some more than others. There is no point in straining to love those you really hate and have no desire to love, or those to whom you are completely indifferent; and there will always be those few whom you spontaneously love. The trick is not to forget your general

306

Part VII

unspecific hatred, or your specific hatred of those for whom you also feel love. Since everyone, with rare exceptions, behaves hatefully some of the time, that should not be difficult. Isn’t this a rather negative religion? Not really: hatred is not a negative emotion—it is perfectly legitimate and normal—and human hatefulness is a fact that has to be accepted. The point is to give it a prominent place in a religious outlook that seeks to overcome natural human bias. It also gives us a religion suitable for the convinced misanthrope, instead of the partial misanthropy of some religions and the hopeless idealism of others. To paraphrase the Beatles: “All you need is hate.”

Index

Abstract–concrete distinction, 121–123, 125, 169 Actions, 41–43 reasons and, 45–48 Adultism. See Ageism Advertising, 235 Aesthetics and truth, 140 Affective conception, 24 Ageism, 278, 279 Altruism, 229–239 Amusement vs. laughter, 289 Analysis and mystery, 193–197 Analyticity, 192, 201n1, 224. See also Analytic– synthetic distinction; Analytic truth(s) apriority and, 157–158, 160 necessity and, 158, 160, 175 Analytic–synthetic distinction, 119, 161, 168. See also Analyticity; Analytic truth(s) Kripke and, 159, 160 necessary–contingent distinction and, 199, 200, 201n1 Analytic truth(s), 115, 117, 162, 166–168, 193, 195, 224. See also Analyticity; Analytic– synthetic distinction meaning and, 136 Animal identity, 208, 209 Antiprivacy argument, 89 Antirealism refuted, 177–179. See also Realism A posteriori fact, concept of an, 126, 169, 175

A posteriori truth. See also A priori truth and a posteriori truth contingent truths as, 157, 159, 199, 201n1 defined, 115–117, 125, 126 necessity and, 159, 163–165, 199 “A Priori and A Posteriori Knowledge” (McGinn), 164n6 A priori as an epistemological vs. a metaphysical concept, 160 A priori fact, concept of an, 126, 157, 168, 169, 175 A priori knowledge a priori truth and, 115–116, 170n8 (see also A priori truth) extramental reality and, 224 A priori truth, 157. See also Apriority conceptual analysis and, 162 defined, 115–117, 122, 125, 126, 157–158 empirical facts and, 168–169 essentially empirical essences and, 166–168 Kripke on, 116–118, 157n1, 158–160, 167n7 (see also Kripke, Saul) necessary truths and, 159–166, 173, 199, 200 A priori truth and a posteriori truth. See also A posteriori truth and the abstract vs. the concrete, 121–123, 125 distinction between, 115–127, 157, 159, 163–164, 169n8, 175, 199 God and, 169, 175

308Index

A priori truth and a posteriori truth (cont.) empirical vs. conceptual necessity and, 164–166 examples of, 115–120 positivism and, 157–160, 201n1 Apriority. See also A priori truth attribute of, 157–158, 160 necessity and, 173 Argument(s). See also specific topics antimaterialist, 3–9 meaning and, 109–112 types of, 150–151 Argument-theoretic semantics, 109 Aristotelian dualism, 45 Aristotle, 19, 20–21, 45 writings, 21, 215 Austin, J. L., 140 Autonomy, 259, 260, 262–266, 279 Behaviorism, 57, 59, 60, 84, 177 Being, the question of, 211–213 Belief formation, 45 Beliefs. See also under God and actions, 45–46 innate, 118, 124, 126–127, 167n7 Beneficence. See Strong principle of beneficence Bible, 137. See also Jesus Christ Black holes, 147n1 Blindsight, 49 Block, Ned, 4–5 Book of Nature, 137 Bowels, nervous system in. See Second mind/ second brain Breeding, selective, 255 Brentano, Franz, 4–5 Cartesian dualism, 26, 45, 53–55 Causal explanation, 199–200, 201n1 Causation, 187–190 Chalmers, David, 4–5 Child liberation, 277–280 Christianity, 303, 305

Cogito ergo sum, 171, 212 Cognitive–affective dualism, 19. See also Emotion: knowledge and Cognitive being, defined, 138 Cognitive gap, 191, 197 Cognitive science, 23 Cognitivism, 140. See also Noncognitivism Cognitivist view of consciousness, 19–20 Color ascriptions, dispositional theory of, 243 Color(s) evolution of, 241–244 ontology of, 243 Color experience, 4–5, 91, 275 Color memory, 4–8 Combination (logic), 221 Comedy, 290 Concept logic, 219 Concept–reality confusion, 20 Concepts. See also Private concepts recycling old concepts in new settings, 197 Conceptual analysis, 162, 193–195, 197 Conceptual augmentation, 195 Conceptual innovation, 197 Conceptual lack, 194, 195, 197 Conceptual necessity, 157, 159–160, 201n1 vs. empirical necessity, 159, 164–166 Conceptual sophistication, 196 Conclusions, 46 Concrete–abstract distinction. See Abstract– concrete distinction Consciousness, 52, 133, 144, 194. See also specific topics cognitivist view of, 19–20 and life having value, 269–270 and light, 49–52 memory and, 4–13, 15–17 nature of, 6–7, 133, 138 Consciousness–brain problem, 3 Consequence (logic), 221

Index309

Consequentialism. See also Utilitarianism problems with, 259–266 Constancy principle, 69–71 Context principle, 69–71, 73, 74 Contingent facts, 158 Contingent truths. See also Necessary– contingent distinction can always be explained, 199–200 as a posteriori, 157, 159, 199, 201n1 defined, 201n1 Kripke and, 158, 159 Correspondence (word meaning), 72, 73 Dark matter and dark energy, 147n1 Darwin, Charles, 245, 248–249, 256 On the Origin of Species, 246–247, 256 Darwinism, immaterial, 251–252 Dawkins, Richard, 235–236 “Arms Races and Manipulation,” 229–231 Deduction, 171 Deontological theories, 264 De re vs. de dicto, 165 essence as, 162–165, 168, 171, 173–175 knowledge as, 175 (see also Knowledge) logical laws as, 222 necessity as, 157–162, 173, 174, 220, 222 self-love as, 285–286 Descartes, René. See also Cartesian dualism cogito ergo sum, 171, 212 Darwinism and, 252 on nature, 53–54 on self-knowledge, 53–55 on thought, 4, 63 on unity, 53–54 Descriptions, theory of, 83–84, 165, 182, 193, 205 Devil, 301–302 Double-action theory of action, 41 Double nature problem, 37 Dreaming, 58n2, 269–270 Dualism, 35–36, 45. See also Universals: particulars and

Cartesian, 26, 45, 53–55 cognitive–affective, 19 inner–outer, 61 modal, 160 ratiocinative, 45, 48 Duty, moral, 259–265, 296–299 Education, 134 Efficacy, power of, 190–191 Eliminativism, 8, 9 Emotion, 20, 275–276, 286–287, 289–290. See also Hate knowledge and, 19–24 double aspect theory of, 20 Emotivism, 19, 20, 22–23, 102, 139, 275, 276 Empirical essences, 166–168. See also Essence(s) Empirical facts, 118, 120, 157, 168–169, 199, 267 Empirical necessity, 160, 164, 166 vs. conceptual necessity, 159, 164–166 Engrams, 16 Enteric nervous system, 25. See also Second mind/second brain Epistemic and the modal, the, 166 Epistemic contingency, 157, 170, 171 Epistemic desire, 24 Epistemic distinctions, 119–121. See also specific distinctions Epistemic emotions, 21 Epistemic faculties, 127, 164, 166–167 Epistemic inversion, 118, 123, 126 Epistemic modality, 157, 171, 172 Epistemic necessity, 157, 170–172. See also Necessity Epistemic notion of truth, 106 Epistemic optimists, 31, 55 Epistemic possibility, 170–172, 302 Epistemic status, 123, 124, 162, 165, 168, 169 is relative to epistemic faculties, 166–167 modal status and, 165, 166

310Index

Epistemic value, 168–169 Epistemology, 58n1. See also specific topics defined, 216 metaphysics and, 118, 120, 121, 159, 160, 162, 163, 166–169, 215–217 Kripke on, 158–160, 166, 168, 170 ontology and, 22–23, 60–61, 169n8, 170, 215–216 Error theory, antirealism as an, 177–179 Essence(s), 34, 157, 160, 161, 174, 175 causal explanation and, 200 as de re, 162–165, 168, 171, 173–175 dualism/duality of, 35–36 essentially empirical, 166–168 nominal vs. real, 166–168 nomological, 172–173 rigidity and, 163 Ethical emotivism, 19, 20, 139, 276. See also Emotivism Ethics. See also Deontological theories cognitivism in, 140 truth, knowledge, and, 140 Evil, God and, 295, 299–302 Evolution. See also Darwinism; Natural selection Darwin’s theory of, 245–247, 249 language of, 245–249 of sensations, 241 (see also Color[s]) Exclusion (logic), 221, 222 Existence, 211–213 nature of, 211–212 vs. nonexistence, 211–212 Experience, education, and learning, 134–135 Explanation, 199–201 External world, proof of an, 129 Extramental reality, 224 Fine, Kit, 174n12 Free will, 194 Frege, Gottlob, 223–224, 301, 302

Freud, Sigmund, 15, 15n1, 19, 143n2 Functionalism, 4, 6, 57, 59 Games, 162. See also Language-games definitions, meanings, and uses of the term, 75–78, 167 Gene selection theory, 253–255, 256n1 “Gene’s-eye view” of evolution, 255 Geometry, 135–136 Gershon, Michael, 25 Gleam theory, 153 God, 10, 189, 203, 205, 237 belief in, 237, 295 (see also Theism) beliefs implanted by, 55n1, 118, 123, 124 (see also Innate beliefs) consciousness and, 31, 52 a deontological theodicy, 295–300 evolution and, 55n1, 124, 251, 252, 255, 269 existence, nonexistence, and, 212 knowledge and, 118, 123, 124, 135, 137, 167 nature of, 212, 237, 295–302 a priori vs. a posteriori truth and, 118, 169, 175 suffering and, 295, 297–299, 302 Goodness, maximizing, 295–296. See also Consequentialism; Suffering: alleviating (vs. tolerating); Utilitarianism Grammatical types of sentences, 71–73 Grice, H. Paul, 65 intentions and, 80, 83, 84, 110 (see also Gricean intentions) meaning and, 80, 83, 84, 92, 204 Gricean intentions, 79–84, 92–96 Gut. See Second mind/second brain Hallucination, argument from, 150–151 Happiness. See Pleasure; Utilitarianism Hate, a religion of, 303–306 Hidden everything is hidden, 143–144 “nothing is hidden,” 143n1

Index311

Higher-order thought theories, 12 Historical-causal explanation, 199–200 Homogenized necessity, 160–164 Humanity. See Person Human rights. See Child liberation; Rights Hume, David, 192 causation and, 187–190 An Enquiry Concerning Human Understanding, 187–189, 192 A Treatise of Human Nature, 190–191 Identity. See also Self-love law of, 116, 220, 221 as a necessary relation, 5, 6 necessity of, 164–165 nonpersonal, 208 personal, 27, 207–210 Identity relation, 38, 165–166, 220 Identity statements, 36, 37, 40, 157n1, 171, 221, 302 mind–brain, 39 for particular events, 39 Identity theory, 6 mind–brain, 39, 40 Illusion, argument from, 150–151 Immaterial Darwinism, 251–252 Inherency, delusions of, 83 Innate beliefs, 118, 124, 126–127, 167n7. See also God: beliefs implanted by; Knowledge: from God Innate ideas, 4, 17, 27 Innate knowledge, 116, 118, 120, 123, 133, 137, 166, 169 Innate mental states, 13 Innate truths, 116, 119, 126–127. See also Innate knowledge Inner, reality of the, 57–61 Inner–outer dualism, 61 Inner process, 57 Instinctual knowledge. See Intuitive knowledge Instructional theory of knowledge, 135 Insufficient goodness, problem of, 295

Introspection, 57–59, 88–89 consciousness and, 7, 16 and the nature of matter, 146 and the unconscious, 7, 8 Introspective faculty, 29 Introspective knowledge, 126 vs. perceptual knowledge, 116, 126–127 Introspective truth defined, 116 vs. perceptual truth, 116, 126–127 Intuitive knowledge, 53–54 Intuitive truths, 115 Involuntary actions, 42 “I think, therefore I am.” See Cogito ergo sum Jackson, Frank, 4 Jealousy, 286–287 Jesus Christ, 297, 303, 305 Justification, 46–48, 101, 120, 124, 133, 139 Kant, Immanuel, 168, 182–184, 305 Knowledge, 116–127, 196, 271–272. See also A priori knowledge as de dicto, 175 definitions, 133, 134 desire for, 21, 24 emotion and, 19–24 from God, 118, 123, 124, 135, 137 implying truth, 139–141 necessity and, 157–174 the riddle of, 133–138 types of, 53–55, 115, 116, 133, 271 (see also A posteriori truth; A priori truth; Innate knowledge) Kripke, Saul, 172 computer case, 116–117, 122, 123 empirical essences and, 166, 168 on epistemology and metaphysics, 158–160, 166, 168, 170 on metaphysical necessity and a priori, 116–118, 157n1, 158–161, 167n7 modal argument, 4

312Index

Kripke, Saul (cont.) Naming and Necessity, 116–118, 122, 123, 157n1, 158–160, 167n7, 170–171 positivism and, 158–160, 201n1 on reference-fixing, 168 on rigidity and rigid designator, 163 Language. See also Private language on the impossibility of a (wholly) public, 91–97 meaning without, 79–85 and persuasion, 235–236 philosophy of, 81–84 thought and, 63–66 ways it can be employed by the mind, 63 Language-games, 69, 71, 74, 109 against, 75–78 meanings of the term, 75–78 Wittgenstein on, 69, 71, 74–77 Laughter, against, 289–291 Learning, 134–135 Lewis, David, 183 Life consciousness and life having value, 269–270 why it is good to be alive, 269–274 Light consciousness and, 49–52 and our knowledge of matter, 145–147 seeing the, 149–153 Light packet theory, 153 Light theory of consciousness, 49, 52 Linguistic acts, 42 Linguistic altruism, 236, 237 puzzle of, 236 Linguistics, 81, 84. See also Language; Meaning Listening, 134 Logic first- vs. second-order, 223 without propositions (or sentences), 219–225 Logical constants, 173 Logical laws, 219–222 types of, 220

Logical necessity, 173, 174, 220, 224 vs. physical necessity, 158 Logical relations, 221–222. See also Logical laws Logical representationalism, 225 Love hate and, 303–306, 303n1 romantic self-love, 285–287 Machines. See under Kripke, Saul; Wittgenstein, Ludwig Manipulation, 229–239 Massive informational simultaneous saturation (MISS), 49–52 Materialism central- vs. external-state, 34 identity and, 36 internal vs. external, 34, 35 memory and, 13 mind–body problem, innerness of the mind, and, 57, 59, 60 and the unconscious, 3–9 Materialistic identity theories, 34–36, 39, 139 Material-object subtraction, argument from, 150, 151 Mathematical facts, 120–121, 136, 169 vs. physical facts, 169 Mathematical information theory, 136n1, 137 Mathematical knowledge, 4, 76, 115, 118, 122, 133, 147, 147n1 a posteriori, 122 empirical evidence and, 117 Mathematical truths, 117, 120–122, 124, 135–136, 203 as a posteriori, 117 as a priori, 115, 120 Mathematics, 83 conceptual innovation in, 197 Kripke and, 117, 122 Matter, light and our knowledge of, 145–147 Meaning

Index313

deciding to mean, 99–103 without language, 79–85 logic and, 111 theories of, 81, 84, 101–103, 107 (see also Truth conditions theories) truth, verification, and, 105–108 types of, 80 Meaning monism, 69–74 defined, 69 Meaning pluralism, 69–71, 74. See also Meaning monism Memories, 15, 208 of private sensations, 87–89 Memory as active/dynamic vs. passive/static, 12–13 consciousness and, 4–13, 15–17 fallibility of, 80, 87, 88 nature of, 10–13, 15–16 perceptual experiences and, 4, 5, 9–10 (see also Color memory) unconscious, 4–13 Memory traces, 16 Metaphilosophy, 193 Metaphysical modality, 172 Metaphysical necessity, 157n1, 163, 170–172 Metaphysics, 174–175. See also specific topics epistemology and, 118, 120, 121, 159, 160, 162, 163, 166–169, 215–217 Kripke on, 158–160, 166, 168, 170 etymology of the term, 215 Metaphysics (Aristotle), 21, 215 Mind. See also Inner concepts of, 15–18 Mind–body problem(s) consciousness and, 3, 8–9, 17 innerness of the mind and, 60 and the paraconscious, 17 as a scientific problem, 193 for the second brain, 26 solving the, 3, 53–55, 60 theoretical and intuitive, 53, 54 Mind–brain identity theories, 33–40

Modal dualism, doctrine of, 160 Modal truth, 203–204 Modesty puzzle of, 281, 282 self-knowledge and, 281–283 Moral advertising, 235 Museum myth, 57 Mystery. See also Hidden; Unconscious analysis and, 193–197 Nagel, Thomas, 3–5 Names and naming, 157n1 Naming and Necessity (Kripke), 116–118, 122, 123, 157n1, 158–160, 167n7, 170–171 Natural selection. See also Evolution; Trait selection agential selection and, 247 vs. artificial selection, 245–249 vs. intentional selection, 249, 255 vs. selective breeding, 248 terminology, 247, 248 unit of, 253 (see also Gene selection theory) Natural vs. non-natural meaning, 80 Nature. See also Book of Nature Descartes on, 53–54 Necessary–contingent distinction, 199–201 analytic–synthetic distinction and, 199, 200, 201n1 Necessity concept of, 160 as de re vs. de dicto, 157–162, 173, 174, 220, 222 empirical vs. conceptual, 159, 160, 164–166 epistemic, 157, 170–172 explanation and, 199–201 homogenized, 160–164 knowledge and, 157–174 necessary truths can never be explained, 199 a posteriori truth and, 159, 163–165, 199

314Index

Nervous system in bowels (see Second mind/second brain) parasympathetic, 15n2 Newton, Isaac, 190 Nominal essences, 166–168 Nominalism, 177, 223 Nomological essences, 172–173 Nomological necessity, 172–173 Noncognitivism, 19 physical, 275–276 Noumenal world, 182, 183 Obligation, moral, 259–265, 296. See also Duty to relieve suffering, 297–299 Ontology. See also specific topics defined, 216 epistemology and, 22–23, 60–61, 169n8, 170, 215–216 “Ontology without Epistemology” (McGinn), 169n8 Pain, 11–12, 19. See also Suffering biological utility, natural selection, and, 241, 242 empathy and, 112, 116, 127 mind–brain identity theories and, 33–40 selfish genes and, 234, 235 Paraconscious, 15–18 Paradox(es), 295 concepts that lead to, 181, 295 the puzzle of, 181–185 semantic, 101, 141, 181, 295 Parasympathetic nervous system (PSNS), 15n2 Particulars. See under Universals Perception subliminal, 11, 12, 49, 51, 52 visual (see Color[s]; Light) Perception-memory-recall system, 11 Perceptual experiences and memory, 4, 5, 9–10. See also Color memory Perceptual vs. introspective knowledge/truth, 116, 126–127 Performing actions, 41–43

Person, concept of a, 207–210 Personal identity. See Identity: personal Persuasion, 235–236. See also Manipulation Phenomenal world, 182–184 Phenotype extended, 241–244 restricted, 241, 242 Philosophers, scientists as, 216n1 Philosophical mysteries, 194 Philosophy, 193. See also specific topics Photon realism, 153 Photons. See Light Physics, 144, 218n2, 243. See also Light entity vs. epistemic, 215 truths of mathematics contrasted with truths of, 115, 120 Physics (Aristotle), 215, 216 Plato, 135 on knowledge, 133 ontological dualism of universals and particulars, 121 Pleasure, 265, 267, 268, 271. See also Utilitarianism Positivism, 100, 101, 157–160, 174, 201n1 Possible worlds semantics, 101, 102, 163, 204 Possible worlds theory of modality, 203 against, 203–205 Postconscious, 11, 15 Practical reason/practical reasoning, 45–48 Pragmatics, 71, 74 Preconscious, 10–11, 15 Predicate logic, 219, 222 Predicates, 222 Priming, perceptual, 11 Privacy, 87–90 language of, 58 (see also Private language) of the mind, 58n1, 60, 65 (see also Inner) Privacy objects, 92–95 Private concepts, 94 Private fragments, 93, 94 Private intentions, 92 Private knowledge argument, 88–89 Private language, 87–96. See also Privacy

Index315

implicit, 94 Wittgenstein on, 87, 88, 90–93, 95–96 Private language argument, 87–90 Private memory, 88. See also Memories: of private sensations Private mental representation argument, 88–89 Private mental states, 92, 94, 95 Private sensations, 87–89 Private thought, 60, 63, 64 Private world, 58–59 Propositional logic, 219–221 Propositions, 219, 221. See also specific topics Proximal stimulus theory, 153 Putnam, Hilary, 4–5 Qualia, 5, 9, 26, 35, 60 Questions, 111 Quiet self-assurance, 283 Quine, Willard Van Orman, 57, 223 Ratiocinative dualism, 45, 48 Rationality vs. emotionality, 19 Rawls, John, 281–283 Realism, 99–102, 126, 178–179, 275 logical, 224 metaphysical, 141 moral, 178, 238 naive, 149, 153, 177 universals and, 224 Reason actions and, 45–48 vs. emotion, 19, 22–23 (see also Emotion: knowledge and) truths of, 115 Reasoning practical, 45–48 theoretical, 45, 48 Reference-fixing, 168 Reflection (physics), 151 Representationalism, logical, 225 Repressed memories, 10. See also Memory: unconscious

Rights, individual, 209. See also Child liberation Rigid designator, 163 Romantic self-love, 285–287 Russell, Bertrand, 54 on knowledge of time, 30 on matter and our knowledge of it, 146 theory of descriptions, 83–84, 165, 182, 193, 205 Ryle, Gilbert, 57 Sadism, 290 Science. See also Physics limitations, 213 as metaphysics, 215–218 nature of, 216 as ontology, 216 Scientists, 217 as philosophers, 216n1 Second mind/second brain, 25–28 definition and nature of, 25 Self-assurance, quiet, 283 Selfish genes and moral parasites, 229–239 Self-knowledge Descartes on, 53–55 modesty and, 281–283 Self-love, romantic, 285–287 Semantic decisions, 101, 102 Semantic emotivists, 102. See also Emotivism Semantic paradoxes, 101, 141, 181, 295 Semantic roles of words, 101 Semantics, 72–74, 99–101. See also Sentences: types of; specific topics ontology and, 74 possible worlds, 101, 102, 163, 204 vs. pragmatics, 71, 74 theories of meaning and, 101–103 types of, 71–72, 100–102, 109 (see also Possible worlds semantics) Semantic theory, dual, 106–107 Semantic tone, 100 Sensations. See also Private sensations evolution of, 241

316Index

Sense datum theory, 153 Sentences. See also Argument(s) function of, 109 meaning of, 109 types of, 69–72, 120 Sentimentalism, 19, 22–23 Sexual selection (Darwin), 247n1 Simulation Game (SG), 131–132 Single essence problem, 35 Single-nature condition, 38, 40 Smiling vs. laughter, 289–290 Speech, 63–65 function of, 235–236 Speech acts, 42 Strong principle of beneficence (SPB), 267 Subliminal perception, 11, 12, 49, 51, 52 Subtraction, argument from, 150, 151 Suffering. See also Pain alleviating (vs. tolerating), 297–300 God and, 295, 297–299, 302 Suits, Bernard, 75, 162n5, 167 Survival of the fittest, 248. See also Natural selection Symbols, 135 Synesthesia, 275 Synthetic inexplicable, notion of, 201n1 Synthetic truth and synthetic knowledge. See Analytic–synthetic distinction Temporal awareness faculty, 29. See also Time Testimony knowledge, 123 Theism, 138n2, 212 Theoretical reason/theoretical reasoning, 45, 48 Thinking aloud, 63–65 Thought. See also Private thought Descartes on, 4, 63 higher-order thought theories, 12 nature of, 63–66 Thought experiments, 9–10, 79–80, 84, 207– 208, 251, 270. See also Zombie argument Time, awareness of, 29–31 Token identity statements, 40

Token identity theory, 36–40 Trait pool, 256n1 Traits, 253 Trait selection, 253–256 Trait selection theory, 253–255 “Trait’s-eye view” of evolution, 254, 255. See also Trait selection Truth knowledge implying, 139–141 logic as the laws of, 223 notions/conceptions of, 140–141 noumenal world and, 183 and quasi-truth, 141 types and classes of, 115, 116, 119–120 (see also A posteriori truth; A priori truth; Innate truths) Truth-based semantics, 101, 102 Truth conditions theories, 70, 99–101, 105–108. See also Meaning monism Truth-theoretic semantics, 71–72 Twin Earth thought experiments, 84 Type identity theory, 33–36, 38–39 Unconscious, 143, 269–270. See also Zombies conceiving the, 7–9 conscious and, 9–13, 15 mystery of the, 3, 7–13 antimaterialist arguments, 3–9 Unconscious memory, 4–13 Unconscious sensory states, 49 Universals, 220–221, 223 particulars and, 122, 223 types of facts about, 220–221 Universe, 272–273 Use-mention confusion, 20 Use theory, 81, 101 Utilitarianism, 261, 295–300. See also Consequentialism absurd, 267–268 Utility quotient (UQ), 267 Verification conditions theory of meaning, 101, 105–107

Index317

Verificationism, 88, 89, 91, 92, 99, 100, 107 Vision. See also Light subliminal, 49, 51, 52 (see also Subliminal perception) Wallace, Alfred Russel, 247 Watson, John B., 57 Wittgenstein, Ludwig, 31, 87 and the hidden, 143 and the inner, 57, 61n3 on language-games, 69, 71, 74–77 on machines and their possible movements, 191–192 meaning monism and, 69 private language and, 87, 88, 90–93, 95–96 on sensations, 87 Word constancy, 70–74 Zombie argument, Chalmers’s, 4, 5