Knowledge, Thought, and the Case for Dualism 9781107037878

Citation preview

KNOWLEDGE, THOUGHT, AND THE CASE FOR DUALISM The relationship between mind and matter, mental states and physical states, has occupied the attention of philosophers for thousands of years. Richard Fumerton’s primary concern is the knowledge argument for dualism – an argument that proceeds from the idea that we can know truths about our existence and our mental states without knowing any truths about the physical world. This view has come under relentless criticism, but here Fumerton makes a powerful case for its rehabilitation, demonstrating clearly the importance of its interconnections with a wide range of other controversies within philosophy. Fumerton analyzes philosophical views about the nature of thought and the relation of those views to arguments for dualism, and investigates the connection between a traditional form of foundationalism about knowledge, and a foundationalist view about thought that underlies traditional arguments for dualism. His book will be of great interest to those studying epistemology and the philosophy of mind.

richard fumerton is the F. Wendell Miller Professor of Philosophy at the University of Iowa. His most recent publications include Realism and the Correspondence Theory of Truth (2002), Epistemology (2006), and The Philosophy of John Stuart Mill (with Wendy Donner, 2009).

CAMBRIDGE STUDIES IN PHILOSOPHY General Editors j o n a t h a n l o w e (University of Durham) n o a h l e m o s (College of William and Mary) Advisory Editors j o n a t h a n d a n c y (University of Reading) j o h n h a l d a n e (University of St. Andrews) g i l b e r t h a r m a n (Princeton University) f r a n k j a c k s o n (Australian National University) w i l l i a m g . l y c a n (University of North Carolina, Chapel Hill) s y d n e y s h o e m a k e r (Cornell University) j u d i t h j . t h o m s o n (Massachusetts Institute of Technology) Recent Titles d a v i d l e w i s Papers on Ethics and Social Philosophy f r e d d r et s k e Perception, Knowledge, and Belief l y n n e r u d de r b a k e r Persons and Bodies r o s a n n a k e e f e Theories of Vagueness john greco Putting Skeptics in Their Place r u th g a r r e t t m i l l i k a n On Clear and Confused Ideas derk perebo om Living Without Free Will b r i a n e l l i s Scientific Essentialism a l a n h . g o l d m a n Practical Rules c h r i s t o p h e r hi l l Thought and World a n dre w n e wm an The Correspondence Theory of Truth i s h t i y a q u e h a j i Deontic Morality and Control wayne a . davis Meaning, Expression and Thought p et e r r a i l t o n Facts, Values, and Norms j a n e h e a l Mind, Reason and Imagination jonathan kv anvig The Value of Knowledge and the Pursuit of Understanding andrew melnyk A Physicalist Manifesto w i l l i a m s . r o b i n s o n Understanding Phenomenal Consciousness d. m. armstrong Truth and Truthmakers keith f rankish Mind and Supermind m i c h a e l sm i t h Ethics and the A Priori n o a h l e m o s Common Sense

j o s h u a g er t Brute Rationality a l e x a n d e r r. p r u s s The Principle of Sufficient Reason fo l k e t e r s ma n Moral Disagreement j o s e p h m e n d o l a Goodness and Justice d a v i d co p p Morality in a Natural World lynne rudd er baker The Metaphysics of Everyday Life sa n fo rd g ol d be rg Anti-Individualism m i c h a e l j. zi m m e r m a n n Living with Uncertainty c r a w f o r d l . e l d e r Familiar Objects and their Shadows j a m i n as a y The Primitivist Theory of Truth

Knowledge, Thought, and the Case for Dualism Richard Fumerton University of Iowa

University Printing House, Cambridge CB2 8BS, United Kingdom Published in the United States of America by Cambridge University Press, New York Cambridge University Press is part of the University of Cambridge. It furthers the University’s mission by disseminating knowledge in the pursuit of education, learning, and research at the highest international levels of excellence. www.cambridge.org Information on this title: www.cambridge.org/9781107037878 © Richard Fumerton 2013 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2013 Printed and bound in the United Kingdom by CPI Group Ltd. Croydon CR0 4YY A catalogue record for this publication is available from the British Library Library of Congress Cataloguing in Publication data Fumerton, Richard A., 1949– Knowledge, thought, and the case for dualism / Richard Fumerton. pages cm. – (Cambridge studies in philosophy) ISBN 978-1-107-03787-8 (hardback) 1. Dualism. 2. Materialism. 3. Mind and body. I. Title. BD331.F87 2013 1470 .4–dc23 2013004140 ISBN 978-1-107-03787-8 Hardback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party internet websites referred to in this publication, and does not guarantee that any content on such websites is, or will remain, accurate or appropriate.

For Alex and Charlie

Contents Preface

1

2

3

page xiii

Setting the stage 1.1 Leibniz’s law and the Cartesian argument for dualism 1.2 Hard vs. easy problems of consciousness 1.3 Summary

1 6 21 27

Distinctions: versions of physicalism and dualism 2.1 Substance physicalism/dualism 2.2 Property physicalism/dualism 2.3 Fact physicalism/dualism 2.4 Event physicalism/dualism 2.5 Proposition physicalism/dualism 2.6 Sentence physicalism/dualism 2.7 Defining “mental” and “physical” 2.8 Physical sciences and physical properties 2.9 Epistemological characterizations of the mental/physical 2.10 Intentionality as the mark of the mental 2.11 Giving up 2.12 Varieties of physicalism 2.13 Logical behaviorism 2.14 The mind/brain identity theory 2.15 Functionalism

29 29 36 39 41 42 44 45 46

Ontological priorities: taking phenomenology seriously 3.1 Foundationalism 3.2 Regress arguments for foundationalism 3.3 Epistemic regress 3.4 The conceptual regress argument 3.5 Acquaintance

91 92 93 94 96 98

ix

49 52 58 60 64 72 76

Contents 3.6 3.7 3.8 3.9

Inferential justification and the physical world Internalism and the threat of skepticism Rejecting methodological naturalism Is there any contribution science can make to the philosophy of mind? 3.10 Summary

119 130 132

4

Knowledge arguments revisited 4.1 The ability hypothesis 4.2 Getting acquainted with color experience 4.3 Mary’s new beliefs 4.4 Different propositions; same truth-maker 4.5 Summary

150 157 161 165 169 181

5

Indirect thought and informative identity 5.1 The direct reference theory 5.2 Two traditional accounts of foundational thought 5.3 Objections to sense theories 5.4 Anti-descriptivist alternatives 5.5 Did Mary always have beliefs about phenomenal color? 5.6 Informative identity

182 183

6

7

8

An ontologically liberating skepticism: the last hope for physicalism 6.1 Indirect understanding of the physical 6.2 An ontologically liberating skepticism 6.3 Panpsychism 6.4 Is this a version of physicalism? 6.5 Summary

143 149

189 192 195 201 206 208 210 218 225 227 231

Objections and replies 7.1 Jackson’s rejection of the knowledge argument 7.2 Foundationalism of knowledge and thought 7.3 The causal theory of objects 7.4 Causal overdetermination

233

The ubiquitous self: a brief postscript 8.1 Summary 8.2 Substance dualism in the background?

257 257 258

x

233 239 245 246

Contents 8.3 8.4 8.5 8.6

The self and the body Humean rejection of the self Reduce where you can: commit where you can’t Diachronic identity

References Index

258 259 264 267 271 279

xi

Preface The task of trying to say something interesting and original about the mind/body issue is daunting to say the least. Furthermore, the issues have become more and more complex as it becomes painfully clear that to settle some of the most fundamental issues in the philosophy of mind, one must reach conclusions on a host of other difficult and fundamental controversies that arise in different, though interrelated, areas of philosophy. These areas include, but are not limited to, epistemology, general metaphysics (including the metaphysics of property exemplification), philosophy of language (particularly those issues concerning intentionality that spill over into the philosophy of mind), and philosophy of science. This relatively short book cannot possibly hope to settle all of the controversies that are relevant to an assessment of various forms of dualism. In fact, I’m convinced that it is only a lifetime of work in all of the relevant areas of philosophy that could be construed as a convincing case for the positions that will be discussed in the book. In what follows I’m interested primarily in the plausibility of so-called knowledge arguments for property dualism. But even here I realize that to fully defend the arguments I offer, I would need to more fully defend some of the epistemological premises I bring to the table. In a number of places I engage in the admittedly annoying practice of referring my reader to other works in which I have tried to provide more complete defenses of premises on which I rely. If the reader hasn’t read xiii

Preface

those other works and has no particular inclination to do so, I still hope that the following discussion will be interesting if thought of as the defense of a very complex conditional. Indeed, I’m most interested in figuring out what views in the philosophy of mind one should adopt if one accepts the rather controversial epistemological positions I have been defending for several decades. In short, I am interested in exploring the implications of a radical empiricist and internalist foundationalism, and an equally radical content internalism, for the philosophy of mind. One might initially suppose that the conditional won’t be that interesting if the antecedent is loaded up in the way described above. But such a conclusion is premature. Indeed, as Russell, Maxwell, Chalmers, and Stoljar have argued, it may be that views most closely associated with traditional dualism offer at least conceptual room for the physicalist. And this is one of the most important issues to be addressed in what follows. There is no getting around the fact that the issues involved are complex. But I’m not sure they need to be quite as complicated as some of the philosophical writing on this topic makes them. I try hard to make the material in this book as accessible as I can. I want the book to be interesting (and no doubt controversial) to the professional philosopher, but readable and thought-provoking for the interested undergraduate. A price for taking this approach is that I do not discuss extensively the voluminous literature on all of the topics I address. There are so many people I should thank for the many philosophical discussions that have influenced my thought on these matters. They include countless colleagues and students – too many to thank individually. But I should probably begin by expressing my appreciation to the late Grover Maxwell for including me in a discussion group when I visited the xiv

Preface

University of Minnesota some thirty years ago. Maxwell was the paradigm of a philosopher who was willing to go wherever argument led him. While he is still perhaps most famous for his (1962) influential paper attacking the idea that there is a clear observation/theoretical distinction, he noted in a footnote (ft. 5) to that paper that there is an extreme form of foundationalism that might put one in a better position to argue for a hard and fast distinction between foundational (observational) knowledge and inferred (theoretical) knowledge. To his great credit he continued to think about that idea and, inspired by Russell and Kripke, started to explore (1978) the very view that he set aside in his earlier work as too extreme. I learned how to do philosophy at Brown University, and I will always be grateful to Chisholm, Sosa, and Van Cleve for the education I received and the examples they set for their students. My views might be wrong, but I always hope that the conclusions I reach and the arguments I give are relatively clear. At Iowa I have learned a great deal and enjoyed discussion with all of my colleagues, but in topics related to the intersections of epistemology with philosophy of mind I should particularly thank Gustav Bergmann, Laird Addis, Panayot Butchvarov, Philip Cummins, Diane Jeske, Gregory Landini, and Evan Fales. My former student John DePoe looked carefully at an early draft of the manuscript and his questions and comments were extremely helpful. Seth Jones helped me enormously working with me on the index and proofreading the manuscript. I also want to thank anonymous referees who gave me all sorts of invaluable advice and criticism. The writing of this manuscript was supported, in part, by a University of Iowa Developmental Leave. I presented some of the material in Chapter 6 at Illinois State in 2011 and at a conference organized by Bruce Russell, March 2011. I thank the participants at both presentations for their probing questions and objections. xv

1 Setting the stage

The relationship between mind and matter, mental states and physical states, has occupied the attention and imagination of the intellectually curious for thousands of years. In most cultures many people are officially committed to religious views that allow for the possibility of our surviving the total annihilation of our bodies. While the answers to questions of diachronic identity (identity through time) are not straightforwardly dependent on questions concerning synchronic identity (the identity of things at a time), it is tempting to think that there might be an intimate connection between the two. In particular, it is tempting to think that if we want to make plans for life after the destruction of the body, we better not tie the existence of our selves at a time too closely to the existence of our bodies at a time. This book is yet another attempt to shed light on the nature of mental states. Despite the fact that volumes have been written on the subject, I believe that there are still contributions to be made. The issues are truly complicated. The primary reason that philosophical debate over the nature of the mental rages on is that, like so many other debates in philosophy, resolution of the issues involves a host of fundamental philosophical controversies on a wide range of philosophical topics. One can’t coherently address problems in the philosophy of mind without working through issues in epistemology, philosophy of language, and broader metaphysical issues concerning the existence and nature of truth, states of affairs, facts, events, 1

Knowledge, Thought, and the Case for Dualism

properties, substances, and identity, both at a time and through time. Lurking in the background are critical metaphilosophical questions concerning the nature and appropriate methods of philosophical inquiry. Consider, for example, the epistemological overtones of historical and contemporary debate concerning the nature of mental states. Certainly from Descartes on, philosophers have been trying to draw conclusions about the nature of the self and its properties by contrasting the ways in which one can know truths about oneself and the ways in which one can know truths about one’s body. Descartes famously thought that the way to identify a secure starting point for knowledge was to include in its foundations only that which cannot be doubted. This methodological suggestion itself can be interpreted in importantly different ways. Descartes could have been talking about the psychological possibility of wondering whether or not a given assertion is true. More plausibly, however, he was concerned with what our justification allows. On this interpretation, S’s belief that P allows for the possibility of doubt when the epistemic justification S possesses for believing P is at least consistent with P’s being false. In Descartes’s search for infallibly justified belief, he hit upon our belief that we exist. The justification we have for believing that we exist is not consistent with that proposition’s being false. The proposition that I exist, therefore, belongs in my foundations. While I have foundational knowledge of my own existence, I have no foundational knowledge of the existence of anything physical. The reconstruction of Descartes’s view is a bit tricky since he eventually retracts many of the arguments he gives (including his famous dream argument). But initially, Descartes appeared to argue as follows. The justification we have for believing any propositions asserting the existence and describing the properties of physical objects is consistent with those propositions being false. Consider, for example, dreams. 2

Setting the stage

In a vivid dream I could have precisely the same justification for thinking that there is a table before me as I would have had were I veridically perceiving a table. But the justification I have in a dream is obviously consistent with the table’s not being there, and so the justification I have in veridical perception leaves open the skeptical possibility as well. Or consider Descartes’s evil demon possibility – the inspiration for many a plot of films and TV series.1 The qualitatively same sensations that come to us in veridical perception, Descartes argued, instead could have been “planted” in us directly by a very powerful being bent on deceiving us with respect to our surroundings. If deceived in this way, we would have the same justification for believing what we do about our surroundings as we would have had were our experiences veridical. The justification we have, therefore, does not preclude the possibility of error.2 The argument does presuppose a version of the view that we might call epistemological internalism.3 On that view, the 1

2

3

The Matrix, Total Recall, What Dreams May Come, and seemingly every other episode of the Star Trek series (old and new). In a popular piece discussing Matrix-like skeptical scenarios, Chalmers (2003) argues that if we had always been living “in a matrix” we shouldn’t take our beliefs about the external world to be false. Rather we would have been (successfully) talking about a reality different in nature from what we took it to be. The argument rests on what I have called a causal theory of objects – we’ll be discussing a view like this later in the book. There are actually many different versions of internalism now. The one I sketch here might usefully be called internal state internalism after its key idea that the epistemic status of a belief supervenes solely on the internal states of the believer. Even here, however, we need a much more careful account of what makes a state internal. Most self-proclaimed internalists of this sort will still want to allow that a given state is internal if it involves a relation between the subject and some abstract entity like a universal. And the kind of epistemologist who thinks that one can be directly acquainted with the surfaces of physical objects might also want to count that relational state of affairs as internal. Internal state internalism must be distinguished from various versions of access internalisms. The access internalist insists on the idea that for something X to justify S in believing P, S must have actual or potential access to X (a relatively weak requirement) or actual or potential access to the fact that X is a justifier for the belief that P (a much stronger requirement). Access requirements on

3

Knowledge, Thought, and the Case for Dualism

justification S possesses for believing a given proposition P at t is solely dependent on S’s internal states at t. The internal state internalist’s slogan is that necessarily, if S and R are in the same internal states, then whatever justification there is for S to believe some proposition P there is for R to believe P. If one embraces the internalist’s doctrine and searches for the justification we have for believing sundry propositions describing our physical world, it will seem obvious that that justification is consistent with our physical object beliefs being false. The conclusion is so difficult to resist because of what we take to be the causal story surrounding experience. Ultimately, we must rely on sense experience to reach any conclusion about the existence of physical objects and their properties. But the immediate causes of sense experiences, we believe, are events occurring in the brain. We believe that there are physical objects, and that when we are before those objects under certain conditions, a long and complex causal chain begins, starting with changes that occur in and around the surface of the object and terminating in a change in the brain which either is (according to the mind/brain identity theorist) or immediately causes (according to the dualist) sensation. In the case of visual experience, we think that light reflects off the surface of an object eventually hitting the retina of the eye and effecting changes there. Of course, changes in the eye aren’t going to result immediately in visual experience. The “information” must be carried all of the way to the relevant region of the brain before the sensation occurs. Given that sensation is the immediate effect of the brain state, the natural thought, then, is that we could, in principle, “break into” the causal chain anywhere and if the causal sequence terminates in the same place, justification are tricky – they almost always lead one to flirt with vicious regress. See Fumerton (1995, Chapter 3 and 4) for attempts to define more carefully different versions of internalism. And see Bergmann (2006, Part I) for a sustained attack on all forms of epistemic internalism.

4

Setting the stage

the same sensation will occur. More often than not, contemporary epistemologists discussing skeptical scenarios replace evil demons with mad neurologists who have stolen brains and who are manipulating those brains so as to produce massive vivid hallucination. Again, the upshot of all this for Descartes was the critical observation that we can know the truth that we exist without knowing any truths about the physical world. One might worry that the arguments for this thesis seemed to presuppose that we have knowledge that allows us to describe the character of both veridical and non-veridical experience. So in his more incautious moments you might find Descartes making a claim about the similarity between the experiences of a dream and the experiences of waking life. And the alert philosopher might pounce on that claim. To know that the experiences of dreams are like the experiences of waking life we would need to know what the experiences of waking life are like. But that would presuppose that we are able to know when we are awake. Similarly, any skeptical argument that proceeds from premises describing the causal history of perception will explicitly invoke information about the physical world, when it is the possibility of getting such information that is coming under skeptical attack. But it would be a mistake to suppose that there is anything problematic in the appeals to possibility that the Cartesian will invoke in attempts to cast doubt on the strength of the justification we have for believing propositions about the physical world. Descartes argues from the possibility of dreaming and hallucinating. One can certainly take advantage of the fact that one’s audience has beliefs that certain possibilities are actual in reaching the conclusion that they are possible – the surest proof that something is possible is that that it is actual. But the conclusion that vivid dreams and hallucination are possible would be true whether or not they ever occurred. 5

Knowledge, Thought, and the Case for Dualism

If one grants Descartes’s claim that one can know that one exists in a way quite different from the way in which one can know any truths describing the physical world, the question then becomes whether or not one can draw any metaphysical conclusions about the nature of the self and its relation to the physical from these epistemological premises. Descartes appeared to conclude that since it is epistemically possible that he exists without his body, then he must be something distinct from his body. But critics, even in Descartes own time, immediately questioned the inference.

1.1

leibniz’s law and the cartesian argument for dualism

If there is any principle that is, or should be, unproblematic concerning identity it is one half of Leibniz’s law, the indiscernibility of identicals. In its metaphysical form the principle states that for any x and any y, if x is identical with y at t then there is only one thing picked out by “x” and “y” and whatever is true of that thing at t is true of that thing at t.4 Leibniz went on to claim much more controversially the converse: that for any x and any y if whatever is true of x at t is true of y at t and vice versa, then x is identical with y at t. In what follows I’m concerned only with the indiscernibility of identicals. And as I said the principle should be uncontroversial. If x and y are 4

The principle gets much more controversial if we don’t restrict it in this way to identity at a time. One might suppose that if one argues that x at t1 is identical with y at t2 only if there is nothing true of x at t1 that isn’t true of y at t2, one will have, implausibly, or at least controversially, eliminated the possibility of a single thing changing over time. One can attempt to avoid that conclusion by insisting that property ascription always be itself time relativized. So I can be identical with the boy who was only 4 feet tall in 1955, because I still have the property of being 4 feet tall in 1955. Temporally relativized properties can’t be lost. If one insists that it is ad hoc to insist on time-relativized property ascriptions, one can simply assert that it is only a time-relativized version of Leibniz’s law that is true.

6

Setting the stage

identical at t then we have only one thing x/y that has properties and stands in relations to other things at that time. The principle amounts to the claim that if a thing has a property it has the property! Indeed, the only reason that anyone might wonder about the principle is if it gets confused with another superficially similar linguistic principle concerning the substitutivity of co-referential expressions. One might find initially attractive the suggestion that if “x” and “y” pick out precisely the same entity, then we ought to be able to substitute “y” for “x” in any assertion without changing the truth value of the assertion. And we might suppose that this is true because if “x” and “y” are co-referential then the substitution won’t make a difference to what is asserted. So, for example, if “my cat” and “Duke” pick out the same cat and it is true that Duke is gray then it is true that my cat is gray. But the principle that one can substitute co-referential expressions salva veritate in all contexts seems to be false. Consider all of the following: (1) (2) (3)

“Duke” begins with the letter “D.” Sue believes that Duke is gray. It is a necessary truth that if Duke exists and is gray then Duke is gray.

It we substitute for “Duke” “my cat” in (1) through (3) (its first occurrence in (3)), we will change the truth value of (1) and (3) and might change it in (2) – we need only imagine a situation in which Sue doesn’t believe that Duke is my cat and further believes that my cat is orange. Now there are all sorts of complications that arise as we think more carefully through the above crude discussion. It is not clear that all, or even any, of (1) through (3) involve genuine instances of failure of referential substitutivity. A great deal depends on how we understand the technical philosophical concept of reference and how we locate the occurrence of a referring expression. So although it might sound initially strange, some philosophers 7

Knowledge, Thought, and the Case for Dualism

(e.g. Quine, 1966, Chapter 15), will deny that the expression “Duke” occurs in (1). Admittedly, the expression “‘Duke’” occurs in (1) but neither “my cat” nor “‘my cat’” refers to the same thing as does “‘Duke’.” It makes no more sense to try to substitute “my cat” for “Duke” in (1) than it would make sense to substitute “automobile” for “car” in “My carpet is dirty.” All we would get is the gibberish, “My automobilepet is dirty.” Whether one takes (2) and (3) to be clear cases exhibiting referential opacity (contexts in which substitution of referring expressions can change truth value) might depend on the verbal question of what one takes “referring” to be. So as we will discuss later, Russell and his followers will almost always take ordinary names to be disguised definite descriptions. “Duke” might mean something like “the cat I bought when I was 15.” “The cat I own” is already an undisguised definite description. At least some of the time Russellians will balk at claiming that definite descriptions refer – indeed they will claim only that the definite description makes a contribution to the meaning of the entire sentence in which it occurs, a sentence which on analysis won’t contain anything that looks very much like the definite description. So “The F is G” becomes “There is one and only one thing F which is G.” Still, the definite description in English clearly picks out some object when it is used successfully, and we need an expression to capture the relation between a definite description and the object it singles out. We can say that a definite description can successfully denote some object a and that it does so just when that object takes the value of the variable that appears in the logically perspicuous “analysis” of the statement. A hard-core Russellian might reserve the term “reference” for the relation that holds directly between a pure name and an object for which it stands (a name whose meaning really is its referent). No one thinks (or should think) that the meaning of a definite description is its denotation. We’ll have much more to say about all this in Chapter 5. 8

Setting the stage

To avoid these terminological matters, we can distinguish the metaphysical principle of the indiscernibility of identicals from controversial principles concerning the substitutivity salva veritate of either co-referential or co-denotational expressions. Again, there are no counterexamples to the principle of the indiscernibility of identicals. Apparent counterexamples are always counterexamples to various principles concerning substitutivity of linguistic terms. So, to illustrate with a much-used example, one might suppose that one can construct counterexamples to Leibniz’s law by thinking about Superman, Clark Kent, and Lois Lane. Superman has the property of being believed by Lois Lane to be really strong, while Clark Kent lacks that property. But if we suppose that there is a property of being believed, what has it? Lois Lane does believe the proposition that Superman is really strong and it is not true that she believes the proposition that Clark Kent is really strong. But all that shows is that there is a property of the one proposition that is not a property of the other (as we’ll see in a moment, some of this gets decidedly more complicated). We haven’t discovered a property of Superman that isn’t a property of Clark Kent. Of course, there is a so-called de re reading of descriptions of belief. They are sometimes marked in ordinary language (or at least ordinary philosophical language) as follows: Lois Lane believes of Superman that he is really strong, but doesn’t believe of Clark Kent that he is really strong. The classic indicator that a description of belief is a de re description is that the relevant description is referentially transparent – one can substitute co-referential (or co-denotational) descriptions without altering the truth of the statement. The following is a statement that in most contexts wouldn’t strike you as all that odd but could only be a plausible claim if it gets a de re reading: Henry Hudson believed that Hudson’s Bay was a passage to the Orient. Upon finding the huge body of water we call 9

Knowledge, Thought, and the Case for Dualism

Hudson’s Bay, Hudson did conclude that he was in the much sought after Northwest Passage. To his great disappointment (followed relatively soon by his death) he was wrong. But however mistaken Hudson might have been, he presumably wasn’t both clairvoyant and crazy. He didn’t believe that the body of water he was in was a bay, and he didn’t believe that it was going to be named after him. Many philosophers think that they can handle the distinction between de dicto and de re reports of belief using scope distinction, at least when the relevant denoting expressions that occur in the belief report are definite descriptions. So when I say that S believes that the F is G, the de dicto reading is: (1) S believes that the one thing that is F is G.

The de re reading is: (2) The one thing that is F is such that S believes that it is G.

I don’t think that this is the way to handle the distinction for the simple reason that (2) really doesn’t identify the content of S’s belief. It is more plausible to suppose that the de re report leaves open the proposition S believes other than to stipulate that it is a proposition whose subject content denotes a thing that is in fact F and whose predicate is G. The rough translation then becomes something like the following: (3) There is just one thing that is F and there is some proposition P such that S believes that P and the subject concept of P denotes a thing that is the one thing that is F and the predicate concept of P picks out G.

Notice that to acknowledge the obvious truth that there are de re/de dicto ambiguities in descriptions of belief does not have any obvious implications for the ontological question of whether there are two quite different sorts of beliefs – beliefs that have as their content propositions and beliefs that 10

Setting the stage

somehow incorporate as constituents the very items the belief is about. More about this later. Back to Descartes. When the smoke clears and the dust settles, what has Descartes shown, if his premises are true? He has concluded that we can know that we exist in a way different from the way in which we can know any proposition asserting the existence of anything physical. Let’s give him that conclusion. So the following are both true: (1) (2)

I know that I exist. It is not the case that I know (in that way) that my body or any of its parts exist.

But how do we get from (1) and (2) to: (3)

My existence is not identical with the existence of my body or any of its parts.

One can imagine that (3) is supposed to follow from (1) and (2) by an application of Leibniz’s law. But here lies the problem. As J. J. C. Smart (1959) argued (long before the great Mary debates), all you get from (1), (2), and Leibniz’s law is that the proposition that I exist is different from the proposition that my body exists. And Smart was willing to concede this while insisting that the two propositions nevertheless describe one and the same reality. As we saw earlier, Smart’s conclusion follows as long as one supposes, plausibly, that being known (in a certain way) or being not known (in that way) is a property of a proposition, where here I am using “proposition” merely as a place holder for whatever you take to be the primary bearer of truth value. Descartes’s famous argument for a dualism between self and body has far less dramatic counterparts with much more obvious premises, arguments that would also appear to get you specific truths about the distinctness of propositions describing mind and propositions describing body, and might 11

Knowledge, Thought, and the Case for Dualism

initially seem to be the same grist for a dualist mill. So let us say a relatively crude type-identity theorist is toying with the idea of identifying my severe pain with the proverbial c-fibers firing (the philosopher’s stand-in for whatever neural activity is the most plausible candidate for the neural counterpart of an occurrent pain). If anything seems uncontroversial, it is that there are all sorts of people who (1) believe that they are in severe pain when they are, but (2) do not believe that their cfiber are firing when they are. Indeed, the vast majority of people on the face of the Earth don’t even know what a c-fiber firing is supposed to be. But again, what kind of thing has the property of being believed or not believed? The most obvious candidate is the bearer of truth value – a representation of the world. So we do get the following: The proposition/assertion/ thought that I am in pain is not identical with the proposition/ assertion/thought that c-fibers are firing in my brain. But, says Smart, that’s not going to get you the conclusion that the severe pain isn’t identical with c-fibers firing. At least it won’t get you that conclusion until you establish that two distinct propositions can’t end up describing and being made true by one and the same fact. Smart (1959) makes an exactly similar sort of move in response to the dualist who argues for dualism by pointing to differences in the modal properties of propositions describing mind versus propositions describing body. It obviously isn’t possible for me to be in pain when I am not in pain, but in some sense it surely is at least conceivable that I could be in pain even though c-fibers aren’t firing in my brain. Now these days, as we will see later, there are plenty of philosophers who won’t make Smart’s move. But Smart was willing to admit that the claim that I am in pain has a different meaning from the claim that c-fibers are firing. Logical relations between assertions are a function of the meanings of those assertions, but statements with different meanings can describe precisely the same event. 12

Setting the stage

Smart puts the dualist’s objection this way (using as his example the dualist who wants to deny that having an afterimage is being in a certain brain state): It is only a contingent fact (if it is a fact) that when we have a certain kind of sensation there is a certain kind of process in our brain. Indeed it is possible, though perhaps in the highest degree unlikely, that our present physiological theories will be as out of date as the ancient theory connecting mental processes with goings on in the heart. (p. 147)

And then he gives his reply: The objection certainly proves that when we say “I have an afterimage” we cannot mean something of the form “I have such and such a brain-process.” But this does not show that what we report (having an after-image) is not in fact a brain process. “I see lightning” does not mean “I see an electric discharge.” Indeed, it is logically possible (though highly unlikely) that the electrical discharge account of lightning might one day be given up. Again, “I see the Evening Star” does not mean the same as “I see the Morning Star,” and yet “the Evening Star and the Morning Star are one and the same thing” is a contingent proposition . . . If the meaning of an expression were what the expression named, then of course it would follow from the fact that “sensation” and “brain-process” have different meanings that they cannot name one and the same thing. (p. 147)

Frank Jackson’s famous Mary argument breathed new life into an old debate. Jackson’s argument was hardly original. As Ludlow, Nagasawa, and Stoljar (2004) point out in their very nice introduction to an anthology on Jackson’s Mary argument, one can find versions of the argument in Russell (1967), Dunne (1958), Broad (1925), Farrell (1950), Feigl (1958), Meehl (1966), Nagel (1974), and others. As we just noted, Smart (1959) replied to a version of it almost thirty years before it resurfaced in Jackson (1982). While Ludlow, 13

Knowledge, Thought, and the Case for Dualism

Nagasawa, and Stoljar (2004) seem to suggest that there is something new in Jackson’s version of the argument, I’m skeptical of that claim. But I don’t say that to cast any aspersions on the importance of Jackson’s version of the argument. Jackson’s statement of the argument was particularly clear and “colorful” and it gave new impetus to some very old arguments about the connections between the epistemology and the metaphysics of mind and matter. So Jackson describes a brilliant neurologist Mary who has been deprived of color experiences. In Jackson’s version of the thought experiment Mary was confined to a black and white room, but it is just as easy to suppose that Mary is afflicted with an actual, but rare, condition – black/white color blindness. Everything appears to her the way images on a black and white film would appear to you and me. For brevity let’s refer to Mary in this state as color-deprived Mary. Despite being color challenged, Mary systematically investigates the physical world and discovers “everything” there is to know about the various objects in that world. Those objects include bodies, brains, and all that goes on in them. On one version of the story, we go on to describe Mary as having systematic and complete knowledge of the truths of physics and biology. Her absence of color experiences is said to be no hindrance to her scientific education. Somewhat more cautiously, we might simply describe her as having exhaustive knowledge of whatever truths can be discovered by someone with the color blindness from which she suffers. The strongly intuitive conclusion is that no matter how thorough and complete her knowledge becomes, she will never discover certain truths. Specifically she will never discover what color experiences are like. And if we add the supposition that she can discover all of the truths of science in her color-blinded condition, we can conclude that there are important truths that science leaves out of a complete world view. To make the thought experiment vivid, we are to imagine that we find a 14

Setting the stage

cure for her radical color blindness, a cure that enables her to discover for the first time what it is like for something to appear colored. Upon her cure she discovers that the world contained kinds of experiences of which she was previously ignorant. Part of the problem with the Mary thought experiment is that philosophers who discuss the thought experiment may be thinking of what Mary is supposed to discover when she gains color experience in quite different ways. So the Mary who is released from the confines of black and white experience can be said to learn what red is. Or she could be described as coming to discover what it is to see red. Or what it is for something to look red. Or what it is for it to seem as if something is red, or be appeared to red-ly (in the language favored by adverbial theorists). How the object of the new knowledge is described might make a big difference when it comes to the plausibility of various attempts to reconcile the new knowledge with the physicalist world view. If redness is a property of physical objects (like tablecloths and shirts), it probably isn’t going to sound right to many philosophers to suppose that Mary discovered what redness is all about. After all most philosophers today think that the property of being red that physical objects exemplify just is the property of having a surface that under certain conditions reflects and absorbs light of a certain wavelength. Some of us still think that is an obviously false view,5 but the resolution of that issue must await our subsequent discussion of informative identity (Chapter 5). 5

G. E. Moore barely thought he needed an argument to refute the idea. Here’s Moore (1903, p. 62): “Consider yellow, for example. We may try to define it, by describing its physical equivalent; we may state what kind of light-vibrations must stimulate the normal eye, in order that we may perceive it. But a moment’s reflection is sufficient to show that those light-vibrations are not themselves what we mean by yellow. They are not what we perceive. Indeed we should never have been able to discover their existence, unless we had first been struck by the patent difference of quality between the different colours. The most we can be entitled to say of those vibrations is that they are what corresponds in space to the yellow which we actually perceive.”

15

Knowledge, Thought, and the Case for Dualism

It is even a bit problematic to view the critical piece of new knowledge that Mary gains as knowledge concerning how physical objects look. That way of talking might suggest that it is a relational state of affairs about which Mary gains knowledge, a relational state of affairs whose relata include the physical objects that look, for example, red. But Mary could, in principle, gain the knowledge that philosophers of mind are so concerned with by having her first color experiences in a dream or in a hallucination – or to be more careful, she could gain such knowledge if the experience gives rise to the relevant phenomenal concepts. At least this is so if, as I shall argue later, the relevant knowledge that Mary lacked in her color-deprived state is that certain sorts of properties are exemplified. It is for that reason that one should always be careful in one’s use of modal operators to characterize the predicament of the colorchallenged Mary. It was never logically impossible for Mary to discover what the relevant color experiences were like when confined to the black and white room (or when afflicted with black/white color blindness). It was always logically possible for her to dream in color (the reverse of what many of my students tell me they do) or to hallucinate color. Pace Hume, it isn’t even clear that it would be impossible for her to conjure up the thought of phenomenal redness (red color experiences) ex nihilo. The irony is that Hume, the empiricist’s empiricist, was obviously committed to the view that there is no incoherence in the hypothesis that a simple idea of color comes into existence without a prior experience of that very color (Hume, 1888, Bk. 1, Pt. 1, sec. 1, pp. 5–6): There is however one contradictory phenomenon, which may prove, that’ tis not absolutely impossible for ideas to go before their correspondent impressions. I believe it will readily be allow’d, that the several distinct ideas of colours, which enter by the eyes, or those of sounds, which are convey’d by the hearing, are really different from each other, tho’ at the same time

16

Setting the stage resembling. Now if this be true of different colours, it must be no less so of the different shades of the same colour, that each of them produces a distinct idea independent of the rest . . . Suppose therefore a person to have enjoyed his sight for thirty years, and to have become perfectly well acquainted with colours of all kinds, excepting one particular shade of blue, for instance, which it never has been his fortune to meet with. Let all the different shades of that colour, except that single one, be plac’d before him, descending gradually from the deepest to the lightest;’ tis plain, that he will perceive a blank, where that shade is wanting, and will be sensible, that there is greater distance in that place betwixt the continguous colours, than in any other. Now I ask, whether’ tis possible for him, from his own imagination, to supply this deficiency, and raise up to himself the idea of that particular shade, tho’ it had never been conveyed to him by his senses? I believe there are few but will be of opinion that he can, and this may serve as a proof, that the simple ideas are not always derived from the correspondent impression.

Interestingly, Hume didn’t seem to think that this observation was all that important to his central claim arguing that “the instance is so particular and singular, that’ tis scarce worth our observing, and does not merit that for it alone we should alter our general maxim” (p. 6). But the “general maxim” in question is that all simple ideas correspond to prior impressions and a “singular” example can destroy a general maxim. Moreover, Hume is committed in his discussion of causation to both the view that it is in principle possible for anything to cause anything, and, indeed, that there is no contradiction in supposing that something comes into existence without a cause (for example, the having of a simple idea). Hume, I believe, was right about all this, and thus the most that Jackson should have been suggesting with his thought experiment is that Mary in her color-deprived state would never have direct knowledge of the intrinsic nature of these color experiences she hears people talk about. Nor is it very likely that she would have even been able to 17

Knowledge, Thought, and the Case for Dualism

form an idea of their intrinsic nature given the limitations of her experiences (she has no “gaps,” for example, in the continua of her color experiences that would allow her to conjure up the idea of a kind of color experience she has never had). If somehow she did find herself able to think of a color experience as it is intrinsically, she would still lack the kind of knowledge that others have that such experiences occur. Jackson and Nagel sometimes talk about people deprived of various sorts of experiences lacking knowledge of what it is like to have those experiences. Nagel famously illustrates the idea with the more exotic question of whether it would be possible for a human to come to know what it is like to be a bat: So if extrapolation from our own case is involved in the idea of what it is like to be a bat, the extrapolation must be incompletable. We cannot form more than a schematic conception of what it is like. For example, we may ascribe general types of experience on the basis of the animal’s structure and behavior. Thus we describe bat sonar as a form of three-dimensional forward perception; we believe that bats feel some versions of pain, fear, hunger, and lust, and that they have other, more familiar types of perception besides sonar. But we believe that these experiences also have in each case a specific subjective character, which it is beyond our ability to conceive. And if there is conscious life elsewhere in the universe, it is likely that some of it will not be describable even in the most general experiential terms available to us. (Nagel, 1974, pp. 439–40)

Jackson (1986, p. 291) uses the same “what it is like” language to describe what Mary didn’t know when she lacked color experiences: It seems that Mary [in her color-deprived] state does not know all there is to know. For when she is let out of the black-and-white room or given a color television, she will learn what it is like to see something red, say. This is rightly described as learning – she will not say “ho, hum.” Hence, physicalism is false. This is the knowledge argument against physicalism in one of its manifestations.

18

Setting the stage

I don’t like the Nagel/Jackson way of putting things. As both acknowledge when they are careful, everything is like everything else in some respects, and I imagine that Mary could figure out some of the ways in which visual sensations of color are like other states of affairs (for example, that they have duration, that they cause people under certain circumstances to say various things). What Mary comes first to know and then can start thinking about is that a certain property is exemplified. Before she had the sensation of redness she didn’t know that the exemplification of that property is part of the structure of reality. To be sure, she probably was in a position to know the existential claim that there is a property satisfying certain descriptions (for example the one that is caused by certain neural activity, or that causes people to say certain things, or that is the change presented by looking at a tomato in good light), but in a perfectly clear sense she didn’t know what that property was. It is tempting to suggest that Mary didn’t have knowledge of the property’s intrinsic nature, but this is dangerous. It is dangerous because talk of natures is ambiguous in this very special context in which we are trying to distinguish direct and indirect knowledge of properties. Ordinarily, when one talks of the nature of something, we talk about properties of the thing. If asked to describe a dog’s nature I might describe it as curious, intelligent, affectionate, and so on. And insofar as properties can exemplify properties we can talk of knowing the nature of a property. But Mary does know the nature of color experiences in this sense. She knows that whatever color experiences are they have a number of different properties (including, of course, the causal properties described above). But she doesn’t know what the property is that has these other, different properties. When I talk of knowing the intrinsic nature of a property, that is just a roundabout way of talking about knowing what the property is as opposed to knowing what properties it has. 19

Knowledge, Thought, and the Case for Dualism

Of course, there is an extensive literature on the Jackson thought experiment. Although Jackson himself has now changed his mind, when he first advanced the argument, he intended to reach a metaphysical conclusion – that there are qualia (his name for the properties that Mary discovered when she had her first color experience). Furthermore, he concluded that these qualia resist identification with any physical properties. Like Descartes, Jackson was trying to reach a metaphysical conclusion based on epistemological premises. Jackson was cagey, however, in his characterization of the epistemological premises. He talked about Mary’s discovering what it is like to experience color. But even among philosophers who agree that Mary with her exhaustive scientific knowledge lacked knowledge of something prior to her gaining color experience, there is radical disagreement about how to characterize the relevant knowledge, disagreement we will need to sift through carefully. The following is (or should be) uncontroversial. Prior to her gaining color experience Mary never had color experiences! And, as Lewis points out (2004) it follows that prior to her gaining color experience Mary never knew that she had color experiences (because you can’t know what isn’t so). So the relevant questions presumably concern color-deprived Mary’s knowledge of the color experiences had by others. Smart, who essentially thought of Jackson’s objection to physicalism long before the appearance of the argument, would concede the following: When color-deprived Mary systematically investigated the brains of other people she successfully entertained propositions which described the color experiences of others. Once having had color experiences herself, Mary discovered new ways of thinking about those color experiences. She entertained new propositions describing color experiences. She formed new beliefs about the color experiences of people. But these new thoughts/propositions/beliefs picked out 20

Setting the stage

precisely the same reality as did her old thoughts/propositions/ beliefs. Lewis (2004), Nemirow (1990, 2007), and others, deny that when Mary acquired color vision she came to believe any new kinds of propositions.6 She acquired a new kind of knowledge, perhaps, but the knowledge wasn’t propositional knowledge – it was a species of knowing how, an ability. By contrast, Churchland (1989), Bigelow and Pargetter (1990), Conee (1994), and others also deny that Mary came to believe any new propositions, insisting that the new knowledge is knowledge by acquaintance. We’ll need to sort through these and other positions in what follows. It should go without saying that there is nothing particularly interesting about color experiences. Color-deprived Mary could have been replaced with pain-free Mary, taste-impaired Mary, olfactory-impaired Mary, or hearing-impaired Mary. Pain-free Mary could study the physiology of people from now until the end of time without ever knowing what pain is like. Hearing-impaired Mary could do the same without knowing what middle C on the piano sounds like. Taste-impaired Mary for all her endless scientific knowledge would never know what peaches taste like, and olfactory-impaired Mary would never know the smell of strawberries.

1.2

h a r d v s . e a s y pr o b l e m s o f consciousness

One finds in the literature a distinction between the so-called hard and easy problems of consciousness. The problem of how to analyze qualia (the “given” character of an experience like pain or visual appearance) is associated with the hard problem. According to some, the problem of how to understand 6

Remember, we are talking about kinds of propositions. She obviously did acquire for the first time belief that she was having a color experience.

21

Knowledge, Thought, and the Case for Dualism

intentional states – states like belief, desire, and fear – is associated with the easy problem. But I’m not sure that there is really much consensus on how to understand the distinction between hard and easy problems. As far as I know, the “hard” vs. “easy” terminology was introduced by Chalmers (1995, 1996, 2004). Chalmers (1995, p. 200) describes the contrast between the easy and hard problems in the following passage: There is not just one problem of consciousness. “Consciousness” is an ambiguous term, referring to many different phenomena. Each of these phenomena needs to be explained, but some are easier to explain than others. At the start, it is useful to divide the associated problems of consciousness into “hard” and “easy” problems. The easy problems of consciousness are those that seem directly susceptible to the standard methods of cognitive science, whereby a phenomenon is explained in terms of computational or neural mechanisms. The hard problems are those that seem to resist those methods. The easy problems of consciousness include those of explaining the following phenomena:  the ability to discriminate, categorize, and react to environmental stimuli;  the integration of information by a cognitive system;  the reportability of mental states;  the ability of a system to access its own internal states;  the focus of attention;  the deliberate control of behavior;  the difference between wakefulness and sleep. All of these phenomena are associated with the notion of consciousness. For example, one sometimes says that a mental state is conscious when it is verbally reportable, or when it is internally accessible. Sometimes a system is said to be conscious of some information when it has the ability to react on the basis of that information, or, more strongly, when it attends to that information, or when it can integrate that information and exploit it in the

22

Setting the stage sophisticated control of behavior. We sometimes say that an action is conscious precisely when it is deliberate. Often, we say that an organism is conscious as another way of saying that it is awake. There is no real issue about whether these phenomena can be explained scientifically. All of them are straightforwardly vulnerable to explanation in terms of computational or neural mechanisms. To explain access and reportability, for example, we need only specify the mechanism by which information about internal states is retrieved and made available for verbal report. To explain the integration of information, we need only exhibit mechanisms by which information is brought together and exploited by later processes. For an account of sleep and wakefulness, an appropriate neurophysiological account of the processes responsible for organisms’ contrasting behavior in those states will suffice. In each case, an appropriate cognitive or neurophysiological model can clearly do the explanatory work. If these phenomena were all there was to consciousness, then consciousness would not be much of a problem. Although we do not yet have anything close to a complete explanation of these phenomena, we have a clear idea of how we might go about explaining them. This is why I call these problems the easy problems. Of course, “easy” is a relative term. Getting the details right will probably take a century or two of difficult empirical work. Still, there is every reason to believe that the methods of cognitive science and neuroscience will succeed. The really hard problem of consciousness is the problem of experience. When we think and perceive, there is a whir of information-processing, but there is also a subjective aspect. As Nagel (1974) has put it, there is something it is like to be a conscious organism. This subjective aspect is experience. When we see, for example, we experience visual sensations: the felt quality of redness, the experience of dark and light, the quality of depth in a visual field. Other experiences go along with perception in different modalities: the sound of a clarinet, the smell of mothballs. Then there are bodily sensations, from pains to orgasms; mental images that are conjured up internally; the felt quality of emotion, and the experience of a stream of conscious

23

Knowledge, Thought, and the Case for Dualism thought. What unites all of these states is that there is something it is like to be in them. All of them are states of experience.

On one natural way of understanding the distinction Chalmers is pointing to, the hard problem is just the distinctively philosophical problem of giving an ontological analysis of what it is to be in a mental state, an analysis that captures the undeniable fact that there is something that it is “like” to be in such a state. The (comparatively) easy problem is the scientific problem of giving a causal account of what brings about the relevant state that can be studied philosophically. Or put somewhat differently, the easy problem is giving a causal account of the body’s interaction with its environment ignoring the mental state that is of such interest to philosophers. An analogy might be useful. The epistemological and metaphysical problems of perception dramatically developed by Descartes, but discussed as early as Sextus, have little or nothing to do with the empirical study of perception. The empirical study of perception brackets both epistemological and metaphysical philosophical problems. That study assumes that we are (somehow) getting knowledge of both macro- and microphysical processes and goes on to use both perception and more theoretical sorts of reasoning to give an account of the way in which our sense organs causally interact with the physical world. The account of visual perception, for example, tells us how light can reflect off the surface of physical objects, can affect the retina of a person’s eye beginning a complex chain of physiological causes and effects, crucial links of which occur in the brain. Progress has been made filling in the details of this account, and there is still much to learn (particularly at the micro level). But, arguably, none of that progress has helped with the philosophical task of analyzing what it is to perceive an object and how we should conceptualize the different ontological constituents of perception. The scientific causal 24

Setting the stage

explanation, for example, leaves open the question of whether one should give a philosophical causal theory of what it is to perceive an object. It leaves open the question of whether any of the changes that take place in the brain can be identified with the change of appearance upon which I rely in forming beliefs and expectations about my physical surroundings (or even whether there is such a thing as an appearance upon which I rely). It also leaves open the question of whether there is a substantive epistemic or ontological common denominator to veridical and non-veridical perception. In precisely the same way, scientific accounts of what produces paradigmatic mental states leave completely open the question of how to understand what those states are. No scientific investigation into the neurological activity accompanying belief, desire, fear, hope, anxiety, pain, or sense experience will ever decide the controversies between dualists, mind–brain identity theorists, or functionalists. These philosophical controversies center around questions that scientists don’t address and aren’t competent, qua scientists, to address. Of course, the scientific attempt to answer the easy questions might presuppose that there is a relatively unproblematic way of accessing when people are in the aforementioned mental states, and, as we shall see in Chapter 3, that itself might raise metaphilosophical questions about where we should put our epistemological priorities. Not everyone understands the “hard”/“easy” distinction the way I just described above. Dennett, for example, sometimes seems to suggest that the easy problem is giving a philosophically adequate physicalist account of intentional states, while the hard problem is giving a philosophically adequate physicalist account of that distinctive qualitative “feel” associated with such mental states as visual experience or pain. It is supposed to be easier according to some physicalists to give a plausible physicalist account of intentional states and even to 25

Knowledge, Thought, and the Case for Dualism

understand how a Mary deprived of a certain sort of intentional state could still figure out through observations of the physical world what the intentional state she never had is like. So I guess the idea is that a fear-deprived Mary could come to know what fear is like without ever having felt fear, or to discover what thirst is like without ever having felt thirst, or what lust is like without ever having experienced lust. The problem of qualia is supposed to be the problem of finding room for “raw” sensory states in a physicalist world view. The quote above from Chalmers might suggest a distinction of the sort just discussed. After all, Chalmers associates the easy problem with giving an account of discrimination of one’s environment and deliberately choosing an action. On any ordinary way of understanding such states they involve paradigmatic intentional states (like belief, desire, and will). And if they don’t present a distinctive philosophical problem it must be because these sorts of states are relatively unproblematic. It is not at all clear to me that there is an important distinction between the challenge of accommodating raw feels and the allegedly easier problem of accommodating various sorts of intentional states into a physicalist world view. More often than not the relative ease with which one can identify intentional states with physical states is supposed to have something to do with the supposed ease with which one can develop a plausible functional account of most intentional states. But there are two major problems with this presupposition. First, functionalist accounts of mental states, properly understood, are no more plausible than the crude logical behaviorisms that preceded them. Second, many philosophers would argue that the exemplifications of qualia are themselves intentional states. I’ve never found this last view very plausible myself, but there are philosophers who think that pain is just a species of intentional state representing damage to the body. Visual sensation is an intentional state representing (or, according to some, 26

Setting the stage

misrepresenting color, and shape). And so on for various other modes of sensation. To be sure, the strategy of assimilating the exemplification of qualia to intentional states is often a strategy designed to transform a hard problem into an “easy problem” but, as the old saying goes, one person’s modus ponens is another’s modus tollens. If one can construe all mental states as intentional, the force of epistemological arguments for various forms of dualism might suggest that the easy problem isn’t an easy problem at all. Again, we’ll discuss these issues in much more detail throughout this book. Still other philosophers (as we shall see, perhaps, Jackson himself after he recants his knowledge argument) find intentional states unproblematic because they have no character other than what they inherit from their “objects.” Experience, some argue (Moore, 1922; Harman, 1990; Tye, 2010), is both intentional and diaphanous. The character of a visual experience is exhausted by what it is an experience of. As long as one can find the object of experience in the world of physical objects one has a clear path to embracing physicalism. Again, we’ll look more at such views later.7

1.3

s ummary

In this chapter we have just tried to lay some groundwork by surveying briefly the knowledge arguments that philosophers 7

And as we shall see the path isn’t all that clear. Those who argue that experience is diaphanous and exhausted by its content obviously face the problem of how to understand hallucination, or more generally, misrepresentation. To secure a character for radically non-veridical experience we will need to find objects for such experiences. And it isn’t going to be easy finding the ghosts, goblins, unicorns, and pink rats about which we dream and hallucinate. The view might inexorably lead one to populating one’s world with physical objects that don’t exist. Based on conversations with him, I believe that Harman is willing to make just such a move, but one must wonder about a “naturalism” that has no place for the irreducibly mental, but is a haven for stuff that has some sort of being but has no existence.

27

Knowledge, Thought, and the Case for Dualism

have employed in trying to undermine physicalism. Much must be done in order to evaluate these arguments. As we shall see, we cannot ignore substantive philosophical issues concerning the nature of knowledge and justification, and hope to get to the heart of the issues raised by knowledge arguments. But even before we look more closely at knowledge and justification, we would be well advised to get clearer about precisely what the debate is between physicalists and dualists. It will become evident that there isn’t just one debate. There are instead a number of different debates. It may be that the knowledge arguments for dualism leave untouched some forms of physicalism but not others.

28

2 Distinctions: versions of physicalism and dualism Many philosophers seem willing to die in the trenches defending versions of physicalism or dualism. It is, then, a matter of considerable irony that there is really no consensus on how to define the terms of the debate. Nor is it all that clear what kinds of entities philosophers are trying to put into the categories of the physical and the mental. Leaving undefined “physical” and “mental” for the moment, we can distinguish all of the following physicalist/dualist controversies.

2.1

substance physicalism/dualism

With the possible exception of abstract entities (e.g. numbers), the substance physicalist thinks that there is only one kind of substance, physical substance. The substance dualist thinks that there are at least two kinds of substances that are mutually exclusive, physical substance and mental substance. As we noted in Chapter 1, Descartes appeared to argue for a dualism about substance (though it is certain he would also embrace a dualism of properties). Of course, one’s going to be neither a substance physicalist nor a substance dualist if one doesn’t think that there is such a thing as substance. A substance is supposed to be something that is not a property but that exemplifies properties. Philosophers also sometimes talk the same way about particulars, also known as bare particulars (Bergmann 1967; Addis 1989; Armstrong 1989). The 29

Knowledge, Thought, and the Case for Dualism

difference between substances and bare particulars isn’t obvious. However, the most famous defenders of bare particulars thought that a particular was a momentary existent whose momentary character is exhausted by its properties at the moment it exists. By contrast, philosophers who posit substances were more apt to think of them as enduring, and sometimes these philosophers seemed to talk as if they have a nature that is independent of the properties they exemplify. Substances were sometimes said to possess causal powers, powers that do not derive from the properties they exemplify (Harré and Madden, 1975).1 But, in the final analysis, it is not at all easy to see what could differentiate one substance from another, qua substance, but a property or properties that substance exemplifies. It is, I will argue later, particularly difficult when it comes to differentiating physical stuff.2 There is another way to think of the distinction between particulars and substances that also relates, at least indirectly, to the idea that substances can endure. Because it makes sense to suppose that a substance can exist at various times and in different places at those different times, it might also make sense to talk about a substance’s essential nature. A property 1

2

It is difficult to understate the impact of the Humean revolution when it comes to thought about causation. With his conviction that causation has something to do with regularity and the apparent impossibility of even specifying regularity without reference to properties, properties started to take center stage in our attempt to understand causation. Even those who reject Hume’s idea that causation supervenes on contingent regularities but who embrace the “generality” thesis that our understanding of causation is parasitic upon our understanding of lawful generality will inevitably be led to properties and relations between properties in their attempt to get an analysis of causation. But prior to Hume, most thought causation was wedded to this difficult idea of substance and its “powers.” Ironically, given current attitudes towards the mental as causally superfluous, the mind itself was often taken to be the paradigm of something with causal powers. Berkeley (1713, First Dialogue) went so far as to define matter as causally inert. In Chapter 8 we’ll turn our attention to one of the other obvious candidates for a substance – the self.

30

Distinctions: versions of physicalism and dualism

of a substance is said to be essential to a substance if the substance could not exist without that property. So Descartes, for example, took it to be an essential property of all physical objects that they are extended (1641, Med. 6, para. 9). Descartes also took it to be an essential property of the kind of substance that he called mind that it thinks (1641, Med. 2, para. 6). He also seemed to think that one and the same substance couldn’t have two such disparate essential properties, though it is not clear what his argument was for this claim. Leaving that aside, though, we can try to explain this distinction between essential and (as they are called) accidental properties of a thing with Descartes’s own famous example of a piece of wax (1641, Med. 2, paras. 11–12). As we ordinarily think of wax there are many properties a piece of wax can gain or lose and still continue to exist. As we bring the wax close to the fire its color, odor, and shape change. But, at least as we ordinarily talk, it still exists albeit with different properties. By contrast, throughout the various changes the wax undergoes, it remains extended, and I suppose there is some intuitive plausibility to the idea that it couldn’t lose that property and still exist. Many contemporary philosophers will argue that the wax couldn’t lose its fundamental molecular structure and continue to exist.3 I’ve been talking as if we should try to understand all this talk about essential versus accidental properties in terms of what could have been true of the wax in the past or what could be true of the wax in the future. These days philosophers are more prone to talk about possible worlds. A given thing X has a property F essentially if there are no metaphysically possible worlds in which X exists without F. I don’t want to get off on too long a rant here, but I should go on record as suggesting 3

Though I’m not sure that this philosophical thesis has received any backing from “common sense.”

31

Knowledge, Thought, and the Case for Dualism

that the use of the possible worlds metaphor to elucidate various modal claims has had a most unfortunate effect on contemporary philosophy. While all but the most metaphysically extravagant will concede that there is really only one world and that all truth, including modal truth, has truthmakers to be found in that one actual world, the easy appeal to possible worlds talk in testing “intuitions” about modality leads to a kind of intellectual sloth. Too often we fail to do the really hard work involved in figuring out precisely what feature of the (one and only) world is making it true that a proposition is contingent or is making it true (or even intelligible) that a thing has one property essentially and another only accidentally. And, despite the post-Kripkean resurgence of essentialism, I’m not convinced that anyone really understands what the difference is supposed to be between the kind of “glue” that won’t allow a property to be detached from a substance without destroying the substance and the weaker glue that allows a property to be stripped from a substance while leaving the substance intact. Earlier, we sharply distinguished the claim that there is a distinction between de re and de dicto reports of beliefs from the claim that there are two ontologically distinct kinds of belief states. Again, a philosopher convinced that the contents of all beliefs are propositions (that all beliefs are de dicto beliefs) can still acknowledge a distinction at the level of language between what we commit ourselves to when we ascribe belief. When we report de re that S believes of the F that it is G, we do commit ourselves to the existence of a unique F, and we do not commit ourselves to S’s believing the proposition that the F is G. After that all bets are off as to how to understand the content of the de re ascription of belief. As I suggested earlier, one could take the de re report to simply leave open in certain respects which proposition S believes (provided that its subject concept picks out the same thing as 32

Distinctions: versions of physicalism and dualism

“the F” and it still attributes the property of being G to that subject).4 One might take the same general approach to understanding de re necessity. We can certainly mark a distinction between (1) The bachelor next door didn’t have to be unmarried and (2) It is not necessarily false that the bachelor next door is unmarried. (1) is true and (2) is false. And we can say that what makes (1) true is that the property of being unmarried isn’t an essential property of that guy who we pick out in part by virtue of his accidental property of being a bachelor. But we don’t have to take the concept of an essential property as one that is irreducible. It may be that we can analyze ascription of de re necessity in terms of de dicto necessity. (1) might be taken to assert that there is just one person who is a bachelor and living next door, and there is a way of picking out that person and attributing the property of being unmarried to him such that the resulting proposition is contingent. The approach would need considerable tweaking if we are to get a reduction of de re necessity assertions that matches our intuitions. The crude account sketched above would allow that the triangle I drew on my blackboard didn’t have to have three sides. After all there is a way of picking out that object (e.g. “the object I just drew on the blackboard”) such that when I attribute the property of having three sides to the object the resulting proposition is contingent. The solution might lie in placing restrictions on the concepts that can be used to single out the thing in question. Where we are dealing with kinds of things (like humans or

4

There is probably also a use of locutions such as “believes of” according to which one can’t truly say of S that he believes of the F that it is G unless S has had some relatively direct “first-hand” experience of the F. But it is not clear how direct the experience would have to be (does television allow us to form de re beliefs about the people we “see” on television?), and what we would take the implications of our assertion to be might vary greatly from context to context.

33

Knowledge, Thought, and the Case for Dualism

triangles) the replacement descriptions need to characterize correctly the relevant kind.5 However we understand talk of the distinction between essential and accidental properties, many philosophers have always been highly suspicious of the categories of both substance and particular. Even the most ardent supporters of substances/particulars most often regard commitment to their existence as a matter of dialectical necessity. Particulars, for example, are often introduced to individuate things that are obviously distinct despite their sharing the same nonrelational properties.6 It’s hard (though not impossible) to find philosophers who take themselves to be phenomenologically acquainted with substances or particulars per se. I must confess that my sympathies lie with the radical empiricists who found literally unintelligible the idea that there is something that remains (even if only in thought) when we strip a thing of all of its properties. Admittedly, this is a slightly misleading way of characterizing the view under attack for, typically, substance/particular theorists will probably admit that substances/particulars necessarily come “clothed” in properties. The distinction between the thing and its properties is what Armstrong (1978) called (following Scotus) a “formal” distinction. Epistemologically, it is surely highly plausible to conclude that when we become aware of a thing it is through our awareness of the instantiation of certain properties. It is that redness and shape or that texture and weight that we see and feel. Indeed, given the epistemological priority of properties over things, it is surely tempting to try to 5

6

But everything falls under indefinitely many categories, is of indefinitely many kinds, so the proposal here still doesn’t give nearly enough guidance as to how to carry out the reduction. See for example, Addis (1967). Gustav Bergmann (1964, p. 278) thinks that one can be directly acquainted with bare particulars, but also seems to suggest that even if that were not so, there might be good dialectical reasons to introduce them.

34

Distinctions: versions of physicalism and dualism

reduce ordinary objects to bundles of properties. It is an understatement, however, to suggest that a successful reduction is hard to come by, and even a cursory treatment of this issue would take us too far afield. The above presupposes that when we are trying to figure out the relationship between properties and the things that have them, we are best off trying to do so in connection with physical objects and their properties. As we shall see in Chapter 6, there is a radical empiricism that we must take very seriously and according to which we have no access to the intrinsic nature of physical objects. If that view turns out to be true, then it might be folly to try to understand the connection between properties and that which has them focusing on ordinary objects like tables and chairs. Indeed, there is a long tradition in philosophy according to which the properties we know best are properties of the phenomenal experiences we ourselves have (alternatively, the experiences themselves if we think of the experiences as properties or modifications of the mind). And if there is a substance we have phenomenological acquaintance with, the best candidate for such a substance is our self. I’ll talk briefly about this rather extreme view in Chapter 8. Whether or not one can get by without a commitment to the existence of substances or particulars, one might insist on the more modest conclusion that it makes no sense to talk of the nature of a substance or particular that is, somehow, independent of the properties it exemplifies. After all, talk about natures is surely just another way of talking about properties. If that’s true, then it might well be folly to try to understand a distinction between kinds of substances that is independent of a distinction between kinds of properties. If the physicalist thinks that only physical substances exist, for example, it is tempting to suppose that in the final analysis this will be, most fundamentally, a claim about what kinds of properties are exemplified. 35

Knowledge, Thought, and the Case for Dualism

2.2

pr o p e r t y p h y s i c a l i s m / d u a l i s m

No form of dualism is all that popular now, but property dualism seems far more in vogue these days than substance dualism. As we just saw, the ontological categories of substance and particular are philosophically controversial. And even if we admit the existence of substances and particulars, our understanding of distinctions between kinds of substances or particulars would seem to be parasitic upon our understanding of distinctions between kinds of properties. So Jackson (when he endorsed the knowledge argument) and those influenced by him often argue that the knowledge argument shows that there are kinds of properties beyond those recognized by the physicalist. Of course this presupposes that we have a principled way of capturing all and only those properties recognized by the physicalist. And as we shall see, it is no easy task to say what makes a property physical, or, for that matter, mental. Until we have a clear way of defining the distinction between two sorts of properties, we won’t know what philosophers are arguing about when they either identify or distinguish physical properties and mental properties. I suggested earlier that the category of substance or particular is philosophically controversial. It would be disingenuous to suggest that the category of property has been enthusiastically endorsed by all philosophers. The irony is that just as some of the empiricists were puzzled by and, even, anxious to eliminate from their ontology, substance and particular, so also, other philosophers have been equally anxious to purge, at least as a fundamental category, properties from their ontologies. Among property theorists we can, of course, distinguish realists and trope theorists. The realist takes properties to be universals. A given property, like triangularity, is a universal if the triangularity of this triangle a is numerically identical with 36

Distinctions: versions of physicalism and dualism

the triangularity of that distinct triangle b. There are many important in-house debates among realists concerning universals. So Platonists think that universals can exist even if they are never instantiated, while so-called in res realists think that only exemplified universals exist. And some philosophers will argue that only determinate universals (this particular shade of red, for example) exist, while others argue that so-called determinable universals (redness, being colored, for example) exist. Some realists are extremely liberal with respect to the universals that exist – they will countenance a universal for virtually every well-formed, meaningful predicate expression. Others are much more ontologically parsimonious. On one relatively popular view, for example, only universals that figure in lawful connections should be construed as real. While one can heroically try to “build” particular things out of bundles of universals the problems are serious indeed, and, as we noted above, many realists will (perhaps grudgingly) introduce substances or particulars to which the universals are “attached” in order to solve the problem of individuation.7 The trope theorist takes properties to be particulars and is likely to “build” individual things out of these “particular” properties. According to the trope theorist the redness of a is not identical to the redness of b even if the former is exactly like the latter. Because the properties of a and b are not identical the philosopher trying to bundle things out of tropes might seem to

7

The classic problem of individuation is still Black’s universe (1952). Max Black asked us to imagine a universe in which there are just two red spheres with precisely the same mass. If we call the one A and the other B we need to explain what makes A distinct from B. By hypothesis all of their non-relational properties are precisely the same, and if the realist about properties is correct, and A and B were nothing but their non-relational properties, A and B would be identical. It is hard to see how we can appeal to relational properties, because we haven’t yet analyzed anything that can stand in the relevant relations. Enter substances or particulars to make A distinct from B.

37

Knowledge, Thought, and the Case for Dualism

have more hope than the philosopher trying to bundle things out of universals. In what follows I’m not even going to try to settle disputes between realists and trope theorists concerning the nature of properties. We can’t talk of the discovery Mary is supposed to make without talking about kinds of properties, but I’ll assume that any plausible trope theorist will have some way of making sense of the claim that two tropes are of the same kind – that, for example, this trope and that trope are both red tropes. I suspect that in the final analysis one will be hard-pressed to avoid Russell’s conclusion (1967, especially Chapters 9–10) that we will be dialectically forced to admit at least some universals, even if it is only the universal of resemblance. But there is no philosophical reason to accept an ontological version of the old adage: in for a penny, in for a pound. There may be good reasons to embrace a mixed view, a view according to which some properties are treated as tropes, while others (perhaps relations) are treated as universals. Both the realist and the trope theorist are, in a sense, nonreductivists when it comes to properties. Each in their own way recognizes properties as part of the ultimate ontological furniture of the world. The term “nominalist” is sometimes used to characterize any philosopher who eschews universals, and in that sense the trope theorist is a nominalist. But there are much more extreme forms of nominalism that we should mention in passing. They are sometimes called “set nominalism” and “linguistic nominalism.” On the crudest versions of their views, these nominalists seek to avoid a commitment to properties by construing subject/predicate assertions as making claims about set membership or linguistic dispositions. On the former view, to say that something is red is to say that it belongs to the class of red things. On the latter view, to say that something is red is to say that it is the kind of thing we are disposed to call “red.” 38

Distinctions: versions of physicalism and dualism

The most unsophisticated undergraduate will, of course, spot the obvious problem with these crude statements of nominalism. One wants to know from the set nominalist what the principle of set membership is, and one wants from the linguistic nominalist an account of what it is about an object that generates the linguistic disposition to call it red. If there is a set of red things it is surely in virtue of the various members of the set exemplifying the property of being red. And it is surely in virtue of our linguistically independent belief that the object is red that we call it “red” when we understand the meaning of “red.” The unsophisticated undergraduate is right in rejecting extreme nominalisms. To be sure, the extreme nominalists will thrash about trying to make their views more palatable, but in the end the views will still seem almost comically false. I haven’t shown that here and I suppose I should apologize for being needlessly polemical. I should instead simply acknowledge that in the rest of this work I will presuppose that these views are false. So if we are to have a well-defined debate between property physicalists and property dualists, we will need to have a welldefined characterization of physical and mental properties – more about that shortly.

2.3

fa c t p h y s i c a l i s m / d u a l i s m

If there are properties, there are almost certainly facts. A fact is best understood in terms of property exemplification. The realist committed to substances or particulars will think of property exemplification as something that always involves a substance or particular that exemplifies the property. The trope theorist is more likely to take property exemplification to be bundle membership of some sort. Again, as I said earlier, I’m assuming that trope theorists will commit to finding some 39

Knowledge, Thought, and the Case for Dualism

way of translating talk about property kinds within the ontological framework they favor. So even a trope theorist will acknowledge that in some sense or other we can count the number of red tropes and, in doing so, we count the number of times redness is exemplified. By now it is becoming a refrain, but the ontological category of fact is another category that is philosophically controversial. One reason is that fact talk is so closely related to truth talk. Indeed, in ordinary discourse to state that it is a fact that P is probably just to assert in a particularly emphatic way that P, or that it is true that P. Still, the fact theorist will insist that we not be misled by ordinary language. We need the category of fact because we need the category of truthmaker. Again, there is no space here to go into all of the thorny issues that arise in defending a plausible conception of truth with a robust role to play for truth-makers. Most philosophers concede that the general idea of a correspondence theory of truth has enormous plausibility and that one should abandon it only in the face of serious objections. I have argued elsewhere (2002) that there are no fatal objections to the right correspondence theory of truth. If one includes in one’s ontology facts, then one has another plausible category about which physicalists and dualists can debate provided that one can define clearly the distinction between a physical fact and a mental fact. Just as was true of substances and individuals, it is plausible to suppose that whatever distinctions we draw between kinds of facts will be parasitic upon distinctions we draw with respect to kinds of properties that are at least partially constitutive of kinds of facts. So if we have a clear definition of a physical property, we can define a physical fact in terms of the exemplification of that physical property. And if we have a clear definition of a mental property, we can define a mental fact in terms of the exemplification of that sort of property. 40

Distinctions: versions of physicalism and dualism

2.4

ev e n t p h y s i c a l i s m / d u a l i s m

At least some philosophers will want to distinguish events from facts (see, for example, Davidson 1993). One can get a feel for what moves such philosophers by thinking of some familiar examples. There does seem to be a relatively clear sense in which, according to the myth, the fact that Oedipus killed his father is a different fact from the fact that Oedipus killed the king of Thebes. The former includes as a constituent the property of being a father – the latter doesn’t. The latter includes as a constituent the property of being a king – the former doesn’t. Still, even if one buys into this claim that the world contains at least two distinct facts (facts which make true two quite distinct thoughts/propositions), there is the nagging sense that only one thing happened. If we asked a cameraman to film both Oedipus’s killing his father and Oedipus’s killing the king of Thebes, we would, subsequently, be looking at only one motion picture. Events understood this way seem to be construed as something observable occupying spatial and temporal location. Even if Oedipus exemplifies the property of having had a father, it is not as if he carries that property along with him wherever he goes, at least in a sense that implies that we can observe his exemplifying it. There was a huge literature on the problem of coming up with identity conditions for events.8 The problem is that there seem to be indefinitely many ways that gerundives can pick out what philosophers want to call the same event. Those same gerundives will pick out different facts insofar as the predicate expressions they contain pick out different properties. So consider the following gerundives: my waking up the neighbors last night; my playing music loudly last night; my playing the piano loudly last night; my playing the Star Spangled Banner loudly last night. There are at least 8

See, for example, many of the essays in Casati and Varzi (1996).

41

Knowledge, Thought, and the Case for Dualism

some event theorists who would argue that all four gerundives might denote one and the same event. Again, it seems to me obvious that in one clear sense the following all pick out different facts: that I woke up my neighbors; that I played music loudly last night; that I played the piano loudly last night; that I played the Star Spangled Banner loudly last night. Certainly the meaning of the noun clauses is different, and the assertoric form of the noun clauses is made true by the exemplification of different properties (if we assume a correspondence conception of truth). But how many events took place? If we use our film metaphor, I suppose one might suppose that at least three of the gerundives pick out the same event – I’m not going to need to film more than one observable spatio-temporal process to film my playing the piano, my playing music, and my playing the Star Spangled Banner. I suppose we might need to film the neighbors’ awakening to record the first event denoted. There is, I think, a reason that the attempt to come up with clear criteria for events eventually died a slow death. It seems in the end utterly hopeless. The very best we can do is make a distinction between properties whose exemplification is observable and those whose exemplification is not and describe events as those consisting of the exemplification of observable properties. But this distinction between observable and unobservable properties, particularly in the domain of physical processes is itself hopeless. As we shall see later on, there are plausible views on which the exemplification of all physical properties is theory-laden – one can never directly observe the exemplification of a physical property.

2.5

pr o p o s i t i o n p h y s i c a l i s m / d u a l i s m

From the time Smart first responded to knowledge arguments, there has been a significant shift in how some philosophers think of propositions. Smart, you will recall, was quick to 42

Distinctions: versions of physicalism and dualism

give up the claim that propositions describing brain states are identical to propositions describing paradigmatic mental states (like pain and after-images). And that was because he probably held the following views that were once considered relatively uncontroversial: (1) (2)

(3)

Propositions are the meanings of sentences or they are sentences-with-a-meaning. We have relatively unproblematic access to the meanings of our sentences and we can discover solely by reflecting on the meanings of our sentences when two sentences share the same meaning. The meaning of, for example, “I am in pain” is obviously different from the meaning of “C-fibers are firing.” Proof: Someone who understands both sentences perfectly well could never discover through reflection that they have the same meaning.

Therefore, (4)

The proposition that I am in pain is distinct from the proposition that my c-fibers are firing.

Since Kripke’s highly influential Naming and Necessity (1980), more and more philosophers have taken up the slogan that meaning is reference (for names and common nouns) or extension (for predicate expressions). Furthermore, only empirical investigation will reveal the meaning/reference/extension of expressions in a natural language. If one combines that view with the view that the meaning of a sentence is the proposition expressed by a sentence, it would seem to follow that it is a matter for empirical investigation what proposition I succeed in expressing using a given sentence. Consequently, I might, in fact, express the same proposition by “My c-fibers are firing” that I express with “I am in pain,” despite the fact that I could not discover that truth by mulling over, from the armchair, the semantic rules I follow in using language. 43

Knowledge, Thought, and the Case for Dualism

As we shall see in Chapter 5, the view of meaning and reference defended by Kripke and his followers is seriously problematic. But given its popularity we have a new entry in the arena of physicalist/dualist disputes. If we can define clearly the expressions “physical proposition” and “mental proposition” we can argue about whether all mental propositions might turn out to be identical to physical propositions. Of course, I haven’t said anything much about how one might construe a proposition, and disputes about the nature of propositions will carry over into corresponding disputes about whether or not we can identify propositions of one class with propositions of another.

2.6

s e n t e n c e ph y s i c a l i s m / d u a l i s m

I include this category only for the sake of completeness. Just as one might have a principled way to define physical-world and mental-world propositions, so also one might have a principled way of defining physical-world sentences or assertions, and mental-world sentences or assertions. And if one does, one could, I suppose, generate a question about whether all mental-world sentences or assertions are, or are not, identical with physical-world assertions. In the end, though, this distinction should collapse into the preceding controversy about propositions. If sentences are construed literally as marks or sounds, there is no interesting controversy concerning how many different kinds of sentences there are – there are as many as there are kinds of shapes and sounds. If there is an interesting controversy about kinds of sentences relating to the nature of the mental and physical it will be a controversy about the content or meaning of sentences. And that is a controversy about propositions. 44

Distinctions: versions of physicalism and dualism

2.7

d e fi n i n g “ m e n t a l ” a n d “ p h y s i c a l ”

We have identified a number of categories about which physicalists and dualists might want to argue. But even if we settle on a category of things about which physicalists and dualists disagree, we still won’t have defined the critical terms of the debate. Whether we are talking about substances, particulars, properties, facts, events, propositions, or assertions, we can’t start arguing about whether we should identify the mental with the physical unless we have clearly defined the respective terms. It is tempting to think that most roads lead back to properties. In our brief discussion of particulars and substances it is hard to see what could distinguish one substance or particular from another except the kind of properties that substance or particular exemplify. And as I understand facts, facts just are the exemplification of properties. Even the elusive event will surely have identity conditions that are at least partially defined by properties “involved” in the event’s occurring. The category of thought or proposition is more problematic, but on at least some conceptions of truth one can plausibly define distinctions between kinds of thoughts or propositions by defining distinctions between the potential truth-makers for those thoughts or propositions. On a correspondence theory of truth, the truth-maker for a given proposition is a fact, and we have already decided that we could identify a fact in terms of property exemplification. So let’s see what we can do by way of defining clearly the distinction between mental and physical properties. The task will prove to be much more difficult than one might have initially supposed. We must walk a rather narrow path between definitions of “physical” and “mental” that guarantee that the physical and the mental are distinct, and definitions that make the controversy uninteresting in the sense that both 45

Knowledge, Thought, and the Case for Dualism

traditional dualists and physicalists will be happy to allow that the properties under dispute fall into the categories of both the mental and the physical.

2.8

ph y s i c a l s c i e n c e s a n d p h y s i c a l properties

Naturalism is all the rage these days. In the next chapter we’ll have occasion to question the philosophical preoccupation with science when it comes to fundamental ontological commitments. In doing so, we’ll have reason to question what the distinction is even supposed to be between a naturalistic ontology and one that is not. But at least some philosophers seem to identify a naturalistic approach to philosophy with some sort of deference to science when it comes to what kinds of things exist. And it seems to me that some philosophers who are concerned to argue that all properties are physical seem to understand that thesis in terms of a claim about the predicate expressions employed by fundamental science. On one approach, for example, one might simply define physical properties as those that are picked out by predicate expressions employed in the statement of certain scientific theories. And physicalism becomes the thesis that all genuine properties are picked out by those predicate expressions. But this statement of the view is obviously far too crude to allow any useful definition of physical properties. For one thing, one would need to know much more about which sciences we are talking. So for example, psychology, cognitive science, and neurology obviously employ in the statement of their theories expressions like “is in pain,” “fears that P,” or “imagines that Q.” These expressions occur in the statement of their scientific discoveries because one of the primary interests of scientists in studying the brain is to relate brain activity to those states with which we are already familiar and about which we already know a great 46

Distinctions: versions of physicalism and dualism

deal. The cognitive scientist and the neurophysiologist want to know what is happening in the brain when one feels pain, is petrified, or is daydreaming. It might be possible to study the brain with the more pristine goal of categorizing as many different types of processes occurring there as one can discover, but we all know that the brain processes of most interest in many contexts are those that can be correlated with mental activity. One might respond that while a science may be concerned with a neural state only insofar as it is presumed to correspond to (correlate with) states with which we are all pretheoretically familiar, it doesn’t follow that the science will employ as a predicate the “folk psychological” predicate expressions of ordinary language. But the science inevitably will contain so-called “bridge” laws that relate neural activity to such states as pain and sensation, or it will be incomplete. It won’t tell us what we are particularly interested in knowing. So if one defines physical properties as those picked out by expressions occurring in the physical sciences and if one includes neurophysiology or cognitive science in the physical sciences (complete with their attempts to connect neural states with ordinarily described mental states), then expressions like “is in pain” will, trivially, pick out physical properties. These predicate expressions may, of course, pick out physical properties in some more robust sense of “physical properties,” but surely we want to characterize the physical in such a way that we don’t trivialize the thesis that all properties are physical. A property dualist will readily agree that being in pain is a physical property if physical properties include all those properties referred to by predicate expressions that appear in a complete, informative statement of scientific theses. While paradigmatic expressions that occur in what is sometimes called “folk psychology” also occur in the statements of many sciences, and, indeed, in the theories of all interesting sciences of the mind, they arguably don’t occur, or at least need 47

Knowledge, Thought, and the Case for Dualism

not occur, in the exposition of fundamental physics. So we might have a more philosophically interesting characterization of physical properties if we define them as those properties referred to by the predicate expressions contained in the exposition of the theories of fundamental physics. To assess such a claim one would presumably want to know more about what fundamental physics is supposed to do. On one interpretation, physics has the task of describing the ultimate constitution of physical objects and to discover the ultimate laws governing the behavior of physical phenomena. But, of course, these characterizations of physics already presuppose an understanding of “physical.” We aren’t saying anything very illuminating when it comes to defining the terms of a physicalist/ dualist debate. We could drop the reference to physical reality, and characterize physics more broadly as a science concerned with describing the ultimate constituents of reality. But property dualists will again now view the claim that all properties are physical as harmless and trivial. If the property dualist is right and there are properties other than the physicalist’s preferred physical properties, then a correct inventory of reality will include the claim that such properties are exemplified. If current physics doesn’t recognize their exemplification, then current physics is incomplete as a description of the world. This last comment should remind us that we have another decision to make if we are trying to characterize the physical in terms of what is recognized by some branch of science. As Hempel (1969) pointed out, we need to decide whether the science we are talking about is the science that exists today, or some utopian science of the future that has managed to get everything right. J. L. Dowell (2006, p. 1) describes the dilemma this way: Briefly, that dilemma begins with the supposition that there are two strategies for defining the physical, one that takes the physical to be

48

Distinctions: versions of physicalism and dualism the posits of current physics, the other which takes it to be the posits of ideal physics. The dilemma arises because if physicalism is the thesis that there’s nothing over and above the posits of current physics, then physicalism is very likely false, but if it is the thesis that there’s nothing over and above the posits of ideal physics, then physicalism lacks determinate content. A satisfactory formulation of physicalism must avoid both horns of this dilemma.

Hempel’s dilemma (as it is known) points out that any actual science is almost certain to embrace at least some false hypotheses as to what exists and, in any event, to be woefully incomplete.9 The idea that our ontological commitments should be decided by current scientific fashion is so odd as to be almost ludicrous. But again if we try to define the physical in terms of a science that doesn’t now exist and almost certainly never will exist, our definition will hardly be useful.10

2.9

ep i s t e m o l o g i c a l c h a r a c t e r i z a t i o n s o f t h e me nt al / p h y s i ca l

There are other classical ways of trying to define one term of the debate concerning the relationship between the mental and the physical. The mental is sometimes identified in terms of the special nature of the access we have to it. So it seems at least initially promising to suggest that a property is a mental property just in case it is the kind of property whose exemplification can be introspectively accessed. Of course, anyone who makes such a suggestion would need to say more about what introspective access is. My own view is that introspective access is best understood in terms of the kind of direct awareness about which we will be talking more in Chapter 3. I have introspective access to my pain states just insofar as I can get myself a 9 10

For an attempt to respond to Hempel’s dilemma, see Melnyk (1997, 2003). For further discussion of this and related issues see Crane and Mellor (1990)

49

Knowledge, Thought, and the Case for Dualism

non-inferentially justified belief that I am in pain through direct awareness of my being in pain (along with direct awareness of my thought that I am in pain corresponding to the pain). On this view, then, a property is a mental property just insofar as at least one person can become directly aware of its exemplification. Any specific account we give of introspection will immediately become controversial for it will invoke implicitly a controversial metaepistemology. There are more than just a few philosophers who think that the concept of direct acquaintance upon which I rely in characterizing non-inferential justification is a chimera. Furthermore, it is no longer the received view that we can only be directly aware of properties that the dualist would regard as mental. Direct realism is making a bit of a comeback, and if the direct realist were prepared to couch direct realism in terms of a view about what one can be directly aware of (many would not), that direct realist will find odd the suggestion that surface properties of physical objects are mental properties. There is also the problem that one should not rule out, a priori, the possibility that there might be mental properties of which we are unaware. While I used to think that the concept of occurrent, but unconscious desire, fear, love, hate, and so on is unintelligible, I no longer see what is so problematic about their existence. It seems to me that one can even countenance the possibility that one is in pain while losing the capacity to become aware of pain. Williamson (2000) has an argument that it is not the mark of the mental that it is luminous. A state is luminous just insofar as anyone who is in the state could, if possessing appropriate concepts, come to know directly and without inference that one is in the state. A weaker concept of luminosity would insist only that anyone in a luminous state could, if possessing appropriate concepts, come to have a noninferentially justified belief that one is in such a state. A weaker 50

Distinctions: versions of physicalism and dualism

concept still would insist that a state is luminous if it is the kind of state such that someone in it could, in principle, come to be non-inferentially justified in believing that one is in such a state. Again, these days most epistemologists think that one can possess non-inferential justification for believing a host of propositions describing the paradigmatically physical world, including propositions describing the physical character of one’s own body – the color and shape of one hands, the texture of one’s skin, and so on. I think that they are mistaken, and as I shall argue in the next chapter, one simply can’t evaluate the debate over the knowledge argument and remain neutral on these sorts of controversial issues. However much disagreement there is on how to analyze introspective access, it does seem to be almost a datum that many paradigmatically mental states are such that the person in such a state has a privileged access to the truth that he or she is in such a state. As I imagine the face of a good friend, I know what is going on in my mind in a way that is far more direct than anyone else could possibly know it. When I feel a sharp pain in my elbow, I typically know more directly and better than anyone else could possibly know that I feel that pain.11 And it is tempting to contrast that sort of knowledge with the kind of knowledge I have of truths about tables, chairs, bodies, 11

One might ask for further elucidation of the concept of “knowing better . . .” Williamson (2000) has argued that the epistemic probability of that which is known is always 1. But so much the worse for Williamson, at least if we allow that our ordinary uses of “know” are more or less accurate. I know that I was born in Toronto, but my knowledge that I know now that I exist is surely better than that knowledge of my birthplace. I would argue that it is better in the sense that the justification I have for the former is much stronger than the justification I have for the latter. But if one is determined to go with Williamson, one will need to find some other intuitively plausible notion of being stronger or more secure. (Williamson, for example, suggests that the “better” knowledge is belief “father away” from the world in which it would fail to be safe.)

51

Knowledge, Thought, and the Case for Dualism

brains, and their micro constituents. In fact, I think that all this is correct, and that we can distinguish the mental and the physical in this way. But I certainly wouldn’t be happy with this way of making the distinction if I were a physicalist. I would feel that I was being set up for a dualist’s argument. In short, if I were a physicalist who got talked into the view that we are sometimes directly aware of properties like being in pain, or sensing red (or sensing redly, as the adverbialist would prefer), I would probably prepare myself to claim that we are sometimes directly aware of physical properties – the very physical properties with which I was planning to identify being in pain, or sensing red. And as we shall see that is precisely what one sort of physicalist inspired by a view that might have been held once by Russell (1948, p. 231) and was definitely held by Maxwell (1978), will do. When one identifies pain with some paradigmatically physical property, one will simultaneously reject the idea that one should contrast the mental and the physical in terms of the critical concept of awareness. If one is seduced into agreeing that the exemplification of mental properties can be an object of direct awareness while the exemplification of physical properties cannot be an object of direct awareness, and one accepts this as the right way to analyze the meaning of mental and physical respectively, then, as we shall see, the property dualist will win the day.

2.10

i n t e n t i o n a l i t y a s t h e ma r k o f t h e me n t a l

Another historically prominent attempt to define the mental appeals to the concept of intentionality. Mental states, some would argue, are essentially intentional. One will still need some independent way of characterizing the physical lest the property dualist win the day through definition. No physicalist 52

Distinctions: versions of physicalism and dualism

(with the possible exception of a hard-core eliminativist materialist) will allow one to define the physical as that which is not intentional. And, almost everyone will allow that in at least some senses the exemplifications of some paradigmatic physical properties are representational states. Following Brentano (1874) one might try to characterize the intentional in terms of states that are directed at an object, but where the object might not exist. One need only state clearly this characterization of the intentional to start worrying about its implications. We seem to be on the verge of taking a dangerous journey into land in which roam Meinongian subsistent but non-existent objects. And that is a place most of us don’t want to go. It is not hard to see how one might get the idea that the intentional should be characterized as essentially relational – as requiring a so-called act/object analysis. One can’t describe beliefs, desires, fears, imaginings, and the like without using transitive verbs that must be completed with an object – either a noun, a gerundive, or a noun clause. One can’t just believe, one must believe that such and such is the case. One can’t just fear, one must fear something or that something is the case. One can’t just desire, one must desire something or that something be the case. But, of course, one can believe that there are unicorns when there are no unicorns, one can fear that there is a hell when there is no hell, and one can desire that one be unbelievably rich when, alas, that will never come to be. One can use a linguistic counterpart to Brentano’s criterion for intentionality, then, by noting that descriptions of intentional states (like belief, fear, and desire) can be true, despite the fact that the object terms completing the transitive verbs describing the respective states don’t succeed in denoting (either an object or a fact). This fact suggests strongly to me that one should treat the surface grammar of such sentences as highly misleading when it comes to characterizing the ontological structure of the state described. I have argued elsewhere 53

Knowledge, Thought, and the Case for Dualism

(2002) that one should endorse a so-called adverbial theory of both sensory states and intentional states – one should treat both my being in a sensory state and my being in an intentional state in terms of my exemplifying certain non-relational properties. In the case of intentional states these properties have the capacity to correspond to the world. The adverbial theory of sensory states was designed to dissolve certain ontological and epistemological puzzles. Consider, for example, the state of affairs that is John’s feeling pain. The surface grammar of the sentence “John feels pain” suggests that the sentence is describing a relational state of affairs – John’s standing in the relation of feeling to pain. Similarly, when we say of John that he is hallucinating a pink rat, the sentence appears to have a subject term, a transitive verb, and an object term (the term picking out a pink rat). If we take seriously that feeling a pain or hallucinating a pink rat is a genuine relational state of affairs, we need to address head on the question of whether pains can exist unfelt or the pink rats hallucinated can exist even after the hallucination ends. Even those philosophers who insist on a relational account of feeling pain or hallucinating rats (as most of the sense data theorists did) typically can’t quite bring themselves to allow for even the intelligibility of unfelt pains or pink rats surviving the intrusive temporary hallucinatory observation. The solution, the adverbial theorist insists, is to regard the surface grammar of our sentences describing such states to be misleading. To feel pain isn’t to stand in a relation of feeling to something else, the pain. To feel pain is to feel a certain way. To visually hallucinate a pink rat is (in part) to visually experience a certain way. One can make the view clear by relying on analogies. As John dances a jig one might, if one were a bit thick, wonder where the jig is that John is dancing. One might ask whether his jig might survive the end of the dance. But, of course, it takes but a little reflection to realize that to dance a jig is just to dance in a 54

Distinctions: versions of physicalism and dualism

certain way. The jig ends when that way of dancing ends. The pain ends when that way of feeling ends. And the hallucinatory object is gone when that way of sensing ends.12 Descriptions of belief, of course, also involve subject terms, transitive verbs, and (typically) noun clauses that are the objects of those transitive verbs. The noun clause completing the description of a belief does, of course, tell us what belief the person is supposed to have, but, the adverbial theorist argues, only in the sense in which the noun in “Jones danced a jig” tells you what kind of dancing Jones is doing. There is no jig upon which dancing is performed – there is just a way of dancing. There is no non-obtaining state of affairs, or non-existent fact, to which false belief is related – there is just a belief with a certain intrinsic nature. Again, when a belief is true, there is a relation between the belief and a representation-independent fact, but we are commenting here on the correct analysis of belief, not the correct analysis of true belief. The adverbial theory of belief (like the adverbial theory of sensory states) is hardly uncontroversial. The adverbial theorist will need to paraphrase heavily to make distinctions that recover apparently valid inferences that act/object theorists seem to deal 12

It is important to be very careful here. Much confusion can be caused by failing to make subtle distinctions. I’ll talk more about this later, but here I want to be clear that in embracing an adverbial account of feeling pain, for example, I’m not suggesting that there is no distinction to be made between being in pain (feeling pain) and being aware of that pain. And both being in pain (feeling pain) and being aware of the pain should also be distinguished from believing that one is in pain or knowing that one is in pain. As I will make clear later, I do not embrace an adverbial account of direct awareness (acquaintance). Acquaintance is a genuine relation and it requires relata. So we can’t be in pain without feeling a certain way, for being in pain just is a kind of feeling. But we can be in pain without noticing it, and without forming a belief that one is in pain. These distinctions allow us to raise all sorts of interesting questions about different sorts of animals. Perhaps some animals feel pain, but aren’t aware of the pain that they feel. There are almost certainly animals that feel pain, perhaps are aware of that pain, but don’t form any beliefs about pain because they lack the conceptual resources necessary to represent the world.

55

Knowledge, Thought, and the Case for Dualism

with much more straightforwardly. Some of these objections will parallel Jackson’s (1977) famous arguments against the adverbial theorist’s attempt to replace the sense-datum language with an adverbial language. So, for example, it seems in English to follow that if Sam believes that there are horses then there is something that Sam believes. The adverbial theorist replaces that inference with the following: If Sam believes that there are horses, then there is some way that Sam believes. It isn’t easy either for the adverbial theorist to translate talk about complex beliefs, e.g. Sam’s belief that there are horses and that there are goats. If this is one belief, we certainly can’t capture its complex content without further artificial use of invented hyphenated expressions – Sam’s believing in the that-there-are-horses way and the that-there-are goats way. But while there is a huge price to pay in terms of engaging in linguistic gymnastics, the ontological pay-off of avoiding unwanted “objects” in the case of false beliefs, unfulfilled desires, unwarranted fears, and the like is also, I believe, huge. Our primary concern here is not with the correct view of intentional states. That issue is going to come up later, because, as we noted earlier, some physicalists, and even many property dualists, want to contrast the so-called hard problem of consciousness with so-called easy problems of consciousness. Giving a physicalist account of intentional states is supposed by some philosophers to be relatively easy. The worry is that there are mental states that resist characterization as intentional, and that, consequently, resist the “easy” physicalist reductions that succeed with respect to states like belief, desire, and fear. As we shall see, Jackson’s repudiation of his influential argument is itself tied to a change of heart as to whether the states associated with qualia are intrinsically representational. In any event, I shall argue later that it is not at all obvious that intentional states are any more receptive to physicalist reduction than any other paradigmatically mental state. 56

Distinctions: versions of physicalism and dualism

Should we assimilate the mental to the intentional? I am not at all comfortable doing so, but I don’t have anything very original to add to the debate. I explain my reservations relying on the same sorts of examples to which many others have appealed. It doesn’t seem to me that pain, general feelings of anxiety, episodic physical pleasure, feelings of euphoria, and the like, are intentional/representational states, though they are paradigmatically mental. And although it is in danger of becoming the received view, I still don’t think that sensations are intentional states – states that intrinsically represent (accurately or inaccurately) the world. To say of X that it intrinsically represents Y is to say that X represents Y just in virtue of its intrinsic character (its non-relational character).13 We should distinguish this sort of representation from both conventional representation and even causal signification. My roommate and I can agree to let a tie hanging on the outside doorknob indicate that the room is occupied. Tracks in the snow might indicate the presence of a rabbit (by being an effect of the rabbit’s movement reliably correlated with rabbits). But neither the tie nor the marks in the snow represent just in virtue of what they are intrinsically – their shapes, color, and so on. When Mary leaves the black and white room and has a visual experience of red for the first time, I believe that it is a mistake to think that Mary enters into a brand new intentional state, a state that represents for her a hitherto unthought of feature of the world. Admittedly, in us the sensation will come with various intentional states. Most of the sensations we experience have associated with them a host of representational states, though many never rise to anything more than expectations of various 13

And again, X can misrepresent. It is perfectly natural to say that S is afraid of ghosts, and on the theory we are considering here it is in virtue of the intrinsic nature of S’s state that it is a fear of ghosts. But there are no ghosts and the representation state is identical with the exemplification of a non-relational property, albeit one with the capacity to correspond.

57

Knowledge, Thought, and the Case for Dualism

sorts, the presence of which are typically noticed only when the expectations are frustrated. These are issues to which we must return later, when we evaluate Jackson’s own repudiation of his knowledge argument for property dualism. If we could characterize the mental as the intrinsically representational, would we have furthered the terms of the debate between physicalists and dualists? Dualists might be happy for almost all of them will think that nothing has the property of being intrinsically representative (what Addis, 1989, calls being a natural sign) other than the very properties that they insist are distinct from physical properties. Put another way, they might be happy to characterize the physical as that which lacks this property. But, of course, physicalists will not typically accept the invitation to characterize the physical this way. Physicalists think that some physical states just are beliefs, desires, fears, and the like, and thus that some physical states are representations. Typically they will reject the view that a given physical state represents in virtue of its intrinsic nature. Crude causal theorists, for example, will understand the capacity of a certain physical state to represent as a function of its causal history (or the causal history of states of that kind). But other physicalists (Searle, 1983, for example) seem happy to embrace something like the idea that complex brain states represent in virtue of their intrinsic character. So we still haven’t found a non-question begging way to frame the terms of the debate. Physicalists are not going be cornered into claiming that nothing physical is fundamentally intentional. And dualists are not going to think that the states that the physicalists identify as intentional are genuinely mental.

2.11

giving up

While it might seem a bit anticlimactic, perhaps we should abandon any attempt to define abstractly the terms “physical” 58

Distinctions: versions of physicalism and dualism

and “mental” in an effort to characterize the debate between property dualists and property physicalists. The astute reader probably noticed that sprinkled through the preceding discussion there were occasional references to paradigmatically physical states, and paradigmatically mental states. We can probably say what separates property dualists and property physicalists by focusing on particular disputes. So, for example, the property dualist thinks that there is such a thing as being in pain. Furthermore, the property dualist denies that we can successfully identify being in pain with being in a brain state (however you care to pick out that brain state), being in a functional state (where the input and output used in the characterization of the functional state involve states of affairs that can be characterized in paradigmatically physical language), being in a disposition to engage in some sort of bodily behavior, and so on. Paradigmatic property dualists will probably accept our epistemological characterization of the mental. Whenever the property dualists find themselves directly aware of the intrinsic phenomenological character of some state, they will also reject identification of that character with the character of anything we discover through the five senses. To be sure, the property physicalist might well insist that there are properties that we can discover both introspectively and through, for example, visual experience. But again, the property dualist will deny these identity claims. We can’t, without begging the question, saddle the physicalist with the claim that physical properties cannot be discovered introspectively. We can, however, contemplate that with which we are phenomenologically acquainted, and consider all of the proposed reductions (or eliminations) advanced by self-proclaimed physicalists. When we reject on principled grounds those reductions, we will proclaim property dualism vindicated. Or if we don’t want to use the expression “property dualism” we can settle for a more modest claim. After the physicalists have 59

Knowledge, Thought, and the Case for Dualism

identified all of the kinds of properties with which they are comfortable, we property dualists will insist that they have an incomplete description of the kinds of properties that exist.14

2.12

varieties of physicalism

I have suggested, in effect, that the property dualist is best understood as a philosopher who rejects certain paradigmatic reductions. There are, the dualist is convinced, such properties as being in pain, having visual, tactile, auditory, gustatory, olfactory, or kinesthetic sensations of various sorts, believing, fearing, desiring, feeling euphoric, feeling uneasy. Furthermore, the dualist argues, self-proclaimed physicalists offer mistaken analyses of what these properties are. But, of course, there are many different forms of physicalism that we should distinguish. Before doing so, however, I should, perhaps, make clear that I am only addressing versions of physicalism that attempt to reduce paradigmatically mental properties to paradigmatically physical properties. The “physicalist” who claims only that mental states supervene on 14

This proposal bears at least a superficial resemblance to Barbara Montero’s (1999) suggestion that we replace controversies defined in terms of a physical/mental dichotomy with a controversy about whether the mental can be identified with the non-mental. On the face of it, such a suggestion leaves the physicalist in the unhappy position of having to argue that the mental is actually not mental (nonmental). Montero’s proposal reminds one a bit of Frankena’s complaint about the naturalistic fallacy. Moore accuses his opponent of identifying the non-natural with the natural – Frankena points out, quite correctly, that the naturalist will prefer to state the naturalist’s thesis as the claim that moral properties just are natural properties. Similarly, if I were a physicalist I would want my claim to be understood as the claim that mental properties just are physical properties (but then we are back to the task of trying to define “physical”). But Montero is, I believe, just emphasizing what we have underscored – we already have a perfectly good grasp of mental states. The real question is not whether we can get rid of them somehow, but whether we can successfully reduce talk about mental states to talk about some other sort of phenomena (as the phenomenalist, for example, sought to reduce talk about physical objects to talk about actual and possible sense data).

60

Distinctions: versions of physicalism and dualism

physical states, or who claims that there are necessary connections between what happens in the physical world and what happens in the mind, is not denying a dualism of properties. An analogy might be helpful. In ethics, Moore was famous for embracing non-naturalism – the view that being intrinsically good is a non-natural property that should not be confused with any of the world’s natural properties. Nevertheless, Moore would be the first to admit that being good strongly supervenes on natural properties. Moore took it to be a necessary truth (knowable a priori) that two things exemplifying precisely the same natural properties would have precisely the same value. It seems to me, however, that it would be horribly misleading to characterize Moore as embracing a form of naturalism because he was committed to these connections between the natural and the non-natural. For precisely the same reason, it simply muddles the debate to include among the physicalists those who are committed to irreducible mental properties. It is even more misleading to characterize as physicalists those who take the mental to supervene lawfully on the physical. The cynic might suggest that the term “supervenience” entered our physical vocabulary when we wanted to talk about a relation between two kinds of things but we didn’t want to be pinned down on just what that relation is. It is, I suppose, harmless to introduce the term, but if one is going to advance a significant claim about supervenience, one simply must make distinctions among kinds of supervenience. One can at the very least talk of analytic, metaphysical, and lawful supervenience. So a property G analytically supervenes on a property F only if it is analytic than anything that is F is G (being a bachelor analytically supervenes on being an unmarried man). G metaphysically supervenes on F only if it is a synthetic necessary truth that anything that is F is G (a’s being colored might supervene in this sense on a’s 61

Knowledge, Thought, and the Case for Dualism

being blue).15 And we might say that G lawfully supervenes on F only if it is a law of nature that anything that is F is G (a’s being metal and expanding might supervene in this sense on a’s being heated). Now the dualism I am examining in this book is absolutely neutral on questions of lawful connection between the exemplification of paradigmatic physical properties and paradigmatic mental properties. Almost all contemporary dualists think that there are causal connections between physical states of the body and the exemplification of mental states. It is not some sort of cosmic coincidence that when I smash my toe into a rock I feel intense pain. And most dualists think that there are causal connections between mental states and physical states – again the intense pain surely has some non-accidental connection to the grimace that follows. But one’s commitment to dualism, particularly a dualism of properties, is quite independent of one’s views about contingent lawful connections. That is not to say that one’s motivations for endorsing physicalism might not have something to do with concern about how to understand causal connections between what the dualist takes to be such different things as states involving mental properties and states involving physical properties. We’ll briefly examine such concerns in Chapter 7. Here, I’m merely identifying the kind of dualism in which I am interested – a dualism that does not deny all sorts of lawful connection between the physical and the mental (and vice versa). The dualist we are addressing in the book is also not necessarily committed to denying stronger than lawful metaphysical connections between the exemplification of physical properties and the exemplification of mental properties, though here most classical dualists will probably either deny the connections or question their intelligibility. I personally think that there 15

Though there are no uncontroversial cases of supervenience of this sort.

62

Distinctions: versions of physicalism and dualism

are synthetic necessary connections and that one can make sense of one property exemplification synthetically entailing another different property exemplification. I think, for example, that being colored does always supervene on being some particular color; being rectangular does supervene on being some particular rectangular shape, and so on. Philosophers have tried to reduce this sort of necessity to analyticity, but nothing seems to work. It would be great if we could analyze being colored into the disjunctive property being red or orange or blue or brown or black or fuchsia or taupe or . . . , but we know that there are too many determinate color properties to complete the analysis in such a way that we could plausibly claim that people have in mind the disjunctive property when they think of what it is to be colored. Moreover, we still have synthetic necessary truths of the form: whatever is red is not blue, and again it is not clear how we could ever construe such a truth as analytic. You could try to define being red as being not blue and not yellow and not . . . , but when you get around to defining being blue as being not red and not yellow and not . . . , you realize that you are stuck in a circle. So like it or not, I think there are non-analytic necessary truths. But none of the paradigms of such truths seem available to use as a model to understand a synthetic necessary connection between the exemplification of physical properties and the exemplification of mental properties. Even if I were wrong about this, however, and we find a way in which mental properties metaphysically supervene on different physical properties, the resulting view is still a form of dualism. Remember the analogy with Moore who rightly characterized his ethical view as rejecting naturalism even though he was quick to acknowledge that ethical properties always metaphysically supervene on natural properties. So with that all said, let us return to our brief survey of physicalists who are committed to the view that the world has 63

Knowledge, Thought, and the Case for Dualism

no room for the property dualist’s irreducible paradigmatically mental properties.

2.13

logical behaviorism

The logical behaviorist has all but disappeared. And I have received feedback on an earlier draft of this manuscript wondering why I take as much time as I do discussing a view that has landed in the dustbin of philosophical history. I shall try to argue, however, that the much more respected functionalism that has replaced behaviorism, when interpreted properly, is vulnerable to precisely the same sorts of problems that behaviorism encountered. Unless we review carefully the problematic past of behaviorism, we are doomed to repeat the view’s mistakes in our pursuit of an equally problematic functionalism. In the heyday of logical positivism, many philosophers found attractive various formulations of the verifiability criterion of meaningfulness. The crude idea was that a given statement has cognitive meaning (the kind of meaning that insures a truth value for the statement) only insofar as it is verifiable. To understand the cognitive meaning of a statement is simply to understand what would verify it. The principle can be interpreted in many different ways, and most verificationists weakened the relevant notion of verification under the pressure of counterexamples. So one might start with the idea that a statement is verifiable, and hence empirically meaningful, only if it allows for the possibility of conclusive verification. One has conclusive justification for believing P when that justification eliminates the possibility of error. The possibility alluded to in the principle can itself be understood in many different ways. So when we talk about the possibility of verification, conclusive or otherwise, we could be talking about possibility that is logical, lawful, or causal relative to 64

Distinctions: versions of physicalism and dualism

circumstances.16 It makes a big difference. If cognitive meaning requires only the logical possibility of verification, and one thinks that one can make sense of a God-like being who is omniscient and outside of time, all bets are off concerning what will satisfy the criterion. Typically, however, the verificationist wanted to reject as meaningless just that sort of theism. With a narrower conception of possibility in place, there was, as I said, considerable pressure to weaken the relevant concept of verification to which the verificationist appeals. The assertion that there are no unicorns seemed to be perfectly meaningful even though it is hard to see how a finite being could accumulate enough evidence to preclude the conceivability of error. If one moves to verifiability or falsifiability as one’s criterion for meaningfulness, one need only disjoin the assertion that there are unicorns to the assertion that all metal expands when heated, to come up with a counterexample. At this point, the obvious move for the verificationist is to retreat to a weakened notion of the possibility of confirmation or disconfirmation as a necessary condition for cognitive significance. A claim admits of the possibility of confirmation or disconfirmation when it is possible for someone to acquire justification that counts in favor of or against the claim in question. Even an epistemic requirement that weak is, however, problematic. Geach (1957) asked us to consider the following sorts of claims: The number of people alive exactly 400,000 years ago was even and no one (past, present, and future) has any reason to believe that this is so. While decidedly odd, the statement seems to have a truth value and thus seems to be cognitively significant. Trivially, however, it is impossible 16

So it is logically and lawfully possible for me to dunk a basketball here on Earth’s surface, but, sadly, it is no longer possible relative to circumstances that include the wretched physical condition I am in right now. In ordinary discourse, when we talk about what people can or can’t do, we are almost always talking about that causal possibility relative to circumstances.

65

Knowledge, Thought, and the Case for Dualism

for anyone to confirm or disconfirm the statement. All is not lost, however, for the verificationist can (and obviously should) retreat to the more modest claim that every cognitively significant atomic proposition17 is confirmable or disconfirmable (in the weak sense). One can combine these meaningful atomic statements in such a way that we create epistemic “dissonance,” but the complex statement will remain meaningful as long as its constituents are meaningful.18 There is considerable irony in the fact that verificationists moved in radically different directions in their zeal to ensure that paradigmatically meaningful statements satisfy their epistemic criteria for being meaningful.19 One group was convinced that the foundations of empirical knowledge and justification consist in knowledge of our subjective, fleeting experience. Faced with the problem of moving from that limited foundation to the rest of what we think we can know or justifiably believe, they found attractive attempts to reduce all meaningful discourse to complex talk of experience. So the 17

18

19

Where minimally, an atomic proposition is one that is not truth-functionally complex. That’s not enough, for we will also want to view as complex subjunctive conditionals, for example – conditionals that have antecedents and consequents but whose truth value doesn’t seem to be a straightforward function of the truth values of those antecedents and consequents (translated into the indicative mood). The move suggested here is obviously reminiscent of the move quickly made by radical empiricists who wanted to assert that all ideas have their origin in experience. A moment’s reflection requires one to give up the thesis as one realizes that one has the capacity to “build” out of simple ideas complex ideas that correspond to nothing we have ever experienced. But even if unicorns never roamed the land, and mermaids never swam the seas, we can insist that the simple ideas out of which such complex ideas are built correspond always to simple elements of experience. We’ll be examining content externalism later on in the book, and there again the content externalist faced with obvious facts about misrepresentation will almost certainly need to borrow a leaf from the book of the radical empiricist. As we shall see, the direction in which one moves here has significant implications for what conclusions one thinks one can derive from the knowledge argument vis-àvis finding room for the exemplification of mental properties by physical phenomena.

66

Distinctions: versions of physicalism and dualism

phenomenalist, for example, suggested that we can literally translate all talk about ordinary physical objects and their properties into complex talk about the kinds of sensations a subject would have were that subject to have other sorts of sensations.20 If effective, the translations might explain how claims about the physical world could be confirmed or disconfirmed even if all we have direct access to are correlations among sensations. A causal theory of objects is a close relative of phenomenalism. The view is one that we will discuss later. Crudely the causal theorist’s hope is to understand physical objects solely in terms of their causal role. Just as I understand Jack the Ripper as the person who committed certain atrocities at the end of the nineteenth century, so, perhaps, I can understand physical objects in terms of their potential role in affecting subjective experience. Unlike the phenomenalist, the causal theorist doesn’t try to eliminate indirect reference to physical objects distinct from experience. But like the phenomenalist the causal theorist argues that our understanding of physical objects is thoroughly parasitic upon our thought of subjective appearance. As we’ll see, this is the kind of reductionism that Russell and Maxwell saw as at least a hope for finding room for the mental in the world of the physical. While one sort of philosopher is trying desperately to reduce talk of the physical to talk of subjective appearance, another group was more worried about the problem of other minds. Taking for granted that we have relatively unproblematic knowledge of the physical world, they were determined to 20

One can find the hint of such a view in Berkeley (1713) when Philonous seems to suggest that we can understand how the world existed before humans by supposing that there were truths about what people would experience had they been around at the time of creation. Mill (1877) much more explicitly endorsed the view. Ayer (1952) gave a swashbuckling (though not all that sophisticated) defense of phenomenalism. And, to my way of thinking, C. I. Lewis (1946) gave the most careful statement and defense of the view.

67

Knowledge, Thought, and the Case for Dualism

secure cognitive meaning for claims about the psychological states of others, particularly when on one traditional view of such states they remain hidden from direct inspection by others. Just as the phenomenalist tried to ease epistemological problems by reducing the content of claims about the physical world to propositions describing relations between actual and possible sensations, so the behaviorist tried to make less epistemologically problematic propositions describing the psychological states of people by reducing the content of such propositions to complicated claims about physical behavior. The crudest form of behaviorism insists that to be in an occurrent mental state is just to behave in a certain way. So my being in pain might be identified with my grimacing, or my complaining (understood in purely behavioristic terms). No serious philosopher, however, would embrace a view that crude. It doesn’t take a genius to realize that people can be in pain and mask that pain with a Spartan exterior. Just as crude Berkeleyean idealism (physical objects are “bundles” of sensations) gave way to more sophisticated phenomenalism (propositions describing physical objects are analytically equivalent to complex subjunctive conditionals describing what sensations a subject would have were he to have others), so also crude logical behaviorism gave way to the more sophisticated (though still decidedly odd) view that propositions describing psychological states are analytically equivalent to propositions describing dispositions to behave in certain ways under certain conditions. And it is tempting to try to capture talk of dispositions using the same subjunctive conditionals that were the building blocks of phenomenalistic reductions. So on this view I might give no outward sign of being in pain, and I might even deny that I am in pain if asked, but, the argument goes, if I really were in pain I would be disposed to grimace absent desires to mislead, and I would be disposed to say that I was in pain if I were asked and I wanted to tell the truth. 68

Distinctions: versions of physicalism and dualism

It is an understatement to suggest that logical behaviorism faces serious objections. One can’t help but notice that when we tried to characterize the relevant dispositions associated with pain, it seemed natural to invoke considerations of what would happen if I wanted to act in certain ways. But the logical behaviorist is committed to translating talk of the problematic mental into talk of unproblematic physical behavior without remainder. So we need to eliminate from the analysans any explicit or implicit reference to unanalyzed mental states. And the problem is that one simply can’t come up with anything even remotely plausible while honoring such constraints. An exactly similar problem seemed to afflict phenomenalistic analyses of propositions describing the physical world. Phenomenalism sounded like a great idea to many philosophers until they were actually challenged to come up with the relevant translations. Mill (1874) said that propositions describing Calcutta were equivalent to propositions describing the sensations a subject would have were he by the banks of the Hoogly. But in his phenomenalistic translation, Mill isn’t allowed to set the stage for the relevant description of sensation by physically situating someone by the banks of the Hoogly. C. I. Lewis (1946) thought that the existence of the doorknob in front of me could be understood in terms of facts about the tactile sensations I would probably have were I to seem to see a doorknob and seem to be initiating a certain grasping motion. But Chisholm (1957) pointed out that you wouldn’t have any tactile sensations if your hand were anesthetized, and the phenomenalist can’t make reference to the absence of this physiological condition in a phenomenalistic translation.21

21

I’m not asserting here that there are no plausible replies to the objection. Lewis (1948) himself thought that he had protected himself from Chisholm’s objection by including a “probability” operator in the consequent of the subjunctive conditionals. Firth (1950) argued that a radical enough empiricism should simply

69

Knowledge, Thought, and the Case for Dualism

Some have thought that one can get around the above problems by endorsing a holistic account of meaning.22 To be sure, the analysis of pain will make reference to desire, and the analysis of desire will make reference to still other desires and beliefs, and the analysis of belief to still more desires and beliefs, but all this is fine as long as we recognize that meaning involves indefinitely complex networks of interlocking thoughts. I’m afraid, however, that I don’t even know what this is supposed to mean. In the end, it just seems like a preface to the admission that we aren’t going to get the promised behavioristic translation of psychological claims. You can’t have networks of logically and probabilistically interconnected thoughts unless the thoughts with their content exist ready to stand in these logical and probabilistic relations. Logical relations are just that – relations. Relations require relata and the relata of such relations are the contents of intentional states. And analyzing the nature of such states is the very problem we are trying to solve. Moreover, if ever there were a plausible example of an internal relation (a relation that necessarily obtains given the intrinsic character of the relata), it would be entailment. So when the holist thinks that the identity conditions of truth bearers is dependent on the logical relations truth bearers have to other propositions, they seem to have things completely backward. Any plausible account of entailment

22

admit that the meaning of physical object claims are in flux and vary from person to person and context to context. I have argued (1985) that the phenomenalist should try to protect the subjunctives with reference to normal conditions where normalcy can be defined statistically. Davidson (1984) and Quine (1951) are two of the most famous philosophers associated with meaning holism. See also Fodor and Lepore (1992) for papers containing extensive discussion of the notion. But the literature on this topic is voluminous. I should also stress that the view I am referring to here is only meaning holism. The idea that one cannot verify any single statement in isolation is a quite different view. Given earlier forms of verificationism, the epistemological thesis might have been linked to the thesis about meaning. But such forms of verificationism about meaning are problematic at best.

70

Distinctions: versions of physicalism and dualism

will seek to understand logical relations in terms of the intrinsic nature of the relata of such relations. Even if one could find subjunctive conditionals that meet the behaviorist’s constraints on an acceptable analysis and that might initially seem to correlate with states like being in pain, the behaviorist’s translations seem pathetically inadequate to the phenomenology of states like pain. When you strike my kneecap as hard as you can, I am directly aware of searing pain. It would literally strike one as a kind of joke to suggest that my awareness of the pain was my discovery that I was suddenly disposed to scream and grimace. In Chapter 1 we have already had occasion to note that the physicalist may well quickly retreat from a claim about meaning to a claim about the underlying truth-makers of claims. But the traditional logical behaviorists were committed to a translation program. They wanted to argue that talk of psychological states could be translated into sophisticated talk about behavior. And as we have just seen, the proposed translation is simply comical when it comes to certain sorts of paradigmatically mental states. Furthermore, although there are somewhat plausible examples of statements with different meanings that nevertheless have the same truth-makers (we’ll look at these later), none of the examples give us a plausible model for understanding the way in which a description of occurrent searing pain might have the same truth-maker as a subjunctive conditional describing behavior one would engage in under certain conditions. Admittedly, there are descriptions of more complex psychological states that might be more amenable to the idea that the claims in question involve ascriptions of dispositions. It is no accident that philosophers like Gilbert Ryle had relatively little to say about severe pain or, for that matter, visual, tactile, olfactory, gustatory, and auditory sensation.23 It probably 23

That might have been his way of implicitly acknowledging that there is a “harder” problem of qualia.

71

Knowledge, Thought, and the Case for Dualism

is highly implausible to claim that S’s loving R is S’s being in an episodic, occurrent state of loving (a sensation-like state). Loving doesn’t seem to have a sharp beginning and an end the way a visual experience has a beginning and an end. It is probably not much more plausible to suggest that loving involves at least a disposition to be in some single distinct intentional state of loving. If one truly loves another, one probably exemplifies a wide range of dispositions to be in various sorts of states, and perhaps even to behave in certain ways under certain conditions. But most of the relevant dispositions can only be characterized in terms of dispositions to be in various psychological states, and the dispositions to behave in certain ways that loving might involve can themselves only be specified relative to conditions characterized in the language of folk psychology. If I truly love Patti, then I must feel bad when she is hurt, and want to be with her (at least some of the time). I must miss her when she has been gone for awhile. I may need to behave in a somewhat kind way towards her, but only when I believe that the behavior in question is conducive to her well-being.

2.14

t h e mi n d / b r a i n i d e n t i t y th e o r y

A radically different approach to securing a physicalist reduction of paradigmatic mental states to paradigmatic physical states is the mind/brain identity theory. The crude statement of the view is that occurrent mental states are identical to brain states. The view also has the resources, of course, to allow for various dispositions associated with being in a mental state and will seat the ground of those dispositions again in states of the brain. The view comes in two forms – type-identity theory, and token-identity theory. The type-identity theorist seeks to identify such properties as being in pain with properties that characterize states of the 72

Distinctions: versions of physicalism and dualism

brain. The canonical form of the theory embraces such claims as that one’s being in pain is identical with having a brain whose c-fibers are firing (again where c-fibers firing can be replaced with the neurologist’s favorite candidate for the kind of brain state correlated with pain). The token-identity theorist need not deny that S’s being in pain at t is identical with S’s having a brain whose c-fibers are firing at t, but they emphasize that the brain state that is identical with any given pain at a given time need not be the same kind of brain state that is identical with another pain (even one qualitatively exactly alike) at another time. As we shall see shortly, the token-identity theorist is, understandably, worried about the empirical fact that brains seem to be extraordinarily malleable when it comes to performing various functions. When part of a brain is damaged, another part can sometimes, it seems, take on the function once carried out by the damaged part. Furthermore, we don’t want to rule out the possibility that the physiological states upon which pain supervenes vary considerably from one kind of creature to another. Still, the identity theorists who reject type-identity will owe us an account of how they can make rational decisions concerning which of the many changes in a brain they identify with pain. The type theorist imagines a utopian neurology in which we have discovered wonderful correlations between kinds of brain states and being in pain. The correlations provide whatever justification there might be for identifying a certain kind of brain state with being in pain. The token theorists won’t let the failure to discover such correlations deter them from concluding that mental states are brain states, but, as I indicated, they need to convince us that they have some principled way of deciding which brain state gets identified with a given mental state on any given occasion of its occurrence. As we shall see, one such approach involves commitment to a 73

Knowledge, Thought, and the Case for Dualism

functionalist account of mental states, but functionalism itself admits of importantly different interpretations. As I indicated earlier, the mind/brain identity theorist takes a radically different approach from that taken by the logical behaviorist. For one thing, as we saw in Chapter 1 with our brief discussion of Smart, the mind/brain identity theorist is unlikely to embrace the view as any sort of meaning analysis. At least that is true if what one means when one uses a given piece of language is something that it is, in principle, accessible to one “from the armchair.” As we shall see in Chapter 5, however, the Kripkean revolution in the semantics of proper names and names for natural kinds complicates considerably the traditional view that when two sentences have the same meaning that fact should be accessible to competent speakers of the language containing those sentences. This much seems utterly obvious. Someone who uses and understands the sentence “I am in pain” and who knows that he is in pain cannot figure out a priori what the state of his brain is. Virtually everyone can identify at least most of their occurrent mental states, and the vast majority of those people have no idea what is happening in their brains. At least that is true if we understand changes in the brain the way the neurophysiologist does. Like most everything else we are saying at this preliminary stage, however, the issue will get much more complicated when we look at certain sophisticated, but highly idiosyncratic, versions of physicalism. In what follows, we won’t be worrying much about what the most plausible candidates are for the neurological correlates of episodic mental states. It just doesn’t matter for the interesting philosophical issues. Some who write on these issues think that the plausibility of the type-identity theory rests on empirical evidence concerning what goes on in the brains of various sorts of creatures. They seem to suggest that the theory is vindicated 74

Distinctions: versions of physicalism and dualism

if they can come up with a description at some acceptable level of generality on which it will be true to say that any creature that is in pain is in a brain state of the “same” kind.24 The problem, of course, is that “same kind of state” is hopelessly vague. I own a Ford Mustang and you own a Ford Mustang. Do we own the same kind of car? In one sense, the answer is obviously “Yes.” But my Mustang is a convertible and your Mustang is not. Does that change the answer to the question? Yours is the muscle V8, mine is a kind of sad 1990 fourcylinder. Same car? Again, I don’t really care how you answer these questions. The two cars have indefinitely many properties in common and are also different in indefinitely many ways. And even if we switch to the question of when two properties are the same, similar ambiguity infects the question. We both own a red shirt. Your shirt is cherry red, and mine is more of a burgundy. Are our shirts the same color? In one sense, yes. In another, no. As I said, we don’t need to worry about empirical questions in assessing the plausibility of mind/brain identity theories. Dualists have always been more than generous in giving the identity theorists whatever empirical information they would care to stipulate. The science of the brain is, in fact, still relatively primitive, but for the most part the dualist is prepared to imagine a utopian science of the future that has discovered everything there is to discover by way of what is happening in the brain when people feel pain, fear ghosts, or are paralyzed by fear. The dualist is convinced that absolutely nothing anyone could discover empirically would ever confirm either the type- or token-identity theories. So in what follows, we’ll consider arguments for or against the mind/brain identity theory on any range of empirical possibilities that the physicalist cares to consider. 24

See Bickle (1992).

75

Knowledge, Thought, and the Case for Dualism

2.15

functionalism

Strictly speaking, functionalism in the philosophy of mind does not entail physicalism.25 But almost every functionalist is, in fact, a physicalist. The view is sometimes introduced as a kind of compromise between behaviorism and the mind/brain identity theory (Armstrong, 1968). Like the behaviorist, at least some functionalists want the view to be an analysis of the meaning of ordinary statements describing the mind and its various states. They will, however, split off into two physicalist camps depending on what they take the ontological implications of their meaning analysis to be. The functionalist is motivated, in part, by considerations that prompted some philosophers to abandon a type-identity theory. We seem to be able to make perfectly good sense of radically different sorts of creatures with radically different physiologies all of whom are, nevertheless, perfectly capable of feeling pain, feeling anger, having beliefs and desires, and being in other intentional states. When we see the Star Trek episode involving the horta,26 a silicone-based creature that eats through rock, we understand perfectly the plot twist that involves Spock discovering that the creature is in intense pain (through a mind meld the mechanics of which has always been a bit obscure). But we don’t even need exotic thought experiments of this sort. It may be that humans, dogs, frogs, and mice all have brains that are in similar states when they feel pain, but we surely wouldn’t be all that surprised to find out that as they 25

26

Strictly speaking, neither does behaviorism. There is logical space for a behaviorist who insists that one can successfully offer behavioristic translations of all of our psychological predicate expressions, while allowing that there are mental states, states for which we have no language. Some have suggested that the enigmatic Wittgenstein was leaving room for the mental even though it couldn’t play any significant role in language games. Episode 25, first aired in 1967.

76

Distinctions: versions of physicalism and dualism

experience pain, their brains are in significantly different states. And as we noted earlier, there is even strong evidence that the kind of brain state that causes pain in a human being at one time might not be the same kind of brain state that causes pain in that person at another time. It is probably no accident that functionalism as a theory of mind gained popularity around the same time as the explosion in computer technology. We describe our computers as being in certain states, or undergoing certain processes even though most of us have very little idea of what precisely is going on in the inner workings of the machine. Consider, for example, spell checking. You and I might have quite different computers with different hardware and different programming. Yet both of our computers might be able to spell check a document. I can truly describe my computer as being in the state of spell checking, while you truly describe your computer as being in precisely that same kind of state. What do we mean when we characterize our respective computers as spell checking the document? It is enormously plausible to suggest that the state of spell checking just is that state, whatever it is, that results from a spell-check command and which results in highlighting misspelled work. Again, I for one have almost no idea what happens inside the computer when I “tell” my computer to spell check a document. I have roughly the same sort of relation to my computer that I have to my car. I turn the key of my car and all sorts of wonderful things happen when I press pedals, push buttons, and shift gears. When the car moves, I know it has something to do with small explosions occurring in cylinders located somewhere in the engine, but I really don’t know the details of what happens in the engine, the transmission, or the body of the car. I know even less about what happens inside a computer when it responds to the manipulation of various keys. But I do know when a computer is spell checking a document, saving a document, printing a document, and so 77

Knowledge, Thought, and the Case for Dualism

on. The state of spell checking is a paradigmatically functional state. It is defined and understood (by most of us) in terms of input and output. The functionalist in the philosophy of mind wants us to believe that paradigmatically mental states should be understood on the model of the functional states of a computer. To say of some being that it is in a paradigmatically mental state is simply to say that it is in a state that results from certain input and results in certain output – it is the state whatever it is that performs a certain functional role. So a really crude functionalist account of pain might suggest that an organism is in pain when it is in the state that normally results from damage to the body and results in behavioral changes that are at least normally conducive to the healing of such damage. As I indicated earlier, the analysis just given formally leaves open the question of what plays that causal role. It could, in principle, be some state of a Cartesian mental substance undergoing change. But almost all functionalists are physicalists who are convinced that what plays the critical causal role, at least in humans, is some modification of the brain. A supposed virtue of the account is that we can understand perfectly well how some other creature might be in pain where what plays the relevant causal role in that creature is a different sort of brain state, or, in principle, some state of the organism other than a brain state. It is important to stress that the functionalist is not merely making the empirical claim that various mental states have functions. That seems almost undeniable, though we will need to do a great deal of philosophical work to remove from the meaning of “function” vestiges of its teleological etymology. But virtually any complete account of explanation should have something to say about the kind of explanation one often finds in the social sciences, and sometimes even in the biological sciences – explanations that seek to understand a 78

Distinctions: versions of physicalism and dualism

phenomenon in terms of the role it plays in securing, or being at least conducive to, the survival or the well-being of the society/culture/organism (see, for example, Nagel, 1953; Scriven, 1962; and Hempel, 1965). It is useful in discussing functionalism to introduce the technical notion of realizing a functional property. The most perspicuous characterization of a functional state will employ a definite description. And the most plausible account of definite descriptions (discussed in more detail later in this book) is Russell’s. On analysis the statement containing the definite description becomes a quantified statement. So on the crude functionalist account of pain, the claim that S is in pain is the claim that there is in S one and only one state x such that x results from damage and produces behavior conducive to healing. If there is something that takes the value of the variable, then we can say of that thing (property, change, state of affairs) that it realizes the functional state. The concept of realization allows one to formulate clearly a fundamental divide among functionalists. As I initially presented the view, the functionalist, like the behaviorist before him, is offering a conceptual analysis. We can capture the content of claims attributing psychological states to organisms by construing them as claims about the functional states of the organism. And on one natural interpretation of such a view, its proponent is committed to viewing these functional states as second-order properties. To say that S is in some functional state F is to say that S has the property of having a property (we don’t say which it is) which property plays a causally defined role (that’s the second-order property). Notice that the property of having a property playing a causal role is quite distinct from the property whose exemplification plays the causal role. The property of being in pain for a functionalist is not, one would naturally think, the property that realizes the functional state. Again, it was one of the alleged advantages of the view 79

Knowledge, Thought, and the Case for Dualism

that we could make this distinction between the functional state of the organism and the realizer of the functional role. That is precisely what allowed us to say of all these different organisms that they were in the same state of being in pain even if what played the causal role that defines the function might be quite different. An analogy might be helpful. Consider the following two statements: (1) Your shirt is my aunt’s favorite color. (2) Your shirt is red.

Suppose that as a matter of empirical fact my aunt’s favorite color is red. The “is” in this last claim is the “is” of identity. The “is” in (1) and (2) is the “is” of predication. Obviously, neither (1) nor (2) should be construed as identifying a shirt with a color. Now here are some critical questions, questions to which we will be returning later in the book: (3) Is the meaning or content of (1) the same as or different from the meaning or content of (2)? (4) Does (1) attribute to the shirt the same property or a different property from that attributed to the shirt in (2)?

The answer to (3) seems to me obvious. The meaning of (1) is not identical with the meaning of (2). Neither proposition entails the other. (4) is trickier. There is surely a sense in which (1) can be construed as attributing to the shirt a different property from that attributed to the shirt by (2). The property (1) says the shirt has the property of having a color property which color property my aunt likes more than any other color property. That’s a property the shirt can’t have in a world in which my aunt doesn’t exist. Put another way, the fact that makes (1) true is a complex fact that includes as a constituent my aunt. (2), by contrast, says nothing about my aunt or what she likes. (2) can be true in a world in which my aunt doesn’t 80

Distinctions: versions of physicalism and dualism

exist. If you agree that (1) and (2) make different assertions, attribute different properties, and are made true by different facts, then if you are a functionalist you should say precisely the same thing about the proposition that says of someone that he or she is in pain, and the proposition that describes the brain state that is, according to the functionalist, a plausible candidate for that which realizes the functional role. Even if it turns out that damage to the body causes c-fibers to fire, which, in turn, together with many other changes, causes the relevant behavior, the property of having a property whose instantiation results from damage and produces damage control behavior is not the same property as the property of having a brain whose c-fibers are firing. The discussion of functionalism, however, is complicated by the fact that some self-proclaimed functionalists want to resist the above argument. Kim (1998), for example, suggests that these “second-order” properties to which I say the functionalist is committed are a kind of illusion – they don’t really exist. If they were to exist then we would have competing candidates for the causal explanation of such behavior as grimaces. On the one hand, the c-fibers firing seem to be a perfectly natural candidate for the antecedent state that causally explains the grimace. But on the functionalist account that treats pain exemplification as the exemplification of this complex secondorder property, we have a second candidate for the cause of the grimace – the pain (the second-order property). This kind of problem is not new. Philosophers of science have been worrying about similar sorts of problem for decades. So consider my explanation of why the vase broke when it was struck ever so slightly. It is, I say, highly fragile. Actually, in most contexts, the explanation will be accepted as perfectly natural. But it is a bit puzzling philosophically. It is tempting to suppose that the property of being highly fragile just is the property of breaking easily. So I explain the fact that the vase 81

Knowledge, Thought, and the Case for Dualism

broke easily by pointing out that it is the kind of thing that breaks easily! Not all that informative. Of course, the disposition in question presumably has a ground. The subjunctive “it would break easily if struck” is made true by structural features of the vase – in the final analysis its molecular structure. More precisely, the subjunctive is made true by the existence of lawful connections between the thing’s having that kind of structure and its breaking easily under certain conditions.27 Most people who understand perfectly well fragility don’t actually know what the underlying grounds of the relevant disposition are. The situation is perfectly analogous to the similar dilemma that faced the functionalist. Just as the functionalist decided that our grasp of pain is through our grasp of something’s having a property (whatever it is) that plays a certain causal role, so also our grasp of fragility is understood in terms of our grasp of that thing’s having a property that plays a certain causal role. And like the functionalist we can allow that there may be more than one sort of property that can ground the relevant dispositions. It is tempting to think that our perfectly natural explanation of the way things behave in terms of their dispositions is simply our way of pointing to the fact that there is nothing unusual about the situation and that this is the kind of thing that has properties that lawfully (perhaps only statistically) correlate with the relevant effect given the relevant stimulus. In fact, when I suggested earlier that it might seem trivial to try to explain the vase’s having broken easily by reference to the fact that it was fragile, that was clearly a bit misleading. Once we know that the vase was fragile we don’t need to look for other explanations. We don’t need to worry, for example, whether a crack in the vase caused the slight blow to break it, or whether the blow was harder than we supposed it to be. We can also understand how 27

See Addis (1981) for a discussion of subjunctives and their grounds.

82

Distinctions: versions of physicalism and dualism

someone might be interested in getting further, and, in some contexts, more useful information concerning precisely what those underlying properties are. Let’s return to the very simple example we used to model the distinction between first- and second-level properties, the distinction between the shirt’s having the property of having a color that my aunt likes best, and the shirt’s being red, again where we stipulate that red is my aunt’s favorite color. While I think that we must make the distinction between the two properties, it doesn’t seem to me that we need pay attention always to the distinction. Suppose I am planning to work near an untethered bull in the field. You tell me not to wear the shirt that is my aunt’s favorite color. I ask you why and you tell me that wearing that shirt will cause the bull to charge. If I’m being a bit obtuse, I might wonder why the bull cares about my aunt’s likes and dislikes. It doesn’t, of course. It responds to the property of being red. But you did, indirectly, make a claim about that redness when you talked about the shirt’s having my aunt’s favorite color. It is red that takes the value of the variable in your specification of the second-order property. And it seems to me that the functionalist should take exactly the same approach with respect to the way in which we might casually use functionally defined predicate expressions in causal explanation. But there is another now common philosophical view that complicates the philosophical picture – the distinction between two roles definite descriptions can have in securing reference. So as I explained functionalism, the specification of the functional property critically involves the use of a definite description embedded in a predicate expression – the definite description that attempts to denote through characterizing a property uniquely playing a certain kind of causal role. Post Kripke many philosophers think that definite descriptions can “fix the reference” of an expression without becoming 83

Knowledge, Thought, and the Case for Dualism

equivalent in meaning to the expression whose reference is fixed. I think the whole notion of reference-fixing is seriously misguided and we’ll talk about this more when we talk about informative identity statements in Chapter 5. But if we assume it is a legitimate concept, then there are two ways to understand the functionalist’s view. When I attribute to some organism the property of exemplifying a property caused by damage, which property exemplification in turn produces damage-control behavior, I could take the definite description contained in the predicate expression to be one that retains its meaning whatever it denotes. The definite description would have precisely the same meaning whether I was describing a person or a mouse even if it turns out that the property denoted when I used it to characterize the person is a quite different property from that denoted when I characterized the mouse. Now the functionalist should probably be more careful than we have been up until now in characterizing the view. We do want the definite description to denote, but definite descriptions denote only when something uniquely satisfies the definite description. Yet the functionalist eager to accomplish this goal typically also wants to allow that quite different states of an organism can realize the functional property. The only way I can see to allow that both desires be satisfied is to admit that the meaning of the functionally defined expressions is species relative. But given that individuals within a species might differ with respect to what realizes the functional state, we would really need to relativize the meaning of our functionally defined expressions to individuals. Continuing to use our crude example of a functionalist account of pain, one could argue that when I say that some organism S is in pain, I mean that S is the kind of state caused in S by damage to S and which in turn produces damage control behavior in S. When I say of R that R is in pain I mean that R is the kind of state caused in R by damage . . . But that probably won’t get the job done either because the 84

Distinctions: versions of physicalism and dualism

functionalist should want to allow for the realizer of the functional role in an individual to change from one time to another. So, more precisely still, our functionalist should say that when I say of S that he is in pain at t, I am saying that S is in the kind of state at t that (plug in your favorite functionalist analysis). When I say of R that he is in pain at t I am saying that R is in the kind of state at t that (plug in your favorite functional analysis). Strictly speaking, on this approach “is in pain” won’t mean the same thing as it is used to characterize any two organisms or the same thing at any two times, but its meaning will shift only in the way that the meaning of other “relative” terms shift. So, arguably when I say of someone that he is tall, I am comparing that person’s height to the height of people in some reference class. But the reference class shifts from context to context. Sometimes the shift is more or less predictably governed by informal rules of communication, but on other occasions there is no hope of getting a precise claim absent a precise specification of the relevant comparison class. None of this stops us from recognizing that there is a sense in which “is tall” retains a common meaning as it performs its function in these relativized height claims. In precisely this way, I would imagine that our more sophisticated functionalist will claim that there is common meaning to “is in pain” in virtue of the common “role” it plays in the implicitly relativized claims we make about people and other creatures being in pain. But the Kripkean still has two ways to go concerning the interpretation of the functionalist’s definite descriptions. Again, he can take the definite descriptions to make essentially the same contribution to the meaning of the sentences whose predicate expressions they are supposed to illuminate. Alternatively, he can take the definite descriptions to have a reference-fixing role where their meaning is “discarded” once the reference is achieved. Remember the Kripkean slogan when 85

Knowledge, Thought, and the Case for Dualism

it comes to ordinary proper names and names for natural kinds. I can introduce an expression like “Fred” to refer to the next president of the USA whoever that is. According to the Kripkean, having used the description to achieve my purpose, I can by fiat make “Fred” a pure referring term denoting whomever the definite description picked out. As I said earlier, I have always found the view fantastic – if ever there were a magical theory of reference this is it (we’ll discuss this more in Chapter 5). But assuming (as most philosophers today seem to) that the view makes sense we could treat the definite description employed in the functionalist account of pain ascription to be one that fixes the reference of “pain” and which is then discarded in the correct account of the meaning of “pain.” “Is in pain” picks out the very property that was denoted by the reference-fixing definite descriptions – c-fibers firing in most people at most times, something else in fish, and something else still in mice. Now all this is a bit tricky. One can’t help but think that this functionalist is trying to have her cake and eat it too. On the one hand, the functionalist was advertising the view as one that can accommodate the idea that “is in pain” means the same thing even if the organisms to which the expression can be truly applied are in radically different kinds of physiological states as they moan, groan, and writhe about. On the other hand, they are not happy with the second-order property that view seems to introduce, so consider the function-specifying definite description as performing only a reference-fixing role. But if it is the latter, we are entitled, presumably, to take seriously the reference fixer’s own slogan that meaning is reference. The meaning of “is in pain” as applied to me at t when “pain” gets its reference-fixing treatment is the property picked out – the c-fibers firing, let’s say. The meaning of “is in pain” when applied to my guppy at t when “pain” gets its reference-fixing treatment is the property, let’s say, of g-fibers firing. Those 86

Distinctions: versions of physicalism and dualism

aren’t the same properties and hence we can’t say that the meaning of “is in pain” is constant. We will have lost one of the crucial motivations for embracing a functionalist view. Moreover, the reliance on reference fixing raises the distinct problem that we might too easily get failure of reference when there is no single, underlying property causally responsible for the observable phenomena to which we refer in our referencefixing description. We’ve already noted that the brain appears to be malleable. Suppose that it turns out that even in humans there is no single kind of neural activity that is denoted by the definite description “the neural activity causally responsible for . . .” (where here one inserts some description of a kind of behavior). One could argue that the attempt at reference was successful because one succeeds in picking out a disjunctive property, but not all philosophers will even recognize the existence of disjunctive properties. Alternatively, one could distinguish implicit species-relative and time-relative acts of reference-fixing, but with the slogan that meaning is reference, it would seem to follow that individual uses of “pain” might pick out radically different sorts of properties and thus these uses might have radically different meanings – consequences the functionalist was anxious to avoid. If we reject the Kripkean view in order to save the “common meaning” that attaches to uses of “is in pain,” it is not clear that functionalism is the slightest bit more plausible than the logical behaviorism it replaces. So suppose that “S in pain at t” means roughly that “S exemplifies the property (whatever it is), exemplification of which resulted from damage to S and which in turn resulted in damage control behavior.” Is that analytic? Hardly. The pain just is that searing sensation which occupies the center of my attention. Perhaps you can claim that the pain is a neural event – that we need to decide. Perhaps the searing pain upon which I am directing my attention is sometimes successfully denoted by a definite description characterizing a 87

Knowledge, Thought, and the Case for Dualism

property in terms of a causal relationship in which it stands. But the claim that I feel searing pain is not the claim that I exemplify a property playing a causal role. That’s every bit as comical as the logical behaviorist suggesting that my pain is a disposition to behave in a certain way. My knowledge of my pain is far more direct and unproblematic than the more theoretical knowledge of causal connection with which the functionalist is concerned. That should be enough to tell you that as an analysis of meaning the functionalist is missing the mark by a mile.28 But, the functionalist will argue, you are simply confusing your intimate knowledge of the realizer of a functional state with the functional state. Well, someone is certainly confused, but it seems to me that it is the functionalist who is confusing the first-order property of being in pain with the much more complicated second-order property of having a property with the property of playing a causal role. As we shall see in the next chapter, at least part of this debate will ultimately hinge on an inescapable phenomenological appeal – it is phenomenology that reveals to one the existence of that occurrent sensation I am calling searing pain. But the conclusion that the sensation is, in fact, pain requires more than phenomenology – that requires employment of a relatively crude thought experiment. 28

One must be careful here. There is a “paradox of analysis” to which one needs a solution. The paradox in short is this: If I successfully analyze some proposition P as analytically equivalent to Q, then P and Q are the same assertion. But philosophical analyses are supposed to be informative and are sometimes controversial. But then some philosophers won’t believe that P and Q make the same assertion and may even believe P when they don’t believe Q. So why should I argue against a proposed analysis of P in terms of Q by pointing out that it is easier to know P than it is to know Q? I have offered my own solution to the paradox of analysis (Fumerton, 1983), but it is probably sufficient to observe here that if the analysis of P in terms of Q is correct, it can’t be any more difficult to come to know Q when one knows P than it is to know that the analysis is successful. And that won’t involve additional empirical knowledge of the sort that distinguishes evidence relevant to being in pain and evidence relevant to establishing causal claims.

88

Distinctions: versions of physicalism and dualism

We can, I assume, imagine a world in which damage to the human body causes the sensation most of us will call euphoria and that the euphoric sensation in turn causes us to stay very still so that we can concentrate on the intense pleasure we feel. We may also suppose that staying very still is what is most conducive to healing the damage that was caused. It seems to me obvious that we should characterize such a world as one in which pleasure plays the role that pain plays in this world. It is not a world in which the sensations we associate with pleasure in this world would be pains. The phenomenal character of a sensation is so obviously that which makes it a pain or pleasure. The causal roles played by sensation are contingent relational properties of sensations. As I indicated earlier, many philosophers, physicalists in particular, are oddly sanguine about the possibility of providing a functionalist account of intentional states like belief, desire, and fear. But if the functionalist is providing a meaning analysis, that functionalist would be well advised to pay attention to the difficulties encountered by the logical behaviorist. If you advertise your view as one that can analyze the meaning of propositions describing intentional states into propositions describing functional states, we will want to see the analysis and we will justifiably insist that your functionalist “translation” not invoke the very expressions of “folk psychology” whose meaning you are trying to illuminate. But it will be no easier for the functionalist to accomplish this task than it was to produce the translations advocated by the logical behaviorist. What is the functionalist analysis of my wanting to become wealthy? I suppose one might be tempted to suggest that my wanting to become wealthy is that state which among other things will result from my believing that X is the way to become wealthy and which in turn causes me to do X. But, of course, the philosopher skeptical of functionalist analyses will insist that you get rid of reference to belief in the analysis. What 89

Knowledge, Thought, and the Case for Dualism

is the functionalist analysis of my believing that X is the way to become wealthy? Is it that state which causes me to do X? Hardly. Lots of different antecedent conditions might cause me to do X. It might be half-way plausible to claim that it is the state that causes me to do X when I want to become wealthy. But again we will justifiably insist that the translation eliminate reference to what you want.29 One can gesture in the direction of some “holistic” theory of meaning in excusing oneself from giving a translation, but such gestures are the last refuge of a philosopher who endorses the possibility of translation without being able to give one. Earlier in this chapter I expressed frustration with the very intelligibility of a holistic theory of meaning. It simply can’t be the case that the meaning of every claim is a function of its logical and probabilistic relations to other claims. And that’s for the simple reason that logical probabilistic relations are just that – relations. They require relata and the obtaining of the relation is parasitic upon the intrinsic nature of the relata. 29

For the same reason it won’t do to include in one’s functionalist analysis reference to “output” in the form of direct knowledge that one is in a given state. Knowledge is, or at least involves, paradigmatically mental states.

90

3 Ontological priorities: taking phenomenology seriously When philosophers deal with an issue in a given area of philosophy they often try to remain as neutral as possible on related controversies. The reason is obvious. If I can convince the reader of my views about the nature of mind and mental properties only by first convincing the reader of my views in epistemology, philosophy of language, and metaphysics, I will obviously lose most of my audience. Any single fundamental claim in philosophy is by its very nature controversial, and any combination of such claims will be even more controversial. It is almost a tautology that most controversial philosophical positions are rejected by most philosophers. Interesting and controversial philosophical positions are virtually always minority positions. So again, it is understandable that a philosopher writing in a given area of philosophy will try to bring to the table as few presuppositions as possible. Nevertheless, I have become convinced that this approach simply won’t work when dealing with the fundamental issues that lie at the intersection of philosophy of mind and epistemology. There is no epistemological Switzerland in the war between dualists and physicalists. To reach philosophical conclusions, one must take sides on important issues across a wide spectrum of philosophical controversies. It should not be surprising that in order to evaluate the attempt to reach controversial metaphysical conclusions from knowledge arguments, one must think long and hard about knowledge. 91

Knowledge, Thought, and the Case for Dualism

In this chapter I’ll sketch in more detail the controversial epistemological presuppositions that philosophers have explicitly or implicitly appealed to in arguing against various forms of physicalism. I’ll begin by outlining the specifics of the foundationalism I defend, but I will also try to show how almost any variation of the view will allow us to reject a methodological approach to the mind/body problem that is inexplicably presupposed by a great many philosophers.1

3.1

foundationalism

Almost anyone who embraces an epistemological argument for some form of dualism will endorse a version of foundationalism. The reason is that dualists are convinced that they have a kind of direct and unproblematic access to the exemplification of paradigmatically mental properties that they don’t have to the exemplification of other sorts of properties. As we shall see later, if one allows that one’s access to the paradigmatically mental is every bit as indirect as one’s access to the paradigmatically physical then one will lose all hope of rejecting Smart’s basic response to the epistemological arguments for dualism. The arguments for foundationalism are formidable. But there are many different versions of foundationalism and, as we shall see, some will be of no use to the philosopher arguing

1

See Stalnaker (2008) for an approach to these issues diametrically opposed to mine. If my reading of Stalnaker is correct, he might accept the conditional that if foundationalism of knowledge and thought of the sort I defend were correct, then the physicalist would be in trouble (though he would also argue a truly radical skepticism also looms). And I would accept the conditional that if externalism about knowledge/justification and thought were correct, then there is no plausible argument for dualism. The real question is which of the antecedents of the two conditionals is true.

92

Ontological priorities: taking phenomenology seriously

for some version of dualism. I have written a great deal elsewhere defending the version of foundationalism I take to be most plausible, so I’ll try to be brief here.

3.2

regress arguments for foundationalism

We should distinguish what some call propositional justification from doxastic justification. Doxastic justification is a property of some belief one has. Trivially, one can’t have doxastic justification for a belief without having the belief. But one might also want to talk about the justification there is for someone to believe a given proposition whether or not the person ends up believing the proposition as a result of that justification. I don’t particularly like the expression “propositional justification” as it might suggest that it is a proposition that either has or lacks the justification. It is still a fact about a cognizer that there is or is not justification for that person to believe a given proposition, and the justification in question must still be relativized to a given potential believer. So there might be justification for S to believe P while there is no justification for R to believe P (because, for example, there is evidence available to S that is not available to R). Although it is controversial, it seems to me that the concept of doxastic justification is parasitic upon the concept of propositional justification. S’s belief that P is epistemically justified when there is justification for S to believe P and S bases his belief that P on that justification. And, although it is again controversial, it seems to me that the basing relation is best understood causally. So S bases his belief on the propositional justification J that there is for S to believe P, only if the fact that S possesses that propositional justification is the causal explanation for S’s believing P. I don’t assert the biconditional because one will need to do some work to rule out “deviant” 93

Knowledge, Thought, and the Case for Dualism

causal chains, a problem that besets almost all causal analyses of problematic philosophical concepts. Unless I indicate otherwise, in what follows I’ll be talking about propositional justification. Let us say that S’s justification for believing P is inferential when that justification is constituted, at least in part, by justification S has for believing some other proposition distinct from P. S has non-inferential justification for believing P when S has justification that is not inferential. Foundationalism is best understood as the thesis that all justification ultimately depends on the existence of non-inferential justification – the justification for all justified beliefs can be traced ultimately to non-inferential justification. The best arguments for the claim that some version of foundationalism is true are still regress arguments. But I have argued (1995, 2006a) that there are actually two regress arguments, an epistemic regress argument and a conceptual regress argument.

3.3

ep i s t e m i c r e g r e s s

The following principle is accepted by most internalists and externalists alike: One has inferential justification for believing P on the basis of E only if one has justification for believing E.

The principle relies on the intuitively plausible slogan: garbage in; garbage out. If you have no justification for believing E, then E can’t confer any positive epistemic status on any proposition you might infer from P. I have argued elsewhere (1995, 2004a, 2004c, 2006a) for a much more controversial principle that I call the principle of inferential justification. That principle adds a clause concerning awareness of evidential connections: 94

Ontological priorities: taking phenomenology seriously (PIJ) One has justification for believing P on the basis of E only if (1) one has justification for believing E and (2) one has justification for believing that E makes probable P (where entailment can be seen as the upper limit of making probable).

If all justification were inferential, then for someone S to have justification for believing some proposition P, S must be in a position to legitimately infer it from some other different proposition E1. But if the first, relatively uncontroversial, clause of the principle of inferential justification were true, then E1 could give S epistemic reason to believe P only if S were justified in believing E1. But if all justification were inferential, the only way for S to be justified in believing E1 would to be to infer it from some other proposition E2 that S had good reason to believe. If all justification were inferential, however, the only way that S could be justified in believing E2 would be for S to justifiably infer it from some other proposition E3 which is justifiably believed, and so on ad infinitum. Finite beings cannot complete an infinitely long chain of reasoning and so, if all justification were inferential, no one would be justified in believing anything at all to any extent whatsoever. This most radical of all skepticisms is absurd (it entails that one couldn’t even be justified in believing it) and so there must be a kind of justification that is not inferential, i.e. there must be non-inferentially justified beliefs which terminate regresses of justification. If the more controversial second clause of PIJ is correct, the looming regresses proliferate. Not only must S above be justified in believing E1, S must also be justified in believing that E1 makes probable P, a proposition that would have to be inferred (if there are no foundations) from some other proposition F1, which would have to be inferred from F2, and so on ad infinitum. But S would also need to be justified in believing that F1 does in fact make likely that E1 makes likely P, a proposition he 95

Knowledge, Thought, and the Case for Dualism

would need to infer from some other proposition G1, which he would need to infer from some other proposition G2 . . . And S would need to infer that G1 does indeed make likely that F1 makes likely that E1 makes likely P . . . Without noninferential justification, it would seem that we would need to complete an infinite number of infinitely long chains of reasoning in order to be justified in believing anything.

3.4

th e c o n c e p t u a l r e g r e s s a r g u m e n t

The epistemic regress argument discussed above relies on the unacceptability of a vicious epistemic regress. But one might also argue, more fundamentally, that without a concept of noninferential justification, one faces a vicious conceptual regress. What precisely is our understanding of inferential justification? What makes the principle of inferential justification true (with or without its controversial second clause)? It is at least tempting to answer that question by suggesting that the principle is analytic. It is just part of what it means, one might argue, to say of someone that he has inferential justification for believing some proposition P, that his justification consists in his ability to legitimately infer P from some other proposition E1 that is justifiably believed. But if this is a plausible suggestion, if this is a plausible account of the very idea of inferential justification, we face another potentially vicious regress, this time a conceptual regress. Our understanding of inferential justification seems to presuppose an understanding of justification. Consider an analogy. The vast majority of things we take to be good (exercise, regular physical checkups, a steady paycheck) we think of as good only as a means. Things are good as means, we might say, when they lead to something that is good. Now suppose a philosopher defines being good as a means this way and then goes on to claim that the only way something can be good is to be good as a means. There seems to be something 96

Ontological priorities: taking phenomenology seriously

seriously wrong with such a view. Of course, on such a view, we might worry about how we could ever know that anything is good as a means, given that it might seem to require an infinitely long search for more and more good things. But there is a more fundamental concern, and that is that if we tried to understand all goodness as merely instrumental goodness, we would never have located the conceptual source of goodness. Our analysis of being good as a means presupposes an understanding of being good. To avoid being charged with a vicious regress, we would need to introduce the notion of something being good in itself. Indeed, it is commonplace in ethics to argue that unless we had an understanding of something’s being intrinsically good, we couldn’t even form the idea of something’s being good as a means (instrumentally good). Similarly, the foundationalist believes that without an understanding of non-inferential justification, we are not even in a position to form the concept of inferential justification. In more technical philosophical language, the foundationalist does (or should) want to understand non-inferential justification as the base clause for a recursive analysis of justification. To be justified in believing P just is to be non-inferentially justified in believing P or for P to be justifiably inferred from some proposition E1 that we are noninferentially justified in believing, or for P to be justifiably inferred from some proposition E1 that is justifiably inferred from some proposition E2 that we are non-inferentially justified in believing, and so on ad infinitum. The conceptual regress argument is not uncontroversial. One might claim that there is a generic concept of justification – one not defined in terms of non-inferential justification, which in turn can be used to set out conditions for inferential justification.2 In like fashion, I suppose one could claim that there is 2

This suggestion was made to me by Peter Klein in defending his infinitist account of justification. The crude idea behind infinitism is that one is justified in believing a

97

Knowledge, Thought, and the Case for Dualism

no such thing as intrinsic goodness. Instrumental goodness is, to be sure, defined in such a way that it presupposes an understanding of being good, but the concept of being good is not itself to be defined by appeal to the idea of intrinsic goodness. Perhaps it is indefinable. Perhaps it has some other definition. One can’t rule out the formal possibility raised by the above response to the conceptual regress argument. But it is just that – a formal possibility. The foundationalist has an exceedingly plausible suggestion for how to define inferential justification based on models of definition we understand very well. It is, at the very least, incumbent upon those who reject the account, but who accept some version of the principle of inferential justification (with or without its controversial second clause) to come up with that generic definition of justification for use in trying to say something useful about the conditions required for inferential justification.

3.5

acquaintance

It is one thing to accept abstract arguments for foundationalism. It is another to come up with a plausible account of what proposition P only insofar as one can offer an argument not only for P but for every premise one uses in arguing for P and every premise one uses in arguing for those premises and so on ad infinitum. Klein argues that it is not only possible to have infinitely many (dispositional) beliefs, but that one actually does have an infinite number of beliefs (e.g. that 2 is greater than 1, that 3 is greater than 1, and so on). But even if one has an infinitely large capacity to present arguments of the sort Klein envisions, it is not clear to me that Klein escapes the conceptual regress. It is trivially possible to present deductively valid arguments for every proposition one believes. If I believe P, I can argue for it by pointing out that (1) either P or not-P and (2) not- notP. And I can argue for these premises in like fashion. The trick is to have justified belief in the premises of such arguments. But if one makes that clear in stating one’s infinitism it is clear that one is relying on an unanalyzed concept of justification in defending the view. The conceptual foundationalist finds not only foundations for knowledge and justified belief in non-inferentially justified belief, but finds the conceptual building block of justification and knowledge in the concept of a noninferentially justified belief.

98

Ontological priorities: taking phenomenology seriously

constitutes non-inferential justification. Descartes sought secure foundations in infallible belief. If infallible belief is understood as belief whose existence entails the truth of what is believed, it is not clear that we should follow Descartes’s lead. Notoriously, if I believe any necessary truth then, trivially, my believing that proposition entails the truth of what I believe. But it is easy to imagine a person who irrationally believes a proposition that nevertheless turns out to be necessarily true. And even when one finds a few empirical propositions that are entailed by my believing them (that I exist, or that I have beliefs), those propositions entail other much more complex contingent propositions that a person may have no reason at all to believe given an inability to grasp the relevant complexity. It is much more plausible, I would argue, to search for a kind of justification that precludes the possibility of error, and does so by including in the justification the very truth-maker for the belief. The traditional foundationalist wants to involve the truth-maker for a belief in its non-inferential justification in such a way that we gain genuine epistemic assurance of truth.3 On the acquaintance theory of non-inferential justification (at least the one I’m interested in defending), one has noninferential justification for believing P when one is directly acquainted with the fact that P, the thought that P, and the correspondence between the thought and its truth-maker. I’m often asked for a further explication of this critical concept of direct acquaintance, and the questioner is usually, and perhaps 3

It is by no means easy to define clearly this notion of epistemic assurance to which I am appealing. It is not merely a psychological state of certainty or lack of doubt. That can be caused by possessing the right sort of epistemic justification, but it can also be caused by a strong enough sedative. In the end the internalist is gesturing towards genuine justification, justification that was supposed to settle for one the truth of a given proposition. But in describing the justification I discuss as genuine and implying that other concepts of justification are spurious, I could easily be accused of simply begging the question.

99

Knowledge, Thought, and the Case for Dualism

understandably, disappointed with my answer: the concept is indefinable. In general, however, a philosopher must get used to the idea that there are fundamental conceptual building blocks which defy further analysis. Just as knowledge and justified belief must have epistemic starting points, so also, conceptual analysis must begin somewhere – there must be conceptual “atoms” so to speak. But to say that acquaintance is indefinable is not to refuse to say philosophically interesting things about it. Furthermore, I would argue that there is a sense in which one can “ostend” acquaintance – one can explain the concept through an act of intellectual pointing. And ostension is one perfectly familiar way of introducing a concept. I think pain, for example, is indefinable, but if someone purports not to understand what a sensation of pain is, I can solve his problem quickly. All I need to do is hit him as hard as I can, asking him to focus on the most dramatic change he notices in the character of his experience. That change, I then tell him, is the kind of thing I call pain. But how might one ostend acquaintance? Well, think again about pain. Most of us remember occasions on which we clearly felt pain – pain of which we were aware – but where we ceased to notice the pain when we became distracted by something else. We had a bad backache, perhaps, and became engrossed in a conversation so interesting that we went for a period of time without even noticing the pain. There are, of course, two possibilities. One is that while we were distracted the pain actually ceased. The other, however, is that the pain continued, but that we simply were unaware of it for a period of time. It seems to me that the latter is every bit as plausible as the former, and on the assumption that it is the correct way to think of what happened, we can now “point” to awareness with a definite description – it is the relation we had to our pain prior to the distraction, a relation which ceased during the distraction, and which began again after the conversation ended. 100

Ontological priorities: taking phenomenology seriously

Allowing that one can be in a psychological state without being aware of that state also allows one to make sense of all sorts of interesting possibilities. When I was younger, I used to think that Freudian talk of the unconscious was either gibberish, “as if” talk, or just a way of talking about complex dispositions to behave. It now seems to me, however, that there is no reason at all to deny the intelligibility of there being occurrent intentional states, states that might have all sorts of behavioral effects, but which have the further feature of being unconscious. Just as an interesting conversation can divert one’s attention from the pain one feels, so also, beliefs, fears, desires, embarrassment – all sorts of factors – might divert one’s attention from other desires, fears, and beliefs. All this might make traditional dualists very nervous. After all, as much of the focus of this book shows, many dualists have held their views, in part, precisely because they thought they had unproblematic access to their mental states. I now seem to be allowing for at least the possibility of mental life to which one has no actual access. I have a colleague, Gregory Landini, who doesn’t understand how one can divorce the genuinely mental from that of which one is conscious. Doing so, he thinks, opens the door to construing literally talk about the mental life of machines – the mental life of anything that can respond systematically to an environment. Again, I want to be very clear about the suggestion I am making. I am not retreating at all from the notion that we get our idea of the nature of mental states from our instrospective access to such states. Through direct acquaintance with my pain I know what pain is. I know what makes the state pain. I can now intelligibly postulate that that very kind of state might occur even when I no longer am aware of it. It is easy to confuse these issues, particularly if one is an adverbialist about pain. You will remember that the adverbialist won’t distinguish between pain and feeling pain. Pain just is a certain kind of 101

Knowledge, Thought, and the Case for Dualism

feeling. But it is only if one equates feeling pain with being aware of pain (being aware of the feeling) that one will think that there is some sort of hopeless confusion in allowing for pain of which one is not aware. Pain of which one is not aware just is a feeling of pain of which one is not aware. It might be extremely difficult to settle the empirical question of whether we ever are in pain while we are not aware of the pain. To do so, we obviously can’t turn to the introspection upon which we normally rely in determining the nature of our mental states. That act of introspection might make us aware of that of which we were previously unaware. But however difficult it might be to settle the empirical question, I can’t see any argument against the possibility of occurrent, genuinely mental states of which we are unaware. And as I noted in an earlier footnote, it is equally important to distinguish both feeling pain and being aware of pain, from having beliefs about one’s pain. Furthermore, because it is not at all clear that forming intentional states requires having a language,4 we must also distinguish having beliefs about one’s pain from being able to describe in language the nature of that pain. Once we make all of these discussions, we must make sure that we attend to them in discussing some of the thought experiments that dualists often employ in trying to direct our attention to the properties or states they want to distinguish from physical properties and states. So philosophical discussions about dualism often invoke the concept of those zombies that figured so colorfully in the plots of B-movies. Can’t we make sense, the dualist argues, of a creature who looks just like us, behaves just like us, is caused to behave just like us by the same physical factors, but who has no “inner life,” who isn’t conscious? The mental is just what is missing from the zombie 4

Unless one makes the connection between thought and language trivial by talking about the “language” of thought.

102

Ontological priorities: taking phenomenology seriously

we are imagining. The thought experiment is, in a way, just a more global version of the Mary thought experiment. Colordeprived Mary lacks a specific kind of mental state (or at least a range of such states) – the color appearances. The zombie lacks the whole kit and caboodle. But given what we said above, there are two importantly different kinds of zombies. There are those who have a relatively rich inner mental life but who lack awareness of it. And there are those who lack any mental life and who also, ipso facto, lack any awareness of it. There are also, of course, zombies who have either mental life or awareness of mental life, but who lack any means of communicating their mental states to others. I would argue that all of the zombies we described above are conceptually and metaphysically possible. And I also think that one can denote what is missing from the physicalist’s world view by talking about that which is missing from the zombie’s life. But much more needs to be said before we are in a position to present such an argument in any sort of detail. Let us return to acquaintance. On my view, then, acquaintance is a real relation that obtains only between existing relata. While one can believe that which is false, desire that which will never happen, and fear that which is not the case, one cannot be acquainted with that which doesn’t exist.5 For this reason, it is highly misleading to describe acquaintance as an intentional state. The genuinely intentional can be characterized by the semantic fact that the sentences describing intentional states can be true even if their grammatical object terms fail to refer. Because acquaintance is a real relation that requires real relata, justification constituted in part by acquaintance guarantees the truth of the belief it justifies. That acquaintance is a relation has 5

Which is not to say that it is impossible to believe that one is acquainted with something possessing a certain characteristic F when one isn’t. That’s a different issue.

103

Knowledge, Thought, and the Case for Dualism

important implications for the role it might play in an argument for dualism. We saw earlier that Smart wanted to dismiss arguments that appeal to Leibniz’s law and the fact that we can form beliefs or have knowledge about pain without forming beliefs or having knowledge about brain states. Smart argued, in effect, that being believed and being known are properties of propositions (more generally, bearers of truth value), not of pains or brain states. But if we stand in a relation of direct acquaintance to pain, our standing in that relation to the pain is a property of the pain. And if a philosopher wants to identify some state of the brain with the pain, that philosopher had better become convinced that we are similarly acquainted with the brain state. But if we divorce acquaintance from propositional knowledge, does acquaintance really bring with it an assurance of truth? Might one start legitimately worrying about whether one really is acquainted with a fact that is the truth-maker for one’s belief? Although he doesn’t couch his discussion in terms of acquaintance, Timothy Williamson (2000) has an argument against the so-called “luminosity” of states commonly thought to be plausible candidates for foundational knowledge. A state of one is luminous, in his sense, if one is always in a position to know through introspection (I would say through acquaintance) that one is in that state whenever it occurs. The modality implicitly referred to is presumably something like causal possibility relative to circumstance – the kind of possibility invoked in my claim that I can raise my right hand any time I want within the next five minutes, but (sadly) I can’t dunk a basketball any time within the remainder of my life. Williamson points out that virtually all paradigmatically mental states can change incrementally where one is unable to notice any given incremental change. Pains can ever so slowly diminish until one is finally no longer in pain. Feelings of warmth can ever so slightly change until one no longer feels 104

Ontological priorities: taking phenomenology seriously

warm. Red appearances can slowly change until they become orange appearances. It seems relatively plausible to suppose that as a matter of empirical fact most people can’t detect extremely slight changes that take place in the character of their experience. If one grants that and one grants that there must come a point at which gradual change does result in one’s no longer feeling pain, or one’s no longer being appeared to redly, then, Williamson argues, certain plausible claims about knowledge will commit us to the view that we can’t always know that we are in pain or being appeared to redly when we are. We need only reflect on the fact that the last stage of the pain or red appearance before it ends is, by hypothesis, indistinguishable to us from the next stage at which we no longer feel pain or are no longer experiencing redly. But knowledge, the argument goes, requires safety. You can’t know P when there is an extremely close possible situation in which not-P and you wouldn’t be able to tell the difference between P and not-P. (Though these days most seem to think that you can know P if there is a remote situation in which not-P which you wouldn’t be able to distinguish from P.) As I said, introspection is a form of acquaintance.6 It should be obvious from what I said earlier that I have no interest in defending the claim that whenever one is in a paradigmatically mental state one can unproblematically access through introspection the fact that one is in such a state. Again, a great deal depends on the interpretation of the modal operator. It is probably always logically possible for us to introspect one of our mental states. But one shouldn’t rule out a priori the 6

One might be able to distinguish different sorts of acquaintance based on the nature of the objects of acquaintance. So we might call direct acquaintance with one’s mental states introspection; direct acquaintance with universals (if there is such a thing) rational intuition. If there were such a thing as visual direct acquaintance with surfaces of physical objects (I don’t think that there is), we could call that a form of perception.

105

Knowledge, Thought, and the Case for Dualism

possibility that one can try and fail to introspectively discover a mental state that one nevertheless has. In my view it is not really a philosophical question, but there may be empirical reasons to suppose that there really is a Freudian unconscious teaming with introspectively hidden desires, resentments, fears, and beliefs. But Williamson’s argument against luminosity might be interesting for a different reason. Consider again the pain that slowly subsides. Let’s suppose that I can’t notice incremental changes, and let’s consider again the pain state immediately preceding the state that is no longer pain. This time let’s add to the story that I still believe at that point that I’m in pain. On a view like mine, I seem to be committed to the view that I’m directly acquainted with a pain state. That acquaintance guarantees the existence of the pain and if it partially constitutes a kind of justification it looks as if it is a pretty good candidate for infallible justification. But when I am that close to error – that close to a situation in which I would have had a false belief – it seems very odd to suggest that I have infallible justification. If I wanted assurance of truth, that justification seems barely better than the externalist’s concept of justification constituted by the fact that my belief that I’m in pain is caused by the fact that I’m in pain. Note, however, that the foundationalism I defend does not identify non-inferential justification with the mere fact that one is acquainted with the fact that is a truth-maker for one’s belief. I have argued that one has non-inferential justification only when one is also directly acquainted with the correspondence between one’s belief (one’s thought) that one is in pain and the fact that makes it true. Correspondence has always seemed to me to be the sort of thing that comes in degrees. While the correspondence theory of truth is often viewed as the most natural companion to a classical two-valued logic, it seems to me that it might actually be the easiest theory of truth to combine with a many-valued logic – a logic that takes there 106

Ontological priorities: taking phenomenology seriously

to be a continuum from paradigmatic falsehood to paradigmatic truth. Williamson exploits the idea that mental states can slowly and imperceptibly change. Through such changes paradigmatic pain can eventually disappear completely. Assuming that there is a point at which one moves from being in pain to not being in pain, that will also often be a point at which one moves from being acquainted with a pain state to being acquainted with a state other than pain. The acquaintance with the state that is so very close to being a state other than pain hardly seems like a very strong source of justification for believing that one is in pain. All this seems plausible. But, of course, nothing follows from this concerning the nature of one’s justification for believing that one is in pain when one is directly acquainted with sharp, searing pain.7 The difference, I would suggest, lies in the nature of the correspondence between the thought and the pain when one has the strong justification provided by acquaintance with searing pain. The justification provided by acquaintance with such correspondence not only precludes the possibility of error, but does so in a way that gives one complete assurance of truth. There is no better justification one could gain. One has all one needs – all one could possibly want by way of justification for one’s belief. A similar response is one (but only one) of the ways in which one might try to respond to the problem of the speckled hen,8 an objection raised by Ernie Sosa (2003a and 2003b) and Peter Markie (2009) to my version of foundationalism. It seems 7 8

On this see Reed (2006). Chisholm’s classic article (1942) presents the problem as one raised by Gilbert Ryle in a discussion with A. J. Ayer. Paul Ushenko (1946, p. 103) claims that the example of the speckled hen was first given by H. H. Price, but that he (Ushenko) raised a variation of the same problem (1937, p. 90). Ushenko also claims that he discussed the problem with Ayer. I thank Steven Bayne for explaining Ushenko’s contribution to the debate (in his history of analytic philosophy electronic mailing). See also Gertler (2011, Chapter 4) for an extended, thoughtful discussion of the problem of the speckled hen.

107

Knowledge, Thought, and the Case for Dualism

plausible to claim that the surface of a hen with forty-eight speckles can present to me a forty-eight speckled appearance. When it does, I can be directly acquainted with the relevant experience but have no justification for believing that I am appeared to forty-eight-speckledly.9 More generally, it seems plausible to claim that when I am acquainted with an experience, that experience might have any number of properties of which I am ignorant.10 And if this is true, the argument goes, we need more to explain non-inferential justification than is offered by the radical foundationalist trying to identify the source of non-inferential justification with direct awareness of experience. Now one might well deny that when confronted by the speckled hen one is appeared to forty-eight-speckledly. One might claim only that one is appeared to many-speckledly. But one must be careful here. There is a familiar use of “appears” that reports tentative belief. When we say in this sense that something appears to us to have a certain property, we are only expressing tentative belief that the thing in question has the property. So I might say of a friend that he appears to be angry, or say of a paper that it appears to have been hastily written. When “appears” is used this way, it is almost obvious that the visual experience about which we have been talking only appears to us to be many speckled, and does not appear to us to be forty-eight speckled. But we are not interested here in our justification for believing that we have a certain belief. Rather, we are discussing the question of whether we have justification for believing that our experience has a certain beliefindependent character. One could still insist that even when talking about the character of the experience, it is only true to say that the experience has a certain “generic” property, but 9 10

In the tortuous language of the adverbial theorist. The most obvious examples are relational properties, but the example here might be better termed a compositional property.

108

Ontological priorities: taking phenomenology seriously

I’m inclined to agree with Markie (2009) that it strains credulity to think that there isn’t some fact of the matter about the determinate, intrinsic character of the experience. It would be like claiming that the surface of an object is colored, even though it doesn’t have any particular color, or that an object has a certain shape, even though it doesn’t have some particular shape. If anything, one should probably be more skeptical of the existence of generic properties than one is of the perfectly determinate properties on which they supervene.11 But even if one admits that the visual experience of the speckled hen has a perfectly determinate character with respect to the spots presented, the acquaintance theorist can retreat to the claim that one is simply not acquainted with a thought corresponding to the perfectly determinate appearance – either because one really doesn’t have such occurrent thoughts (see Feldman, 2004) or because in some circumstances one is only acquainted with a correspondence relation between an indeterminate thought (a thought of being many-speckled) and the determinate fact of the experience being one of being appeared to forty-eight-speckledly. Of course, if one takes the second approach one needs to develop a view about what it is to have an indeterminate thought (the kind of thought that the moderns called an abstract idea). But we surely need such an account no matter what one’s approach is to solving the problem of the speckled hen. After all, we already conceded that it is almost obvious that we can believe that an experience has the generic character of being many-speckled even though we don’t believe 11

The radical empiricists like Locke, Berkeley, and Hume were decidedly more puzzled and concerned about letting into their ontologies properties like being colored, or being triangular. Based on phenomenological considerations they were confident that the data they experience had perfectly determinate colors and shapes (that particular redness, that particular triangularity), but they weren’t at all sure what to make of being colored (or even being red) or being triangular. Their discussion of the problem was usually cashed out in terms of a search for how to understand “abstract” ideas.

109

Knowledge, Thought, and the Case for Dualism

any determinate claim about just how many speckles it has. It seems to follow from this fact alone that we are somehow able to represent the world as having a certain generic property without needing to represent it as having the determinate character upon which the generic property depends.12 The above approach to the problem of the speckled hen focuses on the nature of thought and our acquaintance with correspondence. We rejected as implausible the claim that our experience of the speckled hen doesn’t have a determinate character. But we didn’t reject the claim that in addition to its determinate character, the experience might also have a generic or “determinable” character. Whether there are both determinate and determinable properties simultaneously exemplified is a controversial ontological issue that we cannot hope to decide here. At least some philosophers (see Fales, 1990) commit themselves to the existence of determinables because they claim to be acquainted with determinables. If in addition to having perfectly determinate properties (such as appearing some particular shade of red), experiences also have determinable properties (such as being red, being dark-colored, being colored), there is nothing to prevent the acquaintance theorist from arguing that, for whatever reason, we are sometimes only acquainted with the determinable properties.13 And there lies 12

13

Indeed, it is not all that implausible to argue that virtually all thought of the sort that is expressed in language is precisely of this “abstract” character. Most of us don’t have a vocabulary that would allow us to describe the absolutely determinate character of a visual color experience, for example, with a predicate expression. To be sure, we can characterize an experience as having this character or that character, where the demonstrative is our way of picking out the determinate character. But denoting in this way isn’t exactly the same thing as describing in the ordinary sense of “describing.” Earlier, I argued that it is implausible to claim that determinables are exemplified without some determinate property on which they supervene. But this is, of course, a different view. We are conceding that there can’t be an exemplification of a determinable without an exemplification of the determinate, but are supposing that one might nevertheless be acquainted only with the determinable.

110

Ontological priorities: taking phenomenology seriously

another potential solution to the problem posed by the experience of the speckled hen. There is nothing wrong with the acquaintance theory. One simply needs to pay careful attention to the properties with which one is acquainted. The reason one is not justified in believing that the experience is fortyeight-speckled, is that one is not acquainted with that property. One is only acquainted with the generic property of being many-speckled. And as one has the thought that the experience has that generic property while one is acquainted with the generic property and the correspondence between the thought and the generic property, one does have a non-inferentially justified belief that the experience has the generic property. The reason one doesn’t have a non-inferentially justified belief that the experience has the more determinate property, even if one can be acquainted with the thought that it does, is that one is not acquainted with the fact that is the truth-maker for that more determinate thought. So where are we? We have argued that experience can have both properties with which we are acquainted, but perhaps also properties with which we are not acquainted. In doing so, we have also conceded that our experience can have properties of which we are ignorant. Our inner mental life constitutes a cognitive home not because we always have unproblematic access to our mental states. Our inner mental life constitutes a cognitive home because we sometimes have a kind of justification for believing truths about such states that is better than the justification we have for believing other empirical truths, and that is, in fact, as good as justification gets. But even if one grants that one can gain intellectual justification for believing that one is in pain that is both satisfying and extremely strong, why should one think that such justification is always better than the justification one has for believing, say, certain propositions about the physical world – say the proposition that there is a table before me now? Well, the 111

Knowledge, Thought, and the Case for Dualism

classic argument is still the best argument. A belief of ours is better justified than another when there are fewer ways in which we could go wrong. The traditional foundationalist sought the given by stripping away all that might be in error given the believer’s epistemic perspective. Descartes famously introduced the method in his meditations, but many proponents of the given followed Descartes lead. Consider this wellknown passage from Price (1950, p. 3): When I see a tomato there is much that I can doubt. I can doubt whether it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can doubt whether there is any material thing there at all. Perhaps what I took for a tomato was really a reflection; perhaps I am even the victim of some hallucination. One thing however I cannot doubt: that there exists a red patch of a round and somewhat bulgy shape, standing out from a background of other colour-patches and, having a certain visual depth, and that this whole field of colour is directly present to my consciousness. What the red patch is, whether a substance, or a state of a substance, or an event, whether it is physical or psychical or neither, are questions that we may doubt about. But that something is red and round then and there I cannot doubt.

Here epistemology meets metaphysics. The classical foundationalists argue that there is something experientially common to both veridical and non-veridical (hallucinatory, for example) experience. Admittedly, they don’t always agree on how to characterize that common element. The sense datum theorist claims that whether I veridically see something red and round or am hallucinating something red and round, I am directly aware of a red, round, sense datum. The adverbial theorist claims that in both experiences I am appeared to redly and roundly (where these are construed as non-relational properties of the mind). These days it has become more common to claim that sense experience is a species of intentional state – that visual experiences represent such properties as redness and roundness 112

Ontological priorities: taking phenomenology seriously

(a view we will discuss in more detail in Chapter 7). And it would be natural on such a view to claim that veridical experience and hallucination share these sensory intentional states. Fortunately, we needn’t decide between alternative accounts of the sensory common denominator here to make the relevant epistemological point. Adopting as neutral a terminology as possible, let’s follow a suggestion once made by Ayer and say that whether or not we are veridically seeing a red round object, we at least seem to see something red and round – or it appears to us as if there is something red and round. Now isn’t it obvious that we have stronger justification for believing the proposition that we seem to see something red and round – that we are appeared to a certain way – than we do for believing the proposition that we actually see something red and round? At least that seems true if (a) there is a probability of less than 1 that our perception is veridical and greater than 0 that it is a non-veridical counterpart, and (2) there is a common experiential element to both the veridical experience and its non-veridical counterpart that can be described as seeming to see something red and round. The probability of the disjunction, either I’m having a veridical experience or I’m having a non-veridical counterpart, seems obviously greater than the probability of either disjunct. And the claim about what I seem to see is true on either disjunct. Notice, by the way, that the existence of an experiential common denominator to both veridical experience and its hallucinatory counterpart is perfectly compatible with Williamson’s (2000, Chapter 3) much-discussed claim that perception is prime. Perception is prime in his technical sense when it is not simply the conjunction of an internal state with the obtaining of an external condition. But, of course, almost no one in the history of philosophy ever endorsed the view that one perceives X when one is in a certain internal mental state that occurs when X exists. The standard view required that for there 113

Knowledge, Thought, and the Case for Dualism

to be veridical perception of X there must be a causal connection between the existence of X and the relevant internal mental state. There are a number of philosophers these days who would reject the above argument for thinking that our justification for believing truths about appearance is stronger than our justification for believing truths about external reality. Some reject the supposition that the epistemic probability that a given experience is veridical is always less than 1. Williamson (2000, Chapter 10) famously claims that epistemic probability just is probability relative to evidence. And a person’s evidence, he thinks, should be identified with what that person knows. Thus if a person knows that a given experience is veridical, then trivially the epistemic probability for that person that it is veridical, i.e. the probability relative to what he knows, is 1. Unless we assume at the outset some form of knowledge skepticism concerning the external world, we should never concede that typically the justification for believing the disjunction (either the perception is veridical or it is not) is stronger than the justification for believing either disjunct. But to state Williamson’s position clearly is to see that it can’t be true. There are familiar reasons for thinking that there is a concept of knowledge floating around that really does require for knowledge that P that there is a probability of P relative to evidence that is 1. That’s still the simplest way to deal with the lottery paradox for knowledge. No matter how high the probability is that we just bought a losing ticket, almost everyone is uncomfortable claiming to know that the ticket is a loser. But one must surely choose between understanding knowledge as requiring that what is known has an epistemic probability of 1, and regarding as true most commonplace knowledge claims. We claim to know what day we are leaving England, what time we left the airport in Chicago, what the weather was like when we left, and so on. But surely everyone will concede, at least should concede, that we have 114

Ontological priorities: taking phenomenology seriously

stronger justification for believing that we exist than that we will be in Iowa next week. Surely everyone will concede that we often have stronger justification for believing that I’m wearing a dark shirt than that I’m wearing a black shirt. And one can’t make these obvious concessions if we accept Williamson’s view both that we know all these things, and that propositions known have an epistemic probability of 1.14 There is an older worry about attempts to retreat to the epistemically more secure world – a cognitively safe home – of appearance. Sellars (1963, p. 151) famously argued that in at least one common use of “appears” or “seems” talk, to claim that something appears red, for example, is just to make a certain comparative claim. Something appears red if it appears the way red things appear under certain conditions. Obviously, knowledge that something appears red in this sense can’t be any better than knowledge of at least some truths about the physical world. Knowledge of the truth of comparative appearance claims just is a certain kind of knowledge of the physical world. But while there is a comparative use of “appears” it has always seemed to me that Chisholm (1957, Chapter 4) was absolutely right in stressing that there is also a non-comparative use of “appears.” Contra Sellars we really are entitled to give ontological status to that way red things appear. And there is nothing to stop us from inventing a technical terminology to describe that distinctive way of appearing with which we are all familiar. Now as I emphasized, the preceding does presuppose that there is an experiential common denominator, so to speak, to both veridical experience and its phenomenologically indistinguishable non-veridical counterparts. In some circles, 14

Williamson (2009) tries to soften the counter-intuitive nature of his claim by allowing that while the epistemic probability of all propositions known is 1, some knowledge is “safer” than others. Knowledge is safer when it is “farther away” from possible worlds in which the belief in question would be false.

115

Knowledge, Thought, and the Case for Dualism

however, disjunctivism is in danger of becoming the received view. Many contemporary epistemologists give up on the idea that we can simply add conditions to appearance in order to distinguish between veridical and non-veridical experience. They claim that we simply need to give a quite different account of what’s going on in veridical experience from the account we give of what is going on in non-veridical experience. The disjunctivist’s position strikes most traditional foundationalists as almost preposterous. If there is nothing common to both veridical and hallucinatory experience why are they impossible to tell apart? In response the disjunctivist might remind us of Williamson’s argument that most of us are unable to distinguish at least some pains from states that are not pains. There is presumably a marked difference between being in pain and not being in pain and if that difference can exist without our being able to recognize the difference, why should we get so bent out of shape about the disjunctivist’s claim that something radically different is happening in both veridical and non-veridical experience just because we are unable to distinguish the two kinds of experience? But one doesn’t have to think about this response long before one hits upon the relevant difference. Williamson’s example trades on the existence of a continuum where, by hypothesis, there is only a tiny difference between each change on the continuum. It’s the fact that there is so little difference between the last pain state and the state that is not a pain that explains the difficulty we have distinguishing them. Contemporary disjunctivists are typically committed to giving a radically different account of veridical perception to that they give of non-veridical perception. They offer us, in effect, no account of the phenomenological indistinguishability of the two kinds of experience.15 15

That’s not true of all disjunctivists. I would classify Johnston (2004) as a disjunctivist. He claims that in both veridical experience and its hallucinatory counterpart

116

Ontological priorities: taking phenomenology seriously

And until such an account is forthcoming, we have no alternative but to retreat to the radical empiricist’s idea that there is in fact a common internal state to both veridical and nonveridical experience, a state potential knowledge of which is common to both veridical and non-veridical experience. Once we reject disjunctivism, we will be led inexorably to the conclusion that we have a kind of knowledge of our internal mental states that is better than our knowledge of any propositions describing the physical world. And that’s because if we take the best possible justification we have for believing what we do about the world around us, there is always a safer proposition describing the subjective character of our experience to which we can retreat. It’s there that we find a more secure cognitive home to which we can retire. It is from that home that we can wonder what else might be reasonable to believe. I have tried to illustrate the more secure epistemic status that beliefs about sensations enjoy. In Chapter 2 we talked about various physicalist accounts of mental states that would seem to render introspective knowledge of such states problematic. But if a view can’t explain the special epistemic status that our beliefs about sensations enjoy, that is a reason for rejecting the view. If logical behaviorism or functionalism were a correct account of pain, then we wouldn’t have unproblematic noninferential knowledge of paradigmatic pain when it occurs. But we do have such knowledge and that is a reason for rejecting we are aware of the same universals – exemplified by surfaces of objects in the veridical case and “hanging together” (constituting a profile) somehow without being exemplified by physical objects in the non-veridical case. But I can’t for the life of me understand what this hanging together is supposed to be. Others lean towards the Meinongian view that in both veridical experience and its non-veridical counterpart we are directly aware of objects – the same kind of objects. It’s just that in the non-veridical case the objects don’t exist. But again, although I don’t really understand the view, I would think that there is a huge difference between an existent and a non-existent tree!

117

Knowledge, Thought, and the Case for Dualism

logical behaviorism and functionalism.16 Indeed, if we accept the view that we are directly acquainted with sensations like pain, then, as I indicated earlier, we must reject an identification of pain with anything with which we are not directly acquainted. It sounds odd to suppose that we are directly acquainted with brain states. And it certainly seems true that we can’t get non-inferential introspective knowledge of the truth of propositions describing brain states. But as we will see in Chapter 6, it is not entirely clear that we can’t find room for constituents of brain states with which we are directly acquainted. In trying to illustrate direct acquaintance, I have talked primarily about our direct access to sensations like visual experience or pain. I have just argued that we can reject certain forms of physicalism by arguing that these views cannot accommodate the obvious fact that we are directly acquainted with such states. I have already mentioned in Chapters 1 and 2 that I’m not sure that there is an important distinction for the physicalist between the so-called hard problem of qualia and what some take to be the allegedly easier problem of analyzing intentional states. And one reason I think this is that it seems to me that we have 16

Again, I am presupposing here that the behaviorist or the functionalist in question is providing a conceptual analysis. Propositions describing pain are analytically equivalent to propositions describing dispositions to behave (on the behaviorist’s view) or functional states (on the functionalist’s view). If an analysis is successful then the proposition describing the behavior or the function asserts nothing more (whether we know that or not) than the proposition describing the pain. To be sure there is a paradox of analysis. It does seem that two propositions can be analytically equivalent without one realizing that they are. It therefore doesn’t seem to follow that just because P and Q are analytically equivalent anyone who believes P believes Q. But if P and Q assert the same thing, it should be possible to come to know P in precisely the same way that one comes to know Q and vice versa. And that spells doom for the behaviorist and the functionalist. If one rejects analysis as conceptual analysis, and embraces instead the view that in analysis one discovers a posteriori metaphysically necessary connections between phenomena, then the preceding argument won’t work. We’ll address such views later in this book.

118

Ontological priorities: taking phenomenology seriously

introspective access to at least some paradigmatic intentional states. One knows directly through acquaintance that one fears the dark, or that one is thirsty. If logical behaviorism or functionalism were true, one wouldn’t know directly that one is in such states. Both logical behaviorism and functionalism analyze intentional states in such a way that one must discover causal truths to know that one is in the relevant intentional states and one cannot discover causal truths through an act of acquaintance with the truth-maker for such claims.17

3.6

i nfer entia l j us tific a t i o n a n d th e p h y s i c a l w o r l d

In the preceding section I argued that we have better justification for believing propositions about our occurrent mental life than we have for believing propositions describing the physical world. On some views of non-inferential justification, that is compatible with both sorts of beliefs enjoying non-inferential justification. On process reliabilism for example, the distinction between non-inferential and inferential justification is entirely a function of the input to the belief-forming processes. My introspectively formed belief that I am in pain and my perceptually formed belief that there is a table before me might both be non-inferentially justified on such a view. Both kinds of processes might take as their input data other than (justified) beliefs.18 17 18

Again, for a more detailed argument to this effect, see Fumerton (1995). Though I have argued (1995) that even if process reliabilism were true, it is still highly implausible to claim that ordinary beliefs about our physical environment are formed as a result of a belief-independent process. Admittedly, we don’t seem to consciously entertain premises in forming beliefs (better expectations) about our physical surroundings. But it doesn’t follow from that fact that we don’t have stored a host of dispositional beliefs about such things as conditions of perception – beliefs that are causally operative in our reaching the conclusions we do about the physical world around us.

119

Knowledge, Thought, and the Case for Dualism

Earlier in this chapter I tried to set out my own version of foundationalism. I don’t pretend to have done justice to radical externalist alternatives. The comparison of various metaepistemological accounts of foundational justification would be itself the subject of a rather long book (see Fumerton, 1995). I argued that the traditional foundationalist (inspired by Descartes) sought foundations that bring with them assurance of truth. The fundamental objection to not only reliabilism, but also all paradigmatic externalist accounts of justification, is that on such views, beliefs can enjoy justification without that justification giving one the slightest assurance that what one believes is true. Having a reliably produced belief when one hasn’t the slightest reason to think that the belief is reliably produced, hardly does anything for the philosophers trying to satisfy their curiosity on some intellectually interesting question. There are a great many versions of importantly different externalisms, but all suffer from the above problem. They also virtually all succumb to the so-called new evil demon problem. Just as there are many different versions of externalism, so also there are many different versions of internalism. One important sort of internalism we might call internal-state internalism. The internal-state internalist’s slogan is that if two people are in precisely the same kinds of internal states,19 then whatever justification the one has for believing some proposition the other also has. Now suppose that our experiences are caused in precisely the way we think that they are, and imagine an internal duplicate in a world in which the very same experiences are produced by an evil demon bent on deceiving us. Whatever epistemic status our beliefs enjoy, the argument goes, is also enjoyed by the beliefs of our deceived counterpart. 19

Defining clearly “internal state” is easier said than done, particularly if one allows direct acquaintance with such entities as universals. See Fumerton (1995).

120

Ontological priorities: taking phenomenology seriously

The intuition is powerful. Even Goldman (1988), the (justifiably) most influential externalist, was finally moved to recognize a distinct concept of justification (he called it “weak justification” to try to blunt the force of the argument). Put very crudely, a belief enjoys weak justification when it is formed in a way that conforms to the shared standards of a given epistemic community. The above argument for internalism (like most arguments in philosophy) is far from uncontroversial. Content externalists of a sort we will discuss later might deny that the demon victim we are trying to imagine can have beliefs with the same content as our beliefs (assuming that we are caused to believe what we believe in the way we think we are). Putnam (1981, Chapter 1), for example, famously argued that a brain in a vat couldn’t form the belief that he is a brain in a vat. The basic idea is that unless one interacts with the world in various ways one can’t form beliefs about that world.20 But the thesis must be carefully worded to be in any way plausible. We can clearly form complex thoughts that do not correspond to anything with which we have causally interacted (the thought of mermaids or the thought of unicorns). And even Putnam ends up conceding that if your presently envatted brain was previously in a body that interacted with the world, or was connected in the right way to some other conscious being who has previously 20

To get a feel for the view think about what makes a photograph a photograph of Jones rather than of someone else. It obviously isn’t a question of how much like Jones the photograph is. For one thing pictures aren’t anything like people. But nor is it a function of the similarity of the experience produced by looking at the picture and the experience produced by looking at Jones. The photograph might be a very bad photograph and look more like Smith than Jones. It is entirely plausible in the case of photographs to suppose that a photograph is of Jones just in case Jones is the person who figures in the right way in the causal process that produced the photograph. One can see how a philosopher might entertain (incorrectly I believe) the idea that thought (mental “pictures”) are like photographs in that what the thought is a thought of is a function of its causal history.

121

Knowledge, Thought, and the Case for Dualism

interacted with the world, that envatted brain could still entertain the proposition that it is, in fact, a brain in a vat. Other externalists concede the intelligibility of there being a victim of demonic machination led to believe assertions about a physical world, but think that they can define their externalist conception of justification in such a way as to allow that their cognitive states are justified. So before he introduced “weak justification” to deflect the force of the thought experiment involving the deceiver, Goldman (1986) tried to define justification in terms of reliability, and reliability in terms of the frequency with which in normal worlds the relevant beliefproducing process results in true beliefs.21 Normal worlds are worlds which conform to our fundamental beliefs (true or false) about the world. They are worlds, for example, in which fundamental laws of nature hold, in which we do causally interact with our physical environment, in which there are other minds, and so on. A satisfactory analysis of the concept of normal worlds will, of course, be difficult: the very idea of defining justification in such a way that we lose the core reliabilist intuition that justification should be intimately connected to truth seems to defeat the very idea of a normal worlds “reliabilism.” Still other philosophers (Comesaña, 2002 and Sosa 2009, for examples) endorse a kind of reliabilism that identifies the relevant reliability to be used in the analysis of justification as reliability relative to the actual world (actual world reliabilism). So here the basic idea is that we can say that the demon 21

We are glossing over important distinctions. Goldman distinguishes beliefindependent processes and belief-dependent processes. The former take as their “input” states other than beliefs (or if they do take beliefs, the epistemic status of the belief is irrelevant). The latter take as their input (at least in part) belief states. To produce justified beliefs the belief-independent processes must be unconditionally reliable (result in mostly true beliefs, assuming that we are defining reliability in terms of frequency). The latter need result only in mostly true beliefs when the input beliefs are true. But the input beliefs must also be justified.

122

Ontological priorities: taking phenomenology seriously

victim’s beliefs are epistemically justified because we can think of those beliefs as produced in a way that is reliable in the actual world (again assuming the actual world isn’t a demon world). The demon’s victim isn’t, by hypothesis, forming beliefs in a way that is reliable in the demon world, but even in the demon world he is forming beliefs that are reliable in the actual world. This sort of response frustrates internalists no end. The internalist is trying to ask a question about what the epistemic status of a belief would be if the belief were caused in the way described by the demon scenario. In answering the question, this reliabilist tells us that it depends on what belief-forming processes are reliable in the actual world. The internalist then will ask the following obvious question: Would our beliefs be epistemically justified if they were caused in the actual world (unbeknownst to us) by a demon? Here, both Comesaña and Sosa argue that there is an interpretation of the question such that the answer is still “Yes.” How are we to answer the question according to the actual-world reliabilist? We take ourselves to the nearest possible world in which the actual world is a demon world, and then we ask ourselves whether in that possible world beliefs would be produced in a way that is reliable in the actual world (the world that is still supposed not to be a demon world). Since they are reliably produced in the actual world, the counterfactual question still gets answered affirmatively. All this seems almost willfully designed to misunderstand the thrust of the question. But the issue is complicated by difficult controversies concerning the truth conditions for counterfactual claims. Perhaps the internalist is simply better off avoiding counterfactuals in asking the relevant question. The issue, after all, concerns a disagreement between internalists and externalists about the conceptual connection between various propositions. Consider the following propositions: 123

Knowledge, Thought, and the Case for Dualism (1) We have various sensations and they are caused directly by a very powerful mind. (2) We have a stable disposition to form beliefs about the existence of physical objects and their properties as a result of having those sensations, beliefs that are mostly false.

According to Sosa’s view, do (1) and (2) conceptually entail (3) Our perceptually formed beliefs are unjustified.

The internalist thinks that Sosa and the classic reliabilist must say “Yes.” But most internalists think that the answer is clearly “No.” When the dust settles, it seems to me that the demon described above still plays havoc not only with true belief but with finding a plausible metaepistemology that builds on an externalist, and in particular a reliabilist, account of foundational justification. There are still many variations on a reliabilist theme, some of which are designed to accommodate intuitions about the new evil demon thought experiment. Michael Bergmann (2006, Chapter 5) thinks that if someone is transported to a demon world, that person’s beliefs are still epistemically justified because they are produced by a process that is reliable in the environment for which the process was designed.22 But Bergmann’s approach will still have the counter-intuitive consequence (at least from the internalist’s perspective) that someone unfortunate enough to have been designed by a demon for the purpose of the demon’s demonic machinations would still have epistemically unjustified beliefs. If the experiences of that person are just as ours are, it still seems decidedly odd to the internalist to assign the beliefs a different epistemic status. 22

Theists are likely to try to interpret the talk of design quite literally. Others are invited by Bergmann (in the spirit of Plantinga) to search for some way of understanding “design” in terms of evolution and natural selection.

124

Ontological priorities: taking phenomenology seriously

Lastly we should probably consider a view that is called by its proponents (Henderson and Horgan, 2006) “transglobal reliabilism.” Here the basic idea is that a belief is epistemically justified if it is reliable in most possible worlds. If it is true that a belief is reliable in most possible worlds it is necessarily true (in every world it will be true that the process is reliable in most worlds), and we won’t be forced to assign the demonproduced, perceptually based beliefs a different epistemic status in different possible worlds. But as Henderson and Horgan know full well, the very notion of transglobal reliability is problematic. For any belief-producing process that is only contingently (actually or counterfactually) reliable there are infinitely many possible worlds in which the process is (actually or counterfactually) reliable and infinitely many possible worlds in which that same process is (actually or counterfactually) unreliable. It’s hard to understand the sense in which any such process could be reliable in “most” possible worlds. To be honest, however, I actually agree that there is intuitive sense we can make of the critical concept upon which Henderson and Horgan rely. Suppose, for example, that someone is asked to randomly select a number. Intuitively, there are more possible worlds in which the number selected is not a prime number than there are possible worlds in which the number selected is prime, and that is the case even though there are infinitely many possible worlds in which the number selected is prime, and infinitely many possible worlds in which the number selected is not prime. But I suspect that our intuitive understanding of this sort of talk is parasitic upon our prior understanding of epistemic probability. We believe that it is unlikely relative to our evidence that a person will select a prime number and that is all we mean by saying that there are “more” possible worlds in which that doesn’t happen. If I’m right about this, it is not even clear that justification defined in terms of transglobal reliability is an alternative to the 125

Knowledge, Thought, and the Case for Dualism

internalist’s thesis that justification supervenes on internal mental states that necessarily make probable certain propositions. It is interesting to note that both Horgan and Henderson and the evidentialists are going to end up committed to the claim that claims of epistemic probability are necessary truths. Whether or not this is true, the transglobal reliability upon which Henderson and Horgan rely on defining epistemic justification is similar to normal worlds reliabilism in severing a connection between a belief’s being epistemically justified in the actual world and its being produced in a way that usually gets us to the truth. There is no doubt that most versions of externalism make it much easier to get epistemic justification (provided that the world cooperates). Paradigmatic externalist accounts of justification allow for the possibility of non-inferentially justified beliefs about the physical world. Indeed, paradigmatic externalist accounts of justification allow for the possibility of noninferentially justified beliefs about just about anything.23 But if we want our paradigm of non-inferential justification to bring with it rational assurance of truth, and we do so by trying to build the truth-maker for a belief into the conditions that constitute non-inferential justification, then we are going to be hard-pressed to allow for a non-inferentially justified belief in any proposition describing the physical world. The argument relies again on a rejection of disjunctivism and the intuitively plausible idea that we enjoy precisely the same justification for believing what we do in both veridical cases and their phenomenologically indistinguishable counterparts. Assume that you have justification for believing that there is a physical object with properties F and G before you. Could that 23

Plantinga (2000) successfully exploits this feature of externalism in arguing that on his own version of externalism there is no absurdity in the suggestion that one can have non-inferentially justified beliefs in the existence of God.

126

Ontological priorities: taking phenomenology seriously

justification be constituted by your direct awareness of the object with its properties? No. You would have had precisely the same sort of justification were you vividly hallucinating an object with properties F and G. And in vivid hallucination there is, by hypothesis, no object that is F and G to serve as the object of acquaintance.24 Remember that acquaintance is a real relation that requires existing relata. One cannot be acquainted with what doesn’t exist. If one accepts the kind of radical foundationalism I defend, then, one must inevitably reach the conclusion that if one has justification at all for believing any propositions about the physical world that justification must be inferential. And the only plausible candidates for the premises available to justify such beliefs are propositions describing our phenomenal life. Again, I can’t defend such a view in detail here. I have tried to do so elsewhere (Fumerton, 1995). I will briefly emphasize that many of the reasons for rejecting that classic position are highly problematic. For example, I suspect that at least some people reject the claim that our knowledge of the external world is inferential on the grounds that we don’t move through any conscious process of reasoning in forming beliefs about our environment. That’s no doubt true, and it may well be that important historical figures should be faulted for over-intellectualizing the process of belief-formation. But as I indicated earlier, background beliefs can be causally critical to the formation of beliefs even if they are not consciously entertained as premises. I’m inclined to think that philosophically ideal justification arises for the philosopher only when that philosopher has laid bare the evidential structure of the reasons supporting beliefs, and the nature of the evidential connections holding between one’s evidence and 24

Unless one introduced into one’s “naturalistic” ontology such strange entities as objects that don’t exist. It would be hard for a naturalist of this stripe to get us to take seriously criticisms of the bloated ontology endorsed by the dualist.

127

Knowledge, Thought, and the Case for Dualism

one’s conclusions. But there is nothing to stop the traditional foundationalist from allowing derivative concepts of justification. One might allow, for example, a degenerate concept of inferential justification that is satisfied when one is caused to believe a proposition by justified beliefs that do support the proposition in question (whether one is aware of the evidential connection or not). Or one might even allow that one enjoys a kind of inferential justification when one is caused to believe a given proposition by truth-makers for propositions that do, by themselves, or with background beliefs, evidentially support the proposition believed (see Fumerton, 2004b). But one thing is certain, and that is that we can’t seriously take the absence of conscious inference to be a reliable indicator that the justification, if any, a belief enjoys is non-inferential. Surely, no one thinks that my belief that placing my hand on the red hot poker will cause pain is non-inferentially justified. But I also daresay that virtually no one consciously entertains premises from which they draw the conclusion in question. When you hear the doorbell and expect someone to be at the door, you again rarely, if ever, consciously form a belief in premises describing past correlations between certain sorts of sounds and the presence of people, and then laboriously draw out the conclusion that someone is probably at the door. But again, it is surely absurd to suggest that the justification one has for expecting someone to be at the door is non-inferential. These relatively uncontroversial facts might spell some difficulty for the internalist. There are at least two diagnoses of why we don’t find conscious inference occurring very often, both of which might create problems for the internalist. One (suggested by Hume 1888 in his discussion of skepticism with regard to the senses), supposes that “nature” has taken care of our need to respond quickly and efficiently to data received through the senses. These days, Hume’s talk of nature would be more likely replaced by talk of evolution. The other 128

Ontological priorities: taking phenomenology seriously

response might concede that some sort of reasoning must have occurred at one time, but one obvious function of memory is to “store” conclusions reached for future use. When we store the conclusions, we don’t need to store whatever process was used to reach the conclusions. Both views would leave the internalist in a precarious position. The beliefs produced by “nature” or the beliefs “stored” by memory will not be accompanied by a robust enough inner mental life to provide a rich enough body of evidence to give us the relevant internal justification that these perfectly appropriate beliefs seem to enjoy. Moreover, Goldman (1999) argues, there is typically nothing in our current mental life that would allow us to distinguish stored beliefs that were originally formed in a legitimate way from those that were formed in an illegitimate way. Something I read in the New York Times might now be resting in my mental “filing” system right alongside something I read in the National Enquirer. Goldman wants to suggest that the former is plausibly regarded as a justified belief, while the latter is not. Internalists do not agree with one another concerning how to meet the above objections. The more radical view simply denies the intuition that beliefs produced by “nature” or, for that matter, beliefs stored in memory (unaccompanied by any apparent memory of their source) are epistemically justified. Certainly, it seems right to me that I find myself with all sorts of beliefs cluttering my mind that simply have no serious epistemic backing. They may have once been held for good reason, but I am not even a bit inclined to suppose that just because they were once held rationally that they still are. One can obviously lose one’s justification for believing a given proposition, and it seems entirely plausible to me to think that this happens rather frequently.25 That’s not to say that the belief might not still be useful. It may still play a valuable functional 25

See Conee and Feldman (2001) for an extensive reply to Goldman (1999).

129

Knowledge, Thought, and the Case for Dualism

role in keeping me alive and free from danger, for example, but that’s not the same thing as enjoying epistemic justification.26 Alternatively, one can search one’s mental life for the relevant evidence that would allow one to distinguish the epistemically “respectable” from the epistemically “promiscuous.” It does seem to me plausible that there are introspectively different features of beliefs we have. Perhaps it is nothing other than a difference between the degree of confidence we have in one proposition as opposed to the degree of confidence we have in another. Moreover, while I don’t really remember the exact occasion on which I learned that the Battle of Hastings took place in 1066, that thought is typically accompanied by a gestalt of images involving classrooms, a tall history teacher, and a vivid apparent memory of hearing that the Saxon king was killed by an arrow through his eye. These phenomenological differences would be epistemically useless by themselves, but if I also have apparent memory of general truths about the past reliability of these indicators when it comes to truth, I might get epistemic traction (assuming that I can find an internalist solution to the problem of justifying beliefs about the past based on apparent memory).

3.7

i n t e r n a l i s m a n d th e t h r e a t of skepticism

Some of the above discussion leads directly to a more serious concern for the classic foundationalism that seeks to ground all 26

Alston (2005) argues that it is a mistake for contemporary epistemologists to think that they must choose between internalist and externalist approaches to understanding justification. Rather, he argues, one should simply recognize that there are all sorts of different virtues a belief might have. Some of them are captured by the externalist’s criteria for epistemic justification; some of them are captured by the criteria emphasized by the internalist. There is something plausible about Alston’s suggestion, but I have emphasized all along that I am interested in that special concept of justification that has preoccupied philosophers – the kind of justification that brings with it rational assurance of truth.

130

Ontological priorities: taking phenomenology seriously

justification in knowledge of subjective experience. The worry is that such views face an uphill battle in responding to the skeptic’s attack.27 The fact is that they do. A full discussion of the challenges facing the internalist trying to respond to the skeptic would take us far afield. I have discussed these issues extensively in a number of places (Fumerton, 1992, 1995, 2004a, 2004b, 2005, 2007). The likelihood of one’s being successful in responding to the skeptic will hinge, in large part, on two issues. The first concerns the metaepistemological question of how to understand epistemic probability – how to understand what it is for one proposition to make probable another. The second concerns the question of which epistemic principles, if any, are true. So, if, for example, epistemic conservatism were true, and the mere fact that one finds oneself believing a given proposition makes prima facie probable the truth of that proposition, the refutation of skepticism becomes much easier. If the mere fact that one has a certain visual experience makes likely that one is before a physical object with such and such characteristics, then, again, the rejection of skepticism doesn’t appear to be very hard. Of course, one must have a view about what would justify one’s acceptance of a given epistemic principle, and I have argued that the only real hope for an inferential internalist (one who thinks that one can make a rational inference only when one has reason to believe that the relevant probabilistic connection obtains between one’s evidence and one’s conclusion) is that epistemic principles are necessary truths, knowable a priori. They don’t seem 27

When Quine (1969) first urged us to naturalize epistemology, he was quite candid in suggesting that one of our motivations to accept the suggestion should be the realization that traditional foundationalism had its chance to show us how we can move beyond the foundations to the rest of common sense, and manifestly failed to show us how to succeed. More recently, Goldman’s attack on internalism (1999) consists, in large part, in arguing that the current internal states of a cognizer give us far too little to work with in justifying commonsense beliefs.

131

Knowledge, Thought, and the Case for Dualism

to be analytic so they must be some sort of synthetic necessary truth (of the sort Keynes, 1921, defended). I don’t think that the path to refuting skepticism is easy. I don’t think that one has many non-deductive evidential connections with which to work. In fact, it seems to me that the only real hope for my kind of foundationalist to avoid skepticism is to engage successfully in some radical reductions when it comes to understanding propositions describing the external world (more about this later). And even then, success against the skeptic is hardly clear sailing. The irony, however, is that if there is any hope for the physicalist to find room for phenomenal properties in a broadly physicalist world view, that hope lies in the very radical foundationalism that makes inexplicable the kind of deference to science on matters metaphysical. As we shall see, however, that hope falls far short of anything that would constitute a rational basis for endorsing physicalism.

3.8

re j e c t i n g m e t h o d o l o g i c a l naturalism

I want to conclude this chapter by making explicit the way in which the epistemological conclusions advanced here require us to reject methodological naturalism in the philosophy of mind. As we noted in Chapter 1, a great many philosophers today give their unswerving loyalty to a naturalistic world view. Part of that world view carries with it the presumption that we should let science settle fundamental metaphysical controversies concerning what kinds of things exist. In the next two chapters we will talk more about the suggestion that empirical investigation will settle controversies concerning certain views about the nature of mental states. But if anything is obvious, it is that the epistemological conclusions reached above are incompatible with a deference to science as the arbiter on all matters metaphysical. Even if one rejects the 132

Ontological priorities: taking phenomenology seriously

details of the foundationalism I defend, it seems obvious that the scientific investigation of mental states and their causes always implicitly presupposes some version of the radically egocentric foundationalism sketched above. Let me explain. If we are studying the mind/brain empirically, how would we proceed? How would we conduct our empirical investigation? One thing seems obvious. We can engage in endless empirical study of activity in the brain, scanning that activity in all sorts of exotic ways, all without ever discovering what fundamentally interests us. Insofar as we are interested in understanding mental states, we are going to need to correlate the empirically observed neural activity with a less exotic, but absolutely indispensable source of information about the mind. In short, we will inevitably rely on the first-person reports of people concerning their mental life in trying to correlate observed neural activity with the “relevant” mental state. So if I’m trying to figure out what goes on in the brain when someone feels pain, I’ll need some independent way of discovering that the person in question is in pain before I’m even in a position to isolate the “pain center” of the human brain. There is, of course, more than one way to do this. In humans the easiest way to proceed is to ask the subject what his or her mental state is as the relevant brain scanning occurs. But there are other more or less reliable ways to correlate, say, pain with some state of the brain. If I’m dealing with animals other than humans, and I don’t have PETA looking over my shoulder, I might induce pain in a dog, say, as I monitor the dog’s brain. I’ll be assuming, of course, that I did successfully induce the relevant pain, and about that I could conceivably be mistaken. But then I could conceivably be mistaken in trusting a person’s first-person report that he is in pain as I try to correlate that pain with neural activity. In fact, it seems to me that the best possible empirical investigation would involve the neural scientist monitoring his own brain activity as he introspects the 133

Knowledge, Thought, and the Case for Dualism

character of his mental states. That would seem to me to eliminate potential sources of error in the experiment. One moral to draw from the above platitudes is obvious. Don’t pay the slightest bit of attention to philosophers who talk about ideal theories and methods of cognitive science replacing our inadequate folk psychology replete with its emphasis on the privileged epistemic position we occupy from the first-person perspective. The scientific discoveries that particularly interest us presuppose that we have a way of identifying mental states that is independent of, and at least as reliable as, the ways we have of studying changes that take place in the brain. Indeed, if we couldn’t rely on a relatively unproblematic access to mental states of a sort that we have been presupposing for thousands of years, it’s hard to see how we could even get started in the study of the mind that interests us. Undoubtedly, we could learn all sorts of interesting things about brains and the processes that occur in them,28 but we wouldn’t even have a clue as to which processes were connected to conscious mental states and which were processes that could occur in the absence of any conscious life whatsoever. An analogy might be helpful. One of the more interesting and potentially valuable areas of research in biology is the genome project – an attempt to systematically inventory distinctions among human genes. But the interest of the project is, of course, fundamentally tied to an ability to correlate genes with various “macro” properties of human beings – properties we discover without the more exotic techniques necessary to discover gene types. Unless we can correlate genes with observable human characteristics like eye color, skin color, dispositions to behave, disease, etc., the study of genes loses much of 28

At least we can presuppose this for the moment. In fact, our only access to the external world, including the world of the brain, is through the very subjective experience that is known through introspection.

134

Ontological priorities: taking phenomenology seriously

its importance. Brain states stand to mental states as genes stand to the macro properties of human beings that we want to explain and predict. I said that any plausible conception of the empirical study of mind presupposes that we have an access to mental states that is prior to, and at least as good as, anything we ultimately come up with in cognitive science. If I was right in my earlier suggestion, we should go a step further and concede that the kind of introspective access I have to the intrinsic character of my own mental states is obviously superior to anything I could ever learn about mental states through the empirical study of the brain. Consider again the “ideal” situation of the neural scientist studying his own brain as he introspectively “surveys” his inner mental life. Years ago, David Armstrong (1963) argued against the incorrigibility of our first-person justification for believing that we are in pain. He asked us to imagine adding to our introspective justification the testimony of a scientist well-versed in some utopian cognitive science assuring us, as our brain is monitored, that we aren’t in the right brain state to be feeling pain. It never seemed to me that the argument had even the slightest force. If we imagine the scientist in question monitoring his own brain state as he introspects a stabbing pain, it seems just obvious that when he can’t find the expected brain state he will have no alternative but to reject his theory, or to construe as flawed some other aspect of the monitoring that is supposed to reveal the brain states. No sane person is ever going to turn over first-person knowledge of occurrent conscious mental states to the kind of necessarily indirect knowledge of mental states that might be achieved through empirical study of the brain.29 29

Later I’ll argue that it might be less problematic to defer to neurology when it comes to the question of whether there are occurrent, but unconscious, mental states.

135

Knowledge, Thought, and the Case for Dualism

Notice, by the way, that whether or not you agree with anything I said in the previous couple of paragraphs, it is surely the case that we’re not going to settle any controversies that arise there by turning the question over to empirical science. Perhaps I’m wrong in thinking that introspection of occurrent, conscious mental states would always trump any putative counter-evidence provided by empirical science. But if you are trying to convince me that I’m wrong I wouldn’t suggest bringing a cognitive scientist into the discussion. It is just not their job to answer this sort of question – they haven’t got the right sort of training to even address the question. It’s their job to tell us what’s happening to our bodies as we notice changes in our mental life. I have yet to draw any extravagant metaphysical conclusions from any of the above epistemological considerations. As we have already seen, there are all kinds of knowledge arguments for substance, property, fact, and event dualisms. Whether it is Descartes inferring that the mind is different from the body because we can know of our existence in a way different from the way in which we can know that physical bodies exist, or Jackson wondering if we can establish the existence of qualia by thinking about Mary discovering some radically different fact about sensation when she leaves the black and white room, philosophers have argued about whether we can reflect on differences between our knowledge of mind and body to reach conclusions about the separation of mind and body, or at least mental properties from physical properties. All such arguments are philosophically controversial, and we will be examining them more carefully in the rest of this book. But this much is absolutely certain. The cognitive scientist’s empirical research will result in no information that has any bearing on whether the knowledge arguments are good or bad. Indeed one might well wonder what

136

Ontological priorities: taking phenomenology seriously

philosophical issues, if any, could be settled by our utopian cognitive science. We left our researcher discovering all sorts of interesting correlations between mental states and brain states. (Our imaginary ideal researcher also has all sorts of well-confirmed theories about just what goes on in the brain.) As we noted in our discussion of functionalism, it may or may not turn out that the kind of brain states correlated with pain in people are quite different from the kind of brain states correlated with pain in various other animals. It might even turn out that the kind of brain states correlated with pain in some people under some conditions are different from the kind of brain states correlated with pain in other people under those same or under different conditions.30 Much, of course, will depend on what we count as an interesting difference in kind. But however we answer that question, we should surely recognize that at least these empirical results might bear on the plausibility of certain philosophical accounts of the mind – specifically socalled type-identity theories. That seems right to me, but it is hardly the case that the philosopher needed to wait on empirical research to settle the relevant philosophical issues. We would have to be stupid not to realize that it might turn out that creatures with interestingly different kinds of brains could still have the same mental states that we have. We didn’t need to consult cognitive scientists to reflect on the relevant modal question. All we needed to do was watch enough episodes of Star Trek to realize that we had better understand mental states in such a way that it is at least possible for the same mental state to have a radically different physical base.

30

And not just in the trivial sense in which we have different kinds of pains that would naturally correlate with different kinds of brain states. I’m talking about a specific kind of pain which might have one neural correlate on one occasion and a different neural correlate on another occasion.

137

Knowledge, Thought, and the Case for Dualism

It’s not that science can’t remind us of, or call attention to, modal truths that would be accessible without science. The surest proof of a possibility is an actuality. It would have been a lot harder to come up with a dream argument for skepticism if people didn’t dream, or an argument from the possibility of hallucination if people didn’t have non-veridical experience.31 But it would only have been more difficult. The possibilities upon which the arguments trade are real whether or not they ever materialize. The possibility that the neural activity associated with mental states differs from creature to creature, and even person to person, is (and was always obviously) real whether or not the possibilities materialize. Indeed, after all of the empirical data are in, we might ask again: What distinctively philosophical questions will be, or even could be, answered by science? What distinctively philosophical controversies will be advanced, let alone settled? Suppose that we have obtained exhaustive correlations of the sort described above. Are we any closer to answering any of the following? Can you even imagine any empirical research that would shed light on any of the following? (1) Are knowledge arguments for substance/property/fact/event dualism sound? (2) Are there important distinctions to be made between substance/ property/fact/event dualism? Is there such a thing as substance? What is the connection between differences between kinds of substances and kinds of properties exemplified by substance? What is the difference between a mental property and a physical property? What is the distinction between an event and a fact? (3) Is functionalism a plausible account of the nature of mental states? 31

It’s probably no accident that the preoccupation over a distinction between appearance and reality coincided with a science that concluded that the bread-boxed size objects around us have a quite different nature than anything we would have imagined

138

Ontological priorities: taking phenomenology seriously (4)

(5)

(6)

On a functionalist account of mind, should we identify mental states with that which “realizes” the functional state (i.e. plays the functional role) or should we identify mental states with the exemplification of the second-order functional property? Do we “fix the reference” of predicate expressions picking out kinds of mental states with definite descriptions that nevertheless do not capture the meaning of the predicate expressions, or shall we view terms for kinds of mental states as having the meaning of definite descriptions? If either, what are the relevant definite descriptions? Should we be internalists or externalists about the identity conditions for mental states? If brain states are intentional (representational) what makes them intentional? Is it some feature intrinsic to the states, or has it more to do with the causal origin of the states. If the latter, why were we trusting first-person reports of mental states in trying to correlate neural activity with, say, desires or fears? Why would anyone have privileged introspective access to the character of a mental state whose identity conditions involve facts that are clearly the purview of empirical science?32

Do any of the philosophers partnering with cognitive science think that they’ll get any useful information from empirical science that will help them answer any of the above questions? Don’t misunderstand me. Many of the answers to these philosophical questions will clarify important empirical questions that someone might be interested in answering. If you decide that functionalism is true, you will probably need to turn to empirical science to discover what takes the value of the variable used in the specification of the functional state. If you become really interested in that question, then by all means stop doing philosophy for a while and do (or consult) the relevant empirical research. Alternatively, you could view 32

See the papers contained in Ludlow and Martin (1998), and Nuccetelli (2003), all of which address this question – none of which even appeal to discoveries of cognitive science. I’ll have more to say about this issue in Chapter 7.

139

Knowledge, Thought, and the Case for Dualism

your task qua philosopher as finished once you’ve come up with your functionalist analysis, content then to let the chips fall where they may with respect to the hard-wiring of the brain or the nature of a Cartesian ego. If we decide that either mental predicate expressions, physical predicate expressions, or both have their reference fixed, or meaning given, by definite descriptions, and we can isolate the relevant definite descriptions, it may turn out that only the cognitive scientist can tell you whether or not the predicate expressions have an extension, and if they do what that extension is. I’ve already implied that I think it wildly implausible to suppose that our grasp of pain is somehow indirect (is “by description”), at least as we actually have an experience of pain of which we are consciously aware – we’ll talk about this issue in more detail shortly. I can’t see how it can possibly turn out that “is in pain” fails to pick out a property. Any view that allows for such a possibility is for that reason an implausible view.33 It’s hardly the case, for example, that I think of being in pain as the property I typically exemplify when I’m bleeding all over the floor, or the property exemplification of which causes me to grimace, complain, answer questions in certain ways, or what have you. If such a view were correct, then, to be sure, it would be a matter of empirical investigation as to what that property is, and it might turn out in humans to be a pattern of neurons firing. When I think of Jack the Ripper, I think of a person who is causally responsible for certain atrocities committed at the end of the nineteenth century and there are any number of people who might turn out to be the referent of “Jack the Ripper.” But I already know what property I’m talking about when I talk about pain. I don’t pick it out via 33

The possibility appealed to here is epistemic possibility, and again I’m thinking of what is epistemically possible for me as I actually find myself introspectively aware of pain.

140

Ontological priorities: taking phenomenology seriously

some metaproperty it exemplifies. As we will see in Chapter 5, this will be a critical point to remember in evaluating the idea that the physicalists can rely on empirical evidence in advancing their identity claims about mental states. In the next chapter we will be emphasizing that there must be properties like pain. It can’t be the case that everything I refer to, I refer to only through my grasp of properties exemplified by the referent. There are things and properties like that, but indirect reference is possible only because direct reference is possible. We need to prevent vicious regresses of reference just as we need to prevent vicious regresses of justification. Just as we need to recognize a kind of justified belief that doesn’t presuppose the having of other justified beliefs, so also we need a kind of reference that doesn’t presuppose other different successful reference. And just as we need to end an epistemic regress by recognizing that we can have direct access to truth-makers, so also we need to end a regress of reference by finding something we can confront in thought directly without having first to grasp some other different thing. And there is no better candidate for a property that is directly grasped than a mental property. In Chapter 6, we will discuss a view defended by Maxwell (1978), and inspired by Russell (1948), that our direct access to pain still doesn’t preclude an empirical identification of pain as a property of the brain. While it isn’t the slightest bit plausible to suppose that we think of pain only as that which exemplifies a certain property, it’s not nearly as implausible to suppose that our grasp of brain states and their constituents is indirect. At least it is not implausible if we reject epistemological and ontological direct realism.34 As I argued earlier in this chapter, the truth is that we have better and more direct access to our mental states than we ever could have to our brain states (or to 34

For a recent attempt to explain why we should reject direct realism, see Fumerton (2006b).

141

Knowledge, Thought, and the Case for Dualism

any other physical states of the world). Although the view is now widely rejected, I have also argued that we know the physical world through our five senses only as the potential cause of various sensations and connections between sensations. The brain we are “looking at” through the senses we know only indirectly. Indeed our access to the microstates of the brain involves multiple inferences. Even our knowledge of the presumed effects of brain states on various sophisticated instruments is indirect (again assuming that epistemological indirect realism about perception is correct). And even after we gain access to changes in our scanning equipment, we must employ inference heaped upon inference to get to what is ultimately going on in the physical world, changes understood as the ultimate cause of the way our instruments react. The upshot of all this is that, for all we know, it might be that the ultimate physical reality theorized in our speculation about the brain includes as constituents the very phenomenal properties that are the objects of introspection.35 Is such a view a kind of physicalism? It depends on what you mean. Certainly the idea that physical reality exemplifies phenomenal properties is compatible with a substance monism (on the assumption that we can make sense of a substance ontology). Are the phenomenal properties that might be constituents of the states of affairs postulated by neurology physical properties? It depends on what you mean by physical properties. If we want property dualism to be consistent with substance monism, we don’t want to define a physical property as any property of a physical substance. If we define mental properties in terms of our potentially direct access (our phenomenological access) to such properties, then we should probably embrace a property dualism according to which the world contains both 35

Again, whether this possibility is anything more than a mere conceptual possibility is a matter to be discussed more carefully later in this book.

142

Ontological priorities: taking phenomenology seriously

phenomenologically accessible properties and those that are not phenomenologically accessible. Again it is difficult for me to see how the plausibility of the above view could ever be settled by cognitive science. We will need to engage in philosophical argument of the sort contained in Chapters 6 and 7. Indeed, I think it would be difficult to get the empirical scientists even to understand the rather sophisticated epistemological, phenomenological, and ontological philosophical presuppositions that underlie the intelligibility of such a view. And if we ever could get them to understand it, they would probably rightly shrug their collective shoulders and leave it to the philosophers to argue over the wisdom of endorsing the relevant presuppositions. It is also worth emphasizing that the epistemological and metaphysical presuppositions needed to make sense of the above substance monism are anathema to most physicalists. The naturalistic/physicalist crusade to reduce the mental to the physical is driven by the idea that we have better knowledge and understanding of the physical than we have of the “problematic” and uncomfortably “mysterious” mental. But if what I asserted above is true, the tenability of any sort of physicalism rests on an epistemology that is committed to a view diametrically opposed to this idea. We have a better, and more immediate grasp of the mental than we have of the theoretical posit that is the physical world and its properties. If anything, we should be much more concerned to reduce the physical to the comfortable, more familiar world of the mental than we should be interested in reductions that move in the opposite direction.

3.9 is there any contribution science can m a ke to the ph ilo sophy o f mind? So far I have been more than a little pessimistic about the contributions science might make to the philosophy of mind. 143

Knowledge, Thought, and the Case for Dualism

And I truly am bewildered by philosophers of mind who seek philosophical insight from the empirical study of the brain. On many occasions I have tried, without much success, to get a coherent answer as to what they expect to learn. Again, this is not some sort of anti-science rant. The study of the brain is interesting – indeed it is fascinating. And it is also understandable that philosophers will occasionally feel the frustration of working in a field where the same debates continue without resolution for thousands of years. Still, philosophy is philosophy and you can’t get your philosophical questions answered by asking a different sort of question. But before we leave this topic, let’s try one last time to find some philosophical questions upon which science might cast some light? I can think of at least a couple.

(a)

Disjunctivism

As we saw earlier, in the philosophy of perception philosophers have argued about whether there is an interesting and phenomenologically accessible common denominator to both veridical perception and hallucination. Classical skeptical arguments invoke such a common core to suggest that commonsense beliefs in the external world involve bridging a gap between appearance and reality. To be interesting the common denominator must be something that varies from perceptual act to perceptual act. So, for example, almost everyone would agree that in some sense there is a “self” present in both veridical perception and hallucination, but the view that has come under attack in recent years is the idea that there are sense data or ways of being appeared to that are intrinsically alike in both a veridical experience and its hallucinatory “counterpart.” These sense data or ways of being appeared to were conceived by the tradition as truth-makers for propositions that are the obvious candidates for the premises, the evidence, 144

Ontological priorities: taking phenomenology seriously

upon which we must rely in building our case for truths about the physical world. The threat of skepticism from within such an ontological and epistemological framework is serious, and it is no wonder that philosophers have tried hard to find alternatives to the traditional picture. Contemporary “disjunctivists” try to deny the traditional picture often by embracing something like direct realism as an account of veridical perception, and “something else” (usually severely under-described) as what goes on in non-veridical perception. Here, I can see at least a role for cognitive science to play. On the traditional picture, conscious experience was always thought of as having as its immediate cause neural activity. Skeptical scenarios were easy to imagine – all we need to do is imagine “breaking into” the standard causal chain so as to produce the same subsequent links culminating in the same kind of brain state that is the direct cause of the experience. You understand the plot of The Matrix because you understand how a computer could, in principle, generate the neural activity that we think of as the immediate cause of sensation. In dreams or hallucination we simply imagine the occurrence of the same brain states present in veridical experience, but with a different distant cause. It might seem that this part of the traditional picture could be confirmed or disconfirmed by empirical science. At the very least we can, in principle, find out if the neural activity accompanying vivid hallucination is precisely the same kind of neural activity present in veridical perception (setting aside skeptical concerns about how to tell which perceptions are and are not veridical). Now it’s not clear that the empirical results would settle the controversy. Certainly the property dualist need not abandon the view that there is a common denominator should it turn out that the neural activity is quite different in the “good” and the “bad” case. The principle that different causes always have different effects is problematic at best, and one could insist that 145

Knowledge, Thought, and the Case for Dualism

the phenomenological evidence is so strong that if we discover the neural activity is different in the two cases, then we should conclude that we apparently have quite different brain states resulting in the qualitatively same sensation. By contrast, the principle – same cause; same effect – is far more plausible. On some views of causation (Hume’s or Mill’s, for example) it’s analytic. So one might be tempted to suppose that cognitive science can, in principle, shut the door on the disjunctivist’s hopes. But even here the issue is complicated. Many disjunctivists embrace relatively extreme forms of content externalism and also construe sensation as fundamentally intentional in character. On their view, the sensations of which we are aware are a kind of intentional state whose intentional character is determined by (perhaps even constituted by) its cause. So on such a view one could concede that the same neural activity occurs in both veridical and non-veridical experience, but claim that the neural activity is an intentional state with some external content only in the veridical case, and is something else entirely in the nonveridical case. It is easiest to make such a move only with the crudest (and most implausible) forms of content externalism – views according to which the actual cause of a neural event determines its content. Any content externalism sophisticated enough to allow for misrepresentation (and no content externalism worth considering seriously is unable to allow for misrepresentation) can’t deny to hallucinatory experience its status as an experience just because it doesn’t have the cause we take it to have. As always then, there are a host of philosophical issues we’ll have to settle before we decide on the significance of any empirical study of hallucination. But at the very least, I suspect it might become increasingly more difficult to embrace disjunctivism as the empirical evidence suggesting common neural causes for conscious experiences comes in. 146

Ontological priorities: taking phenomenology seriously

(b)

Unconscious mental states

Philosophers of mind have often argued about whether or not we can make sense of unconscious mental states. Here I’m not interested in whether one can provide halfway plausible behaviorist “translations” of talk about the unconscious. It seems relatively unproblematic to suppose that people can act as if they have certain desires, beliefs, and fears of which they are unaware. And it is equally obvious that other animals, even insects, behave as if they have a wide and rich array of mental states of which they might be unaware.36 The more interesting question is whether they can actually have occurrent desires, beliefs, and fears of which they are unaware, of which they are not conscious. The debate on this issue is sometimes confusing, partly for terminological reasons. One sometimes gets the feeling that philosophers use “occurrent mental state” as both the contrast term for “dispositional mental state” and as a synonym for “conscious mental state.” But we want to keep the concepts of occurrent mental state and conscious mental state distinct. On the most natural interpretation, a conscious mental state would be thought of as a mental state of which we are conscious, of which we are aware. And on the most natural view, awareness is a kind of relation that we bear to that of which we are aware. So if my pain, for example, is a pain of which I am conscious, then I stand in a relation of being aware of my pain. And that naturally raises the question of whether the pain of which I am aware can continue (with precisely the same 36

We earlier pointed out that on the acquaintance theory we should distinguish being in pain, for example, from being aware that one is in pain, and both of these from believing that one is in pain. It would certainly be nice if cows, chickens, and the like didn’t feel pain (for those of us who enjoy steak and chicken wings). It doesn’t seem likely, but there might be more hope that even if they feel pain, they aren’t aware of the pain that they feel (even if the pain still manifests itself in various sorts of behavior).

147

Knowledge, Thought, and the Case for Dualism

intrinsic character) even when I am no longer aware of it. Whatever we say of pain presumably will apply to other occurrent mental states of which we are aware. Earlier, I tried to ostend acquaintance by asking you to consider pain of which we become unaware. Let’s consider that hypothetical situation again. I suggested that most of us are familiar with having a headache or a backache and becoming so engrossed in an interesting conversation or activity that we no longer notice the pain. I conceded that there are, of course, two descriptions of the situation. On the one, the interesting conversation causes the pain to stop for a while. On the other, the interesting conversation only causes awareness of the pain to stop. Again, one might wonder if we couldn’t bring to bear the resources of neurology to help decide the issue. To be sure, the relevant experiments would be more than a bit difficult to perform. Ideally, we would scan someone’s brain who was complaining about pain, and continue to scan the brain as we try to distract the subject in some way. But again let us imagine some wildly successful utopian cognitive science. And let us suppose that we have already been dramatically successful correlating pain with certain patterns of neurons firing. There is also some neural activity in a different region of the brain and we notice that it picks up in intensity when we ask our subjects to focus their attention on their pain. We naturally wonder whether this second sort of brain activity might have something to do with our “monitoring” of the pain. Now sure enough, when we distract our pained subject as we monitor her brain we find that the neural activity associated with the pain continues in precisely the same way as before, but that the neural activity tentatively associated with the monitoring of the pain ceases entirely. I suppose I would take that to be rather strong empirical evidence that the very pain of which we were aware sometimes continues even when we are not aware of it. The available empirical evidence surely 148

Ontological priorities: taking phenomenology seriously

wouldn’t entail this result – the Quine–Duhem thesis37 is almost obviously correct when it comes to the interpretation of these sorts of empirical data – but the evidence surely might be more than a bit suggestive.38 So science could shed light on at least this sort of philosophical controversy, right? Perhaps. But to be honest I’d be more inclined to use the above thought experiment as evidence that the empirical question of whether there are unconscious occurrent mental states isn’t really philosophical. I’m inclined to think that we are done with our philosophizing when we do the kind of thing I tried to do in the last few paragraphs. Once we figure out what the kind of evidence science could or couldn’t bring to bear on a question, we’ve finished our task qua philosopher. We might be interested in how things turn out after that, but if we are even moderately well-adjusted inquisitive people, we are interested in the answers to questions that go beyond philosophy.

3.10

summary

In this chapter I first tried to lay out the basics of the acquaintance-based foundationalism that I bring to an investigation into the nature of mental states. That discussion paved the way for a thoroughgoing rejection of the naturalist’s approach to the study of mind. I have no objection to cognitive science, but I’m convinced that it has no light to shed on the fundamental philosophical questions about the nature of mind – questions that are the focus of this book. 37

38

The basic idea behind the thesis is that we can always make our favored theory consistent with observational data provided that we are willing to make enough adjustments in our auxiliary hypotheses. We would still need to be careful to distinguish awareness of pain from belief that one is in pain. So our hypothetical researcher might find it possible to distinguish three “levels” of neural activity – one corresponding to pain, another corresponding to awareness of pain, and yet a third corresponding to belief that one is in pain.

149

4 Knowledge arguments revisited

In the last chapter I outlined a specific view about the nature of non-inferential justification and knowledge. I also discussed in a very general way the implications of such a view for our knowledge of our mental states. Specifically, I argued that we often have non-inferential knowledge that we are in certain mental states. I also suggested that a metaphysics that defers to the discoveries of science has epistemology completely backwards. The knowledge we have of the existence and intrinsic nature of our mental states is far more secure than what typically falls under the heading of scientific knowledge. Indeed all scientific knowledge of the physical world (including the physiology of our brains) is utterly parasitic upon this more secure knowledge we have of our subjective states. Let us now return to the knowledge arguments sketched in Chapter 1 and see precisely what epistemological conclusions we can reach and use as premises in a knowledge argument for the rejection of physicalist accounts of mental states. As you will recall, the Mary who was deprived of color experiences was energetically and successfully investigating the physical world in general, and the workings of the brain in particular. We stipulate that color-deprived Mary came to know all of the following: (1) She knows that people have experiences they would describe using color expressions. She knows that people say things like “this looks red” or “it looks to me as if this is red.” She has even

150

Knowledge arguments revisited

(2)

(3)

(4)

occasionally run into philosophers who describe themselves as adverbial theorists and who insist that whether or not they are dreaming or hallucinating they are quite sure that they are appeared to redly! She knows how to classify all sorts of physical objects – she recognizes such mundane objects as apples and shirts, and she knows that people describe them using expressions like “red.” She has theoretical knowledge of light and knows the differences between the way in which the things people call “red” reflect and absorb light. She can consequently develop concepts that are co-extensive with the color concepts we apply to physical objects. She has extensive knowledge of what goes on in the brains of people when they say that something looks red. She knows that the relevant brain states are typically caused by those objects people call “red” and she knows that those objects include such things as apples and tomatoes. She also knows that the very brain states people are in when they say something looks red occur sometimes in the context of what she recognizes as hallucination. She can consequently develop concepts that are coextensive with concepts such as the concept of looking red that people apply to their experience.

As Jackson describes Mary’s hypothetical situation, she knows everything there is to know about the physical world. But, for reasons that will become clear, we cannot stipulate that without begging important questions. We cannot stipulate, for example, that Mary knows everything that happens in someone’s brain as various objects look red to that person. There is, of course, a trivial sense in which we can’t say that Mary knows all truths about brain states, at least if we include among the truths about brain states propositions describing their relations to other states of affairs. The property dualist who employs the knowledge argument will not allow that Mary knows that as brains undergo certain changes, experiences of a certain sort occur! Shall we stipulate that Mary 151

Knowledge, Thought, and the Case for Dualism

knows all of the intrinsic properties exemplified by brain states? No, because that will again beg certain questions concerning the neural “panpsychism” discussed by Maxwell (though not under that label), Chalmers, Stoljar, and others, a view that we will need to address carefully in more detail later. One might want to allow that phenomenal redness is itself a property, indeed an intrinsic property of some physical process occurring in the brain. But on the most plausible version of the view that locates phenomenal properties in the brain, it may be the case that through her paradigmatically scientific investigation of the physical world, Mary never came to know any of the intrinsic properties of the brain, or, for that matter, any other physical object. Here’s another way of characterizing the knowledge that we are giving color-deprived Mary. We are simply presupposing that Mary gets all knowledge of truths describing the workings of the brain and its relation to the physical world that she will ever or could ever get investigating the brain in her colordeprived state. But even here we must be careful because the modal operator can be interpreted in a number of different ways. In Chapter 1, I conceded that even a radical empiricist like Hume should allow that it is conceivable that someone come to have the idea of phenomenal redness without ever having the relevant experience to which the idea corresponds. One might suppose that a crude content externalist won’t allow that someone could have an idea of redness without experience of the exemplification of that property, but any content externalism that crude will be hopelessly implausible. Content externalists typically allow that a brain state can acquire its representational content via rather circuitous causal chains that extend indefinitely into the past.1 1

So, for example, content externalists emphasizing the importance of causal connection to representation would probably insist that a name or, for that matter, an idea,

152

Knowledge arguments revisited

But as we noted earlier, even content internalists should be cautious before they embrace the view that it is metaphysically impossible, or inconceivable, that someone should develop an idea of, say, a red appearance, without actually having had the experience. It depends, in part, on how the internalist understands having an idea. At least some content internalists seem to understand having an idea of F, where F is a property, as having F “in” mind. An acquaintance theorist, for example, who is a realist about properties, might argue that to have the idea of being appeared to redly is just to stand in a relation of direct acquaintance to the property. Presumably, that’s not the same thing as experiencing the property, and if it isn’t, it is hard to see how one can conclude on a priori grounds that it is impossible to stand in such a relation to the property without having actually experienced the property. If it involves no contradiction to suppose that someone could acquire the idea of phenomenal redness without having the relevant experience, it would also involve no contradiction to suppose that such a person could also entertain propositions asserting that people exemplify such properties when their brains are in certain states. Then we need only conjure up an imaginary situation in which that person gets inferential justification strong enough for knowledge to believe that proposition. We could imagine, for example, the person relying on an oracle who proclaims that what the person currently believes is true, an oracle that is in fact trustworthy, and that the person has extraordinary good evidence to accept as trustworthy. David Lewis (2004, pp. 78–79) makes a similar point: can represent Caesar in virtue of the use of the name or the occurrence of the idea being the last link in a long causal chain that began with Caesar’s being called something or Caesar’s having done something. Putnam (1975) and others also emphasize the importance of a division of linguistic labor. I can successfully use a name, but also deploy an idea to pick out something or someone, where I got the name or idea from someone else.

153

Knowledge, Thought, and the Case for Dualism Having an experience is surely one good way, and surely the only practical way, of coming to know what that experience is like. Can we say, flatly, that it is the only possible way? Probably not. There is a change that takes place in you when you have the experience and thereby come to know what it’s like. Perhaps the exact same change could in principle be produced in you by precise neurosurgery, very far beyond the limits of present-day technique. Or it could possibly be produced in you by magic. If we ignore the laws of nature, which are after all contingent, then there is no necessary connection between cause and effect: anything could cause anything. For instance, the casting of a spell could do to you exactly what your fist smell of skunk would do. We might quibble about whether a state produced in this artificial fashion would deserve the name ‘knowing what it’s like to smell a skunk’, but we can imagine that so far as what goes on within you is concerned, it would differ not at all.

So when we imagine what color-deprived Mary comes to know, we are probably just imagining her acquiring the knowledge we believe that she would acquire, as a matter of contingent fact, if she were to use all of the techniques for acquiring knowledge available to her. If this seems a bit vague, it doesn’t really matter for, in the final analysis, we don’t need to worry about making all of this much more precise. In particular, we don’t need to have some principled way of characterizing everything Mary came to know through her scientific study of the world.2 Remember that in Chapter 2 we decided that it might be best to understand the property dualist as someone who rejects various proposals put forward by self-proclaimed physicalists. So, for 2

So Dennett (1991, pp. 398–406) complains about the sheer impossibility of conjuring up in our imagination the hypothetical situation discussed in the Mary thought experiment. Mary is described as knowing everything there is to know about the physical world. Dennett seems to think we couldn’t even imagine what it is to know everything there is to know about the physical world. I’m trying to argue here that Dennett’s complaint misses the mark.

154

Knowledge arguments revisited

example, our property dualists will reject the suggestion that there is no more to being in pain than having a brain whose c-fibers are firing. Our property dualist will reject the suggestion that there is no more to being in pain than being in a state, whatever it is, that results from damage to the body and typically causes behavior conducive to healing. Consider every proposed physicalist reduction where the left-hand side of the identity claim employs some predicate expression of folk psychology and the right-hand side employs the paradigmatic predicate expressions favored by the physicalist in question. In effect, we are granting that color-deprived Mary has discovered all of the truths expressed in the physicalist’s analysans. In the case of the eliminativist, we are imagining Mary coming to know all of the truths that the eliminativist thinks will “replace” the “theory” of folk psychology. That’s what color-deprived Mary knows. The next step in the knowledge argument, however, focuses on what colordeprived Mary doesn’t know. And perhaps the most promising approach to characterizing the knowledge that she is supposed to lack is by thinking about the kind of knowledge that Mary gains with her new ability to experience color. Jackson describes this new knowledge as knowing what it is like to experience red, for example.3 And Nagel (1974) talks about the subjective character of experience, contrasting our idea of that with the “schematic” understanding that is the only understanding that would be available to someone who never had that kind of experience. But while it seems harmless enough to describe her new knowledge that way, there is much disagreement on how to gloss more precisely what that sort of knowledge is. Virtually every philosopher 3

He says that upon her release “she will learn what it is like to see something red, say” (Jackson, 1986, p. 291). As we’ll see in a moment, I would prefer to describe the relevant experience in terms of seeming to see something red.

155

Knowledge, Thought, and the Case for Dualism

discussing the knowledge argument agrees that something rather dramatic changes in Mary when she experiences color for the first time. They don’t agree, however, on how to characterize the change. And, in particular, they don’t agree on how to characterize the dramatic change in Mary’s epistemic position. This much seems unproblematic. When Mary experiences color for the first time Mary’s brain undergoes certain kinds of changes it had never before undergone. All but a wildeyed eliminativist also agree that objects look red (and other colors) to Mary for the very first time. As Lewis (1988, p. 82) points out, it follows trivially from these admissions that Mary can come to know for the very first time that objects look red to her, and that her brain has undergone the relevant changes. Knowledge is factive and Mary couldn’t have known those propositions describing herself when they weren’t true. No physicalist is interested in denying these truisms. To characterize an interesting truth that might be of use to the property dualist we must move to a claim about property exemplification. So the basic idea is that when Mary acquires color vision she comes to know that a certain kind of property is exemplified in the world, one of which she was previously ignorant. She knew about the properties exemplified by brain states. She had successfully discovered causal connections between various states of the world and changes in the brain. She knew that people were disposed to use the words “red” and “looks red” under various conditions, and she could identify conditions that typically trigger the dispositions to manifest the linguistic dispositions. Whether or not she exemplified these properties herself, she knew that those kinds of properties under those descriptions have been exemplified. She knew, in other words, that there are properties that have certain properties, just as I might know that there is a person 156

Knowledge arguments revisited

who is the tallest person in the world. But just as I don’t know who that tallest person is, so also Mary in her color-deprived state never knew that there was this property of looking red that people exemplify when they are confronted by physical objects of a certain sort under the right conditions – that’s the heart of the dualist’s claim. Now as everyone knows, there has been strong resistance to this way of characterizing Mary’s new propositional knowledge. And we need to very briefly survey the main strands of this resistance.

4.1

th e ab i l i t y h y p o t h e s i s

Nemirow (1990, 2007) and Lewis (1988), among others, argued that the non-trivial change in Mary’s knowledge resulting from her newly acquired color experience is best understood in terms of a newly acquired ability – a newly acquired know how rather than newly acquired propositional knowledge. So Nemirow (1980, p. 475) tells us that: As for understanding an experience, we may construe that as an ability to place oneself, at will, in a state representative of the experience. I understand the experience of seeing red if I can at will visualize red.

And Lewis (2004, p. 98) describes the change that occurs when you have a new experience (in his example, tasting Vegemite) this way: If you have a new experience, you gain abilities to remember and to imagine. After you taste Vegemite, and you learn what it’s like, you can afterward remember the experience you had. By remembering how it once was, you can afterward imagine such an experience. Indeed, even if you forget the occasion itself, you will very likely retain your ability to imagine such an experience.

157

Knowledge, Thought, and the Case for Dualism

He goes on: Further, you gain an ability to recognize the same experience if it comes again. If you taste Vegemite on another day, you will probably know that you have met the taste once before. And if, while tasting Vegemite, you know that it is Vegemite you are tasting, then you will be able to put the name to the experience if you have it again. (p. 98)

And: As well as gaining the ability to remember and imagine the experience you had, you also gain the ability to imagine related experiences that you never had. After tasting Vegemite, you might for instance become able to imagine tasting Vegemite ice cream. (p. 99)

That there is a general distinction to be drawn between knowing that . . ., and knowing how . . ., seems relatively uncontroversial. Unfortunately, I can have extensive propositional knowledge concerning truths about what would allow me to drive a golf ball 300 yards without knowing how to drive a golf ball 300 yards.4 And at least some sorts of know how don’t seem to involve any sort of propositional knowledge at all. Mice know how to swim, and they have that ability even if they are completely unable to entertain truth bearers. But what abilities could possibly exhaust the new knowledge that Mary acquires upon gaining color experience? Lewis suggests that when Mary acquires color vision, she comes to know how to imagine color experiences in their absence, to remember what they are like, and to recognize and re-identify 4

Though even here I think the language is ambiguous. As someone continues to instruct me on how to get more yardage out of my drive, I might snap back “I know how to do it – I just can’t get myself to follow the instructions.” Relying on these sorts of examples Stanley (2011) suggests that we might be able to reduce knowing how to knowing that. If one could, that would throw a bit of a monkey wrench into those who try to advance the ability hypothesis as an alternative to conceding that Mary acquires knew propositional knowledge.

158

Knowledge arguments revisited

them in new ways. Conee (2004, p. 200) argues that the abilities in question are neither sufficient nor necessary for the kind of knowledge we have in mind when we characterize Mary as coming to know what experiencing red is like. Consider imagination. Earlier we conceded that it is at least conceivable that someone could come to acquire the thought of redness without experiencing redness. Such a person would certainly know more about redness than color-deprived Mary did (as we were imagining her), but, Conee argues, she would still learn something different when she gets her first introspective awareness of a color experience – it is only then that she would fully know what it is like to have the experience. Now I’m not sure how much force this argument has. A great deal depends, I suspect, on the nature of this hypothesized ability to imagine. We can certainly sympathize with Conee’s general observation that there is a world of difference between thinking of an experience and actually being directly aware of the qualities of that experience. If that weren’t true, we could save a great deal of money on our wine bill. We could get the wonderful flavors and aromas of the expensive cabernet just by concentrating our imagination on the relevant experience. But we also remember how the truly vivid imagination of a sour taste sensation can literally make our lips pucker. In any event, we are trying to get at what is involved in Mary’s coming to know what looking red is like. That, presumably, is something that, at least typically, Mary carries with her after the color sensation is over, and if that is true, it fundamentally has to do with the thoughts that Mary can entertain, and the justified beliefs she can have about those thoughts being true. If one can know what a given experience is like long after one has been acquainted with the experience, then it is not clear to me that the relevant knowledge is constituted by an act of direct acquaintance with the relevant experience. 159

Knowledge, Thought, and the Case for Dualism

Conee’s arguments against the necessity of other abilities for knowing what an experience is like are more persuasive. Although this requires a more exotic thought experiment, we could imagine a person who is locked in a specious present, who lacks the ability to remember or even to re-identify a color experience after she has it. Still, while she has the experience she knows what it is like, and she would know that for the first time even if she were cursed-Mary whose newfound ability to experience color comes with the unfortunate loss of short-term memory. It seems to me that there is a more fundamental problem with the ability theorist’s suggestion for how to understand the new knowledge that comes with Mary’s color experience. In one clear sense the suggestion seems to be incompatible with the position of the physicalist seeking to identify the property of having a color experience with some physical property. We can illustrate the point with a crude typeidentity theory that seeks to identify having the color experience with r-fibers firing, but the same point will apply mutatis mutandis to any other proposed physicalist reduction of the mental to the physical. We have stipulated that Mary knows that r-fibers fire in people’s brains. We also stipulate that Mary can recognize, imagine, identify and re-identify these states of a person’s brain. So if looking red is nothing more than r-fibers firing, she already had even before her color experience the abilities that Nemirow and Lewis say constitute knowing what a given property is like. Nemirow and Lewis would need to argue that Mary with her newfound color experiences has a new way of imagining and recognizing color experience. But that view will morph into the view that Mary acquires new representations of color experiences, a view we will discuss in more detail below.

160

Knowledge arguments revisited

4.2

ge t t i n g a c q u a i n t e d w it h c o l o r ex p e r i e n c e

Churchland (2004) and Conee (2004) both appear to argue that the relevant change in Mary is a new relation she comes to bear to a property about which she previously had only propositional knowledge. When she experiences red for the first time, she becomes directly acquainted with redness – the very redness that she successfully discovered and successfully described through her empirical investigations of the brain and its relation to its environment. While they both talk about acquaintance, it seems pretty obvious that Churchland and Conee aren’t thinking of this acquaintance in anything like the same sort of way. Churchland’s “acquaintance” with redness is really nothing more than the having of the experience. The critical neural change that occurs in Mary with that experience, according to Churchland, is a triggering of all sorts of complex dispositions to discriminate more finely colors (2004, especially pp. 166–168). It seems to me, then, that Churchland’s view is best thought of as a version of the new ability response to the knowledge argument. I don’t know the science to which Churchland appeals and, frankly, I don’t really think that it has any relevance to our discussion. We can imagine a hypothetical color-deprived Mary who is able to discriminate as finely as you like the physiological conditions that correspond to subtle differences in color experiences. If there is any power to the knowledge argument it won’t be affected by this fact. The property dualist is still convinced that discriminating the physical states is radically different from discriminating the mental states. When Conee talks about acquaintance he probably has in mind something close to the kind of relation I talked extensively about in characterizing non-inferential justification. When one experiences color in the context of veridical 161

Knowledge, Thought, and the Case for Dualism

experience, a dream, or a hallucination, one does typically stand in a relation of direct awareness to the state one is in. When one experiences pain one is directly aware of the pain.5 However much one visually “peers into” the brain, even one’s own brain, one can never in that way become directly aware of the pain state, though one can have propositional knowledge that the person whose brain one is investigating is in pain. But, Conee argues, direct acquaintance is not a kind of propositional knowledge. Now as I tried to make clear in Chapter 3, I do agree with Conee that direct acquaintance, by itself, is not a kind of propositional knowledge. It has always been unfortunate that Russell (1910, 1967) talked about knowledge by acquaintance and knowledge by description. I don’t think that even Russell wanted the acquaintance about which he spoke to be a kind of knowledge. Rather he wanted it to be a source of knowledge. The expression “knowledge through acquaintance” would better capture the idea he was after. Even though acquaintance itself is not a kind of propositional knowledge, knowledge achieved through acquaintance and knowledge through description are both species of propositional knowledge. They are distinguished by the nature of the justification we possess for the relevant belief. When I know through acquaintance that I am in pain, I am directly acquainted with the pain. But as Sellars (1963, pp. 131–132) pointed out, we haven’t yet introduced into the picture a truth bearer. We haven’t yet introduced into the picture propositional knowledge. But, I would argue, when one is directly acquainted with the fact that one is in pain, a fact constituted in part by the property of being in pain with which one is also 5

Or at least that’s what usually happens. I argued earlier that one should distinguish, even if only conceptually, feeling pain from being directly acquainted with the pain that one feels.

162

Knowledge arguments revisited

acquainted, while one has the thought that one is in pain and while one is directly acquainted with the correspondence between the thought and its truth-maker, then one possesses all of the justification necessary to convert the thought that one is in pain into knowledge that one is in pain (provided that the relevant belief is based on the justification one possesses). That is precisely how the property dualist conceives of Mary’s new propositional knowledge. Conee is certainly right in pointing out that Mary does become acquainted with color experience for the first time. But, I would argue, that acquaintance gives her non-inferential justification for believing that she exemplifies the property with which she is acquainted. She didn’t know the proposition that phenomenal redness is exemplified in the world before she became directly aware of that property’s exemplification in her own case. (We need to be careful here. The property dualist should admit that Mary could have known other propositions whose subject concepts denote phenomenal redness, but those subject concepts do not pick out the relevant property in terms of its intrinsic character.6) I spent so much time in the last chapter detailing the account of acquaintance-based knowledge because I think that one can’t even get a feel for the force of the knowledge argument without something like that view of knowledge in place.7 If one embraces various versions of externalism, one can mimic the acquaintance theorist’s characterization 6

7

And again we need to be careful. When I talk about the intrinsic character of a property, I’m not talking about a property of the property. The intrinsic character of a property is just the “what it is” to be that property – it is probably what proponents of the knowledge argument have in mind when they talk about the “what it is like to X.” In Chapter 7, I’ll consider again not just which presuppositions are sufficient to get the knowledge argument off the ground, but which are necessary. Strictly speaking, knowledge by acquaintance is necessary, I think, to reach the conclusion that mental

163

Knowledge, Thought, and the Case for Dualism

of Mary’s epistemic situation without getting any immediate metaphysical dividend. So a process reliabilist, for example, might understand a non-inferentially justified belief as one produced by the kind of process that Goldman (1979) once described as a belief-independent, unconditionally reliable process. There is nothing problematic about the suggestion that the very same proposition can come to be believed via different methods. Of course, that is true on any plausible account of justification. I can have testimonial evidence that Chicago won the game, while I also saw the game on TV. But on classical versions of foundationalism, there is something special about the kind of justification provided by direct access to the truth-maker for your belief. The acquaintance that is the source of non-inferential justification gets the relevant aspect of the world directly and unproblematically before consciousness. For the reliabilist, introspection will be understood as a causal process that moves from the mental state to the belief that one is in that state. From the perspective of the believer the resulting belief gets one no closer to the truth-maker of one’s belief than if one were directly caused by God to believe that one is in the mental state. That the reliabilists are wrong in their characterization of introspective knowledge seems obvious to me (though the underlying philosophical issues are obviously a matter of some controversy). Compare the hypochondriac who is caused by a belief-independent process to believe that he is in pain (when there is no pain and no direct awareness of the pain) with the person who believes that he or she is in pain while being directly aware of the pain. Wouldn’t there be an obvious difference from properties are exemplified. The foundationalism of thought, about which I’ll say more later, can by itself get one knowledge of the conditional claim that if there are mental properties they are something over and above paradigmatic physical properties.

164

Knowledge arguments revisited

the internal perspective between the two belief states? And how can one account for the difference if the believer has no access to the cause of the respective belief states? From the perspective of the believer, after all, the causes of the belief are indistinguishable, if only in the sense that the believer is quite ignorant of the causal process that yields the belief.

4.3

ma r y ’ s ne w b e l i e f s

Evaluation of the knowledge argument focuses on Mary’s purported new propositional knowledge. And that’s perfectly natural since the property dualists will claim that Mary comes to know a proposition of which she was ignorant in her colordeprived state. But as we saw in Chapter 1, if all we are interested in is an argument for the conclusion that the proposition that one is appeared to in a certain way is distinct from various propositions describing one’s physical states, one actually needn’t worry about what Mary knows or how she knows it. If Mary with her new color experiences comes to believe a kind of proposition that she didn’t believe in her color-deprived state, then we can conclude, on the basis of Leibniz’s law, that the propositions describing the phenomenological character of color experience are quite different from any propositions describing physical reality. As we will see in the next chapter, an externalist about belief content might be able to claim that even if Mary doesn’t realize it, the proposition she comes to believe about the phenomenological character of a subject’s color experience is the very same proposition that she believed when she reached conclusions about the physiological states of the subjects who describe themselves as experiencing color. But such a philosophical move comes with an enormous cost. One will end up having to admit, implausibly, that a subject has no privileged access to 165

Knowledge, Thought, and the Case for Dualism

what that subject believes. Mary can think that she believes two quite distinct propositions even when there is, in fact, only one proposition believed. Again, we’ll need to examine such views more closely in the next chapter. Without a radical externalism about content, it is difficult to resist the conclusion that when Mary acquires color vision, she also gains new concepts, concepts that allow her to entertain new propositions. Nida-Rümelin (2004) makes the point nicely when she asks us to consider the following sort of thought experiment.8 Color-deprived Mary, you will recall, has all sorts of beliefs about changes that occur in people as they are confronted by ordinary objects like tomatoes. She knows that as light reflects off the surface of the tomato it effects changes in the eyes of an appropriately situated observer. She also knows that the retinal changes in turn bring about complex neural changes, and she can describe in considerable detail precisely what those neural changes are. Furthermore, she knows that people will typically say, when asked, that the object looks red to them under those sorts of conditions. Now suppose that after she gains color experience, an investigator working on arguments for dualism confines Mary to the infamous black and white room and shows her two slides, one a vivid red and one a vivid blue. There is the way the red slide looks to her and the way the blue slide looks to her, and Mary is asked to speculate on which of those two ways tomatoes looked to the subjects she studied in her color-deprived state. To facilitate discussion, Mary is asked to call the way the red slide looked, looking phenomenally R, and the way the blue slide looked, looking phenomenally B, and let’s suppose that for all her extensive scientific knowledge Mary has no common sense and doesn’t clue in to the fact that the investigator’s suggestion that we use 8

I change the example slightly so that it fits my description of Mary as moving from being color blind to having color experiences.

166

Knowledge arguments revisited

the label “phenomenally R” might be tied to that redness everyone keeps talking about. In fact, let’s suppose that, for whatever reason, Mary has a strong hunch that tomatoes look phenomenally B. She is then shown tomatoes and she realizes that her hunch was wrong – she had a false belief. The thought experiment seems to suggest strongly that Mary must have acquired a new concept when she was shown the slides. She might have used the word “red” before in characterizing the kind of experiences people have upon seeing tomatoes. And as she used that term, she obviously knew perfectly well that tomatoes look red – i.e. they look the way tomatoes look when people say that they look red. But she obviously didn’t know that tomatoes look phenomenally R, indeed she explicitly formed the belief that they look phenomenally B and not phenomenally R. Her new concept of looking phenomenally R is a different concept from whatever concept of redness she deployed before. Now again, this relies on the intuitively plausible suggestion that Mary should know if she came to believe something about the appearance of tomatoes that she didn’t believe in her color-deprived existence, and however obvious that might seem to some, it has become an enormously controversial claim. Post-Kripke, a great many philosophers will claim that there is no distinction between the meaning of certain sorts of expressions and their referents or extensions. The meaning of “Hesperus,” the slogan goes, is its referent (the planet Venus), and the meaning of “Phosphorus” is its referent (also the planet Venus), and therefore the meaning of “Hesperus is Phosphorus” is the same as the meaning of “Hesperus is Hesperus.” If the proposition asserted by a sentence is a function of what the sentence means, then while there are two sentences, those two sentences express only one proposition. Undoubtedly, astronomers thought that they were making a significant discovery when they learned that Hesperus is 167

Knowledge, Thought, and the Case for Dualism

Phosphorus, but they didn’t really come to learn a new astronomical truth. After all they already knew that Hesperus is Hesperus. At best they learned something about the truth of sentences. They learned that the sentence “Hesperus is Phosphorus” expresses a truth. To state the view clearly comes close to offering a reductio of it. The enormous attraction of Russell’s view that ordinary names have the meaning of definite descriptions is the ease with which the view explains precisely why the sentence “Hesperus is Phosphorus” doesn’t make the same assertion as does “Hesperus is Hesperus.” And despite the cartwheels performed by direct reference theorists, it seems to me that they never have explained plausibly the data that Russell’s theory explains.9 The direct reference theorist’s view about proper names was also extended to common nouns – names for kinds. “Water is H2O” expresses the same proposition as does “Water is Water.” To be sure, people didn’t know that until relatively recently. They didn’t even have the expression “H2O” in their vocabulary until relatively recently. And even after some people acquire the use of both “water” and “H2O” they might not have realized that the expressions have the same extension – pick out the same kind of stuff. But if meanings “ain’t in the head” (according to Putnam’s famous slogan from 1975), there is no particular reason to draw any inferences about whether or not two expressions mean the same thing from the fact that people who use them don’t realize that they 9

It is not as if they haven’t tried. Some introduce something quasi-meaning-like standing between the term and its referent. They don’t want it to be full-blooded meaning lest the view collapse into the Russellian view. But they also can’t live with the fact that “Hesperus is Phosphorus” is no more significant than “Hesperus is Hesperus.” If they don’t introduce something mental that distinguishes the two sentences, the only other real alternative is to go metalinguistic, and claim that the significance of “Hesperus is Phosphorus” is a function of what we learn about words and sentences. One could, of course, allow that mental states, like images, for example, accompany sentences without constituting the meaning of those sentences.

168

Knowledge arguments revisited

mean the same thing. Again, we will need to discuss all this in more detail in the next chapter.

4.4

different propositions; same t r u t h - m ak e r

Let’s assume for now that proponents of property dualism have made a powerful case for the conclusion that the proposition or the thought that someone seems to see something red is distinct from all of those propositions or thoughts whose content concerns the physiology of subjects who have color experiences. As we saw in Chapter 1, it’s not clear that this is going to get the property dualist any metaphysical conclusions. Remember that Smart, writing before the rise of meaning externalism, was quite willing to give the dualist the idea that the meaning of “I am in pain” is quite distinct from the meaning of “My brain is in state X.” Nevertheless, Smart thought that we can quite plausibly construe those two sentences with their different meanings as describing one and the same reality. From the fact that I represent the Eiffel Tower in my painting, while you represent the tower quite differently in your statue, it surely doesn’t follow that there is more than one tower represented in these quite different ways. Why should the medium of thought be any different? Why can’t one concede that when Mary gains color experience she represents the world in a way quite different from the way she represented it in her colordeprived state, while insisting that what she represents in these two quite different ways is precisely the same reality? Now it does seem to me that the physicalist shouldn’t be too quick to allow that Mary gains a radically different way of representing color experience after she experiences color for the first time. There are, after all, plausible theories of truth that tie the nature of truth bearers to the nature of truthmakers. The most dominant conception of truth in the history 169

Knowledge, Thought, and the Case for Dualism

of analytic philosophy is the correspondence theory of truth. There are many different versions of the theory. The correspondence theorist needs to give us an account of truth bearers, truth-makers, and the relation of correspondence that holds between truth bearers and truth-makers. Critically, one must also allow for the possibility of false propositions. Most correspondence theorists will distinguish primary and secondary bearers of truth value. On my own view, for example, sentence tokens can certainly be described as true or false, but only derivatively. I have argued elsewhere (2002) that the primary bearers of truth value are best construed as thoughts. And the most obvious candidates for truth-makers are facts. As we said in Chapter 1, a fact is constituted by the exemplification of properties – monadic, relational, or both. While it would certainly be nice if we could offer an informative analysis of correspondence, the prospects are, I believe, dim. The best hope would be an account that piggybacks on Dretske’s (1981) naturalistic account of representation. The crude idea behind such accounts is that a given kind of state X of an organism represents some property exemplification Y when under normal conditions Y causally explains the occurrence of X, X is strongly correlated with Y (as the rings of a tree are correlated with its age), and X leads to an evolutionarily advantageous interaction between the organism and Y, one that continues to explain the relevant goal-directed behavior whenever X occurs (with or without its usual cause, Y). Of course, one must allow for the possibility that an X state occurs without Y – one must allow for misrepresentation. One sort of misrepresentation seems to be handled, in principle, by the Dretske-style account. Some state other than Y can cause the occurrence of X, while X still represents Y in virtue of the causally explanatory role Y plays in connection with the past reinforced behavior caused by X. But naturalistic accounts of representation must explain much more radical misrepresentation. Humans have the 170

Knowledge arguments revisited

capacity to represent kinds of things with which they have never causally interacted. We think about unicorns, mermaids, gods, non-natural properties of goodness, and assorted other categories of properties which might never have been exemplified. At this point the naturalist would be well-advised to borrow a leaf from the book of the radical empiricist. A crude statement of radical empiricism includes the doctrine that all ideas are copies of prior impressions. It took but a moment, however, for the empiricist to recognize that the doctrine had better be modified to apply only to simple ideas. There is no prior impression of a mermaid for the idea of a mermaid to copy. But complex ideas, the empiricist insisted, are built up out of simple ideas that do obey the copy principle. The philosopher relying on our notion of direct acquaintance might put the principle in terms of simple ideas being copies of property exemplification of which we have had direct acquaintance. Analogously, the naturalist could give a causal/evolutionary story of representation for simple ideas, while allowing that we need a different account to accommodate the representational content of complex ideas. But it is far from clear how the account would go for the complex ideas. The naturalist cannot capture the representational content of a complex idea simply by reference to the content of the constituent ideas. The thought of John’s loving Mary is a different thought than is the thought of Mary’s loving John, and that is so even if the ideas of John, loving, and Mary out of which both thoughts are “built” have the same content. Even if the naturalist can come up with an account of the capacity of complex thoughts to represent that is parasitic upon a successful naturalistic account of the representational capacity of simple thought, it is far from obvious that one can come up with that plausible naturalistic account of simple human thought. There are well-known and much-discussed 171

Knowledge, Thought, and the Case for Dualism

problems with how to select from among the relevant items in the causal chain the kind of property exemplification that gets the privilege of constituting the content of the neural item produced. Furthermore, as I have argued elsewhere (2003), it is not clear that one can reconcile any naturalistic account of thought content with the fact that we seem to have an unproblematic introspective access to the contents of our thoughts. As many have pointed out, if we can introspectively know that we have a simple thought of F and that thought has its content only if we have directly or indirectly interacted with F’s, then we can know through introspection, and our knowledge of that philosophical truth about the nature of thought, that there are F’s.10 And that is surely a reductio of the view. I don’t intend to explore this controversy in detail here, but we will need to return to this issue later in assessing the plausibility of the claim that Mary with her new color experiences simply has acquired new ways of representing old properties. I suggested that one might hope to come up with a naturalistic account of correspondence that derives its plausibility from a naturalistic account of representation. The crude idea would be that the thought of F’s being exemplified is true when it corresponds to F’s being exemplified. And the thought of F’s being exemplified corresponds to F’s being exemplified when it represents F’s being exemplified, when F is exemplified. False thoughts still represent but they fail to correspond. If one abandons naturalism, one’s best alternative is the theory that Putnam (1975) derided as “magical.” True thought stands in an unanalyzable relation of 10

I am assuming here that the externalist’s philosophical thesis is not one that requires empirical investigation. It is a classic example of an a priori truth (or at least a truth discoverable from the philosopher’s proverbial armchair). I’ll return to this reductio of content externalism in Chapter 7 when I consider objections to the views I defend. In particular, I’ll respond to the externalist who claims that external conditions determine content without being constitutive of that content.

172

Knowledge arguments revisited

correspondence to fact – false thought fails to correspond to anything (though it, or its constituents, has the capacity to correspond). Thoughts, on this view, are what some philosophers (e.g. Addis 1989) have called “natural signs.” It is part of their intrinsic nature that they have the capacity to correspond. Marks and sounds can also represent, of course, but they represent only because they go proxy for thoughts that have intrinsic intentionality. Proponents of this sort of correspondence conception of truth have often likened the primary bearers of truth value to pictures and have likened the correspondence relation to one of picturing. The metaphor is, of course, potentially misleading. It calls to mind the view that thoughts are literally images, images that can correspond by resembling in some sense that to which they correspond. And while the metaphor might get some purchase if we think about visual images, it is not even clear how we are supposed to think of images of sounds, feels, smells, pains, and other mental states. Still, the point of the metaphor might be to suggest that just as one can read off of an accurate picture features of the person or thing pictured, so also one can read off of a true thought the features of the world that make that thought true. And therein lies the danger for a physicalist who embraces too quickly Smart’s concession that while the meaning of “I am in pain” is quite distinct from the meaning of “My brain is in state X” one can nevertheless think that the truth-maker for these two distinct claims is one and the same fact. The correspondence theorist who endorses the slogan: different thoughts; different facts is a potential threat to this sort of move. But how plausible is the view that one cannot have distinct representations of reality without having distinct aspects of reality making true those representations? The view is certainly not compelling. There is clearly a sense in which these two sentences have different meaning: 173

Knowledge, Thought, and the Case for Dualism (1) Paris is in France. (2) The moon is made of green cheese or Paris is in France.

In one fairly clear sense the thought expressed by (1) is a different thought from the thought expressed by (2). The truth conditions for (1) are distinct from the truth conditions for (2). But it is surely tempting to think that there is only one fact that is making both (1) and (2) true. The disjunctive thought might be a bit like a thought morphing between two representations and the rules for a disjunction require for its truth correspondence only between either of the two thoughts and the world. In a similar sense one might hold that if Fred is the only person who is F then Fred’s being F is the truth-maker for both: (3) Fred is F

and (4) Someone is F.

Again, (3) and (4) obviously don’t have the same meaning. They don’t express the same thought and, in one sense, they don’t have the same truth conditions. But in the world in which only Fred is F, one might suppose (3) and (4) have the same truth-maker. Or consider just one more example. The radical empiricists were always highly suspicious of generic or determinable properties, like being red or being colored. While they found themselves directly acquainted with perfectly determinate shades of red, they weren’t at all sure that the very same object simultaneously exemplified the distinct properties of being this cherry red, dark red, red, dark colored, and colored. To be sure, it is not at all clear that one should reject the existence of determinable or generic properties. At least some philosophers would claim that they find themselves phenomenologically acquainted with 174

Knowledge arguments revisited

determinable properties (see Fales 1990, Part 2, Chapter 9). But if one thought that there was no “generality” in the world, but only the generality that mind brings to the world through thought, one would again find naturally attractive a view that distinguishes difference in truth bearers from differences in truth-makers. While the thought that some particular thing is cherry red is clearly a distinct thought from the thought that the thing in question is colored, one might suspect that there is only that perfectly determinate property of being cherry red whose exemplification makes both thoughts true. So there may be plausible examples of distinct thoughts made true by one and the same fact. But it is not clear that any of this helps the physicalist seeking a single truth-maker for the thought that I am in pain and the thought that my brain is in some state X. The pain is not a determinate property of which the brain state is a determinable. The two thoughts can’t be contrasted as a subject/predicate thought of the form [a is F] and an existentially quantified thought of the form [There exists an F]. And we don’t seem to have a distinction between a “truth-functionally complex” molecular thought and one of its atomic constituents. There are, however, two promising avenues for the physicalist to explore. One we have discussed briefly above and will return to again in the next chapter. The physicalist can argue that a state of a person X represents a state of affairs Y when there is the right causal/evolutionary connection between the occurrence of Y and the occurrence of X. The account leaves open the possibility that two quite distinct states X and Z both represent precisely the same state of affairs Y. This is so because the two distinct states X and Z might both be tied in the right sort of way causally to Y. Y under one set of conditions C1 can cause X, and Y under another set of conditions C2 can cause Z. The result might be that both X and Z can come to represent Y. So a physicalist might understand the 175

Knowledge, Thought, and the Case for Dualism

evolution of Mary’s representational capacities this way. In her color-deprived state Mary is causally connected to the occurrence of brain states of type X via perception. The last link in the relevant causal chain is a brain state of Mary’s, call it alpha. Alpha is a representation of brain state X in virtue of standing in the right sort of causal connection to brain states of that sort. After gaining color experience, Mary is caused by her own brain state X to be in another brain state, call it beta, where this new causal connection establishes a representational connection between beta and X states. The causal path establishing representational content was quite different between the kind of brain state represented by alpha and that same kind of brain state represented by beta, but the two different causal paths might both be sufficient to establish the relevant representational content for both alpha and beta. An analogy might be helpful. A causal theory of representation probably gets off the ground by relying on crude analogies. Consider photographs. What makes a given photograph a picture of Fred? It is hardly that there is a general resemblance between the photograph and Fred. For one thing, photographs don’t actually resemble people very much at all. The photograph is relatively thin, is rectangular, has a glossy finish, and so on. Fred is comparatively bulky, isn’t rectangular, and (probably) doesn’t have a glossy finish. At best it is the experience of looking at the photograph that might be rather like the experience of looking at Fred himself (if the photograph is a good one). But even here, there might well be a photograph of Harry that looks (in this sense) like Fred more than the rather bad picture of Fred. What makes the photograph a picture of Fred is, arguably, just that Fred himself played the right sort of causal role in the image produced on the film (though it is by no means easy to specify what the “right” sort of causal role is). Now if Fred were swimming under water, a sonar blip on a surface ship might also “represent” Fred. Once again Fred is 176

Knowledge arguments revisited

playing a critical causal role in the blip’s appearing on the sonar. Blips on sonar and Kodak prints don’t have all that much in common, but they can both represent Fred in virtue of Fred’s being a common link in the two distinct causal chains leading to both the blip and the print. That’s precisely how a physicalist with a naturalistic account of representation can think of the two quite distinct representational states of Mary both representing a certain kind of color experience. Again, all of this presupposes the success of a naturalistic account of representation. As you must realize by now, I think that one’s contact with the world is much more direct than anything suggested in such accounts of representation. I think that we have direct awareness of the exemplification of certain properties and that it is such awareness that gives rise to an understanding of what such a property is intrinsically. Direct awareness grounds not only non-inferential knowledge, it also grounds direct thought. But the grounding is different. Direct awareness enters into the analysis of what noninferential knowledge is. Direct awareness is the causal grounding of direct thought. Both direct knowledge and direct thought play a role in the argument for dualism. Direct thought tells us what certain properties must be like if they exist. Direct knowledge tells us that such properties are exemplified. If one rejects a naturalistic account of representation, how else might one argue for the view that we can have radically different thoughts both of which turn out to be about one and the same neural state? Well, we can return to our simple example of color ascription discussed earlier. You will recall that we can describe a given shirt in these two different ways: (1) (2)

This shirt is red. This shirt is my aunt’s favorite color.

Again, for (1) and (2) to make sense, the “is” must be construed as the “is” of predication. If my aunt’s favorite color 177

Knowledge, Thought, and the Case for Dualism

denotes (as a matter of contingent fact) the property of being red, we certainly don’t intend to assert by (2) that this shirt is identical with that property. Now this much seems obvious. There is at least a sense in which my thought that this shirt is red is a quite different thought from the thought that this shirt is my aunt’s favorite color. (1) doesn’t entail (2) and (2) doesn’t entail (1). The thought that the shirt is red is a simpler thought than is the thought that the shirt is my aunt’s favorite color. The latter thought involves my aunt and her preferences. The former does not. We suggested earlier that there might be a sense in which when Fred is the only F, Fred’s being F might be regarded as the truth-maker for the claim that there are F’s. Might one analogously argue that when red is my aunt’s favorite color, the shirt’s being red is the truth-maker for the claim that the shirt is my aunt’s favorite color? Here the move seems wrong. The existence of my aunt and her preferences is absolutely critical to (2)’s being true. If the world didn’t contain my aunt, or if my aunt had no preferences among colors, then (2) couldn’t be true. Put another way, the truth-maker for the shirt’s having my aunt’s favorite color critically involves my having an aunt who likes one color more than any others. Donnellan (1977) argued some time ago that we should distinguish between a referential and an attributive use of definite descriptions. When we are at a party and I say to you “The man in the corner holding the martini is my uncle,” there is, I suppose, a sense in which you won’t regard what I said as false if the drink in question was a margarita instead of a martini. Donnellan, at least, thought that was the right thing to say, and that, consequently, we shouldn’t take the use of the definite description in this context as asserting that there is one and only one man who is in the corner and who is holding a martini, and who is my uncle. The definite description is successfully used in this context if it achieves my goal of drawing 178

Knowledge arguments revisited

your attention to the person about whom I wanted to assert something. There are other contexts in which it might matter to me whether the definite description denotes in the Russellian sense, but this isn’t one of them – this isn’t a context in which the definite description is used attributively. There was considerable debate concerning Donnellan’s suggestion. Philosophers chose sides in much the way they did concerning the older debate between Strawson (1950) and Russell (1957) about how to characterize sentences in which definite descriptions fail to denote anything. On the Russellian analysis “The present king of France is bald” is simply false. Strawson wanted to say that the sentence “misfires” – it fails to make a genuine assertion. It always seemed to me that Strawson’s suggestion was unmotivated by anything other than the observation that if we simply say that “The present king of France is bald” is false we will have violated a Gricean conversational rule that we convey as much information as we think our listener would be interested in having. The sentence in question might be false because there is a unique king of France who isn’t bald, or it might be false because there is no unique king of France. Similarly, if a third party remarked only that what I said wasn’t true when I said “the man in the corner holding the martini is my uncle” that person would have been guilty of transgressing pragmatic conventions governing conversation. My listener might well jump to the conclusion that the guy wasn’t my uncle. The inference, strictly speaking, wouldn’t be legitimate, but there would be no reason to leave room for my audience to make the mistake. The point is similar to one concerning complaints about so-called compound questions. There is, in fact, no fallacy involved in asking for a “yes/ no” answer to the question “Have you stopped beating your wife?” It is sometimes said that the question is problematic because it can’t be answered correctly with either a “Yes” or a “No” by someone who has never beaten his wife. But that’s 179

Knowledge, Thought, and the Case for Dualism

quite wrong. The question always has a definitive answer. The answer “No” is correct when either you never did beat your wife (and thus haven’t stopped), or when you did beat your wife and you still do. The answer “Yes” is correct when you beat your wife at one time and no longer do. The only right people have to complain about someone’s requiring merely a “Yes/No” answer to the question is that the “No” answer doesn’t discriminate between two possibilities in a context in which the person giving the answer might very much like the situation to be clear. If we really are focused on the meaning of “The man in the corner holding the martini is my uncle,” it seems simply obvious that the descriptive content of expressions like “the martini” is critical to the sense of the entire sentence. Someone who doesn’t understand the expression “martini” doesn’t understand what was said. And for precisely the same reason, if the man wasn’t holding a martini, then what was said isn’t true. We won’t care that what was said isn’t strictly speaking true. In all probability I will have successfully conveyed the information I was interested in conveying with my false statement. But that doesn’t make the statement any less false. I raised Donnellan’s distinction because it obviously has the potential to bear on the question of whether (1) and (2) above can be plausibly construed as having the same truthmaker. If we thought that “The man in the corner holding the martini is my uncle” and “The man in the corner holding the margarita is my uncle” both made a true assertion, then we would probably concede that there was a single truth-maker for the two quite different sentences. That the sentences were distinct and that they in some sense had different meanings would still allow us to focus on a single fact as the truthmaker for the distinct sentences. In a context in which you and I both know that my aunt’s favorite color is red, then someone sympathetic to Donnellan’s distinction might regard 180

Knowledge arguments revisited

(1) and (2) as accomplishing precisely the same goal – as making the same substantive assertion about the world, and as being made true by precisely the same fact. But once we distinguish between the information someone cares about conveying in making an assertion and information that is, nevertheless, asserted, there is no reason to think that (1) and (2) ever make the same assertion or have the same truthmaker. The above might seem like an extended digression from the primary topic of this book, but, as we shall soon see, it is absolutely critical to a careful assessment of the only halfway plausible physicalist response to the force of knowledge arguments.

4.5

s ummary

In this chapter we have tried to get clear about what precise premises are furnished by the knowledge argument for use in an argument for property dualism. We have rejected the ability hypothesis and the acquaintance hypothesis as radically incomplete. The argument seems to give us a powerful reason to think that when Mary acquires color experience she comes to believe and know a new proposition – the proposition that a certain kind of property is exemplified. At this point, the physicalist has two hopes. One is to reject the appearance that Mary comes to believe a new proposition. One can defend a radical version of content externalism that allows us to reject the conclusion that propositions that will seem quite distinct even to Mary, really are distinct. The other is to concede that the propositions are quite distinct, but to find a way of reconciling that admission with the view that there is only one feature of reality that makes both representations true. We’ll return to these two suggestions in the next chapter as we look more carefully at what makes informative identity statements possible. 181

5 Indirect thought and informative identity

In the last chapter we tried to identify more precisely the premises which might be provided by the knowledge argument. We concluded that Mary certainly seems to acquire new propositional knowledge when she acquires color vision. Specifically, Mary seems to discover that a certain sort of property is exemplified, one of which she was hitherto ignorant. In discovering that truth, she was also positioned to discover connections between the exemplification of that property and the various physiological changes she observed in people as they were confronted by such objects as tomatoes and apples, and when they described the experiences as those of looking red. There remains the nagging suspicion, however, that one can’t really generate any robust metaphysical conclusions about the nature of mental properties from the fact that Mary came to entertain, and even came to know a new proposition. Furthermore, we still need to address the more extreme position that Mary didn’t really come to entertain a distinct proposition – that her discovery should really be viewed as a discovery about which proposition she believed antecedently as she described various people as having that red experience she associated with physiological changes in the brain. Let’s consider this latter question first. 182

Indirect thought and informative identity

5.1

the direct reference theory

It would be easy to write an entire book on theories of reference. This isn’t going to be that book, but we simply can’t get to the bottom of some of the issues that arise in the philosophy of mind without addressing various claims about meaning in general and the content of intentional states in particular. As we noted earlier, the direct reference theorist embraces at least two related slogans. One is that the meanings of certain expressions are their referents or the kinds denoted by those expressions. The other is that meanings aren’t “in the head.” The theses are related, because if the meaning of a term like “Paris” really is the city referred to by that expression, then it surely isn’t going to be plausible to locate meanings “in the head.” Paris is a bit too large to “fit” in my brain, or, for that matter, in my mind. The idea behind the direct reference theory is not new. One can find passages in which John Stuart Mill seemed to suggest that there is no more to the meaning of a genuine proper name than the thing to which the name refers. So he does say of proper names that they “denote the individuals who are called by them, but they do not imply any attributes as belonging to those individuals” (1906, p. 20). One can also find passages in Mill, however, that are quite inconsistent with that idea. So he suggests that naming something is analogous to a thief putting a chalk mark on a house to remind himself that he intends to rob it. So far so good for the direct reference theorist. But then he goes on to say: We put a mark, not indeed upon the object itself, but, so to speak, upon the idea of the object. A proper name is but an unmeaning mark which we connect in our minds with the idea of the object, in order that whenever the mark meets our eyes or occurs to our thoughts, we may think of that individual object. (1906, p. 22)

183

Knowledge, Thought, and the Case for Dualism

Note here that Mill has clearly not only room for the meanings of names, but that he also puts such meanings “in the head.” And, furthermore, it becomes clear from Mill’s (1874) full discussion of the content of propositions about the physical world that physical objects, including the kind of physical objects to which we often give proper names, are really nothing but what he calls “permanent possibilities of sensations.” On analysis, the sentence containing a name disappears in favor of complex subjunctive conditionals describing the conditions under which one would have certain sensations were one to have others.1 Although the now popular versions of the direct reference theory arose out of Kripke/Putnam-style arguments against descriptivist alternatives, there is a more fundamental reason for thinking that the meanings of at least some expressions must be direct. The argument resembles closely the regress arguments for epistemological foundationalism discussed in Chapter 3. We can, the argument goes, pick out things through description. Put more perspicuously, we can succeed in referring to, and thinking about particular things, properties, facts, and kinds of things, properties, and facts, as the bearers of certain properties. I can talk about and think about the person who has the property of being taller than any other person. I can talk about and think about the person who stole my wallet from my office. I can talk about and think about the color that President Obama likes more than any other color. But it seems plausible to suppose that my ability to think about things this way is parasitic upon my ability to think about the various properties picked out by the predicate expressions that occur in the definite descriptions. If I couldn’t think about the relation of being taller than any other person, I couldn’t use 1

See Fumerton and Donner (2009, pp. 187–189) for a more complete discussion of these issues.

184

Indirect thought and informative identity

that property to think about the tallest person. If I couldn’t think about being a color liked by Obama more than any other color, I couldn’t single out in that way a given color property. Now the properties I think about when indirectly thinking about some particular thing or property, or kind of thing or kind of property, might themselves be thought about only through properties that they in turn exemplify. But this process can’t continue forever or I couldn’t successfully think of anything. There must be some things or properties that I succeed in thinking about without thinking about other things – there must be some things or properties that I think about without having to think about them through thinking of their properties. Put another way, there must be foundational thought. It will be useful here to introduce a distinction between direct and indirect reference, and direct and indirect thought. Just as there must be direct knowledge if there is to be indirect knowledge, so also there must be direct reference/thought if there is to be indirect reference/thought. And just as the regress argument for epistemological foundationalism had two versions, so also this regress argument for foundational reference and thought has two versions. The first regress argument purports to establish that without foundations for reference and thought there would be no indirect reference or thought. The second argues, more strongly, that the very concept of indirect reference and thought requires the concept of direct reference or thought. Just as an abstract commitment to epistemological foundationalism leaves open the question of where to locate the foundations of knowledge and justification, so also an abstract commitment to a foundationalism with respect to reference and thought leaves open the question of where to locate direct reference and thought. While direct reference theorists today are thought of as diametrically opposed to Russell and his 185

Knowledge, Thought, and the Case for Dualism

“descriptivism,” it is worth remembering that Russell also was the paradigm of a direct reference theorist for certain expressions. He thought that if there are to be meaningful terms, then there are, and must be, logically proper names, names whose meanings are their referents. And, in a perfectly clear sense, he believed that the meaning of a predicate expression occurring in an ideal (logically perspicuous) language just is the property picked out by that predicate expression. Furthermore, he believed that all thought was parasitic upon our ability to think of the kinds of things that could be referred to by logically proper names and predicate expressions. But how are we to locate the most plausible candidates for entities of which we can think directly? For Russell, the answer to this question is tied to direct acquaintance. One can directly refer to, and directly think about, only those entities with which we have direct acquaintance. How do we discover with what we are directly acquainted? Well, here we employ the same sort of thought experiments we employ in trying to discover those facts acquaintance with which yields non-inferential justification. Acquaintance is a genuine relation, you will recall, that obtains only when its relata exist. So if one is wondering whether one has knowledge through direct acquaintance of the existence of a physical object, say, one simply asks whether the justification one possesses for thinking that the object exists is consistent with the object’s not existing. If the answer is yes, then the justification isn’t constituted by a relation obtaining between you and the physical object. If you are trying to decide whether you are thinking about some object directly or only indirectly, ask whether the existence of the thought is perfectly compatible with its failing to pick anything out. If it is, then you don’t stand in the kind of direct relation to the object of thought that allows you to think about it directly. So, for example, when I think about Obama’s favorite color, I realize that my forming the thought is perfectly compatible with there being nothing to 186

Indirect thought and informative identity

which the thought corresponds – Obama might not have a favorite color. When I think about the severe pain I have right now, however, there is no possibility that the thought fails to correspond to that very property I have before my mind – that very property of which I am thinking directly. There is another way that the Russellian will test for the directness of reference and thought. If you really are using “a” and “b” as pure referring terms, then the statement “a = b” will strike you as utterly trivial – the thought expressed by that sentence will be utterly trivial. By contrast, when you think about some x as simply whatever it is that is uniquely F and you are thinking about some y as simply whatever it is that is uniquely G, then the claim that the thing that is F is the thing that is G will strike you as informative. It is, often, a contingent feature of the world that there is one thing that both uniquely exemplifies F and uniquely exemplifies G. And truths of that sort can often be discovered only through empirical investigation. Relying on this test, Russell concluded that most ordinary proper names are not pure referring terms. Statements like “Samuel Clemens is Mark Twain,” “Hesperus is Phosphorus,” “Deep Throat is Mark Felt” are all potentially informative. It was an interesting and important astronomical discovery that Hesperus is Phosphorus. People had been curious for some time about the identity of Deep Throat and the discovery that it was Mark Felt was a matter of considerable interest to many people. Even many uses of demonstratives fail the test for being pure referring terms. I pass the window of a store catching in my peripheral vision the image of a person and think to myself “That guy looks really old.” I realize to my horror that the reflection was mine and I also realize that it certainly hadn’t occurred to me before just then that I look really old. Both Frege’s and Russell’s genius was to recognize that the only way to accommodate the data described above is 187

Knowledge, Thought, and the Case for Dualism

to recognize that ordinary proper names and most uses of demonstratives have a sense that is distinct from their referents, a sense that (along with the world) determines the referent of a name, a sense that gives the name semantic life even if it fails to correspond with anything in reality. With such a view we immediately understand how there can be informative identity claims, and, in the process, solve other philosophical puzzles that have often led philosophers to ontological excess. In the grips of the idea that the meaning of a name is its referent, philosophers struggled hard to answer the following questions: (1) How can it be true that Pegasus doesn’t exist? After all, to be true, it would have to be meaningful, but if it were meaningful “Pegasus” would need to refer. But if “Pegasus” refers then it can’t be true that Pegasus doesn’t exist.

To answer this question within the confines of the Millian theory of proper names, Meinongians actually sought to populate our world with things that have being but don’t exist. There is a Pegasus – he just doesn’t exist! And we succeed in referring to that non-existent Pegasus when we say of him that he lacks existence. Even Mill, the empiricist’s empiricist, a philosopher who was suspicious of any kind of thing that took him beyond ideas received through the senses, seemed driven to introduce into his ontology things that don’t exist: I may think of a hobgoblin, as I may think of the loaf which was eaten yesterday, or of the flower which will bloom to-morrow. But the hobgoblin which never existed is not the same thing with my idea of a hobgoblin, any more than the loaf which once existed is the same thing with my idea of a loaf, or the flower which does not yet exist, but which will exist, is the same with my idea of a flower. They are all, not thoughts, but objects of thought; though at the present time all the objects are alike non-existent. (1906, p. 33)

188

Indirect thought and informative identity

If one must include in one’s ontology things that have being but do not exist, one is paying a high price, indeed, for a direct reference theory. Or consider, (2)

I believed that Deep Throat was Woodward and Bernstein’s informant, but I didn’t believe that Mark Felt was the informant. How is that possible if the meaning of a name is its referent and the two names have the same referent? My belief that Deep Throat is the informant literally has the same content as does the belief that Mark Felt was the informant.

The reasons for attributing a sense distinct from reference to ordinary proper names might seem to apply equally to names for so-called natural kinds. Just as we made the informative discovery that Hesperus is Phosphorus, so also we made the informative discovery that water is H2O. The kind of stuff that we picked out using the word “water” is the same kind of stuff that another group picked out using the expression “H2O.” Some time ago people labeled a disease that afflicted some legionnaires vacationing in Philadelphia, “Legionnaire’s disease.” They wanted a way of talking about the virus or bacteria that caused similar symptoms in a group of people. We have no difficulty understanding the claim that Legionnaire’s disease doesn’t exist. And again once we understand the term as equivalent to that description, the common disease responsible for those similar symptoms, we have no difficulty seeing how it might turn out that there is no such disease – that in fact it was a number of quite different illnesses that caused similar symptoms.

5 . 2 tw o t r a d i t i o n a l a c c o u n t s of foundational thought In the preceding discussion I have been referring to both Frege and Russell as paradigms of philosophers who embrace the 189

Knowledge, Thought, and the Case for Dualism

concept of foundational thought and who also reject the view that ordinary names, many demonstratives, and names for natural kinds have a meaning that can be identified with their referents/extension. But the two theorists who have come under attack from contemporary direct realists have quite different accounts of foundational thought. As we have seen, Russell wanted to locate foundational thought with thought that is constituted by direct acquaintance with its object. So the paradigm of the subject matter of foundational thought was, for example, either a sense datum with which one is acquainted, or a universal with which one is acquainted. One can be acquainted with both universals and sense data as constituents of facts (property exemplification) with which one is acquainted. But Russell also seemed to think that even without acquaintance of this sort, one can be acquainted “in thought” with the abstract universal. So right now I can have the direct thought of phenomenal redness where that thought consists of my being directly acquainted with the relevant universal. I don’t need to be phenomenologically aware of that property’s being exemplified (by a sense datum) to be directly aware of the property. It might be true as a matter of contingent fact that my ability to “contact” the universal now is causally parasitic upon my past encounter with the property as exemplified by a given sense datum. But, it seems to me that Russell should allow for the possibility of one’s being acquainted in thought with the relevant universal even without the prior experience. As we saw earlier, even Hume, the empiricist’s empiricist, seemed to allow that one can think of a shade of blue of which one never had experience.2

2

Though again it didn’t seem to bother Hume much. He didn’t seem to consider this a serious objection to the fundamental idea that all simple ideas are copies of prior impressions. It may be that Hume thought that one could easily modify the view so that it became the thesis that all simple ideas are copies that resemble more or less prior impressions.

190

Indirect thought and informative identity

So Russell had a fundamentally relational account of foundational thought. And foundational thought is what allows language to have meaning. That’s the sense in which Russell really was the paradigm of a philosopher who was committed to the view that all meaningful language begins with genuine names whose meaning is their referents. The Fregean, by contrast, doesn’t identify meaning with referent, but also has the resources to recognize a foundationalism of thought. It might be helpful to contrast the Fregean and Russellian foundationalism of thought with an example. And let’s take as our example the (admittedly controversial) example of the missing shade of blue. If Russell acknowledges that we can think of the missing shade of blue, he is committed to the existence of that property (universal) as one of the relata of the relation of thinking of that holds between a subject and the property. The view is, no doubt, dialectically attractive. But suppose that one is deeply suspicious of the idea that there are universals that exist even if they are not instantiated, and suppose further that the missing shade of blue about which we are thinking has never been instantiated. Obviously, with this combination of views one must give up on the relational conception of thought. To be sure, the thought of this shade of blue has the capacity to correspond. That is what makes a state intentional – its capacity to correspond. But one needn’t take the state of thinking of X to be a genuinely relational state of affairs. Frege believed that the thoughts or senses that have the capacity to correspond are themselves Platonic entities whose existence is not tied to the contingent existence of conscious beings. But one can adopt something like the Fregean idea and hold that thoughts are contingent properties that exist only when exemplified by conscious beings. Again when one has a thought (exemplifies the relevant thought property) one is in a state that has the capacity to correspond, but its status as thought is independent of whether or not the world contains a referent for the thought. 191

Knowledge, Thought, and the Case for Dualism

From within the framework of such a view, one can still make a distinction between foundational and derivative thought. As I am imagining it, the thought of that missing shade of blue is foundational. I am not thinking of the property by thinking of whatever it is that exemplifies a property. I am, so to speak, thinking of the property directly. The existence of the property is not guaranteed by the existence of the thought, but if the property exists, its intrinsic character will be correctly represented by the thought. Again, it may or may not be the case that it is lawfully possible for one to form a foundational thought of the intrinsic nature of a property with which one has never been acquainted. I suspect that when people imagine color-deprived Mary they just take for granted that she doesn’t have the kind of foundational thought of phenomenal color that she acquires as soon as she gains color experiences of which she is aware. But again, David Lewis (1988) was right to remind us that there is no a priori absurdity in conceding the intelligibility of a world in which Mary never has color experiences but has foundational color thought.

5.3

objections to sense theories

Given the enormous power of Frege’s and Russell’s approach to solving the puzzles mentioned above, one might well suppose that one needs powerful reasons indeed to abandon the approach. But many, probably most, philosophers now think such reasons exist. No one was more influential in attacking the Russellian approach than Kripke in his highly influential Naming and Necessity (1980). While acknowledging the attraction of views like Russell’s, Kripke systematically attacked the details of the theory. Painting with a broad stroke, we can summarize the problems allegedly exposed with descriptivism about many demonstratives, ordinary proper names, and names for kinds (water, heat, tree, animal, etc.) as follows: 192

Indirect thought and informative identity (1)

(2)

(3)

For many names we successfully use to make assertions, we simply can’t think of any definite description that has a chance of uniquely denoting the individual or kind about which we make the assertion. Even when we can successfully associate with names definite descriptions that denote the referent of the name, the statement containing the name has different modal properties than the statement containing the definite description. Scope ambiguities that occur with the use of definite descriptions do not arise concerning statements that are made using names instead.

Consider each in turn. Somewhere in the back of your mind you might remember that Khufu was a famous Egyptian. If you were on a quiz show and were asked about the nationality of Khufu you might, consequently, assert that Khufu was an Egyptian. And you would have been right – you would have said something meaningful and true. But let us stipulate that you have no idea which Egyptian Khufu was. You can’t come up with a definite description that denotes the person about whom you made your assertion. I guess that you might try “the famous Egyptian named ‘Khufu’” or “the most famous Egyptian,” or “the person I seem to remember being an Egyptian.” But you know perfectly well that none of these are likely to do the trick. For all you know there was more than one famous Egyptian named “Khufu,” that the one you made a claim about wasn’t the most famous Egyptian, and most of us remember Cleopatra after all. Let’s turn to the modal arguments. Consider the definite description(s) you associated with a name like “Aristotle” (to take a well-worn example). Suppose you have always thought of Aristotle as the teacher of Alexander the Great. If you used “Aristotle” and “the teacher of Alexander” synonymously, then the following claim would be analytic: Aristotle (if he existed at all) taught Alexander. But we would have no 193

Knowledge, Thought, and the Case for Dualism

difficulty making sense of an historian who purported to discover that Alexander had no connection at all with Aristotle, and had simply invoked a fictional association as a way of magnifying his fame. No one in his right mind would reject the historian’s claim as literally self-contradictory. It is important to distinguish this last argument from a superficially similar, but far less powerful objection. In Naming and Necessity Kripke often seems to play on de re/de dicto ambiguity. Aristotle, the argument goes, didn’t have to have taught Alexander. Sure enough. But that, on the face of it, does nothing to show that “Aristotle” doesn’t have the same meaning as “the teacher of Alexander” for there is no contradiction in the following de re assertion of possibility: The teacher of Alexander didn’t have to have taught Alexander (had he, for example, become a recluse early in life). Kripke was always sensitive to the charge that he was milking de re/de dicto ambiguity in making his arguments, and strongly denied it. But the cleanest way to make the relevant modal argument is to focus on de dicto modality. For any definite description “the F” you associate with a proper name “P” ask yourself honestly if you would reject as self-contradictory an assertion that P wasn’t, after all, the F. If the answer is “No,” then you don’t really regard “P” as equivalent in meaning to “the F.” We discussed a moment ago de re/de dicto distinctions and the scope ambiguities that come with them. When I say that “The F might be G” and we give that statement its usual Russellian translation we discover that it is ambiguous. We can interpret it as claiming that it is possible that the one thing that is F is G. Or we can interpret it as claiming that the one thing that is F might be G. But it is not just modal operators that create scope ambiguity. The Russellian has two choices concerning the interpretation of the following: “The F is not G.” The sentence can be taken as asserting that it is not the case that there is just one thing F that is G, or it can be taken as 194

Indirect thought and informative identity

asserting that there is just one thing that is F and it is not the case that it is G. The sentence “Fred is not hungry” doesn’t naturally lend itself to these alternative readings and that might be thought to cast doubt on the idea that one can capture the meaning of “Fred” with a translation that opens up scope for ambiguities.

5.4

anti-descriptivist alternatives

So-called direct theorists are better with objections to descriptivist alternatives than they are in putting forth their own positive theories. In the Kripke-inspired tradition, one can distinguish two main strands of direct reference theory. One makes the referent/meaning of a name or common noun on a given occasion of its use a function of that use’s causal ancestry. There was some guy named something like “Julius Caesar” and when I use the name “Julius Caesar” in a sentence, I am referring to the guy who was once called “Julius Caesar” and whose being called that was a critical link in the causal chain leading (in the right way) to this use of the expression by me. Causal theorists always need to add the proviso “in the right way” to protect the account from so-called deviant causal chains – my aunt greatly admired Julius Caesar and names my cousin “Julie” as her way of honoring Caesar. I’m obviously not referring to Caesar when I tell my wife that Julie is coming over to watch TV. A causal theory of reference for how my use of a name refers relies on the past existence of successful reference. As such it is incomplete. And there are two ways in which the direct reference theorist (who rejects the approaches of both Frege and Russell) might move at this point. One is to try to extend the causal theory to an ostensive “baptism” ceremony. The other is to turn to a reference-fixing “baptism” that employs definite descriptions. Let’s consider the latter. 195

Knowledge, Thought, and the Case for Dualism

It is part of Kripke’s view that one can introduce a directly referring term into one’s language through use of a definite description. While looking at Lake Michigan, I can baptize a term like “water” with the definite description “the structurally distinct kind of stuff presenting such and such an appearance in that lake.” Both detectives and the general public in late nineteenth-century London found it useful to dub the person presumed to have killed several prostitutes “Jack the Ripper.” The idea is that one can use a definite description to attach a name to the thing or kind named, but do so without assigning the name the meaning of the definite description. The name retains its character as a name – a term whose meaning is its referent, despite the fact that it was a description that breathed referential life into the referring term. I’ve always thought that the whole idea of reference-fixing is a magical illusion. I can’t for the life of me see how one can build reference on the scaffolding of description, destroy the scaffolding and expect the reference to continue standing. Let’s begin with a pure case. Suppose that having read Kripke I decide I want to create a name for the next Republican President of the United States. I use the definite description to breathe referential life into the name “Rep” that I introduce. I get some of my friends to agree to use the name the same way as I do and we carry on natural enough conversations concerning the identity of Rep. You think Rep will be Ryan and I think Rep will be Romney. Now is there any plausibility to the claim that “Rep” is used by us as anything but a synonym for “the next Republican president of the USA”? As Kripke would be quick to concede, we can know a priori the falsity of the proposition that the next Republican President of the USA will not be Rep. But, of course, the Kripkean revolution brings with it a distinction between the a priori and the necessary – one can know a priori that which is only contingent (e.g. that Rep, if he exists, is a Republican) and, furthermore, one can 196

Indirect thought and informative identity

know only empirically that which is necessary (e.g. that Hesperus is Phosphorus or that water is H2O). Still when I think to myself that Rep will be Romney, what on earth am I thinking other than that the next Republican president of the USA will be Romney? Undoubtedly, I can entertain the de re modal thought that Rep wouldn’t have to become a president. Even if Rep is the next president of the USA he didn’t have that property essentially – he could exist without becoming president. But then, as we saw before, I can make that claim about the next Republican president of the USA. The next Republican president of the USA didn’t have to have pursued a career in politics either – he could have done something else with his or her life. In the technical language of philosophers, the next Republican president of the USA doesn’t have the property of being a Republican president essentially.3 If we are to highlight a relevant modal difference we need to turn again to de dicto necessity. After I dub the next Republican president of the USA “Rep,” is it, or is it not, analytic that Rep, if he exists at all, is the next Republican president of the USA? And by any test I can think of, it is. If this is right, then we are left with the causal theory doing the work of distinguishing direct reference theories from the Russellian alternative. But as I have argued elsewhere (Fumerton, 1989) the Russellian always has the resources to steal whatever good work the causal theorist does in coming up with a plausible account of how we succeed in referring with ordinary names and names for natural kinds, despite the fact that we have very little in our descriptive inventory to infuse meaning into the proper and common nouns.4 In our crude characterization of 3

4

Again, you will recall from Chapter 2 that the acknowledgment of a distinction between de re and de dicto ascriptions of modality doesn’t commit one to a significant underlying metaphysical distinction between two kinds of modality. Ironically, perhaps, I believe that I first got this idea from an informal conversation I had with David Kaplan.

197

Knowledge, Thought, and the Case for Dualism

causal theories we used, as we must, definite descriptions to denote the referents of proper names and names for kinds. The causal theorist hasn’t got a theory at all until he coughs up the definite description that on his view correctly identifies the referent of a name. And if the definite description is part of an adequate philosophical account of reference, it won’t pick the referent out contingently. It is a semantic rule that the causal theorist is trying to capture concerning how reference works. So it is not hard to see what is coming next. Suppose that the causal theorist comes up with a half-way plausible view about what determines the referent of my use of “Aristotle” in “Aristotle was a Greek.” Aristotle, the causal theorist says, is the person whose baptismal ceremony is causally connected in the right way to my use of “Aristotle” in “Aristotle was a Greek.” If we can think of a possible situation in which that definite description gets the intuitively wrong answer to the question “What is the referent of ‘Aristotle’?,” then it is back to the drawing board for the causal theorist. But let’s suppose that we convince ourselves that there are no counterexamples. The Russellian need only thank the causal theorist for the work and identify that definite description as the one that captures the meaning of “Aristotle.” “Aristotle was a Greek” our Russellian now proclaims is equivalent in meaning to “The guy whose baptismal ceremony resulted in my use of ‘Aristotle’ in this sentence was a Greek.” One can also add to this account whatever conventions one thinks plausibly capture our use of modal operators. So it might be plausible, for example, to suggest that whenever we use modal operators in the context of making assertions using names, our convention is that the ascription of modality always be construed de re. It may also be plausible to incorporate similar restrictions concerning the scope of “not” in sentences using names. So it may well be that in all (or most) sentences using names for individuals or 198

Indirect thought and informative identity

kind, e.g. N is not F, we always (or usually) understand “not” in such a way that it gets narrow scope in the reduction of that sentence that replaces “N” with a definite description. There are only two obstacles to the above attempt at “Russelling” causal theories. The first concerns criteria for the plausibility of a meaning analysis. The second is a more technical problem. Consider the first. One might worry that the proposed analysis is wildly implausible. If my use of “Aristotle” had the meaning of that causal definite description, why did it take even sophisticated philosophers so long to come up with the idea? If “X” as used by me means the same as “Y” as used by me, then surely it shouldn’t have been that difficult for me to discover that “X” and “Y” have the same meaning. But it took someone with the philosophical skill of a Saul Kripke or a Hilary Putnam to come up with the idea. Traditional analytic philosophers should be very cautious before accepting this sort of argument against a proposed meaning analysis. We all know how difficult it can be to discover the semantic rules we follow in using language – it had better be difficult if we are to justify our existence. There is, to be sure, a bit of puzzle here. How can we follow rules if we don’t already know what the rules are? But the puzzle has an obvious answer. The following of semantic rules is best identified with complex linguistic dispositions. In this sense we can successfully follow a rule without having the slightest idea what rule it is that we are following. Consider the useful analogy of syntactic rules. We follow rather complex grammatical rules from a very young age. We correct young children when they use the wrong tense of a verb, or don’t have the proper subject/verb agreement. We teach them how and when to use the possessive, and eventually even teach them how to use participles and gerunds (though these days we rarely call these parts of speech by their names). When things go well, the teaching has its effect and the linguist behavior ends up being 199

Knowledge, Thought, and the Case for Dualism

in accord with the relevant rules. But, as I said, it is hardly the case that the three-year-old child has internalized the rules in the form of even a dispositional belief that such and such are the correct rules of grammar. Sadly, these days even university students would be hard pressed to identify the many and subtle syntactic rules they follow when speaking. Academics in linguistics departments spend countless hours arguing with each other about which sentences are well-formed and which are not. And they rely heavily on counterexample, just as philosophers do, in criticizing each other’s positions. So yes, I certainly didn’t think that I was following the causal theorist’s semantic rule when using names, but I have come around to the idea that the view has at least prima facie plausibility. I might be discovering something about the semantic rule governing my use of ordinary names and names for natural kinds. That it took me a while had better not be an obstacle to the view being correct. People thought that by “knowledge” they meant “justified true belief.” Along comes Gettier (1963) and they decided they were wrong. In the wake of Gettier counterexamples, they considered all sorts of radically different proposals concerning what they meant by “know,” some of them very complex indeed.5 The technical difficulty with the Russellian attempt at theft over honest toil is a bit more serious. You will recall that the Russellian is contemplating viewing “Aristotle” in “Aristotle was a Greek” as analytically equivalent to “The person whose baptismal ceremony is causally responsible in 5

To get just a feel for some of the cycles and epicycles of attempts to Gettier-proof an analysis of knowledge just take a look at some of the selections in Pappas and Swain (1978). It would take a lot of philosophical nerve to reject analyses of this nature just because they are complicated. Even Williamson (2000) who officially recommended giving up on attempting to analyze knowledge (by taking it to be indefinable) never argues for that conclusion based on the observation that proposed analyses are simply too byzantine.

200

Indirect thought and informative identity

the right way for this use of ‘Aristotle’.” But it is interesting to note that one couldn’t replace “Aristotle” with the definite description. The resulting sentence “The guy whose baptismal ceremony resulted in this use of ‘Aristotle’ was a Greek” is false. The definite description fails to denote because there is no use (though there is a mention) of “Aristotle.” It is tempting to think that if two expressions have the same meaning they should be substitutable salva veritate, at least in extensional contexts. And we cannot make the substitution without altering truth value. But all this really shows is that we have another interesting counterexample to principles of substitutivity. There is no obstacle to the view that “Aristotle” goes proxy for a thought that contains demonstrative reference to the use of “Aristotle” in the sentence. Names, then, might be like a number of other indexicals. Consider our use of “now.” It is entirely plausible to suppose that a token of “now” is our way of referring to the time at which that use of the token “now” occurs. Interestingly enough, we couldn’t replace “now” with that definite description because in making the replacement we would have lost the critical use of “now” (though the term would still be mentioned). But again I don’t see why that should stop us from identifying the relevant thought expressed by “now” as the one captured by the definite description.

5.5

did mary always have beliefs about phenomenal color?

The previous discussion was necessary background for our evaluating the suggestion that Mary didn’t really come to believe any new propositions when she gained her color experience. Rather, the suggestion is, she came to discover which propositions she was believing all along in her color-deprived state. If the idea behind reference-fixing made sense, then we could imagine that Mary fixed the reference of expressions 201

Knowledge, Thought, and the Case for Dualism

like “looking red” using various definite descriptions, e.g. the state that person is in when confronted by a tomato and when reflected light brings about the relevant change in a subject’s brain. Mary believed in her color-deprived state that things look red to people, and her belief was about that very property her definite description denoted. She seemed to have a different belief about that property when she became directly acquainted with it, but it was just the same old belief. The difference, if any, is a difference in knowledge of which sentences express the same proposition. So in our NidaRümelin thought experiment Mary thought she believed that tomatoes look B* – she also thought that looking red was looking B*. But she was wrong. She was just like the person who thought that Hesperus wasn’t Phosphorus even though it is, and even though every time she had a belief about Hesperus she had (unbeknownst to her) a belief about Phosphorus. It should be obvious from my earlier discussion that I have no sympathy for this sort of move. Mary’s thought in her color-deprived situation was no more than the thought that people are in a certain state, whatever state it is, that results from those neural changes. That thought is not the same thought as the thought that people are appeared to in a certain way. There is, no doubt, a sense in which color-deprived Mary’s thought turned out to be about looking red. It is precisely the sense in which if I think of a shirt as having Obama’s favorite color, when I have no idea which color is Obama’s favorite, then that thought can turn out to be about the color red. It is also the sense in which a functionalist could allow that when one thinks of a given creature’s being in pain, one’s thought is about whatever property realizes that functional state (plays the functionally defined role). But the thoughts about Obama’s favorite color and the realizer of the functional states are paradigmatically indirect. One thinks about the color and the realizer through thinking of a property 202

Indirect thought and informative identity

exemplified by the color and the realizer of the functional state. As we saw earlier, our thought of that other individuating property might itself be indirect, but we must eventually hit upon a property that is thought of directly and not through thinking of some property it exemplifies. In the preceding comments I have argued for the radical conclusion that one can’t make sense of the attempt to convert an expression into a directly referring expression through a ceremony of reference-fixing. I have also argued for the equally radical suggestion that the apparent referential relevance of the causal ancestry of an expression’s use can be captured perfectly well by the descriptivist. It’s not clear, however, that one can’t make many of the same points on which I have insisted from within a framework that allows for reference-fixing and causally determined reference. Chalmers (1996, 2004), for example, distinguishes the epistemic from the counterfactual content of expressions. If we allow that Mary could fix the reference of “looking red” using a definite description that denotes the experience through causal properties of the experience, then we still need to distinguish the counterfactual and the epistemic content of the expression “looks red” for colordeprived Mary. While the details of the “two-dimensional” logic Chalmers puts forth seem initially complicated, I think the basic idea is no more problematic than the following. If we fix the reference of an expression “R” using the definite description “the property that is F,” then on the assumption that we have succeeded in converting “R” into a directly referring rigid designator, “R” picks out the same property in all possible worlds. If R is that property of looking red that Mary becomes aware of after gaining color experience, then we must say that “R” picks out that property of looking red in all possible worlds. The proposition that “R is the property of looking red” is a necessary truth, albeit one that could be discovered only a posteriori. But we simply must also 203

Knowledge, Thought, and the Case for Dualism

accommodate and explain the obvious fact that for Mary, in Nida-Rümelin’s thought experiment, it was epistemically possible both that R is looking red and that R is looking blue – those two phenomenal color properties she became acquainted with by being shown color samples. As a first stab, we might say that a given proposition P is epistemically possible for S when P’s truth is consistent with everything S knows. There arises, however, an immediate objection. No necessary falsehood is consistent with anything S knows, but one surely wants to allow that a proposition P might be epistemically possible for S even if it is (unbeknownst to S) necessarily false. By the same token there may be contingent propositions that are inconsistent with P’s evidence, but where P doesn’t realize that fact. Again such propositions are surely, in some sense, epistemically possible for S. In the final analysis, it is probably best to say that P is epistemically possible for S when P has a truth value and S is not in a position to know not-P. Now Kripke and company want to allow that propositions such as the proposition that heat is not molecular motion are epistemically possible for most people. But if propositions are the meanings of sentences, and terms like “heat” are rigid designators whose reference has been fixed by definite descriptions, then there is a sense in which the proposition that heat is not molecular motion is the same proposition as the proposition that molecular motion is not molecular motion. But that proposition is not epistemically possible for anyone with a modicum of common sense. One can retreat to sentences as the secondary bearers of truth value and make the claim that it was only epistemically possible that the sentence “Heat is not molecular motion” expressed a truth. The truth about the sentence was epistemically possible only because we didn’t know that that sentence expressed the same proposition as the sentence “Molecular motion is not molecular motion.” But as I have been arguing above, that move seems to

204

Indirect thought and informative identity

miss the character of the thought a person has as they entertain the possibility that heat is not, in fact, molecular motion. In any event, when Chalmers talks about epistemic content in contrast to counterfactual content, he seems to be replacing the Kripkean proposition (expressed by the sentence containing reference-fixed rigid designators) with the corresponding proposition expressed by a sentence containing the floating, non-rigid definite descriptions that were supposed to have done their reference-fixing work. So, even if there is a sense in which it is necessary that heat is molecular motion, the epistemic content of one’s belief that heat is molecular motion is not relevantly guaranteed by anything we know. And that just amounts to the claim that the cause of sensation S (the reference fixer for heat) being molecular motion is not guaranteed by anything we know. Chalmers wants to argue that in the case of phenomenal color, the epistemic and counterfactual possibilities come together, and that’s why we can’t make sense of Mary’s making an a posteriori discovery that the redness she was talking about in her color-deprived condition is in fact phenomenal redness. The argument is similar to Kripke’s own argument for property dualism. Kripke has an explanation (one we reject) of how we can find informative a claim like water is H2O, or like heat is molecular motion. The explanation is that the left-hand term flanking the identity statement is one whose referent is determined by a reference-fixing definite description, one that is not equivalent in meaning to the term, but which converts the term into a rigid designator for a kind. Furthermore we don’t know what, if anything, we have succeeded in referring to with our new rigid designator. We can successfully use the term “heat” to pick out a kind without knowing what kind we have picked out. So we can make the empirical and informative discovery that what we were talking about when we talked about heat is, in fact, molecular motion. The explanation for 205

Knowledge, Thought, and the Case for Dualism

the informative character of the identity statement is, in fact, Russell’s explanation, albeit with the direct reference theorist’s wrinkle that converts the definite description into a tool for securing reference, rather than a locution that captures the meaning of the expression. Again, I have argued that insofar as we think of the meaning of a sentence as a function of the thought expressed by the sentence, the Kripkean explanation of the informative character of an identity statement should convince us that the statement in question really has the sort of meaning that Russell said it had. But in any event, I’m now stressing that when it comes to understanding the informative character of an identity claim, Kripke is offering essentially the same explanation as Russell’s, an explanation that relies on our distinction between direct and indirect thought. Kripke is conceding, in effect, that however direct the referential relation might be between “heat” and its referent, our access to heat in thought is still indirect – we are still thinking of heat as whatever it is that causes that familiar sensation of heat. Science later tells us that the critical cause of the sensation is molecular motion and that’s what we find informative. But the case is different when it comes to phenomenal states like pain. To put Kripke’s argument in our own terms, our thought of phenomenal pain is direct – we are not thinking of pain through thinking of properties it exemplifies. Rather, we are thinking of the pain directly. As a result, we simply can’t offer the same explanation of the informative nature of an identity claim. Chalmers is making the same point with a different terminology.

5.6

informative identity

With our distinction between direct and indirect thought we are in a position to make the following claim about informative identity. The discovery that X is identical with Y (or that X’s 206

Indirect thought and informative identity

are identical with Y’s) can be informative when either the thought of X or the thought of Y (or both) is indirect in the sense explained above. When we are thinking of both X and Y directly – when our thought directly represents X and Y we can read off the intrinsic nature of X and Y (if they exist) from the nature of the thought. On the Russellian view, we can do this because X and Y are constituents of the direct thought. On the Fregean view, we can do this because direct thoughts “picture” the property as it is in itself, not as the bearer of higher level properties. We can make informative identity claims about any sort of thing, and that is so because there is no restriction on the kind of thing that we can think of indirectly. I can think of a person indirectly (e.g. as the tallest person in the USA). I can think of a color indirectly (e.g. as the color that my aunt likes more than any other color). I can think of a proposition indirectly (e.g. as the proposition expressed by the next assertion Diane makes). I can think of a fact indirectly (e.g. as the fact that Russell liked best as an example of a fact). There is nothing that exists that cannot be thought of indirectly. As a result, there is no sort of thing about which we cannot make an informative identity claim. But if Russell, Frege, and Kripke are correct, a necessary condition of a thought concerning identity being informative is that the thought involves at least one way of thinking about the subject of the identity judgment that is indirect. The Kripkean argument for property dualism is, however, potentially flawed. The principle he relies on is correct. If A = B is to be an informative identity claim either our thought about A or our thought about B must be indirect. But Kripke focused on only one half of the claim.

207

6 An ontologically liberating skepticism: the last hope for physicalism In Chapter 5 we suggested that the key to understanding informative identity claims is to recognize the distinction between direct and indirect thought. It is because we can think of individuals, properties, propositions, and facts, indirectly, that we can make informative discoveries concerning the intrinsic character of the objects of our thought. It looks as if this might be the key commitment that drives the property dualist to the idea that we must reject an identification of paradigmatically mental states with physical states. As Kripke argued, we don’t think of a state like pain (or like looking red) indirectly. We don’t pick out such properties by thinking indirectly of whatever it is that exemplifies certain properties. But even if one accepts the above distinction between direct and indirect thought, and the diagnosis of informative identity that relies on that distinction, one has a potential response to the property dualist. It may be that our thought of paradigmatically mental properties is direct, but we haven’t yet discussed the question of whether our thought of the paradigmatically physical properties with which the physicalist wants to identify mental properties is direct. There is a considerable irony in the fact that the kind of radical empiricism favored by most property dualists is also the very view that leads naturally to the conclusion that we have only 208

An ontologically liberating skepticism

indirect thought of the properties that (partially) define not only brain states, but also that characterize physical objects in general. The possibility we are about to explore will be anathema to most self-proclaimed physicalists. More often than not, physicalists embrace their reductions because they are deeply suspicious of the properties introduced by the property dualist. As we saw earlier in this book (Chapter 2), reductionists in general are moved to reduce in the direction of what they take to be epistemically and metaphysically unproblematic. So, again, phenomenalists were confident that they understood and had unproblematic knowledge of the phenomenal and found highly problematic the suggestion that there is a material world radically different from the world of the phenomena with which they are directly acquainted. By contrast, the behaviorists thought they had a relatively unproblematic understanding and knowledge of the physical world, and found philosophically suspicious the “hidden” world of mental phenomena that would make so problematic knowledge of other minds. So in the history of philosophy we find phenomenalists who were bound and determined to reduce the physical to the mental, and we also find behaviorists who were bound and determined to reduce the mental to the physical. When we try to understand how there can be informative identities between the mental and the physical by introducing the idea that we have only indirect thought of the physical, we have, in effect, conceded that it is the physical that is more problematic than the mental when it comes to our understanding its nature. And, as I suggested, this will be anathema to physicalists bound and determined to philosophically explain the world in terms of the “unproblematic” category of the physical. I have already suggested in Chapter 3 that the physicalist has things completely backward when it comes to both our understanding of, and our knowledge of, the physical. 209

Knowledge, Thought, and the Case for Dualism

That understanding and knowledge is obviously parasitic upon our understanding and knowledge of subjective appearance. And if we are to find a defensible form of physicalism, we must do so within the confines of a view that respects the phenomenological priority of the subjective.

6.1

in d i re c t un d e r s t a n d in g o f t h e ph y s i c a l

The very view that insists that our understanding and knowledge of the mental is more fundamental than our understanding and knowledge of the physical also might seem to open the door to a physicalist world view that finds room for the mental. The crude idea is simple. The extreme empiricist should claim that all thought of the physical is parasitic upon thought of the mental. To explain this view further it might be useful to begin with properties of physical objects that were once widely viewed as secondary properties. Think of the sourness of a lemon. Now ask yourself whether that thought involves thought of the familiar gustatory sensation we call tasting sour. It seemed to many in the history of philosophy utterly obvious that when we think of a lemon’s being sour we are thinking about a certain sour taste sensation. But we also know that we can’t identify a lemon’s being sour with its actually tasting a certain way to some subject. For one thing we can obviously make sense of a lemon’s being sour even if no one actually bites into the fruit and is caused to have the relevant sensation. Moreover, we all realize that something can be sour even if it doesn’t affect us with the relevant sour taste sensation when we bite into it. We understand that there are conditions that can interfere with the normal causal chain that leads from biting into the lemon to the sour taste sensation. If you have a bad enough cold, for example, and bite into a lemon, then you won’t taste much of anything. So the classic secondary quality 210

An ontologically liberating skepticism

theorist turns to counterfactual conditionals to explain what it is for a lemon to be sour. A lemon is sour just insofar as it would produce in a normal person under normal conditions the sour taste sensation were the person to bite into the lemon. A great deal would need to be said before we had an analysis that could meet all potential objections, but the basic idea is both clear and, at least initially, plausible. Again, these days, many philosophers are more likely to argue that the reference of “sourness” is fixed by a definite description that denotes the properties of the lemon causally responsible for the sour taste sensation. The sourness of the lemon then gets identified with the microstructure playing a critical role in causing the sensation. The view is analogous to that held by the functionalists we talked about in Chapter 2 who want to combine their functionalism with the view that the mental state functionally analyzed is identical with the realizer of the functional role. I tried to convince you that the consistent functionalist shouldn’t identify the mental state that receives the functional analysis with its realizer. The functionalist should want the property of being in pain to be the same property even as that which plays the functional role changes. As you know by now, I don’t think that functionalist accounts of mental states have the slightest plausibility, but the secondary quality theorist can learn from the conceptual points made above. Different kinds of things can all be sour, and they can all be sour even if the underlying structure that causally grounds the disposition to taste a certain way under certain conditions is quite different in the different sour things. If our goal is to find what is common to all those things in virtue of which they are sour, we shouldn’t look to the ground of the disposition. It is the disposition to taste a certain way that constitutes the sourness. And even if I am mistaken about all this, I have also tried to convince you that the “reference fixers” should allow that at the level of thought the critical 211

Knowledge, Thought, and the Case for Dualism

thought of sourness is precisely what is expressed by the subjunctive conditional. So the basic idea, again, is that our thought of a thing’s being sour just is the thought of the thing’s having the power to produce a certain taste sensation under certain conditions.1 Most people understand perfectly what it is for something to be sour even though they haven’t got the slightest clue as to what properties of the thing play the relevant causal role in producing the familiar sour taste sensation. If one becomes convinced that the sourness of the lemon is only “in” the lemon as the power to produce a sour taste sensation, it isn’t hard to adopt the same analysis of other properties that we attribute to objects. So the redness of the apple was thought by the radical empiricist to be nothing but the power the apple has to affect conscious beings with that familiar visual experience (the one that color-deprived Mary never had). And Berkeley (through his spokesman Philonous) dragged poor Hylas from one property to another, getting Hylas to concede that his thought about sensible qualities always turned out to be a thought that critically involved a characteristic sensation that the object had the power to produce. The extreme view is that all properties of physical objects revealed to us through the senses are thought of by us as nothing but powers to produce sensations. The view might be called the causal theory of objects.2 It bears a close resemblance to the even more radical 1

2

The power referred to is causal power, but in using this term I should make clear that I am not endorsing anything like an Aristotelian view of hidden causal powers, the kind of view that was held also by most modern philosophers until Hume’s revolutionary thoughts about causation. I’m using causal power as just an abbreviation for whatever the correct view is of what makes one thing the cause of something else. It is critical that we distinguish the causal theory of objects from the causal theory of perception. The causal theorist of perception offers a highly plausible account of what it is for someone S to perceive a physical object O. The crude idea is that S perceives O when S has an experience that is caused in the right way by O. As always

212

An ontologically liberating skepticism

phenomenalism that sought to reduce all talk of physical objects to talk about the sensations a subject would have were the subject to have others. But it is critically different. The causal theory of objects doesn’t deny that when we think about physical objects we are thinking (albeit indirectly) about mind-independent objects and their properties. But the view is that our understanding of objects and their mind-independent properties is thoroughly parasitic upon our understanding of phenomenal qualities. When we think and talk about physical objects and their properties we are always thinking about those objects and their properties in terms of the causal role they play in affecting our conscious life. The view has surfaced occasionally in the history of philosophy. It was Hylas’s last ditch attempt to save materialism from Philonous’s relentless march to idealism. In the second of Berkeley’s Three Dialogues, Hylas (1713, p. 202) finally suggests that we have only an indirect idea of matter: I find myself affected with various ideas, whereof I know I am not the cause, neither are they the cause of themselves or of one another, or capable of subsisting by themselves, as being altogether inactive, fleeting, dependent beings. They have therefore some existence distinct from me and them: of which I pretend to know no more than that it is the cause of my ideas. And this thing, whatever it be, I call matter.

Although in the end, Hume is probably best thought of as a radical skeptic who despaired of not only knowing, but making intelligible, claims about a perceiver-independent physical world, he, like Hylas, toyed with the idea of allowing a relative the problem of deviant causal chains presents no end of headaches for the causal theorist trying to revise the view so as to avoid counterexamples. But whatever form the analysis eventually takes, it is important to note that one can adopt a causal theory of perception while one leaves completely open the correct analysis of physical objects and their properties. The causal theory of objects is an analysis of what it is for a physical object with certain properties to exist.

213

Knowledge, Thought, and the Case for Dualism

idea of the external world. He says in Book I, Part II, Sec. VI of the Treatise (p. 68): The farthest we can go towards a conception of external objects, when suppos’d specifically different from our perceptions, is to form a relative idea of them, without pretending to comprehend the related objects.

In our terminology, Hume appears to be suggesting that we can form an indirect, but not a direct thought, of physical objects and their properties. Again, in the Enquiries (p. 155), Hume alludes to a view like the one put forth by Hylas: Bereave matter of all its intelligible qualities, both primary and secondary, you in a manner annihilate it, and leave only an unknown, inexplicable something, as the cause of our perceptions; a notion so imperfect that no skeptic will think it worth to contend against it.

The last sentence in the above quote strongly suggests that Hume didn’t really think the suggestion under discussion could be employed in an effective defense of common sense against the skeptic’s attack, but it is interesting that there is no real argument advanced by Hume to that effect. John Stuart Mill was in many ways the ultimate culmination of the various strands that made up the heart of British empiricism. There are certainly passages like the following where he clearly takes very seriously a view of matter as the unknown and unknowable cause of sensation. Consider the following from Mill (1877): External things exist, and have an inmost nature, but their inmost nature is inaccessible to our faculties. We know it not, and can assert nothing of it with a meaning. Of the ultimate Realities, as such, we know the existence, and nothing more. But the impressions which these Realities make on us – the sensations they excite, the similitudes, groupings, and successions of those sensations, or to sum up all this in a common though improper expression, the

214

An ontologically liberating skepticism representations generated in our minds by the action of the Things themselves – these we may know, and these are all that we can know respecting them. (pp. 9–10)

But in other passages it is clear that Mill is more comfortable moving to a version of phenomenalism, a closely related but different view:3 the sensations which, in common parlance, we are said to receive from objects, are not only all that we can possibly know of the objects, but are all that we have any ground for believing to exist. What we term an object is but a complex conception made up by the laws of association, out of the ideas of various sensations which we are accustomed to receive simultaneously. There is nothing real in the process but these sensations . . . but we have no evidence of anything which, not being itself a sensation, is a substratum or hidden cause of sensations. (p. 8)

The view Hylas toyed with, the view that all we have is a relative idea of physical objects, found its way into twentiethand twenty-first century thought as well. It is an early forerunner of the famous Ramsey sentence (Ramsey, 1929) that was also the building block of a view in the philosophy of science called structural realism. Ramsey basically suggested a way in which one could eliminate sentences in which 3

So both phenomenalism and the causal theory sketched here recognize that our thought of the physical world is conceptually parasitic upon our thought of appearance. Each recognizes that we think of objects and their properties through thinking about sensations and connections between sensations. The causal theorist, however, employs quantifiers in the analysis whose variables can range over something other than sensations. The phenomenalist is trying to avoid that. The distinction is captured rather nicely by the two passages from Mill quoted above. The first describes a causal theory; the second describes phenomenalism. It is also helpful to notice a structural analogy between phenomenalism and the causal theory on the one hand, and between logical behaviorism (the analogue of phenomenalism) and functionalism (the analogue of the causal theory) on the other. The realizer of the functionalist’s causal input/output descriptions is analogous to Hylas’s “matter” playing the causal role in the causal theorist’s analysis of the meaning of physical object statements.

215

Knowledge, Thought, and the Case for Dualism

theoretical terms appear to refer by translating them into quantified statements where entities can take the value of a variable but where the only genuine predicate expressions that occur are those that pick out sensible properties. If one locates the genuinely “observable” at the level of the phenomenal, and one “Ramsifies” ordinary talk about the physical world, one, in effect, gets the causal theory of objects. I don’t believe that Bertrand Russell had a consistent view about these matters, but by his own admission (1959) he never wavered from a commitment to direct acquaintance as the ultimate source of ontological commitment. In Russell (1948) he defends what amounts to Hylas’s “last stand”: I conclude that while mental events and their qualities can be known without inference, physical events are known only as regards their space-time structure. The qualities that compose such events are unknown – so completely unknown that we cannot say either that they are or that they are not different from the qualities that we know as belonging to mental events. (p. 231)

As we will see in a moment, it is sympathy with the kind of view inspired by Russell (but going back much farther, at least to Berkeley) that leads philosophers like Maxwell, Chalmers, and Stoljar to hold out some hope that the physical world might still find room for the exemplification of phenomenal properties. I’m not about to convince you here that the causal theory of objects is true. (I have tried to argue for such a view in Fumerton 1995, 2003, and 2006b.) To take the view seriously you would almost certainly need to feel the pressure of the skeptic’s arguments. And to feel that pressure you would probably need to accept a traditional form of foundationalism of the sort I defended in Chapter 3 – you would need to abandon epistemological direct realism, and feel the force of 216

An ontologically liberating skepticism

the idea that when all is said and done our only access to an external world, both conceptual and epistemic, is through the appearances it produces. Here, I merely want to point out that the causal theory of objects is completely agnostic with respect to the intrinsic (non-relational) properties of physical objects. Given that our only access in knowledge and thought to physical objects and their properties is through our thought about that which plays causal roles in producing phenomenal experience, we should be genuine skeptics about the intrinsic character of the objects (if any) to which we succeed in referring. That skepticism, of course, does not automatically translate into a skepticism concerning the external world. On this sort of view, one could consistently argue that our beliefs about physical objects are perfectly justified (though it certainly won’t be easy to deal with all variations on skeptical arguments). When we believe that a given physical object exists, the content of our belief is exhausted by the postulation of an entity with properties that have the potential to play a causal role in our phenomenal life. Again, an analogy might be helpful (one we used earlier in explaining functionalism). I have very little idea what is going on in my computer as it spell checks a document. I realize that there are incredibly many changes taking place at an amazing speed, but I really don’t know how the thing works in any sort of detail. But that doesn’t stop me from believing, perhaps justifiably, that my computer is in the process of spell checking a document. The content of my thought might make indirect reference to the inner workings of the computer but the descriptive content of the thought makes reference only to the input/output mechanism (whatever it is) that results in highlighted misspelled words. So also, on the extreme empiricism we have put forth above, we don’t know anything about the intrinsic character of physical objects. Nor do we have any occasion to think about such things (outside of 217

Knowledge, Thought, and the Case for Dualism

a philosophical context). The thought with which we are comfortable is thought that might best be characterized in the words of John Stuart Mill: it is thought of a permanent possibility of sensation.4 No doubt, the ordinary person would be taken aback at the suggestion that we neither think about directly, nor have any knowledge of, the intrinsic character of physical objects, but then the ordinary person wouldn’t understand what we are taking about when we use “intrinsic character” in this philosophically loaded sense. Daniel Stoljar (2001, 2006) makes what I think is essentially the same point when he characterizes two conceptions of the physical. Stoljar argues that we can think of the t-physical properties of an object, which he contrasts with the o-physical properties of the object. The o-properties are what I have been calling the intrinsic properties of the object. The t-properties of an object include its dispositional properties – its causal properties. When we conceptualize an object through its t-properties we are thinking of the object through our thought of its causal role in affecting the world in various ways. The radical empiricism described above essentially conceptualizes all of the physical properties discovered through the five senses as what Stoljar calls t-properties.

6.2

an ontologically liberating skepticism

This much is certain. We are directly and immediately aware of paradigmatically mental properties such as visual appearance and pain. Through that awareness we gain non-inferential knowledge that such properties are exemplified. This 4

Though as we saw earlier, Mill makes clear that his permanent possibilities of sensation are to be understood as the phenomenalist understands them. The permanent possibilities to which we refer here are the external objects thought about through their causal role.

218

An ontologically liberating skepticism

knowledge is the best sort of knowledge imaginable. There is no surer place to start one’s ontological commitments. The awareness that allows foundational knowledge of these properties also allows one to think directly about those properties. When one thinks of searing pain, one is not thinking of the property indirectly through some property it has. When one thinks of searing pain, for example, one is (typically) not thinking of it as that state, whatever it is, that results from damage to tissue and produces pain-healing behavior. One is not thinking about it as whatever it is that causes people to grimace. One is not (merely) thinking about the property that typically results from laying one’s hand on a red-hot burner. Rather, one is thinking of the searing pain as the property it is (not through some property it has). It is the physical world that is epistemologically and conceptually more problematic. We know the world of mindindependent, enduring objects only through the world of subjective and fleeting experience. Our thought isn’t quite limited to such experience, but it is limited to thought that we have through thinking about such experience. We can, as Berkeley suggested, imagine that we are thinking of a tree unperceived, but when we frame such a thought we are inevitably thinking of the way a tree looks.5 That is not to say that the thought of the tree is nothing more than the thought of an appearance. It is rather to say that the thought of a tree is nothing more than the thought of that which has the capacity to produce various experiences (under a variety of conditions). Hume was almost right when he said: 5

Unfortunately, Berkeley seemed to infer the tree’s mind-dependence from the fact that it was thought of. I have always wondered whether he wanted instead to remind us that in thinking of the tree unperceived we are still thinking of perceptions (ideas). Try Berkeley’s thought experiment. Think of the Eiffel Tower, but don’t think of how it would look, or feel, from some perspective. Can you follow the instruction?

219

Knowledge, Thought, and the Case for Dualism Let us chace our imagination to the heavens, or to the utmost limits of the universe; we never really advance a step beyond ourselves, nor can conceive of any kind of existence, but those perceptions, which have appear’d in that narrow compass. This is the universe of the imagination, nor do we have an idea but what is there produc’d. (Hume 1888, pp. 67–68)

If Hume restricted his comment to what we can imagine directly he would have been correct (at least with respect to the idea of those things that exist contingently).6 But as we saw earlier, even Hume allowed for the possibility of a relative idea – an idea of something thought of only as that which stands in a relation to something else. And one of the most familiar sorts of relative ideas is built on the relation of causation. We can think of something merely as the cause of something else. Again, we can’t think of everything this way or we would never be able to give thought a beginning. But we can think of most things this way, and if a Humean position were correct, we would think of everything physical that way. So how can any of this be of help to the physicalist? Well, we reached the point at which we were going to rest our property dualism on the critical observation that our thought of paradigmatic occurrent mental states is direct. As a result, we were tempted to conclude that we could give neither a Russellian/Fregean nor a Kripkean account of the informative nature of the claim that mental states are identical with certain physical states. But it is obvious that our radical empiricism opens the door again to informative identity claims because the thought of the physical is radically indirect. Why can’t the mental property whose existence we are sure of, and whose nature we understand completely, be an intrinsic property exemplified by some part of the brain or by some process occurring in the brain? 6

With the possible exception of the self – see Chapter 8.

220

An ontologically liberating skepticism

There was an old objection to the mind/brain identity theory. A version of it was suggested by Leibniz (1714) but was also considered by Smart (1959). The objection went something like this. Take something paradigmatically mental – a yellow after-image in the visual field say. Now look for it in the brain. Let’s imagine that as Jones has his yellow afterimage we lop the top of his head off and start peering around at his brain. We use all of the instruments currently available and we let our imagination roam concerning a utopian scientific future in which we have still other instruments of detection yet to be designed. Does anyone in his right mind think that in this way we will come across the yellow after-image? Does anyone in his right mind expect to discover something round and yellow in the brain of the person who has such an experience? Does anyone in his right mind think that one could figure out what experience the person was having solely through intensive examination of changes occurring in the brain? The answer to all of these questions is a resounding “No.” That is precisely what Jackson and his many predecessors were trying to stress in expressing their misgivings about physicalism. But then if the radical empiricist is right, one shouldn’t reach any dramatic conclusions about the intrinsic nature of brain states from any sort of perceptual observations we make of what’s happening in a brain. After all, through sense experience, the empiricist argues, we are never directly aware of the intrinsic nature of anything physical, and that includes, of course, the brain and the physical changes occurring in that brain. When studied scientifically, the brain (like tables, trees, rocks, and everything else physical) is known to us through the experiential effects it has on us. And the brain (like tables, trees, rocks, and everything else physical) is thought of by us through our thought of those effects. Perceptual knowledge leaves open the intrinsic nature 221

Knowledge, Thought, and the Case for Dualism

of brain states – the non-relational properties exemplification of which is involved in the occurrence of brain states. Again, we must be careful not to misunderstand the point. I am not here suggesting any sort of interesting skepticism concerning justified belief about the occurrence of brain states.7 In judging that the brain is in a certain state, I am judging that the brain is in that state whatever it is that plays a causal role in affecting sentient beings in various ways. In judging that the brain is in a certain state, I take no position on the intrinsic nature of that which plays the causal role. But introspection is not perception. There is nothing to stop me from wondering whether the brain states I’m thinking about contain as constituents, the very kind of property exemplifications of which I am directly aware in introspection. To be sure, the thought is initially a bit strange. But it might only seem so very strange because we tend to lapse into a crude epistemological direct realism (what Hume called the view of the vulgar) when we think about perceptual knowledge. We think that if there is a yellowish expanse in the brain, we ought to become directly aware of it when we empirically investigate the brain. But if the radical empiricist is correct, we never become aware of any external things or intrinsic properties of those things through perception. In perceiving the external world we become directly aware of our own phenomenal properties – properties we take to have an external cause. So it is at least intelligible to suppose that when we perceive someone’s brain, we become directly aware of phenomenal properties that our own brain exemplifies. To make the point vivid, we can imagine empirically investigating our own brains intent on discovering what is 7

I’m also not arguing that one won’t encounter interesting skepticism down the road. Like most other philosophers who take skepticism seriously, I am here “bracketing” the very real skeptical challenges that philosophers must meet in turning back external world skepticism.

222

An ontologically liberating skepticism

happening in those brains as we experience intense pain, or as we have a yellowish visual experience. Consider the latter. We introspect the phenomenal quality as we visually examine our brain (with or without the aid of instruments – it doesn’t matter). Through perception we don’t detect anything that looks yellow. What we are directly aware of are various shapes and colors – phenomenal properties. We take these to be indicators that various brain states are occurring. On the proposal we are considering, we are introspectively aware of yellowness while we are also introspectively aware of all sorts of other colors and shapes (the experiences produced as we visually perceive the brain). All of these phenomenal properties, we are now supposing, might be properties of brain states. We have postulated processes occurring in the brain that are causally responsible for the phenomenal states associated with visually perceiving the brain. Our judgment that those physical processes are occurring is agnostic with respect to the intrinsic character of the cause. But we are now speculating that the cause indirectly referred to in the judgment about the occurrence of the physical process might include as a constituent something that exemplifies that very yellowness of which we are also introspectively aware. The story seems to be perfectly intelligible. But what precisely is supposed to be exemplifying all of these phenomenal properties? Given the view under consideration, we can only guess. In order to answer this question we would need to reach some heavy-duty ontological conclusions about what sorts of things exemplify properties. Although I canvassed a number of views in Chapter 2, I took no position on whether there are such things as substances or particulars distinct from properties. If, however, there are such things, then there is no reason why some particular or substance that is a constituent of the larger substance that is the brain might not exemplify a phenomenal property such 223

Knowledge, Thought, and the Case for Dualism

as phenomenal yellowness. But substances aren’t the only kinds of things that can exemplify properties. Both properties and states of affairs can exemplify properties. Perhaps that phenomenal yellow is a property of a property exemplified in the brain. Or, perhaps it is a property of the occurrence of some state of affairs that involves the exemplification of nonphenomenal properties. If one were going to be an agnostic in general about the non-relational properties of physical objects and processes, one would be a fool to be all that confident about even the category of thing that might exemplify phenomenal properties. The view we are discussing here is at best a possibility. And we’ll point out shortly that it is highly misleading to think of it as a version of physicalism. Moreover, one must surely admit that it sounds more than a little strange to attribute phenomenal properties to brains or brain states. It is surely I who feels pain, who feels hungry, who seems to remember putting the keys on the mantel, who has visual experiences of a certain sort. It is not my brain, let alone some micro-event in my brain, that exemplifies these properties. We’ll talk more about the difficulty of eliminating the self from one’s ontology in Chapter 8. Certainly, if one can’t get rid of the self and one is a property dualist, one might as well attribute phenomenal properties to that self. But without the “I,” the property dualist had probably better soft-pedal complaints about how odd it sounds to attribute phenomenal properties to brain states, or to properties of brain states. After all, while it is undeniable that I describe myself as feeling pain, undergoing experiences, having desires, and the like, I also describe myself as having brown eyes, being six feet tall, having grey hair, and so on. Even Descartes will admit that I am six feet tall just in virtue of something that is not my self (my body) having certain dimensions. I’m brown eyed just in virtue of my eye having certain pigments. 224

An ontologically liberating skepticism

6.3

pa n p s y c h i s m

The “ontologically liberating” skepticism concerning the intrinsic nature of physical objects (including physical states of the brain) might seem to leave open the door for the possibility that phenomenal properties are, after all, exemplified by physical phenomena (where those physical phenomena are conceptualized in terms of their causal powers). The view is sometimes discussed under the rubric “panpsychism.” But this label is a bit unfortunate. I actually don’t think we have any positive reason to suppose that phenomenal properties are exemplified by brain states, but even if they were, it surely doesn’t follow that all of physical reality has a phenomenal character. And panpsychism is most naturally interpreted as a view about all of physical reality. To be sure, one could go there. Once one embraces agnosticism about the intrinsic character of the physical, one will be hard-pressed to justify too much outrage at allowing the conceptual possibility that atoms, rocks, and trees exemplify phenomenal properties.8 Despite the fact that Russell (1948) is often cited as one of the inspirations for the view, it seems clear that he didn’t take it all that seriously. Using “thought” as a generic term for mental events, Russell does say: When we come to events in parts of physical space-time where there are no brains, we have still no positive argument to prove that they are not thoughts, except such as may be derived from observations of the differences between living and dead matter coupled with inferences based on analogy or its absence . . . Just as we cannot be sure that the sun is not bright, so we cannot be sure that it is not intelligent. (Russell, 1948, pp. 230–231) 8

Timothy Sprigge (1983) not only allows for the possibility, but seems to think that one will inevitably be forced to such a position. But I can’t figure out what the argument is. For recent discussions of Sprigge’s view, see McHenry (2010) and Basile (2010).

225

Knowledge, Thought, and the Case for Dualism

But he quickly adds in a footnote: I do not wish the reader to take this possibility too seriously. It is of the order of “pigs might fly,” dealt with by Mr. CrawshayWilliams in the Comforts of Unreason. (p. 231)

Searle (2004) argues that panpsychism isn’t even up there with pigs flying: Aside from its inherent implausibility, panpsychism has the additional demerit of being incoherent. I do not see any way that it can cope with the problem of the unity of consciousness. (p. 149)

The charge of incoherence is strong, but we will return to the problem of the unity of consciousness to which Searle alludes in Chapter 8. In any event, whether panpsychism is incoherent or not, there doesn’t seem to be much to say for the view. Even if one does have some initial reason to wonder whether complex neural events might exemplify mental properties, it is hard to see why we would have the same sort of reason to consider seriously the possibility that all physical objects or events exemplify such properties. You will recall that on the view we sketched above, neural events (like all physical events) will be defined in terms of their powers to affect sentient beings. Return again to that neuroscientist investigating the workings of his own brain. He has the kind of visual experiences associated with c-fibers firing and postulates the occurrence of the “permanent possibility” of that kind of sensation. He also feels pain, let us suppose. At least he has a correlation between the theoretical change postulated (the neural event) and the introspectable pain. When he observes (indirectly) these same c-fibers firing in another brain, he doesn’t experience pain, so there is no reason to even contemplate the possibility that his phenomenal pain is a property of those c-fibers firing. By contrast, when a physicist observes a rock, say, the properties 226

An ontologically liberating skepticism

of the rock defined in terms of their causal power are, let us suppose, relatively stable, even as his visual and tactile experiences dramatically change as his situation relative to the rock changes. There seems no reason even to entertain the possibility that those phenomenal properties are properties of the rock. In addition, setting certain global skeptical problems aside, we can discover, as Russell suggested above, non-accidental correlations between the exemplification of certain phenomenal properties and behavior. If there is a solution to the problem of other minds, it lies with the argument from analogy. We know that there are connections between our grimacing and our being in pain, our moving away from something and our being afraid of that thing, and so on. When we conclude that other bodies are behaving in similar fashions we infer that there is a good chance that that behavior is also connected to the exemplification of phenomenal properties. We simply don’t have that kind of reason for thinking that stones or trees have phenomenal states. Again, that’s not to say they couldn’t. Property dualists are committed to the view that there can be phenomenal states without any behavioral indication that those states occur. But to allow for the possibility is one thing, to acquire any evidence that such states actually occur is a quite different proposition.

6.4

is th i s a v e r s i o n o f ph y s i c a l i s m ?

I’ve described the view that neural events themselves exemplify phenomenal properties as though it might be a view friendly to physicalism. But there is also a sense in which it is still a paradigmatic version of property dualism. You will recall that in Chapter 2 we eventually decided to define property dualism in terms of its rejection of various paradigmatic physicalist reductions of the mental to the physical. On the view outlined above, there is a sense in which properties like being 227

Knowledge, Thought, and the Case for Dualism

appeared to in a certain way, or feeling pain do not get reduced to paradigmatic physical properties. The property of being in pain, for example, is not the having of a disposition to behave in a certain way. The property of being in pain is not the second-order property of having a property that plays a causal role. The property of being in pain is not the property of one’s brain being in a certain physical state, and that’s true even if phenomenal properties are exemplified by brain states. This last claim, however, requires a very subtle treatment. On the extreme empiricism outlined above, the physical characteristics we attribute to brain states based on perception of brains are causally defined characteristics. To describe the brain as being in a certain state is to describe the brain as having that property whatever it is that plays the critical causal role affecting perceivers. To be clear it is not the causal role it plays in the conscious life of the person whose brain state it is. It is the causal role that the state plays in affecting the way in which normal people are appeared to perceptually as they observe the brain.9 Now attribution of a brain state to a person will involve a kind of indirect reference to whatever plays the causal role, but the characterization of the brain state is silent on what that property is. Put linguistically, when one ascribes to the brain a certain property, one will employ variables leaving open the question of what takes the value of the variable. The view sketched above allows for the possibility that what takes the value of the variable either is a phenomenal property, contains a phenomenal property as a constituent, or exemplifies a phenomenal property. But we do not ascribe the phenomenal property to the brain when we characterize that brain as being in a certain state. 9

The more technical and theoretical a term is, the more likely it is that the causal powers that define the term are powers to affect sentient beings whose perception is aided by instruments.

228

An ontologically liberating skepticism

It might be useful to compare the view of physical object descriptions sketched above, with the (mistaken) functionalism discussed in Chapter 2. You will recall that the functionalist should say that when we ascribe a mental state to a person we are asserting that the person is in that state whatever it is that plays a causal role. To say of a person that he is in pain, on this view, is to say something like that the person is in that state whatever it is that results from damage to the body and that in turn causes behavior conducive to the healing of that damage. The functionalist introduces the technical concept of a realizer of the functional state. The realizer of the functional state is that state that plays the causal role. It can be different in different creatures. But if we want to allow that the property of being in pain will be the same property even if it is realized by different states of different organisms, we should not as functionalists identify being in pain with its realizer (a realizer that may well be even radically different in different organisms). The philosopher who treats all ascriptions of properties to physical objects as, in effect, secondary properties, as powers to affect sentient beings in certain ways, has a view about the ascription of properties to physical objects that is just like the view that the functionalist brings to understanding the ascription of mental properties. In describing that lemon as sour we are saying of the lemon that it has a property that has the capacity to play a critical causal role in producing that familiar sour taste sensation. If we want to allow, as we should, that the physical realizer of the causal role in sour objects can change dramatically from sour object to sour object, then again, we don’t want to identify the property of being sour with its realizer in a given sour object. In an ascription of properties to the brain we will again be able to distinguish the property of being in a certain brain state, from whatever property it is that plays the role of realizer. The characteristic of the brain state 229

Knowledge, Thought, and the Case for Dualism

will be the causally defined property. It will not be the realizer of the causally defined property. An even simpler analogy might be helpful. Let’s consider one more time that indirect way in which we can describe an object’s color. In addition to saying of an object that it is blue, red, yellow, and so on, I can say of the object that it has my aunt’s favorite color. In describing the object as having my aunt’s favorite color, I am clearly not saying of the object that it is red, blue, yellow, or any other specific color. All I assert is that it has whatever color it is that my aunt likes more than any other color. If my aunt’s favorite color is blue, then I am still not saying of the object that it is blue when I say that it has my aunt’s favorite color. Or at the very least, we should view that as a highly misleading way of characterizing the content of my claim. The property of being my aunt’s favorite color is a property that can’t be exemplified in a world in which my aunt doesn’t exist. The property of being blue can be exemplified in a world in which my aunt doesn’t exist. But isn’t there any sense in which if the definite description picks out blue, I am ascribing blueness to the object? Probably there is. We can certainly explain the sense in which blue “realizes” the property of being my aunt’s favorite color, and we can certainly say that when one ascribes a second-order property to an object, there is also a sense in which one ascribes the first-order property that exemplifies the second-order property – this is a matter of terminological decision (and won’t be settled by ordinary language considerations). One might also suppose that the above discussion simply makes obvious that there is a scope ambiguity in the ascription of my aunt’s favorite color to an object. The statement “The shirt has my aunt’s favorite color” can be read in one of two ways: (1) there is a color property that my aunt likes more than any other color and the shirt has it; (2) the shirt has that property whatever it is that my aunt likes more than any 230

An ontologically liberating skepticism

other color. But I’m not sure that in the final analysis (1) can be read differently from (2). If you could use definite descriptions to secure reference, discard the reference-securing definite description after it does its work, and then do things with the referent (like predicate it of things), there would be the relevant ambiguity. But the anaphoric reference of “it” in (1) seems to me to refer right back to the definite description. What makes (1) true is still the existential fact that my shirt has a color that my aunt likes more than any other color. If the realizer of the physical property we ascribe to the brain is, includes, or exemplifies a phenomenal property, shall we say of that property that it is physical? It doesn’t matter at this point. We have said what it is, and we have said what it is not. If (and as we have emphasized, this as a huge “if”) phenomenal properties get picked out indirectly in the way we have described and you want to call them physical properties, go ahead. What is critical is that they be included in one’s ontological inventory, and that we reject the direction of the physicalist’s reductions. The existence of the phenomenal properties is known better and understood better than any of the scientifically understood physical properties on which the physicalist wants to model our knowledge and understanding of the phenomenal.

6.5

s ummary

There are phenomenal properties and there is non-inferential propositional knowledge that they are exemplified. That knowledge can be reached only through the introspective act of acquaintance with those properties being exemplified. The same act of acquaintance that gives us knowledge that phenomenal properties are exemplified also generates the capacity to think of such properties directly. The phenomenal properties with which we are acquainted and about which we think 231

Knowledge, Thought, and the Case for Dualism

directly are not functional properties, dispositions to behave, or properties of the brain as those properties are usually thought of. There is, however, room for the possibility that those phenomenal properties are exemplified by the same things that exemplify the properties of the brain, or are exemplified by the complex state of affairs that is the brain’s exemplifying whatever other intrinsic properties it has. I emphasize, however, the limited nature of this last claim. It is hard to know how to bundle the phenomenal properties with other properties. And it is particularly hard if the radical empiricist is right and the intrinsic properties of physical objects (including the brain) are, in general, hidden from perception. That some aspects of the physical world (individuals, non-phenomenal intrinsic properties of physical objects, or states of affairs exemplifying such non-phenomenal properties) might themselves exemplify phenomenal properties is a hypothesis that remains only an abstract epistemic possibility.

232

7 Objections and replies Given that it is so obvious that we have introspective knowledge of the exemplification of phenomenal properties, and that after gaining such knowledge we have the capacity to think directly of such properties, what is really curious is that philosophers have so resisted allowing them into their ontology. Some of this resistance can, no doubt, be blamed on the curious naturalism that has gained so much acceptance among contemporary philosophers. That naturalism gives a kind of priority and respect to scientific knowledge over what the naturalist takes to be more problematic phenomenologically based knowledge. But as we saw in Chapter 3, scientific knowledge in general, and of the mind in particular, ultimately rests on the kind of knowledge about which the naturalist is suspicious. One’s fundamental ontological commitments must always be based on phenomenological acquaintance. So much for rhetorical flourish. But physicalists must have reasons for rejecting qualia from their world view. Let’s examine some of those reasons here.

7 . 1 ja c k s o n ’ s r e j e c t i o n of the knowledge argument It was disconcerting to some property dualists to learn that Frank Jackson, one of the most influential of the “qualia freaks,” has abandoned the view that there are qualia and, with it, the view that the knowledge argument establishes that such properties are exemplified (Jackson, 2004). If I understand his position correctly, it is grounded on his acceptance of 233

Knowledge, Thought, and the Case for Dualism

a radical intentionalism about sensory states – the sort of states that he used to think involved the exemplification of qualia. The basic idea behind intentionalism is that sensations, including visual, tactile, auditory, olfactory, gustatory, kinesthetic, and even pain, are intentional states like belief, desire, and fear. We noted earlier in this book that many contemporary philosophers of mind contrast the so-called hard problem of qualia with what is supposed to be the easier problem of analyzing intentional states. The idea is usually that intentional states like belief, desire, and fear more easily admit of a successful reduction to functional states. That presupposition, I argued in Chapter 2, is mistaken. Analytic functionalism is no more plausible than logical behaviorism and logical behaviorism isn’t the slightest bit plausible. I’m not sure, however, that Jackson’s conception of intentional states is functional. Rather, I think that his view is a species of the kind of view advanced by Moore (1922) and, later, Harman (1990). On this view of intentionality, intentional states are relational and “diaphanous.” Their character is exhausted by their “object.” One might get a feel for what this means by thinking about belief. When one believes that one’s car is in the parking lot, can one find, phenomenologically, the believing as something that is distinct from what is believed? Like relations in general, one might suppose that an attempt to locate the “act” without focusing on its object will be hopeless. Of course this can’t be the whole story or we wouldn’t have an account of the distinction that obviously exists between different intentional states with the same content. There is a world of difference between believing that there is a God, fearing that there is a God, and hoping that there is a God. Yet all three intentional states have precisely the same content – that there is a God. Jackson now seems convinced that when Mary acquires her color experience she gains a new intentional state which represents physical objects as being a certain way. When she is 234

Objections and replies

appeared to as if there is something blue in front of her, she is in a representational state whose content is that there is something blue. That representational state can correctly represent the physical world or it can misrepresent the physical world. Either way, the character of the intentional state is exhausted by its content, by what it represents. Although the attempt seems to me almost desperate, one might try to take the same tack in analyzing searing pain. Searing pain, on this sort of view, is also an intentional state that represents damage to the body. I don’t just feel pain. Typically I feel pain as located in a certain part of my body. Philosophers of mind got very used to thinking that if pain has a location at all, it is in the brain. To be sure, those philosophers usually admit, we say that the pain is “in” our elbow, our back, or our foot, but we need only think about phantom-limb pain phenomena to realize that we can’t really place the pain where we describe it as being. It is a well-known fact that a person whose foot has been amputated will still often report feeling pain “in that foot.” But the foot has long been dumped into an incinerator – there is no foot for the pain to be in. The pain obviously still exists and thus we need to locate it somewhere other than the foot. The Cartesian “places” it in the mind. The mind/brain identity theorist places it in the brain. But the intentionalist or, as we might also call this philosopher, the representationalist, thinks that the pain is a representation of tissue damage and the phenomenon of phantom limb pain is no more problematic than the phenomenon of false belief, unfounded fear, or unfulfilled desire. Intentional states can represent things as being a certain way even when things aren’t that way. So Jackson thinks that Mary with her newfound color experiences does not exemplify or find exemplified a phenomenal property of which she was previously ignorant. Rather she represents (in a new way) color properties as properties of 235

Knowledge, Thought, and the Case for Dualism

physical objects Describing the intentionalist’s view of how to understand Mary’s new experience of the phenomenal “property” of being red, Jackson (2004) says: Intensionalism tells us that there is no such property. To suppose otherwise is to mistake an intensional property for an instantiated one. Of course, when I sense red and you sense red, there is something in common between us that we English speakers report with descriptions that include the word “red.” But what is in common is not the property tagged with the word “red” but, first, how things have to be for our experiences to represent correctly, and, second, our both being in states that represent things as being that way. (p. 430)

Although he doesn’t definitively commit himself, Jackson also seems clearly sympathetic to the idea that the representations in question are misrepresentations – that there really is no property of the sort represented: Intensionalism means that no amount of tub-thumping assertion by dualists (including by myself in the past) that the redness of seeing red cannot be accommodated in the austere physicalist picture carries any weight. That striking feature is a feature of how things are being represented to be, and if, as claimed by the tub thumpers, it is transparently a feature that has no place in the physicalist picture, what follows is that physicalists should deny that anything has that striking feature. And this is no argument against physicalism. Physicalists can allow that people are sometimes in states that represent that things have a nonphysical property. Examples are people who believe that there are fairies. What physicalists must deny is that such properties are instantiated. (p. 431)

And he also says: In consequence, color experience presents to us as if it were the acquisition of information about highly salient, more or less intrinsic features of our surroundings. But there are no physical features fitting this characterization; in consequence color experience presents itself to us as if it were information about certain

236

Objections and replies nonphysical features. Indeed, we may want to go so far as to say that sensing red misrepresents how things are. If this is right we should say that nothing is red, for nothing would be as our experience of red represents things as being; we should be eliminativists about red and about color in general. (pp. 431–432)

Because of Jackson’s emphasis on misrepresentation, his view is quite different from those who think that Mary, with her new color experiences, has found a new way of representing the same old states of light reflection and absorption that she had already successfully represented in her color-deprived condition.1 Because the intrinsic nature of an intentional state is exhausted by its intentional content (in the case of color experience, a content that might be a non-existent state of affairs) our phenomenal properties (the qualia) have disappeared. In the case of accurate representation the phenomenal qualities are replaced by the exemplification of the physical properties represented. In the case of misrepresentation the phenomenal qualities are replaced by the non-existent, or non-obtaining state of affairs represented. Before evaluating this sort of position in more detail, I can’t resist marveling at what some naturalists will find natural! At a Rutgers Epistemology Conference (2006) I was talking to Gil Harman about the sort of view I just described, and wondered how philosophers so anxious to avoid sense data and qualia seem so willing to embrace a realm of Meinongian non-existent beings. The only way you can really hang on to an account of intentionality that makes sensory states both intentional and diaphanous 1

Though Jackson (2004) also opens the door to a view according to which color experience is a combination of misrepresentation and “close enough” representation of certain physical properties: “A more moderate position is that although our experience of color contains substantial degree of misrepresentation . . . there are complex physical properties ‘out there’ that stand in relations near enough to those captured by the color solid for us to be able to identify them with the various colors” (p. 432).

237

Knowledge, Thought, and the Case for Dualism

is to find “objects” for non-veridical experience. And if one rejects sense data and their qualia, intent on finding some “respectable” physical object content to the non-veridical sensory experience, one will need to embrace these non-existent, or non-occurrent states of physical objects. Harman’s view is, I think, that one should go fearlessly where the phenomenology leads one, even if that is into the land of Meinong. Now as you must have gathered by now, I have no use for the distinction between natural and non-natural properties in general. If a property is exemplified, its exemplification is part of reality, the reality that is nothing other than the exemplification of properties. The exemplification of phenomenal properties is just as natural as the exemplification of any other sort of property. But one doesn’t need an appetite for desert landscapes to worry about introducing into one’s ontology things that don’t exist. To be sure, some find a non-occurrent state of affairs less troubling than a non-existing Pegasus,2 but it has never seemed to me to make that much of a difference. I have no idea how to even understand the difference between the state of affairs that doesn’t exist (or, if you prefer, doesn’t obtain) and the state of affairs that is “infused” with reality. If one accepts an intentionalist account of representation and one accepts a relational account of intentionality that construes the character of the intentional state as exhausted by its intentional object, one will never be able to avoid the cluttered landscape of Meinong’s exotic world. But one needn’t accept a relational account of intentionality. On my own view genuine intentional states are non-relational properties that are sui generis and that alone among property exemplifications have the 2

Chisholm, for example, throughout his philosophical writings was always anxious to avoid commitment to non-existent things, but built a good part of his ontology around the supposition that there are non-obtaining states of affairs. Non-obtaining states of affairs were, in effect, identified with false propositions. Obtaining states of affairs were true propositions.

238

Objections and replies

capacity to correspond. When there is misrepresentation there is no correspondence – one needn’t look for a non-existent object that is related by the intentional act. But if one construes intentional states this way, their existence just is the exemplification of certain properties – the very properties that the property dualist will seize on as defying physicalist reduction. That color experience that Mary has is the exemplification of a property. I don’t think that the exemplification of the property is an intentional state (though in us it is almost always and spontaneously accompanied by intentional states). But even if it were an intentional state it is not diaphanous in the way that someone like Harman supposes. There are no Meinongian non-existent objects to give Mary’s visual experience its “color.” If we are going to find the dramatic change in Mary’s experience, we must look to Mary’s experience. To repeat a refrain the reader might well be tired of, we must rely on our direct awareness of color experience to find the phenomenal property of which we are speaking. Without phenomenology, I repeat, philosophy is blind. But when we locate the property, we will be committed to its being exemplified and we will be able to think directly of that property in the absence of its being exemplified. Perhaps the exemplification of that property constitutes a state that has the capacity to correspond to the world. If it does then we can call the resulting state an intentional state. But we cannot make the property that is doing the representing disappear. And if it doesn’t disappear, we still have to figure out how to make room for it in a physicalist world view. After all, Mary never ran across that property when confined to her black and white room.

7.2

fo u n d a t i o n a l i s m o f k n o w l e d g e a n d th o u g h t

I have tried to make clear throughout this book that the case for property dualism rests on a foundationalism of knowledge 239

Knowledge, Thought, and the Case for Dualism

and thought. Put as simply as possible, our direct thought of phenomenal properties tells us that if such properties are exemplified they resist all paradigmatic physicalist reductions. Our direct acquaintance with the exemplification of such properties (when we have the relevant thoughts and are acquainted with the correspondence between such thoughts and the exemplification of phenomenal properties) gives us direct knowledge that phenomenal properties are actually exemplified. I acknowledged in the introduction to this book that in making the case for a philosophical thesis, one should try to minimize the number of controversial philosophical premises upon which one relies. But I also confessed that what makes the philosophy of mind so difficult is that the issues in this field are so intimately connected to other difficult philosophical issues in metaphysics and epistemology. Years ago I presented a paper arguing for property dualism relying on at least some of the premises defended in this book. In the question period, a prominent philosopher asked me if I thought that the argument I presented was a decisive argument for dualism. I was a bit taken aback. Obviously, I represented the argument as being sound. But there is another sense in which no philosophical argument that rests, in part, on other controversial philosophical views, is decisive. The problem is not that there aren’t answers to philosophical questions that are either true or false. The problem is that there are so many twists and turns on the road to answering fundamental philosophical questions that one would expect rational people to disagree.3 3

The literature on philosophical disagreement and its epistemological significance is important and fascinating. We know that philosophers who we regard as intellectual peers often have diametrically opposed positions to ours. The question is whether we should regard this knowledge as defeating whatever other evidence one might have had for the positions one is inclined to accept. For a range of views on this question, see Feldman and Warfield (2010). For my own view see Fumerton (2010) in that volume.

240

Objections and replies

In any event, I’m prepared to concede that if one couldn’t make sense of direct acquaintance and direct thought, one has no reason to accept dualism. To repeat again the refrain, without phenomenology philosophy is blind. It seems to me crucial that acquaintance not be construed as a species of intentional state, that it be construed as the act by which one gets a property before mind. If one argues, alternatively, that the only thing one can do in trying to locate a property (whether it is a property the exemplification of which constitutes an intentional state or not) is move up a level to get oneself in yet another intentional state, regress awaits. Access to the metastate would be through yet another higher-level state and so on ad infinitum. The resulting regress really does seem to me vicious. It makes the “veil of perception” transparent by comparison. If all access to everything is through intentional states, then properties and states of affairs we try to grasp will always recede beyond our reach, hidden from us by a veil of representation, a veil that has infinitely many layers blinding us from each of the levels below that it screens. A mind that faces such a regress is a mind that is just as blind as one incapable of representing reality at all. As I argued earlier, it is direct awareness, a real relation, that enables us to get that critical grip on some aspects of reality, a grip that can get us started in our attempt to represent the rest of reality. I have argued earlier that the key to understanding informative identity claims lies in the distinction between thinking of some thing, state, or property directly, and thinking of that same thing, state, or property only indirectly. While it might not be impossible to think of a phenomenal property directly without having first been acquainted with it, the actual world is such that our direct thought of qualia would never occur absent our direct awareness of qualia. But does one need to embrace content internalism of the sort I defended in this book in order to have reason to accept property dualism? I’m afraid 241

Knowledge, Thought, and the Case for Dualism

the answer is “Yes.” The fundamental idea behind content externalism is the idea that a thought, or at least a certain kind of thought (perhaps a simple thought), has its content determined in part by the causal history of the thought. The view is still capable of making a distinction between direct and indirect thought, but it will remain externalist if direct thought is understood in an externalist’s framework. If such a view were correct, and if there is more than one causal route to some state’s becoming a direct thought of some property z, then there is no reason why there couldn’t be two quite distinct representational states, x and y, both of which directly represent the same property z. But then as I also argued, on a view like that it seems to me that one wouldn’t be able to read off introspectively the content of one’s thought. And that I take to be a reductio of content externalism. Admittedly, this last claim is hardly uncontroversial. Relying on an analogy, David Henderson once tried to convince me that acknowledging external constraints on the content of thoughts is perfectly compatible with allowing that one has direct introspective access to the content of those thoughts. He held up a quarter and asked if we agreed that a metaphysically necessary condition for this being a quarter is that the piece of metal had a long and rather complex history that probably involved among other things the creation of the Department of Treasury (not to mention the creation of the United States of America). After securing that agreement, he pointed out that I hardly viewed my knowledge that he was holding a quarter in his hand as inferential knowledge, or, at least, as inferential knowledge involving that much complex background knowledge of other propositions. The argument is similar to one put forth by Donald Davidson (2003) when discussing what it is to be sunburned. Like Henderson, Davidson gives an example that is supposed to help reconcile externalism with the possibility of introspective 242

Objections and replies

knowledge of intentional states. He argues that just because my being sunburned implies the existence of the sun, it surely doesn’t follow that being sunburned isn’t a condition of my skin. In the context of the present debate, this might seem to imply that we can surely notice (be aware of) someone’s sunburn without having to reach conclusions about the existence of the sun or the sun’s causal role in producing the burn. And this is true despite the fact that one couldn’t develop a plausible story about what sunburn is without talking about the way in which the sun causes the skin to change color. Presumably, the pink skin color is supposed to be analogous to the internal state that we know through introspection. The causal origin of the skin color, a necessary condition for the person’s being sunburned, is supposed to be analogous to the causal origin of the internal state, a necessary condition for its being an intentional state with the content it has. Are either Henderson’s or Davidson’s examples instructive? For many of us both analogies are a bit problematic because we don’t think that we are directly aware of physical objects like quarters, or the color of skin in the same way that we are directly aware of our mental states. But let’s leave that aside. The awareness of the coin and the sunburn is still supposed to be more direct than knowledge of the complex causal history that determines the metal’s being a coin and the skin color’s being a sunburn. So do the analogies help the externalist reconcile content externalism with introspective access? I think not. To see that someone is sunburned, to be aware of the fact that someone is sunburned, does involve knowing that the skin color has a certain causal origin. If, for example, I see someone with reddish skin that is, in fact, caused by a rash, it is simply false that I am aware of the fact that the person is sunburned. Without some reason to believe that the skin color is caused by exposure to the sun, I have no reason to conclude that the person is sunburned and I won’t know that 243

Knowledge, Thought, and the Case for Dualism

the person is sunburned. Without employing the concept of causation, I can’t even see the person as sunburned. In such a case, I might still be described as seeing a sunburn. But notice that the sense in which someone can see an F without even believing that the thing is F is irrelevant to knowing that an F exists. The alleged claim about intentional states is not merely that one is aware of them in the sense in which one can be aware of a coat rack without seeing it as a coat rack. It is, rather, that one is aware of them as intentional states – one is aware of them in a way that can give rise to propositional knowledge of what one is thinking. And the same is true of becoming aware that someone is holding up a quarter. Either the causal history of the piece of metal is conceptually critical to its being a quarter or it is not. If it is, then knowing that the metal is a quarter will involve knowing that it has the relevant causal history. One might worry that people don’t have the conceptual sophistication even to realize that something’s being a quarter involves having the relevant causal history, but that worry is surely implausible. Ordinary people understand perfectly the concept of counterfeit money. They know that something’s being a dollar bill, or a quarter, is not simply a function of its non-relational character4 – its shape, its color, its weight, etc. To be sure, they might have only an incomplete grasp of what the relevant causal history is, but in judging something to be money they are implicitly judging that it has the “right” causal history.5 What all this does show is that ordinary judgment, and, in particular, judgment about artifacts often involves far more than some philosophers wish to acknowledge. But the content 4

5

Here I am using “non-relational” loosely. Strictly speaking, none of these properties should be viewed as non-relational. And recall that we have allowed that we can “borrow” whatever is plausible in a causal theory of reference, both for names and kind expressions to find descriptivist content for expressions about whose referent people know very little.

244

Objections and replies

internalist’s reductio of externalist views still stands. If the very concept of an intentional state’s content requires reference to the state’s causal history, then we wouldn’t be able to know introspectively what the content of the state is. But we can. And, on the externalist’s view, if one could know introspectively what the content of the state was, then one could use that introspective knowledge with the a priori knowledge of alleged philosophical insights about conditions determining content to discover from the armchair complicated facts about the historical causal chains. But we can’t.

7.3

t h e c a u s a l th e o r y o f o b j e c t s

In Chapter 6, I defended the view that our thought of the physical is parasitic upon our thought of sensation. I expressed some sympathy for that last ditch attempt of Hylas’s to make intelligible a purely relative idea of matter. It is an understatement to admit that it is no small task to develop a plausible “causal theory” of physical objects. As I indicated earlier, I tried to defend such a view in Fumerton (1985). Does the case I have presented for property dualism stand or fall on a successful defense of a causal theory of the physical? I don’t really think that it does. There are at least three (and probably more) possibilities. One is that there is a plausible causal theory of objects that will accommodate commonsense views about the nature of physical objects. A second is that our thought of the physical is through thought of sensation, but it is confused thought and there is no plausible analysis of the meaning of physical objects in terms of their potential causal role in producing sensations. This is one way of understanding Hume. Unlike Berkeley, the phenomenalist, or the causal theorist, Hume concluded, I think, that while our thought of the physical is through our thought of (what he called) perceptions, there is no way of reconciling this fact with our 245

Knowledge, Thought, and the Case for Dualism

commonplace beliefs about physical objects.6 Lastly, one might reject entirely the idea that our thought of the physical and its properties is always through thought about the phenomenal. One might argue that one has direct thought of at least some physical properties that is entirely distinct from thought of subjective experience. I don’t think that view is plausible. I can’t form the thoughts in question. But I also don’t see how the view could help those who want to reject property dualism if my earlier arguments about the nature of informative identity claims are correct. Direct thought about phenomenal qualities and direct thought about physical properties will expose the fact that if such properties are exemplified they will be intrinsically different.

7.4

causal overdetermination

Hostility towards dualism in general, and property dualism in particular, has often been grounded, in part, by the physicalist’s deep suspicion that the exemplifications of such properties couldn’t possibly causally interact with the physical world and are, thus, “nomological danglers.” Ockham’s razor is then pulled out to tidy up our ontology. Smart’s attitude is paradigmatic:

6

This may have been Hume’s approach to a number of metaphysical issues. Consider his views about causation, for example. Hume offers both an “objective” and a “subjective” definition of causation. Crudely put, the objective definition is in terms of brute regularity, the subjective definition appeals to “determinations” of the mind produced by custom. He knows perfectly well that his readers won’t like either definition. The “objective” definition leaves out the necessary connection that ordinary folk think is crucial to causation. The “subjective” definition has the apparently bizarre consequence that there would be no causation without conscious beings. Hume’s attitude seems to be that there simply is nothing else that could plausibly be meant by “cause,” so we’ll just need to get used to a “revisionist” story about the meaning of causal claims.

246

Objections and replies That everything should be explicable in terms of physics (together of course with descriptions of the ways in which the parts are put together – roughly, biology is to physics as radio-engineering is to electro-magnetism) except the occurrence of sensations seems to me to be frankly unbelievable. Such sensations would be “nomological danglers,” to use Feigl’s expression. (Smart, 1959, p. 142)

In approaching this issue, I should again emphasize what is by now obvious. There is no stronger reason to commit oneself to a property than the fact that one finds oneself phenomenologically acquainted with the property. Acquaintance with F at t is not an intentional state caused by the exemplification of the property (though it might be a state caused, in part, by the exemplification of F at t – 1). Acquaintance is a real relation and its obtaining is contemporaneous with the existence of its object. One can’t use a razor to slice away from one’s ontology one’s searing pain. If you can’t find a causal role for it to play, then it is obviously a mistake to use the fact that something plays a causal role as a criterion for whether or not it exists. Having indicated that it would be an almost comical error to use causal efficacy as a general criterion for existence, we might still wonder what kind of causal role the phenomenal properties to which we are committed might play. In discussing these issues, it might be helpful to become a bit clearer about what bothers the physicalist so much concerning the postulation of qualia. Sometimes the physicalist gestures vaguely in the direction of ghosts in machines and the utter absurdity of supposing that entities in space could causally affect and be affected by entities that have no spatial location. Of course, the first point that should be noted is that objects don’t cause objects. To be sure, Hume sometimes talked that way, but when being careful, even he was clear that it is only an object’s exemplifying a property or properties that, properly speaking, should be viewed as the relata of a causal connection (Hume 1888, Bk. 1, XV). 247

Knowledge, Thought, and the Case for Dualism

Still, in searching for causal connection, one might, as Hume did initially, hit upon spatial contiguity as at least a half-way plausible necessary condition for a’s being F directly causing b to be G. If one is at all tolerant of our ordinary way of speaking, one must emphasize the “directly.” Outside of a philosophical context, no one is the slightest bit fazed by someone’s identifying the harsh armistice agreement after World War I as one of the causes of World War II. In any event, it is hard to see what spatial contiguity has going for it as a conceptually necessary condition for causal connection. Although there probably is no such thing as mental telepathy, it also seems unreasonable to reject its existence on conceptual grounds. Gravitational attraction is a paradigm of causal connection, and although one might postulate entities to “fill” the space between objects exercising causal influence at a distance, it doesn’t seem to me inconceivable that there should be genuine, ineliminable, action at a distance. To be sure, however, action at a distance is different from action without any spatial relations. To get to the heart of the issue of whether one should blanch at the thought of causation without spatial relations one must get clearer about the relata of causal connection and the nature of causation itself. On a regularity theory of the sort inspired by Hume, the most obvious candidates for the relata of causal connection are the exemplifications of properties. The most perspicuous form a causal claim will take is the following: a’s being F caused a to be G, or a’s being F and standing in R to b caused b to be G. Of course, the complete cause and the complete effect might be extraordinarily complex. Reference to property exemplification is certainly critical in the characterization of the relata of causal connection on a regularity theory. It is through reference to the properties by which we characterize cause and effect that the relevant regularity falls out. If the claim is that a’s being F caused a to be G, then the 248

Objections and replies

relevant regularity will be that all F’s are G. A similar point applies to what we might call generality theories of causation. Like the regularity theorist, the generality theorist thinks that causal connection is constituted in part by lawful regularity. But unlike the regularity theorist, the generality theorist leaves open the question of whether one can analyze lawful regularity in terms of brute contingent regularity. The most straightforward generality theory that rejects a regularity theory takes necessary connection between universals to be the essence of causal connection. If such a view were correct, then particular cases of causal connection would hold when those universals are exemplified by particulars. I’m not about to argue here for either a generality theory of causation, or the species of generality theory that is the Humean regularity theory. Both theories, but particularly the Humean regularity theory, face formidable objections. But it is hard to deny the plausibility of the fundamental thought underlying the generality theory, namely that it is the exemplification of properties that generates causal connection. But nothing in the generality theory per se speaks to the question of whether there can be causal connection occurring between states of affairs even when none of the particulars involved in those states of affairs are spatially contiguous, or even spatially related. So one is still left searching for a reasonable explanation for the physicalist’s antipathy towards the idea of mind/ body or mental property/physical property causal interaction.7 One source of potential concern relates again to a worry about overdetermination. It is taken by most philosophers to be an unassailable truth that the universe is physically causally closed. That thesis amounts minimally to the claim that every 7

And I’m not admitting that alternatives to generality theories present any difficulties for causal interaction between mental and physical phenomena. Davidson (1993) seems to think that causation is a primitive relation holding between events (where events are also taken to be primitive bearers of certain properties).

249

Knowledge, Thought, and the Case for Dualism

physical occurrence has a physical cause. A stronger claim is that every occurrence has a physical cause. Now I’m not sure how precisely to understand this claim in a universe that is also described by many physicists as radically indeterministic. It wasn’t long ago that we used to hear a great deal of talk about the singularity – the Big Bang that was supposed to create matter and energy (and space and time). As it was sometimes described, it seemed to be a physical occurrence that had no cause.8 But leaving aside such grand occurrences, I also gather that many are convinced that at the micro-level there is a radical indeterminacy with respect to such events as radioactive decay. Some philosophers of science think that radical indeterminacy is perfectly compatible with causal connection (Armstrong, 1983; Fales, 1990; Anscombe, 1993), but others (famously Hempel, 1965) are equally convinced that where there is genuine indeterminacy things happen without a causal explanation. It is probably still going to be possible to put forth modified, more sophisticated versions of a causal closure principle. One might restrict one’s principle to macro-events. Alternatively, one might argue that, at the very least, antecedent conditions place limitations on what can happen, and that nothing other than a physical antecedent condition can place such limitations.9 But let’s set aside these technical worries and concede, for the sake of argument, that the exemplification of mental properties to which we are committed is a part of a deterministic system. And let’s suppose further that we want to say that the severe pain I feel has a physical cause. The more remote 8

9

I just don’t know enough about the physics to make informed claims about some of this, but I gather that while some physicists asserted that the singularity had no cause, others remained completely agnostic about what might have happened “before” the singularity. I am indebted to my colleague Evan Fales for helpful discussions I have had concerning these matters.

250

Objections and replies

cause is damage to the body; the proximate cause is a brain state. Before we became involved in abstract philosophical discussion, we wouldn’t hesitate to identify the pain we feel as itself the cause of a great many other changes – changes in both our mental life and our physical behavior. The pain causes me to become depressed, to complain, to grimace, to worry, to seek medical help, to lose concentration, etc. Let’s take the simple example of grimacing. Unless we identify the pain state with some brain state we have two candidates for the cause of the grimace. We can say that the grimace is caused by the proverbial c-fibers firing, or we can say that the grimace is caused by the excruciating pain. Alternatively, we can say, of course, that both claims are perfectly true. On one view that we sketched in the last chapter, a brain state is picked out only indirectly. In characterizing the brain as being in that state, we make no assertion about the intrinsic properties exemplified in the brain. So we could treat the causal claim as in some respects analogous to a familiar sort of claim that we make about many other objects. I just barely tap the vase on your desk with my pen and it shatters into a thousand pieces. “Why did that happen?” I ask. “I just barely touched it.” You respond that it is made of a special sort of glass that is highly fragile. Again, while philosophers might start worrying about such explanations, ordinary people find them perfectly acceptable. What would be the philosophical worry? Well, it is an old problem. On a natural explanation of fragility, explaining why something breaks by reference to its fragility might seem vacuous. After all, to say that X is fragile is just to say that X would break easily if hit. I want an explanation of why it broke so easily and you tell me that it is the kind of thing that breaks easily! Not much of an explanation. Certainly, not as good as a chemical analysis of the glass replete with some detailed explanation of how certain structures change upon impact. But then we wouldn’t think we had to choose between 251

Knowledge, Thought, and the Case for Dualism

the fragility explanation and the more detailed chemical explanation. Why do we find the fragility explanation helpful at all? A great many answers have been proposed, but it seems to me that the most obvious is this. We are perfectly aware that there is almost certainly a ground for the disposition to break easily. There is some property of the glass and some sort of lawful connection between things having that property and being struck even lightly and those things subsequently breaking. Most of us don’t know what the property is. But when we are told that something broke because it was fragile, we have useful information. We know that the underlying explanation lies in the glass (as opposed to, for example, the pen, the fact that I don’t know my own strength, the fact that the glass was broken before and as a result had a structural weakness, and so on). Most philosophers would use the above analogy coupled with a functionalist account of pain to explain why we view both the pain explanation and the neural explanation as acceptable. As you might imagine after reading the last chapter, I’m inclined to reverse the analysis. I think we find the neural state explanation satisfactory, but do so without knowing what the intrinsic properties of the neural state are that might be causally relevant. The phenomenal property of being in pain is, perhaps, one of those intrinsic properties that, in conjunction with others, produces the grimace. Both explanations are perfectly acceptable. None of this will assuage the worry of the physicalist. The physicalist doesn’t expect phenomenal properties to have a place in the physical explanation of grimaces. They aren’t the right sorts of properties to fit into a physicalist world view, a view according to which matter and energy interact so as to produce changes in that matter and energy. But again, it is not clear to me that we are intruding in the least on the world with which the physicist deals. That world is probably best defined epistemologically. Physicalists don’t deal with introspectable 252

Objections and replies

properties – they deal with objects known through, and perhaps even defined in terms of, their causal roles. And I don’t see any reason why they need to reject or to modify any of the explanations they offer in light of the obvious fact that phenomenal properties are exemplified. We described the radical empiricism that allows room for the phenomenal in a physical world as ontologically liberating. We are free to mull over various epistemic possibilities any one of which might conceivably be the correct story of the way in which mental properties interact with physical properties. Nothing I said earlier is incompatible with the view that there are intrinsic, non-phenomenal physical properties exemplified by physical objects in general, and brain states in particular. Those physical properties might conceivably be co-exemplified with phenomenal properties. And if this is so there might be lawful connections between the exemplification of the intrinsic physical properties and the exemplification of the phenomenal properties. Alternatively, there might be non-phenomenal intrinsic properties, call them P, of brain states such that when a region of the brain a exemplifies P, a’s being P exemplifies the phenomenal property F. And, for all I know, it might be lawfully impossible for a to be P without F’s being exemplified by a’s being P. In general, I don’t find the concept of an emergent property the least bit odd or problematic. There are different senses in which a property F might be emergent. As we just saw, F might be emergent in the straightforward sense in which, when a complex of properties G are exemplified, F might be exemplified as a lawful consequence. There are all sorts of properties that are emergent in this sense. A given substance with F, G, and H might not be soluble, but when it becomes J as well it suddenly acquires the property of being soluble. A kind of stuff with properties A, B, and C might not be combustible, but, when it acquires the additional property D, it becomes combustible. 253

Knowledge, Thought, and the Case for Dualism

As we saw above, F might be an emergent property of a state of affairs. The state of affairs that is the exemplification of G and H might not have the property F, but the state of affairs that is the exemplification of G, H, and I might have the property F (where, again, there is, we might suppose, a lawful connection between a’s being F, H, and I and the state of affairs that is a’s being G, H, and I exemplifying the property F). Suppose that there is a lawful connection between something’s having physical properties G, H, and I and its also exemplifying the phenomenal property of being in pain. Indeed suppose, for the sake of argument, that the exemplification of G, H, and I is not only lawfully sufficient but lawfully necessary for the exemplification of pain. Further suppose that there is a lawful connection between the exemplification of G, H, and I and some behavior, say that grimace we were discussing earlier. It might be tempting to suppose that the phenomenal property of being in pain is a nomological dangler. After all, we already have all we need to explain the occurrence of the grimace in the exemplification of physical properties. But then, one might argue, the pain also gives us all we need to explain the grimace. But surely, the physicalist might argue, there is an asymmetry. After all, if G, H, and I were to occur in the absence of pain the grimace would still occur. It is not true, the physicalist might argue, that if the pain were to occur in the absence of the physical properties the grimace would occur. But how, precisely, are we to evaluate such counterfactuals? The following all seem equally true, given the hypothesis of lawful connection stated above. If G, H, and I were not to occur then neither pain nor the grimace would occur. If pain hadn’t occurred neither G, H, and I nor the grimace would have occurred. If G, H, and I were to occur in the absence of pain, the laws of nature would have been different. If the laws of nature were different then . . . Then what? Who knows what would have happened 254

Objections and replies

had fundamental lawful connections been other than they are.10 Once we postulate fundamental lawful connections connecting phenomenal properties to other properties, why does the exemplification of the physical properties lawfully connected to the exemplification of pain have any explanatory priority over the pain when it comes to explaining such things as behavior? Still, wouldn’t it be odd if nature conspired to create the sort of nomological connections postulated above? Well, I must confess that there is always the nagging suspicion that there is something decidedly odd about conscious experience. It seems obvious that we can imagine a world just like ours but without consciousness – a world in which there are bodies responding to physical stimuli with behavior but without an inner conscious life. A world in which human bodies were more like sophisticated mobile plants is perfectly conceivable. Why is there consciousness – what evolutionary purpose does it serve? While it is tempting to think that the question is interesting, indeed profound, I suspect that there might be no interesting or profound answer that one can give. That there are phenomenal properties is a proposition known better than any other contingent proposition we know. Why is the exemplification of some properties lawfully tied to the exemplification of other properties? That is a kind of question that must eventually have no answer. To be sure we can sometimes derive some laws from others (either alone or together with empirical facts). When we can, it is perfectly appropriate to say that we have explained the former in terms of the latter. 10

One of the best discussions of the truth conditions for counterfactuals is still Chisholm (1955). Chisholm points out that there is no determinate fact of the matter when it comes to what we take to be the truth conditions for a given counterfactual. The critical questions are what we are holding constant as we imagine a world in which the conditions described by the antecedent of the counterfactual obtain. And there simply are no hard and fast rules governing such matters.

255

Knowledge, Thought, and the Case for Dualism

To take an utterly trivial example, you can explain the fact that copper expands when heated by appealing to the more general law that metal expands when heated, together with the truth that copper is a metal. And the law that metal expands when heated might itself follow from some still more general law together with a claim about the microstructure of metal. But this can’t go on forever. Whatever one’s account of law is, one is going to face the fact that there are fundamental laws – laws that can’t be explained by more general laws. And those fundamental lawful regularities have no explanation. Even if there were a God who was capable of willing lawful regularities into existence, there would be a lawful generalization about the connection between God’s will and the existence of regularities and that lawful regularity would have no explanation. There is no point in lamenting this fact or in insisting that there simply must be an answer to every “Why?” question. There is no answer, for example, nor could there be any answer, to the question of why there is something rather than nothing. Any answer would need to appeal to the existence of something before there was something and such an answer would be self-contradictory. And there is no answer to the question of why the fundamental lawful regularities exist. Those fundamental lawful regularities appear to include lawful connections between physical properties and phenomenal properties. If they do, that’s just the way the world is.

256

8 The ubiquitous self: a brief postscript 8.1

s ummary

The focus of this book until now has been property dualism. As I indicated in Chapter 2, it seems as if the status of mental or phenomenal properties has dominated most recent discussion in the philosophy of mind. And there are good reasons for this. No one will even take seriously the idea that there is a mind or self distinct from physical stuff unless they become convinced that there is a kind of property that is different from the paradigmatic physical properties recognized by the physicalist. I have argued that there is no denying the fact that there are phenomenally given properties that resist classic reductions to physical properties. Moreover our phenomenological acquaintance with such properties gives us both propositional knowledge that such properties are exemplified and also the capacity to represent directly such properties in thought. I have conceded that there is an odd sense in which one might still find room for such properties within a physicalist world view. The reconciliation will not be attractive to most physicalists for it gives epistemic and conceptual priority to the phenomenal. But there is the epistemic possibility that the very phenomenal properties that the physicalist wants to reduce to paradigmatic physical properties are (1) the realizers of the dispositional properties that define for us physical objects and their properties, (2) properties that are co-exemplified by the same processes that exemplify intrinsic physical properties (whatever 257

Knowledge, Thought, and the Case for Dualism

they are), or (3) properties exemplified by the state of affairs that it is the exemplification of those intrinsic physical properties (whatever they are). I emphasized that if the causal theory of physical objects and their properties is correct, it is highly misleading to describe the phenomenal properties as physical properties, even if one of the three possibilities sketched above were true. And I also emphasized that I don’t think we have any positive reason to believe that any of those three possibilities is actual.

8.2

su b s t a n c e d u a l i s m i n the background?

I indicated in Chapter 2 that I am not all that comfortable with the category of substance. For one thing, historically, it has been most closely associated with physical objects. But the radical empiricism I find attractive makes me largely agnostic with respect to the intrinsic properties of physical objects, and without direct awareness of what they might be, it would be impossible to reach any phenomenologically based ontological conclusions about the bearers (if any) of such properties. But I can’t say the same thing about phenomenal properties. I have been claiming throughout this book that we find ourselves directly acquainted with phenomenal properties. Do we have any phenomenological basis for supposing that there is a self that is the bearer of those properties?

8.3

t h e s e l f a n d th e b o d y

Before proceeding, I should emphasize at the outset that if we were to have phenomenological evidence for the existence of a self, that self wouldn’t be, or even include, our physical body. At least that is true on the combination of epistemological views that lead one to property dualism. To be sure, we all 258

The ubiquitous self: a brief postscript

sometimes (perhaps usually) use the term “I” in such a way that our referent includes our physical body. I say that I am six feet tall, that I have bad knees, that I have brown eyes, and so on. But if there is a good argument from the possibility of hallucination against epistemological direct realism, then we have no non-inferential knowledge of the existence of anything physical, and that includes our body. Even if one of the famous skeptical scenarios turned out to be true (a scenario in which in fact I have no body that is six feet tall, I have no knees, I have no eyes, etc.), I would still have access to my self and its experiences. It is that self whose existence Descartes knew better than anything else that we are now looking for phenomenologically.1

8.4

h u m e a n re j e c t i o n o f t h e s e l f

In what may still be the received view among empiricists, David Hume purported to have phenomenological evidence that there is no such thing as the self. As we saw earlier in this book, Hume was officially committed to the view that every simple idea is a copy of a prior impression. Complex ideas can be built up out of simple ideas and complex ideas need not correspond to prior impressions. So for Hume to be able to make sense of the concept of self, he must locate an idea of the self. But when he searches for such an idea, all he finds are ideas of prior impressions: For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or

1

So yes it is closer to what Ryle derisively called the “ghost in the machine,” though the “in” is more than a bit misleading. And we are, of course, taking an approach that Strawson (1959) viewed as fatally wrong-headed. Strawson thought that we should take our discourse at face value and that we should assume that persons have both physical and psychological characteristics.

259

Knowledge, Thought, and the Case for Dualism cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe any thing but the perception. (1888, p. 252)

He goes on to say of people in general that: they are nothing but a bundle or collection of different perceptions, which succeed each other with an inconceivable rapidity, and we are in perpetual flux and movement. (p. 252)

While he finds it tempting to appeal to the metaphor that we are but a theatre in which “several perceptions successively make their appearance” (p. 253), Hume also emphatically warns us not to be misled by the metaphor. We shouldn’t think that the self stands to perceptions as the audience in the theatre does to performing actors. Perceptions are distinct existences that have no theatrical home and there is no audience to observe their phenomenal performance. While I have argued repeatedly throughout this book that there is, in the final analysis, no alternative but to rely on phenomenology, there is a considerable irony in the fact that philosophers employing the phenomenological method can reach diametrically opposed conclusions. Moore was quite confident, for example, that if we “looked” for an objective property that all intrinsically good things have in common, we would find it. Hume was equally convinced that you wouldn’t find anything remotely plausible as a candidate for value until you turned your attention inward and found in yourself certain sentiments arising from the contemplation of actions and characters. As Butchvarov (1982) points out in connection with Hume’s search for objective value, one must be certain that one is looking for the right sort of thing in the right sort of “place.” Hume is clearly looking for the self among the perceptions – he is looking for an idea of self that is a copy of a prior impression. That he won’t find. But if there is a self that is something other 260

The ubiquitous self: a brief postscript

than a bundle of perceptions, it would, presumably, be a radically different sort of entity. Although I am sympathetic to Hume’s claim that we will search in vain for a self distinct from various perceptions, I am genuinely torn. It also seems to me that there is a sense in which one is always aware of one’s self as one is aware of perceptual properties of that self. Hume (1888) himself seems to express strong reservations about his bundle theory in an appendix (p. 635). While he still complains that all he can find when he looks for a self are various perceptions, and while he still can’t imagine how he could be something other than his perceptions, he is genuinely and candidly worried about coming up with a plausible bundling principle. There must be a distinction between those perceptions that are mine and those that are yours. And if we are bundle theorists, we must identify the relevant “glue” that ties the perceptions that are mine together. Furthermore, and herein lies the rub, we must do so without presupposing an understanding of a self distinct from those perceptions. Hume’s concern here mirrors Kant’s (1929, First Division, Book I) extended argument for the transcendental ego (albeit without the bells and whistles that always seem to accompany Kant’s discussion of any topic). Let’s suppose for simplicity that there are two and only two people in the world – you and I. We exist at the same times and have various perceptions. I am in pain while I have a visual image of a red apple. You experience euphoria as you have the visual image of an orange. The bundle theorist says that the pain and the visual redness go together to make up me, while the euphoria and the visual orangeness go together to make up you. Why is that true? What do the redness and the pain have in common that makes them both my experiences? Of course, Hume was limited in the kind of answer he could give to the above question because he was committed to an ontology in which nothing exists but perceptions. So he would 261

Knowledge, Thought, and the Case for Dualism

have to find a critical connection among such perceptions if he were to answer our question. If one embraced an ontology of physical objects and causal connections between changes in physical objects and perceptions, one might argue that the experiences that are mine are mine because they are the immediate effects of my brain states. Those that are yours are yours because they are the immediate effects of your brain states. Of course, our critic won’t like the use of the possessive pronouns “my” and “your.” We are supposed to be giving a reductive account that tells us what all this “me” and “you” is about. But we can, of course, point to the different brains, or label them differently. There is the person whose experiences are those caused (immediately) by states of brain A and there is the person whose experiences are caused (immediately) by states of brain B. Will this approach work? Well it won’t for me. As has become obvious by now, I rest all contingent ontological commitment ultimately on phenomenological appeal. I am trying to figure out how I get a sense of self from having various experiences all of which I seem to know unproblematically are mine. I would surely know that the experiences I am having right now are mine even if I had no idea that brain states are the immediate causes of such experiences.2 I am certainly not directly aware of the immediate physical causes of experience. And yet there doesn’t seem to be any question for me that the pain, the visual, the auditory, and tactile experiences I am having right now are all mine. And as I indicated earlier, I am inclined to think that we can even make sense of a disembodied person whose experiences are still given to that person as his (although this possibility is, of course, much more controversial). What is the correct story that makes sense of this phenomenology? 2

Egyptians apparently thought that the (causal) seat of the mind was the heart. They nevertheless almost certainly had exactly the same concept of self that we have.

262

The ubiquitous self: a brief postscript

It is tempting to suppose that the experiences that are mine are mine precisely because I am, or can easily become, introspectively aware of their occurrence. Against this idea, one might suppose that it is at least intelligible to imagine a Corsican twin who is directly aware of his brother’s pain whenever it occurs, but as one entertains the possibility it seems equally plausible to describe the situation as one in which the pain is shared. And the sharing of the pain can be construed as both parties feeling (the same kind of) pain – exemplifying the same pain property.3 If one is trying to avoid a commitment to a self or an ego, it seems hopeless to rely on anything like compresence to consciousness as the bundling criterion for experiences. The experiences, on this view, are mine, just insofar as I am aware of them, or have the capacity to become aware of them. We need something that looks suspiciously like a self – the I – to serve as one of the relata in this relation of awareness. Of course, this is the dilemma that faces any philosopher who is writing in the philosophy of mind and who is hanging his ontological hat on the relation of phenomenological awareness. Relations require relata. I can be directly aware of the exemplification of this or that property. The reason for calling it a phenomenal property is just that fact – that it is phenomenologically presented to me. But we can’t say any of this without using the personal pronoun “I” (or one of its grammatical variations). It wouldn’t even help the phenomenological project to introduce a property as mysterious as a haecceity or individual essence. Your haecceity, on some views, is a property that is essential to you and that only you have. It is a property that makes you the person that you are, and it always 3

If one is a realist about properties, the brothers can literally exemplify one and the same property. If one is a trope theorist the sameness of the property will be analyzed somewhat differently.

263

Knowledge, Thought, and the Case for Dualism

accompanies the other experiences that you have. When it is no longer exemplified, you no longer exist. But even if there were such a property, it wouldn’t do you the slightest bit of good in bundling experiences that consist in the exemplification of various phenomenal properties. The haecceity would be just one more phenomenal property that needs to be bundled. There may be good philosophical reasons for introducing the concept of a haecceity. Perhaps it helps with problems of diachronic identity. It is notoriously difficult to give a satisfactory philosophical account of the necessary and sufficient conditions for the survival of a person through time – particularly if one is trying to capture, in part, the phenomenology of survival. But I don’t see how it helps with the unity of experience at a time. Even if I were aware of a haecceity accompanying my other experiences, it would be my awareness of the haecceity as mine and as co-present with my other experiences that is doing the bundling. And we still haven’t come close to eliminating the need to use that personal pronoun “my.”

8.5

re d u c e w h e r e y o u c a n : c o m m i t where you can’t

It seems to me that the slogan that titles this section is the right rule to follow when doing ontology. We do need to be constantly on guard against needlessly introducing kinds of things in taking stock of our ontological commitments. There may be lots of men in the world, but we don’t need to add to those men the average man. To be sure, we use a definite description in sentences like “The average man is 5 feet 10 inches tall.” And if one were slightly dense, one might initially suppose that the definite description denotes in the same way that other paradigmatic definite descriptions denoting men denote. But it doesn’t take a genius to figure out that the meaning of the 264

The ubiquitous self: a brief postscript

sentence in question can be analyzed in such a way that one successfully captures what is asserted and does so without any apparent reference to a man who satisfies the description “the average man” in the same way that some man might satisfy the description “the tallest man.” Similarly, Russell (and Quine) taught us how to avoid positing a Pegasus to go along with “Pegasus,” or a present king of France to go along with “the present king of France.” The success of a reduction is obviously not always something about which consensus is reached. So, for example, while no ontologist in his or her right mind is hanging on to the average man, at least some philosophers seem oddly attached to fictional characters. And the story there does need to become complicated. If we don’t want the world to contain Sherlock Holmes smoking his pipe and solving crimes, we need to analyze the sentences that occur in the fiction in such a way that we avoid problematic ontological commitment, but can also understand how people get engrossed in the stories. And we must also analyze the meanings of sentences made by people outside of the story when they are talking about the fictional character, many of which strike us as true. Speaking about Conan Doyle’s character, we say, and say truly, that Sherlock Holmes was a detective. That same sentence as it occurs in one of the stories is not obviously true (nor, according to some, is it obviously false). Again, there are well-known attempts to provide a philosophical analysis of the meanings of both uses of the sentence type. Some of them seem almost obviously right to me – that, for example, the statements made by us about the fictional characters are complex claims about what was said in the stories and what can reasonably be inferred from what was said in the stories. But it is an understatement to suggest that what seems obvious to me doesn’t always seem obvious (or even the slightest bit plausible) to many others. 265

Knowledge, Thought, and the Case for Dualism

Earlier in this book I suggested that the attempts to eliminate apparent commitment to occurrent mental states by analyzing statements that seem to make reference to them as complex statements about dispositions to behave are, at best, quixotic. I really do think that no one in his or her philosophically right mind can think that searing pain is nothing more than a disposition to grimace, groan, complain, etc. As I indicated elsewhere in this book, I don’t really think it is much more plausible to treat such states as functional states. Though searing pain no doubt has all sorts of evolutionarily advantageous functions, one cannot identify it with a functional state. One can imagine creatures in states that perform exactly the same functions as searing pain but who are fortunate enough never to feel pain. If there is such a thing as the self, it is decidedly and radically different from anything else of which one can think. My first instinct is Humean. I would try hard to get rid of apparent reference to the self by successfully translating statements in which such reference appears to occur into statements in which there is no apparent reference to the self. If I could effect the relevant translations, then I would gladly bid philosophical farewell to the self. If I can’t, though, I have no business taking anything but an agnostic attitude towards the problematic ontological commitment. As I indicated above, when I was young, I thought Hume’s phenomenological appeal was correct. But I’ve been in the business of philosophy a long time and I can’t come up with the translations that would justify my reaching the conclusion that there is no self. I can’t come up with reductions that seem to me even remotely plausible. It is absolutely certain that phenomenal properties exist. The self may not exist, but I’m in no position to make that claim. With remarks as tentative as those I just made, it probably isn’t useful here to engage in a further discussion of what that 266

The ubiquitous self: a brief postscript

phenomenally given self might be like. Some are sure that the self they find is simple, for example. But I’m not even sure what this means. If the self is what exemplifies phenomenal properties, it is certainly not simple in the sense that there is only one property it has. It probably won’t have any spatial parts, but it would be misleading, of course, to suggest that it is spatially simple. It would be best to think of attributing spatial properties to the self as a kind of category mistake. But again, this book was primarily focused on the nature of mental properties. This very brief discussion of the self is just what was advertised by the chapter heading – a postscript.

8.6

d i a c h r o n i c id e n t i t y

In Chapter 1, I very briefly introduced the idea that theists might be particularly interested in defending some version of dualism in order to make conceptual sense of our surviving the destruction of our bodies. There might seem to be some prima facie credibility to the idea that if we are to make sense of life after physical death, we had better not tie our mental life too closely to the existence of the brain and changes that occur in that brain. It is not clear, however, that any of the positions considered in this book have much bearing on the intelligibility of a self surviving the destruction of the body causally associated with it. There is even less of a connection between the issues discussed here and the empirical question of whether a self ever does survive the destruction of a body. To settle such issues we would need to answer very difficult questions concerning criteria for identity through time. Those questions are thorny enough when we are searching for persistence conditions of physical objects, and they become even harder when we search for the identity conditions of a person over time. It certainly is true that most forms of dualism will resist the idea that the conceptually necessary and sufficient 267

Knowledge, Thought, and the Case for Dualism

conditions for the survival of a person over time include the survival of that person’s body over time. Perry (1978) aside, even most physicalists countenance the possibility that one could house my brain in a different body and ensure my survival that way. Moreover, even with the brain itself we can deploy exotic thought experiments resembling the famous ship of Theseus problem to question whether the survival of a brain, even on a physicalist world view, is necessary for the survival of the person. On one version of the ship puzzle, Theseus’s ship is gradually modified. To keep his ship as seaworthy as possible, Theseus each couple of weeks replaces a plank, discarding the old planks in a shipyard. At the end of a few years he has a ship that he still regards as his ship but which has none of the original parts. Meanwhile, an enterprising young shipbuilder has taken all of the discarded planks and has built a ship that he dubs “Old Theseus.” Which of the two ships at the end of this process is identical with the ship Theseus owned at the beginning of the process? The ship Theseus continues to regard as his underwent a very gradual change – the kind of change that we typically think is compatible with the survival of that which changed. But the ship of the scavenger has exactly the same parts in exactly the same sort of relation to one another as does the original ship. Intuitions obviously move in both directions. One can imagine neurology developing to the point where we can replace parts of the brain with synthetic components that take over the functions of the organic parts they replace. And we can imagine the process taking place ever so gradually and in such a way that we eventually end up with a completely synthetic brain. Meanwhile the philosopher’s favorite mad scientist has kept the organic parts alive and has reconstructed a brain out of the very parts that had been discarded in the process of manufacturing the artificial brain. If the survival of 268

The ubiquitous self: a brief postscript

the brain is critical to the survival of the person, then which brain goes with which person? The answers to these questions are difficult, and they make difficult any questions concerning the survival of people over time – even if we assume the truth of physicalism. Matters are arguably just as bad, if not worse, on a version of property dualism, and even on a view that postulates an unanalyzable self that is the bearer of those properties. The survival of a self presumably has something to do with property exemplification. And there is no shortage of views as to which properties must be exemplified by a self at time t2 in order for that self to be identical with a self at t1. On this issue it doesn’t seem to me to make much of a difference if one embraces endurantism (the view that the self that exists at t1 is literally numerically identical with the self that exists at t2) or perdurantism (the idea that the self at t1 is not identical to the self at t2 but are two stages of a four-dimensional entity that is properly viewed as the person). Both sides of this controversy still need to figure out the relevant characteristics of the self at t1 and the self at t2 that allows that self to endure (on the endurantist’s view) or that make the two stages relevantly connected so as to be part of the relevant four-dimensional space/time “worm.” I’m not going to enter the fray over the above issues. I think memory must be brought into the picture somehow, and I think that if one is an endurantist, one had better give up on the transitivity of identity as a principle that governs selves not just at a time but through time. Fission cases will make hanging on to the principle wildly implausible. But it doesn’t seem to me all that problematic to suppose that my fission products are both identical with that earlier self called Fumerton even if they are not now identical with each other. (The perdurantist can hang on to transitivity of identity but only because distinct stages are not strictly identical either with each other or with the whole – the person – of which they are part.) 269

Knowledge, Thought, and the Case for Dualism

The above issues relate only to the conceivability of survival of the self in various hypothetical scenarios. Such questions are in the province of philosophy to answer. Empirical questions concerning causally necessary or sufficient conditions for the obtaining of the conceptually necessary and sufficient conditions for survival are outside the area of philosophical competence to answer (though of course there is nothing to stop a philosopher from gaining knowledge in some other field that will enable that philosopher to engage in the relevant empirical investigation). The philosophical literature on the nature of the self and its survival through time is enormous. This chapter is but the most superficial excursion into these topics. I do believe, however, that the questions about mental and physical properties that are the focus of this book must be answered in order to even begin a meaningful discussion about the nature of the self.

270

References Addis, Laird. 1967. “Particulars and Acquaintance,” Philosophy of Science 34(3): 251–259. 1981. “Dispositions, Explanations, and Behavior,” Inquiry 24(1): 205–227. 1989. Natural Signs: A Theory of Intentionality. Philadelphia: Temple University Press. Alston, William. 2005. Beyond Justification. Ithaca: Cornell University Press. Anscombe, Elizabeth. 1993. “Causality and Determination.” In Causation, ed. Ernest Sosa and Michael Tooley. Oxford University Press, 88–104. Armstrong, David M. 1963. “Is Introspective Knowledge Incorrigible?” Philosophical Review 72: 417–432. 1968. A Materialist Theory of Mind. New York: Routledge. 1978. A Theory of Universals, Volume II. Cambridge University Press. 1983. What is a Law of Nature? Cambridge University Press. 1989. Universals: An Opinionated Introduction. Boulder, CO: Westview Press. Ayer, A. J. 1952. Language, Truth and Logic. New York: Dover. Basile, Pierfrancesco. 2010. “It Must be True – But How Can it Be? Some Remarks on Panpsychism and Mental Composition.” In Basile, Kiverstein, and Phemister (2010), 93–112. Basile, Pierfrancesco, Julian Kiverstein, and Pauline Phemister, eds. 2010. The Metaphysics of Consciousness. Cambridge University Press. Bergmann, Gustav. 1964. Logic and Reality. Madison: University of Wisconsin Press. 1967. Realism: A Critique of Brentano and Meinong. Madison: University of Wisconsin Press. Bergmann, Michael. 2006. Justification without Awareness. Oxford University Press. Berkeley, George. 1713. Three Dialogues between Hylas and Philonous. In Philosophical Works: Including the Works on Vision, ed. M. R. Ayers. Totowa, NJ: Rowman & Littlefield, 1975, 129–207. (Page references follow the standard of referring to the original pagination, which most versions note on the side margin.)

271

References Bickle, John. 1992. “Multiple Realizability and Psychophysical Reduction,” Behavior and Philosophy 20(1): 47–58. Bigelow, John and Robert Pargetter. 1990. “Acquaintance with Qualia,” Theoria 61(3): 129–147. Black, Max. 1952. “The Identity of Indiscernibles,” Mind 51: 153–164. Brentano, Franz. 1874. Psychology from an Empirical Standpoint, trans. T. Rancurello, D. Terrell, and L. McAllister. (English translation published by New York: Routledge, 1973.) Broad, C. D. 1925. The Mind and its Place in Nature. London: Routledge & Kegan Paul. Butchvarov, Panayot. 1982. “That Simple, Indefinable, Nonnatural Property Good,” Review of Metaphysics 36: 51–75. Casati, R. and A. C. Varzi, eds. 1996. Events. Aldershot: Dartmouth. Chalmers, David. 1995. “Facing Up to the Problem of Consciousness,” Journal of Consciousness Studies 2(3): 200–219. 1996. The Conscious Mind. Oxford University Press. 2003. “The Matrix as Metaphysics.” Matrix website (http://consc.net/ papers/matrix.html). 2004. “Phenomenal Concepts and the Knowledge Argument.” In Ludlow, Nagasawa, and Stoljar (2004), 269–298. Chisholm, Roderick. 1942. “The Problem of the Speckled Hen,” Mind 51: 368–373. 1955. “Law Statements and Counterfactual Inference,” Analysis 15(5): 97–105. 1957. Perceiving. Ithaca, NY: Cornell University Press. Churchland, Paul M. 1989. “Knowing Qualia: A Reply to Jackson.” In Paul M. Churchland, A Neurocomputational Perspective: The Nature of Mind and the Structure of Science. Cambridge, MA: MIT Press, 67–76. Reprinted in Ludlow, Nagasawa, and Stoljar (2004), 163–178. Comesaña, Juan. 2002. “The Diagonal and the Demon,” Philosophical Studies 100: 249–266. Conee, Earl. 1994. “Phenomenal Knowledge,” Australasian Journal of Philosophy 72(2): 136–150. Reprinted in Ludlow, Nagasawa, and Stoljar (2004), 197–215. Conee, Earl and Richard Feldman. 2001. “Internalism Defended,” American Philosophical Quarterly 38(1): 1–18. Crane, Tim and D. H. Mellor. 1990. “There is No Question of Physicalism,” Mind 99: 185–206. Davidson, Donald. 1984. Inquiries into Truth and Interpretation. Oxford: Clarendon Press. 1993. “Causal Relations.” In Causation, ed. Ernest Sosa and Michael Tooley. Oxford University Press, 75–87. 2003. “Knowing One’s Own Mind.” In Nuccetelli (2003), 87–110. Dennett, Daniel. 1991. Consciousness Explained. New York: Little, Brown.

272

References Descartes, René. 1641. Meditations on First Philosophy. (Many editions and translations.) Donnellan, K. S. 1977. “The Contingent A Priori and Rigid Designators,” Midwest Studies in Philosophy 2: 12–27. Donner, Wendy and Richard Fumerton. 2009. The Philosophy of John Stuart Mill. Malden, MA: Blackwell. Dowell, J. L. 2006. “Physicalism: Empirical not Metaphysical,” Philosophical Studies 131 (October): 1–40 Dretske, Fred. 1981. Knowledge and the Flow of Information. Cambridge, MA: MIT Press. Dunne, J. W. 1958. An Experiment with Time. London: Faber and Faber. (Originally published 1927.) Fales, Evan. 1990. Causation and Universals. New York: Routledge. Farrell, B. A. 1950. “Experience,” Mind 59: 170–198. Feigl, Herbert. 1958. “The Mental and the Physical.” In Minnesota Studies in the Philosophy of Science II: Concepts, Theories, and the Mind–Body Problem, ed. Herbert Feigl, Michael Scriven, and Grover Maxwell. Minneapolis: University of Minnesota Press, 370–497. Feldman, Richard. 2004. “The Justification of Introspective Belief.” In Earl Conee and Richard Feldman, Evidentialism. Oxford University Press, 199–218. Feldman, Richard, and Ted Warfield, eds. 2010. Disagreement. Oxford University Press. Firth, Roderick. 1950. “Radical Empiricism and Perceptual Relativity,” Philosophical Review 59: 164–183, 319–331. Fodor, Jerry and Ernest Lepore. 1992. Holism. Oxford: Blackwell. Fumerton, Richard. 1983. “The Paradox of Analysis,” Philosophy and Phenomenological Research 43(4): 477–497. 1985. Metaphysical and Epistemological Problems of Perception. Lincoln, NE: University of Nebraska Press. 1989. “Russelling Causal Theories of Reference.” In Rereading Russell, ed. C. Wade Savagem. Minneapolis: University of Minnesota Press, 108–118. 1992. “Skepticism and Reasoning to the Best Explanation,” Philosophical Issues 2 (ed. Enrique Villaneuva): 149–169. 1995. Metaepistemology and Skepticism. Lanham, MD: Rowman & Littlefield. 2002. Realism and the Correspondence Theory of Truth. Lanham, MD: Rowman & Littlefield. 2003. “Introspection and Internalism.” In Nuccetelli (2003), 257–276. 2004a. “Epistemic Probability,” Philosophical Perspectives 19: 121–139. 2004b. “Achieving Epistemic Ascent.” In Ernest Sosa and his Critics, ed. John Greco. Oxford: Blackwell, 72–85. 2004c. “Inferential Internalism and the Presuppositions of Skeptical Argument.” In Current Issues in Theoretical Philosophy, vol. 2: The

273

References Externalist Challenge, ed. Richard Schantz. Berlin: Walter de Gruyter, 157–168. 2005. “The Challenge of Refuting Skepticism.” In Contemporary Debates in Epistemology, ed. Mathias Steup and Ernest Sosa. Oxford: Blackwell, 85–97. 2006a. Epistemology. Malden, MA: Blackwell. 2006b. “Direct Realism, Introspection, and Cognitive Science,” Philosophy and Phenomenological Research 73(3): 680–695. 2007. “Epistemic Conservatism: Theft or Honest Toil,” Oxford Studies in Epistemology 2: 64–87. 2010. “You Can’t Trust a Philosopher.” In Feldman and Warfield (2010), 91–110. Geach, Peter. 1957. Mental Acts: Their Content and Their Objects. London: Routledge & Kegan Paul. Gertler, Brie. 2011. Self-Knowledge. London and New York: Routledge. Gettier, Edmund. 1963. “Is Knowledge Justified True Belief?” Analysis 23: 121–123. Goldman, Alvin. 1979. “What is Justified Belief?” In Justification and Knowledge, ed. G. Pappas. Dordrecht: D. Reidel, 1–23. 1986. Epistemology and Cognition. Cambridge, MA: Harvard University Press. 1988. “Strong and Weak Justification.” In Philosophical Perspectives 2: Epistemology, ed. James Toberlin. Atascadero: Ridgeview Publishing, 51–69. 1999. “Internalism Exposed,” The Journal of Philosophy 96(6): 271–293. Harman, Gilbert. 1990. “The Intrinsic Quality of Experience,” Philosophical Perspectives 4: 31–52. Harré, Rom and Edward H. Madden. 1975. Causal Powers: A Theory of Natural Necessity. Lanham, MD: Rowman & Littlefield. Hempel, Carl. 1965. Aspects of Scientific Explanation and Other Essays in the Philosophy of Science. New York: Free Press. 1969. “Reduction: Ontological and Linguistic Facets.” In Philosophy, Science, and Method: Essays in Honor of Ernest Nagel, ed. Sidney Morgenbesser, Patrick Suppes, and Morton White. New York: St. Martin’s Press, 179–199. Henderson, David and Terence Horgan. 2006. “Transglobal Reliabilism,” Croatian Journal of Philosophy 6(2): 171–195 Hume, David. 1888. A Treatise of Human Nature, ed. L. A. Selby-Bigge. Oxford University Press, 1951. 1972 (1777). Enquiries Concerning Human Understanding and Concerning the Principles of Morals. London: Oxford University Press. Jackson, Frank. 1977. Perception. Cambridge University Press. 1982. “Epiphenomenal Qualia,” Philosophical Quarterly 32: 127–136. 1986. “What Mary Didn’t Know,” The Journal of Philosophy 83(5): 291–295.

274

References 1998. From Metaphysics to Ethics. Oxford University Press 2004. “Mind and Illusion.” In Ludlow, Nagasawa, and Stoljar (2004), 421–442. Johnston, Mark. 2004. “The Obscure Object of Hallucination,” Philosophical Studies 103: 113–183. Kant, Immanuel. 1929 (1781). Critique of Pure Reason, trans. Norman Kemp Smith. New York: St. Martin’s Press. Keynes, John Maynard. 1921. A Treatise on Probability. London: Macmillan Kim, Jaegwon. 1998. Mind in a Physical World. Cambridge, MA: MIT Press. Kripke, Saul. 1980. Naming and Necessity. Cambridge, MA: Harvard University Press. Leibniz. Gottfried. 1714. Monadology. Section 17, trans. Paul Schrecher and Anne Martin Schrecher. Indianapolis: Bobbs-Merrill, 1965. Lewis, C. I. 1946. An Analysis of Knowledge and Valuation. LaSalle, IL: Open Court. 1948. “Professor Chisholm and Empiricism,” The Journal of Philosophy 45(19): 517–24. Lewis, David. 2004. “What Experiences Teaches.” In Ludlow, Nagasawa, and Stoljar (2004), 77–103. Ludlow, Peter and Noah Martin, eds. 1998. Externalism and Self-Knowledge. Stanford, CA: CSLI Publications. Ludlow, Peter, Yujin Nagasawa, and Daniel Stoljar, eds. 2004. There’s Something about Mary: Essays on Phenomenal Consciousness and Frank Jackson’s Knowledge Argument. Cambridge, MA: MIT Press. Markie, Peter. 2009. “Classical Foundationalism and Speckled Hens,” Philosophy and Phenomenological Research 79(1): 190–206. Maxwell, Grover. 1962. “The Ontological Status of Theoretical Entities.” In, Minnesota Studies in the Philosophy of Science III: Scientific Explanation, Space, and Time, ed. Herbert Feigl and Grover Maxwell. Minneapolis: University of Minnesota Press, 3–15. 1978. “Rigid Designators and Mind-Brain Identity.” In Perception and Cognition: Issues in the Foundations of Psychology, ed. C. W. Savage. Minneapolis: University of Minnesota Press, 365–405. McHenry, Leemon. 2010. “Sprigge’s Ontology of Consciousness.” In Basile, Kiverstein, and Phemister (2010), 5–20. Meehl, Paul E. 1966. “The Complete Autocerebroscopist.” In Minnesota Studies in the Philosophy of Science II: Concepts, Theories, and the Mind–Body Problem, ed. Herbert Feigl, Michael Scriven, and Grover Maxwell. Minneapolis: University of Minnesota Press, 103–180. Melnyk, Andrew. 1997. “How to Keep the ‘Physical’ in Physicalism,” Journal of Philosophy 94(12): 622–637.

275

References 2003. A Physicalist Manifesto. Cambridge University Press. Mill, John Stuart. 1877. An Examination of Sir William Hamilton’s Philosophy, and of the Principal Philosophical Questions Discussed in His Writings. New York: Henry Holt and Company. 1906. System of Logic. New York: Longmans, Green, and Co. Montero, Barbara. 1999. “The Body Problem,” Noûs 33(2): 183–200. Moore, G. E. 1903. Principia Ethica. Cambridge University Press. 1922. “The Refutation of Idealism.” In G. E. Moore, Philosophical Studies. London: Kegan Paul, 1–20. Moreland, J. P. 2003. “The Knowledge Argument Revisited,” International Philosophical Quarterly 43(2): 219–228. Nagasawa, Yujin and Daniel Stoljar. 2004. “Introduction.” In Ludlow, Nagasawa, and Stoljar (2004), 1–36. Nagel, Ernest. 1953. Vision and Action. New Brunswick: Rutgers University Press. Nagel, Thomas. 1974. “What is it Like to be a Bat?” Philosophical Review 83(4): 435–450. Nemirow, Lawrence. 1980. “Review of Nagel’s Mortal Questions,” Philosophical Review 89: 475–476. 1990. “Physicalism and the Cognitive Role of Acquaintance.” In Mind and Cognition: A Reader, ed. William Lycan. Oxford: Blackwell, 490–499. 2007. “So This is What It’s Like: A Defense of the Ability Hypothesis.” In Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism, ed. Torin Alter and Sven Walter. Oxford University Press, 32–51. Nida-Rümelin, Martine. 2004. “What Mary Couldn’t Know: Belief about Phenomenal States.” In Ludlow, Nagasawa, and Stoljar (2004), 241–265. Nuccetelli, Susanna, ed. 2003. New Essays on Semantic Externalism and Self-Knowledge. Cambridge, MA: MIT Press. Pappas, George, and Marshall Swain, eds. 1978. Essays on Knowledge and Justification. Ithaca, NY: Cornell University Pres. Perry, John. 1978. A Dialogue on Personal Identity and Immortality. Indianapolis: Hackett Publishing Company. Plantinga, Alvin. 2000. Warranted Christian Belief. Oxford University Press. Price, H. H. 1950. Perception. London: Methuen. Putnam, Hilary. 1975. “The Meaning of ‘Meaning,’” Minnesota Studies in the Philosophy of Science 7: 131–193. 1981. Reason, Truth and History. Cambridge University Press. Quine, W. V. O. 1951. “Two Dogmas of Empiricism,” Philosophical Review 60(1): 20–43. 1966. The Ways of Paradox and other Essays. Cambridge, MA: Harvard University Press.

276

References 1969. Ontological Relativity and Other Essays. New York: Columbia University Press. Ramsey, F. P. 1929. “Theories.” In The Foundations of Mathematics and Other Logical Essays, ed. R. B. Braithwaite. Paterson, NJ: Littlefield and Adams, 212–236. Reed, Baron. 2006. “Shelter for the Cognitively Homeless,” Synthese 148(2): 303–308. Russell, Bertrand. 1905. “On Denoting,” Mind 14: 479–493. 1910. “Knowledge by Acquaintance and Knowledge by Description,” Proceedings of the Aristotelian Society 11: 108–128. 1927. The Analysis of Matter. London: Kegan Paul. 1948. Human Knowledge: Its Scope and Limits. New York: Simon & Schuster 1957. “Mr. Strawson on Referring,” Mind 66: 385–389. 1959. My Philosophical Development. London: George Allen & Unwin. 1967. The Problems of Philosophy. Oxford University Press. Ryle, Gilbert. 1949. The Concept of Mind. Chicago: University of Chicago Press. Scriven, Michael. 1962. “Explanations, Predictions, and Laws.” In Minnesota Studies in the Philosophy of Science III: Scientific Explanation, Space, and Time, ed. Herbert Feigl and Grover Maxwell. Minneapolis: University of Minnesota Press, 170–230. Searle, John. 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge University Press. 2004. Mind: A Brief Introduction. Oxford University Press. Sellars, Wilfrid. 1963. Science, Perception and Reality. New York: Routledge. Smart, J. J. C. 1959. “Sensations and Brain Processes,” Philosophical Review 68(2): 141–156. Sosa, Ernest. 2003a. “Privileged Access.” In Consciousness: New Philosophical Perspectives, ed. Q. Smith and A. Jokic. Oxford University Press, 273–292. 2003b. “Beyond Internal Foundations to External Virtues.” In Laurence Bonjour and Ernest Sosa, Epistemic Justification. Malden, MA: Blackwell, 99–170. 2009. Reflective Knowledge: Apt Belief and Reflective Knowledge, Volume II. Oxford University Press. Sprigge, Timothy. 1983. The Vindication of Absolute Idealism. Edinburgh University Press. Stalnaker, Robert. 2008. Our Knowledge of the Internal World. Oxford University Press. Stanley, Jason. 2011. “Knowing (How),” Noûs 45(2): 207–238. Stoljar, Daniel. 2001. “Two Conceptions of the Physical,” Philosophical and Phenomenological Research 62(2): 253–281. 2006. Ignorance and Imagination. New York: Oxford University Press.

277

References Strawson, Peter. 1950. “On Referring,” Mind 59: 320–334. 1959. Individuals: An Essay in Descriptive Metaphysics. London: Methuen. Tye, Michael. 2010. “The Puzzle of Transparency.” In The Norton Introduction to Philosophy, ed. A. Byrne, J. Cohen, G. Rosen, and S. Shiffrin. New York: W. W. Norton. Ushenko, Andrew Paul. 1937. The Philosophy of Relativity. London: George Allen & Unwin. 1946. Power and Events: An Essay on Dynamics in Philosophy. Princeton University Press. Williamson, Timothy. 2000. Knowledge and its Limits. Oxford University Press. 2009. “Probability and Danger,” The Amherst Lecture in Philosophy 4: 1–35.

278

Index acquaintance, 50, 55, 98–119, 127, 148, 153, 161–165, 171, 174, 186–187, 216, 231, 233, 239–241, 247, 257 See also foundationalism speckled hen problem, 107–111 Addis, Laird, 29, 58, 173 adverbial theory, 15, 52, 54, 108, 151 of intentional states, 55–56 of sensory states, 54–55, 101, 112 Alston, William, 130 Anscombe, Elizabeth, 250 Armstrong, David, 29, 34, 76, 135, 250 Ayer, A. J., 67, 113 Basile, Pierfrancesco, 225 behaviorism, 64–72, 76, 87, 117, 118, 119, 209, 215 Bergmann, Gustav, 29 Berkeley, George, 67, 68, 212, 213, 216, 219 Bickle, John, 75 Bigelow, John, 21 Black, Max, 37 Brentano, Franz, 53–54 Broad, C. D., 13 Butchvarov, Panayot, 260 causal overdetermination, 246–256 causal theory of perception, 4–5, 24–25, 114, 176, 212

causal theory of physical objects, 67, 175–177, 195, 197, 198, 212–217, 245–246, 258 causation, analysis of, 17, 30, 248–249 Chalmers, David, 3, 22, 24, 26, 152, 203, 205–206, 216 Chisholm, Roderick, 69, 115, 238, 255 Churchland, Paul, 21, 161 Conee, Earl, 21, 159–160, 161–163 consciousness, 21, 102, 134, 135, 140, 145, 147, 255, 263 See also acquaintance, dualism, intentional states, introspective access, panpsychism, phenomenalism, qualia, self hard v. easy problem, 21–27, 56 correspondence theory of truth, 40, 42, 45, 106–107, 169–173 See also fact Crane, Tim, 49 Davidson, Donald, 41, 242 de dicto ascription of belief, 10–11, 32–34 de dicto modality, 194–195, 197 de re ascription of belief, 9–11, 32–34 de re modality, 194–195, 197, 198 Dennett, Daniel, 25, 154

279

Index Descartes, René, 2–6, 11–12, 20, 24, 29, 31, 99, 112, 120, 136, 259 descriptivism, 7–11, 79, 83–85, 100, 162, 168, 178–180, 184, 186, 192–195, 198, 231, 264 direct acquaintance See acquaintance direct awareness See acquaintance direct realism disjunctivism, 144–146 epistemological, 141, 216, 222, 259 metaphysical, 50, 141 Donnellan, K. S., 178–181 Dowell, J. L., 48 Dretske, Fred, 170 dualism, 6–21, 62, 92, 102, 166, 246 See also phenomenalism, qualia event, 41–42 fact, 39–40 property, 36–39, 48, 58–60, 142, 154–157, 169, 220, 239 proposition, 42–44 sentence, 44 substance, 29–35, 258 Dunne, J. W., 13 epistemic possibility, 140, 232, 253, 257 evidentialism, 94, 126, 127, 132 fact, 3, 40, 80, 104, 106, 173, 174, 180, 207, 243 See also truthmaker and properties, 45, 162, 170 dualism, 136 physicalism, 40 vs. event, 41 vs. proposition, 9, 10–11, 32, 45, 128, 163–164, 167–169, 169–181, 182 Fales, Evan, 110, 175, 250 Farrell, B. A., 13

Feigl, Herbert, 13, 247 Feldman, Richard, 109, 240 Firth, Roderick, 69 Fodor, Jerry, 70 folk theory, 47, 72, 89, 134, 155 See also dualism, physicalism foundationalism, 92–93, 98–119, 127, 132, 216 See also internalism vs. externalism, justification, regress argument epistemic vs. conceptual regress arguments, 93–98, 184–186 foundational knowledge, 2, 239–245 Fumerton, Richard, 4, 88, 120, 127, 128, 131, 197, 216, 245 functionalism, 26, 76–90, 117, 118, 119, 137, 139, 202, 211, 215, 229 See also behaviorism Geach, Peter, 65 Gertler, Brie, 107 Gettier, Edmund, 200 Goldman, Alvin, 121, 122, 129, 131, 164 Harman, Gilbert, 27, 234, 237–239 Harré, Rom, 30 Hempel, Carl, 48–49, 79, 250 Henderson, David, 125–126, 242 Horgan, Terence, 125–126 Hume, David, 16–18, 30, 128, 146, 152, 190, 213–214, 219–220, 222, 245, 246, 247–249, 259–262 identity See also reduction, reference, reference fixing direct vs. indirect thought, 206–207 foundationalism of thought, 189–192

280

Index indiscernibility of identicals See Leibniz’s law intentional states, 21–27, 53, 89, 101, 113, 119, 234–239, 244 intentionality, 58, 234–239 See also conciousness, hard v. easy problem, intentional states, internalism vs. externalism, content controversies as mark of the mental, 52–58 internalism vs. externalism content controversies, 146, 153, 181, 241, 243, 245 epistemic controversies, 119–130, 163–165 introspective access, 49, 51, 119, 135, 172, 242, 263 See also acquaintance Jackson, Frank, 13–21, 27, 36, 56, 151, 155, 221, 233–239 Johnston, Mark, 116 justification, 2–4, 64, 94–119, 164, 186 See also acquaintance, foundationalism, internalism vs. externalism doxastic vs. propositional, 94 internal vs. external, 119–130 Keynes, John Meynard, 132 Kim, Jaegwon, 81 knowledge argument, 18–21, 57, 150–157 responses to ability hypothesis, 157–160 acquaintance hypothesis, 161–165 Jackson’s rejection of the knowledge argument, 233–239 new modes of representation, 169–181 Kripke, Saul, 43, 74, 85–86, 87, 184, 192–194, 196–197, 204–206, 207, 208

Leibniz, Gottfried, 6, 221 Leibniz’s law, 6, 9, 11, 165 Lepore, Ernest, 70 Lewis, C. I., 67, 69 Lewis, David, 20, 21, 153, 156, 157–159, 192 Ludlow, Peter, 13 luminosity, 50–51, 104–106 Madden, Edward, 30 Markie, Peter, 107, 109 Martin, Noah, 139 Mary thought experiment See knowledge argument Maxwell, Grover, 52, 67, 141, 152, 216 McHenry, Leemon, 225 Meehl, Paul E., 13 Mellor, D. H., 51 Melnyk, Andrew, 49 Mill, John Stuart, 69, 183–184, 214–215 mind–brain identity theory, 72–75, 221 See also physicalism token, 73 type, 72, 160 Montero, Barbara, 60 Moore, G. E., 15, 27, 61, 63, 234, 260 Nagasawa, Yujin, 13, 14 Nagel, Ernest, 79 Nagel, Thomas, 13, 18, 23, 155 naturalism See physicalism Nemirow, Laurence, 21, 157 Nida-Rümelin, Martine, 166, 202, 204 Nuccetelli, Susanna, 139 panpsychism, 152, 225–231 Pappas, George, 200 Pargetter, Robert, 21 phenomenalism, 67, 69, 213, 215

281

Index phenomenology, 59, 71, 88, 142, 174, 233, 239, 247, 257, 259–264 See also acquaintance, causal theory of objects physicalism, 46, 49, 60–64, 132–143 See also behaviorism, functionalism, mind-brain identity theory, panpsychism event, 41–42 fact, 39–40 property, 36–39 proposition, 42–44 sentence, 44 substance, 29–35 Plantinga, Alvin, 124, 126 property essential vs. accidental, 30–34 intrinsic vs. relational, 17, 19, 59, 70–71, 90, 135, 148, 150, 151–152, 163, 177, 192, 207, 210–218, 237, 252, 253, 257 mental vs. physical, 36–39 realist vs. trope theorist, 39 Putnam, Hilary, 121, 153, 168, 172, 184, 199 qualia, 20, 21 arguments for, 102–103 See also knowledge argument rejection of, 233–239 Quine, W. V. O., 8, 131, 265 radical empiricism, 171, 208, 210–218, 220 Ramsey, F. P., 215 reduction, 60, 67, 68, 72, 143, 209, 264, 265 See also behaviorism, causal theory of objects, functionalism, phenomenalism,

physicalism, self, Humean idea of informative identities, 88, 187–189, 205–206, 206–207, 209, 220, 241 Reed, Baron, 107 reference See also descriptivism, reference fixing direct, 141, 168, 183–189 indirect, 67, 83, 141, 185, 217, 228 opaque contexts, 8 transparent contexts, 9 reference fixing, 85–87, 196–197, 205–206 regress arguments for epistemic foundationalism, 93–98 for foundational thought and reference, 184–186, 239–241 Russell, Bertrand, 8, 13, 38, 52, 67, 79, 141, 162, 168, 179, 185–188, 190–191, 194, 197–199, 206, 216, 225, 265 Ryle, Gilbert, 71 Scriven, Michael, 79 Searle, John, 58, 226 self, 2, 224, 269–270 See also unity of consciousness Humean idea of, 259–264 problem with reductive analyses, 266–267 Sellars, Wilfrid, 115, 162 skepticism, 2–5, 95, 128, 130–132, 144–145, 259 with respect to the intrinsic nature of physical objects, 210–218, 221–224 Smart, J. J. C., 11–13, 20, 42, 104, 169, 173, 221, 246 Sosa, Ernest, 107, 122–124

282

Index Sprigge, Timothy, 225 Stalnaker, Robert, 92 Stanley, Jason, 158 Stoljar, Daniel, 13, 14, 152, 216, 218 Strawson, Peter, 179, 259 substance vs. bare particulars, 29–35 supervenience, 3, 60, 61–63, 126 Swain, Marshall, 200

truth-makers, 32, 40, 45, 99, 119, 141, 144, 164 See also fact unity of consciousness, 226, 264 Ushenko, Andrew Paul, 107 verificationism, 64–67, 70 Williamson, Timothy, 50, 51, 104–107, 113–115, 116

283