前言

Supplement to 'Philosophy' Royal Institute of Philosophy Supplement: 43 CURRENT ISSUES IN PHILOSOPHY OF MIND Edited by Anthony O'Hear Contributors Ned Block, Bill Brewer, John Campbell, Quassim Cassam, Andy Clark, Tim Crani Martin Davies, Naomi Eilan, Jane Heal, Ted Honderich, Jennifer Hornsby, Jaegwon Kim, Cynthia Macdonald, M. G. F. Martin, Gregory McCulloch, Michael Morris, Christopher Peacocke, Tony Stone, Crispin Wright

Current Issues in Philosophy of Mind ROYAL INSTITUTE OF PHILOSOPHY SUPPLEMENT: 43 EDITED BY Anthony O'Hear CAMBRIDGE UNIVERSITY PRESS

PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE The Pitt Building, Trumpington Street, Cambridge, CB2 1RP, United Kingdom CAMBRIDGE UNIVERSITY PRESS The Edinburgh Building, Cambridge CB2 2RU, United Kingdom 40 West 20th Street, New York, NY 10011-4211, USA 10 Stamford Road, Oakleigh, Melbourne 3166, Australia © The Royal Institute of Philosophy and the contributors 1998 Printed in the United Kingdom at the University Press, Cambridge Typeset by Michael Heath Ltd, Reigate, Surrey A catalogue record for this book is available from the British Library CIP data is available from the Library of Congress ISBN 0 521 63927 1 paperback ISSN 1358-2461

Contents Notes on Contributors v Introduction 1 ANTHONY O'HEAR The Mind-Body Problem After Fifty Years 3 JAEGWON KIM How to Find the Neural Correlate of Consciousness 23 NED BLOCK Embodiment and the Philosophy of Mind 35 ANDY CLARK Folk Psychology and Mental Simulation 53 MARTIN DAVIES AND TONY STONE Understanding Other Minds from the Inside 83 JANE HEAL Self-knowledge: the Wittgensteinian Legacy 101 CRISPIN WRIGHT Joint Attention and the First Person 123 JOHN CAMPBELL Consciousness as Existence 137 TED HONDERICH Setting Things before the Mind 157 M. G. F. MARTIN Perceptual Intentionality. Attention and Consciousness 181 NAOMI EILAN Experience and Reason in Perception 203 BILL BREWER 111

Contents Intentionality as the Mark of the Mental 229 TIM CRANE Intentionality and Interpretation 253 GREGORY McCULLOCH Externalism and Norms 273 CYNTHIA MACDONALD Mind, World and Value 303 MICHAEL MORRIS Mind, Knowledge and Reality: Themes from Kant 321 QUASSIM CASSAM The Modality of Freedom 349 CHRISTOPHER PEACOCKE Dualism in Action 377 JENNIFER HORNSBY Index 403 IV

Notes on Contributors Jaegwon Kim—William Faunce Professor of Philosophy at Brown University. He is the author of Supervenience and Mind (Cambridge: Cambridge University Press, 1993), Philosophy of Mind (Boulder, Colorado: Westview Press, 1996), and Mind in a Physical World (forthcoming). Ned Block—Professor of Philosophy and Psychology, came to NYU in 1996 from MIT where he was Chair of the Philosophy Program. He works in philosophy of mind, metaphysics and foundations of cognitive science and is currently writing a book on consciousness. He has been a Guggenheim Fellow, a Senior Fellow of the Center for the Study of Language and Information, a Fellow of the Centre de Recherche en Epistemologie Appliquee, a faculty member at two NEH Institutes and two NEH Seminars and the recipient of grants from the American Council of Learned Societies and the National Science Foundation. He is a past president of the Society for Philosophy and Psychology and a past Chair of the MIT Press Cognitive Science Board of Syndics. The Philosophers' Annual selected papers (as one of the 10 best for that year) in 1983, 1990, and 1996. Two volumes of his collected papers are forthcoming from MIT Press. Andy Clark—Professor of Philosophy and Director of the Philosophy/Neuroscience/Psychology Program at Washington University in St Louis, USA. His latest book is Being There: Putting Brain, Body and World Together Again (MIT Press, 1997). Martin Davies—Wilde Reader in Mental Philosophy and Professor of Philosophy in the University of Oxford, and a Fellow of Corpus Christi College. He is the author of Meaning, Quantification, Necessity: Themes in Philosophical Logic (1981) and of many papers in the philosophy of language, mind and psychology, and was the Executive Editor of the interdisciplinary journal Mind and Language from 1989 to 1994. He co-edited Consciousness: Psychological and Philosophical Essays (1993, with Glyn W. Humphreys) and Folk Psychology and Mental Simulation (1995, with Tony Stone). Tony Stone—Head of the Division of Psychology at South Bank University, London. He co-edited (with Martin Davies) Folk Psychology and Mental Simulation (1995), and has published papers on philosophy of mind and psychology, cognitive neuropsychiatry, and political philosophy. Jane Heal—Reader in Philosophy at the University of Cambridge. She is the author of Fact and Meaning: Quine and Wittgenstein on Philosophy of Language (Oxford: Blackwell, 1989) and of articles on philosophy of mind and philosophy of language.

Notes on Contributors Crispin Wright—Professor of Logic and Metaphysics and Wardlaw Professor at the University of St Andrews, and also the holder of a Leverhulme Personal Research Professorship. He was previously Nelson Professor of Philosophy at the University of Michigan and is a regular Visiting Professor at Columbia University. His principal publications are Wittgenstein on the Foundations of Mathematics (Duckworth, 1980), Frege's Conception of Numbers as Objects (Aberdeen University Press, 1983), Truth and Objectivity (Harvard, 1992), and Realism, Meaning and Truth (Blackwell second edition, 1993). He is a Fellow of the British Academy and of the Royal Society of Edinburgh. John Campbell—Fellow and Tutor in Philosophy at New College, Oxford. His publications include Past, Space and Self (MIT, 1994). He was British Academy Reader 1995-7. Ted Honderich—Grote Professor of the Philosophy of Mind and Logic at University College London, and the author of A Theory of Determinism: The Mind, Neuroscience, and Life-Hopes, of various papers in the Philosophy of Mind, and of How Free Are You? He recently edited The Oxford Companion to Philosophy. M. G. F. Martin—Lecturer in Philosophy at University College London. He has published papers on bodily awareness and sensation, self-knowledge and perception. He is currently working on a book on perception and naive realism. Naomi Eilan—Director of the British Academy Humanities Research Board project Consciousness and Self-Consciousness, an Inter-institutional project based at Warwick University. Her main research interests and publications concern the relation between theories of content and explanations of consciousness and self-consciousness. She is co-editor of Spatial Representation: Problems in Philosophy and Psychology (Blackwell, 1993) and The Body and the Self (MIT, 1995). Bill Brewer—Lecturer in Philosophy at Oxford University and a Fellow of St Catherine's College; he is also on the Steering Committee of the British Academy Project on Consciousness and Self-Consciousness at Warwick University. He has published articles on issues in epistemology and philosophy of mind, focusing recently upon the nature of empirical knowledge, and was an editor of Spatial Representation (Blackwell, 1993). His book, Perception and Reason, is forthcoming with Oxford University Press. Tim Crane—Reader in Philosophy at University College London. He is the author of The Mechanical Mind (1995) and the editor of The Contents of Experience (1992). Gregory McCulloch—Educated at the University of Leicester and Oriel College, Oxford. Held posts in Leicester and Nottingham before being

Notes on Contributors appointed to the Chair of Philosophy in Birmingham in 1995. Author of The Game of the Name (Oxford University Press, 1989), Using Sartre (Routledge, 1994), The Mind and its World (Routledge, 1995) and numerous articles on the philosophies of mind and language. Cynthia Macdonald—Reader in Philosophy at the University of Manchester. She is author of Mind-Body Identity Theories and Foundations of Contemporary Metaphysics (forthcoming), and co-editor of Philosophy of Psychology: Debates on Psychological Explanation, Connectionism: Debates on Psychological Explanation, Knowing Our Own Minds, and Reading in the Foundations of Contemporary Metaphysics. Michael Morris—Reader in Philosophy at the University of Sussex. He is the author of The Good and the True (Oxford University Press, 1992). Quassim Cassam—Lecturer in Philosophy at Oxford University, and Fellow and Tutor in Philosophy at Wadham College, Oxford. He is author of Self and World (1997), and the editor of the volume on Self Knowledge in the Oxford Readings in Philosophy series (1994). Christopher Peacocke—Waynflete Professor of Metaphysical Philosophy at the University of Oxford, Leverhulme Research Professor, Visiting Professor at New York University, and a Fellow of the British Academy. He is the author of Thoughts: An Essay on Content and A Study of Concepts. He is currently writing a book on the integration of metaphysics and epistemology. Jennifer Hornsby—Professor of Philosophy, Birbeck College, University of London. She is author of Actions (1980) and of Simple Mindedness: Essays in Defense of a Naive Realism (1997), and of articles in philosophy of language and on feminism. V l l

Introduction The philosophy of mind is one of the most exciting and innovative areas in philosophy at the current time. Necessarily, much of the work in the area is highly specialized, but as a consequence it is not widely available or accessible. By bringing together some of the leading figures in the field, we hope in this volume to fill what is often perceived both inside and outside philosophy to be a gap. Contributors have attempted in their papers to give an idea of their current concerns, to indicate the directions in which their work is taking them, and to suggest how it relates to other issues both in the philosophy of mind and in philosophy generally. After a general review of work on the mind-body problem over the last 50 years, the collection focuses on various aspects of neural activity and embodiment, on mental simulation, on the first person, on consciousness (including a new approach to the topic), on intentionality, on perception, on the mind as generating norms, on its connection to the world outside, on free will and on action. The papers in the volume are based on the lectures given in the Royal Institute of Philosophy's annual lecture series 1996-7. Thanks are due to all the contributors, and especially to Christopher Peacocke and Ted Honderich for their help in planning the series. I would also like to thank James Garvey for preparing the index, and for help with editing the volume. Anthony O'Hear

The Mind-Body Problem after Fifty Years JAEGWON KIM It was about half a century ago that the mind-body problem, which like much else in serious metaphysics had been moribund for several decades, was resurrected as a mainstream philosophical problem. The first impetus came from Gilbert Ryle's The Concept of Mind, published in 1948, and Wittgenstein's well-known, if not well-understood, reflections on the nature of mentality and mental language, especially in his Philosophical Investigations which appeared in 1953. The primary concerns of Ryle and Wittgenstein, however, focused on the logic of mental discourse rather than the metaphysical issue of how our mentality is related to our bodily nature. In fact, Ryle and Wittgenstein would have regarded, each for different reasons, the metaphysical problem of the mind-body relation as arising out of deplorable linguistic confusions and not amenable to intelligible discussion. There was C. D. Broad's earlier and much neglected classic, The Mind and Its Place in Nature, which appeared in 1925, but this work, although robustly metaphysical, failed to connect with, and shape, the mind—body debate in the second half of this century. It is fair to say that the mind-body problem as we know it today had its proximate origins in a trio of papers published in the late 1950s: U. T. Place's 'Is Consciousness a Brain Process?',1 in 1956, and J. J. C. Smart's This paper derives in part from my 'The Mind-Body Problem: Taking Stock After 40 Years', forthcoming in Philosophical Perspectives, 1997. 1 U. T. Place, 'Is Consciousness a Brain Process?', British Journal of Psychology 47/I (1956), 44-50. There were even earlier modern statements of the identity approach: e.g. Samuel Alexander, Space, Time, and Deity (London: Macmillan, 1920), vol. II, p. 9, where he says, 'The mental process and its neural process are one and the same existence, not two existences'; the psychologist Edwin G. Boring states, 'If we were to find a perfect correlation between sensation A and neural process a, a precise correlation which we had reason to believe never failed, we should then identify A and a ... it is scientifically more useful to consider that all psychological data are of the same kind and that consciousness is a physiological event' {The Physical Dimensions of Consciousness (New York: Dover reprint, 1963), p. 14). Boring's book was first published in 1933.

Jaegwon Kim 'Sensations and Brain Processes' and Herbert Feigl's 'The "Mental" and the "Physical"', published in 1958 and 1959 respectively.2 In these papers, Place, Smart and Feigl proposed an approach to the status of mind that has been variously called 'the mind-body identity theory', 'central-state materialism', 'type physicalism', and 'the brain-state theory'. In particular, it was the papers by Smart and Feigl that had a major philosophical impact, launching the debate that has continued to this day. For those of us who came of age philosophically in the 1960s, the brain-state theory was our first encounter with the mind-body problem as a problem in systematic philosophy. We were impressed by its refreshing boldness, and it seemed in tune with the optimistic scientific temper of the times. Why can't mentality turn out to be brain processes just as heat turned out to be molecular motion and light turned out to be electromagnetic waves? But the brain-state theory was surprisingly short-lived — its precipitous decline began only several years after its initial promulgation — and by the late sixties and early seventies it had been abandoned by almost all philosophers working in philosophy of mind and psychology. This was more than the fading away of a bold and promising philosophical theory: the demise of the brain-state theory gave a bad name to all forms of reductionism, turning the term 'reductionist' into a distinctly negative, often disdainful, epithet. In most academic and intellectual circles these days, calling someone a reductionist has become more than saying that he or she holds an incorrect view; it is a thinly disguised putdown that labels the targeted person as intellectually backward and simplistic. It is clear in retrospect, though, that in spite of its brief life, the Smart—Feigl physicalism made one crucial contribution that has outlived its reign as a theory of the mind. What I have in mind is the fact that the theory helped set the basic parameters for the debates that were to follow - a set of broadly physicalist assumptions and aspirations that still guide and constrain our thinking today. One indication of this is the fact that when the brain-state theory collapsed philosophers didn't lapse back into Cartesianism or other serious forms of dualism. Almost all the participants in the debate stayed within the physicalist framework, and even those who had a major hand in the demise of the Smart-Feigl materialism continued their allegiance to a physicalist worldview. And this fact 2 J. J. C. Smart, 'Sensations and Brain Processes', PhilosphicalReview 68 (1959), 141-56. Herbert Feigl, 'The "Mental" and the "Physical"', in Minnesota Studies in the Philosophy of Science, vol. II, eds. Herbert Feigl, Grover Maxewell and Michael Scriven (Minneapolis: University of Minnesota Press, 1958).

The Mind-Body Problem After Fifty Years has played a central role in defining our Problematik: through the seventies and eighties and down to the present, the mind—body problem - our mind-body problem - has been that of finding a place for the mind in a world that is fundamentally and essentially physical. If C. D. Broad were writing his 1925 book today, he might well have given it the title The Mind and its Place in the Physical World. What made the demise of the brain-state theory so quick and seemingly painless, causing few regrets among philosophers, was the fact that the principal objection that spelled its doom, the socalled multiple (or 'variable', as they say in Britain) realization argument, first advanced by Hilary Putnam,3 contained within it seeds for an attractive alternative approach, namely functionalism. The core thesis of functionalism, that mental kinds are 'functional kinds', not physical or biological kinds, was an appealing and eyeopening idea that seemed to help us make sense of 'cognitive science', which was being launched around then. The functionalist conception of the mind seemed tailor-made for the new science of mentality and cognition, for it appeared to posit a distinctive and autonomous domain of mental/cognitive properties that could be scientifically investigated independently of their physical/biological embodiments - an idea that promised both legitimacy and autonomy for psychology as a science. Functionalism made it possible for us to shed the restrictive constraints of physicalist reductionism without returning to the discredited dualisms of Descartes and others. Or so it seemed at the time. The functionalist conception of mentality still is 'the official story' about the nature and foundation of cognitive science.4 But functionalists, by and large, were not metaphysicians, and few of them were self-consciously concerned about just where functionism stood in regard to the mind-body problem. Some functionalists, like David Armstrong and David Lewis, thought that they were defending physicalism, whereas others, like Hilary Putnam and Jerry Fodor, claimed that functionalism delivered a decisive refutation of physicalism. The key term they used to describe the relation between mental properties (kinds, states, etc.) and physical properties was 'realization' (or sometimes 'implementation', 'execution', etc.): mental properties are 'realized' or 'implemented' by (or in) physical properties, though not identical with them or reducible to 3 In 'Psychological Predicates' first published in 1968 and later reprinted with a new title, 'The Nature of Mental States', in Hilary Putnam, Collected Papers II (Cambridge: Cambridge University Press, 1975). 4 See, e.g., Zenon Pylyshyn, Computation and Cognition (Cambridge, MA: MIT Press, 1985).

Jaegwon Kim them. But the term 'realization' was introduced5 and quickly gained currency, chiefly on the basis of computational analogies (in particular, mathematically characterized computing machines being realized in physical computers), and few functionalists, especially in the early days, made an effort to explain what the realization relation consisted in — what this relation implied in terms of the traditional options on the mind—body problem. I believe that the idea of 'supervenience' came to the fore in the seventies and eighties in part to fill this void. The doctrine that mental properties are supervenient on physical properties seemed nicely to meet the needs of the post-reductionist physicalist in search of a metaphysics of mind; for it promised to give a clear and sturdy sense to the primacy of the physical domain and its laws, thereby vindicating the physicalist commitments of most functionalists, while freeing them from the burdens of physical reductionism, thereby protecting the mental as an autonomous domain. Further, by allowing multiple physical bases for supervenient mental properties, it was able to accommodate the multiple realizability of mental properties as well. Many philosophers, especially those who for one reason or another had abandoned hopes for a physicalistic reduction of the mental, sought in mind-body supervenience a satisfying metaphysical statement of physicalism without reductionism. By the late seventies, what Ned Block has aptly called 'the antireductionist consensus',6 was firmly in place. This has helped to enthrone 'nonreductive physicalism' as the new orthodoxy not only on the mind-body relation but, more generally, on the relationship between 'higher-level' properties and underlying 'lower-level' properties in all other domains as well. Thus, the approach yielded as a bonus a principled general view about the relationship between the special sciences and basic physics. One side effect of the entrenchment of the antireductionist consensus has been the return of emergentism — if not the full-fledged doctrine of classic emergentism of the 1920s and 30s, at least its characteristic vocabulary and slogans. When positivism and the idea of 'unity of science' ruled, emergentism was often regarded with undisguised suspicion, as a mysterious and possibly incoherent metaphysical doctrine. With reductionist physicalism out of favour, 5 The first philosophical use of this term, roughly in its current sense, that I know of occurs in Hilary Putnam's 'Minds and Machines', in Dimensions of Mind, ed. Sydney Hook (New York: New York University Press, 1960). 6 In his 'Antireductionism Slaps Back', forthcoming in Philosophical Perspectives, 1997.

The Mind-Body Problem After Fifty Years emergentism appears to be making a strong comeback,7 and we now see an increasing and unapologetic use of terms like 'emergence', 'emergent characteristic', 'emergent phenomenon', 'emergent cause' and the like, roughly in the sense intended by the classic emergentists, not only in serious philosophical writings8 but in primary scientific literature in many fields.9 To sum up, then, three ideas have been prominently on the scene in recent discussions of the mind-body relation: the idea that the mental is 'realized' by the physical, the idea that the mental 'supervenes' on the physical, and the idea that the mental is 'emergent' from the physical. In this paper I want to explore the interplay of these three ideas, and the roles they play, in current debates over the mind-body problem, and, in the process, to indicate where I think we now stand with this problem. II Let us begin with supervenience. It is convenient to construe supervenience as a relation between two sets of properties, the supervenient properties and their 'base' properties. As is well known, a variety of supervenience relations is available, but for our present purposes fine-grained distinctions won't matter. The core idea of mind-body supervenience is that the mental properties or states of something are dependent on its physical, or bodily, properties, in the sense that once its physical properties are fixed, its mental properties are thereby fixed. This implies that if two things - organisms, persons or electromechanical systems - have identical physical properties, they must have identical mental natures as well; that is to say, exact physical twins are ipso facto exact mental twins. Mind-body supervenience can be equivalently formulated in the following useful way: if an organism instantiates a psychological property M (say, pain) at a time, it has at that time some physical 7 In addition to a number of recent journal titles, the signs of the return of emergentism include a recent collection of new essays on emergence, Emergence or Reduction? ed. A. Beckermann, H. Flohr and J. Kim (Berlin: de Gruyter, 1992), two volumes of essays on emergence being prepared in Europe as of this writing, and the 1997 Oberlin Philosophy Colloquium on the topic 'Reductionism and Emergence'. 8 See e.g., John R. Searle, The Rediscovery of the Mind (Cambridge, MA: MIT Press, 1992). 9 E.g., Francisco Varela, Evan Thompson and Eleanor Rosch, The Embodied Mind (Cambridge, MA: MIT Press, 1993). See especially Part IV entitled 'Varieties of Emergence'.

Jaegwon Kim property P on which M supervenes, in the sense that necessarily if anything has P, it has M. Thus, if you experience pain at a time, you must instantiate a certain physical property at the time (presumably, some neural property) on which pain supervenes. No mental property can be instantiated in an organism unless that organism instantiates some suitable physical property that serves as its physical base. In this way, mind-body supervenience promised to give sense to the physicalist idea that the physical enjoys ontological primacy over the mental, and the idea that physics is the most basic, and most comprehensive, of our sciences, all other sciences being 'special sciences' over restricted domains. Moreover, and this was of crucial importance to the nonreductive physicalist, supervenience prima facie did not seem to commit us to reductionism: after all, many moral theorists, like G. E. Moore and R. M. Hare, believed in the supervenience of the moral on the nonmoral, but rejected the reducibility of the former to the latter. And accepting the idea that aesthetic properties of works of art are supervenient on their physical characters doesn't seem to lead to the position that aesthetic properties are reducible to physical properties. In mind-body supervenience, then, we seemed finally to have found a metaphysical basis for nonreductive physicalism; supervenience seemed to be just the metaphysical relation of dependence that would enable us to understand how the mental, in spite of its dependence on the physical, could still remain irreducible to it, forming its own autonomous domain. Much of the discussion that followed the introduction of the supervenience idea into the mind-body debate was over the question whether supervenience was indeed free of reductionist implications. The question is still unsettled, but it has become clear that this was really a non-issue. The real issue, I believe, is whether or not the doctrine of mind-body supervenience itself can claim to be a distinctive position on the mind-body problem. The question, then, is this: do we have in mind-body supervenience an account of how our mentality is related to the physical nature of our being? That is, can we use supervenience itself to state a philosophical theory of the way minds are related to bodies? Brief reflection shows that the answer is no, that mind-body supervenience by itself cannot constitute a theory of the mind-body relation. There are two related reasons for this. First, mind-body supervenience is consistent with a host of classic positions on the mind-body problem; it is in fact a shared commitment of many mutually exclusionary mind-body theories. Take emergentism, for example: emergentism is a dualistic theory that stress-

The Mind-Body Problem After Fifty Years es the irreducibility of the emergents, including mental properties, to more basic physicochemical conditions, and yet it respects supervenience. On emergentism, the emergents necessarily emerge when, and only when, appropriate 'basal conditions' are present; when identical basal conditions are present, identical emergents must emerge. The functionalist view that the mental, when it is realized, must be physically realized, too, entails mind body supervenience: the same physical conditions, the same functional properties. What is more obvious, mind-body supervenience is a trivial consequence of type physicalism (for example, the brain-state theory), which reductively identifies mental properties with physical properties. Even epiphenomenalism is committed to supervenience; if two things differ in some mental respect, that must be because they differ in some physical respect - it must be because the physical cause of the mental respect involved is present in one and absent from the other. If mind—body supervenience is a commitment of each of these conflicting approaches to the mind—body problem, it cannot itself be a position on this issue alongside these classic alternatives.10 What this shows is that the mere fact (assuming it is a fact) of mind-body supervenience leaves open the question of what grounds or accounts for it — that is, why the supervenience relation obtains between the mental and the physical.11 To see the general issue involved here, consider normative supervenience, the widely accepted doctrine that normative or evaluative properties supervene on nonnormative, nonevaluative properties. Various metaethical positions are committed to normative supervenience but offer differing accounts of its source. According to ethical naturalism, the supervenience holds because normative properties are definable in terms of nonnormative, naturalistic properties; that is, normative properties turn out to be naturalistic properties. Ethical intuitionists, like G. E. Moore, would see normative supervenience as a primitive synthetic a priori fact not susceptible to further explanation; it is something we directly apprehend through our moral sense. R. M. Hare, a noncognitivist, would attempt to explain it as a form of a consistency condition essential to the regulative charac10 Mind—body supervenience is not excluded even by Cartesian substance dualism. See my 'Supervenience for Multiple Domains', reprinted in Supervenience and Mind (Cambridge: Cambridge University Press, 1993). 11 On the need for explaining supervenience relations see Terence Horgan, 'Supervenience and Cosmic Hermeneutics', Southern Journal of Philosophy 22 (1984), Supplement, 19-38, and Terence Horgan and Mark Timmons, 'Troubles on Moral Twin Earth: Moral Queerness Revisited', Synthese 92 (1992), 221-60.

Jaegwon Kim ter of the language of commending and prescribing. Still others may try to explain it as arising from the very idea of normative evaluation, maintaining that evaluative or normative properties must have descriptive criteria. It is clear, then, that the thesis of normative supervenience by itself does not serve to characterize a distinctive position in metaethics. Similarly, it is useful to think of the diverse mind-body theories as offering competing explanations of mind-body supervenience: the explanation offered by type physicalism is parallel to the naturalistic explanation of normative supervenience: mind-body supervenience holds because mentality is physically reducible and mental properties turn out in the end to be physical properties. Emergentism, like ethical intuitionism, takes mind-body supervenience as a brute fact not amenable to explanation, something that should be accepted, as Samuel Alexander urged, with 'natural piety'. In contrast, epiphenomenalism invokes the causal relation (the 'same cause, same effect' principle) to explain supervenience, and on functionalism, as we will see, mind—body supervenience is a consequence of the view that mental properties are functional properties with physical properties as their realizers. We must conclude, then, that mind-body supervenience itself is not an explanatory theory; it merely states a pattern of property covariation between the mental and the physical, and points to the existence of a dependency relation between the two. Yet it is wholly silent on the nature of the dependence relation that might explain why the mental supervenes on the physical. Supervenience is not a metaphysically deep, explanatory relation; it is merely a phenomenological relation about patterns of property covariation. Mind—body supervenience, therefore, states the mind—body problem — it is not a solution to it. Ill Cartesian substance dualism pictures the world as consisting of two independent spheres, the mental and the material, each with its own distinctive defining properties. There are causal interactions across the domains, but entities in each domain, being 'substances', are ontologically independent of those of the other, and it is metaphysically possible for one domain to exist in the total absence of the other. What has replaced this picture of a dichotomized world is the familiar multi-layered model that views the world as stratified into different 'levels', 'orders' or 'tiers', organized in a hierarchical structure. The bottom level is usually thought to consist of elemen10

The Mind-Body Problem After Fifty Years tary particles, or whatever our best physics is going to tell us are the basic bits of matter out of which all material things are composed.12 Higher up on the ladder, we find atoms, molecules, cells, larger organisms and so on. The ordering relation that generates the hierarchical structure is the mereological relation: entities belonging to a given level, except those at the very bottom, have an exhaustive decomposition, without remainder, into entities belonging to the lower levels. Entities at the bottom level have no physically significant proper parts. What then of the properties of these entities? It is part of this layered conception that at each level there are thought to be properties, activities and functions that make their first appearance at that level (we may call them the 'characteristic properties' of that level). Thus, among the characteristic properties of the molecular level are electrical conductivity, inflammability, density, viscosity and the like; activities and functions like metabolism and reproduction are among the characteristic properties of the cellular and higher biological levels; consciousness and other mental properties make their appearance at the level of higher organisms. For much of this century, a layered picture of the world like this has formed an omnipresent, if only implicit, background for debate over the mind-body problem, emergence, reductionism, the status of the special sciences and related issues, and has exerted a pervasive influence on the way we formulate philosophical problems and debate their solutions. Sometimes, the layered model is couched in terms of concepts and languages instead of entities in the world and their properties. Talk of levels of descriptions, levels of analyses, levels of concepts, levels of explanations and the like is rampant everywhere — it has thoroughly pervaded primary scientific literature as well as philosophical writings about science.13 Now we come to a critical question: how are the characteristic properties of a given level related to the properties at the adjacent levels - in particular, to those at the lower levels? How are biologi12 The layered model as such of course does not need to posit a bottom level; it is consistent with an infinitely descending series of levels. 13 In his work on vision David Marr famously distinguishes three levels of analysis: the computational, the algorithmic and the implementational. See his Vision (New York: Freeman Press, 1982). The emergentists, early in this century, appear to have been first to give an explicit formulation of the layered model; see, e.g., C. Lloyd Morgan, Emergent Evolution (London: Williams and Norgate, 1923). For a particularly clear and useful statement of the model, see Paul Oppenheim and Hilary Putnam, 'Unity of Science as a Working Hypothesis', Minnesota Studies in the Philosophy of Science, vol. II, ed. Feigl, Maxwell and Scriven. 11

Jaegwon Kim cal ('vital') properties related to physicochemical properties? How are consciousness and intentionality related to biological/physical properties? How are social phenomena, phenomena characteristic of social groups, related to phenomena involving individual members? As you will agree, these are some of the central questions in philosophy of science, metaphysics and philosophy of mind. Possible answers to these questions define the classic philosophical options on the issues involved. Some of the well-known major alternatives include reductionism, antireductionism, methodological individualism, functionalism, emergentism, neo-vitalism and the like. You may attempt to give a single uniform answer applicable to all pairs of adjacent levels, or you may take different positions regarding different levels. For example, you might argue that properties at every level (higher than the bottom level) are reducible, in some clear and substantial sense, to lower-level properties, or you might restrict the reductionist claim to certain selected levels (say, biological properties in relation to physicochemical properties) and defend an antireductionist stance concerning properties at other levels (say, mental properties). Moreover, it isn't necessary to give a uniform answer in regard to all characteristic properties at a given level; concerning mental properties, for example, it is possible to hold - and some have done just that — that phenomenal or sensory properties, or qualia, are irreducible, while holding that intentional properties, including propositional attitudes, are reducible (say, functionally or biologically). Let us now turn to the reductionist approach to the question of interlevel property relationships. As I said, reductionism, in particular mind-body reductionism, suffered massive defections during the 1970s and 1980s, with the result that there are hardly any reductionists left anywhere in sight in philosophy of mind.14 This, I think, is generally true in all areas of philosophy; there may still be reductionisms or reductionist programmes (and I believe there are), but I don't know anyone who advertises him/herself as a reductionist about anything. But what is reduction, to begin with? 14 Andrew Melnyk writes: 'Indeed, it seems to be a little-known law governing the behavior of contemporary philosophers that whenever they profess faith in any form of materialism or physicalism they must make it absolutely clear that they are, of course, in no way endorsing anything as unsophisticated, reactionary, and generally intolerable as reductionism', in 'Two Cheers for Reductionism: Or, the Dim Prospects for Non-Reductive Materialism', Philosophy of Science 62 (1995), 370-88. According to Melnyk there are only two reductionists left on the scene; he says, 'The law holds ceteris paribus; for example, it does not apply if your name is Jaegwon Kim or Patricia Churchland.' 12

The Mind-Body Problem After Fifty Years And what has made the reductionist a persona non grata in philosophy of mind? The concept of reduction that has served as a shared background in the discussion of physical reductionism was derived from a model of reduction elaborated by Ernest Nagel in the 1950s.15 Nagel was mainly interested in intertheoretic reduction, as a relation between two scientific theories, and his principal focus was on the logical relationship between the theory to be reduced and the theory serving as the reduction base. According to Nagel, reduction is fundamentally a proof procedure, consisting in the logical/mathematical derivation of the laws of the reduced theory from those of the base theory, taken in conjunction with 'bridge laws' connecting the predicates of the two theories. Nagel thought that these intertheoretic linkages were necessary to secure logical/derivational connections between the two theories, since the theories may be couched in entirely distinct descriptive vocabularies. Standardly, these bridge laws are taken to be biconditionals in form ('if and only if statements), providing each property in the domain of the theory to be reduced with a nomologically coextensive property in the reduction base. For mind-body reduction, then, the Nagel model requires that each mental property be provided with a nomologically coextensive physical property; that is, a law of the following form must hold for every mental property M: (BL) M<r*P where P is some physical property. This bridge-law requirement made mind-body reductionism - in fact, all reductionisms - an easy target. As noted earlier, the most influential antireductionist argument, one that had a decisive role in establishing the antireductionist consensus, was the multiple realization argument based on the observation that, on account of their multiple realizability, mental properties fail to have coextensions in the physical domain, and that this makes mind—body bridge laws unavailable for Nagelian reduction. This argument was then extended in defence of a general antireductionist position in regard to all special sciences.16 This has made bridge laws the focal point of debates on reduction and reductionism: for three decades the battles over reductionism have been fought on the question whether bicon15 See The Structure of Science (New York: Harcourt, Brace & World, 1961), chapter 11. The model had been developed in Nagel's earlier papers published during the 1950s. 16 See J. A. Fodor, 'Special Sciences (or The Disunity of Science as a Working Hypothesis)', Synthese 28 (1974), 97-115. 13

Jaegwon Kim ditional bridge laws are available to connect the mental with the physical domain. But this is the wrong battlefield on which to contest the issue of reduction. The Nagel model is the wrong model of reduction for discussions of mind—body reduction, and bridge laws are linkages of the wrong kind to induce reduction. My view is that bridge laws are neither necessary nor sufficient for reduction. I think it is easy to see that derivation via bridge laws is not sufficient for reduction. There are two reasons for this. The first has to do with the explanatory import of a reduction: a reduction must explain how, and why, the reduced phenomena ('higher-level' phenomena) arise from the processes at the level of the reduction base ('lower-level' phenomena), and this explanatory demand is not met when, as in Nagel reduction, the bridge laws are assumed as unexplained primitives of reductive derivations. A bridge law of the form (BL) only tells us that mental property M (say, pain) co-occurs, as a matter of nomic necessity, with a physical property P (say, C-fibre activation), and Nagel reduction simply does not address the question why this is so. Why is it that you experience a pain, rather than an itch or tickle, whenever your C-fibres are activated? Why don't you experience pain when your A-delta fibres are firing? Why does any conscious experience arise when these neural fibres are firing? When the emergentists claimed that the properties of consciousness are irreducible emergent properties, it was because they despaired of ever answering these explanatory questions. They accepted both a fundamental physicalist ontology and the supervenience of higher-level properties on the lower-level ones; and they were not concerned about the multiple realizability of the former in relation to the latter. The availability of biconditional correlation laws was the least of their concerns. The intelligibility of these laws was what agitated the emergentists. It is the phenomena of emergence, codified in our bridge laws, that they advised us to accept as brute facts — 'with natural piety'. As far as the emergentists were concerned, we were welcome to help ourselves to as much Nagel reduction of the mental as we pleased, but this would only be so much logical exercise - it would not advance by an inch our understanding of why, and how, mentality makes its appearance when certain propitious configurations of biological conditions occur. Attaining such an understanding is exactly the same task as explaining the likes of (BL), that is, mind-body bridge laws. A second reason that Nagelian derivation via bridge laws does not suffice for reduction is ontological: we expect our reductions to simplify - simplify our scheme of concepts or scheme of entities. But 14

The Mind-Body Problem After Fifty Years bridge laws are standardly taken as contingent and empirical, not analytic or a priori, and this means that the concepts M and P in (BL) remain distinct. Hence, Nagel derivation does nothing for conceptual simplification. Moreover, a bridge law expresses only nomic coextensivity of properties, not their identity, which means that M and P remain distinct properties. Hence, Nagel derivation via bridge laws doesn't do anything for ontological simplification either. It does give us some simplification of laws - the laws of the reduced theory have been absorbed into the reducer — but this, too, may be largely illusory, since we are forced to inflate our base theory by adding the bridge laws as new primitive laws. Introduction of these new laws can represent a significant expansion of both the ideology and the ontology of the base theory, since these laws bring with them concepts and properties alien to the original base theory. At this point, the standard move we see in the philosophical discussion of theory reduction is to consider how, and under what conditions, correlations of the form (BL) can be enhanced into identities of the form: (1)M=P This strategy is proper as long as we work within the Nagelian paradigm. I believe, however, that the Nagelian conception of reduction should be jettisoned, and that our thinking about reduction needs to be reoriented if reduction is to remain a philosophically significant factor in our reflections on the mind-body problem and related issues concerning interlevel relationships of properties. For the philosophical poverty of Nagel reduction is easy to see. As I have already pointed out, there is nothing in emergentism that rules out a Nagel reduction of psychology to physical theory. Not even substance dualism needs to preclude Nagelian mind-body reduction. Furthermore, some forms of dualism actually entail the Nagel reducibility of psychology to physical theory: for example, the double-aspect theory and the doctrine of pre-established harmony each would provide us with mind-body bridge laws in abundant numbers to enable a Nagelian mind-body reduction (and reductions in the opposite direction as well!). It is clear, then, that any doctrine of mind-body reductionism couched in the Nagelian conception of reduction cannot be a significant claim about the status of the mind. If so, the refutation of mind-body reductionism in that sense of reduction cannot be regarded as a significant philosophical accomplishment either. 15

Jaegwon Kim IV I will now sketch a model of reduction that I believe is more appropriate for both science and philosophy. If M and P in bridge law (BL) are both intrinsic properties, the correlation between M and P must be taken as a brute fact about the two distinct intrinsic properties, and no amount of philosophical legerdemain can turn it into an identity. The only way we can go beyond such brute correlations is to interpret, or reinterpret, the reduction target, M, as an extrinsic/relational property. Let us look at some examples. Consider temperature: to reduce temperature, we first must think of it relationally and characterize it in terms of its relation to other properties. Temperature is that property of an object, or system, which is such that its magnitude increases when the object is in contact with another with a higher magnitude of it; when it is sufficiently high, it can cause wood and coal to burn; when it is extremely high, it can cause iron to turn into a molten state; when it is sufficiently low, it causes water to freeze - well, you get the idea. What is being done is to understand temperature as a property characterized in terms of its causal/nomological relations to other properties; that is, it is given an extrinsic characterization as a 'causal role'. Consider another example: the reduction of the gene. To get started we must construe the concept of a gene in terms of its causal function: the gene is that mechanism in an organism that is causally responsible for the transmission of heritable characteristics. Reduction of temperature is achieved when we can identify the property that fills the causal specification: for gases the property turns out to be the mean kinetic energy of the molecules; for solids and plasmas, and in vacuums, it turns out to be different properties. The reduction of the gene is accomplished when we identify the mechanism that fills the causal role specified: it turns out to be the DNA molecule — at least, for earthly organisms. On this view of reduction,17 then, the reduction of a property M 17 The ideas involved here go back to David Lewis, 'An Argument for the Identity Theory', Journal of Philosophy 63 (1966), 17-25 and David Armstrong's argument for central-state materialism in his A Materialist Theory of Mind (New York: Humanities Press, 1968). See also Robert Van Gulick, 'Nonreductive Materialism and the Nature of Intertheoretical Constraint', in Emergence or Reduction? ed. Beckermann, Flohr and Kim, and Joseph Levine, 'On Leaving What It Is Like', in Consciousness, ed. Glyn W. Humphreys and Martin Davies (Oxford: Blackwell, 1993). Relevant also are David Chalmers' discussion of 'reductive explanation' in The Conscious Mind (New York: Oxford University Press, 1996), and Frank Jackson's views on the role of conceptual analysis in metaphysics, in, e.g., Armchair Metaphysics', in Philosophy in Mind, ed. J. O'Leary Hawthorne and M. Michael (Dordrecht: Kluwer, 1993). 16

The Mind-Body Problem After Fifty Years consists in two steps: (i) the conceptual step of construing M in terms of its causal/nomological relations to other properties; and (ii) the empirical-theoretical step of identifying M's 'realizers' - that is, properties, or mechanisms, in the reduction base domain that have the specified causal/nomological characteristics. We can expect the second step to involve a theory that explains just how these realizers get to have these causal/nomological properties (such a theory will almost certainly be involved in the process of identifying the realizers of targeted functional properties). Step (i) is in effect the process of 'functionalizing' the target property, that is, defining it as a causal role. More specifically, it is useful to think of functionalization in terms of second-order properties: to have M is to have the second-order property of having some property, Q, meeting specification C. This is second-order in the sense that it involves quantification over first-order properties (that is, the properties already given). When the specification C involves causal/nomic relations, we can call the second-order property a 'functional' property.18 At step (ii), multiple realizers have to be expected as a rule. We now must face the following question: does the phenomenon of multiple realizability of the target property present difficulties to this account of reduction as well? Answering this question is a somewhat complicated affair that to some extent depends on a decision as to what we want to call 'reduction'. However, a persuasive case can be made for a negative answer: there is no need to fear multiple realization. Suppose that M has two realizers, Q, and Q2. For something, x, to have M is for x to have Q, or Q2. (Notice that this does not introduce a disjunctive property, having Q, or Q2; the 'or' here is a sentence disjunction, not a predicate disjunction with disjunctive properties as semantic values.) That is to say, each Minstance is either a Q,-instance or Q2-instance, and there are no Minstances over and above these Q-instances. Suppose M is instantiated on a given occasion in virtue of its realizer <3, being instantiated on that occasion. This M-instance, then, is identical with this Qx instance, and they have exactly identical causal powers: no new causal powers can magically accrue to the M-instance that are not 18 The notion of a second-order property in the present sense is due to Hilary Putnam, 'On Properties', in Essays in Honor of Carl G. Hempel, ed. N. Rescher et al. (Dordrecht: Reidel, 1969). It's interesting to note that although the inventor of functionalism also introduced the concept of a second-order property, which is tailor-made for a perspicuous explanation of 'realization', no functionalist, to my knowledge, took advantage of it until Ned Block did so in his 'Can the Mind Change the World?', in Meaning and Method, ed. George Boolos (Cambridge: Cambridge University Press, 1990). 17

Jaegwon Kim had by the Q\ -instance. The realizers of M do all the work that M does, and M does not represent a net addition to the ontology or causal structure of the world. But some might object: 'But what about M itself} M is identical with neither Q\ nor Q2, and therefore must be counted as a property distinct from each and every property in the reduction domain.' This is the same ontological challenge we posed for Nagel reduction: where is the ontological payoff as far as M is concerned? So does M still need to hang around in our ontology? I think we can handle M in one of two ways. One simple way is to identify M as the disjunctive property, Q\ v Q2. If the Qs are diverse realizers of M, their diversity must mean something, and the only thing it could mean is causal/nomological diversity. If they are causally and nomologically identical or quite similar, there would be no reason for counting them as distinct realizers. It is generally accepted that kinds in science are primarily individuated on the basis of causal powers. So M as a disjunction of causally diverse properties, will be a causally heterogeneous kind, and it will have only limited usefulness as a scientific kind.19 The second way of handling M to view it only as a concept, not a property. By forming a second-order expression of the form 'having some property or other, Q such that C(Q)', we cannot literally bring a new entity into our ontology. All we are doing is to introduce a way of picking out certain first-order properties by specifying a condition they need to meet; we might say that a second-order expression of this form refers indifferently to members of a class of first-order properties, namely those that satisfy the specified condition. By mere linguistic operations like quantification we can neither expand nor contract our ontology; what we expand is our linguistic repertoire. This, I believe, is a sufficient answer to the ontological question. It shows how a functional reduction gives us a simplified ontology. But how does the functional model of reduction meet the explanatory demands on reduction? In what way is a functional reduction an explanatory reduction? I believe that this question has a satisfying answer. Why is M instantiated in systems of kind S whenever Qi is instantiated by these systems? Because having M just is having some property meeting causal specification C, and Qi is a property that realizes M — that is, meets the specification C - in systems of kind S. Why does this particular system instantiate M on this occasion? Because it is instantiating Qi, one of NTs realizers. Why does this M-instance cause an effect of kind E? Because it is in fact a Qiinstance, where Qi is a realizer of M, and Qi-instances have effects 19 I argue that such properties are not inductively projectible in 'Multiple Realization and the Metaphysics of Reduction', in Supervenience and Mind (Cambridge: Cambridge University Press, 1993). 18

The Mind-Body Problem After Fifty Years of kind E. Since the causal powers of M-instances are identified with those of their realizers, all questions about the causal relations involving iW-instances are answerable at the level of M's realizers. What more can we ask from an explanatory reduction of M? When functionalism was introduced as an alternative to classic type physicalism, that is, the brain-state theory, it was thought, and is still widely thought, that it was a form of antireductionism - in fact, the principal version of antireductionism about the mental. What I am advocating is the exact opposite: the functionalizability of mental properties is necessary and sufficient for reduction {sufficient pending successful scientific discovery of their realizers for domains of interest to us). This is not merely a redefinition of the term 'reduction'; I hope I have persuaded you that the functional model points us toward the right way to think about reduction. On this model of reduction, then, emergent properties are easily characterized: a property M is emergent relative to a given domain D of properties just in case M is not functionalizable in terms of properties in D. In assessing where we now are with the mind-body problem, therefore, we must know where we stand with the functionalist approach to the mental. It has been customary to distinguish between two broad categories of mental phenomena, the intentional and the phenomenal, without excluding those that have aspects of both (for example, emotions). Intentionality is particularly evident in propositional attitudes, like belief, desire and intention. There has been much scepticism about the viability of a functionalist account of intentionality; in particular, Hilary Putnam, the father of functionalism, has recently mounted sustained attacks on the causal/functionalist accounts of content and reference, and John Searle has also vigorously resisted the functionalization of intentionality.20 However, I remain unconvinced by these arguments; I don't see unsurmountable obstacles to a causal/functional account of intentionality. Let me just say here that it seems to me inconceivable that a possible world exists that is an exact physical duplicate of this world but lacking wholly in intentionality.21 Such a world must be identical with ours in all intentional-psychological aspects.22 20 See Hilary Putnam, Representation and Reality (Cambridge, MA: MIT Press, 1988); Searle, The Rediscovery of the Mind. 21 I believe others (perhaps Shoemaker and Block) have made a similar observation. 22 A position like this is explicitly defended by David Chalmers in The Conscious Mind. 19

Jaegwon Kim The trouble comes from qualia. For, by contrast with the case of intentional phenomena, we seem able to conceive a physical duplicate of this world in which qualia are distributed differently or entirely absent (a 'zombie world' as some call it). To get to the point without fuss, it seems to me that the felt, phenomenal qualities of experiences, or qualia, are intrinsic properties if anything is. To be sure, we commonly refer to them using extrinsic/causal descriptions; e.g., 'the colour of jade', 'the smell of ammonia', 'the taste of avocado', and so on. However, this is entirely consistent with the claim that what these descriptions pick out are intrinsic qualities, not something extrinsic or relational. (Arguably it is because they are intrinsic and subjective that we need to resort to relational descriptions for inter subjective reference.) Compare our practice of ascribing intrinsic physical properties to material objects by the use of relational descriptions; e.g., 'two kilograms', '32 degrees Fahrenheit', etc. To say that an object has a mass of 2 kilograms is to say that it will balance, on an equal arm balance, two objects each of which would balance the Prototype Kilogram (an object stored somewhere in France). That is the linguistic meaning, the 'concept' if you prefer, of '2 kilograms'; however, the property it designates, having a mass of two kilograms, is an intrinsic property of material bodies. If the qualitative properties of consciousness are intrinsic, they will resist functionalization and hence reduction. My doubts about the functionalist accounts of qualia are by and large based on the well-known, and not uncontested, arguments from qualia inversions and the familiar epistemic considerations. In any case, it seems to me that if emergentism is correct about anything, it is more likely to be correct about qualia than about anything else.23 This is what makes the stance you take on the problem of qualia a decisive choice point with respect to the mind—body problem. Let me close by noting how the question of reducibility relates to another central problem in the metaphysics of mind, namely the problem of mental causation. If mental property M is functionally reducible in our sense, there is an easy answer to the question how M can have causal powers in the physical domain. As we noted, the causal powers of any given M-instance are identical with the causal powers of the particular physical realizer of M on that occasion, there being no net addition of causal powers beyond those of physical properties. But if M is not functionally reducible it is difficult, in fact not possible, to see how M, or iW-instances, can exercise causal powers 23 This position on qualia and reductionism bears close similarity to the positions defended by a number of philosophers - in particular, Joseph Levine, Frank Jackson, David Chalmers and perhaps Ned Block. 20

The Mind-Body Problem After Fifty Years in the physical domain if we assume, as I believe we should, that the physical domain is causally closed. Thus, the price we may have to pay - I believe it is the price we must pay - for the irreducibility of qualia is their causal powers: if they are irreducible, they are threatened with causal impotence - at least, in the physical domain.24 In this way, the two central problems in the philosophy of mind, the problem of consciousness and the problem of mental causation, come together in the same arena. The only visible way of explaining consciousness physically - that is, finding a place for it in the physical world - is to functionalize it in the physical domain. If that could be done, we could also solve the problem of its causal powers. If, as seems likely, it cannot be done, consciousness is threatened with epiphenomenalism. It seems, then, that we can preserve consciousness, or any other aspect of our mentality, as something distinctive and autonomous only if we are willing to accept their causal impotence. In short, the two problems make each other insoluble.25 Fifty years of debate have shown, I believe, that the central core of the mind-body problem is constituted by two great and deep puzzles, consciousness and mental causation. And these two puzzles turn out to be intimately intertwined - the key to both is the question whether phenomenal properties of consciousness can be functionalized. I believe that is where we now stand with the mind-body problem, half a century after its reintroduction into philosophy by Ryle, Smart, Feigl and others. 24 I believe the irreducibility leads to causal impotence tout court, but a detailed argument must await another occasion. 25 This way of putting it was suggested by David Chalmers in conversation. 21

How to Find the Neural Correlate of Consciousness* NED BLOCK There are two concepts of consciousness that are easy to confuse with one another, access-consciousness and phenomenal consciousness. However, just as the concepts of water and H2O are different concepts of the same thing, so the two concepts of consciousness may come to the same thing in the brain. The focus of this paper is on the problems that arise when these two concepts of consciousness are conflated. I will argue that John Searle's reasoning about the function of consciousness goes wrong because he conflates the two senses. And Francis Crick and Christof Koch fall afoul of the ambiguity in arguing that visual area VI is not part of the neural correlate of consciousness. Crick and Koch's work raises issues that suggest that these two concepts of consciousness may have different (though overlapping) neural correlates — despite Crick and Koch's implicit rejection of this idea. I will start with two quotations from Searle. You will see what appears to be a contradiction, and I will later claim that the appearance of contradiction can be explained if one realizes that he is using two different concepts of consciousness. I'm not going to explain yet what the two concepts of consciousness are. That will come later, after I've presented Searle's contradiction and Crick and Koch's surprising argument. Searle's Contradiction Searle discusses my claim that there are two concepts of consciousness, arguing that I have confused modes of one kind with two kinds: *This is a substantially revised version of a paper in Trends in Neuroscience 19/2 (1996). I am grateful to audiences at the 1996 consciousness conference in Tucson, at the 1996 cognitive science conference at the University of Sienna, at the University of Oxford, Department of Experimental Psychology, at Union College Department of Philosophy and the Royal Institute of Philosophy. I am grateful to Susan Carey, Francis Crick, Martin Davies, Christof Koch, David Milner and to the editor of Trends in Neuroscience for comments on a previous draft. 23

Ned Block There are lots of different degrees of consciousness, but door knobs, bits of chalk, and shingles are not conscious at all ... These points, it seems to me, are misunderstood by Block. He refers to what he calls an 'access sense of consciousness'. On my account there is no such sense. I believe that he ... [confuses] what I would call peripheral consciousness or inattentiveness with total unconsciousness. It is true, for example, that when I am driving my car 'on automatic pilot' I am not paying much attention to the details of the road and the traffic. But it is simply not true that I am totally unconscious of these phenomena. If I were, there would be a car crash. We need therefore to make a distinction between the center of my attention, the focus of my consciousness on the one hand, and the periphery on the other.... [Italics added].1 Note that Searle claims that if I became unconscious of the road while driving, the car would crash. Now compare the next argument. the epileptic seizure rendered the patient totally unconscious, yet the patient continued to exhibit what would normally be called goaldirected behavior ... In all these cases, we have complex forms of apparently goal-directed behavior without any consciousness. Now why could all behavior not be like that? Notice that in the cases, the patients were performing types of actions that were habitual, routine and memorized ... normal, human, conscious behavior has a degree of flexibility and creativity that is absent from the Penfield cases of the unconscious driver and the unconscious pianist. Consciousness adds powers of discrimination and flexibility even to memorized routine activities ... one of the evolutionary advantages conferred on us by consciousness is the much greater flexibility, sensitivity, and creativity we derive from being conscious.2 Note that according to the first quotation, if I were to become unconscious (and therefore unconscious of the road and traffic), my car would crash. But in the second quotation, he accepts Penfield's description 'totally unconsciously' as applying to the case of the petit mal patient who drives home while having a seizure. Thus we have what looks like a contradiction. 1 John Searle, 'Who is Computing with the Brain?' Behavioral and Brain Sciences 13/4 (1990), 632-4. 2 John Searle, The Rediscovery of the Mind (Cambridge, MA; MIT Press, 1992). 24

How to Find the Neural Correlate of Consciousness Crick and Koch's Peculiar Argument I will now shift to Crick and Koch's recent article in Nature1 arguing that VI, the first major way station in the brain for processing visual signals, is not part of the neural correlate of consciousness (what they call the NCC). Crick and Koch say that VI is not part of the neural correlate of consciousness because VI does not directly project to the frontal cortex. (They extrapolate (tentatively) from the fact that no direct connections are known in macaques to no connections in humans.) Their reasoning makes use of the premise that part of the function of visual consciousness is to harness visual information in the service of the direct control of reasoning and decision-making that controls behaviour. On the hypothesis that the frontal areas are involved in these mental functions, they argue that a necessary condition of inclusion in the NCC is direct projection to frontal areas. Though something seems right about their argument, it has nonetheless puzzled many readers. The puzzle is this: why couldn't there be conscious activity in VI despite its lack of direct connection to frontal cortex? This is Pollen's4 worry: 'I see no a priori necessity for neurons in perceptual space to communicate directly with those in decision space.' The possibility of conscious activity in VI is especially salient in the light of Crick and Koch's suggestion that visual consciousness is reverberatory activity in pyramidal cells of the lower layers of the visual cortex involving connections to the thalamus.5 For one wonders how they have ruled out the possibility that such activity exists in VI despite the lack of direct connection between VI and frontal cortex. They do not address this possibility at all. The overall air of paradox is deepened by their claim that 'Our hypothesis is thus rather subtle; if it [no direct connection] turns out to be true it [VI is not part of the neural correlate of consciousness] will eventually come to be regarded as completely obvious' (p. 123). But the reader wonders why this is true at all, much less obviously true. When such accomplished researchers say such puzzling things, one has to wonder if one is understanding them properly. I will argue that once the two concepts of consciousness are separated out, the argument turns out to be trivial on one reading and not clearly compelling on the other reading. That's the critical part of my comment on Crick and Koch, but I have two positive points as well. I argue that nonetheless their conclusion about VI should 3 F. Crick and C. Koch, 'Are We Aware of Neural Activity in Primary Visual Cortex?' Nature 375 (11 May 1995), 121-3. 4 D. Pollen 'Cortical Areas in Visual Awareness', Nature 111 (28 September 1995), 293-4. 5 F. Crick, The Astonishing Hypothesis (New York: Scribners, 1994). 25

Ned Block be accepted, but for a different reason, one that they implicitly suggest and that deserves to be opened up to public scrutiny. Further, I argue that the considerations that they raise suggest that the two concepts of consciousness correspond to different neural correlates despite Crick and Koch's implicit rejection of this idea. The Two Concepts The two concepts of consciousness are phenomenal consciousness and access-consciousness6. Phenomenal consciousness is just experience; access-consciousness is a kind of direct control. More exactly, a representation is access-conscious if it is poised for direct control of reasoning, reporting and action. One way to see the distinction between the two concepts is to consider the possibility of one without the other. Here is an illustration of access without phenomenal consciousness. In Anton's Syndrome, blind patients do not realize that they are blind (though implicit knowledge of blindness can often be elicited). Hartmann et al.7 report a case of 'Reverse Anton's Syndrome' in which the patient does not realize that he is not really blind. The patient regards himself as blind, and he is at chance at telling whether a room is illuminated or dark. But he has a small preserved island of VI which allows him to read single words and recognize faces and facial expressions if they are presented to the upper right part of the visual field. When asked how he knows the word or the face, he says 'it clicks' and denies that he sees the stimuli. There is no obvious factor in his social situation that would favour lying or self-deception. In addition to the damage in VI, he has bilateral parietal damage, including damage to the left inferior parietal lobe. Milner and Goodale8 have proposed that phenomenal consciousness requires ventral stream activity plus attention, and that the requisite attention can be blocked by parietal lesions. So perhaps this is a case of visual access without visual phenomenal consciousness. (Note that Milner and Goodale's account is not in conflict with Crick and Koch's claim that VI is not part of the NCC if activity in VI is not the object of attentional processes.) So we see that access-consciousness without phenomenal con6 N. Block, 'On a Confusion about a Function of Consciousness', Behavioral and Brain Sciences 18/2 (1995), 227-47. 7 J. A, Hartmann et al., 'Denial of Visual Perception', Brain and Cognition 16 (1991), 29-40. 8 A. D. Milner and M. A. Goodale, The Visual Brain in Action (Oxford: Oxford University Press, 1995). 26

How to Find the Neural Correlate of Consciousness sciousness makes sense and may even exist in a limited form. What about the converse, phenomenal consciousness without access? For an illustration at the conceptual level, consider the familiar phenomenon in which one notices that the refrigerator has just gone off. Sometimes one has the feeling that one has been hearing the noise all along, but without noticing it until it went off. One of the many possible explanations of what happens in such a case illustrates phenomenal consciousness without access-consciousness: before the refrigerator went off, you had the experience (phenomenal consciousness) of the noise (let us suppose) but there was insufficient attention directed towards it to allow direct control of speech, reasoning or action. There might have been indirect control (the volume of your voice increased to compensate for the noise) but not direct control of the sort that happens when a representation is poised for free use as a premise in reasoning and can be freely reported. (It is this free use that characterizes access-consciousness.) On this hypothesis, there is a period in which one has phenomenal consciousness of the noise without access-consciousness of it. Of course, there are alternative hypotheses, including more subtle ones in which there are degrees of access and degrees of phenomenality. One might have a moderate degree of both phenomenal consciousness of and access to the noise at first, then filters might reset the threshold for access, putting the stimulus below the threshold for direct control, until the refrigerator goes off and one notices the change. The degree of phenomenal consciousness and access consciousness may always match. Although phenomenal consciousness and access-consciousness differ conceptually (as do the concepts of water and H2O) we don't know yet whether or not they really come to the same thing in the brain. Once one sees the distinction, one sees many pure uses of both concepts. For example, the Freudian unconscious is access-unconscious. A repressed memory of torture in a red room could in principle be a phenomenally vivid image; what makes it unconscious in the Freudian sense is that it comes out in dreams, slips, fleeing from red rooms and the like rather than directly controlling behaviour. Thus in principle an image can be unconscious in one sense (not poised for access), yet experienced and therefore conscious in another sense (phenomenally). Searle's contradiction Let's go back to Searle's contradiction. You will recall that he says that if he were to become unconscious of the details of the road and 27

Ned Block traffic, the car would crash. 'When I am driving my car "on automatic pilot" I am not paying much attention to the details of the road and the traffic. But it is simply not true that I am totally unconscious of these phenomena. If I were, there would be a car crash.' But he also says that Penfield's famous unconscious driver is 'totally unconscious', yet manages to drive home. Note that there is no room for resolving the contradiction via appeal to the difference between 'conscious' and 'conscious of. If Penfield's driver is 'totally unconscious', then he is not conscious of anything. And thus we have a conflict with the idea that if one were to become unconscious of the road and traffic, the car would crash. Can we resolve the contradiction by supposing that what Searle thinks is that normally if one were to become unconscious of the road the car would crash, but the Penfield case is an abnormal exception? Not likely, since Searle's explicit conclusion is that consciousness adds flexibility, creativity and sensitivity to action - suggesting that he thinks that consciousness is simply not necessary to routine activities like driving home. I think that appeal to the access/phenomenal distinction does serve to resolve the contradiction. The resolution is that Searle is presupposing that the Penfield petit mal seizure case loses phenomenal consciousness but still has sufficient access-consciousness to drive. But when he says that if he were unconscious of the road the car would crash, he is thinking of loss of both phenomenal and access-consciousness - and it is the loss of the latter that would make the car crash. I find that audiences I have talked to about this issue tend to divide roughly evenly. Some use 'conscious' to mean phenomenal consciousness - to the extent that they control their uses. Others use 'conscious' to mean either access-consciousness or some kind of self-consciousness. But Searle's error suggests that he - and I don't think he is alone - mixes the two concepts together. How Crick and Koch's Argument Depends on a Conflation Crick and Koch argue that VI is not part of the neural correlate of consciousness because VI does not project to frontal cortex. Visual consciousness is used in harnessing visual information for directly guiding reasoning and decision making and direct projection to frontal cortex is required for such a use. But what concept of consciousness are Crick and Koch deploying? They face a dilemma. If they mean phenomenal consciousness, then their argument is extremely interesting but unsound: their conclusion is unjustified. If 28

How to Find the Neural Correlate of Consciousness they mean access-consciousness, their argument is trivial. Let me explain. Let us look at their argument more closely. Here it is: 1. Neural machinery of visual consciousness harnesses visual information for direct control of reasoning and decision making. 2. Frontal areas subserve these functions. 3. VI does not project directly to frontal cortex. 4. So VI is not part of neural correlate of consciousness. Note that the 'direct' in premise 1 is necessary to generate the conclusion. But what reason is there to suppose that there cannot be some neural machinery of visual consciousness — VI, for example — that is part of the machinery of control of reasoning and decision making, but only indirectly so? If by 'consciousness' we mean phenomenal consciousness, there is no such reason, and so premise 1 is unjustified. But suppose we take 'consciousness' to mean access-consciousness. Then premise 1 is trivially true. Of course the neural machinery of access-consciousness harnesses visual information for direct control since access consciousness just is direct control. But the trivial interpretation of premise I trivializes the argument. For to say that if VI does not project directly to areas that control action, then VI is not part of the neural correlate of access-consciousness is to say something that is very like the claim that if something is a sleeping pill, then it is dormitive. Once Crick and Koch tell us that VI is not directly connected to centres of control, nothing is added by saying that VI is not part of the neural correlate of consciousness in the access sense. For an access-conscious representation just is one that is poised for the direct control of reasoning and decision making. On this reading, we can understand Crick and Koch's remark about their thesis that 'if it [VI is not directly connected to centres of control] turns out to be true it [VI is not part of the neural correlate of consciousness] will eventually come to be regarded as completely obvious'. On the access-consciousness interpretation, this remark is like saying that if it turns out to be true that barbiturates cause sleep, their dormitivity will eventually come to be regarded as completely obvious. To avoid misunderstanding, I must emphasize that I am not saying that it is a triviality that neurons in VI are not directly connected to frontal areas. That is an empirical claim, just as it is an empirical claim that barbiturates cause sleep. What is trivial is that if neurons in VI are not directly connected to frontal areas, then neurons in VI are not part of the neural correlate of access-consciousness. 29

Ned Block Similarly, it is trivial that if barbiturates cause sleep, then they are dormitive. That was the 'access-consciousness' interpretation. Now let us turn to the phenomenal interpretation. On this interpretation, their claim is very significant, but not obviously true. How do we know whether activity in VI is phenomenally conscious without being access-conscious? As mentioned earlier, Crick and Koch's own hypothesis that phenomenal consciousness is reverberatory activity in the lower cortical layers makes this a real possibility. They can hardly rule out this consequence of their own view by fiat. Crick and Koch9 say, 'We know of no case in which a person has lost the whole prefrontal and premotor cortex, on both sides (including Broca's area), and can still see.' But there are two concepts of seeing, just as there are two concepts of consciousness. If it is the phenomenal aspect of seeing that they are talking about, they are ignoring the real possibility that patients who have lost these frontal areas can see. Crick and Koch attempt to justify the 'directly' by appeal to representations on the retina. These representations control but not directly; and they are not conscious either. Apparently, the idea is that if representations don't control directly, then they are not conscious. But this example cuts no ice. Retinal representations have neither phenomenal nor access-consciousness. So they do not address the issue of whether VI representations might have phenomenal but not access-consciousness. So Crick and Koch face a dilemma: their argument is either not substantive or not compelling. Is the Point Verbal? Crick and Koch often seem to have phenomenal consciousness in mind. For example, they orient themselves towards the problem of 'a full accounting of the manner in which subjective experience arises from these cerebral processes ... Why do we experience anything at all? What leads to a particular conscious experience (such as the blueness of blue)? Why are some aspects of subjective experience impossible to convey to other people (in other words, why are they private)?'10 Crick and Koch often use 'aware' and 'conscious' as synonyms, as 9 F. Crick and C. Koch, untitled response to Pollen, Nature 377 (28 September 1995), 294-5. 111 F. Crick and C. Koch, 'Why Neuroscience May Be Able to Explain Consciousness', sidebar in Scientific American, December 1995, 92. 30

How to Find the Neural Correlate of Consciousness does Crick in The Astonishing Hypothesis. For example, the thesis of the paper in Nature11 is that VI is not part of the neural correlate of consciousness and also that VI is not part of the neural correlate of visual awareness. But sometimes they appear to use 'awareness' to mean access-consciousness. For example, 'All we need to postulate is that, unless a visual area has a direct projection to at least one of [the frontal areas], the activities in that particular visual area will not enter visual awareness directly, because the activity of frontal areas is needed to allow a person to report consciousness' (p. 122, emphases added). What could 'consciousness' mean here? 'Consciousness' can't mean access-consciousness, since reporting is a kind of accessing, and there is no issue of accessing access-consciousness. Consciousness in the sense in which they mean it here is something that might conceivably exist even if it cannot be reported or otherwise accessed. And consciousness in this sense might exist in VI. Thus when they implicitly acknowledge an access/phenomenal consciousness distinction, the possibility of phenomenal without access consciousness looms large. My point is not a verbal one. Whether we use 'consciousness' or 'phenomenal consciousness', 'awareness' or 'access-consciousness', the point is that there are two different concepts of the phenomenon or phenomena of interest. We have to acknowledge the possibility in principle that these two concepts pick out different phenomena. Two vs. one: that is not a verbal issue. Are the Neural Correlates of the Two Kinds of Consciousness Different? Perhaps there is evidence that the neural correlate of phenomenal consciousness is exactly the same as the neural correlate of accessconsciousness? The idea that this is a conceptual difference without a real difference would make sense both of much of what Crick and Koch say and of much of the empirical work on consciousness. But paradoxically, the idea that the neural correlates of the two concepts of consciousness coincide is one which Crick and Koch themselves actually give us reason to reject. Their hypothesis about the neural correlate of visual phenomenal consciousness is that it is localized in reverberatory circuits involving the thalamus and the lower layers of the visual cortex12. This is a daring and controversial hypothesis. But it entails a much less daring and controversial conclusion: that the localization of visual phenomenal consciousness does not involve 11 Crick and Koch, 'Are We Aware?' 12 Ibid. 31

Ned Block the frontal cortex. However, Crick and Koch think that the neural correlate of access-consciousness does involve the frontal cortex. Even if they are wrong about this, it would not be surprising if the brain areas involved in visual control of reasoning and reporting are not exactly the same as those involved in visual phenomenality. One way for Crick and Koch to respond would be to include the neural correlates of both access- and phenomenal consciousness in the 'NCC. To see what is wrong with this, consider an analogy. The first sustained empirical investigation of heat phenomena was conducted by the Florentine Experimenters in the seventeenth century. They didn't distinguish between temperature and heat, using a single word, roughly translatable as 'degree of heat', for both. This failure to make the distinction generated paradoxes. For example, when they measured degree of heat by the test 'Will it melt paraffin?' heat source A came out hotter than B, but when they measured degree of heat by how much ice a heat source could melt in a given time, B came out hotter than A.13 The concept of degree of heat was a mongrel concept, one that lumps together things that are very different.14 The suggestion that the neural correlate of visual consciousness includes both the frontal lobes and the circuits involving the thalamus and the lower layers of the visual cortex would be like an advocate of the Florentine Experimenters' concept of degree of heat saying that the molecular correlate of degree of heat includes both mean molecular kinetic energy (temperature) and total molecular kinetic energy (heat). The right way to react to the discovery that a concept is a mongrel is to distinguish distinct tracks of scientific investigation corresponding to the distinct concepts, not to lump them together. Another way for Crick and Koch to react would be to include both the frontal lobes and the circuits involving the thalamus and the lower layers of the visual cortex in the neural correlate of phenomenal consciousness. (Koch seems inclined in this direction in correspondence.) But this would be like saying that the molecular correlate of heat includes both mean and total molecular kinetic energy. The criteria that Crick and Koch apply in localizing visual phenomenal consciousness are very fine grained, allowing them to emphasize cortical layers 4, 5 and 6 in the visual areas. For example, they appeal to a difference in those layers between cats which are awake and cats which are in slow wave sleep, both exposed to the 13 M. Wiser and S. Carey, 'When Heat and Temperature Were One', in Mental Models, ed. D. Gentner and A. Stevens (Hillsdale, N.J.: Erlbaum, 1983). 14 See N. Block, 'On a Confusion'. 32

How to Find the Neural Correlate of Consciousness same visual stimuli. No doubt there are many differences between the sleeping and the waking cats in areas outside the visual cortex. But we would need a very good reason to include any of those other differences in the neural correlate of visual phenomenology as opposed, say, to the non-phenomenal cognitive processing of visual information. A Better Reason for not Including VI in the NCC Though I find fault with one strand of Crick and Koch's reasoning about VI, I think there is another strand in the paper that does justify the conclusion, but for a reason that it would be good to have out in the open and to distinguish from the reasoning just discussed. (Koch tells me that what I say in this paragraph is close to what they had in mind.) They note that it is thought that representations in VI do not exhibit the Land effect (colour constancy). But our experience, our phenomenal consciousness, does exhibit the Land effect, or so we would all judge. We should accept the methodological principle: at this early stage of inquiry, accept what people say about their own experience. Following this principle and assuming that the claim that cells in VI don't exhibit colour constancy is confirmed, then we should accept for the moment that representations in VI are not on the whole phenomenally conscious. This methodological principle is implicitly accepted throughout Crick's and Koch's work. An alternative route to the same conclusion would be the assumption that the neural correlate of phenomenal consciousness is 'part of the neural correlate of access-consciousness (and so there can be no phenomenal without access-consciousness). Phenomenal consciousness is automatically 'broadcasted' in the brain, but perhaps there are other mechanisms of broadcasting. (Blindsight would be a weak example.) So even if the 'reverse Anton's syndrome' case turns out to be access- without phenomenal consciousness, Crick and Koch's conclusion might still stand. This is a weaker argument than the one just given because of the possibility that colour nonconstant information is actually broadcast in the brain but 'swamped' by colour constant information from higher visual areas. Note that neither of the reasons given here make any use of the finding that VI is not directly connected to frontal areas. The assumption that phenomenal consciousness is part of access consciousness is very empirically risky. One empirical phenomenon that favours taking phenomenal without access-consciousness seriously is the fact that phenomenal consciousness has a finer grain 33

Ned Block than access-consciousness based on memory representations. For example, normal people can recognize no more than 80 distinct pitches, but it appears that the number of distinct pitch-experiences is much greater. This is indicated (but not proven) by the fact that normal people can discriminate 1400 different frequencies from one another.15 There are many more phenomenal experiences than there are concepts of them. Despite these disagreements, I greatly admire Crick's and Koch's work on consciousness and have written a very positive review of Crick's book.16 Crick has written 'No longer need one spend time .... [enduring] the tedium of philosophers perpetually disagreeing with each other. Consciousness is now largely a scientific problem.'17 I think this conceptual issue shows that even if largely a scientific issue, it is not entirely one. There is still some value in a collaboration between philosophers and scientists on this topic. 15 D. Raffman, 'On the Persistence of Phenomenology', in Conscious Experience, ed. T. Metzinger (Place: Schningh, 1995). "' N. Block, Review of Francis Crick, The Astonishing Hypothesis, Contemporary Psychology, May 1996, pp. 427—9. " F. Crick, 'Visual Perception: Rivalry and Consciousness', Nature 379 (2 August 1996), 485-6. 34

Embodiment and the Philosophy of Mind ANDY CLARK I Introduction: The Rediscovery of the Body and of the World Cognitive science is in some sense the science of the mind. But an increasingly influential theme, in recent years, has been the role of the physical body, and of the local environment, in promoting adaptive success. No right-minded cognitive scientist, to be sure, ever claimed that body and world were completely irrelevant to the understanding of mind. But there was, nonetheless, an unmistakeable tendency to marginalize such factors: to dwell on inner complexity whilst simplifying or ignoring the complex inner-outer interplays that characterize the bulk of basic biological problem-solving.1 This tendency was expressed in, for example, the development of planning algorithms that treated real-world action as merely a way of implementing solutions arrived at by pure cognition (more recent work, by contrast, allows such actions to play important computational and problem-solving roles2). It also surfaced in David MarrV depiction of the task of vision as the construction of a detailed threedimensional image of the visual scene. For possession of such a rich inner model effectively allows the system to 'throw away' the world 1 Notable exceptions to this trend include work such as J. J. Gibson, The Ecological Approach to Visual Perception (Boston: Houghton-Mifflin, 1979) and, in a more philosophical key, Maurice Merleau-Ponty's, La Structure du Comportment (Paris: Presses Universitaires de France, 1942). Recent work in Animate Vision and ecological optics (see Section II below) is clearly influenced by Gibsonian ideas, while treatments such as F. Varela, E. Thompson and E. Rosch, The Embodied Mind (Cambridge, MA: MIT Press, 1991) explicitly acknowledge Merleau-Ponty. There is a brief discussion of these historical roots in chapter 8 of my own Being There: Putting Brain, Body and World Together Again (Cambridge, MA: MIT Press, 1997). 2 See, e.g., P. Agre and S. Rosenschein (eds.) Computational Theories of Interaction and Agency (Cambridge, MA: MIT Press, 1996); D. Kirsh and P. Maglio 'On Distinguishing Epistemic from Pragmatic Action', Cognitive Science 18 (1995), 513-49; and E. Hutchins, Cognition in the Wild (Cambridge, MA: MIT Press, 1995). 3 See D. Marr Vision (San Francisco, CA: W. H. Freeman, 1982). 35

Andy Clark and to focus subsequent computational activity on the inner model alone.4 More generally, the whole vision of cognition as inner operations on internal world models reflects an explanatory strategy which might reasonably be dubbed isolationism:5 (Isolationism) The world is (just) a source of inputs and an arena for outputs, and the body is (Just) an organ for receiving the inputs and effecting the outputs (actions). The task of early processing is to render the inputs as an inner world-model of sufficient richness to allow the bulk of problem-solving activity to be defined over the inner model alone. Isolationism, it is fair to say, is in increasing disrepute. But the precise shape of an alternative approach remains unclear. Anti-isolationist assertions range from the relatively innocent insistence that we won't achieve a balanced vision of what the brain does until we pay more heed to the complex roles of body and world, to the selfconsciously revolutionary accusation that mind itself is not, after all, a special inner arena populated by internal models and representations but is rather the operation of a profoundly interwoven system, incorporating aspects of brain, body and world — a system which resists informative analysis in terms of the old notions of inner models, representations and computation.6 The most radical anti-isolationist vision thus depicts human beings as a species of (so-called) 4 This tradition is nicely critiqued in P. S. Churchland, V. Ramachandran and T. Sejnowski, 'A Critique of Pure Vision' in C. Koch and J. Davies (eds.), Large-Scale Neuronal Theories of the Brain (Cambridge, MA: MIT Press, 1994). 5 Roboticists refer (usually disparagingly) to this isolationist vision as the idea of a linear Sense-Think-Act Cycle, See, e.,g., C. Malcolm, T. Smithers and J. Hallam, An Emerging Paradigm in Robot Architecture', Edinburgh University Department of Artificial Intelligence Technical Report, 1989. 6 Major statements of this view include J. Haugeland, 'Mind Embodied and Embedded' in Y.-H. Houng and J.-C. Ho (eds.), Mind and Cognition (Tapei, Taiwan: Academia Sinica, 1995), pp. 3-38, and T. Van Gelder, 'What Might Cognition Be, If Not Computation?' Journal of Philosophy 92/7 (1995), 345—81. Closely related claims and arguments appear in T. Van Gelder and R. Port. 'It's About Time' Introduction to R. Port and T. Van Gelder (eds.), Mind as Motion: Dynamics, Behavior, and Cognition (Cambridge, MA: MIT Press, 1995). E. Thelen and L. Smith, A Dynamic Systems Approach to the Development of Cognition and Action. (Cambridge, MA: MIT Press, 1994), and Varela, Thompson and Rosch, The Embodied Mind. 36

Embodiment and the Philosophy of Mind post-Cartesian agent.1 The post-Cartesian agent is a locus of knowledge, acts for reasons and has beliefs and desires. Yet she harbours no internal representations and resists analysis in terms of any cognitely important distinctions between inner and outer processes, between perception, cognition and action, or between mind, body and world. I shall argue that the post-Cartesian vision is unconvincing and that a key move in the argument (a move I dub the 'cognitive-tocoping shift') is both dialectically suspect and empirically unsound. More positively, I shall argue for a much weaker but still anti-isolationist stance: one that nevertheless suggests the need for some deep revisions in our understanding of the nature of internal representations and inner world models. The foundational and conceptual challenges thus prove real enough, even when stripped of their radical post-Cartesian trimmings. II Inner Symbol Flight The outright rejection of the notion of internal representation is just the extreme limiting case of a marked tendency that might be dubbed 'inner symbol flight'. This flight involves the progressive rejection of more and more of the apparatus and assumptions associated with the vision of cognition as the manipulation of chunky inner symbols. According to one simple (and historically important) vision, semantically sensible transitions between mental states are best explained in terms of syntactically constrained transitions between inner symbol strings. These symbol strings contained discrete elements corresponding rather closely to the semantic elements identified in sentential descriptions of the relevant mental states. Thus, the thought that John loves Mary is imagined to be realized as a complex inner symbol string that incorporates distinct and independently manipulable elements standing for 'John' 'loves' and 'Mary'.8 This vision of simple inner symbolic atoms (unstructured base items) corresponding rather closely to the familiar concepts and 7 This vision is clearly contemplated in Haugeland 'Mind Embodied' and in Van Gelder, 'What Might Cognition Be?' Both authors, however, recognize the large space of intermediate possibilities. The term 'postCartesian agent' is from Van Gelder, p. 381. See also Thelen and Smith , A Dynamic Systems Approach, p. 338, Van Gelder and Port, It's About Time', p. ix. 8 See J. Fodor and Z. Pylyshyn, 'Connectionism and Cognitive Architecture: A critical analysis', Cognition 28 (1988), p. 13. 37

Andy Clark relations enshrined in daily discourse was challenged by the development of distributed connectionist9 models. The sentential paradigm10 was replaced, in this research, by a vision of internal representations as distributed patterns of activity across a whole array of simple processing units. Such distributed patterns were allowed to overlap in semantically significant ways, giving rise to a variety of computationally significant side-effects including free generalization, damage-resistance, etc." More recently still, we have witnessed increased attention to the temporal dynamics of the inner representational vehicles. The use of (for example) simple recurrent neural networks12 allows information to be encoded not just in instantaneous patterns of activity but in temporally extended processing trajectories. In these networks, much of the information-processing power resides in the way a current state allows or restricts future change and evolution. The progression has thus been from a view of simple, atomistic inner symbols to a notion of spatially distributed patterns, to a notion of spatially and temporally distributed patterns. The inner vehicles of content, courtesy of this progression, have come to look less and less like simple inner states and more like increasingly complex inner processes. This metamorphosis, moreover, is probably still incomplete. Future developments look set to include seeing many inner vehicles as multiply functional and seeing aspects of the inner architecture as dynamically reconfigurable. Multiple functionality would mean that one and the same inner resource may play a variety of contentbearing roles13 (perhaps varying in accordance with local context). Dynamic reconfigurability would mean that the inner architecture is itself subject to rapid change and reorganization, as when the release of a chemical neuromodulator causes two neural networks to temporarily fuse and behave as one. 9 See D. Rumelhart, J. McClelland and the PDP Research Group, Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Vote. I and II (Cambridge, MA: MIT Press, 1986). 10 P. S. Churchland, Neurophilosophy (Cambridge, MA: MIT Press, 1986). 11 The details need not concern us here. But see e.g. my Microcognition: Philosophy, Cognitive Science and Parallel Distributed Processing (Cambridge, MA: MIT Press, 1989), for discussion. 12 J. Elman, 'Representation and Structure in Connectionist Models', in G. Altman (ed.), Cognitive Models of Speech Processing (Cambridge, MA: MIT Press, 1991). 13 For some hints of such content-sensitive complexity, see J. Knierim and D. Van Essen, 'Visual Cortex: Cartography, Connectivity and Concurrent Processing', Current Opinion in Neurobiology 2 (1992), 150—5. 38

Embodiment and the Philosophy of Mind The moral, then, is that our understanding of the nature of the (putative) inner vehicles of content is in a state of extreme flux, characterized by a rapid flight from the initial image of static, chunky unstructured inner symbols. This flight has a content-related aspect too. For as the inner vehicles have become more complex, so the characteristic contents seem to have become more partial and fragmentary. This is because the emphasis has shifted from isolationist forms of problem-solving towards iterated series of agentenvironment interactions. This shift is nicely exemplified by recent work in the field known as Animate Vision.14 Recall Marr's depiction15 of the task of vision. The task, according to Marr, is to construct a rich inner model of the three-dimensional visual scene on the basis of the available (two-dimensional) input information. Recent work in the field known as Animate Vision takes a very different tack. This work depicts the task as, simply, the use of visual strategies to control behaviour, in realworld contexts, at as low a computational cost as possible. To this end, Animate Vision avails itself of three central ploys. 1. The use of task-specific cues and shortcuts. 2. The use of body-centred (egocentric) strategies. 3. The use of repeated environmental interactions. Task-specific cues and shortcuts include, for example, the use of personalized idiosyncratic strategies such as searching for bright yellow (a cheap, easy visual cue) when searching for my coffee cup (which just happens to be canary yellow). Egocentric strategies include the use of so-called deictic pointers (explained below). Repeated environmental interactions include, for example,the use of repeated visual saccades to visit and re-visit different aspects of a scene retrieving specific information only as and when required. The case of deictic pointers can serve as a general illustration. A pointer in classical Artificial Intelligence is an inner state which can function in self-contained computational routines but which can also point to other data structures.16 This pointing allows the retrieval, when required, of more detailed information, and the effective binding of certain items of information to others. Such binding often needs to be temporary, as when we bind certain features (e.g. bright yellow) to certain current (but clearly temporary) visual locations (e.g. 'y e n o w detected at the top left of visual field'). Deictic pointers, however, are actual bodily orientations (such as 14 D. Ballard, 'Animate Vision', Artificial Intelligence 48 (1991), 57-86. 15 D. Marr, Vision (San Francisco, CA: W. H. Freeman, 1982). 16 See, e.g., Z. Pylyshyn (ed.), The Robot's Dilemma: The Frame Problem in Artificial Intelligence (Norwood: Ablex, 1987). 39

Andy Clark saccadic eye movements) that play the same kind of functional role. The idea is that the system is set up so that the very act of fixating a particular aspect of a visual scene implements a kind of temporary variable binding in which the detected features are bound to a given spatial location. A related example concerns binding a 'reachingand-grasping' routine to a target object. Here too the binding may be cheaply implemented using what is informally called a 'do-itwhere-1'm-looking' strategy. Here, the system is set up so that the grasping routine is automatically directed to the currently fixated visual location. In all these cases, the authors comment: The external world is analogous to computer memory. When fixating a location, the neurons that are linked to the fovea refer to information computed from that location. Changing gaze is analogous to changing the memory reference in a silicon computer.17 One important thrust of the Animate Vision research, then, is that bodily actions (such as saccadic eye motions) can play vital computational roles. Another is that repeated agent-environment interactions obviate much of the need to create all-purpose, detailed internal world models. By visiting and re-visiting different aspects of the current scene as and when required, we allow the world to function as 'its own best model'. The research programme is thus staunchly anti-isolationist. But it is not by any means 'post-Cartesian' - it does not reject the very ideas of internal models and representations, so much as reconfigure them in a sparser and more interactive image. We still read of 'inner databases' e.g. ones that associate small objects, such as my car keys, with larger, easily detectable locations, such as on the kitchen table), of 'internal featural representations' (of colour, shape, etc.), of 'indexical representations' that specify locations relative to bodily position and so on. What is being rejected is emphatically not the notion of inner content-bearing states per se, but rather the much more specific notion that we construct rich, memory-intensive internal representations of all aspects of the current visual scene. A similar profile is presented by much actual research into realworld robotics. A good example is work18 in real-world robotic navigation in which knowledge of location is directly encoded as a perceptuo-motor routine: a routine that actually specifies how to move 17 D. Ballard, M. Hayhoe, P. Pook and R. Rao, 'Deictic Codes for the Embodiment of Cognition', Behavioral and Brain Sciences, forthcoming. 18 M. Matraric, 'Navigating with a Rat Brain: A Neurobiologically Inspired Model for Robot Spatial Representation', in J.-A. Meyer and S. Wilson (eds.), From Animals to Animals 1 (Cambridge, MA, MIT Press, 1991). This work is further discussed in Clark, Being There, chapter 2. 40

Embodiment and the Philosophy of Mind the robot from its present position to the target location. In these models the inner map is itself the controller of the appropriate action. There is no need for a further system to access the map and to plan a route. The robot's knowledge is thus both descriptive and prescriptive19 - a dual nature that affords great economies both in terms of response-time and computational effort. Once again, we see interesting work that is not so much anti-representational as sparsely representational. The crucial distinction, it seems to me, is not between representational and non-representational solutions so much as between rich and expensive forms of internal representation (which may increase flexibility but which often require additional computational work to specify a behavioural response) and sparser, more action-oriented forms of representation. The most convincing work in Animate Vision and real-world robotics thus stops well short of the full 'post-Cartesian' rejection of inner models and representations. Why, then, have some theorists gone on to question the idea of internal representations and inner models tout court? Ill Radical Interactionism The leading anti-representationalist argument20 seems to turn on the impact of dense, reciprocal causal exchanges uniting agent and environment in a complex web of mutual influence. Under such conditions, it is argued, the kinds of representational decomposition and analysis that work so well for many contemporary computer models of intelligent processes simply get no foothold. The problem (it is argued) is that the notion of x representing y is too oneway and too simplistic to do justice to cases in which x is continuously affecting and being affected by y and vice versa. Yet typical 19 For more on this theme, see R. Millikan 'Pushmi-pullyu Representations', in L. May, M. Friedman and A. Clark (eds.), Mind and Morals (Cambridge, MA: MIT Press, 1996). 20 This argument is the centrepiece of Van Gelder, 'What Might Cognition Be?' where we read, for example, that: 'The core dynamical hypothesis ... goes hand in hand with a conception of cognitive systems ... as complexes of continuous, simultaneous and mutually-determining change. [ ] In this vision, the cognitive system is not just the encapsulated brain; rather, since the nervous system, body, and environment are all constantly changing and simultaneously influencing each other, the true cognitive system is a single unified system embracing all three' (p. 373). The argument is also visible in Van Gelder and Port, 'It's About Time', pp. 23-5, in Thelen and Smith, A Dynamic Systems Approach, p. 27, and in Varela, Thompson and Rosch, The Embodied Mind, pp. 172—5. 41

Andy Clark agent—environment interactions, the argument continues, often present just such a complex and circular causal profile. Consider ballroom dancing. As you dance, your motions (if you are a good dancer!) are both continuously influenced by and an influence upon, those of your partner: the two sets of motions 'coevolve' in a highly interdetermined way. Nor is the presence of two human agents essential to the phenomenon. The same complex relation obtains between (for example) an experienced windsurfer and her rig: the windsurfer constantly affects and is affected by the set of the rig. Van Gelder makes the same point using the extended example of the Watt (or centrifugal) governor - a device which maintains a steam engine at a steady speed by both affecting and being affected by the engine speed.21 Such episodes of mutual influence were much discussed both in early cybernetics22 and in the work of the French phenomenologist Maurice MerleauPonty.23 Where such continuous, dense, circular causal influence obtains, it is argued, the tools of representational (and computational) analysis run aground. The idea of explaining the shape of these complex ongoing agent-environment interactions by depicting inner states as representing outer ones is rejected as coarse and unilluminating. Instead, inner and outer co-evolve in a mathematically precise way that is best captured (so the argument goes) by the use of coupled differential equations in which the current values of certain internal variables appear as parameter settings in the evolution equation for the external system and vice versa.24 Fortunately, the details of such a dynamical systems model are unimportant for present purposes.25 What matters is rather the general shape of the argument. Van Gelder puts it well: 21 Here Van Gelder 'What Might Cognition Be?' (p. 353) notes that: 'arm angle and engine speed are at all times both determined by, and determining, each other's behavior ... there is nothing mysterious about this relationship ... Yet it is much more subtle and complex than the standard concept of representation can handle.' This example is treated in detail in A. Clark and J. Toribio, 'Doing Without Representing?' Synthese 101 (1995), 401-31. 22 For example, in W. Ross Ashby's Introduction to Cybernetics (New York: Wiley, 1956). 23 Maurice Merleau-Ponty, The Structure of Behavior (New York: Beacon, 1963). Originally La Structure du Comportment (Paris: Presses Universitaire de France, 1942). 24 For an accessible introduction to these dynamical approaches, see S. Kelso Dynamic Patterns (Cambridge, MA: MIT Press, 1995). A classic text is R. Abraham and C. Shaw Dynamics — The Geometry of Behavior (Redwood, CA: Addison-Wesley, 1992). 42

Embodiment and the Philosophy of Mind The internal operation of a system interacting with an external world can be so subtle and complex as to defy description in representational terms, (p. 381.) Before responding to this argument, it is worth pausing to clarify the challenge. For what is at issue is not the status of certain systems (ourselves, for example) as representers. That is a given. We surely do represent our world, our past, our possible futures, our absent friends and so on. We think of these things and states of affairs and to that extent we clearly represent them to ourselves. What is not a given (and what is at issue here) is that we use internal representations to do so. The point is that the scientific claim that cognition involves internal representations (and computations defined over them) is meant not as a simple rehearsal of the fact that we are thinkers, but as a substantial and explanatorily potent empirical hypothesis: the kind of thing that could indeed turn out to be false. The falsifiable claim, to a first approximation, is that there exist distinct, identifiable inner states or processes whose systemic or functional role is to stand in for specific features or states of affairs. This notion of internal stand-ins is, however, itself ambiguous. It is ambiguous26 between a weak notion in which x 'stands in' for y iff x is an inner resource that (a) carries information about y and (b) is used to control behaviour, and a much stronger notion in which the inner resource must be capable of functioning as a genuine surrogate, i.e. be capable of systematically controlling appropriate behaviour even if y is absent or non-existent. A neural population27 closely keyed to bodily orientation and used to control on-line skilled action may thus be counted as a system of weak stand-ins. And even here the representational gloss seems to tell us something useful about the purpose of the neuronal population. But such a population, though it engages in the information-based control of action, need not be capable of driving appropriate actions in the absence of the (weakly represented) state of affairs. It is this latter, and surely less common, capacity to act as an inner surrogate in the absence of direct environmental control that, I suggest, characterizes the strongest 25 For a fuller discussion, see Clark, Being There, chapters 5, 6 and 8. 26 See A. Clark and R. Grush, 'Towards a Cognitive Robotics' (submitted). 27 For example, the posterior parietal neuronal population in the rat which encodes information about which way the rat's head is facing and which is exploited in radial maze running - see B. Naughton and L. Nadel, 'Hebb-Marr Networks and the Neurobiological Representation of Action in Space', in M. Gluck and D. Rumelhart (eds.), Neuroscience and Connectionist Theory (Erlbaum, 1990). 43

Andy Clark and most conceptually unequivocal cases of internal representation.28 The problem then is that the entire argument concerning the circular causal complexity of rich agent—environment interactions is vitiated by its failure to engage the real issue of strong representation. All the examples share (and must share) a certain problematic feature, namely, they are all cases in which the target behaviour is continuously driven and modified by the relevant environmental parameter. Yet one major motivation for positing internal representations in the first place was to explain our puzzling capacity to go beyond such tightly coupled agent-world interactions and to coordinate our activities and choices with the distal, the modal and the non-existent. The original notion of internal representation is thus grounded in the notion of strong inner surrogates and is merely extended (perhaps problematically) to the case of (merely) information-bearing inner states used for the control of action. This helps to explain why the best cases for the argument-from-continuousreciprocal-causation may well strike us as rather poor examples of traditionally cognitive phenomena. For they depend crucially on the constant presence of the relevant environmental factors and thus do not strike us as especially representation-hungry29 in the first place. Properly representation-hungry scenarios would be, for example, planning next year's vacation, using mental imagery to count the number of windows in your old house (this example is from Dan Dennett, in conversation), doing mental arithmetic, dreaming, etc., etc. The dialectical situation is, however, rather delicate. For the antirepresentationalist may now reply that the point of her argument, in part, was to suggest that these traditional cases (of what might be termed 'environmentally de-coupled' reason) are in fact empirically marginal and that the bulk of daily intelligent response displays precisely the richly interactive profile the argument highlights. Environmentally de-coupled reason, it is claimed, is at best a tip-ofthe-iceberg phenomenon. What is being promoted is thus a shift of emphasis away from off-line cogitation and onto real-time interactive engagement30 — a kind of cognitive-to-coping shift. This shift in emphasis is in one sense welcome. From both an 28 David Israel 'Bogdan on Information', Mind & Language 3/2 (1988), 123-40 makes essentially the same point. See also Brian Cantwell Smith, The Origin of Objects (Cambridge, MA: MIT Press, 1996). 29 The phrase is from Clark and Toribio, 'Doing Without Representing?'. 30 This move is explicitly made in Haugeland, 'Mind Embodied' and is also clearly in evidence in van Gelder and Port, 'It's About Time'. 44

Embodiment and the Philosophy of Mind evolutionary and a developmental31 point of view, real-world realtime responsiveness is clearly in some sense primary. But as we shall now see, the notion that the richly interactive case is in some way biologically basic is in fact perfectly compatible with the claim that off-line environmentally de-coupled reason is not the mere tip of the adaptive iceberg. Indeed, the way to forge a genuinely cognitive science of embodied, environmentally embedded agency is, I believe, precisely to target the relations between densely coupled and more strongly representationally mediated forms of adaptive success. This is the project that I dub Minimal Cartesianism, and to which we now turn. IV Minimal Cartesianism Minimal Cartesianism seeks to locate the roots of strongly representational reason in the richly interactive settings emphasized in work on embodied cognition. Thus consider the phenomenon of skilled reaching.32 Smooth, skilled reaching involves the use of proprioceptive feedback - signals that tell the brain how the arm is oriented in space. But the timing of these signals poses a problem. The minimal delay between the onset and the use of such information looks to be between 200 and 500 milliseconds.33 Yet we make essential trajectory corrections, that look to be governed by such feedback, within the first 70 milliseconds34 of reaching. How does nature turn the trick? This problem of requiring feedback before it is practically available crops up in industry too: in chemical plants, bioreactors and so forth.35 One common solution, in these cases, is to add a forward 31 See Thelen and Smith, A Dynamic Systems Approach. 32 I borrow this case from R. Grush, 'Emulation and Cognition' (Ph.D. Dissertation, University of California at San Diego, 1995). A further treatment is available in Clark and Grush, 'Towards a Cognitive Robotics'. 33 This figure is established by, for example, using artificial vibrators strapped to the tendons to disrupt proprioceptive signals arriving from the muscle spindles, and timing the gap between such disruptive input and alterations to the arm motion itself (see C. Redon, L. Hay and J. L. Velay, 'Proprioceptive Control of Goal Directed Movements in Man, Studied by Means of Vibratory Muscle Tendon Stimulation', Journal of Motor Control 23/2 (1991), 101-8). 34 See J. van der Meulen, R. Gooskens, J. J. Dennier van der Gon, C. C. A. M. Gielen and K. Wilhelm, 'Mechanisms Underlying Accuracy in Fast Goal-directed Arm Movements in Man', Journal of Motor Behavior 22/1 (1990), 67-84. 35 See Grush, 'Emulation and Cognition' for a review. 45

Andy Clark model or emulator into the systems. This is a circuit that takes as input a specification of both the previous state of the system and the commands just issued, and that gives as output a prediction of the feedback that should later arrive. The emulator thus generates a kind of mock feedback signal available substantially in advance of the real thing. Nature, it now seems, may deploy much the same strategy. There is a growing body of neuroscientific evidence36 that suggests that neural circuitry spanning the cortico-spinal tract, the red nucleus, the inferior olive, the contralateral dentate and cerebellar cortex may be playing just such a role. Such circuitry looks to take a copy of the afferent motor command and to output a fast prediction of the feedback later due arrive by the slow 200-500 millisecond route. The same trick has been replicated in a variety of neural network37 models. What matters for our purposes, however, is an additional conjecture. It is the conjecture38 that the biological emulator circuit plays a dual role. This dual role involves both the fine tuning of on-line reaching (the normal case, in which the emulator circuit acts as an aid to smooth real-time reaching) and the production of visuo-motor imagery allowing the off-line mental rehearsal of motor routines. In the latter case, the emulator circuit is now running alone, de-coupled from the real-world action system. Such an additional role for the very same emulator circuitry implicated in daily skilled reaching looks evolutionary plausible and helps to explain some otherwise puzzling results. These include the robust finding that mental rehearsal can actually improve sports skills39 and the surprising activity of the cerebellum (generally thought of as a motor area) during mental imagery.40 36 See M. Ito, The Cerebellum and Neural Control (New York: Raven Press, 1984), M. Kawato, K. Furukawa and R. Suzuki, 'A Hierarchical Neural Network Model for the Control and Learning of Voluntary Movement', Biological Cybernetics 57 (1987), 169-85, and D. Wolpert, Z. Ghahramani and M. Jordan, 'An Internal Model for Sensorimotor Integration', Science 269 (1995), 1880-2. 37 E.g. M. Kawato, 'Computational Schemes and Neural Network Models for Formation and Control of Multijoint Arm Trajectory', in W. T. Miller III, R. Sutton and P. Werbos (eds.), Neural Networks for Control (Cambridge, MA: MIT Press, 1990), Wolpert et al., 'An Internal Model'. 38 Grush, 'Emulation and Cognition'. 39 See e,g. D. Fetz and D. Landers, 'The effects of Mental Practice on Motor Skill Learning and Performance: A Meta-Analysis', Journal of Sport Psychology 5 (1983), 25-57. 40 J. Decety, H. Sjoholm, E. Ryding, G. Stenberg and D. Ingvar, 'The Cerebellum Participates in Cognitive Activity', Brain Research 535 (1990), 313-17. 46

Embodiment and the Philosophy of Mind Motor emulation circuitry, if this is correct, is both an aid to fluent, real-world action and a support for independent, environmentally decoupled mental rehearsal. It is thus a minimally Cartesian mental tool, but one that is parasitic upon adaptations closely geared to the promotion of smooth real-time agent-environment interactions. As a result, even the modestly Cartesian phenomenon of visual imagination remains closely tied to the biomechanics and actiontaking profile of the agent. Given this profile, we can see why isolationist methodologies and assumptions may prove inadequate even in the case of certain kinds of environmentally decoupled cognitive skills. For such skills may remain action-oriented at one remove, courtesy of the constraints on the original endowment: an endowment that is now redeployed to serve 'off-line' 'Cartesian' ends. Crucially, this failure of isolationism should not be seen as an invitation to scepticism about internal representation and inner models. In the emulator case, at least, it is clearly apparent that we are now dealing with identifiable circuitry whose functional role is, at times, strongly representational. Yet the account is perfectly compatible with the various morals and emphases suggested by the action-oriented research discussed in section II above. The conciliatory position that I favour thus involves combining the stress on real-world, real-time action with a search for the biologically basic roots of more decoupled forms of thought and problem-solving. For it is only by confronting the latter class of cases that representationalism can be given a fair trial. V Scaling, Rationality and Complexity Minimal Cartesianism aims to build bridges between the recent emphasis on richly interactive tasks and the more traditionally cognitive focus on decoupled reason. To that end it stresses the use of multiple, partial, action-oriented inner models and of deictic, idiosyncratic and action-oriented internal representations. The compelling question at this point becomes whether we can really hope to explain the full gamut of human cognition without at some point just reinventing the classical image of context-neutral, rich, actionindependent, highly manipulable inner symbolic structures. In short, can Minimal Cartesianism scale up so as to account for the full complexities of 'higher cognition' ? Such 'scaling up', if it is to have a reasonable chance of success, must give due credit to the way external structures, linguistic actions and cultural practices all conspire to effectively re-configure the shape of the computational spaces we must negotiate in order to 47

Andy Clark solve more complex and abstract problems. Complex human cognition is best depicted as occurring at the fecund interface between a variety of action-oriented internal resources and a larger scaffolding of external structures, tools and practices: a supportive web that acts so as to substantially alter the computational spaces that can be explored by our form of basic, on-board biological reason. A classic example41 is the use of pen and paper to do (e.g.) long multiplication: a trick that allows us to use an iterated sequence of simple inner computations (such as 7 x 7, 4 x 4) and a sequence of externally stored and manipulated inscriptions so as to solve much more complex problems (such as 777 x 444). Public language, I elsewhere argue, plays a wide variety of similar roles.42 The mere act of labelling, as Dennett43 points out, affords great economies of search and classification, while the capacity for linguistic rehearsal may, according to Ray Jackendoff,44 be what enables us to attend to the details of our own thoughts — thus opening up vast new possibilities of reflection and analysis.45 External artefacts and social organizations likewise alter and transform the tasks that individual brains need to perform. The cognitive anthropologist Ed Hutchins46 offers, in this vein, a wonderfully detailed and persuasive account of the process of ship navigation in which it is the overall system comprised of multiple brains, bodies and instruments that solves the navigation problem. Each crew member within this larger nexus merely monitors and responds to certain simple environmental conditions. The responses alter a few aspects of the shared work space and thus promote and support similar forms of responsiveness among the others. The whole process constitutes an environmentally extended computational flow in which multiple agents, simple 41 D. Rumelhart, J. McClelland, P. Smolensky and G. Hinton, 'Schemata and Sequential Thought Processes in PDP Models', in D. Rumelhart, J. McClelland and the PDP research Group (eds.), Parallel Distributed Processing: Explorations in the Microstructure of Cognition, vol. II (Cambridge, MA: MIT Press, 1986), pp. 7-58. 42 A. Clark, 'Magic Words: How Language Augments Human Cognition', in P. Carruthers and J. Boucher (eds.) Thought and Language: Interdisciplinary Themes (Cambridge: Cambridge University Press, forthcoming). 43 D. Dennett, 'Labelling and Learning', Mind and Language 8 (1994), 540—8. See also Chapter 13 of his Darwin's Dangerous Idea (New York: Simon & Schuster, 1995). 44 R. Jackendoff, 'How Language Helps Us Think', Pragmatics and Cognition 4/1 (1996), pp. 1-34. 45 Ibid., pp. 19-22. 46 E. Hutchins, Cognition in the Wild (Cambridge, MA: MIT Press, 1995). 48

Embodiment and the Philosophy of Mind routines, and a variety of external props and artefacts (such as nautical slide rules) all combine to solve a complex problem. Even a minimally Cartesian treatment of basic biological reason may thus hope to scale up so as to illuminate the full panoply of human thought and reason. The trick is to take the issue of external scaffolding very seriously indeed (and especially to recognize the computational virtues of public language: the one action-neutral symbolic code we already know ourselves to possess). One implication of this approach to the scaling problem is that we will need, at times, to study these larger systems (of multiple communicating brains and artefacts) as organized wholes and to recognize extended computational processes spanning the boundaries between brain, body and world. Such assertions can easily be mistaken for antipathy towards the study of inner resources and processes. But the real challenge, once again, is to interlock the two approaches and thus to relocate individual human reason in its proper ecological niche. The project also raises questions about the notion of human rationality itself. Isolationist cognitive science tended to depict rationality in terms of semantically apt transitions between inner mental states. Turing's achievement, as repeatedly stressed by Jerry Fodor,47 was to show how such transitions could be supported by a purely mechanical process. The environmentally extended approach just mooted need not reject that account. It may (and should) incorporate Turing's central idea of inner processes whose syntactic48 properties preserve semantic relations. But this will be just part of a more encompassing theory that allows rational behaviour to supervene on wider webs of structure involving other agents, artefacts and aspects of the local environment. There remains a worry about complexity. Even if the general project sketched in this paper proves attractive (the project of bridging between interaction-based models and more environmentally decoupled forms of reason), it could still turn out that the inner vehicles of content prove too spatially and temporally complex to figure in illuminating accounts of mental processes. Such a worry gains some force from recent demonstrations of the role of complex recurrent connections49 in modulating the information-processing profile of neuronal populations and from the sheer difficulty of 47 For example, see the comments on pp. 277—8 of his 'Replies to Critics', in B. Loewer and G. Rey (eds.), Meaning in Mind: Fodor and his critics (Oxford: Blackwell, 1991), pp. 255-319. 48 Syntactic properties are any non-semantic properties that can be directly exploited by a physical system. Temporally extended processes, as described in section II, are in this sense syntactic too. 49 Knierim and Van Essen, 'Visual Cortex', 150-5. 49

Andy Clark assigning specific content-bearing roles to tracts of neural machinery. These complexities and difficulties can lead to a subtly different kind of scepticism in which it is the complexity of the inner story itself (not the inner-outer interaction) that is supposed to make trouble for the representational analysis. The issues here are more purely empirical and it is impossible, given the current state of research, to make any firm predictions. But one interesting possibility is that new analytic tools may yet provide the means to identify functionally important patterns of activity. Dynamical systems analyses, of the kind sometimes promoted as an alternative to the representational approach, may in fact help us to identify tractable inner vehicles despite the presence of burgeoning spatial and temporal complexity. This possibility is clearly noted by van Gelder50 himself, who allows that 'an exciting feature of the dynamical approach is that it offers opportunities for dramatically reconceiving the nature of representation in cognitive systems'. Internal representations, then, may be realized not as simple inner states but as dynamical patterns of just about any conceivable kind. Such patterns may, in addition, be transient entities that form only in response to the details of current context. We thus better appreciate the limits of the inner vehicle metaphor itself. Such vehicles need be neither simple nor static in order to play a representational role. Van Gelder's observation is important. He does not take himself to have shown that there are no internal representations: just that there might not be any, and that if there are they may take a very different form to the one we once expected. I have tried to show that some of the more specific sceptical considerations he advances (concerning the potential complexity of agent-environment interactions) fail (and must fail) to make contact with the original pro-representationalist argument: an argument grounded directly in our capacities for environmentally decoupled reason. The revisionary representationalist option, however, remains open, appealing and increasingly in evidence in actual cognitive scientific applications.51 In sum, our vision of biological reason is rapidly changing. There is a growing emphasis on the computational economies afforded by real-world action and a growing appreciation of the way larger structures (of agents and artefacts) both scaffold and transform the space of individual reason. These twin forces converge on a rather more minimalist account of individual cognitive processing - an account that tends to eschew rich, all-purpose, action-neutral internal models and sentential forms of internal representation. Such 50 Van Gelder, 'What Might Cognition Be?'. 51 See papers in Port and Van Gelder (eds.), Mind as Motion. 50

Embodiment and the Philosophy of Mind minimalism, however, has its limits. Despite some rather ambitious arguments, there is currently no reason to doubt the guiding vision of individual agents as both loci of modest internal representations and users of a variety of inner world models. Rather than opposing representationalism against interactive dynamics, we should be embracing a broader vision of the inner representational realm and seeking the crucial continuities between tightly coupled behavioural strategies and the more 'Cartesian' space of environmentally decoupled reason. Our reward will be a better vision of rational agency itself. 51

Folk Psychology and Mental Simulation MARTIN DAVIES AND TONY STONE This paper is about the contemporary debate concerning folk psychology - the debate between the proponents of the theory theory of folk psychology and the friends of the simulation alternative} At the outset, we need to ask: What should we mean by this term 'folk psychology'? Shall we perhaps say that folk psychology is just what the folk know (or believe) about psychological matters? The problem with this putative definition is that, if folk psychology is a body of known or believed propositions about psychology, then it may be said that folk psychology is a psychological theory. This would threaten to render invisible even the possibility of an alternative to the theory theory of folk psychology. Someone might respond to this problem by saying that not just any collection of propositions about psychology deserves to be called a theory. Only a set of propositions organized around generalizations that support counterfactuals and are appropriately Some of the material in this paper was presented at the Pacific Division meeting of the American Philosophical Association and at the University of Michigan, as well as at seminars in Canberra, Melbourne, Oxford, Paris and Sydney. We are grateful to many friends and colleagues, including Ned Block, Greg Currie, Allan Gibbard, Robert Gordon, Paul Harris, Jane Heal, Frank Jackson, Janet Levin, Christopher Peacocke, Philip Pettit, Huw Price, Peter Railton, Ian Ravenscroft, Michael Smith, Dan Sperber, Stephen Stich and Kendall Walton, for comments and conversations. MD is pleased to acknowledge financial support from the Australian National University and the Humanities Research Board of the British Academy and is especially grateful to members of the Philosophy Department at the University of Michigan for the opportunity to visit as the James B. and Grace J. Nelson Philosopher in Residence. 1 Much of the relevant literature is gathered in three collections: Folk Psychology: The Theory of Mind Debate, ed. M. Davies and T. Stone (Oxford: Blackwell Publishers, 1995); Mental Simulation: Evaluations and Applications, ed. M. Davies and T. Stone (Oxford: Blackwell Publishers, 1995); and Theories of Theories of Mind, ed. P. Carruthers and P. K. Smith (Cambridge: Cambridge University Press, 1996). 53

Martin Davies and Tony Stone objective will earn that title.2 So, folk psychology will be a theory only if what the folk know or believe about psychology has something of the character of a science. This response has some plausibility. There is surely something to be said for this restrictive use of the term 'theory', and it will be important in Section III of this paper, when we consider explanation and understanding. But many of the participants in the debate between the theory theory and the simulation alternative have used the term 'theory' in an extremely inclusive way. For example, Stephen Stich and Shaun Nichols adopt a 'wide interpretation' of the term on which 'just about any internally stored body of information about a given domain [counts] as an internally represented theory of that domain'.3 Our initial aim is to describe the debate - or at least one aspect of the debate - in a way that takes account of the use of the term 'theory' to include any body of knowledge, belief or information. Instead of beginning with folk psychology as what the folk know or believe about psychology, we do better to start with folk psychological practice — a practice in which we all engage on an everyday basis. We describe people as bearers of psychological states. We explain people's behaviour (or decisions, or judgements or other psychological states) by appeal to their psychological states. We predict people's behaviour (or decisions, or judgements or other psychological states) by relying on assumptions about their psychological states. The debate between the theory theory and the simulation alternative can then be seen as a debate about this three-stranded practice.4 2T. Nagel, The View from Nowhere (Oxford: Oxford University Press, 1986), p. 5: 'A view or form of thought is more objective than another if it relies less on the specifics of the individual's makeup and position in the world, or on the character of the particular type of creature he is.' 3 S. Stich and S. Nichols, 'Folk Psychology: Simulation or Tacit Theory?' in Folk Psychology, ed. Davies and Stone, p. 133. See also S. Stich and S. Nichols, 'Second Thoughts on Simulation', in Mental Simulation, ed. Davies and Stone; S. Nichols, S. Stich, A. Leslie and D. Klein, 'Varieties of Off-Line Simulation', in Theories of Theories of Mind, ed. Carruthers and Smith. 4 The debate (particularly in its early stages) seems to have been conducted under two assumptions. One is that there is a single question to be asked about folk psychology. The other is that the theory theory and the simulation alternative offer the only two viable approaches to answering that question. But both of these assumptions are flawed. As against the first assumption, we would say that there are many different, and fairly independent, questions to be asked about folk psychological practice, each one of which might be given a theory theory or a simulation theory style of answer. (See T Stone and M. Davies, 'The Mental 54

Folk Psychology and Mental Simulation Amongst the many questions that can be asked about folk psychological practice, one question that has been central in much of the recent literature is the basis question: What is the basis of our ability to engage in folk psychological practice?5 Indeed, a great deal of attention has been focused on the basis question applied to just the prediction strand of folk psychological practice. The greater part of this paper shares this relatively narrow focus (sections I and II). Only in the final section do we move to consider explanation and understanding. I Prediction, Theory and Simulation What would be the theory theory's account of folk psychological prediction, and what alternative account would the simulation theory offer? We approach the question indirectly by considering first a case of prediction in a straightforwardly physical domain. How could someone predict the change in pressure of the gas in a cylinder when its temperature is raised? Prediction in a physical domain One possibility would be to use an empirical generalization about the way in which the pressure of a volume of gas increases as its 5 Elsewhere (Stone and Davies, 'The Mental Simulation Debate: A Progress Report', p. 120), we have put the question this way: 'What resources do mature adult humans draw upon as they go about the business of attributing mental states, and predicting and explaining one another's mental states and actions?' We called it the explanatory question about normal adult folk psychological practice. We have now opted to call it the basis question lest the term 'explanatory' suggest that the question relates only to the explanation strand of folk psychological practice. Simulation Debate: A Progress Report', in Theories of Theories of Mind, ed. Carruthers and Smith, pp. 119-20, for nine such questions. No doubt there are more.) As against the second assumption, we would make two brief points. One point is that we cannot simply assume that the two terms, 'theory theory' and 'simulation theory', even when quite generously construed, cover the whole space of possible answers to the questions that are at issue. The other point is that, even for a single question, and even when the theory theory and the simulation alternative are the only approaches in view, the correct answer might be a hybrid, drawing on both approaches. 55

Martin Davies and Tony Stone temperature increases.6 In this case, the predictor would be drawing on a body of information about gases, in line with a theory theory account. Another possibility would be to draw on a theory about the movement and energy of gas molecules. By considering the forces exerted on the walls of the cylinder, the predictor might arrive at a prediction of increased pressure without actually having antecedent knowledge of the temperature-pressure law. Or again, the predictor might not draw on any knowledge about gases in general, but simply make use of a formula relating the temperature and pressure of the gas in this particular cylinder. Given the inclusive notion of theory that is in play, this would count as use of a theory. There is, of course, an alternative to these theory-based strategies for arriving at a prediction about the pressure of the gas in a cylinder, A, after its temperature is raised. We could take another similar cylinder of gas, B, heat it to the temperature in question, and measure the pressure. Provided that the cylinder B really is relevantly similar to cylinder A, this method is liable to yield an accurate prediction. By using the behaviour of the second cylinder of gas as a simulation of the behaviour of the first cylinder, we can make a prediction about cylinder A in the absence of any antecedent empirical knowledge about changes in the behaviour of gases under increases in temperature. In order to use simulation in this way to predict the pressure in gas cylinder A, we need to use another real gas cylinder and we need to raise its temperature in reality. This simulation in reality provides for prediction in the absence of antecedent empirical knowledge about the behaviour of gases. A predictor who did not have a second cylinder to hand could, of course, imagine having a second gas cylinder. Or a predictor who was armed with a second gas cylinder but did not want to heat it could imagine its temperature being raised. But in order for either of these imaginative strategies to yield a prediction about the pressure in cylinder A, the predictor would need to develop the imagined gas cylinder narrative beyond its starting point ('There is a cylinder of gas. It is heated up. And then...'); and to do this, the predictor would need to draw on some theory - some body of information - about the behaviour of gases.7 6 The general principle is that pressure is proportional to (absolute) temperature and inversely proportional to volume. In the present context, the volume is constant. If, instead, the temperature is regarded as constant then the resulting principle, that volume is inversely proportional to pressure, is known as Boyle's law. 7 This kind of prediction by simulation in imagination is closely connected with the use of thought experiments in science. Thought experiments are often important in the development of theory, and so it may 56

Folk Psychology and Mental Simulation As a strategy for predicting the pressure in cylinder A, simulation in imagination must deploy essentially the same resources as those that are used according to the theory theory account. So, in this case at least, simulation in imagination is theory-driven simulation.8 It is only simulation in reality that constitutes a genuine alternative to the use of empirical theory in prediction. Psychological prediction In the folk psychological case, it is clear enough how knowledge of an empirical theory about psychological matters can yield predictions. The body of theory drawn on might consist of some relatively superficial generalizations about (personal level) psychological properties (cf. the laws relating temperature, pressure and volume of gases) or postulates about (subpersonal level) information processing apparatus (cf. postulates about the movement and energy of gas molecules); or it might be information that is specifically about a particular individual (e.g. someone whom the predictor knows well; cf. a formula linking temperature and pressure in cylinder A). It is also clear that, in the psychological case, simulation in reality can be an effective way of generating predictions without relying on knowledge of empirical theory. Suppose that I want to predict (i) how a person C will feel (or how soon C will fall over) after drinking a pint of whisky, or (ii) how the Miiller-Lyer illusion will look to C, or (iii) how C will feel and what he will decide to do if he is suspended over a cliff on a rope and he cannot find a foothold and his hands are starting to slip, or (iv) whether C will draw the conclusion that something is white from his belief that snow is white.9 8 See A. I. Goldman, 'Interpretation Psychologized', in Folk Psychology, ed. Davies and Stone, p. 85, for the distinction between theory-driven and process-driven simulation. 9 The whisky example is discussed by Jane Heal, 'How to Think About Thinking', in Mental Simulation, ed. Davies and Stone, p. 48, and by Richard Moran, 'Interpretation Theory and the First Person', seem implausible to say that simulation in imagination draws on theory. We need to note, once again, that an inclusive notion of theory is in play, and that in some cases the propositions drawn on will simply be intuitive assumptions about what kinds of thing do, or do not, tend to happen in the physical world. See, for example, R. Sorenson, Thought Experiments (Oxford: Oxford University Press, 1992), pp. 52-4, for an account of Stevinus's use (in 1605) of a thought experiment to determine the force needed to keep a ball from moving down an inclined plane. One of the assumptions at work in this case was that perpetual motion does not happen. 57

Martin Davies and Tony Stone In each case, I can use the strategy of placing another person, D, into the same situation and observing D's reactions. This may well yield a correct prediction about C, provided that C and D are relevantly similar ((i) in the way that alcohol affects their bodily constitution; (ii) in the way that their visual systems work; (iii) in the way that they experience and act on emotions; (iv) in the way that they reason). To the extent that I, myself, am relevantly similar to C, I have an option that is not available to me in the case of gas cylinder simulation in reality; namely, I can place myself into those situations and observe my own reactions. I drink a pint of whisky, or look at the two lines, or dangle perilously over a cliff, or draw out some simple inferences from my belief that snow is white. Indeed, in discussions of mental simulation in reality, it is usually this option of using oneself in a simulation that is considered.10 But it is mental simulation in imagination that is central for the simulation theory. We saw that gas cylinder simulation in imagination needs to be driven by empirical theory. Does the same go for 10 S. Stich and S. Nichols, 'Cognitive Penetrability, Rationality and Restricted Simulation', Mind and Language, 12 (1997), p. 302, call this 'actual-situation-simulation'. It is important to avoid a possible confusion here. In some important examples, a protagonist has a false belief about her situation: there is a difference between the situation as it actually is and the situation as the protagonist takes it to be. A subject who is asked to predict what the protagonist will think or do may make an incorrect prediction by focusing on the situation as it actually is rather than the situation as the protagonist takes it to be. (This is what very young children tend to do. There is a substantial empirical literature on the false belief task. See, for example, H. Wimmer and J. Perner, 'Beliefs About Beliefs: Representation and Constraining Function of Wrong Beliefs in Young Children's Understanding of Deception', Cognition, 13 (1983), pp. 103-28.) But this predictive strategy is not what Stich and Nichols mean by 'actual-situation-simulation' (and not what we mean by 'simulation in reality'). Rather, actual-situation-simulation would involve placing myself into the same situation as the protagonist and making myself (perhaps per impossibile) subject to the same false belief. Philosophical Quarterly, 44 (1994), p. 163. The Miiller-Lyer illusion is discussed by Robert Gordon, 'Reply to Stich and Nichols', in Folk Psychology, ed. Davies and Stone, pp. 175-6. The example of emotional response to a story is discussed by Kendall Walton, 'Spelunking, Simulation and Slime: On Being Moved by Fiction', in Emotion and the Arts, ed. M. Hjort and S. Laver (Oxford: Oxford University Press, 1997), and Ian Ravenscroft, 'What Is It Like to be Someone Else?: Simulation and Empathy', Ratio, 11 (1998). The case of inference is central in Heal's discussions. We take the example from Allan Gibbard, 'Brains, Thoughts, and Norms', unpublished manuscript. 58

Folk Psychology and Mental Simulation mental simulation? It seems clear that if, with a view to making a prediction about C, I imagine placing another person D into the same situation, then I shall need to draw on theory in order to develop the simulation beyond this starting point. But if what I imagine is actually being in the situation," then simulation in imagination might allow a prediction about C to be generated. What this prospect seems to depend on is the possibility that my imagining being in a situation engages the same psychological or mental processes in me as would be operative if I were really in that situation. Consider, then, the conditions under which simulation in imagination would yield correct predictions in the four sample cases that we have mentioned, (i) If simulation in imagination is to yield a correct prediction about how C will feel after drinking a pint of whisky, then imagining drinking a pint of whisky must produce in me feelings of giddiness leading to a fall - or at least imagined feelings of giddiness leading to an imagined fall, (ii) In the case of the MiillerLyer illusion, imagining the two lines and the arrowheads must lead to a visual experience — real or imagined — as of one line being longer than the other, (iii) When I imagine being suspended over a cliff on a rope, this act of imagination must lead to real or imagined feelings of fear and panic, (iv) When I imagine believing that snow is white (or, more to the point, when I imagine believing that, say, butter is white - something that I do not, in reality, believe), this must lead to the real or imagined act of judging that something is white. We take it that the facts about these cases are roughly as follows, (i) Imagining drinking a pint of whisky does not, in and of itself, produce real or imagined feelings of giddiness. The bodily processes that lead up to a feeling of giddiness are not engaged by the imagined consumption of alcohol in the same way that they would be engaged by the real consumption of alcohol. If my simulation in imagination does move forward from the drinking to the feelings, then this is because I am bringing to bear some empirical knowledge about how people typically feel — or about how I usually feel — after consuming large quantities of alcohol. (ii) Imagining the lines and the arrowheads does not, in and of itself, generate the Miiller-Lyer illusion in imagination. The visual processes that give rise to the illusion are not engaged by the imagined confrontation with that array of lines and arrowheads in the same way that they would be by the real presentation of the array, (iii) On the other hand, imagining being in that dangerous situation, 11 B. A. O. Williams, 'Imagination and the Self, in Problems of the Self (Cambridge: Cambridge University Press, 1973). 59

Martin Davies and Tony Stone dangling at the end of a rope, may well lead to real feelings of fear or panic, without my drawing on any empirical theory about how people in that kind of situation typically feel. Imagined danger may engage a range of bodily and emotional processes in somewhat the same way that real danger does. (iv) Finally, imagining believing the premises of an argument (that butter is white) certainly can lead me to an imagined judgement of the conclusion (that something is white), without my using any antecedently known empirical theory about how people typically reason. There is an important contrast between the case of reasoning from imagined beliefs and the case of emotional response to imagined danger. The bodily symptoms of fear or panic may well be real, even though the danger is only imagined. But in the case of reasoning, if my commitment to the premises is only an imagined commitment, then my judgement of the conclusion is similarly imagined rather than real. The process leading from one to the other is, however, real, and not merely imaginary, reasoning; and that real reasoning may also prompt a real judgement, namely, the conditional judgement that if the premises were true then so would be the conclusion. What all this suggests is that the prospects for psychological prediction by simulation in imagination, without the use of empirical theory, are not utterly forlorn. It may also seem to suggest that we need to set about the task of cataloguing which psychological processes are engaged in the same way by imagined inputs as by real inputs. But while real interest and importance would attach to that cataloguing project, it is also important to note that it is not just a brute fact that imagining premises engages our reasoning abilities in the same way that really believing the premises does.12 Rather, the explanation of this fact is that reason relations (such as entailment relations) obtain, and are known by any thinker to obtain, amongst imagined or hypothesized thought contents, in just the same way that they obtain amongst believed thought contents. When I simulate C's reasoning in imagination, a theory may well be used. But it is not an empirical theory about how people happen typically to reason. Rather, it is a normative theory about right reasoning; and it is the very same theory that I can use when I engage in reasoning from premises that I actually believe.'3 Although the simulation of reasoning may involve deployment of normative principles, the simulation theory is not (even when 12 This point is stressed, for example, by Goldman, 'Interpretation Psychologized', p. 85, and by Heal, 'How to Think About Thinking', pp. 34-5. 13 See Stone and Davies, 'The Mental Simulation Debate: A Progress Report', pp. 136-7. 60

Folk Psychology and Mental Simulation restricted to reasoning) to be equated with what might be called the normative theory theory. It is possible to know normative principles relating to an activity in which one does not, oneself, engage. But the simulation theory is clearly not proposing that we make predictions by the disengaged use of a set of normative principles about reasoning.14 Rather, normative principles may be used in simulation because they are already available to us when we ourselves engage in reasoning. When we use those normative principles, our reasoning becomes what Tyler Burge describes as critical reasoning. Critical reasoning is reasoning that involves an ability to recognise and effectively employ reasonable criticism or support for reasons and reasoning. It is reasoning guided by an appreciation, use, and assessment of reasons and reasoning as such. As a critical reasoner, one not only reasons. One recognises reasons as reasons... Essential to carrying out critical reasoning is using one's knowledge of what constitutes good reasons to guide one's actual firstorder reasoning.15 Not all reasoning is critical reasoning. But it is arguable that the possibility of critical reasoning is an essential part of normal adult reasoning as we know it.16 The point we have reached is that predicting the conclusions of another person's (theoretical or practical) reasoning appears to be a particularly favourable case for a simulation theory answer to the basis question about the prediction strand of folk psychological practice. Of course, in order to reach a correct prediction about C's conclusions by simulating his reasoning in imagination, I need to take account of differences between C and myself. If C believes that butter is white, while I do not, then C may arrive at the judgement that butter and snow are the same colour, given that snow is white, 14 See S. Blackburn, 'Theory, Observation and Drama', in Folk Psychology, ed. Davies and Stone, p. 283: 'Theorizing under a normative umbrella is still theorizing. It could, it seems, be done quite externally, in the light of a sufficient stock of principles telling what it would be right or wrong to think or feel in some situation...'. Janet Levin, 'Folk Psychology and the Simulationist Challenge', Ada Analytica, 14 (1995), p. 91, also makes the point that if we use a normative theory to predict what inferences a person will make then this does not yet seem to involve anything that is 'in any serious sense a simulation'. 15 T. Burge, 'Our Entitlement to Self-Knowledge', Proceedings of the Aristotelian Society, 96 (1996), pp. 98-9. 16 Burge, 'Our Entitlement to Self-Knowledge', p. 99: 'A non-critical reasoner reasons blind, without appreciating reasons as reasons. Animals and children reason in this way.' 61

Martin Davies and Tony Stone whereas I would not myself draw that conclusion. But I can take account of this difference between C and me within the simulation, without needing to draw on any empirical information about how people who believe that butter is white tend to reason. Rather, in imagination I take on the belief that butter is white and then, given the premises that snow is white and that butter is white, I conclude that butter and snow are the same colour. That is what right reasoning requires. Predicting how someone will feel after drinking a pint of whisky, in contrast, is a good case for a theory theory answer to the basis question. Consequently, predicting the conclusions that will be reached by someone reasoning after drinking a pint of whisky also depends on at least some contribution from empirical theory. If C has just drunk a pint of whisky and I have not, then I need to take account of this difference between him and me when I try to simulate his reasoning in imagination. Even if I correctly take on C's premises in imagination and imagine drinking a pint of whisky, still my predictions about C's conclusions are liable to be incorrect, unless I bring to bear some empirical information about how whisky affects (C's) reasoning. Here, correct prediction requires an intrusion of theory. But this is not to say that, in the case of the inebriated C, my prediction strategy must owe everything to empirical theory and nothing to mental simulation. The empirical information that I draw on might take the form of information about the ways in which someone in C's condition is liable to depart from right reasoning. In that case, I could first use my own reasoning ability to work out what would be a correct conclusion to draw from C's premises and then tweak my prediction in the light of that empirical information. The epistemology of prediction by simulation Let us now consider, in a little more detail, how prediction by simulation would work. We have already noted that, in the case of the gas cylinders, prediction by simulation in reality relies on some assumption of relevant similarity between cylinder B and cylinder A. One form that this assumption can take is that cylinder B is a typical member of a class, G, of gas cylinders of which A is also a member. Heating the gas in cylinder B and measuring its pressure can then be conceived as an experiment, licensing a general claim about temperature and pressure in gas cylinders in the class G. Since cylinder A is assumed to be a member of this class, the experimentally licensed generalization can be applied to it. Essentially the same kind of account could be given, in the psy62

Folk Psychology and Mental Simulation chological case, of the role of the assumption of relevant similarity between person D and person C. And if, in a case of mental simulation in reality, I use myself instead of another person D, then an assumption of relevant similarity between me and C plays the same role again. Placing myself in the situation can be conceived as an experiment. It would seem plausible, then, that there is no very deep difference between the epistemological status of predictions based on simulation and predictions that rely on experimentally licensed knowledge of empirical generalizations. Furthermore, it would appear that, in the case of mental simulation in imagination, much the same account would be given, but with an extra empirical assumption to the effect that the processes in me that are engaged by imagined inputs work in the same way as the processes in me, and in C, that are engaged by real inputs. The cataloguing project mentioned on p. 60 above would then be seen as the project of assessing the extent to which that empirical assumption is warranted. However, the account that we have sketched of simulation of reasoning in imagination may open the possibility of a distinctive epistemology of psychological prediction. What the normative theory of right reasoning tells the simulator is that the conclusion - say, that something is white - is the right thing to think, given the premise - say, that snow is white, or that butter is white. This normative judgement about what is the thing to think does not, by itself, yield a prediction about C, of course. The simulator also needs an assumption that C will think the thing that is the thing to think. That is a defeasible assumption in any given case. But it may enjoy a default status, nevertheless, since unexplained departures from these normative requirements of reasoning call in question our attributions to C of thoughts with such contents as that snow is white or that butter is white.17 This route to prediction goes via a normative judgement (this is the thing to think in such-and-such a situation) and an assumption about interpretation (C will think the thing that is the thing to think). It is to be contrasted with a route 17 The general idea here is familiar from discussions of the principles involved in radical interpretation. Some advocates of mental simulation contrast the simulation approach with the rationality approach, and so would not adopt the account of the epistemology of psychological prediction that is sketched here. See, for example, Goldman, 'Interpretation Psychologized'. On the other hand, R. M. Gordon, 'Simulation Without Introspection or Inference from Me to You', in Mental Simulation, ed. Davies and Stone, can be seen as resisting the idea that the epistemology of prediction by simulation is the same as that of prediction by way of empirical theory. 63

Martin Davies and Tony Stone that goes via an empirical judgement (this is what I think when placed experimentally in such-and-such a situation) and an assumption of similarity (C is relevantly like me). II Prediction Failure We have distinguished between simulation in reality and simulation in imagination as methods of prediction. Simulation in reality can certainly be an effective way of generating predictions without relying on empirical knowledge. But the prospects for prediction by simulation in imagination depend on the possibility that imagining being in a situation should engage the same psychological or mental processes as would be operative if one were really in that situation. We considered a range of examples and concluded that predicting the results of another person's reasoning is a good case for simulation in imagination while predicting how someone will feel after drinking a pint of whisky is not. But while it might be agreed that predicting the conclusions of reasoning could be achieved by mental simulation, this does not settle the question whether prediction is in fact achieved in that way. Perhaps, despite the availability of simulation, we ordinarily make such predictions by relying on an empirical theory about how people reason. The basis question with which we began is an empirical question about our three-stranded folk psychological practice, and we have been focusing on the question as it applies to the prediction strand. But we have so far said nothing about the kinds of empirical evidence that would count in favour of one or another answer to the basis question. In a series of important papers, Stich and Nichols have urged that the phenomenon of prediction failure is strong evidence in support of a theory theory answer to the question about the basis of our prediction practice.18 In our everyday folk psychological practice, we sometimes make wrong predictions. Stich and Nichols argue that this happens because our prediction method is cognitively penetrable — that is, our psychological predictions are influenced by our antecedent knowledge or beliefs about the psychological domain. This kind of explanation of prediction failure is available to the theory theorist but unavailable, Stich and Nichols say, to the friend of mental simulation. So the existence of prediction failure is a crucial test of the empirical adequacy of the two competing accounts of the causal 18 'Folk Psychology: Simulation or Tacit Theory?', 'Second Thoughts on Simulation', and 'Varieties of Off-Line Simulation'. We note again that Stich and Nichols use the term 'theory' in an extremely inclusive sense. 64

Folk Psychology and Mental Simulation basis of our prediction practice, and favours the theory theory account. Thus, on the one hand: One virtue of using a simulation to predict the behavior of a system is that you need have no serious idea about the principles governing the target system. You just run the simulation and watch what happens ... In predictions based on simulation, what you don't know won't hurt you ... If there is some quirk of the human decision-making system, something quite unknown to most people that leads the system to behave in an unexpected way in certain circumstances, the accuracy of prediction based on simulations should not be adversely affected. If you provide the simulation with the right pretend input, it should simulate (and thus predict) the unexpected output.19 But, on the other hand: Just the opposite is true for predictions that rely on a theory. If we are making predictions on the basis of a set of laws or principles, and if there are some unexpected aspects of the system's behavior that are not captured by our principles, then our predictions about those aspects of the system's behavior should be less accurate. Theory based predictions are sensitive to what we know and don't know about the laws that govern the system; they are cognitively penetrable.20 The dialectical situation that Stich and Nichols sketch is especially clear when we contrast theory-based prediction and prediction by simulation in reality. Thus, consider again our prediction of the pressure in gas cylinder A. If someone has a false theory about the behaviour of gases, then a theory-based prediction about cylinder A is liable to be false. But, if the predictor uses the behaviour of cylinder B as a simulation of the behaviour of cylinder A, then the prediction arrived at should be correct. Because the prediction method does not draw on any antecedently believed empirical theory about the behaviour of gases, the prediction can, in principle, be insulated from any false theoretical beliefs that are antecedently held by the predictor.21 If someone makes an incorrect prediction about the pressure of the 19 'Folk Psychology: Simulation or Tacit Theory?' p. 150. 20 Ibid. 21 In section I, we noted the similarity between gas cylinder simulation in reality and the use of experiments to establish generalizations about how gas cylinders in a certain class generally behave. The present point, that simulation in reality yields predictions that are insulated from antecedently held theory, is analogous to the point that experiments are apt to yield results that conflict with antecedently held theory. 65

Martin Davies and Tony Stone gas in cylinder A after it has been heated then either the predictor is not using simulation as the prediction method or else the simulation is flawed in one of two ways. It may be that cylinder B is not relevantly similar to cylinder A or it may be that the gas in cylinder B was not heated to the correct temperature.22 In the psychological case, just the same points can be made. If, in order to arrive at a prediction about C, I use D (or myself) for a simulation in reality, then the prediction should be correct. If it is incorrect then either D is not relevantly similar to C (or I am not similar to C), or else D (or I) was not placed into the correct situation (that is, the simulation was not provided with the correct inputs). But the central case of mental simulation is simulation in imagination. Is the dialectical situation the same here? There is some reason to allow that it is. Someone who claims that mental simulation provides even a possible account of folk psychological prediction relies on the idea that imagining being in a situation may engage the same psychological or mental processes as would be operative if one were really in that situation. For some examples, such as the situation in which one drinks a pint of whisky, the idea has no plausibility. But the advocate of mental simulation has to maintain that there are other cases where the idea is plausible, and we have suggested that these would include cases of theoretical and practical reasoning. So, it appears that prediction failure relating to reasoning would present a problem for anyone offering a mental simulation answer to the basis question about folk psychological prediction. Certainly, this is what Stich and Nichols have argued; and they have gone on to present examples of this kind of prediction failure. Examples of prediction failure There is no shortage of surprising experimental psychological data about conclusions that people draw and decisions that they take. The very fact that we find the data surprising indicates, of course, that we ourselves would have made incorrect predictions about what the subjects in the experiments would conclude or what they would decide. We shall describe two of these examples.23 22 Someone using simulation in reality as a prediction method may, of course, refuse to accept the result of a simulation if it conflicts with an antecedently held theory, and may judge that the simulation must be flawed in some way. The same goes for experimentation. 23 These two examples are discussed by Stich and Nichols, 'Folk Psychology: Simulation or Tacit Theory?' p. 151, along with the example of belief perseverance; see R. Nisbet and L. Ross, Human Inference (Englewood Cliffs, NJ: Prentice Hall, 1980), pp. 175-9. In 'Second 66

Folk Psychology and Mental Simulation Position effects: right bias in selecting goods Shoppers are presented with a display of what are, in fact, identical samples of some product. They are asked to assess the quality of these samples and then - by way of payment for participating in the survey - to select one sample to keep. The result is that the shoppers' selections show a bias towards samples near the right-hand end of the display over samples near the left-hand end.24 Most people are surprised to hear the result of this experiment; they would predict that shoppers' selections would be random. If these predictions are arrived at by mental simulation, then simulation is generating incorrect predictions. Yet it is reasonable to assume that the people who are asked to predict the outcome of the experiment are relevantly similar to the subjects in the experiment (the shoppers). The Langer effect Two groups of subjects are sold lottery tickets for $1 each. Subjects in one group are allowed to choose their lottery ticket (choice condition); subjects in the other group are simply given a ticket (nochoice condition). Subjects are then (under some pretext or other) asked to be ready to sell their ticket back to the experimenter, and are asked to set a sell-back price. The result is that subjects in the choice condition set very much higher prices on average than subjects in the no-choice condition (over $8 versus just under $2).25 Most people are surprised to hear the result of this experiment. For example, Stich and Nichols report anecdotal evidence of presenting undergraduate students with a description of the experiment and asking them to predict what the subjects would do. The students failed to predict the difference between the sell-back prices set by subjects in the choice condition and subjects in the no-choice condition. If these predictions are arrived at by mental simulation - the students simulating first being in one condition and then in the other - then simulation is generating incorrect predictions. Yet it is reasonable to assume that the students who are 24 Nisbet and Ross, Human Inference, p. 207. 25 E. Langer, 'The Illusion of Control', Journal of Personality and Social Psychology, 32 (1975), pp. 311—28. The example is discussed at length in Nichols et al., 'Varieties of Off-Line Simulation'. Thoughts on Simulation', pp. 101-2, they introduce the further example of failure to predict how subjects will behave in Milgram's obedience experiment. S. Nichols, S. Stich and A. Leslie, 'Choice Effects and the Ineffectiveness of Simulation', Mind and Language, 10 (1995), pp. 442—4, also discuss an example of subjects' failure to predict how they themselves will behave when asked to put a price on an article that they own. 67

Martin Davies and Tony Stone asked to predict the outcome of the experiment are relevantly similar to the original subjects. Response on behalf of the simulation theory Given the way that the argument about prediction failure has been set up, it will appear that the defender of mental simulation is bound to make a move analogous to saying that gas cylinder B was not heated to the correct temperature. That is, the defender of simulation must say that, in these cases of prediction failure, the (pretended) inputs to the predictor's simulation in imagination are in crucial respects different from the inputs that engaged the psychological processes of the subjects in the real experiments. This is, indeed, the way in which advocates of mental simulation have responded. Thus, for example, Robert Gordon comments on the example of right bias in selecting goods that, 'unlike the subjects in the original experiment, the subject in the imagination experiment [the person trying to predict how shoppers will behave] must be told that the items on display are identical (and thus of equal quality)'.26 In a similar vein, Paul Harris notes that a person trying to predict the outcome of the Langer experiment using simulation: needs to simulate the vacillation and eventual commitment of the free-choice subjects. Moreover, in making that simulation they must also set aside the tacit reminder embedded in a narrative that juxtaposes the two groups of subjects, namely that any lottery ticket whether selected or allocated, has the same likelihood of winning. Subjects in the experiment who were offered a free choice had no knowledge of the other group, and by implication, no such tacit reminder.27 This is a good initial move to make on behalf of the simulation theory. Someone who is aiming to make a prediction by simulation in imagination must imagine being in the very situation that the subjects in the original experiment were in. And this must be done in such a way as to offer the simulator's psychological processes inputs that are equivalent to the inputs that engaged the original subjects' psychological processes. In a case of simulation of reasoning, the simulator must take on in imagination the very same premises that were available to the original subject. But, as Gordon and Harris point out, the way in which the experimental situation is described may prevent this condition from being met. 26 Gordon, 'Reply to Stich and Nichols', p. 176. 27 P.L. Harris, 'From Simulation to Folk Psychology: The Case for Development', in Folk Psychology, ed. Davies and Stone, p. 218. 68

Folk Psychology and Mental Simulation There is a quite general point here; namely, that the way in which the situation to be imagined is described can make a huge difference to the prospects for successful simulation. Consider the case of a lexical decision experiment. Letter strings are flashed up on a computer screen — some strings form real words, and some form (pronounceable) non-words. The subject has to decide whether each letter string is a word or a non-word and press one or another button to indicate this decision. Suppose that I am asked to predict what decisions subjects will make. Simulation in reality is no problem here: I can just sit in front of the screen myself. But if I have to simulate this experimental regime in imagination, then some ways of describing the input make my task nearly impossible. I might, for example, be given a description of the screen display in terms of the pattern of light and dark pixels that form the image of the letter string. If, on the other hand, the screen display is described by the letters being named in order, then I may very well be successful in simulating the performance of subjects in the experiment and thus predicting their responses. This successful prediction would not seem to depend on antecedent knowledge about how normal subjects respond to this or that letter string in a lexical decision experiment. Rather, I would just make what I take to be the correct decision about each imagined letter string, and then assume that other subjects would make the correct decision too. In doing this, I would make use of stored information; but it would be information about lexical items, not information about normal subjects' lexical decisions. In our view, this line of response (in terms of wrong inputs) enables the simulation theorist to fend off the initial versions of the objection from prediction failure. But it does not resolve the debate in favour of either side because the response simply invites the theory theorist to improve the design of the prediction experiment so as to rule out the wrong inputs response. Thus, for example, Nichols, Stich, Leslie and Klein report a prediction experiment in which subjects in one group watch a videotape of a subject in the choice condition of a Langerstyle experiment and are asked to predict the subject's sell-back price, while subjects in another group similarly watch a videotape of a subject in the no-choice condition.28 As in the original Langer experiment, subjects in the choice condition set significantly higher sell-back prices than subjects in the no-choice condition. But there was no significant difference between the prices predicted by subjects shown the choice condition videotape and the prices predicted by subjects shown the video of the no-choice condition. Thus, even with a videotape to help them imagine the experimental situation, subjects are not reliably able to reach correct predictions. 28 Nichols et al., 'Varieties of Off-Line Simulation', pp. 49-52. 69

Martin Davies and Tony Stone There is no doubt that discussion of these examples can be continued, with the defender of prediction by mental simulation in imagination deploying variations on the wrong inputs theme.29 But there is a slightly different kind of response to these examples that is suggested by our earlier reflections on the prospects for psychological prediction by simulation in imagination (pp. 57-62 above). The circumscribed domain of prediction by mental simulation There are all kinds of factors that may affect a person's theoretical or practical reasoning, such as whether the person believes that butter is white, or whether the person has just drunk a pint of whisky. Some of these factors can readily be taken into account by someone attempting a prediction by mental simulation in imagination, while others cannot. Showing me a videotape of a subject drinking a pint of whisky before engaging in some reasoning will not enable me to predict the outcome of the subject's reasoning, however accurately I may imagine the subject's situation. As we noted on p. 59 above, what I need is empirical information about the effects of whisky drinking.30 (Recall, too, that the use of this empirical information need not wholly supplant engagement in mental simulation.) The fact that prediction by mental simulation in imagination 29 See for example, A. Kiihberger, J. Perner, M. Schulte and R. Leingruber, 'Choice or No Choice: Is the Langer Effect Evidence Against Simulation?', Mind and Language, 10 (1995), p. 433: '[I]t is difficult to ensure that simulator participants are provided with sufficient information about exactly the right combination of factors that produces the Langer effect.' Kiihberger et al. refer to the requirement that 'the imagined situation captures the relevant features of the simulated person's actual situation' as the requirement of imaginative adequacy (p. 424). A theory theorist may object that the use of the wrong inputs response by the friend of mental simulation is ad hoc and that the defender of the simulation theory in the face of examples of prediction failure should be willing to state in advance under what conditions the requirement of imaginative adequacy would be met. (See Stich and Nichols, 'Second Thoughts on Simulation', p. 102.) But it is not clear that the theory theorist's own approach to examples of prediction failure is any more principled. The theory theorist's explanation of prediction failure is in terms of the predictor's use of an incomplete or false theory about psychological matters, or the predictor's use of incorrect initial conditions to instantiate correct generalizations. But no independently motivated account of the exact nature of the predictor's failure is provided. (This point is made in an unpublished paper by Ian Ravenscroft, and also by Stich and Nichols, 'Cognitive Penetrability, Rationality and Restricted Simulation', p. 323, who credit it to Meredith and Michael Williams.) 30 Alternatively, I could drink a pint of whisky myself, combining simulation of the subject's beliefs in imagination with simulation of the subject's 70

Folk Psychology and Mental Simulation requires an intrusion of theory in such cases of 'non-rational' influences has been recognised from the beginning of the contemporary debate.31 Furthermore, it seems quite likely that some of the factors at work in producing the Langer effect or the right bias in selecting goods may be more like whisky than like reasons. For example, most people who are told about the position effects experiment find it surprising that the shoppers' selections show a bias towards samples near the right-hand end of the display. They would predict a random distribution of selections. A plausible explanation for this prediction is that there is no evident reason to make one selection rather than another; the fact that a sample is towards the right-hand end of the display scarcely constitutes a justification for selecting that sample rather than any of the others.32 It is not ad hoc, then, to maintain that these examples of prediction failure fall outside the proper domain of prediction by mental simulation unaided by empirical theory. A narrow circumscription of this domain is explicit in Heal's work: The kind of simulationism I would like to defend says that the only cases that the simulationist should confidently claim are those where (a) the starting point is an item or collection of items with content, (b) the outcome is a further item with content, and 31 J. Heal, 'Replication and Functionalism', in Folk Psychology, ed. Davies and Stone, p. 48; Harris, 'From Simulation to Folk Psychology: The Case for Development', p. 219. 32 So-called non-rational influences may have their effects in a very direct way - by-passing reasoning altogether - as, perhaps, in the case of the shoppers. But they may also work by making something that is not in fact a reason for acting in a certain way nevertheless appear to be a reason. We are not committing ourselves to any specific account of the various examples of prediction failure. Indeed, we are not even committed to the idea that the examples of prediction failure all involve non-rational influences. Perhaps subjects in the Langer-style experiment have good reasons for setting their sell-back prices, but those reasons are somehow obscured from subjects in the prediction experiment. In that case, a defender of simulation will, in the end, be right to use some version of the wrong inputs response. What we are pointing out is just that there is a different kind of response — in terms of non-rational influences - that is, in principle, available to the simulation theorist. See J. Heal, 'Simulation and Cognitive Penetrability', Mind and Language, 11 (1996), pp. 60-6. whisky drinking in reality. This might enable me to make a correct prediction, if whisky has the same effect on my reasoning from hypothesised contents as on the subject's reasoning from believed contents. However, it is important to note that the effects of my drinking the whisky will not be restricted to my simulation of the subject; my reasoning in my own person will be perturbed as well. This might be a disadvantage if I need to think carefully and accurately about how best to act towards the subject. 71

Martin Davies and Tony Stone (c) the latter content is rationally or intelligibly linked to that of the earlier item(s).33 But her proposal faces an objection. In many cases of prediction failure, the subjects about whom the predictions are made seem to depart in some way from right reasoning. But, in some other cases of equally flawed reasoning, correct prediction seems to be quite straightforward. In these latter cases, why do not the non-rational influences put the reasoning beyond the range of prediction by mental simulation? Consider an example discussed by Daniel Kahnemann and Amos Tversky.34 At a flying school, instructors adopt a policy of responding positively to good performance (such as successful execution of complex manoeuvres) and negatively to bad performance. When reviewing this policy, they note that pilots who do particularly well and are praised are likely to perform less well next time, while pilots who perform particularly badly and are criticized are likely to do better at their next attempt. The instructors conclude that, contrary to what psychologists tell us about positive reinforcement, rewarding good performance is not an effective training method. Most people find the flight instructors' conclusion to be quite unsurprising; it is just as they would predict. Yet the instructors' reasoning is flawed; it overlooks the phenomenon of regression towards the mean. (A pilot who has reached a certain level of competence and performs outstandingly well on one trial is likely to perform less well on the next trial, independently of the reaction of the instructor.) Is this not a problem, the objector asks, for the idea that the proper domain of prediction by mental simulation is the domain of rational linkages? It is clearly relevant to note, here, that the reasoning of the people who successfully predict the instructors' conclusion is flawed in just the same way as the reasoning of those instructors. But that point is liable to suggest, again, that there is something wrong with the proposal to circumscribe the proper domain of prediction by mental simulation in terms of the contrast between right reasoning and non-rational influences. What matters for mental simulation, the objector may say, is not rationality but similarity.35 Prediction by 33 Heal, 'Simulation and Cognitive Penetrability', p. 56. 34 D. Kahnemann and A. Tversky, 'On the Psychology of Prediction', in Judgment Under Uncertainty: Heuristics and Biases, ed. D. Kahnemann, P. Slovic and A. Tversky (Cambridge: Cambridge University Press, 1982), pp. 67—8. The example was used by Ned Block (in conversation) to make the objection under discussion here. Essentially the same objection against Heal's circumscribed version of simulation theory is pressed by Stich and Nichols, 'Cognitive Penetrability, Rationality and Restricted Simulation'. 35 Dan Sperber (in conversation) pressed the objection in this form. 72

Folk Psychology and Mental Simulation mental simulation will be successful just where a process that operates in imagination in the predictor is relevantly similar to the process operating in reality in the subject about whom the prediction is being made. By that account, probabilistic reasoning that overlooks regression towards the mean falls squarely within the proper domain of prediction by mental simulation, since the error is one that virtually everyone is disposed to make. Our view is that it is possible to defend Heal's circumscription of the proper domain of mental simulation by drawing on two ideas from section I: the idea of a normative theory and the idea of a distinctive epistemology of psychological prediction. First, as critical reasoners, we are each in possession of a normative theory of right reasoning (p. 61 above). We are also subject to non-rational influences and so we are all liable, on occasion, to reason in ways that are out of line with our normative principles. However, some departures from right reasoning may actually be sanctioned by our normative principles; that is, our normative theory may itself be flawed. Second, in virtue of our possession of a normative theory, we can arrive at judgements about what is the thing to think in a certain situation; and we can use those judgements, coupled with an assumption that the subject will think the thing that is the thing to think, in order to arrive at predictions. This predictive strategy can bestow a distinctive kind of epistemic warrant (p. 63 above). When a subject departs from right reasoning in a way that is out of line with our normative theory, this strategy will yield a wrong prediction, and will need to be augmented by empirical information about the non-rational influences at work on the subject. When the subject departs from right reasoning in a way that is sanctioned by our normative theory, in contrast, this strategy will yield a correct prediction. But it will be a prediction that does not constitute knowledge, since it is based on two false claims - that this is the thing to think, and that the subject will think the thing that is the thing to think - where the error in the second claim compensates for the error in the first. On this account, if the predictor and the subject share an incorrect normative theory then the predictions arrived at will be fortuitously, rather than knowledgeably, correct. The narrowly circumscribed domain of mental simulation is the domain of knowledgeable predictions that are arrived at by that epistemologically distinctive route. However, we should also consider what happens if the predictor learns about the importance of not ignoring regression towards the mean. For now the predictor will, provided that he or she is properly attentive, arrive at a correct judgment about what is the thing for the flight instructors, for example, to think. But the predictor may also recognize that, in this kind of case, most people are apt not 73

Martin Davies and Tony Stone to think the thing that is the thing to think. So, the predictor will take this empirical information into account when arriving at a prediction about the flight instructors. It may be that the informed predictor characterizes the way in which most people depart from right reasoning simply as the way in which he or she used to reason before learning about regression towards the mean. Perhaps, indeed, the predictor still finds it all too easy to slip back into that flawed pattern of reasoning. In that case, the predictor may suspend his or her recently acquired normative knowledge, and engage in a piece of not wholly critical reasoning, so as to arrive at a correct - and knowledgeably correct — prediction about the flight instructors. This is quite properly regarded as a piece of prediction by mental simulation. But the route that it takes is via an empirical judgement (this is what I used to think — how I used to reason) and an assumption of similarity (the flight instructors are relevantly like me as I used to be). So, while the distinctive epistemology of prediction that goes with the idea of a normative theory of right reasoning is of some importance, it would be wrong for us to suppose that all knowledgeable prediction by mental simulation exhibits that distinctive epistemology. Ill Simulation, Explanation and Understanding We began with the three strands of folk psychological practice - description, explanation, and prediction - but we have been almost exclusively concerned with folk psychological prediction, and with the basis question concerning that strand of our practice. In this final section, we turn briefly to folk psychological explanation. Explanation and generalizations Suppose that we want to explain the increase of pressure in our gas cylinder that results from an increase in temperature. The theory theory account of prediction (pp. 55—6 above) can readily be converted into an account of explanation by subsumption.36 The conjunction of an increase in temperature and an increase in pressure is subsumed under the temperature-pressure law. The truth of this generalization is not, however, something brute. The relatively superficial temperature-pressure law belongs, not only with a more 36 C. Hempel, Aspects of Scientific Explanation (New York: The Free Press, 1965), who provides the seminal account of the deductive-nomological model of explanation, regards the distinction between prediction and explanation as being merely pragmatic. 74

Folk Psychology and Mental Simulation general principle relating temperature, pressure and volume, but also with a body of empirical theory about the movement and energy of gas molecules, and about forces exerted on the walls of the cylinder. In terms of this theory, it is possible to give a mechanistic account of how it is that the relatively superficial law is true — of how the temperature-pressure connection is implemented. In short, according to the theory theory account, prediction and explanation go naturally together, and a predictor who knows not only the superficial generalization but also the broader body of theory is able to achieve an explanatory understanding of the predicted increase in pressure. In the folk psychological case, too, the theory theory account of prediction goes along with an account of explanation. Knowledge of a body of psychological theory provides the resources for explanations that work by subsuming events under causal generalizations. There may be variations on this theme. Some theory theorists will regard knowledge of relatively superficial psychological generalizations as the visible tip of an iceberg of more elaborated tacit knowledge, while others will commit themselves only to knowledge in the ordinary sense of the term. Some theory theorists will regard cognitive scientific theories about subpersonal level information processing machinery as offering deeper explanations of psychological matters, while others will hold hard to the personal level. But the general picture is clear. Given that familiar picture of explanation by subsumption, it may seem obvious that the basis question about the explanation strand of folk psychological practice has a ready answer in terms of the theory theory, but cannot be answered in terms of the simulation alternative. Explanation requires generalizations; but mental simulation is supposed not to depend on antecedent knowledge of psychological generalizations. However, what seems obvious is not quite correct. Consider again the case of the gas cylinders. We have noted already (p. 62 above) the similarity between prediction by simulation in reality and the use of experiments to license generalizations. So, gas cylinder simulation, carried out without antecedent knowledge about the behaviour of gases, could yield knowledge of generalizations that could, in turn, be used in subsumptive explanations. Gas cylinder simulation in reality would naturally be called black box simulation; we simply give the simulation device (gas cylinder B) a temperature as input and receive back from it a pressure as output. Consistently with that description, the experimentally licensed generalizations would be superficial and, because of the lack of explanatory depth, the simulation would scarcely provide any explanatory understanding of the predicted increase in pressure. But still, the basic point remains. Simulation, conceived as experi75

Martin Davies and Tony Stone ment, may yield knowledge of generalizations under which events can be subsumed. We could call this simulation-driven theory. So also, in the folk psychological case, simulation can be regarded as experiment and may yield knowledge of empirical generalizations. This is particularly clear in the case of simulation in reality. By drinking pints of whisky, looking at pairs of lines, dangling on ropes and drawing inferences, I may not only arrive at predictions about another person C (pp. 57—62 above). I may also, by induction from these bouts of simulation considered as experiments, arrive at empirical generalizations under which events in the mental life of C may be subsumed. This is also true - though over a circumscribed domain - for mental simulation in imagination. When I simulate C's reasoning in imagination, I draw on a normative, rather than an empirical, theory about reasoning. But I may arrive at empirical generalizations by induction on the results of simulation in imagination; and, to the extent that mental simulation may yield correct predictions, it may also yield correct generalizations.37 Simulation and understanding If explanation is conceived as subsumption under generalizations, then the debate initiated by the basis question about the explanation strand of folk psychological practice will take a course that is essentially parallel to the debate over the prediction strand. But in fact, many advocates of the simulation alternative would defend the idea that there is a distinctive - not straightforwardly subsumptive - kind of explanation involved in folk psychological understanding. Thus, for example, Heal says: The difference between psychological explanation and explanation in the natural sciences is that in giving a psychological explanation we render the thought or behaviour of the other intelligible, we exhibit them as having some point, some reason to be cited in their defence.38 37 We are committed to the possibility that there may be both normative and empirical principles cast in very similar terms. Both kinds of principle would make use of ceteris paribus clauses; but those clauses would be interpreted differently in the two cases. 38 Heal, 'Replication and Functionalism', p. 52. See also, J. McDowell, 'Functionalism and Anomalous Monism', in Actions and Events: Perspectives on the Philosophy of Donald Davidson, ed. E. LePore and B. McLaughlin (Oxford: Basil Blackwell, 1985), p. 389: '[T]he concepts of the propositional attitudes have their proper home in explanations of a special sort: explanations in which things are made intelligible by being revealed to be, or to approximate to being, as they rationally ought to be.' 76

Folk Psychology and Mental Simulation This kind of normative explanation reveals what someone thought or did as having been the rational thing to think or do, or the thing that it made sense to think or do, given the circumstances and the agent's beliefs and preferences. Clearly, explanation in this style fits together with our account of prediction by mental simulation (in a circumscribed domain) in somewhat the way that explanation by subsumption is the natural companion of prediction that draws on empirical generalizations. But we do not get an adequate view of the distinctive kind of psychological understanding that might be provided by mental simulation if we focus only on the normative aspect. For, as we have noted (p. 61 above), it is possible to deploy a normative theory about an activity in which one does not, oneself, engage. What mental simulation promises is a kind of understanding that is not only normative but also first personal.39 We see the combination of these two aspects most vividly in the simulation of reasoning in imagination; and the idea that mental simulation can provide a distinctive kind of understanding of another person's reasoning is strikingly similar to R. G. Collingwood's claim that historical understanding is to be achieved by the re-enactment of the historical character's thought: But how does the historian discern the thoughts which he is trying to discover? There is only one way in which it can be done: by rethinking them in his own mind. The historian of philosophy, reading Plato, is trying to know what Plato thought, when he expressed himself in certain words. The only way in which he can do this is by thinking it for himself. This, in fact, is what we mean when we speak of 'understanding' the words. So the historian of politics or warfare, presented with an account of certain actions done by Julius Caesar, tries to understand these actions, that is, to discover what thoughts in Caesar's mind determined him to do them. This implies envisaging for himself the situation in which Caesar stood, and thinking for himself what Caesar thought about the situation and the possible ways of dealing with it. The 39 Thus, for example, Gordon, 'Simulation Without Introspection or Inference from Me to You', p. 56 quotes Kant (Critique of Pure Reason, A353) approvingly: 'It is obvious that, if I wish to represent to myself a thinking being, I must put myself in his place, and thus substitute, as it were, my own subject for the object I am seeking to consider (which does not occur in any other kind of investigation).' For an illuminating discussion of issues not far removed from those of the present section, see Moran, 'Interpretation Theory and the First Person'. 77

Martin Davies and Tony Stone history of thought ... is the re-enactment of past thought in the historian's own mind.40 Indeed, just as the domain of prediction - and correlatively of understanding - by mental simulation may be narrowly circumscribed (pp. 71—2 above), so also understanding by re-enactment may seem to be restricted to right thinking. This would be a severe limitation on historical understanding. Patrick Gardiner considers this objection to Collingwood in a recent paper: [I]t may ... be objected that the re-enactment conception of understanding remains unrealistically restrictive in the amount it seems to exclude from the historian's proper scope. However scrupulous the care taken to judge an action from the agent's own standpoint, there can be no a priori guarantee that the reasoning ascribable to him will turn out to have been cogent or sound; as Francis Bacon once remarked, 'it is a great mistake to suppose men too rational'. It is always conceivable in principle, and it is surely often the case in practice, that there is a lack of coincidence between the conclusions people actually draw on the basis of their beliefs and purposes and the conclusions that rationally they should have drawn. Thus in history as elsewhere people may engage in faulty practical thinking, whether because of such things as haste or unimaginativeness or as a result of underlying emotional factors that sway or distort their judgement. But when that happens - the objection may continue - it does not follow that their behaviour is unintelligible in terms of reasons, only that the reasons are liable to be poor or inadequate ones.41 Gardiner's response to this objection is to note that 'Collingwood would be less vulnerable to some of the criticisms brought against him on the present score if his conception [of re-thinking] were interpreted in a more flexible manner.' On such an interpretation, re-enactment of thought would cover not only right reasoning but also, for example, 'empathetically appreciating how an agent could w R. G. Collingwood, 'Human Nature and Human History', in The Idea of History (Oxford: Oxford University Press, Revised Edition 1992), p. 215. As is quite widely remarked, the simulation approach to psychological understanding has marked affinities with the hermeneutic tradition of Vico, Herder, Dilthey, Weber and Croce, as well as Collingwood. See Verstehen and Humane Understanding: Royal Institute of Philosophy Supplement 41, ed. A. O'Hear (Cambridge: Cambridge University Press, 1997). 41 P. Gardiner, 'Collingwood and Historical Understanding', in Verstehen and Humane Understanding, ed. O'Hear, pp. 117-18. 78

Folk Psychology and Mental Simulation have been tempted or misled into accepting a particular practical conclusion without recognising the faultiness of the reasoning involved'.42 Is it possible for a friend of mental simulation to expand the domain of understanding by simulation in a similar way? In the case of prediction by simulation (p. 62 above), we saw that there could be an intrusion of empirical theory without the prediction strategy coming to owe everything to theory and nothing to mental simulation. The possibility that we mentioned there was that the empirical information drawn on might be information about how particular influences (such as drinking a pint of whisky) lead to departures from right reasoning. However, there is no guarantee that, if we modulate the re-enactment of thought in the light of empirical information, then the resulting first person narrative (in imagination) will be one that we find intelligible. Thus, for example, Simon Blackburn considers the case of deliberating about what is the thing to do if one is a subject in Milgram's obedience experiment, and then taking account of the empirical evidence about what subjects actually tend to do. The simulator can modify his or her narrative in the light of the empirical information. But, 'this need have no tendency to make the behaviour of Milgram's subjects intelligible. I might still feel quite baffled, both by them, and if I am like them, by me.'43 An intrusion of empirical theory may, then, bring with it a loss of intelligibility. But it would not be right to conclude that there is no prospect of a more flexible conception of the domain of understanding by simulation. Consider, for example, the predictor who now knows about regression towards the mean but who still finds it all too easy to slip back into flawed reasoning (p. 74 above). This predictor will surely not be baffled by the reasoning of the flight instructors. Their reasoning does not measure up to the informed predictor's normative theory; but their first person narrative is nevertheless one that the predictor will find intelligible. There are other cases, too, in which it may be possible, without simply being driven by an empirical theory, to re-enact thinking that departs from right reasoning. Let us return to one of our earlier examples. I want to predict how C will feel and what he will decide to do if he is suspended over a cliff on a rope and he cannot find a foothold and his hands are starting to slip (example (iii) on p. 57 above). Seized by fear or panic, C may not think or do the best thing, the most rational thing. Yet, by simulating C's situation in imagination, I might reach a correct prediction about C without drawing on any empirical theory about how people dangling over 42 Ibid., p. 118. 43 Blackburn, 'Theory, Observation and Drama', p. 283. 79

Martin Davies and Tony Stone cliffs on ropes tend to think. For imagining the situation might be enough to produce in me physiological and emotional responses that perturb my reasoning in imagination in just the way that C's reasoning in reality is perturbed.441 might arrive at a correct prediction about C; and if I regard the simulation exercise as an experiment I might arrive by induction at some generalizations about how people think and act in dangerous situations.45 But there is something more. By re-enacting C's desperate thinking, struggling to maintain a grip, deciding to make another attempt to find a foothold - all in imagination, of course - I surely gain some kind of empathetic understanding of the thoughts, feelings and decisions that I predict. This is not a case of theory-driven simulation; and it is not black box simulation either. It is simulation that, in Gordon's phrase, 'essentially engages [my] own practical and emotional responses'.46 There is an alternative way in which I can gain a measure of first personal understanding of C's thoughts, feelings and decisions — a way that does not require actual physiological and emotional responses in me at the moment of understanding. I may take into account my own remembered similar experiences.47 In this case, I draw on stored information — about how I felt, physically and emotionally, and about how this affected my thinking and decision taking - and I use this information to help me imagine what it is like to be C. (I may also draw on memories of imaginings in which I was fully physiologically and emotionally engaged.) Producing a correct narrative about another person is not always sufficient for finding what that person thinks and does to be intelligible. But it is plausible that in some cases we can make sense of what someone thinks and does by drawing on memory to help us imagine being in the 44 These responses may have consequences, not only for my reasoning within the scope of my simulation of C, but also for my reasoning in my own person. Cf. footnote 30 above. 45 If my prediction about C's thoughts and actions is to count as knowledge then it should not depend on the flawed normative judgement that this is the thing to think, or to do, in these circumstances. In this case, knowledgeable prediction seems to require some recognition of the fact that one's reasoning is indeed being perturbed. 46 R. M. Gordon, 'The Simulation Theory: Objections and Misconceptions', in Folk Psychology, ed. Davies and Stone, p. 103. Since understanding is a kind of knowledge, there will once again be a need for me not to be wholly in the grip of the re-enactment (cf. footnote 45 above). 47 Nichols et al., 'Varieties of Off-Line Simulation', pp. 59-67, discuss empathy and in particular the role of memory in empathetic emotion. What we are concerned with here, however, is remembered emotion, not emotion aroused by memory. 80

Folk Psychology and Mental Simulation other person's situation - indeed, to help us imagine being that person. This is an intrusion of empirical theory, given the inclusive way in which the term 'theory' has been used. But it does not obstruct first personal understanding, and it does not move us back towards explanation by subsumption. Conclusion In the first two sections of this paper we were concerned with the prediction strand of folk psychological practice. The theory theory and the simulation alternative agree about what folk psychological prediction is; but they disagree about its basis. According to the theory theory, the predictor draws on a body of information about psychological matters. According to the simulation alternative, prediction is sometimes possible by simulation in imagination without the aid of empirical psychological theory. However, the domain of prediction by mental simulation - particularly if the epistemology is to be different from the epistemology of empirical theory — is rather closely circumscribed: it is the domain of reason. When we turn to the explanation strand of folk psychological practice, we find that the contours of the debate are very different. For there is a disagreement about what folk psychological explanation is. According to the theory theory, it is explanation by subsumption under causal generalizations. So, if the basis of the explanation strand of folk psychological practice is to be knowledge of a psychological theory, then that theory must contain generalizations of the right kind - objective, counterfactual supporting - to figure in subsumptive explanations. It is a theory in a more restricted sense. According to the simulation alternative, folk psychological explanation is normative and first personal; it is a matter of finding the other person's life intelligible 'from the inside'.48 This is an imaginative project; and understanding involves not only reasoning in imagination but also emotion and memory. What is remembered is, of course, information about psychological matters. So, if psychological understanding is to range beyond the domain of reason then, even by the lights of the simulation account, it must draw on psychological theory. But this does not constitute a victory for the theory theory, because the psychological theory on which simulation and understanding draw is theory in the inclusive sense, but not in the restricted sense that is relevant to the theory theory's account of psychological explanation. If we do not distinguish the inclusive sense of 'theory' which is 48 See Jane Heal's paper in this volume. ^- 81

Martin Davies and Tony Stone relevant to the debate about prediction from the restricted sense of 'theory' which is relevant to the debate about explanation, then we may obscure from ourselves the role of empathy and emotion in commonsense psychological understanding. 82

Understanding Other Minds from the Inside JANE HEAL Can we understand other minds 'from the inside'? What would this mean? There is an attraction which many have felt in the idea that creatures with minds, people (and perhaps animals), invite a kind of understanding which inanimate objects such as rocks, plants and machines, do not invite and that it is appropriate to seek to understand them 'from the inside'. What I hope to do in this paper is to introduce and defend one version of the so-called 'simulation' approach to our grasp and use of psychological concepts, a version which gives central importance to the idea of shared rationality, and in so doing to tease out and defend one strand in the complex of ideas which finds expression in this mysterious phrase.1 Let us here recap the salient ideas of the simulation approach.2 Simulationism is to be contrasted with another approach to philosophy of mind which has, at least among Anglo-American analytic philosophers, been the dominant one of the last decades and which has also been an important influence on psychologists and cognitive scientists. We may call this familiar alternative the theory theory. The version best known to philosophers is functionalism in philosophy of mind. This says that to grasp psychological notions is to grasp that there are certain inner states of persons which are typically caused by such and such external events, which interact among themselves to cause further inner states and events, and which finally combine to cause behaviour.3 To possess the concept of some particular mental state is to grasp the particular causal-explanatory role associated with that state. When we use our understanding of psy1 Talk of persons 'having a point of view' and of there being such a thing as 'what it is like to be that person' are also parts of the same set of ideas. But I would like to stress that the whole issue of the existence of 'qualia' is not touched on at all in what follows. 2 For more on this see M. Davies, in this volume and also the two collections Folk Psychology and Mental Simulation, ed. M. Davies and T. Stone (Oxford: Blackwell, 1995) and Theories of Theories of Mind, ed. P. Carruthers and P. K. Smith (Cambridge: Cambridge University Press, 1996). 3 For a collection which includes many of the classic papers arguing for this view see Readings in the Philosophy of Psychology, vol. 1, ed. N. Block (Cambridge, MA. Harvard University Press, 1980). 83

Jane Heal chological notions, for example in predicting what another will think or do, we deploy this theoretical knowledge. There are many ways of spelling out this general idea, depending on what view we take of how we acquire and represent this supposed theoretical knowledge - whether it is innate or learnt, whether explicitly or tacitly known etc. But let us leave all these issues on one side, just noting however that the theory theory does not seem at all hospitable to the 'from the inside' idea. Indeed part of its motivation is to find an account of the psychological which is naturalistic, i.e. which does away with certain deeply suspect forms of dualism and sees human beings as part of the natural order. Theory theory does this precisely by claiming the similarity of psychological concepts to non-psychological concepts such as those of natural science, presenting the former merely as particular complex and interesting cases of the general style of thought invoked in the latter. So much for theory theory. Now for a thumbnail sketch of its rival, the simulation approach. This is by no means an entirely new idea. A version of it goes back to Vico in the early eighteenth century; it gets a passing mention in Kant; it is associated with Dilthey and is forcefully defended by Collingwood.4 And under the name 'Verstehen' one broadly simulation-style approach is familiar, and has been extensively debated, in the philosophy of history and social science. But in the last ten years the idea has been revived in the context of psychology and philosophy of mind. And here it provides a new perspective on a great number of familiar topics — for example the nature of imagination, the differences between practical and theoretical reasoning, the nature of emotion — as well as initiating an interesting body of empirical work in psychology and a suggesting new models in cognitive science.5 I shall touch on only a very small part of this. A way of putting the central idea of the simulation approach is this. When we think about another's thoughts or actions we somehow ingeniously exploit the fact that we ourselves are or have minds. What we do is to make our own mind in some way like the mind of the one we seek to predict or understand. We simulate his or her thoughts, we recreate in ourselves some parallel to his or her 4 I. Berlin, Vico and Herder (London: Hogarth Press, 1976); I. Kant, Critique of Pure Reason, trans. Norman Kemp Smith (London: Macmillan, 1953), p. 336; W. Dilthey, Selected Writings, H. P. Rickman (ed.) (Cambridge: Cambridge University Press, 1976), R. G. Collingwood, The Idea of History, (Oxford: Oxford University Press, 1946), esp. pp. 282-302. 5 These themes are all illustrated in the collections mentioned above in footnote 2. 84

Understanding Other Minds from the Inside thought processes. Many simulationists further articulate this by talking of my having 'pretend beliefs' or 'pretend desires' and of my 'inference mechanisms' being run 'off-line'. But others (and here I include myself) would prefer to use more everyday vocabulary and to talk of my using my imagination and of my thereby entertaining the same thoughts and making the same inferences as the other. We shall come to the significant differences here in due course. I shall not here consider in detail the reasons for preferring a simulation approach to a theory theory one. Let me just indicate one central and immediately apparent advantage of simulationism. It is this. Others' thoughts are very varied and numerous and interact with each other in countless different ways. The remarkable thing is how successfully we deal with this, correctly adjusting our expectations of others' thoughts, feelings and actions in an immense variety of circumstances. Clearly any theory adequate to systematize our competence here would itself be immensely complex. But simulation can explain our competence without crediting us with knowledge of any such vast, and very probably unwieldy, body of information. Rather in thinking about other's thoughts, in order for example to predict their intentions, we harness our own cognitive apparatus and make it work in parallel with that of the other and then use the result we arrive at to ground our prediction. And for this to occur all that is required is, first, that we have cognitive apparatus which is sufficiently similar to that of the other to produce usefully similar results and, secondly, that we can make it work in a parallel way. It is not required that we have some representation of the apparatus itself or its workings. We do not need to possess a 'know that' about the processes of thinking, what thoughts lead to what others and so forth, provided that we can harness relevantly our own 'know how' of doing the thinking itself and can thus follow through in ourselves the same train of thought as the other has pursued. The economy of the proposal is striking. The phrase 'thinking about others' thoughts' covers a great variety of importantly different kinds of reflection which we now need to distinguish. There is something very importantly right in the overall picture painted by functionalism, namely the facts which it highlights that psychological states may be caused by events in the world, that such states interact with each other to give rise to further states and that they may give rise to bodily behaviour. This gives us a useful framework for considering the different sorts of issue which may arise for me concerning another's thoughts. 1. I may wonder what effect the circumstances around her will have on the psychological states of another person. (E.g. She is being whirled round in a fairground ride: will she feel sick? There is 85

Jane Heal a disturbance going on in the corner; will she notice it?) Let us call the connections I focus on here 'world-mind links'. 2. I may wonder what further thoughts she will have, given some thoughts about which I already know. (E.g. She believes thus and so about the cash flow of our firm: will that lead her to think that we are about to go bankrupt? She endorses these and those principles: what decision will she reach in this particular case?) Let us call these 'mind-mind' or 'intrapsychic links'. 3. Thirdly, I may wonder what behaviour, i.e. what actual bodily movements, she will exhibit, given her thoughts. (E.g. She will hear a balloon popped behind her: will she jump? She intends to smash her opponent's ball away to the side line: will she succeed in jumping high enough to get the needed angle?) Let us call these 'mind—world' links. But let us also note that there is a fourth thing I may be doing under the general heading 'thinking about others' thoughts'. 4. I may try to work back from the behaviour she has produced to a view about the psychological states from which the behaviour arose. (E.g. She pulled a funny face: was she really amused? She said such and such: was she annoyed?) These are four extremely different contexts in which psychological concepts are used: and competence at each may call upon different aspects of the skill which is grasp of these concepts. We should beware of lumping them all together and supposing that a philosophical account of our competence with such concepts, whether simulationist or otherwise, should say the same about each. And the claim I want to make about simulationism is that it is particularly at home, its strengths and plausibility particularly apparent, in the second of the listed circumstances, i.e. in an account of our grip of mind-mind or intrapsychic links. And this will be mainly what I shall discuss below. I do not think that simulationism has anything distinctive to say about our ability to answer the third sort of question. It does have distinctive ideas to contribute on the first and fourth, but I shall not discuss them in detail here. However, a few remarks about the fourth may help to ward off some misunderstandings. It is important to note how, on any view, the fourth context — that of interpreting and explaining behaviour - must be a very different matter from the others. All philosophers, whatever their theory of mind, acknowledge that many alternative explanations of the same behaviour are possible. For example, even if we can identify something with fair confidence as an intentional raising of the arm, when we move to identify the purposes behind the raising, and to the feelings, goals, beliefs which in turn lie behind that purpose, it is clear 86

Understanding Other Minds from the Inside that many accounts could be given. Even on a functionalist or theory theory view there is no such thing as just 'applying the theory' in some fixed and algorithmic way to derive an interpretation. In the other cases (those falling under 1—3 above), as conceived by functionalism, there is such a fixed and algorithmic procedure; if one has sufficient information about prior conditions then one just has to identify the bits of the theory which deal with those conditions and apply them. One's prediction may be hedged because one knows that further information may reveal the case as more complex than it at first appears or as requiring adjustments in the light of subsidiary theoretical principles. But (these kinds of complications aside) forward-moving theory-invoking prediction is quite a different matter from backward-looking theory-invoking explanation. An account of how we do the latter cannot just call on 'our knowledge of the theory' but must also tell some story about how we generate a range of possible explanations compatible with the theory and how we assess them. The same general kind of point needs to be made in connection with simulationism. Even if we accept a simulationist account of how prediction about others' thoughts or behaviour are arrived at (e.g. in cases of type 2 above), this does not of itself tell us how backward-looking interpretations and explanations are arrived at. So we should beware of the idea that simulationism is the proposal that mere awareness of another person - of his or her circumstances and behaviour - automatically produces in the observer, via some natural sensitivity, a simulation of the other's mental state. Simulationism ought not to be the claim that we have some kind of quick route to knowledge of other minds, or that we can empathetically 'tune in' to others, or anything of this kind. Perhaps such a thing exists in a few basic cases. But patently other people are often difficult to understand; often we know that we are ignorant of their thoughts and feelings or we have little confidence in our conjectures about what they may be. Simulationism is not the promise of some easy answer to these difficulties.6 6 Some simulation theories do postulate a natural tendency to 'catch' others' mental states. For example in normal infants we find very early a disposition to attend to what others attend to, to be frightened if adults in their company are frightened and the like. It is extremely plausible that we do have some such basic pattern of response and that this is central to our ability to understand other minds. The point I am emphasizing, however, is that this does not take us very far. How we build on it to arrive at interpretations in the more complex cases is something about which simulationists have some proposals but not fully worked out ideas. See the papers by Gordon, Goldman and Heal in the Davies and Stone collections. 87

Jane Heal Let us now turn to a consideration of the distinctively simulationist story about mind-mind links, i.e. about how I might come to some prediction or further belief about another's thought on the basis of knowledge of some subset of her thoughts. To take a very schematic case, suppose the other believes pl-pn and is interested in whether or not q. How might I work out her likely opinion on whether or not q? Theory theory of course says that I, so to speak, look up what the theory tells me about what is the likely upshot of the combination of believing pl-pn with an interest in whether q; I look up the relevant axioms about beliefs and interests of that kind and apply them to this particular case; so it is by applying my knowledge about thoughts and their effects that I work out what to expect. Simulationism will say something different. But there are two contrasted ways in which the simulationist story can be told. One story starts with a picture of the mind which is very congenial to the theory theorist and derived to a considerable extent from cognitive science. The mind, on this picture, consists of a number of subsystems which perform various functions. For example there are two stores in which beliefs and desires are kept; there are various processors which produce beliefs and put them in the belief store; these include a sensory analysing system, which takes sensory inputs and transforms them into beliefs; they include also some inference mechanisms, which take beliefs and derive other beliefs from them; there is also a practical reasoning system which takes beliefs and desires as input and produces intentions as output; and so forth. Each processing system is designed to accept certain kinds of input; receiving input of the appropriate kind causes it to go through its distinctive evolutions and to produce output of distinctive kinds. These inputs and outputs - sensory states, beliefs, desires, intentions and so forth — are realized or coded in vehicles which are, in fact, brain states, for examples neuronal patterns described at some suitable level of abstraction. And, on this picture, what really drives the evolution of the inference mechanisms, practical reasoning system and the like are the intrinsic properties of the vehicles, the brain states or neuronal patterns, which are the beliefs, desires etc. Given this view of what goes on in the mind, simulationism is now spelt out in the following way. Suppose, as in the schematic example above, that I wish to work out what O, the other, is likely to think about whether or not q, given that she believes pl-pn, when I myself do not share her beliefs. What I do first is construct some 'pretend' beliefs that pl-pn. These are items which do not, in my mental architecture, play the role of beliefs; they do not come from my belief store. Nevertheless they are like beliefs in the nature of 88

Understanding Other Minds from the Inside the vehicle in which they are coded. I now take my inference mechanisms 'off-line' - that is I detach them from their usual links with my belief store. I feed in the pretend beliefs I have constructed, at the same time making some adjustment to the mechanisms to make them search for q-relevant consequences; I then wait to see what the mechanisms produce as output. If they output a pretend belief that q then I attribute to O the belief that q. Of course I do not do all this consciously. Nevertheless this is what is going on at the level of the operations of my subpersonal cognitive machinery. Simulation theory presented in these terms is conceived of as an empirical hypothesis. Those who articulate it like this suppose that it has empirical consequences different from those of the theory theory, that we can already see what these consequences are and that we can set about testing them. But I would like to suggest that there are considerable problems with this conceptualization of the issue. Consider first the fact that we do not have any well-backed-up and detailed view about what kind of functional 'systems' are to be found in the brain or of how the various kinds of mental state and process recognized in commonsense are in fact implemented at the sort of level envisaged. Many kinds of architecture are imaginable other than the one sketched above. For example, is it necessary to distinguish between theoretical and practical reasoning, in the way proposed? To insist that we should is to make substantive and controversial philosophical assumptions about the relation of belief, desire and value. Another, and for our purposes more important, question is whether we have to take it that 'inference mechanisms' operate on beliefs, i.e. the whole complex state including both content and the attitude to it. A different articulation would take it that 'inference mechanisms' operated on mental representations minus their attitude determiners - on, so to speak, 'thought radicals'. There is surely some case for thinking that we can reason with representations which we do not believe. How do we explain what we are doing in arguing by reductio ad absurdum or reasoning hypothetically if every piece of reasoning needs a belief as starting point? But if we thus reconceptualize the 'inference mechanisms' as operating on thought radicals, then simulationism, formulated in terms which presuppose the existence of inference mechanisms operating on beliefs, turns out to involve a false picture of the mind and so to be worthless. Consider also the fact that (even supposing that the original sketch of the architecture of our cognitive machinery is the right one) we have very little idea of what would be involved, neurophysiologically or functionally, in taking a system 'off-line'? We do not know what features of operation would remain the same and what 89

Jane Heal would be different. It may be that we are seduced by the image of a machine made of cogs, levers and pistons, where we can make sense of things like disengaging the gears, detaching the drive belt and so forth. But is the brain in any sense like that? Who knows? I am not here seeking to make difficulties for whole project of seeking to understand the mind by breaking up its overall operation into various different functions and looking for the biological structures and processes which subserve those functions. Good luck to the cognitive scientists, psychologists, neuroanatomists and so forth who grapple with these fascinating and difficult tasks. The issue is rather that we do not yet have enough grip on how that project might actually work out in detail to have any confidence that we are working in terms of the right architecture when we talk of 'pretend beliefs' and 'off-line running'. Nor do we have enough understanding of how that proposal could be implemented, to see what the talk of 'off-line running' would actually amount to. The latter point means that we do not really know how to test simulationism, regarded as this empirical hypothesis. The former means that simulationism, if it is articulated in terms of this particular architecture, is made hostage to future discoveries in brain science and might, given unfavourable developments there, turn out to be a total mistake. But it seems that the simulation hypothesis has considerable plausibility quite independent of any empirical developments in brain science, a plausibility noted by Kant, Dilthey, Collingwood and others, who were not at all in the business of speculative cognitive science or high-level neuroanatomy. This suggests that there ought to be a reading of the simulation proposal in which it is articulated in quite different terms, terms which place it much nearer the a priori end of the spectrum and on which it is effectively insulated from how things turn out in neuroanatomy and the like. Let us also note at this point that the idea of 'off-line' use of inference mechanisms and the like does not offer any particularly congenial setting for the idea of 'understanding from the inside'. The attraction of the idiom is not at all illuminated by the simulationist story as spelled out above. If I wish to predict how another person will react to some new supposed cholesterol-lowering medication I may try to find out its effects on her by taking a dose myself and observing the results. Or (indulging in some science fiction) I might be able to unhook a part of my circulatory system and run an experiment on that. In either case I would 'simulate' in myself the operation of the drug on her. But the fact that it is a bit of my own bodily apparatus which is being run in experimental fashion gives no special insight 'from the inside' into the workings of the drug. And we have been told nothing which entitles us to 90

Understanding Other Minds from the Inside think the case of the mind - i.e. the brain - to be any different. But the idea of simulation did seem to have some resonance with the idea of 'understanding from the inside'. So again, we are led to the idea that there may be an alternative way of conceptualizing the idea. So now let me sketch such an alternative. Consider a normal person who is capable of having beliefs about a certain subject matter, i.e. of forming them appropriately and reasoning from them appropriately, among other things. Let us take Charles as an example; he is an investment expert and can form the belief that the base rate will rise on seeing evidence that it will and can make sensible inferences from this, for example to a fall in the value of shares. Now we take it entirely for granted that if Charles is capable of doing these things then he is also capable of reasoning hypothetically about what would happen if base rate were to rise. It is difficult to make any sense of the opposite supposition. Remember that Charles is a normal human being, so that in dealing with most subject matters, cups of tea, rain, buying a house and so forth, Charles can cope with both actual and hypothetical. Suppose that we now try to graft on the supposition that, for example, when faced with sentences beginning 'Suppose that base rate were to rise...' Charles goes deaf or berserk or in some other way just fails to cope, although he does respond normally to the straight assertion 'Base rate has risen'. Or suppose we try to add on the idea that Charles can appreciate the need for contingency planning in connection with most kinds of events but never seems to indulge in any kind of contingency planning about base rate rises, although, remember, he copes with great competence when they actually occur. Can we really fill in the details of such a story in a coherent way? I do not say that it is provably impossible that we should do so. We are familiar with the extremely bizarre and disconcerting way in which what are normally treated as unitary abilities can unravel in cases of brain damage and disease (agnosias, aphasias and the like). But the interesting point is that such cases are extremely rare and that our ordinary psychological concepts do not allow for them. Our ordinary concepts are, quite properly I suggest, tailored to the outward, behavioural contours of the normal case, to the kind of successful performances and achievements one can regularly expect of persons. They are not tailored to respect or record the structure of the machinery which realizes these abilities. In our thinking about other people one fundamental question we can and often do ask is what subject matters they are familiar with, i.e. roughly what concepts they possess and in what kind of detail. Do they understand about tables and chairs? About royalty? About 91

Jane Heal snow? About car engines? About income tax? And how well do they understand about each? If a person is familiar with a subject matter and understands it to some roughly indicated level then we take it for granted that this ability to think about the subject matter will manifest itself just as much in coming to counterfactual beliefs, in considering possibilities and their upshots, as it does in forming and reasoning from categorical beliefs. The ability will show itself also in desires, intentions, emotions, dreams and fantasies. Competence in thinking about a subject matter is a multifaceted ability. It is an error, a distortion, of our central psychological notions, to think of concept possession as something which shows up only or centrally in the formation of categorical beliefs. Rather, belief formation is just one facet of an ability which naturally manifests itself also in other kinds of thinking. Note here a further important point, implicit in what has been said already but needing emphasis. A parallelism between certain psychological processes is already presupposed in the everyday conception we have been spelling out, namely a parallelism between, on the one hand, the inferences a person makes with categorical beliefs in virtue of his or her grasp of a subject matter and, on the other, the counterfactual conditional beliefs he or she would form as a result of making suppositions and the like. So Charles infers from 'Base rate will rise' to 'Share prices will fall.' But it is also the case that if he wonders 'What if base rate were to rise?' then he will come to the conditional belief 'If base rate were to rise then share prices would fall.' This parallelism must stay more or less in place on pain of our losing our right to describe the content of Charles' wondering as 'What if base rate were to rise?' It cannot be base rate and its possible rise that he is wondering about if he does not at this point come up with the same idea, to figure in the consequent of his conditional belief, as he would come up with in straight belief to belief inference. The fact that this parallelism exists is what makes viable the whole conceptual structure we use in talking of others' thoughts, plans, desires, reasonings etc. It is the idea of the multifaceted ability which is, in effect, the idea that the same content can be identified as playing a role embedded within other contents and as the object of various different attitudes. Someone might here offer a hypothesis about how it is that we have such a multifaceted ability, i.e. about the nature of the systems or devices in which the ability is realised. Perhaps what goes on when I wonder 'What if p?' is that I take some inference mechanisms 'off-line' and feed in a pretend belief that p?7 But to pursue 7 See, e.g. the paper by Nichols et al. In Theories of Theories of Mind, ed. Carruthers and Smith. 92

Understanding Other Minds from the Inside the line of thought I am proposing we do not need to get embroiled in issues like this at all. The parallelism between thinking about what is taken to be actual (having a belief) and thinking about what is taken to be merely possible (wondering, hypothesizing, imagining and the like) exists, whatever its underpinnings turn out to be. And we are entitled to invoke it in our account of thought about other minds. So now back again to simulation and other minds. We can now present the simulation hypothesis like this; ability to think about another's thoughts, e.g. to reason from the existence of those thoughts to conclusions about the existence of further thoughts, is an extension or redeployment of ability to think about the subject matter of the other's thoughts. How does this work? Let us take the following way of spelling things out. Let us revert to our schematic example in which the other believes that pl-pn and is interested in whether or not q. I know this and I am interested in whether or not she comes to believe that q. What she will do is wonder 'In the light of pl-pn is it the case that q?' i.e. she will direct her thought to answering the question whether q, having in mind the evidence that pl-pn. If the propositions that pl-pn imply that q, and she comes to be aware of them as so doing, then she will come to believe that q, taking this to be a belief to which she is entitled, in the light of the facts (as she sees them) that pl-pn. What will I do? If I share her beliefs I may, in effect, pose myself just the same question, viz. 'In the light of the facts that pi - pn is it the case that q?' But if I do not share her beliefs then the question I should address is, rather, 'If it were that pl-pn, would it be that q?' But in either case the other person and I share a central aim, namely trying to get a sense of the relations of implication or otherwise between pi—pn and q. And we carry out this aim by exercising our ability to think about the subject matter of pl-pn and q. And if it comes to me to seem that if it were that pl-pn then it would be that q then I attribute to the other belief that q. Let us reflect now on the concepts implicit in the story I have just sketched. I have spoken of us as having 'a sense of some thoughts as implying or being implied by others. Much recent philosophy, influenced perhaps by cognitive science models, tacitly operates with a picture of the progress of thoughts through time, as when a person is reasoning and reflecting, as a matter of there being one thought (perhaps quite a complex one) occupying the conscious mind at one instant and of its being entirely replaced by another thought at the succeeding moment. So, for a schematic example, at first I think 'p, p—>q, q?' and then this complex thought is swept away and replaced by 'q'. But this is surely a distortion of our expe93

Jane Heal rience as thinkers. A slightly more accurate narrative is one in which I first think 'p, p—>q, q?" and then next think 'Well, clearly q, since p and p—>q'. That is, I judge that q in the light of a sense of it as following from p and p.—>q. Relatedly I take this latter belief of mine to be justified by my beliefs that p and that p—>q. So far only beliefs have been considered. But the above is a structure which we find in numerous intrapsychic connections, for example between desires, intentions and emotions (or at least some important aspects of them) and other contentful states. So my fear of something consists, at least in part and in central cases, of my taking it to be dangerous or threatening. But when I so take it, it is in the light of my belief that it may explode, or may bite. So my fear, insofar as it is to be identified with taking the feared thing to be dangerous, is experienced by me as justified or appropriate in the light of other thoughts. Similarly I may take a resolution to perform an action in the light of that action seeming to me to be advantageous and to have no drawbacks. Again it is not just that first I think about the advantages and lack of drawbacks and then next instant those considerations are entirely swept from my conscious mind and replaced by the thought 'I'll do it!' Rather, the ensuing thought is more like 'I'll do it (since it is advantageous and has no drawbacks).' And what goes for me goes for others, on the account of the use of psychological concepts which I am sketching. We do not think of others primarily or solely as extremely complex biological machines, with many physical structures inside interacting in elegant ways; in thinking of a person as a person, these aspects of human existence are not to the fore. Of course there is complexity in others' psychological states and this complexity is implicated in temporal development which it is quite proper to think of as causal, in some sense of 'causal'. This is what makes the 'biological machine with complex innards' story, and the related functionalist view in philosophy of mind, seem plausible at all. But when we think of persons the complexity we are aware of is unified in a particular way. It is not unified just as 'the states of the bits of stuff inside that skull' but rather as 'the elements of the coherent world view constructed by the person whose body that is'. And the person is unified inasmuch as her mind is unified, i.e. inasmuch as the elements of it are seen as cohering and are brought to bear on one another, to suggest new conjectures, to correct misconceptions, to provide mutual support through their rational connection and so forth. A person becomes aware of her world and builds up a picture of it, through perception, memory and reasoning. And that view must be unified in the way sketched. But let us note also that her view will 94

Understanding Other Minds from the Inside necessarily include, woven in among the rest, many indexical thoughts, defining her beliefs about herself, her placement, role, capacities and so forth. For example they will include beliefs of the form 'I am in such and such a location'. 'I am capable of these or those actions.' 'I occupy such and such a role.' 'These and those achievements, dangers, disappointments or pleasures are possible for me' and so forth. These elements may be said to define a 'point of view' on the world, in both a literal and a metaphorical sense. So when I attempt in simulationist style to recreate another's thoughts, insofar as such indexical thoughts are included, then I have, to some greater or lesser extent, attempted to recreate her point of view. The suggestion I would like to pursue now is that it is this complex of ideas which makes the adoption of the idiom of 'understanding from the inside' so natural and attractive. There are a variety of strands in this metaphor. The mind of the other is 'inside' in the sense that (sometimes at least) it is not immediately apparent in behaviour what a person thinks and so we need to reflect on what her thoughts are. It is also 'inside' in that mental events and states are capable of moving the body to spontaneous (i.e. not immediately externally caused) movement. But there is also the fact that, on the view sketched, when I consider the nature of what is 'inside' another person, in the senses suggested by these two points, what I find myself postulating is a set of thoughts which represent the world from a point of view. So the 'inside' which I find is not mere mechanical or biological complexity. If the inside were of that kind there would not be any question of anything being 'from' it. But things can be 'from the inside' with a person because what is 'inside' is itself outwardly directed. It is an interlocking complex of items with indexical representational content concerning the world around that person. The existence of this kind of outward-directed content is bound up with the person's ability to respond to changing perceptions and reasoning by modifying and enlarging the world view in rationally intelligible ways. We think of the content as having been built up by exercise of the person's cognitive capacities, her perceptual awareness and her abilities to remember and reason. The idea of reason then provides a further strand which enriches the 'from the inside' metaphor, inasmuch as in deploying it I represent other people as beings capable of recognizing and responding to norms, whose thoughts and behaviour therefore have sense and can be justified, in ways which have no analogue in the explanations provided for the behaviour of inanimate items. What is the status of all this, you might ask. I have just outlined very roughly a picture which we have of ourselves and others - each of us a rational subject with a point of view having multifaceted 95

Jane Heal abilities to think effectively about many subject matters and so forth. And this picture is, I have suggested, the one presupposed by the form of simulationism which I have tried to outline and defend. I would also like to suggest the converse, namely that this kind of simulationism is the natural theory of the understanding of other minds for someone who conceives of persons as unified rational subjects. It is clear, then, that a presupposition of rationality, ability to appreciate what follows from what, to respond to reasons by grasping their force, is central to this whole complex of views. But could it be that this presupposition might turn out to be recognizably false? If so either we must say that, contrary to what I have urged, there is no conceptual link between the mental and rationality or we shall have to reconfigure our idea of the mental so as to extrude the rationality assumption. Or might it turn out that mental notions are inextricably bound up with the illusory idea of rationality and so need eliminating altogether? This is too big a topic to address here. But I would like to conclude by offering a few reflections. The idea that we are rational has received some excellent probing and clarification from philosophers; and psychologists have also undertaken fascinating empirical investigations bearing on the actu- \ al workings of our inferential processes. The joint upshot of this j philosophical and empirical work is that it is quite clear that there > are a number of grand and demanding senses of 'rational' in which ; we cannot properly claim to be rational. Such ability as we have is imperfect, limited by the finitude of our memories and by the amount of complexity we can take in. We do not have the time, energy or attention even to do all of the comparatively simple thinking and inferring which would be useful to us, let alone many elaborate reasonings, and letting even further along the grandiose projects of achieving total consistency or coming to recognize all the logical consequences of what we accept. And, worse, we seem to be prone to systematic errors in elementary reasoning; there seem to be inferential versions of perceptual illusions such as the Muller-Lyer case, where we are gripped by the conviction that something follows from something else when it does not.8 So our rationality, if it exists, does not amount to anything very grand. But then we do not need anything very grand to defend the picture sketched above, any more than we need to credit ourselves with illusion-resistant eyesight of eagle-like acuity in order to 8 Two excellent books, the second of which also provides references to much other recent work, are C. Cherniak, Minimal Rationality (Cambridge, MA: MIT Press, 1986) and E. Stein, Without Good Reason (Oxford: Oxford University Press, 1996). 96

Understanding Other Minds from the Inside defend the claim that in vision we have a sense which enables us to become aware of the placement and properties of things about us. Sight is reliable enough for us to be able to become aware, when we reflect, of when it is prone to illusion. So we can use in it increasingly subtle and well-focused ways (involving cross-checking, selfcritical awareness of possible sources of error, helpfully devised instruments and the like) to progressively improve our grasp on the layout and properties of objects. The central claim we need about rationality is closely analogous. We need to be entitled to the assumption that thinking about a question, deploying all that we know which bears on it, will generally tend to improve our grasp on that issue rather than the contrary. And as in the case of vision, what makes this central idea defensible is that we are capable of such things as reflecting on our reasoning practices, of recognizing mistakes through cross-checking and of turning, where need be, to various aids. And thus we are capable of progressively improving our sense of what follows from what. No empirical evidence currently to hand shows that we are not entitled to the assumption that our basic thinking capacity is not fundamentally pointed in the right direction, i.e. in the direction of leading us, when we employ it, to a better grip on things. Indeed the empirical studies which identify our inferential shortcomings are precisely evidence to support the assumption. And how would anyone who did not make it proceed with his or her thinking? What is the practical alternative to making it? There is none. Consider finally something about our relations with other people. It is often taken for granted in the discussions of philosophers and psychologists that the central role of psychological concepts is to enable us to predict inner states in others so that we can, in turn, predict and sometimes influence the behaviour those states bring about. But this is a serious distortion. Our relations to other people do not have the same structure as our relations with inanimate objects, plants or machines. We do not deal with our family members, friends, colleagues or fellow citizens as we do with volcanoes, fields of wheat or kitchen mixers, namely by trying to figure out the nature and layout of their innards so that we can predict and perhaps control them. Prediction and control may (sometimes rightly and sometimes wrongly) be the name of the game for psychiatrists, prison governors or dictators, in some of their dealings with some people. But it is not the name of the game for most distinctively human interactions. A much more central pattern occurs where one person offers to another some articulation of how things strike him or her - a remark, gesture, action or expression — in the course of pursuing 97

Jane Heal some more or less well-defined joint project. Certainly this will be offered in the expectation (or at least the hope) that it will be identified for what it is. Thus far a prediction will probably be made. And also a prediction will be made that a response of a certain very broad class will be forthcoming. So in a philosophical debate one will expect to get back a philosophical question or observation, in a chess game one will get a chess move, in a game of mud pies one will get an elaboration of the mud pie world, in a courtship one will get a move to deepen the intimacy and so forth. But the specific nature of the response is not predicted. Social life would be utterly boring, completely different from the communicative reality we experience, if it were. What we hope of another with whom we interact is not that he or she will go through some gyrations which we have already planned in detail, but that he or she will make some contributions to moving forward the joint and co-operative enterprise on which we both are, more or less explicitly, engaged. There will be in any particular case many moves which would fit the general bill; which move an individual makes depends upon his or her individual appreciation of the situation, to which he or she brings not only differences in temperament, inventiveness and the like but also, nearly always, differences in awareness both of empirical facts and of what follows from what. In a philosophical discussion the parties will probably share a good deal of common ground; but they will not be, psychologically speaking, identical twins. That is why there is a point in discussion; we engage in it so that we can pool our knowledge, insights, inventiveness and so forth. This is one way of combating our finitude, namely by having different of us pursue different lines of thought, since there is typically more labour in discovery than in appreciation of the discovery once made. Division of intellectual labour is not something which comes on the scene merely with a large accumulations of knowledge and specialization in the sciences. It is built right in to the idea of conversation and co-operation in the most everyday activities and plans. And the way in which we carry on such activities shows that we presuppose the rationality of others, presuppose that we share standards of what follows from what and what is relevant to what. Our first move, on finding another's response not immediately intelligible and helpful, is to search round for an interpretation which makes it so. And if others disagree with us about what constitutes good reasoning, making moves which show that at some level they do not share standards of what follows from what, then we seek to put them right in the expectation that they will acknowledge the mistake (or perhaps they will show that it is we who have made the mistake). 98

Understanding Other Minds from the Inside Let us further note that when a mistake is agreed to have been made we will often look for, and find, a reason why it was made, not just in the sense of a cause or regularity in its making but in the sense of some excuse which reconciles the mistake with the idea that, even in making it, the perpetrator was exercising his or her rationality. This may be done by pointing to the false presuppositions which were accepted, the misleading analogy which was unduly prominent or some similar factor. Few mistakes, whether factual errors or mistakes in reasoning, are just opaquely and blankly completely unintelligible when reflected on. Some shred of justification can nearly always be found. And how well this general orientation to others serves us, how well things work out, on the whole. And how completely lacking we are of any conception of how things could be differently conducted. Empirical studies of our limitations and proneness to error (together with such things as awareness of the differences of our own outlook from those of other cultures and times or Freudian insights into the deeper workings of our motivations and self-conceptions) may all enrich the mixture and make us aware that the intelligibility we seek is not always to be found easily or on the surface. But such things do not prevent us looking for reason and intelligibility or stand in the way of our thinking, in most cases, that we have found it. So I conclude that rationality, in the schematic but still powerful sense sketched, is a very deeply entrenched assumption in our picture of human beings, ourselves and others, and hence that the understanding of other minds which calls upon the simulationist framework is not to be easily dislodged or replaced. 99

Self-knowledge: the Wittgensteinian Legacy CRISPIN WRIGHT It is only in fairly recent philosophy that psychological self-knowledge has come to be seen as problematical; once upon a time the hardest philosophical difficulties all seemed to attend our knowledge of others. But as philosophers have canvassed various models of the mental that would make knowledge of other minds less intractable, so it has become unobvious how to accommodate what once seemed evident and straightforward — the wide and seemingly immediate cognitive dominion of minds over themselves. In this paper1 I'll begin by trying to characterize this dominion with some care. We need to have it as clear as possible why one traditional way of thinking about the matter has seemed so attractive - even unavoidable - and what a satisfactory account of the issues in this region has to accomplish. However my overarching concern will be with the bearing of later Wittgensteinian materials on the question. Ultimately I think we can get an insight into the intended force of something which I do not think has so far been sufficiently well understood: the anti-explanatory motif that runs through the pronouncements on philosophical method occurring in the Philosophical Investigations. People can be variously deluded about themselves, self-deceived about their motives, for instance, or overly sanguine, or pessimistic, about their strengths of character and frailties. But it is nonetheless a truism that for the most part we know ourselves best - better than we know others and better than they know us. In one kind of case, the explanation of this would seem straightforward. It is (merely) that our own presence is, for each of us, a constant factor in the kind of situation, usually but not always social, in which the evidence emerges which bears on various of our psychological characteristics. No-one else is so constantly around 1 This is an edited version of my contribution to Knowing Our Own Minds, ed. C. Macdonald, B. Smith and C. Wright (Oxford: Oxford University Press 1998). 101

Crispin Wright us. So no-one else observes as much of us or is as much observed by us. Selves have the best evidence about themselves. Evidently, however, this form of explanation of the truth in the truism can run only in cases where one's own and another's knowledge of oneself must draw on the same kind of evidence. So it is restricted, it would seem, to broadly dispositional characteristics like honesty, patience, courage and conceit - cases where there is no essential self/other asymmetry in the means of knowledge. And this is not, of course, the most salient type of case. In the most salient type of case, we not merely know ourselves best but also differently from the way in which we know others and they know us. The distinction is complicated, admittedly, by the fact that many apparently dispositional psychological characteristics are distinctively manifested not by raw behaviour, as it were, but by psychological performance in respects that may themselves exhibit self/other epistemological asymmetries. Conceit, for instance, will be, inter alia, a disposition to form certain kinds of belief. It remains that the type of case that sets our problem is that which gives rise to the phenomenon of avowal — the phenomenon of authoritative, non-inferential self-ascription. The basic philosophical problem of self-knowledge is to explain this phenomenon - to locate, characterize and account for the advantage which selves seemingly possess in the making of such claims about themselves. The project will be conditioned by whatever more precise characterization we offer of the target phenomenon. It seems safe to suppose that we must begin by distinguishing two broad classes of avowal. The first group — what I will call phenomenal avowals — comprises examples like 'I have a headache', 'My feet are sore', 'I'm tired', 'I feel elated', 'My vision is blurred', 'My ears are ringing', 'I feel sick' and so on. Such examples exhibit each of the following three marks: First, they are groundless. The demand that somebody produces reasons or corroborating evidence for such a claim about themselves - 'How can you tell?' - is always inappropriate. There is nothing they might reasonably be expected to be able to say. In that sense, there is nothing upon which such claims are based. Second, they are strongly authoritative. If somebody understands such a claim, and is disposed sincerely to make it about themselves, that is a guarantee of the truth of what they say. A doubt about such a claim has to be a doubt about the sincerity or the understanding of its author. Since we standardly credit any interlocutor, absent evidence to the contrary, with sincerity and understanding, it follows that a subject's actually making such a claim about themselves is a criterion for the correctness of the corresponding third-personal claim made by someone else: my avowal that I'm in pain must be 102

Self-knowledge: the Wittgensteinian Legacy accepted by others, on pain of incompetence, as a ground for the belief that I am. Finally, phenomenal avowals exhibit a kind of transparency. Where P is an avowal of the type concerned, there is typically something absurd about a profession of the form, 'I don't know whether P' - don't know whether I have a headache, for instance, or whether my feet are sore. Not always: there are contexts in which I might be uncertain of a precondition (for instance, whether I have feet). But in the normal run of cases, the subject's ignorance of the truth or falsity of an avowal of this kind is not, it seems, an option. None of the examples listed is an avowal of a content-bearing state. It is the hallmark of the second main group of avowals - what I shall call attitudinal avowals — that the psychological characteristics, processes and states which they concern are partially individuated by the propositional content, or intentional direction, which they contain - for instance, 'I believe that term ends on the 27th', 'I hope that noise stops soon', 'I think that professional philosophers are some of the most fortunate people on earth', 'I am frightened of that dog', 'I am thinking of my mother.' In order to see what is distinctive about an author's relation to avowals of this kind, we need first to take account of the fact that such claims can also be made as part of a process of self-interpretation — in the kind of context when we say that we have learned about our attitudes by finding that certain events cause us pleasure, for instance, or discomfort. Consider the following passage from Jane Austen's Emma: Emma's eyes were instantly withdrawn; and she sat silently meditating in a fixed attitude, for a few minutes. A few minutes were sufficient for making her acquainted with her own heart. A mind like hers, once opening to suspicion, made rapid progress. She touched - she admitted - she acknowledged the whole truth. Why was it so much the worse that Harriet should be in love with Mr Knightley than with Mr Churchill? Why was the evil so dreadfully increased by Harriet's having some hope of return? It darted through her, with the speed of an arrow, that Mr Knightley must marry no-one but herself.2 Here Emma has just been told of the love of her protegee, Harriet, for her - Emma's - bachelor brother-in-law, a decade older than Emma, a frequent guest of her father's, and hitherto a dependable, somewhat avuncular part of the background to her life of whom she 2 Jane Austen, Emma (London: Penguin Books, 1987), p. 398. This nice passage was drawn to my attention by Julia Tanney, who uses it for her own purposes in her 'A Constructivist Picture of Self-Knowledge', Philosophy 71 (1996), 405-22. 103

Crispin Wright has entertained no romantic notions. But now she realizes that she strongly desires that he marry no-one but her, and she arrives at this discovery by way of surprise at the strength and colour of her reaction to Harriet's declaration, and by way of a few minutes' reflection on that reaction. She is, precisely, not moved to the realization immediately; it dawns on her first as something she suspects and then recognizes as true. It explains her reaction to Harriet. In such self-interpretative cases, none of the three features we noted of phenomenal avowals is present. There is no groundlessness: the subject's view is one for which it is perfectly in order to request an account of the justifying grounds. There is no strong authority: mere sincerity and understanding will be no guarantee whatever of truth - it is for Jane Austen to stipulate, as it were, that Emma's self-discovery is the genuine article, but in any real context such a conclusion could be seriously mistaken. Finally, there is no transparency: within a context of self-interpretation, it is no way incongruous if the subject professes ignorance of particular aspects of her intentional psychology. However, what it is vital to note for our present purpose is that such self-interpretative cases, although common, cannot be the basic case. For the body of data on which self-interpretation may draw is not restricted to recollected behaviour and items falling within the subject matter of phenomenal avowals. When Emma interprets her reaction to Harriet's declaration as evidence that she herself loves Knightley, there is an avowable ground - something like 'I am disconcerted by her love for that man and (more so) by the thought that it might be returned' - which is a datum for, rather than a product of self-interpretation. Selfinterpretation, that is to say, will typically draw on non-inferential knowledge of a basic range of attitudes and intentionally characterized responses. These will not be distinguished, I think, from nonbasic, interpretative cases by any generic features of their content; rather they will reflect matters which, for the particular subject in the particular context, happen to require no interpretation to be known about - matters which are precisely avowable. It is these basic examples which comprise the attitudinal avowals. Such avowals will have the same immediacy as phenomenal avowals and will exhibit both groundlessness and transparency — groundlessness rather trivially in so far as, any interpretational basis having been excluded, there will naturally be nothing a subject can say to justify such a self-ascription; transparency in the sense that, except where the matter is one of interpretation, we think a subject ought to know without further ado what she believes, or desires, etc., so that any profession of ignorance or uncertainty, unless coupled with a readiness to allow the matter is not basic but calls for 104

Self-knowledge: the Wittgensteinian Legacy (self-)interpretation, will seem perplexing. However, attitudinal avowals do not exhibit the strong authority of phenomenal avowals: to the extent that there is space for relevant forms of self-deception or confusion, sincerity-cum-understanding is no longer a guarantee of the truth of even basic self-ascriptions of intentional states. Any avowal may be discounted if accepting it would get in the way of making best sense of the subject's behaviour. But with attitudinal avowals it is admissible to look for other explanations of a subject's willingness to assert a bogus avowal than those provided by misunderstanding, insincerity or misinterpretation. This is indeed the space occupied by the ordinary notion of self-deception; but the more general idea is just that we can be caused to hold mistaken higher-order beliefs in ways - wishful thinking, for instance - which do not go through misguided self-interpretative inference. It is striking that attitudinal avowals would appear to exhibit a form of weak authority nevertheless: that is, at least in basic, nonself-interpretative cases they provide empirically assumptionless justification for the corresponding third-person claims.3 Other things being equal, I ought to know what my beliefs, desires and hopes, etc., are, even if sincerity and understanding alone do not guarantee the truth of what I say about them. 3 Since it cannot be attributed, as with phenomenal avowals, to the fact of sincerity-cum-understanding guaranteeing truth, it is an interesting question what this weak authority should be taken to consist in. It might be suggested that it is nothing other than the presumptive acceptability of testimony generally. And certainly that proposal would be enough to set our problem: for the presumptive acceptability of original testimony — testimony for which the source is not itself testimony - extends no further than to subject matters which an informant is deemed competent to know about. So the question would recur: how is it possible for subjects to know about their intentional states in ways that involve no consideration of the evidence on which a third-party must rely? Actually, however, I think the suggestion is wrong. What distinguishes the presumptive acceptability of attitudinal avowals from anything characteristic of testimony generally, is that the authority which attaches to them is, in a certain sense, inalienable. There is no such thing as showing oneself chronically unreliable in relation to the distinctive subject matter of attitudinal avowals. I may have such poor colour vision that you rightly come to distrust my testimony on matters of colour. I may, unwittingly, have a very bad memory and, learning of this, you may rightly come to a state of wholesale suspicion about my testimony on matters of personal recall. But no corresponding wholesale suspicion concerning my attitudinal avowals is possible. You may not suppose me sincere and comprehending and yet chronically unreliable about what I hope, believe, fear and intend. Wholesale suspicion about my attitudinal avowals - where it is not a doubt about sincerity or understanding-jars with conceiving of me as an intentional subject at all. 105

Crispin Wright II We now have a sufficient focus for our central question. The cardinal problem of self-knowledge is that of explaining why avowals display the marks they do - what is it about their subject matter, and the subject's relationship to it, which explains and justifies our accrediting her sincere pronouncements about it with each of groundlessness, transparency and strong authority in the case of phenomenal avowals, and with groundlessness, transparency and weak authority in the case of attitudinal avowals? How is it possible for subjects to know these matters non-inferentially? How is it (often) impossible for them not to know such matters? And what is the source of the special authority carried by their verdicts? There is a line of response to these questions that comes so naturally as to seem almost irresistible - indeed, it may even seem to ordinary thought to amount merely to a characterization of the essence of mind. According to it, the explanation of the special marks of avowals is that they are the product of the subject's exploitation of what is generally recognized to be a position of (something like) observational privilege. As an analogy, imagine somebody looking into a kaleidoscope and reporting on what he sees. No-one else can look in, of course - at least while he is taking his turn. If we assume our Hero perceptually competent, and appropriately attentive, his claims about the patterns of shape and colour within will exhibit analogues of each of the marks of phenomenal avowals: 1. The demand that he produces reasons or corroborating evidence for his claims will be misplaced - the most he will be able to say is that he is the only one in a position to see, and that is how things strike him; 2. granted his proper perceptual functioning, it will be sufficient for the truth of his claims that he understands them and is sincere in making them; so that for anyone who understands the situation, our Hero's merely making such a claim will constitute a sufficient, though defeasible reason for accepting its truth; and 3. where P is any claim about the patterns of shape and colour visible within, there will be no provision — bearing in mind Hero's assumed perceptual competence and attentiveness - for his intelligibly professing ignorance whether or not P. This analogy isn't perfect by any means. In order to construct it, we have had to assume normal perceptual functioning and full attentiveness on the part of our observer. And no such assumption conditions our reception of others' avowals. But once into one's stride with this type of thinking, this difference will not seem 106

Self-knowledge: the Wittgensteinian Legacy bothersome. The line will be that in the inner observational realm, in contrast to the outer, there is simply no room for analogues of misperception or of oversight or occlusion - for the objects and features there are necessarily salient to the observing subject. Or at least they are so in the case where they are objects and features recordable by phenomenal avowals. In the case of the subject matter of attitudinal avowals, by contrast, space for an analogue of misperception can and should be found - that will be what explains the failure of strong authority in those cases. In brief: this - Cartesian — response to the problem of avowals has it that the truth values of such utterances are non-inferentially known to the utterer via her immediate awareness of events and states in a special theatre, the theatre of her consciousness, of which others can have at best only indirect inferential knowledge. In the case of phenomenal avowals, this immediate awareness is in addition, infallible and all-seeing; in the case of basic attitudinal avowals, it is merely very, very reliable. So presented, the Cartesian picture, of the transparency of one's own mind and, by inevitable contrast, the opacity of others', emerges as the product of a self-conscious attempt at philosophical explanation. That may seem congenial to John McDowell's claim that 'We need to be seduced into philosophy before it can seem natural to suppose that another person's mind is hidden from us.'4 McDowell recoils against the idea that anything like the Cartesian picture might be part of ordinary unphilosophical thought. But I think he is wrong about this, the theoretical setting I have given to the picture notwithstanding. To be sure, it is unclear what should count as a 'seduction into philosophy'. But if every manifestation of the Cartesian picture is to rate as the product of such a seduction, then the seductive reach of philosophy is flatteringly wide. I do not imagine, of course, that people typically self-consciously follow through the train of thought I outlined. But we ought not to balk at the notion that no intellectual routine characteristically pursued by those in its grip should capture exactly the best reconstruction of why an idea appeals. The privacy of the inner world is a recurrent idea in literature.5 It is arguably a presupposition of the whole idea of the continuation of one's consciousness after death. The thought 4 From p. 149 of John McDowell, 'Intentionality and Interiority in Wittgenstein', in Meaning Scepticism, ed. K. Puhl (Berlin: de Gruyter, 1991), pp. 148-69. 5 To take another nineteenth-century example, it is, in a sense, the entire subject matter of George Eliot's novella, The Lifted Veil. The Cartesian character of that writer's notions about the mental is explored in depth in Catherine Wright's 'The Unseen Window: Middlemarch, Mind and Morality' (Ph.D. dissertation, University of St Andrews, 1991). 107

Crispin Wright of the undetectable inverted colour spectrum is something which can engage quite young children without too much difficulty. And in each of these cases what comes naturally is essentially nothing other than the notion of a kind of privileged observation of one's own mind which works, in the ways we have reviewed, to explain the first-thirdperson asymmetries in ordinary psychological discourse. The privileged-observation explanation is unquestionably a neat one. What it does need philosophy to teach is its utter hopelessness. One very important realization to that end is that nothing short of full-blown Cartesianism can explain the asymmetries in anything like the same way - there can be no scaled-down observational model of self-knowledge which preserves the advantages of the Cartesian account while avoiding its unaffordable costs. The problem, very simply, is that the kind of authority I have over the avowable aspects of my mental life is not transferable to others: there is no contingency (anyway, none of which we have any remotely satisfactory concept6) whose suspension would put other ordinary people in position to avow away on my behalf, as it were - would transfer, or extend my advantage to them. So the conception of avowals as reports of inner observation is saddled with the idea that the observations in question are ones which necessarily only the subject can carry out. And once that conception is in place, others' means of access to the states of affairs which their subject (putatively) observes is bound to seem essentially second-rate by comparison and to be open to just the kinds of sceptical harassment which generate the traditional problem of other minds - the unaffordable cost referred to. Ill If this is right, then a deconstruction of the privileged-observation solution to the problem of self-knowledge is an indispensable prerequisite of an overall satisfactory philosophy of mind. It seems to me that the accomplishment of such a deconstruction was one major achievement of Wittgenstein's later philosophy - though it would take another paper, or series of papers, properly to fill out how it 6 In particular, I do not think that we have any satisfactory concept of what it would be to be in touch with others' mental states telepathically. I do not mean, of course, to rule it out that someone might prove, by dint of his own occurrent suspicions and afflictions, to be a reliable guide to the states of mind of another. But that possibility falls conspicuously short of the idea that a subject might share direct witness of another's mental states. 108

Self-knowledge: the Wittgensteinian Legacy goes.7 In essentials, what he does is to mount a two-pronged attack on the Cartesian picture, with the prongs corresponding to the distinction between the two main kinds of avowals. The idea that phenomenal avowals serve as inner observation reports is challenged by the so-called 'private language argument' — the battery of considerations that surface in §§243 to the early 300s in the Investigations. The attack is multi-faceted but the famous central strand is that the Cartesian picture implicitly surrenders the resources needed for a distinction which is essential if such 'reports' are to have anything of the objectivity implicit in the very idea of an observational report: the objectivity implicit in the idea of successful representation of some self-standing aspect of reality, which demands a potential contrast between how matters seem to an observer and how they really stand. The corresponding conception of attitudinal avowals, by contrast, is challenged by the various phenomenological and other considerations which Wittgenstein marshals in the, as we may call them, 'not a mental process' passages recurrent throughout the text.8 A central problem with the idea that attitudinal avowals describe introspectable mental occurrences concerns the answerability of ascriptions of intentional states like expectation, hope and belief to aspects of a subject's outward performance that may simply not be available at the time of avowal. If an expectation, say, were a determinate, dated occurrence before the mind's eye, then in any particular case it would either have taken place or not, irrespective of how I subsequently went on to behave. So we ought to be guilty of a kind of conceptual solecism if we hold claims about expectation to be answerable to subsequent sayings and doings in a fashion broadly akin to the way in which the ascription of dispositional states is so answerable. Yet that is exactly what we actually do. The conception of attitudinal avowals as reports of inner observation thus stands at odds with another, fundamental feature of their grammar - their essential answerability, in broadly the fashion of dispositions, to matters which may be unobservable at the time they are asserted.9 7 For further indications, see my 'Wittgenstein's Later Philosophy of Mind: Sensation, Privacy and Intention', in Meaning Scepticism, ed. Puhl, pp. 126-47. 8 See e.g. Philosophical Investigations §§34, 146, 152, 154, 205, 303, 330-2, 427, 577, 673; also part II §vi p. 181, and §xi pp. 217-18. The distinction is prominent in the Remarks on the Philosophy of Psychology as well, where Wittgenstein uses the terminology of dispositions versus states of consciousness; see, for instance, vol. II, §§45, 48, 57 and 178. 9 This idea is elaborated a little at pp. 237ff. Of my 'Wittgenstein's RuleFollowing Considerations and the Central Project of Theoretical Linguistics', in Reflections on Chomsky, ed. Alexander George (Oxford: Basil Blackwell, 1989), pp. 233-64. 109

Crispin Wright The pursuit of these ideas of Wittgenstein leads one to recognize deep incoherences in the Cartesian response to our problem - incoherences that are prior to its inordinate sceptical costs. Note, moreover, that if what I said earlier is right - viz. that there is no alternative for one disposed to pursue the privileged-observation route than to see the privilege as necessarily the exclusive property of the observing subject - then the incoherence of the Cartesian response is the incoherence of any broadly observational model of a subject's relation to her ordinary psychological states. That's a crucial lesson. IV But if not an observational model, then what? There is a proposal \ about our problem that for a time was widely accepted as Wittgen- jj stein's own. At Investigations §308 he writes ! How does the philosophical problem about mental processes and states and about behaviourism arise? - The first step is the one that altogether escapes notice. We talk of processes and states and leave their nature undecided. Sometime perhaps we shall know more about them - we think. But that is just what commits us to a particular way of looking at the matter. And a little earlier (§304) he urged that we need to make a radical break with the idea that language always functions in one way, always serves the same purpose: to convey thoughts — which may be about houses, pains, good and evil, or anything else you please. These sections advance the diagnosis that our difficulties in this neighbourhood are generated by 'the grammar which tries to force itself on us here' (§304). They go, Wittgenstein suggests, with a conception of avowals as reports and the associated conception of a self-standing subject matter which they serve to report. We take it that there are mental states and processes going on anyway, as it were - the 'first step' that escapes notice - and that each person's avowals serve to report on such states and processes as pertain to her. The immediate effect is to set up a dilemma. How, in the most general terms, should we think of the states of affairs which confer truth on these 'reports'? There is the Cartesian - events-in-anarena-accessible-only-to-the-subject - option; this does a neat job of explaining the distinctive marks of avowals, at least at a casual muster, but it relies on an 'analogy which... falls to pieces' (§308) - 110

Self-knowledge: the Wittgensteinian Legacy the analogy between avowals and observation reports made from a privileged position. But the only other option seems to be to 'go public': to opt for a view which identifies the truth-conferring states of affairs with items which are somehow wholly manifest and available to public view — an option which Wittgenstein expects, writing when he did, will naturally take a behaviourist shape so that 'Now it looks as if we had denied mental processes.' Of course, a philosopher who takes this option - whether in behaviourist or other form - will want to resist the suggestion that she is denying anything, according to her recommended understanding of 'mental process', just as Berkeley resisted the suggestion that he was denying the existence of matter. But the manifest problem is to reconcile any such conception of the truth conditions of avowals with their distinctive marks: for as soon as you go public, it becomes obscure what advantage selves can enjoy over others. This line of difficulty may seem to point to an obvious conclusion. Conceiving of avowals as reports of states and processes which are going on anyway appears to enforce a disjunction: either accept the Cartesian view, which cannot accommodate ordinary knowledge of others, or accept some form of externalization - perhaps behaviourist, nowadays more likely physicalist - which cannot sustain the special place of self-knowledge. So we should reject the parent assumption. And one tradition of commentary, encouraged especially by Investigations §244,10 interprets Wittgenstein as doing this in a very radical way: as denying that avowals are so much as assertions (that they make statements, true or false), proposing to view them rather as expressions of the relevant aspects of the subject's psychology." 'Expression'? To give expression to an aspect of one's psychology just means, presumably, to give it display, in the way in which wincing and a sharp intake of breath may display a stab of pain, or a smile may display that one is pleased, or a clenching of the teeth that one is angry. Propositional attitudes too can be open to natural expression of this kind: a prisoner's rattling the bars of his cell is a natural expression of a desire to get out. (It is not a way of acting on that desire, of course - it is not rationalized by it.) Wittgenstein's famous suggestion in §244 is that we should see the avowal of pain 10 But see also Remarks on the Philosophy of Psychology, vol. I, §§450, 501, 593, 599 and 832. 11 The sometime popularity of this interpretation is traceable to its being advanced by several of the first reviewers: P. F. Strawson, for instance, in his critical study of the Investigations in Mind 63 (1954), 70-99; and Norman Malcolm in his 'Wittgenstein's Philosophical Investigations' in The Philosophical Review 63 (1954), 530-59. I l l

Crispin Wright as an acquired form of pain behaviour: something one learns to use to supplant or augment the natural expression of pain and which (the expressivist tradition of commentary suggests) is no more a statement - something with a truth-evaluable content - than are such natural forms of expression. The immediate question is how well an expressivist treatment of avowals can handle their distinctive marks. And the answer appears to be: not badly at all. For instance, if the avowal 'I am in pain' is not a statement, true or false, then naturally it is inappropriate to ask its author for grounds for it (groundlessness) and naturally there is no question of her ignorance of its truth value (transparency). And if, when uttered with proper comprehension, it is to be compared to an episode of pain behaviour, then only its being a piece of dissimulation — not sincere — can stand in the way of a conclusion that the subject really is in pain (strong authority). (And of course it will provide a criterion for the subject's being in pain in just the way that ordinary pain behaviour does.) Nevertheless the expressivist proposal has come to be more or less universally viewed as a non-starter, for reasons preponderantly to do with the perceived impossibility of making coherent philosophy of language out of it. The claim that the avowal 'I am in pain' serves to make no statement, true or false, has to be reconciled with a whole host of linguistic phenomena whose natural explanation would exploit the opposed idea that it is, just as it seems, the affirmation of a truth-evaluable content. Here are four of the snags: 1. What has the expressivist proposal to say about transformations of tense - 'I was in pain' and 'I will be in pain'? If either is a genuine assertion, doesn't there have to be such a thing as an author's making the same assertion at a time when doing so would demand its present-tense transform? If on the other hand they are regarded likewise merely as expressions, what do they serve to express? (Doesn't an expression accompany - hence have to take place at the same time as — what it expresses?) 2. How is the proposal to construe a locution like 'He knows that I am in pain'? If there is a use of the words 'I am in pain' so embedded, which I can use to express the content of someone else's possible knowledge, why may I not assert that very same content by the use of the same words? 3. There are genuine - for instance quantified - statements which stand in logical relations to 'I am in pain.' It entails, for instance, 'Someone is in pain.' How can a genuine statement be entailed by a mere expression? 4. 'I am in pain' embeds like any normal assertoric content in logical constructions such as negation and the conditional. 'It's not the 112

Self-knowledge: the Wittgensteinian Legacy case that I am in pain' and 'If I am in pain, I'd better take an aspirin' are syntactically perfectly acceptable constructions. But how can a mere expression, in contrast to an assertion, be denied.} And doesn't the antecedent of a conditional have to be understood as the hypothesis that something is the case} This kind of point - I shall dub the whole gamut 'the Geach point"2 - has often been used as a counter to various forms of expressivism, notably in ethics, and much ingenuity has been expended (squandered?) by philosophers of expressivist inclination in the attempt to meet it. But in the present case I don't think it ought to have been influential at all. In the ethical case, the expressivist thesis is, crudely, that there are no real moral states of affairs; so the occurrence of what are apparently truth-evaluable contents couched in distinctively moral vocabulary has to be some kind of illusion. In that case the Geach point represents a very serious challenge, since it seems to show that everyday moral thought, in exploiting perfectly standard syntactic resources like those afforded by ordinary sentential logic, requires to the contrary that truthevaluable moral contents exist. By contrast, it is no part of the present, allegedly Wittgensteinian expressivist proposal that there is no such thing as a statement of ordinary psychological fact. No-one is questioning that 'He is in pain' is an assertion. The expressivist thesis distinctively concerns avowals. How does that difference help? Well, it is clear that we have to draw a distinction in any case between the question whether an indicative sentence is associated with a truth-evaluable content and the question whether its characteristic use is actually assertoric. For the two notions routinely come apart in the case of standard performatives like 'I promise to be on time', 'With this ring, I thee wed', 'I name this ship...', and so on. Each of these locutions embeds in all the ways the generalized Geach point focuses on; and none of them is standardly used, in the atomic case, as an assertion. We should conclude that what the Geach point signals is merely the presence of truth-evaluable content. It is powerless to determine that the standard use of a locution is to assert such a content. And how the expressivist thesis about avowals can be merely that the typical use of such sentences is as expressions rather than assertions. There need be no suggestion that one cannot make assertions about one's own psychology. But the suggestion - now initially rather exciting - will be that the appearance of the epistemic superiority of the self which avowals convey is an illusion created by 12 After P. T. Geach's emphasis of such difficulties for moral expressivism, Austin's performatory account of knowledge, etc. See Geach's Assertion' in The Philosophical Review 74 (1965), 449-65. 113

Crispin Wright attempting to find a home for features of such utterances which they carry qua expressions in the context of the mistaken assumption that they are ordinary assertions. When selves do make strict assertions about their own psychology, the story should continue, any epistemic advantages they enjoy are confined to those of superiority of evidence which I briefly noted at the beginning. That, it seems to me, is, in outline, how the best expressivist proposal should go. Now to its real problems. Perhaps the most immediate awkwardness, if a general account of avowals is to be based upon the §244 idea, is that, even in the case of sensations, the range of cases where there are indeed natural, non-linguistic forms of expression — cases like pains, itches and tickles — is very restricted: contrast for instance the sensation of coolness in one foot, or the smell of vanilla. In the latter kind of case, the suggested model of the acquisition of competence in the avowal simply won't grip, and the theorist will have to try to live with the idea of a range of sensations whose only expression consists in their avowal. The same is evidently true in spades of psychological items other than sensations. This threatens a worrying dilution of the key notion of expression. That's a worry that might, I suppose, be worked on. But the next one seems decisive. Suppose a highly trained secret agent under torture resolutely gives no ordinary behavioural sign of pain. However, his torturers are men of discernment, with subtle instruments, who know full well of his agony nonetheless: they know the characteristic signs - patterns on the electro-encephalograph, raised heart rate, activation of reflexes in the eye, changes in surface skin chemistry, etc., etc. If the suggestion really is to be that the superiority of the first-person viewpoint is wholly an artefact of a grammatical misunderstanding - the misconstrual of expressions as assertions - then any knowledge, strictly so conceived, which the victim has of his own pain has to originate in the same way as that of his tormentors. But by hypothesis he isn't expressing pain behaviourally. And the signs that leave them in no doubt are things which, in his agony, he may not be attending to, or which, like the print-out on the electroencephalograph, he may not be able to see or interpret if he could see. So in such a case, when it comes down to knowledge, it looks as though the expressivist account must represent the victim as actually at a disadvantage. That's evident nonsense. In general, merely to conceive of avowals as expressive does not, when it goes in tandem with an acceptance of the reality of the states of affairs which they express, provide any way of deflecting the question: how, broadly speaking, should we conceive of the kind of state of affairs which is apt to confer truth on psychological 114

Self-knowledge: the Wittgensteinian Legacy ascriptions, and in what sort of epistemological relationship do their subjects themselves in general stand to such states of affairs? If this relationship is in any way more than evidentially privileged, we have our original problem back. If it isn't, we seem to get absurdities like that just illustrated. A different way of seeing the ultimate unplayability of the expressivist position is to reflect that the content of an avowal is always available to figure just in a subject's thoughts, without public expression. You may sit reading and think to yourself, 'My headache has gone', without giving any outward sign at all. And anyone versed in ordinary psychology will accept that if you have that thought, not by way of merely entertaining it but as something you endorse, then you will be right (authority); that there is no way that your headache could have passed unless you are willing to endorse such a thought (transparency); and that your willingness to endorse it will not be the product of inference or independently formulable grounds (groundlessness). Thus analogues of each of the marks of avowals that pose our problem engage the corresponding unarticulated thoughts. It must follow that the correct explanation of the possession of them by avowals cannot have anything to do with illocutionary distinctions. We should conclude that while the expressivist proposal flies rather further than is usually thought, it is a dead duck all the same. For sure, the textural evidence for attributing the expressivist view to Wittgenstein was always pretty exiguous. Investigations §244 in particular should be contrasted with the much more cautionary and nuanced remarks elsewhere.13 Such apparent equivocations, of course, are fuel for the common complaint that while Wittgenstein has suggestive criticisms to offer of certain tendencies in the philosophy of mind, he left any intended positive contribution shrouded in fog. What exactly - or even roughly - is Wittgenstein saying about avowals, if he is not advancing the expressivist view? How exactly does he propose we should liberate our thinking from Cartesian tendencies? What did he think we should put in their place? 13 For instance Investigations part II, section ix: 'a cry, which cannot be called a description, which is more primitive than any description, for all that it serves as a description of the inner life. A cry is not a description. But there are transitions. And the words, "I am afraid", may approximate more, or less, to being a cry. They may come quite close to this and also he far removed from it.' 115

Crispin Wright Well actually, I don't think it is all that difficult to glean what his positive recommendation is, at least in general outline. The difficulty is, rather, to settle for it. The first essential in interpreting him here is to give due prominence to the Investigations' explicit conception of the genesis of philosophical problems and of proper philosophical method. Wittgenstein wrote, recall, that we may not advance any kind of theory. There must not be anything hypothetical in our considerations. We must do away with all explanation, and description alone must take its place... [Philosophical problems] are solved, rather, by looking into the workings of our language, and that in such a way as to make us recognise those workings: in despite of an urge to misunderstand them... Philosophy is a battle against the bewitchment of our intelligence by means of language.14 And, very famously, Our mistake is to look for an explanation where we ought to look at what happens as a 'proto-phenomenon'. That is, where we ought to have said: this language game is played." The bearing of these strategic remarks is immediate if we reflect that our whole problem is constituted by a demand for explanation. We are asking: what is the explanation of the characteristic marks of avowals? And we easily accept a refinement of the question along such lines as: what is it about the subject matter of avowals, and about their authors' relation to it, which explains the possession by these utterances of their characteristically effortless, non-inferential authority? Cartesianism takes the question head on, giving the obvious, but impossible, answer. And the expressivist proposal, radical though it is in its questioning of the assumption that the authority of an avowal is the authority of a claim to truth, is not so radical as to raise a question about the validity of the entire explanatory project. But Wittgenstein, seemingly, means to do just that. Against the craving for explanation, he seemingly wants to set a conception of the 'autonomy of grammar'.16 The features of avowals which set our problem - the features which seem to betray something remarkable about self-knowledge - do so only if we suppose that they are in some way consequential upon something deeper, for instance the nature of their subject matter and of their author's relationship to 14 Philosophical Investigations §109. 15 Philosophical Investigations §654. It doesn't matter that this is said in the context of discussion of a different issue (recollection of the content of a prior intention). 16 As Baker and Hacker style it. 116

Self-knowledge: the Wittgensteinian Legacy it. But what imposes that way of looking at the matter? Why shouldn't psychological discourse's exhibition of these features be regarded as primitively constitutive of its being psychological, so that the first-/ third-person asymmetries that pose our question belong primitively to the 'grammar' of the language game of ordinary psychology, in Wittgenstein's special sense - 'grammar' which 'is not accountable to any reality' and whose rules 'cannot be justified by showing that their application makes a representation agree with reality' ?17 What did Wittgenstein suppose entitled him to this? In his later work, as everyone knows, he radically rethought his early conception of the relation between language and reality. It is to this readjustment, I suggest, that we must look if we are to understand the doctrine of the 'autonomy of grammar'. As I read the early 300s, the obstacle which Wittgenstein sees as lying in the way of our philosophical understanding of 'mental processes and states' is not the assumption of the truth-evaluability of avowals, as the expressivist interpretation has it, but rather a general picture of the working of all truth-evaluable language. Wittgenstein means to reject a certain picture of what truth-evaluability involves: the picture gestured at in §304, that our statements always serve 'the same purpose: to convey thoughts - which may be about houses, pains, good and evil, or anything else you please'. This picture involves thinking of assertions as expressing propositions which are laid over against reality in the manner of the Tractatus, so that there have to be selfstanding states of affairs to correspond to avowals, when they are true, and it has therefore to be possible to raise general questions about the nature of these self-standing states of affairs, and the nature of the subject's knowledge of them. And then, when we are mindful of the distinctive marks of avowals, it appears that the states, and the mode of knowledge, must be something rather out of the ordinary — the relevant states of affairs have to be conceived as somehow especially transparent to the subject, or, at the least, as working on her by some form of curiously reliable 'blindsight' (whose curious reliability, moreover, would have to be common knowledge if the authority credited to avowals is to be explained). Wittgenstein's diagnosis is that the 'philosophical problem about mental processes and states and about behaviourism' arises because we insist on interpreting the truth-evaluability of avowals — the source of the linguistic features on which the Geach point fastens - as imposing a conception of their being true, when they are, in terms which have to raise these constitutive questions about nature and access. But these are the very questions — Wittgenstein is saying 17 Philosophische Grammatik, (Oxford: Blackwell, 1969), section X, §§133 and 134. 117

Crispin Wright - which we must free ourselves of the temptation to raise; they are the questions which lead to the fast-track into the fly-bottle. Of course, the conception of truth and truth-makers which, in Wittgenstein's diagnosis, is here at the root of our difficulty is the core of the outlook which Hilary Putnam has called metaphysical realism. Perhaps his single most significant departure from the metaphysics of the Tractatus was Wittgenstein's coming to believe that we have to stop thinking about the relationship between language and reality, and about truth, in that kind of way. VI What is involved in this re-orientation deserves a more refined depiction that I can attempt here. If we abstract from the globally anti-explanatory background mantra, the cash-value of the proposal, just for the issue of self-knowledge, involves a generalization to all avowable subject matter, phenomenal and attitudinal, of a view which might be characterized like this: the authority standardly granted to a subject's own beliefs, or expressed avowals, about his intentional states is a constitutive principle: something which is not a consequence of the nature of those states, and an associated epistemologically privileged relation in which the subject stands to them, but enters primitively into the conditions of identification of what a subject believes, hopes and intends.18 I'll call this general viewpoint the Default View. According to the Default View, it is just primitively constitutive of the acceptability of psychological claims that, save in cases whose justification would involve active self-interpretation, a subject's opinions about herself are default-authoritative and default-limitative: unless you can show how to make better sense of her by overriding or going beyond it, her active self-conception, as manifest in what she is willing to avow, must be deferred to. The truth conditions of psychological ascriptions are primitively conditioned by this constraint. In particular, it is simply basic to the competent ascription of the attitudes that, absent good reason to the contrary, one must accord correctness to what a subject is willing to avow; and limit one's ascriptions to her to those she is willing to avow. It would be a great achievement of Wittgenstein's discussion if it made it possible to understand how the Default View might be the 18 From p. 142 of my 'Wittgenstein's Later Philosophy of Mind: Sensation, Privacy and Intention', in Meaning Scepticism, ed. Puhl. 118

Self-knowledge: the Wittgensteinian Legacy last word on the issue. But it is anything but clear, actually, how a repudiation of the metaphysical realist picture of truth could just by itself directly enjoin this conception. Moreover it is difficult to rest easy with the general anti-explanatory mantra, which is seemingly in tension with a diagnostic thought which is very important to Wittgenstein himself: that philosophical problems characteristically arise because we are encouraged by surface-grammatical analogies to form expectations about an area of discourse which are appropriate only for a particularly salient surface-grammatical analogue of it. That is exactly Wittgenstein's diagnosis in the present case: the target analogy is that between the use of avowals and ordinary reports of observation. So then that diagnosis itself requires that the explanatory questions which we are required not to press in the case of avowals are, by contrast, perfectly properly raised, and unanswerable, in the case of ordinary reports of observation. There cannot, accordingly, just be a blanket prohibition against explanatory questions of that kind. Put that thought alongside the plausible claim that there are perfectly legitimate modes of conceptual explanation - informal mathematics, in particular, is full of them - and it appears that it cannot in general be merely a confusion to seek to explain features of the practice of a discourse a priori by reference to our concepts of the kind of subject matter it has and of the epistemic capacities of speakers. Thus the insistence that these questions are misplaced in the target case of psychological self-ascriptions begins to seem merely dogmatic. Is there any way this impression of dogmatism might be dispelled? In the analogy of the kaleidoscope, our conception is that of a range of independent features and events: evolving patterns of shape and colour to which the privileged observer is sensitive — responsive - by dint of his situation and his possession of certain germane cognitive capacities, notably vision. There is a story to be told about the kind of things on display and how things of that kind can elicit a response from someone with a suitable cognitive endowment. Now, one way to try to exculpate the Default View from the charge of dogmatism, it seems to me, is to seek a framework which places controls on the relevant idea of responsiveness." One form of control might be elicited from pursuing recently prominent issues to do with judgement-dependence and the Euthyphro contrast: we may pursue the details of the relations, in different regions of thought, between best opinion and truth, attempting thereby to arrive at a conception of what it is for them to relate too closely, so to speak, for their congruence to count as a success in tracking. 191 suppose this is a programme of what McDowell has disparagingly called 'constructive philosophy'. 119

Crispin Wright Another control might emerge from consideration of the question how wide the potential explanatory range has to be of a certain type of states of affairs if we are to think of our judgements about them as genuinely responsive to their subject matter at all {Width of Cosmological Role).20 We can seek a general framework of such controls and try to show that first-person psychological discourse emerges on the wrong side of the tracks, so to speak, under their application. Then, if its apparent urgency does indeed derive from a tacit assumption of the responsiveness of selves to their own psychological states, the general explanatory question about selfknowledge, which official Wittgensteinian philosophical method would have us ignore, might emerge as something which we can understand why we ought not to ask. By contrast, if that kind of project is dismissed, it is hard to see how the Default View can come to much more than a take-it-orleave-it recommendation: a mere invitation to choose to treat as primitive something which we have run into trouble trying to explain, and to do so just on that account. Wittgenstein notoriously came to view philosophical problems as akin to a kind of self-inflicted intellectual disease; they would thus contrast starkly with mathematical problems as traditionally viewed (not by Wittgenstein, of course) - a kind of sublime, objective puzzle whose force can be felt by any rational intellect. If philosophical problems are justly deflated in Wittgenstein's way, then a kind of 'Here: think of matters this way, and you'll feel better' remedy might be the best we can do. But the remedy seems enormously disappointing, intellectually. For most of us, after all, the attraction of philosophy is all about gaining understanding. Except in cases where one can explain a priori why the quest is inappropriate, it is apt to seem a mere abrogation of the subject to be told there is nothing to understand. VII Let me try to draw some strands together. We owe to Wittgenstein the insight that we are making an assumption in regarding it as a deficiency of understanding to lack a satisfactory explanation of the distinctive marks of avowals. The assumption is, roughly, that those distinctive marks must be consequential: that either they must derive from the nature of the subject matter — something which therefore drives our discourse about it into the relevant characteristic turns — or else they must derive from some unobvious feature of the seman20 Both these ideas are explored in my Truth and Objectivity (Cambridge, MA: Harvard University Press, 1994). 120

Self-knowledge: the Wittgensteinian Legacy tics of first-person psychological discourse (its being, for instance, expressive rather than assertoric). So, according to the assumption, there must be an explanation - which we have yet to assemble and get into focus. There is a frontal collision between this way of thinking and the conception of the nature of legitimate philosophical enquiry seemingly quite explicit in Wittgenstein's later official methodological pronouncements. According to Wittgenstein, the limit of our philosophical ambition should be to recognize the assumptions we are making in falling into philosophical difficulty and to see our way clear to accepting, by whatever means, that nothing forces us to make them. It is, for Wittgenstein, with the very craving for legitimizing explanations of features of our talk about mind, or rules, or mathematics, that we are led into hopeless puzzles about the status - the epistemology and ontology - of those discourses. Philosophical treatment is wanted, not to solve these puzzles but to undermine them - to assuage the original craving that leads to the construction of the bogus models and interpretations by which we attempt to make sense of what we do, but which are the source of all our difficulties, and yet whose want is felt as a lack of understanding. The problem of self-knowledge is a signal example. It can have — I believe Wittgenstein thought — no solution of the kind we seek; for that very conception of a solution implicitly presupposes that there must be a something-in-virtue-of-which the distinctive marks of avowals are sustained. But those marks are part of 'grammar' and grammar is not sustained by anything. We should just say 'this language game is played'. The generalization of this position - the 'estoppel' of all philosophical explanation - seems to me vulnerable to a version of what one might call the Paradox of Postmodernism. The paradox is that while, like all deflationists, Wittgenstein needs to impress us of the illegitimacy of more traditional aspirations, argument for that is hard to foresee if it is not of the very coin which he is declaring to be counterfeit. For what is needed here is precisely a philosophical explanation. To be sure, what belongs to 'grammar', in Wittgenstein's special sense of that term, requires no explanation. Of course; that's a matter of definition. But even a sympathetic reading of him will find a frustrating inattention to the question when something may legitimately be taken to be part of 'grammar'. It may be a crucial first step to recognize that the problem of self-knowledge is occasioned by an assumption of explicability - an assumption that may be discarded with a clear conscience if the special position of subjects in determining what is true of their psychology is indeed 'grammatical'. But, one wants to say, what shows that? Once one 121

Crispin Wright recognizes the Default View as a possibility, the immediate instinct is to ask: what might justify the idea that it is the whole truth? That is the instinct to attempt to understand when and why it is a good move to dismiss the attempt to understand. To succumb is to reenter the space of explanatory philosophy. To resist is to have no reason for the Default View. To feel this dissatisfaction is not to have a reason to deny the insight that in a wide class of cases philosophical perplexity does indeed take the form of a casting about for what strike us as satisfying explanations of features of our language and of failing to find any that do not generate singularities, of one sort or another. (Just briefly to mention a second prominently Wittgensteinian example: how are we to make sense of the intelligibility of the distinction between whether a statement is really true and whether anybody ever takes it to be true unless the rule incorporated in its truth condition may be thought of as issuing its verdict autonomously and independently of any human judgement? So isn't the very idea of unratified truth an implicit commitment to 'rules-as-rails' platonism?!) But to accept Wittgenstein's insight, that some of the hardestseeming philosophical problems take this form, is not a commitment to an explanation-proscribing view of philosophy. Even if it is misguided to persist in assuming that there must be something satisfactorily to take up the explanatory slack left by the demise of platonism, or Cartesianism, it may yet be possible to explain why such an assumption needn't be true in particular cases. It does not seem merely confused to seek, in particular, to characterize with some care the conception we have of the kinds of ways the marks of avowals might in principle be explained. It is even foreseeable that such a characterization might lead to a clear-headed realization that nothing could fulfil it. That would be - at least in this area - the discovery 'that gf-ves philosophy peace'.21 21 The phrase, of course, is from Philosophical Investigations §133. 122

Joint Attention and the First Person JOHN CAMPBELL I Models of Joint Attention It is sometimes said that ordinary linguistic exchange, in ordinary conversation, is a matter of securing and sustaining joint attention. The minimal condition for the success of the xionversation is that the participants should be attending to the same things. So the psychologist Michael Tomasello writes, 'I take it as axiomatic that when humans use language to communicate referentially they are attempting to manipulate the attention of another person or persons." I think that this is an extremely fertile approach to philosophical problems about meaning and reference, and in this paper I want to apply it to the case of the first person. So I want to look at the case in which you tell me something about yourself, using the first person, and we achieve joint attention to the same object. But I begin with some remarks about how this approach applies to proper names and to perceptual demonstratives. Suppose, for example, that you are explaining to me the role that a particular person plays on a particular committee. You may be presupposing an enormous amount of knowledge on my part about the other people on the committee and the work that it does. You are focusing on the history of the committee, how it was set up and who the founding members were, as well as the idiosyncrasies of the current members. We are achieving joint attention here, as you manage to highlight the way this individual fits into the structure. There is nothing particularly perceptual about this exercise of joint attention, beyond what is required to hear each other's speech; the operation of attention here has to do with the way in which background knowledge we have is being mobilized and put to work. It is a familiar idea that understanding a proper name involves having some dossier of information, which may or may not be correct information, about the bearer of the name. Using the name of the person we are talking about, in the above example, is a way of activating the right dossier, so that we each have the relevant dossier in play for the purposes of the conversation. Joint attention is not just a matter of us both, coincidentally, 1 Michael Tomasello, 'Joint Attention as Social Cognition', Joint Attention: Its Origins and Role in Development, ed. Chris Moore and Philip J. Dunham (Hillsdale, NJ: Erlbaum, 1995). 123

John Campbell attending to the same object. Rather, we know we are attending to the same thing, and aim to make it so. So we can use proper names and brief descriptions to make sure we are activating the relevant dossier, as when I say, 'John Wayne, the actor from all those Westerns'. For mutual understanding it is not essential that we have exactly the same information about John Wayne; it is enough if our dossiers derive from the same source and we know that they do. Contrast the case of perceptual demonstratives. If you say to me, 'Isn't that man over there an actor?', you have to point or look in marked kind of way, or otherwise give me some perceptual cue as to who you are talking about. The most common kind of cue we use is spatial: pointing is the most basic kind of perceptual demonstration. So achieving joint attention in this kind of case is a matter of spatial co-ordination. I may direct my gaze so that its line intersects with the direction in which you are pointing or the direction in which you are looking. Of course, spatial cues are not the only ones we ever use. If you say, 'What is that noise?', we may manage to achieve joint focus on the noise though neither of us can tell where it is coming from; it has other features which make it salient for both of us. Even in these demonstrative cases, though, where one component of shared attention is perceptual, the sharing of attention is not exhausted by the perceptual. There is more to joint attention than spatial co-ordination of gaze, for example: we have to be mobilizing common background knowledge and some attunement to one another's interests to be attending to the same thing. I turn now to my main topic, the case of joint attention achieved by the use of the first person. Suppose, for example, that you are telling me some story about yourself, about something that happened to you; or perhaps you are telling me about your current emotional instability, or perhaps about whether you would like tea or coffee. In these cases, you use the first person, and if the communication is effective, we achieve, however briefly, joint attentional focus on the same thing. My question is what we are doing when we do this; what it comes to, that we have achieved attentional focus on the same object. I am not questioning whether we actually do manage the thing; it seems to be a datum that we do, and it seems so easy and effortless, in the short term at any rate, that it can seem completely unproblematic. What is there to ask^about joint attention achieved through the use of the first person? Suppose we consider the two models that we have so far for joint attention: proper names and perceptual demonstratives. Can we apply either of them to the case of the first person? Suppose we look at the model of proper names. Here joint attention is a matter 124

Joint Attention and the First Person of our both having activated the relevant dossiers. Suppose ti*at you tell me, 'I have been hurt.' Is the joint focus here a matter of our having activated the relevant dossiers? For me to interpret your remark, I do have to know who it was that spoke. But I may never have met you before, so I may not have anything in the way of a previous dossier on you. There may be no dossier for me to activate; this looks much more like the case of a perceptual demonstrative. Certainly my identification of you may be perceptual-demonstrative identification. There is, indeed, the case in which I read some first-person remarks made by Bill Clinton, and to interpret them I have to bring my Clinton dossier to bear; but that is not the general case. And what about your own understanding of your statement, 'I have been hurt'? Is this a matter of you activating the relevant dossier? Of course, typically you will have some background knowledge about yourself. Typically you will have some kind of 'self dossier. But it is hard to believe that your understanding of your own remark can consist in your activation of the right dossier, as if T were a kind of proper name that you reserved for application to the person whose dossier this was. One reason is that although the case is unusual, you might be unable to activate any relevant dossier, yet still make comprehending use of the first person. If you are coming round after an accident, for example, you might say, 'Where am I?', or 'I have been hurt', without having any relevant dossier active at all. You still understand your own use of the first person, and you and your interlocutor may be achieving joint attention to the same person. This may make it seem that the model of the perceptual demonstrative is more to the point here. Does the joint attention consist in something like spatial co-ordination of gaze, as when you and I focus on the same place? It certainly is true that I have to know who is speaking to interpret your remark, and that may involve my focusing on the person at the right place; and even if we are talking on the telephone, I will be using some perceptually salient cue, like the fact of your speaking on the line, to direct my attention. But your attention to yourself is not of this type. You do not have to direct attention to any particular place in interpreting your own uses of the first person, and you do not have to be using any perceptually salient cue, like the pitch of your voice, to direct your attention to one person rather than another. So the model of perceptual demonstratives also seems to fail us. Joint attention in the case of the first person is not achieved by spatial co-ordination, nor is it a matter of the joint use of any other perceptually salient cue. At this point, far from seeming unproblematic, the securing of joint attention in the use of the first person may seem quite impos125

John Campbell sible. Your attention to yourself seems to be a kind of 'inner pointing', whereas my attention to you is either perceptually based, in the case in which we are talking, or draws on my knowledge of a proper name, as in the case in which I read Clinton's remarks. How could two such different types of attention be co-ordinated? So perhaps we do not in fact ever achieve joint attention through the use of the first person. But something has gone wrong here; joint attention achieved using the first person is commonplace. II Reference, Input and Output Let us draw back, and ask how we do understand the first person. Part of understanding is an ability to use the term. There is an enormous variety of inputs to and outputs from first-person judgements. We use the first person in judgements about our current location, current psychological states and objectives, memories, physical condition, and so on, which are made on a wide variety of different bases. And accepting a first-person proposition, like 'I've been swindled', or whatever, may have any of a wide range of implications for action or further thought. There ought to be more to our grasp of the first person than merely knowing how to use it, in grasping the inputs and outputs. There should be knowledge of the semantic value of the term, which would rationalize our use of those input and output procedures, which would explain why those are the correct input and output procedures to be using. The first person is a singular term, so a specification of its semantic value will be a matter of specifying the reference of the term. The truth or falsity of a judgement using the first person will be determined, in part, by the reference of the term. On the face of it, we have a very straightforward and absolutely reliable way of specifying the reference of the first person. The reference of the first person is given by the token-reflexive rule: The Token-Reflexive Rule: Any token of 'I' refers to whoever produced it. The problem now is to explain how the token-reflexive rule relates to the inputs and outputs of first-person judgements, the bases on which they are made and the consequences we draw from them. Straight off, it seems too much to ask that inputs and outputs to first-person judgements should in general be truth-preserving. We accept that we are fallible in many of the first-person judgements we make, in that even if we were right about the data on which we 126

Joint Attention and the First Person base our judgement, the judgement itself may be wrong. Similarly, even if the judgement is correct, the consequences we draw from it may be mistaken. But surely we could ask that the inputs to a firstperson judgement should be capable of yielding knowledge of the truth of the judgement, given that its truth-value is determined in the way suggested by the account of semantic value. This suggests what I will call the principle of concord: Concord: The bases on which judgements using a singular term are made must yield knowledge of the object assigned as reference. We might expect there also to be a complementary principle of effective action: Effective Action: The consequences of judgements using a singular term must be actions effectively directed to the object assigned as reference. And then the problem is to understand how the use of the first person can be rationalized in the sense explained by these two principles; how we can show that the assignment of semantic value made by the token-reflexive rule explains the correctness of the inputs and outputs we actually use in making first-person judgements. This problem is not just a problem for the theorist. Someone who understands the term T also grasps the semantic value of the term and finds the input-output procedures intelligible because he grasps the semantic value of the term. So an ordinary understanding of the first person does not just include an ability to make firstperson judgements in appropriate circumstances, and to act on them appropriately. On the face of it, you might do all that without having the slightest glimmering of the point of all these procedures; you might have no sense whatever of the reason why these procedures were correct. An ordinary understanding of the first person includes knowledge of the semantic value of the term, and knowledge of how that rationalizes the ordinary procedures of verification and consequence-drawing. It is when you have that understanding of the term that you can be said to be self-conscious; that is what self-consciousness is. So how are we to describe our ordinary knowledge of the semantic value of the first person? It is just at this point that it can seem so compelling to appeal to 'inner attention', which provides you with conscious awareness of the self as an object. Inner attention, so conceived, would let you know what you were about in making first-person judgements, it would let you know which thing it was that you were making judgements about. Once you had that knowl127

John Campbell edge, you would be able to see why the procedures you used actually conformed to these principles of Concord and Effective Action. And as I said, this idea of inner attention seems to make the possibility of joint attention to the self quite problematic. Ill Attention and Knowledge of Semantic Value It is a basic point of contrast between the first person and perceptual demonstratives that there is that single reference rule, the token-reflexive rule, which fixes the reference of the term over time, so that whenever the term is used the rule operates, and across speakers, so that whoever uses the term, the rule operates. Moreover, the input and output procedures that we use for firstperson judgements are constant across time and across speakers. So when we ask how the reference rule can rationalize the input and output procedures we use, we are not talking about a rationalization that has to be available at a single moment or for a single speaker; we can be talking about a rationalization that is essentially available only across time in a situation in which there are many speakers using the first person. To find an alternative to the 'inner attention' approach, we have to find an alternative way in which the subject could find intelligible the particular input and output procedures we use for first-person judgements. It is easiest to explain the alternative I have in mind if we have before us a concrete example of the kind of input and output procedures I have in mind. So I will first state a particularly simple and central case: the inputs and outputs from judgements about one's own location, and then consider how knowledge of the reference of the first person can rationalize the use of these procedures. There is a basic distinction that we have to draw between what I shall call relational and what I shall call monadic egocentric spatial notions. Relational egocentric notions are those that we use when we say, for example, 'he is sitting on my left', 'the chasm yawned before him', 'look behind you', and so on. These notions specify the person whose right or left, up or down is in question. They are twoplace notions: 'x is to y's left', 'x is below y\ and so on. Now in stating the spatial content of vision, we do not seem to need these relational notions. We do not need the general conception of something's being to the right or left of an arbitrary subject. Rather, we need the much more primitive monadic egocentric terms. These are notions such as 'x is to the right', 'x is below', and so on. An animal could quite well have spatial vision even though it did not have the 128

Joint Attention and the First Person relational egocentric notions; it could not represent anyone else's left or right, only its own. But it is not even as if its vision makes explicit the spatial relations that things bear to it; it is not always itself an object in its own visual field. Its vision represents things as 'to the right' or 'above'; it does not seem correct to say it represents things as 'to my right' using the relational notion, because of the lack of generality in whose right or left can be represented. And the same seems to be true of ordinary human vision. It represents things as 'to the right' or 'above' using the monadic egocentric notions, rather than the relational terms. When we learn the first person, we learn a procedural rule: that if vision represents an object as to the right, we are then in a position to say, 'The object is to my right.' We learn how to make relational judgements involving the first person on the basis of monadic spatial input. We also learn how to act on the basis of relational judgements involving the first person. This at some point will involve the production of a 'motor image' detailing the spatial organization of the planned movement; and we may use relational judgements involving the first person in constructing such an image. For example, if I have formed the judgement that x is to the left of me, and I want to pick up x, then I have to form a motor image which directs reaching to the left. In effect, then, we have an introduction and an elimination rule for the use of the first person. The introduction rule takes us from the monadic egocentric representation we have in vision, to the relational judgement involving use of the first person: VISION: x is to the left . hence, x is to the left of me The elimination rule takes us from a relational judgement involving use of the first person, to the construction of a motor image using only monadic egocentric notions: x is to the left of me . MOTOR IMAGE: hence, to pick up x, move to the left The question now is whether we can find a semantic justification for these rules. If we say that the first person is subject to the tokenreflexive rule, can we use that point to rationalize our use of these procedures for making first-person judgements and drawing implications for action from them? In the terms I just used, we have to show that the principles of concord and effective action are observed here. Suppose someone has, in vision, a monadic spatial representation 129

John Campbell of an object as being to the left. If he consequently forms a judgement, 'the object is to my left', does the token-reflexive rule give any reason to think that his judgement may constitute knowledge? If someone has, in vision, a monadic spatial representation of an object as being to the left, does that in any way confirm the idea that the person who has the visual representation, if he forms a judgement, 'the object is to my left', will be forming a true judgement? By the token-reflexive rule, what this requires is that the person who forms the judgement, the person who produces the token of 'I', should have the object on his left. Evidently the critical point here is that the person who has the visual representation should be the same as the person who makes the first-person judgement, and that visual-monadic right and left should in general be good guides to what is to the right or left of the person who has the visual representation. How can you achieve an understanding of why this is in general true? In your own case, you can determine what is monadically right or left, but you have no better check on what is relationally to your right or your left than by the use of the rule. So there does not seem to be any understanding here of why monadic right and left are good guides to relational right and left. And in the case of other people, you can check what is relationally to their right or left by using visual-attentional skills to set up a frame of reference centred on the other object, but how can you check which monadic spatial representations the other person has? The whole point about a monadic spatial representation is that it is available only to the person who has that particular perceptual system. Suppose you are trying to grasp the egocentric positions of things from someone else's point of view. One way to proceed is to use relational egocentric notions. That is what we are doing when we say, 'it's behind you!', and so on. But there is another possibility. You might form an image of the scene from the other person's position. The spatial content of the image is given in monadic spatial terms; that the image is from the other person's position is not itself part of the content of the image. You are forming the image as part of the project of grasping the indexical facts from another viewpoint, but that this is the project is not itself part of the content of the image you form. To say that, you have to step back from the formation of the image, and use some other method of representation. So here there is a two-level construction: at one level there is the imagistic representation of the monadic egocentric facts, and at the other level there is the statement of which person it is that has the image. Why should we bother with this kind of two-level construction? It has to do with our need to use ways of representing the world, 130

Joint Attention and the First Person such as monadic egocentric spatial notions, that involve a viewpoint only implicitly, where the viewpoint from which the world is being described cannot be made part of the description. These ways of representing the world include perceptual or imagistic representations, where which person is having the image is not itself part of the content of the image. So if you are to use these ways of representing the world to indicate how things stand from viewpoints other than your current viewpoint, you will need a two-stage construction. At the first stage you describe which viewpoint is going to be represented, and at the second stage you use the implicitly viewpointed system to describe how the world is from that viewpoint. The two stages cannot be collapsed together, because the implicitly viewpointed system will resist any attempt to make it incorporate an identification of which viewpoint is being described. We are all part of a community capable of this kind of imaginative construction, and that is what we use to explain and understand the correctness of the transitions back and forth between monadic and relational egocentric spatial notions, the introduction and elimination rules for the use of the first person. We can use this procedure to achieve an understanding of why it is that the introduction and elimination rules being used by another person do yield knowledge and facilitate action, in the way demanded by the principles of concord and effective action, with respect to the object assigned as reference by the token-reflexive rule. Each of us can use this imaginative procedure to understand why in general the pattern of use of the first person is correct, given that it is governed by the tokenreflexive rule. And this understanding does not need to appeal to any conception of the reference of the first person other than that given by the token-reflexive rule. In particular, there is no need to appeal to 'inner attention'; it does no work in rationalizing the input-output procedures we use for the first person. In fact, it does no work at all. IV Inner Attention When we make perceptual-demonstrative reference to an object, we do so on the basis of attention to it. If I want to refer to an object we can both see, I have to draw your attention to it, and my own reference to the object depends on my attending to it. When we think about self-consciousness and self-reference, it is easy to think of this in terms of attention too. Self-reference, on this view, depends on attention to the self. Views of self-consciousness then divide. Is reference to the self a matter of attending to a physical object, or is it 131

John Campbell a matter of attending to the psychological? As I have said, I think that we should resist the assimilation of the first person to the perceptual demonstratives, and in particular, to resist the idea that the use of the first person should be thought to depend on attention to the self, whether understood as attention to the physical or attention to the mental. This seems to me to be the deep divide between perceptual demonstratives, where there is no systematic rule determining the reference of the term, only one's attention to the object, and token-reflexive terms, such as the first person, which are governed by systematic rules, such as 'Any token of the first person refers to whoever produced it.' What is wrong with the idea that reference to yourself is a kind of 'inner pointing'? One possibility, which a Cartesian would favour, is that attention to oneself is attention to a purely psychological object. This would be a non-spatial, non-physical object which bears only psychological properties. But it is not easy to see how you yourself could ever attend selectively to such an object. How could you know that you were singling out one rather than any other such object? How do you know you always single out the same such object in an act of inner attention? And how do you know there is only one such object to be the bearer of all the psychological properties you take yourself to have? It seems impossible that we could find any account of the principles by which such inner attention might work. There is a certain suspension of disbelief when we are considering this kind of picture of self-reference, anyhow, because as Hume remarked, we do not ordinarily take ourselves to be encountering any such exotic object; hence his comment that when he entered into himself, he encountered only particular perceptions. This might suggest an alternative picture of inner attention, on which it is not a matter of attending to a purely psychological substance, but a matter of attending to what Hume called particular perceptions. Since Hume thinks of the self as whatever it is that is the object of our attention in introspection, he concludes that the self is nothing but a bundle of perceptions. Wittgenstein's radical contribution to the subject was to remark that attention to one's own perceptions is not a matter of attention to a particular kind of thing. Rather, it is a matter of attending to the ordinary physical objects around you; what turns your report into a report of your own psychological state is a particular twist that you give it. If you look at the sky and say 'How blue the sky is!', it is quite clear that your attention is directed to the common sky; and if you give that report the twist that turns it into a report of your perception, by saying, for example, 'I see the sky as blue', this 132

Joint Attention and the First Person is not a matter of you finding some new, inner object of attention. Rather, you have continued to attend to the sky, but in your report you have made use only of your current vision, giving the report you would have made without the benefit of any other information you might have. Prefacing your report with 'I see...' is just the way in which you signal that the following judgement was made on a purely visual basis. The problems of privacy, according to Wittgenstein, all arise because we mistake this kind of twist in the report for the discovery of a new, inner object of attention, rather than the external sky. Once we recognize that the only object of attention to be found is the external scene, the problem of the inverted spectrum, for example, vanishes; for the colour of the sky can be seen by all. A similar point can be made about belief: if I am trying to find out whether I believe that p, I do not do it by attending to a special kind of inner object, which somehow indirectly bears on how things are in the world around me. Rather, to determine whether I believe that p, I use the following procedure. First, without stirring, I ask whether p is so. If the answer is that p is so, I make the report: 'I believe that p.' This only involves attention to whether p. At no point do I have to exercise some inner attention on a purely psychological realm. Any account of self-reference is going to have to give a central place to this ability to twist our reports of what is around us into reports of psychological state. For the moment, though, the point is that self-reference cannot be thought to be a matter of introspective attention to a particular kind of psychological object, the self. You might object that we can continue to think of the first person as like a perceptual demonstrative, so long as we abandon the idea of psychological-introspective attention, and view it instead as a matter of inner attention to your own physical states — to the way your body is. There certainly is such a thing as somatosensory attention, and it certainly is linked to our sense of ownership of our bodies. One way to bring this out is to remark that there are some neuropsychological conditions in which the patient does not recognize that he or she is the owner of various body parts - the left arm, for instance, may seem alien. What makes the difference between the case in which you do experience the limb as yours and the case in which you do not experience the limb as yours? It has been suggested what makes the difference is the possibility of somastosensory attention to the limb.2 This link between somatosensory attention 2 Marcel Kinsbourne, 'Awareness of One's Body', in The Body and the Self, ed. Anthony Marcel and Naomi Eilan (Cambridge, MA: MIT Press, 1995). 133

John Campbell and the sense of ownership may seem to support the idea that grasp of the first person depends on somatosensory attention. In fact, though, the relation of dependence seems to be round the other way. We can make a distinction between non-conceptual and conceptual somatosensory attention. Non-conceptual somatosensory attention is available to an animal or child, when pain directs its attention to an injured body part, for example. This need not involve thinking about the body part at all, and it may involve no exercise of self-consciousness; it is just that the animal is now poised to protect the body part. This kind of non-conceptual attention is evidently not enough for first-person thought. In contrast, it can happen that someone capable of thought has somatosensory attention directed to a body part, and thinks about that body part. But this kind of somatosensory attention does not make available new ways of thinking about body parts, as 'that hand', for example. (In contrast, seeing a hand really does make it possible to identify the thing in a new way, as 'that hand'.) Somatosensory attention to one's hand, if it involves thought about the thing at all, involves thought of it not as 'that hand', but as thought about, e.g., 'my right hand'. Somatosensory attention uses a system for identification of the body parts which is available prior to the individual acts of attention, and which presupposes, and so cannot explain, our use of the first person. If we suppose that the reference of the first person is fixed by somatosensory attention, we would have to think of T as meaning something like 'whoever's hand this is', 'whoever's foot this is', and so on. But that would mean that the notion of ownership could be explained prior to an understanding of the first person; that in advance of an understanding of the first person, you could know that it means to say that this or that body part belongs to one or another person. And this does not seem to be correct: the only conception of ownership we have that the owner of the hand is the one who is entitled to speak of it as 'my hand', and so on. V Joint Attention and the Token-Reflexive Rule I began with the problem of describing what it consists in, that joint attention can be secured in an ordinary conversation by the use of the first person. We saw that the models of co-ordination of background knowledge and perception do not seem to work in this case, and turned to look in detail at how self-reference and attention to oneself might be related. I have argued that we cannot view self-reference as depending on inner attention, whether that is thought of 134

Joint Attention and the First Person as psychological or as somatosensory attention. Rather, there is only the use of the first person, subject to the token-reflexive rule, and our imaginative understanding of each other to rationalize the input and output procedures we use. The question I have been addressing has to do with how deep the token-reflexive rule goes in describing our understanding of the first person. According to the view I have been opposing, there is nothing deep about the token-reflexive rule. It is true that our use of the first person is governed by this rule, but, on this view, it tells us nothing about what joint attention achieved through the use of the first person consists in. On this view, the use of the token-reflexive rule is only instrumental to achieving an understanding of particular uses of the first person. When you hear someone else using the first person, you know that the person is using a term subject to the token-reflexive rule, but that tells you nothing until you know who it is that is speaking; until you have some other way of identifying the speaker. Then you can use that identification to interpret the first-person statement. Suppose you hear someone saying, 'I have been hurt'. You find that the speaker was Gustav Lauben. Lauben said, 'I have been hurt', so to interpret the use of 'I', you strike out the 'I' and write in 'Lauben'. Similarly, to understand your own use of the first person, it is not enough that you should be using a term subject to the token-reflexive rule, and employing the right input-output procedures. You need to have some further identification of yourself, so that you can interpret your own uses of 'I', using some 'inner demonstrative' like 'this person'. It is when you are following this line of thought that the appeal to inner attention seems absolutely inescapable. It is also this line of thought that makes the phenomenon of joint attention achieved through the use of the first person seem quite impossible. The alternative I have been recommending is that the speaker using the first person does not need any other kind of self-identification than is provided by the use of the first person, subject to the token-reflexive rule, and grasp of the ordinary grounds on which we make first-person judgements and their consequences for action. The puzzle this raises is how we can use knowledge of the reference of the first person to explain the correctness of the input and output procedures we ordinarily use. The desire to understand this is, I think, at the heart of the appeal to 'inner attention'. But I have tried to show that we can explain the correctness of the input and output procedures we actually use by employing our capacity for an imaginative understanding of each other, and our knowledge that we are ourselves targets of imaginative understanding. 135

John Campbell On this view, an understanding of someone else's use of 'I' does depend on knowledge of who it was that spoke; you need some such further identification if you are to attend to the right person. But an understanding of your own use of T does not depend on any further identification at all, and in particular, not on an identification that appeals to inner attention. Rather, your attending to yourself just consists in your use of 'I', subject to the token-reflexive rule, and in accordance with our ordinary input-output procedures. So the phenomenon of joint attention achieved through the use of the first person is straightforward. All that is required is a use of the first person in a context in which it is manifest to the audience who the speaker is; a context which immediately makes available some further identification of the speaker. It is by finding this further identification that the audience manages its side of the joint attention. But the speaker has no further identification to find; the speaker discharges his side of the joint attention simply by continuing to use the first person. 136

Consciousness as Existence TED HONDERICH I Leaving Consciousness Out, or Trying to The difference for present purposes between ourselves and stones, chairs and our computers is that we are conscious. The difference is fundamental. Being conscious is sufficient for having a mind in one sense of the word 'mind', and being conscious is necessary and fundamental to having a mind in any decent sense. What is this difference between ourselves and stones, chairs and our computers? The question is not meant to imply that there is a conceptual or a nomic barrier in the way of non-biological things being conscious. It may happen one decade that the other minds problem will shoot up the philosophical agenda and get a lot of attention as a result of a wonderful computer attached to perceptual and behavioural mechanisms, and that the thing will in the end be taken as conscious, rightly. Our question is not what things can be conscious, but what the property or nature of consciousness is. Conscious or mental events, as we know them now, are in some kind of necessary connection with neural events. This fact of psychoneural intimacy, which is consistent with what has just been said of the possibility of non-biological things being conscious, provides the best argument for strict or true identity theories of consciousness. These take the property of consciousness to be a neural property, or, as we can say instead, take conscious events to have only neural properties. The objection to these theories seems to me not that they make conscious events physical. I take it that in a good sense of 'physical', definitely not the indeterminate one relativized to the science of the moment or to future unknown science, conscious events are indeed physical. That is, they are either in the category of things that are spatio-temporal and perceived or the category of things that are spatio-temporal and are nomically connected with spatio-temporal things that are perceived. Stones and the like are in the first category, particles and the like in the second.' My thanks for comments on an earlier draft of this paper are due to Bill Brewer, John Campbell, Geert Engels, Alastair Hannay, John Heil, Jennifer Hornsby, Bob Kirk, Jonathan Lowe, Paul Noordhof, Ingmar Persson, Ingrid Coggin Purkiss, and Barry Smith. They have not yet been converted to Consciousness as Existence. 1 For an exposition of this fundamental conception of physicality, see Anthony Quinton, The Nature of Things (London: Routledge & Kegan Paul, 1973) pp. 46-53. 137

Ted Honderich What disposes very many people against strict identity theories is of course that our experience of conscious events, the having of them, leaves us thinking that they have a property or nature other than the properties or nature had by wholly neural events - transmitter-substance properties and so on. Strict identity theories leave something out. They seem to me to leave out not something elusive, or something diaphanous, or something peripheral, but the reality of our mental lives. They leave out the most immediate of all the facts we know. Hence many of us feel that psychoneural intimacy must be accommodated by a means other than asserting that conscious events have only neural properties. All other identity theories, the lenient or arguable ones such as Donald Davidson's Anomalous Monism and also the Union Theory on which I am keen,2 all of which bear the slight burden of being called propertydualisms, raise the very question we are considering. They allow that consciousness brings in something non-neural. What is it? Conscious or mental events as we know them also have causal roles. That is, they stand in many kinds of necessary connections with input and output. Some desires stand in necessary connection with things that have been perceived and with subsequent acquisition-behaviour. Some pain stands in necessary connection with certain sensory stimuli and avoidance-behaviour. There are also distinctive connections in the case of thinking, perception and so on. Here is a respectable and daunting subject-matter in itself, worth the diligence invested in it by cognitive scientists. But the basic fact, conscious events being many kinds of effects and causes, is also used to provide the argument for what can be called strict or philosophical cognitive science and functionalism. These doctrines are distinguished by taking conscious events in general to be nothing more than the many kinds of effects and causes. It is fundamental to these doctrines, so long as they remain philosophically distinctive, that there is nothing further to be said of the nature of these events, anything else they have in common. Stated in this summary way, deprived of obscuring elaboration, strict cognitive science and functionalism are open not only to the objection that they leave out the reality in or of our conscious lives, but to another objection that seems insuperable. Stones, chairs and our computers, considered in themselves, also involve events which are kinds of effects and causes. There is also the event of, say, my own unnoticed little gain in weight, which 2 Donald Davidson, 'Mental Events', in Essays on Actions and Events (Oxford: Clarendon Press, 1988); Ted Honderich, A Theory of Determinism: The Mind, Neuroscience and Life Hopes (Oxford: Clarendon Press, 1988), chapters 2—3, or Mind and Brain (Oxford: Clarendon Press, 1990), chapters 2-3. 138

Consciousness as Existence never comes to my mind. To put the objection in one way, how did strict cognitive science and functionalism then find, or discriminate from other things, their general subject-matter, the large but of course limited range of kinds of effects and causes with which they are concerned? How did these doctrines separate off irrelevant events, in particular irrelevant events in us? Nothing is more essential to the doctrines. We need to know what they are talking about. Put differently, what is their general conception of consciousness? Evidently to speak only of kinds of effects and causes will not work. There are very many kinds, of which I have just mentioned a few, that do not and must not get into the story. It therefore seems these doctrines were covertly from the beginning, or must now collapse into, the view that conscious events are events in only some causal sequences — not any old causal sequences, Which ones then? The only possible answer seems to be the sequences involving consciousness in another sense. What is necessary is a characterization of conscious events in addition to the insufficient one which almost all of us accept, that they are certain effects and causes. This raises exactly the question we are considering. On reflection, this same objection of incoherence can be made to strict identity theories. Which wholly neural events are the conscious ones.3 You may be made uneasy by all this, on account of what it could lead to. Your uneasiness will not be reduced if the property of conscious events missed out by the theories we have glanced at is named in advance the property of real subjectivity. The picture in the offing, given my physicalism avowed at the start, seems to be of our heads having two kinds of properties or events in them, the first being neural and the second non-neural although physical, Properties of the second kind are perhaps not rightly to be abused as ghostly stuff, but they are bad enough. The idea is that there exist properties or events which, although physical and in the causal and nomic web, are not at all akin to kinds now accepted, but properties or events whose actual discovery would transform or overturn neuroscience as it is. If this idea is perhaps not an awful one, it is alarming enough.4 3 For more of the incoherence objection to functionalism, see 'Functionalism, Identity Theories, The Union Theory,' in The MindBody Problem: The Current State of the Debate, ed. T. Szubka and R. Warner (Oxford: Blackwell, 1994). 4 I was driven, alas, to tolerate or anyway contemplate this sort of thing in 'Consciousness, Neural Functionalism, Real Subjectivity', American Philosophical Quarterly 32/4 (October 1995), 379. It is to be confessed, too, despite pp. 64—5 of 'Seeing Things,' Synthese, 98 (1994), that in that paper there is a great deal that seems, as you might say, in another world from the view that follows here. But it wasn't all wrong. Some of 'Seeing Things' would have echoes in a further development of the present view. 139

Ted Honderich ' I hope in the end to be able to reassure you. That is, I hope it will be possible to maintain what is unquestionable, that conscious events have more than neural properties and particular causal relations, without adding to the kinds of properties or events we already know to be in our heads. Those are the kinds allowed in contemporary neuroscience. Further, since conscious events will be taken to involve more than what goes on inside heads, I hope the view will not add to the kinds of properties and events we already know to be outside our heads. Those are roughly the kinds allowed by ordinary experience and contemporary science. What is needed is not more things, but a different way of looking at or categorizing the ones we have. II The Existence of a World The difference between me now and a chair in this room, it can be said, is that for me a world exists, and for the chair a world does not exist. Or rather, as I prefer to say, my consciousness now consists in the existence of a world. The rest of this paper will have to do with understandings of this seemingly metaphorical sentence. It is owed to contemplating consciousness directly, despite its obscurity. This policy of mental realism may unsettle some philosophers in the current philosophy of mind, since they are averse not only to dualisms which no one should contemplate,5 but also to the mystery which is the fundamental question of the nature of consciousness. But you, if you are in a way stronger-minded, may share the hope that the sentence points in the right direction, does indicate the nature of consciousness. That is, the sentence may express more than one proposition, and the hope is that one of them, a literal one, will really shed light on the nature of consciousness. It seems to me that trying to dissipate this mystery is better than recoiling from it. The sentence can naturally be taken for another one. It is that all of my consciousness now, including any thoughts unprompted by this room, maybe some day-dreaming, consists in the existence of a world. Perhaps we can on some other occasion get to an account of consciousness generally, or all of the consciousness of one person, which is pointed to by that sentence. But on this occasion let me limit myself to something else, my perceptual consciousness 5 The most numbing of these dualisms, perhaps, well beyond ghostly stuff, is to be found in K. R. Popper and J. C. Eccles, The Self and Its Brain (Berlin: Springer, 1977), where to conscious and neural events a Self is added. 140

Consciousness as Existence now — my consciousness in so far as it consists in my seeing, hearing and so on. So what we have is that my perceptual consciousness now consists in the existence of a world. Let us think about only this, and trust, as seems reasonable, and in accord with several philosophical traditions, that perceptual experience is the base of all consciousness, and that on some other day an understanding of it will be used to explain the rest. Thinking of my perceptual experience now as consisting in the existence of a world needs to be distinguished from and may be more promising than another piece of mental realism, a well-known one. Here, my perceptual consciousness is characterized as part or most of what it is like to be something, or what it feels like to be something.'' What strikes me as wrong with these locutions, if they are intended seriously, as being on the way to a general understanding of consciousness, and of course not just about differences between conscious things or states, is that the analysandum is right there in each analysans. The locutions surely presuppose and depend for their understanding on what some supporters of them assert, that there is not something which is what it is like to be a stone, chair or computer, and of course not something which is what it feels like to be one.7 Does the familiar piece of mental realism not come to this, then, that my perceptual consciousness is characterized as part or most of what it is like to be conscious, or to feel conscious} This is of no use to us, no analytic help. But does the sentence I am promoting, 'My perceptual consciousness now consists in the existence of a world', share a different disability with talk of what it is like or feels like to be something? (I postpone for a little while the very large question of whether it shares the first disability, being no analytic help.) You may grant that conceiving of my perceptual consciousness as amounting to the existence of a world points at something, indicates the nature of something. But, you may say, that thing, as in the case of talk of what it is or feels like to be something, is only the phenomenology of consciousness. It is only consciousness as it seems or appears to be, not the reality of it. This objection may amount to one of several things. It may simply be insistence on strict identity theory or strict functionalism and cognitive science as the truth about conscious6 Thomas Nagel, 'What Is It Like To Be A Bat,?' The Philosophical Review 83 (1974), reprinted in his Mortal Questions (Cambridge: Cambridge University Press, 1979). See also Timothy Sprigge, 'Final Causes.' Supplementary Proceedings of the Aristotelian Society 45, (1971). 7 John Searle, The Rediscovery of the Mind (London and Cambridge, MA: MIT Press, 1992), p. 132. 141

Ted Honderich ness. Or perhaps insistence on those doctrines lightly amended by something about qualia. The latter over-worked items, I take it, are elusive differences between kinds of perceptual consciousness.8 They are 'feels' rather than contents, or more of the nature of 'feels' than of contents, and very evidently not the character of all of consciousness. These amended doctrines, as you will anticipate, have not been sufficiently amended to satisfy me. They too leave out an explicit and general account of what is fundamental about consciousness. As you will gather, my sentence about the existence of a world is not an assertion of the existence of qualia. No doubt they exist, but they are not the general nature of perceptual consciousness. If the phenomenology objection fails when taken in this way, is there a better way? Is there a better reason than the given doctrines for dismissing my sentence as only talk of the appearance of consciousness? Well, the dismissal may be misleadingly expressed, but be intended as conveying that there is some other fact about consciousness more important than anything conveyed by the sentence - say the relation of conscious to neural events, or the causation of consciousness, or the role of consciousness in the explanation of behaviour. There is also the truth already indicated, which strict functionalism and cognitive science wonderfully exceeded, that kinds of conscious events, say desire, pain and thinking, and subkinds of them, are differentiated by their causal connections, and could not be characterized adequately without reference to those connections. But surely none of this, although it involves disagreements about what is important, amounts to the proposition that the general conception of consciousness we are contemplating is of only the appearance of it, not its reality. In fact this proposition, if taken literally and not as a misleading expression of other things, seems to presuppose a falsehood. It is that we can attach sense to talk of a reality-behind with respect to consciousness itself. Things, say stones, chairs and computers, may of course be otherwise than they seem, but that is not a distinction within consciousness. The distinction presupposes consciousness, our having different views of things, but what it has to do with or is about is the chairs, stones and computers. If we stick to consciousness, is it not the case that all there is, in so far as it itself is concerned, is what is being misdescribed as an appearance? Is it not the case that all there is, in so far as consciousness itself is concerned, is what is pointed to by my sentence and perhaps related ones, and also, despite its disability, by talk of 8 Thomas Nagel, 'Qualia', The Oxford Companion to Philosophy (Oxford: Oxford University Press, 1995). 142

Consciousness as Existence what it it is or feels like to be something? Consciousness, after all, is what we have. And what we don't have in this sense isn't consciousness. Also, we don't have it in two ways. Certainly we can't get behind or beyond consciousness itself by introspection or recollection and bring back a hidden part of it. There isn't any other experiential access to it than the single one we've all got. The only conceivable other access, so to speak, would be a theory about it. But, so far as I know, we haven't had any philosophically successful theory about it, the reality of it, as distinct from about its causation, explanatory role, other relations, kinds of it and their differentiation, secondary features of it, and so on.9 The theories that do seem to be about consciousness itself, having to do with about - ness or intentionality, cannot be regarded as successful. None has come to the fore. It is worth adding, finally, something implied by what has just been said, that if there are ways or techniques of bringing things into consciousness, perhaps dispositions of ours of which we have been unaware, these are not a different access to consciousness itself. As for those very dispositions, often called the subconscious or the unconscious, evidently they are not in or part of consciousness. No doubt they are neural. To repeat, what we don't have isn't consciousness, and we don't have it in more ways than one. So much for the objection that my sentence points at only the phenomenology of consciousness. Let me now make a start on the inevitable objection that it is of no analytic help. Saying that my perceptual consciousness now consists in the existence of a world, if this is understood in certain ways, will indeed be of no help. For a start, it cannot usefully come to just this, that my perceptual consciousness consists in my seeing, hearing and otherwise sensing what exists around me spatio-temporally. If the sentence is taken this way, it will be useless, a really overt instance of the analysandum turning up as the analysans, the analysans being no advance on the analysandum. We already understand perceptual consciousness to be seeing, hearing and otherwise sensing spatio-temporal things. That is the ordinary content of talk about perceptual consciousness. That is what we are trying to improve on. This objection of uselessness, of course, is likely to come up for a particular reason. Perceptual consciousness, according to the sentence, is the existence of a world. Furthermore, it was first said above 9 It is the burden of my 'Consciousness, Neural Functionalism, Real Subjectivity' that Searle's admirable attempt to characterize consciousness in The Rediscovery of the Mind does not come to grips with the fundamental reality of it. 143

Ted Honderich that the difference between me and a chair is that for me a chair exists. Both those sentences can indeed be taken to suggest that the idea is that perceptual consciousness is more than a world - it is the world's existing. And, the thought continues, all that that can mean is that the consciousness consists in a world's being seen, heard, etc. Well, that is not the hope. There is some heuristic advantage in saying, as I shall sometimes persist in saying, the perceptual consciousness is the existence of a world - that might just wake us up to something we have been missing or mislaying - but no more is being suggested than is also suggested by saying, simply, that perceptual consciousness consists in a world. There is something else to be put aside. We would not get anything useful by interpreting my sentence as taking perceptual consciousness to consist in awareness of subjective things — representations, sense-data or the like. This would amount to giving the particular account of perceptual consciousness which is the representative theory of perception or phenomenalism. My reason for saying that giving this account would not help is not that to do so would be to impose on the sentence a theory supported only by doubtful arguments, although this is surely true, and my thinking so will inform some later comments. The reason we would get nowhere is that in this interpretation of the sentence, what we would have is that perceptual consciousness is awareness, if in an obscure sense, but it is indeed awareness that we are trying to understand. 'Awareness' in the obscure sense is not synonymous with 'perceptual consciousness', but it is too close for comfort. We would get no understanding of perceptual consciousness itself by being directed away from certain objects of it, objective ones, and towards other supposed objects of it, subjective ones. This reason for not imposing the representative theory on our sentence is equally a reason for not imposing on it direct or naive realism, the theory which grows out of what was mentioned a moment ago, the ordinary content of talk about perceptual consciousness. It is to the effect that perceptual consciousness consists not in the awareness of subjective but rather of objective things. Let us say that such things, unlike representations and sense-data, are public, which is to say perceivable by more than one person, and are perceivable by more than one sense, and also exist unperceived.10 Plainly this different theory, first of all in speaking of awareness, also contains the problem. We need to approach the problem on our own. 10 The distinction is taken from A. J. Ayer, The Central Questions of Philosophy (London: Weidenfeld & Nicholson, 1973). 144

Consciousness as Existence III A Mental World? It happens near the start of our lives that each of us does what each continues to do afterwards, distinguish herself or himself in a particular way from all else, all other things and persons. Each of us comes into possession of the fact of something unique and persistent in a life, certainly not a body, or all of a body. Each of us comes to have some kind of sense of subject, self or person - a sense of oneself, as we can say. This claim can be true, of course, without our having a respectable theory of consciousness or relying on daring philosophical theories of the self. We do not have to swallow Descartes in order to have senses of ourselves. More will be said about a subject later but let us for the moment rely on what we all have in order to state the first of five considerations bearing on what has been said so far and seeming to point in a particular direction. 1. The particular subject each of us senses enters into the existence of a world, a person's perceptual consciousness as I understand it, in a certain way. The essential thing for now is that this state of affairs could not exist in the absence of the subject. The particular subject is a necessary condition of the state of affairs. It is so because it is in some manner a part of it. But that is not all. In the absence of the subject, there would not exist anything of the world whose existence is what perceptual consciousness consists in. Such a dependency on a particular subject is not true of three larger worlds, the first being the one that is physical in the sense mentioned earlier.11 This is the world, of which much will be said, that is spatio-temporal and has perceived properties or is in nomic connection with things with perceived properties. This world does not have the mentioned dependency. The part that is perceived is not dependent on any particular subject. And the part that isn't perceived is also not dependent on any particular subject — it doesn't enter into perceptual consciousness at all. There is the same want of dependency on a particular subject with a second world, with which we shall also be concerned. This is one lately in view, the objective world. It has in it things perceivable by more than one person, and perceivable by more than one sense, and such as also to exist unperceived. Evidently it shares a feature or two with the physical world as defined. Finally, there is the same want of dependency on a particular subject with a third world, also in view earlier. This is the world of things in current or anticipated science, an indeterminate world to say the least. Let us call these three worlds mind-independent worlds. I trust it will be clear, incidentally, that speaking of these various II See above, p. 137. 145

Ted Honderich worlds, of which we now have four, is not to be taken as indulgence in any sort of ontological extravagance. In the primary and most ordinary sense of the word, there is but one world. Of it or of some of it, we have different conceptions. What falls under a conception is, in my secondary sense of the word, a world. My endeavour in this paper, as will become plain, is to see relations between several conceptions and worlds, and to recommend one conception and world in connection with consciousness. 2. There is also another dependency that needs to be attended to. If I take my perceptual consciousness now to consist in the existence of a world, this necessarily is a world which also has a second dependency seemingly different in kind from the one on a particular subject. It is hard for me to resist the conclusion that the correct understanding of the fact of psychoneural intimacy mentioned at the start is not the strict identity theory but the theory that consciousness is in nomic or lawlike connection with neural events, events with only neural properties. Although the story of the Union Theory gets complicated, part of it is that my perceptual consciousness has a dependency on, has a kind of nomically necessary condition in, my simultaneous neural events.12 So the world that is my perceptual consciousness, for this second reason, cannot be the physical world as understood, or the objective world, or the world indicated by science. 3. There is something else, another part of the story of lawlike connection between consciousness and simultaneous neural events. A neural event is not only a kind of necessary condition but also a nomic correlate of a conscious event. That is, although the conscious event is not an effect of the neural event, it is true that given the occurrence of the simultaneous neural event, the conscious event necessarily happened. In a traditional terminology, the neural event was not only a kind of necessary condition for the conscious event, but also a kind of sufficient condition. We can say the neural event was a guarantee of the conscious one. Of course these considerations having to do with the brain, together with what should be added about dependencies in the other direction, of brain on consciousness, go against some engrossing and influential doctrine which includes a denial of the existence of psychoneural laws.13 But allow me to take psychoneural lawlike connection for granted on this occasion. If it does not exist, by the way, that will certainly be bad news for neuroscience, since standard neuroscience certainly presupposes it. Shouldn't that fact give pause 12 A Theory of Determinism or Mind and Brain, chapters 1, 2. 13 Donald Davidson, 'Mental Events' and other papers in his Essays on Actions and Events. 146

Consciousness as Existence to any philosopher of mind who wants to keep an eye on science? And on the most relevant part of science, which certainly is not physics? To stick to my subject, however, we have in the neural guarantee a third reason for supposing the world which is my perceptual consciousness cannot be identical with the physical world as understood, or the objective world, or the world indicated by science. What you will now suppose is that the world in which my perceptual consciousness is being said to consist must be a mental world, an interior world, a mind-world. It is what the dictionary calls the totality of my thoughts and feelings, or all of a class of them. Certainly you need a particular subject for one of these. Maybe such a world is a subject. It is such a world, too, that has a person's neural events as a kind of necessary and sufficient condition. And you will say, very truly, that if this is what the speculation about perceptual consciousness comes to, we are back in the debacle of having the analysandum in the analysans. To say that my perceptual consciousness consists in a mental world would be no help at all. In addition to the three dependencies, another reason or two are likely to occur to you for your disappointment, or maybe Schadenfreude. 4. Is it not implicit in what has been said of my world of perceptual consciousness, most notably about dependence on a subject, that this is a private world? Is it not the case that what is being postulated, despite the rhetoric, is no more than a multitude of worlds each private to their owner? Well, part of the answer is yes, in a way. You do not have access to my perceptual world. That seems to me no deep proposition, incidentally, nor one that necessarily will be true of those who come after us. There seems no conceptual impossibility or incoherence in the speculation that in some future decade one member of our species will have replicated in her head the neural events of someone else, and so by guarantee have access to what otherwise would have been only the other person's experiences. Still, the point remains that a world of consciousness is in this way private. This, at least in an ordinary understanding of them, is not true of any of the three mind-independent worlds. The objective world is explicitly said to be perceivable by more than one person. 5. Finally, although it may seem that no more needs to be said in support of the supposition that a world of perceptual consciousness is a mental world, there is the idea that a world of perceptual consciousness, because of the three dependencies, does not exist unperceived. The point, is worth separating out from (1) the necessity of a subject to a world. But unperceived existence is an explicit feature of the objective world, and of one part of the physical world, and it 147

Ted Honderich is fundamental although implicit in what was said of the world indicated by science. One burden of five considerations, then, is that the worlds I am promoting, the worlds of perceptual consciousness, are not identical with any of three other worlds. Is there another burden - that the worlds being promoted are no more than mental worlds? And hence that we get no useful understanding, certainly no analysis, of perceptual consciousness? Just talk? IV My World of Perceptual Consciousness and the Physical World I wonder. There is a troublesome fact. The world in which my present perceptual consciousness seems to consist is surely spatial. That chair over there is bigger than that other one, and to the left of it, and I can measure the distance between them. It's not a representation of the first chair that is bigger than a representation of the second, and it's not representations that are relatively positioned in that way or the given distance apart. So with time. In this world of my perceptual consciousness, one thing happens before, simultaneous with or after another, and things come out of the future into the present and then go into the past. It's not thoughts of them that do this. Nor, does this world have in it only sense-data or ideas or whatever of other properties of things. It has in it the solidity and brownness of the chair. In short, despite all that has been said, it seems this world at least resembles something else we have noticed regularly on our way. It seems to resemble the physical world in one of its two parts: spatiotemporal things that are perceived as against spatio-temporal things in nomic connection with the perceived things. Having arrived at this proposition about resemblance, fundamental to this paper, it is my aim in what follows to clarify and defend it, and, above all, to draw a proposal from it. As you will gather, the proposition about resemblance is not the weak one that the world of perceptual consciousness has in it representations of what is in the given part of the physical world. The point, strongly put, is that both worlds have chairs in them. Let me pass by what I hope is the battered idea that the troublesome fact and the proposition of resemblance are just a matter of the phenomenology of perceptual consciousness, not the real fact of it, and also put aside for a while (1) the dependency of my world on a subject, and attend to something else. It is the second consideration going against the resemblance, the fact that my world is dependent on 148

Consciousness as Existence my neural events. However much it may seem to have chairs in it, not representations of chairs, must this neural dependency not destroy any talk of real resemblance between my world of perceptual consciousness and the perceived physical world? And must any lingering hope not be finished off by the third consideration, that this world of consciousness is no less than guaranteed by my neural events? Several things need to be recalled or taken on board at this stage, and in particular in connection with the second consideration. One is that it is no part of what has been suggested that only the worlds of perceptual consciousness exist. There is the unperceived part of the physical world, and the objective world, and the world indicated by science. It has certainly not been doubted that these conceptions are true of what there is, or anyway of some of what there is. Something of their sort is undeniable. Also, these conceptions evidently overlap to certain extents, and will overlap with other mind-independent conceptions of what there is. Let us now focus on one fundamental overlap. It is asserted or implied in at least two of these conceptions that part of what exists is not perceived, not in perceptual consciousness. It will be convenient to have a name for this. Let us have one last world, the world-in-itself or noumenal world, but leave out any implications from the past, notably the doctrines of Kant and Plato. Think of the world-in-itself, if you like, in scientific terms, perhaps as a world of particles in fields of force, or, of course as spatio-temporal events in nomic connection with spatio-temporal events that are perceived. The principal role of the unperceived part of the physical world as we have understood it is to do some explaining with respect to the perceived part. That is also the principal role of the world indicated by science. We carry over this idea, of course, to our world-initself. What we then get is that my world of perceptual consciousness, while having a dependency on my neural events, also has a dependency on the world-in-itself, How this works is clear enough. My neural events do not come out of nowhere, If they are in a way the necessary conditions of my conscious events, they are also effects of something else. Each neural event is the upshot of a causal sequence, every stage of which is a causal circumstance or kind of causally sufficient condition for what follows. Of what initial causal circumstance is my neural event at some time an effect? Well, some will simply say the world-initself. I have in mind particularly those who take the world-in-itself to be a world of science, and in particular of physics, But, to be more cautious, it must surely be that there is a causal circumstance for my neural event in which the world-in-itself plays at least a large part. 149

Ted Honderich These propositions are of importance to us. We are considering the argument that since my world of perceptual consciousness is dependent on my neural events, has a kind of necessary condition in them, it must be merely a mental world, and not something that importantly resembles the perceived part of the physical world. Is that a good argument if my world is also dependent in the way outlined on the world-in-itself ? It seems not to be. I say so because the perceived part of the physical world, as we ordinarily understand it, has the same dependency. We do not sub- 1 tract the chair from the physical world, and, so to speak, put it in the mind, on account of our undoubted personal contribution to it. This contribution has to do with our perceptual apparatus and our con- ] ceptualizing and so on, and in particular the physical chair's neural i dependency. The relationship between the physical chair and the ] chair-in-itself gets in the way of putting the physical chair in the ] mind. This consideration against identifying my perceptual world | with a mental world does seem persuasive. Why should the neural ] dependency of my perceptual world degrade it into being 'mental' if the same fact does not degrade part of the physical world? In both cases the second dependency, on the world-in-itself, makes for an independence that is lacked by what we have been calling a mental world. What of the third consideration, that my perceptual world has not only a kind of necessary condition in my neural events but also what was called a guarantee? My neural events are a kind of sufficient condition for my world. Is that not a disaster? If my neural events stand in this relation to that chair over there, how can it be other than in my mind? I certainly grant that our conception of the perceived part of the physical world does not include the proposition of psychoneural correlates, of a neural guarantee for what is in this part of the physical world. But, as it seems to me, this is not essential to my line of argument. It is part of our conception of the given part of the physical world, as just noticed, that the world-in-itself is in some way necessary to it. There is, as we also know, this same dependency with respect to my perceptual world. The world-in-itself is a necessary condition for my neural events, the correlates of my conscious events. Evidently this provides a response to the argument that if my world of perceptual consciousness is guaranteed by my neural events, it must be merely a mental world and not something that substantially resembles the perceived part of the physical world. The world-in-itself is necessary to the guarantee. What of the fourth consideration, about privacy? Does what has been admitted as to the privacy of my perceptual world stand in the 150

Consciousness as Existence way of claiming that it substantially resembles the part of the physical world? Well, what has been admitted is that in a sense you do not have access to my perceptual world. Such a thing could happen in the future, but it is not a possibility now. That does make a difference between the two worlds. What size is the difference? One thing that wouldn't help my claim of substantial resemblance would be something about my perceptual world now and yours: their having numerically different things in them. Do they? As you may anticipate from my earlier scepticism about the representative theory of perception or phenomenalism, the answer seems to be no. To revert to our ordinary talk about perception, it does not follow from the fact that you and I have different accesses to a chair that we are aware of two things. More particularly, it does not follow from our perceiving a chair differently that we are not perceiving just one thing. What is a chair? What is one of these things? It is something that looks different from different points of view or angles. If something didn't look different from different points of view, it wouldn't be a chair. It would be something like a number or a concept or a proposition, or maybe the Eternal Idea of Chair, but not a chair. So, as it seems to me, we two are aware of the very same thing.14 There is something else in this neighbourhood that would do more damage to my claim of substantial resemblance between my perceptual world and the perceived part of the physical world. That would be these two worlds' having numerically different things in them. Do they? There seems no good reason for saying so. Why should the chair in my perceptual world not be the very same thing as the chair in the physical world? What is relevant about my perceptual chair, so to speak, is unique access to it. What is relevant about the physical chair is that it is perceived. But cannot these two descriptions be true of just one thing? The fifth consideration was that because of the three dependencies my perceptual world is something that does not exist unperceived, but that unperceived existence is an explicit feature of the objective world, and fundamental if implicit in what is said of the world indicated by science. And, most relevantly, as I remarked,15 it is an explicit feature of the physical world - that part that is spatiotemporal but only in lawlike connection with what is spatio-temporal and perceived. My remark, since we were then reflecting on differences rather than resemblances, left something out, maybe a 14 For, further arguments against a revised phenomenalism, see 'Seeing Qualia and Positing the World', in A. J. Ayer: Memorial Essays, ed. A. Phillips Griffiths (Cambridge: Cambridge University Press, 1991). 15 See above, pp. 147-8. 151

Ted Honderich little heuristically or even deceptively - the part of the physical world we are now interested in above everything else, the perceived part. It is there that we need to look for likeness or the lack of it to my perceptual world. Does this stuff, the perceived part of the physical world, exist unperceived? What does it mean to ask if anything exists unperceived? It is the question, presumably, of whether any properties can be assigned to a thing when it is unperceived. What property is assigned to the perceived part of the physical world if it is said that it also exists unperceived? That is not very clear to me, but perhaps the answer is that it is capable of being perceived. Can any properties be assigned to the items in my perceptual world when they are not in it? That is, taking my perceptual world to be a large temporally discontinuous particular, can any properties be assigned to an item in it in the times when this world is not in existence? Well, there seems to be a good deal to say. This item too has a capability - of being in my world when my world reappears. Also, it may now be in your perceptual world. Both of these facts are tied up with another, the item's relation to the thing-in-itself. Further, since we have lately identified the chair in my perceptual world with the one in the perceived part of the physical world, my chair when unperceived by me still has whatever properties give it membership of the perceived part of the physical world. Finally, if this is a different fact, my chair when unperceived by me remains spatio-temporal. One final remark here. My perceptual world was casually described a moment ago as a discontinuous particular. Like a club, it pops in and out of existence over time. That idea, it might seem, itself stands in the way of asserting any important resemblance between my world and the perceived part of the physical world, Still, something can be said. The latter thing is, as we can say, 'a world as perceived' or 'a world as experienced'. In that case, it too is a discontinuous particular. It is not dependent on any particular subject, but it is not there when we are all asleep, and parts of it are not there when they are in nobody's experience. This fifth consideration about unperceived existence gets us into deep or anyway troubled waters. Let us emerge from them with only the proposition that the consideration does not easily defeat my claim about resemblance, and turn back to what was passed by, the first consideration, about my perceptual world and a subject, self or person. Here too, as elsewhere, there certainly is a difference between my perceptual world and the perceived part of the physical world. It is perhaps the main difference. The perceived part of the physical 152

Consciousness as Existence world has no dependency on a particular subject. But the extent of the difference between the two worlds will depend on how we try to understand the subject, the fact of real subjectivity with respect to my world. I have no full and satisfactory understanding of the fact, needless to say. But, since the matter of a subject is bound up with the matter of perceptual consciousness, there is the consolation of being able to say something, and of the idea that it is possible to come to a tentative conclusion about perceptual consciousness without being able to say more. One thing that can be said is that the view of perceptual consciousness being contemplated allows for a literal understanding of some common philosophical talk about a subject: that it is or involves a point of view, a view from somewhere rather than nowhere, a perspective. There is one of those, literally, in the world of my perceptual consciousness. It is the point of view from where my head is, This is a little blessing - an escape from metaphor, the besetting problem of the Philosophy of Mind when it does not abandon its mission. Furthermore, it is possible on the view we are contemplating to start to explain what was remarked on earlier, that a subject not only is a necessary condition of perceptual consciousness in the sense of somehow being a part of it, but is such that the state of affairs would not exist at all in its absence. The explanation is that a point of view, literally speaking, is constitutive of the state of affairs. There could be no understanding of it which left out a real point of view. It is all a matter of the way things are from here, where my head is. It would be rash to suppose that all that there is to the fact of subjectivity is a real point of view. I have left out what is true, that my world is a matter of my particular conceptualizations. This fact enters into subjectivity, as does the fact of privacy, and no doubt a person's feelings and desires. I shall not take these reflections further, but merely remark that the view we are contemplating of perceptual consciousness gives some promise of a satisfactorily naturalistic conception of a subject. In so far, as it does that, we get some consonancy between my perceptual world and the perceptual part of the physical world. I also pass by the role not of a particular subject but of subjects in the perceived part of the physical world. Some subject is necessary to it. And the role of conceptualization. These help too. V Consciousness as Existence Let me sum up the comparison now completed between my world of perceptual consciousness and the perceived part, of the physical 153

Ted Honderich world. My world has dependencies on (1) a subject, myself, and (2, 3) on my neural events. It is (4) in a way private and (5) is said not to exist unperceived. This is enough, certainly, to make a difference from the mind-independent worlds and in particular the perceived part of the physical world. However, and to be brief, my world has chairs in it. Also, there is more to be said about the perceived part of the physical world. This (2', 3) shares a good deal of the neural and thing-in-itself dependencies with my world, and (4') it has in it, among other things, the very same things that are in my world. (5') In the matter of unperceived existence, it is not all that far from my world, and (1) it can be said to be consonant with my world's dependency on a subject. These propositions, in my submission, amount to an important resemblance between the two worlds. That is to say that my world cannot be regarded as just what was called a mental world - a totality of thoughts and feelings of mine. More particularly, my world is not being conceived in a useless, pre-analytic way. On the contrary, what we have, by way of the resemblance with part of the physical world, is an articulated and relatively rich conception. My perceptual consciousness, my world of perceptual consciousness, is an articulated state of affairs. I own up to doubts about the details of all this, and a residual worry that some inconsistency has gone unnoticed. But not enough doubts and worry to stand in the way of my main proposal in this paper. It is in part that in thinking about the mind and what exists, we have been stuck with two categories. These are, in the most general terms, the mind-independent worlds and mental worlds. It is not only philosophers of the mentally realist kind16 who have been stuck with not only mind-independent but also mental worlds. Philosophers sceptical about mental worlds, indeed with some reason disdainful of them, have nevertheless not escaped them, but write more and more books trying to accommodate them. To come to the very nub, what we need, in order to deal first with perceptual consciousness and thereafter with all of consciousness, is a new category: worlds of perceptual consciousness. They take a good deal from both mind-independent and mental worlds. We do not need new kinds of properties or events. We need this different way of looking at what we have got. Or, to remember my doubts and worry, and to be properly hesitant, we need some new way like this, something along these lines, We need some view of perceptual consciousness as existence, or, if you like, existence as perceptual consciousness. We need an idea to the effect that for something to be conscious is for a world to exist, although certainly not a world 16 See above, p. 140, and A Theory of Determinism, pp. 77-83. 154

Consciousness as Existence wholly dependent on it, This, in my submission, is what we have missed out in being anchored in the two categories of mind-independent and mental worlds. Is there not much to be said for this different category? Four more things come to mind. The category is not factitious. Our worlds of perceptual consciousness, in fact, are the only worlds that are not worlds of theory. They are not got by inference or speculation, however well-founded or even coercive the inference or theory. They are epistemically and perhaps conceptually prior to all other worlds, notably the objective and scientific ones. It is not clear, since the idea of ontological priority is more difficult than sometimes supposed, that they are not ontologically prior to the rest.17 Does the category of worlds of perceptual consciousness offend against a commitment to physicalism, taking the latter to be a commitment to the physical world and the world indicated by science, and perhaps the objective world, and at least a scepticism about mental worlds? The answer is that what has been proposed is a kind of physicalism. One reason is that our worlds of consciousness are approximate to the perceived part of the physical world.18 Nothing has been said until now of what seems to be a fact about consciousness and in particular perceptual consciousness. It is that it itself has a role in the explanation of behaviour. Conscious events are are ineliminable parts of full explanations of our actions. Accounts of the mind must fail or be incomplete, it seems, if they entail or allow for epiphenomenalism. My world of perceptual consciousness has no such shortcoming. Far from it. It has in it the very things that can most naturally be said to motivate us, chairs for a start. Finally, one more word about the crux of all of the philosophy of mind that deserves the name. That is the fact of our real subjectivity. Something was said earlier19 of how the proposed view of perceptual consciousness contributes to understanding here. It seems to me that consciousness as existence gives us more than other views of what we want, and more than has been mentioned. For one thing, subjectivity has to do with reality and immediacy. My world of perceptual consciousness is very real and very immediate. 17 'Dependence,' in Cambridge Dictionary of Philosophy, ed. Robert Audi (Cambridge: Cambridge University Press, 1995). 18 Cf., alas, 'Seeing Things', p. 52. 19 See above, pp. 152-3. 155

Setting Things before the Mind M. G. F. MARTIN Listening to someone from some distance in a crowded room you may experience the following phenomenon: when looking at them speak, you may both hear and see where the source of the sounds is; but when your eyes are turned elsewhere, you may no longer be able to detect exactly where the voice must be coming from. With your eyes again fixed on the speaker, and the movement of her lips a clear sense of the source of the sound will return. This 'ventriloquist' effect reflects the ways in which visual cognition can dominate auditory perception. And this phenomenological observation is one what you can verify or disconfirm in your own case just by the slightest reflection on what it is like for you to listen to someone with or without visual contact with them. A common assumption in most philosophical discussions of appearances and experience is that, when one does engage in just such reflection, the character of how things appear to one is just obvious to me. Just this assumption seems to lie behind Ned Block's comment what is it that philosophers have called qualitative states? As Louis Armstrong said when asked what jazz is, 'If you got to ask, you ain't never going to get to know.' It is implicit in much of the recent debate about the problems of explaining consciousness, in particular what has come to be called phenomenal consciousness, in purely naturalistic terms: although we may not be able to explain how such consciousness can arise without a physical world, we have a clear sense of what the problematic subject matter is just by focusing on one's own case. Now while the assumption is widespread, and in many ways seems sensible, it does raise a deep puzzle concerning the ways in which philosophers debate the nature of perception and perceptual appearances. For it is clear in such debate that philosophers disTalks from which this paper was drawn were given at a conference in Miskolc Tapolca, University College Dublin and the Institut fur Philosophic Universitat Miinchen; I am grateful to those audiences and the one at the Royal Institute for their questions and comments. I would also like to thank Paul Boghossian, Tim Crane, Naomi Eilan, Norbert Niclauss and Scott Sturgeon for detailed comments and discussions of these matters. Work for this paper was carried out while on research leave sponsored by the British Academy. 157

M. G. F. Martin agree, and that they disagree about the nature of appearances. Some philosophers claim that it is just obvious that there are aspects of your experience, say of your currently looking at this page, which are entirely independent of any aspect that you may perceive the mind-independent world to have. Others, however, are insistent that it is just obvious to us that our perceptual experiences of the world are purely representational or intentional, and that what it is like to be in such states is a matter of no more than how things are represented as being by those states. It is difficult to interpret these disagreements as other than being disagreements about the nature of appearances, how things look, or feel, or taste to us when we explore the world around us. Yet, if the nature of appearances really is just open to simple reflection, how can there be room for any serious disagreement? Surely one can confirm or disconfirm any theory of appearances straight off. The persistence of disagreement would suggest that either the inner lives of philosophers are much more varied than we previously had reason to suspect, or that at least one party to the debate must be deeply confused. Instead, I suggest that the fact that such disagreement does occur indicates that even if the character of experience is obvious to us, it is not obvious how obvious it is. To make sense of these different theories, we must interpret them as able to draw a contrast between the real nature of appearances and how their opponents may be misled in describing how such appearances seem. And this thought raises the question whether we can find an appropriate common ground among parties to the dispute: some description of what experience is like which neutrally expresses how appearance seem to us. We could then see the competing parties as attempting to give competing explanations of this common ground. It is this interpretative task that I attempt to undertake in this paper. I shall not be offering any final or definitive account of the nature of perceptual experience or the relation between experience and perception in this talk. On the other hand, the reader is sure to be able to find many such accounts in other discussions of perception. It is more difficult, however, to discern the common root or the starting point for these incompatible accounts of the supposedly obvious, and that is why I trust there is sufficient interest in trying to find a suitable overview of the disagreements here. In recent discussion, the notion of qualia has dominated debate about the nature of sensory consciousness. This has occurred par158

Setting Things before the Mind ticularly in the context of debate about the viability of a purely physicalist understanding of the mind. A common view is that the intentional or representational properties of mind, those in virtue of which our thoughts are about objects or properties in the world around us, present no insuperable problem to a physicalist account of the mind. In contrast, it has been suggested that the fact that we are conscious, and more specifically that we have sensory phenomenal consciousness, has been thought inexplicable given the state of neurosciences and cognitive psychology. Associated with posing the problem in this way is the thought that if we do have phenomenal consciousness, then such consciousness is not to be understood in representational terms. I want first to focus critically on the notion of qualia since it stands in the way of our getting a proper over-view of the disagreements concerning the nature of perceptual experience. Although many philosophers write as if it is simply obvious to us that there are qualia, and that we know what they are, I shall argue instead that this is all chimerical. For the most common usage of the term 'qualia' is equivocal, and the most familiar means of elucidating the term, by a kind of inner ostension of one's conscious states, simply fails to pick out a unique target. Furthermore, lying behind this confusion is a long-standing dispute about the nature of experience and our knowledge of it which needs to be made explicit before we can advance in our task of setting up a common framework for understanding the debate about perception. The term has been used in a number of different ways, but we would do well to start with the usage found in this passage from David Chalmers: a mental state is conscious if there is something it is like to be in that mental state. To put it another way, we can say that a mental state is conscious if it has a qualitative feel — an associated quality of experience. These phenomenal feels are also known as phenomenal qualities, or qualia for short.' So used, the term is intended to pick out in the most general and neutral way the various aspects of conscious episodes. In taking conscious experience to be suitably evident to a reflective audience, philosophers often avoid any explicit or informative definition of the term 'qualia'. Indeed, it is sometimes suggested that no informative definition could be given. Instead, we are often offered a verbal equivalent of an inwardly directed gesture, which in the context of the discussion is intended to direct one's attention on the appro1 D. Chalmers, The Conscious Mind (Oxford: Oxford University Press, 1996), p. 4. 159

M. G. F. Martin priate subject matter. While it is generally assumed that it is simply obvious to us then what qualia are to be taken to be, I shall argue that in fact the term is generally used equivocally, and that independent of some further clarification, we cannot determine how people are using the term. We can trace the equivocation to the way in which we are introduced to the term. A notable such example is provided by Daniel Dennett, in a discussion which more generally is hostile to the notion of qualia. Despite Dennett's hostility to the notion, his opponents have been happy to accept his initial elucidation of the notion right at the outset of his paper. It is worth looking at in some detail: 'Qualia' is an unfamiliar term for something that could not be more familiar to each of us: the ways things seem to us ... Look at a glass of milk at sunset; the way it looks to you - the particular, personal, subjective visual quality of the glass of milk is the quale of your visual experience at the moment. The way the milk tastes to you then is another, gustatory quale, and how it sounds to you as you swallow is an auditory quale. These various 'properties of conscious experience' are prime examples of qualia.1 Now this gloss on what Dennett complains is 'frustratingly elusive' contains a central, and, I shall argue, significant problem: Dennett equivocates on the term 'qualia' even as he introduces it. As the last sentence of the passage makes clear, and as the course of the paper it comes from also indicates, Dennett assumes, with many other authors, that we should use the term 'qualia' to pick out 'properties of experience'. We may think of seeing a glass, or more neutrally having a visual experience as of a glass, as being a state of mind, the having of an experience. Someone who has such an experience thereby has the property of having an experience of a glass. Qualia are then to be seen either as properties of properties - that is, what it is like to have an experience of a glass is a property of having the property of having an experience of the glass. Alternatively, we can think of the ways in which things seem to one as further determinations or specifications of the determinable, having an experience. Each of the specific experiences that you might have - the feeling of the hardness of the chair beneath you, hearing the rustle of frustration around you — are different ways of having an experience. Qualia are then just these different ways of having experience. However, Dennett does not stick with this usage, for the moment he gives us any concrete examples of qualia we seem to shift to something of an 2 D. Dennett, 'Quining Qualia', in Consciousness and Contemporary Science, ed. A. Marcel and E. Bisiach (Oxford: Clarendon Press, 1988), p. 42. 160

Setting Things before the Mind entirely different order to that of a property of an experience: for Dennett's examples are themselves not properties of experiences, but properties of the objects we come to perceive. He writes first of the way the glass of milk looks to one, where the object which has the property is itself part of the world around us and not part of the mind, namely a glass. Likewise, it is the particular quantity of milk which tastes some way to one, and the milk, one's throat (and lack of manners), all together which are responsible for the sound which Dennett picks out as an auditory quale. But surely nothing can be both the property of an object independent of the mind and at the same time a way of having an experience. So Dennett seems to have introduced just the wrong examples to indicate as 'properties of experience'. It is not difficult to see where the problematic ambiguity is introduced in the discussion. For the phrase 'the way things seem to us' is itself ambiguous. Dennett, and others, seek to introduce the term 'qualia' by reference to such English locutions for how things look, feel, sound or more generally appear. But appearance talk is itself complex and hence allows for abstraction of terms in more than one way. The different instructions for fixing on an example of a quale result from abstracting now in one way, and now in another. For example, when I tell you: It looks to Dan as if there is a rosy-hued glass of milk before him I may intend to emphasize how things are with Dan, and to contrast the fact that Dan has a certain kind of experience with the fact that Mary is asleep, or that Ben has an altogether different kind of experience. So we can imagine that the following underlined aspect of the sentence would be up for substitution in contrasting the way Dan is, with how else he might have been: 1. It looks to Dan as if there is a rosy-hued glass of milk before him On the other hand, given that this is in fact a case in which Dan is perceiving the glass of milk, we might rather be interested in what aspects of the milk are evident to Dan. In this case we may be interested that it is the specific shade that the milk has that is manifest to him, in contrast to the maker's mark on the glass. In that case, the following underlined aspect of the sentence would be open to substitution to contrast ways in which the situation might have differed: 1". It looks to Dan as if there is a rosy-hued glass of milk before him So in moving from talk of something appearing F to someone, to talk of appearances, qualities of experience or qualia, the loss in 161

M. G. F. Martin complexity of the semantic structure leaves one open to equivocation between properties of what appears and properties of what is appeared to. Just such slippage occurs in the passage quoted from Dennett: within one paragraph we move from properties of experience to properties of the object of experience, the glass of milk, back to properties of experience again. Dennett is not an isolated example of this shift, but perhaps we can make do with just one other more recent example. Fred Dretske, like Dennett, is hostile to a tradition of thought which sees qualia as presenting an insuperable problem for a naturalistic account of the mind. In his monograph Naturalizing the Mind, Dretske puts forward 'the Representational Thesis' as his account of how the mind can be part of the natural order. The thesis itself consists of two claims: '(1) All mental facts are representational facts, and (2) All representational facts are facts about informational functions'* When Dretske turns to the issues raised by conscious experience in the third lecture, he makes the following claim: The Representational Thesis identifies the qualities of experience - qualia - with the properties objects as representeds as having.4 Whatever one things of the Representational Thesis itself, one ought to hesitate before accepting this identity claim as a consequence of it. As the first half of the identity claim makes clear, qualia are assumed to be properties of experiences, properties of properties of one's mind, or ways in which one may come to have an experience. But Dretske, like most philosophers who ascribe a representational content to experience, supposes that our experiences represent how objects independent of the mind are. Such mind-independent objects cannot have properties which are properties of states of mind. So it is implausible to suppose that our experiences should represent mind-independent objects as having properties of states of mind. Yet this manifest absurdity is what Dretske claims in this passage. Well, if we try to reconstruct what Dretske might be trying to say here, we can see the same equivocation in play as in the Dennett paper. Although Dretske starts the sentence by talking about qualities of experience, the only intelligible claim he could be making is one which identifies qualia understood as the properties objects 3 F. Dretske, Naturalizing the Mind (Cambridge, MA: MIT Press, 1995), p. xiii. 4 Ibid., p. 65: 'represents' is Dretske's term for sensory or phenomenal representation as opposed to conceptual representation - the details of the distinction he draws has no import for the point made in the text. 162

Setting Things before the Mind appear to have with the properties our experiences represent those objects as having. This thesis is perfectly intelligible, even if some people might find it mildly controversial. On the other hand, one might think that this identity claim alone falls short of telling us much about what experiences are like, and how Dretske's position differs from those who insist that there are qualia but who reject the Representational Thesis. But the need for Dretske to link up claims about how objects may come to appear to have properties with claims about what our experience can be like is obscured for him by use of the equivocal term 'qualia'. Since he can now use it in one sense, now in another, it may seem as if he covers all angles at once. This example not only increases our sample of equivocal uses, but directs us towards the significance of this slip of the pen. For it would be mistaken to respond to this problem by claiming that we can easily re-interpret both authors so as to avoid any such equivocation and ambiguity. A charitable response to these problems would no doubt be one which understood both authors as intending strictly just to talk about the properties of what experiences are like when they talk of qualia, and hence to re-interpret any passages where they slip into talking instead of the properties that objects appear to have. But one could undertake this interpretative task only if we could reconstruct the theses put forward solely in terms of properties of experience on the one hand, and properties that objects appear to have on the other. Once we make the distinction we can see that the theories do not offer us any explicit account of how the two sets of properties are related, even though the equivocation between the two suggests that in interpreting the notion of qualia we are to understand that there should be some important relation between them. Indeed, the need to do so can be made even more explicit by setting this issue in an historical context with which it is not normally associated, that between sense-datum theories of perception and socalled adverbial approaches. Consider first this notorious passage from H. H. Price: 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 a material thing there at all ... 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 presented to my consciousness ... that something is red and round then and there I cannot doubt ... that it now exists, and that / am conscious 163

M. G. F. Martin of it - by me at least who am conscious of it this cannot possibly be doubted... This peculiar and ultimate manner of being present to consciousness is called being given, and that which is thus present is called a datum,} Perhaps the most salient aspect of this picture of perception and experience are those aspects of it which used to be called 'the actobject' model of experience. According to Price the occurrence of an experience involves a subject, a relation of being given which relates that subject to various objects, and the data, which are presented or given to her. Furthermore, Price is insistent that such objects will be present even in cases of illusion or hallucination, so at least some of these data are non-physical. However, the passage is of most concern to us for the kind of view of knowledge of experience that it expresses. For Price seems to be of the view that one knows about the character of one's experience, that some red bulgy thing is present to one's mind, through attending to the object which is given in the experience, the red bulgy thing itself. Indeed, like Moore before him, Price thinks that consciousness is entirely diaphanous, and hence that all differences between conscious states of mind are differences in the objects which those states can have.6 So, when one comes to know what one's experience is like, and how it may differ from other conscious states one could have come to have, one does so through attending to the objects of awareness given to one through having such states. In the middle of the twentieth century, sense-datum theories, as 'act-object' accounts of experience, provoked an alternative kind of account normally known as 'adverbial' theories of perception. The epithet comes from a suggestion first made by C. J. Ducasse, in response to Moore that 'blue', 'bitter', 'sweet', etc., are names of objects of experience nor of species of objects of experience but of species of experience itself. What this means is perhaps made clearest by saying that to sense blue is then to sense bluely, just as to dance the waltz is to dance 'waltzily' (i.e., the manner called 'to waltz') to jump a leap is to jump 'leapily' (i.e., in the manner called to leap) etc.7 The primary motivation for such adverbialism is to avoid any commitment to the existence of non-physical objects of the sort that 5 H. H. Price, Perception (London: Methuen, 1932), p. 3. 6 Moore's opinion can be found in G. Moore, 'The Refutation of Idealism', in Philosophical Studies (London: Routledge & Kegan Paul, 1922); Price commits himself to the view in Perception, p. 5. 7 C. Ducasse, 'Moore's "Refutation of Idealism'", The Philosophy of G. E. Moore, ed. P. A. Schilpp (La Salle, IL: Open Court, 1942), pp. 232-3. 164

Setting Things before the Mind Price is happy to accept. The assumption of such discussions is that a commitment to ways in which one experiences, as opposed to objects which one senses, cannot be thought objectionable since we will be committed to the existence of such states of mind, as long as we are not eliminativist about the mind or sensory consciousness. But the adverbialism which Ducasse favours goes beyond the purely negative thesis that we should not commit ourselves to the existence of non-physical objects of sense, to a contrasting picture both of the role that experience plays in our perception of the world, and of how it can be that we come to be aware of our own experiences. The key idea is that we should principally think of our experiences as effects upon us by the environment; effects which have a distinctive qualitative character, and which are such that they bring about beliefs about the environment. Such states have sufficient dimensions of variation that there can be a reliable connection between environmental conditions which bring them about. In turn such states will act as the causes of beliefs about the presence of such environmental conditions which reliably correlate with the states of affairs they are about. We can think of our descriptions of experience as being of red, or of green triangles, or of musk, all as indicating the kind of cause which brings them about and correlatively the belief which they could reliably fix. On this view, awareness of the objects of perception and how they appear to be is one thing - the mind is directed out at the world - and attention to one's own experience another thing. The experience is a merely a causal intermediary between world and our knowledge of it: our awareness of experience requires directing attention not at the objects of sense, but rather within the mind.8 Now if we bracket for the moment a concern with the metaphys8 Ducasse's main concern, it must be said, is with Moore's contention that the object of consciousness in sensing is independent of the mind - and the dispute between Moore and Ducasse involves much talking past each other. For a further development of adverbialism which takes on the elements described in the text, see R. Chisholm, Theory of Knowledge (Englewood Cliffs, NJ: Prentice Hall, 1966), pp. 91-8 and M. Tye, 'The Adverbial Approach to Visual Experience', Philosophical Review 93 (April 1984), 195-225. This approach has its roots in Thomas Reid, see T. Reid, 'Essays on the Intellectual Powers of Man', in Inquiry and Essays, ed. R. Beanblossom and K. Lehrer (Indianapolis: Hackett Publishing Co., 1983). Note that not all philosophers whose views on sensation have been classified as adverbialist have made the assumption about our knowledge of experience mentioned in the text. The most notable exception is Wilfrid Sellars: see 'Empiricism and the Philosophy of Mind', in Science, Perception and Reality (London: Routledge & Kegan Paul, 1963), and Science and Metaphysics (London: Routledge & Kegan Paul, 1968). 165

M. G. F. Martin ical status of the objects of sense, whether they can be non-physical or not, the contrasting attitudes towards knowledge of experience on Price's view and on the adverbialism opposed to it offer us contrasting interpretations of the connection between the properties of what experiences are like, which we can come to be aware of, and the properties which objects appear to us to have. For Price, there is nothing more to learn about the nature of one's experience than to learn what objects, and what qualities of objects, are given to one. To learn about the properties of one's experience just is to learn what properties objects are presented as having. We might put this in terms of qualia by saying that on this view, qualia is the sense of the what-it-is-like properties of experience, qualiaj as one might say, are partly constituted by the properties which objects appear to have or are presented as having, qualia2- And one comes to know what the qualiaj of one's experience are, through knowing what the qualia2 of one's experience are. In contrast, for the adverbialist, properties of one's experience need to be sharply distinguished from properties that objects appear to have: the properties objects appear to have, on the whole, are those which our experiences are liable to cause us to believe that they have. The properties our experiences have, qualiaj, are the properties which are responsible for our coming to acquire these beliefs, but they are distinct and our awareness of them is distinct from our awareness of the properties that objects appear to have, qualia2. If we look back to Dretske and Dennett, then we can see this controversy mirrored in what they have to say. In Dretske's case it is clear that the conception of knowledge of experience is closest to the sense-datum approach, although he is surely keen to avoid the metaphysical extravagances of that view: why the identity of qualia2 for him with the properties objects are represented as having may be relevant to the Representational Thesis is simply that if one accepts with Price that qualia2 determine qualiaj, and that we have knowledge of qualiaj through knowledge of qualia2, he can claim that our knowledge of what experience is like is simply knowledge of how it represents things to be, and hence knowledge of its representational properties. With Dennett, on the other hand, it is clear what we could interpret the passage in either way. For if one sides with the adverbialist then, given the close correlations between properties objects can be perceived to have and the experiences to which those objects give rise, one might imagine that thought of the one would be liable to bring to mind the other. Dennett can be seen as employing a form of metonymy: in mentioning the properties the glass of milk may be perceived to have, he enables his audience to latch on instead to the 166

Setting Things before the Mind distinct set of properties which one's experiences would have, were one perceiving the milk. Furthermore, since there is no obvious vocabulary for the qualities of experience so conceived, one might think that this is the most natural and obvious way to introduce such ineffable aspects of the mind into conversation. It is clear that there is a substantive disagreement here over the nature of qualia, even when we restrict that term simply to mean the what-it-is-like properties of experience. The instructions provided for simply directing one's attention to these elusive properties are inadequate to the task of settling which account is the right one; yet the terms in which philosophers discuss these matters tend to equivocate between talk of properties of experience and talk of properties of objects; and the ways in which they talk slip between supporting now one account of the relation of these properties and now the other.91 suggest it is no accident that these two things come together. As long we simply equivocate over the use of the term 'qualia' we can hide from ourselves the need for answering the difficult question how the properties of experience relates to the properties which things appear to have. II The issue here is substantive, but how are we to settle it? Well, the idea that our conception of what our experience is like and our conception of what properties objects appear to us to have might be separate is attractive only as long as we look at the simplest of descriptions of experience: for example a visual experience of a red bulgy thing; an experience of a bitter or tangy thing. There seems 9 Paul Boghossian suggested to me that one could define a perfectly good notion of qualia without this threat of equivocation: qualia just are the non-representational properties of the mind which make a difference to what it is like to be one. We can determine whether there are any qualia, simply by asking whether two individuals could differ with respect to what it is like to be them without differing in their representational properties. However, the problem with this suggestion concerns how we are to apply the test: for in order to use the test within a thought experiment we need to determine when two individuals are to be considered as sharing all the same representational properties. This we cannot do without attending to the properties which things appear to them to have. This, somewhat indirect, test for the existence of qualia implicitly exploits the kind of direct test discussed in the text: we are either meant to recognize those aspects of consciousness which are purely representational or those which are not. So the problem of what one is to direct a subject's attention to, when their attention is directed to the qualitative aspects of sensory experience remains. 167

M. G. F. Martin to be nothing about these descriptions that should make us prefer one account over another. But when we look to more complex cases, we see that the description of what is apparent to us is not independent of our appreciation of what experience is like, and that for some aspects of experience it is difficult to conceive of how they could be independent of how things appear to us. Proper attention to experience, I suggest, shows that the adverbialist conception of our knowledge of experience is in the end unintelligible. To focus on a concrete example, consider the following passage from a discussion of the nature of shadows by the art theorist Michael Baxandall: I am writing this at a table with a wall each side of it, on a day of mixed sun and cloud. The wall on the right is modern, made of brick, and painted white with a matte but even emulsion paint. At the base of the wall the paint is blistering from damp. The wall on the left is much older, rough-cast rendering over undressed sandstone masonry, and there have been various attempts to patch gaps in the rendering with cement of various consistencies. It too is painted white, but with a rougher sand-textured stuff. This is flaking off in places due to an impermeable white flint element in the rough-cast; and in some but not all of these places desultory touching up has been done with a different, slick and clinging white paint, some of it applied by a roller and some boldly by a brush. The conspectus of the walls to left and right is almost as monochrome white, nevertheless... As the sun comes and goes the various kinds of radiation change level by a large factor, certainly to the point of discomfort - there are windows on three sides - and yet the walls remain white: brightness constancy, of course. But, partly because of these shifts between direct strong light and diffused weak light on the monochrome walls, partly because of a special interest, I am very aware of being in an indescribably intricate ambience of microshadow. It may usually be called texture, a word that somehow invokes the sense of touch, but it consists visually of almost pure shadow - very small self-shadows, derived shadows, and slant/tilt shadings ... It is almost purely from shadow that my visual access to the microstructure of the two plane surfaces of the walls derives. I do not think stereopsy is helping much. What I do not do, or would not be doing but for a special interest, is to attend to the individual microshadows as shadows or as objects of perception in their own right. If I attend to part of a wall I get a sense of its surface quality and that seems enough. Even with a special interest, it takes an effort of will, a decree of 168

Setting Things before the Mind the mind, to attend to the same area of wall, to categorize its shadow types, and read the bearing of their lighting. It is not an optical problem of acuity, in this strong light; rather, it seems to go against the grain of the perceptual process...10 Baxandall is concerned with the question whether 'we can [attend to individual shadows] and at the same time preserve the pattern of our more usual utilization of the same shadow in the course of normal variously directed perception'.11 His concern is with the ways in which we can attend to shadows, the difficulty in doing so, and the ways in which our perception of our environment may subtly change as we do so. In the description of his study, we are given familiar types of description of his surroundings, intermingled with observations about the existence and nature of certain types of shadow and visual phenomena, together with some technical commentary on the physical nature of the light array. These three elements mingled together may give one a greater or lesser sense of what it must have been like for Baxandall glancing over his study and staring out at the countryside beyond. The more one knows the kind of room discussed, the more one can link it with one's own knowledge of what it must have been like; the more one follows Baxandall in attempting to attend to elements of the visual array, and discern the structure of shadows, the more one has the sense of what he has done, and how one can do it well or badly. However, the passage is also a bravura display of how one might try to describe a visual scene combining such elements: Baxandall draws our attention at least as much to what he is reporting himself as doing and how he is reporting it, as to what he discerns; we have the sense of what it is like keenly to attend to the visual world, so as to discern various of its elements, and the difficulty and effort involved in drawing out the role of shadow in our visual perception of the world. One might react to this passage by wondering what its bearing is on the question we are interested in, namely the nature of experience and our first person access to it. One might think that while it tells us much, more than we wished to know, about what its author perceived that afternoon in the environment around him, it does not tell us about his experience. But such a response, I suggest, would be wrong: what Baxandall does here, and reports himself as doing, is to attend to what it is like for him to look out at the world around him, and attend now to the objects he recognizes, now to the shadows by which they come to be visually defined for him. 10 M. Baxandall, Shadows and Enlightenment (London: Yale University Press, 1995), pp. 125-6. 11 Ibid., p. 128. 169

M. G. F. Martin When we follow the passage and see some surface now as textured and now as covered in a skein of shadows, we learn not only something about the object we are attending to but also how we learn things visually about that object. The relation between the shadows and the texture seem to be ones which are forged within one's experience. It is this type of phenomenological fact which Baxandall focuses on. For this reason, if we are to find anything which deserves the epithet of description of what it is like for one to see, then Baxandall's account deserves such a title. It is, of course, a fragmentary such account, offering only a limited such description, partial in what it highlights and what it omits, and undoubtedly in much of its description highly theory-laden. None of that, I suggest, can take away from the clear sense a reader has that what Baxandall does in the passage, and can be taken as intending to do, is describe his visual experience of the world, and not merely the objects of perception. But if it is a description of his experience it also has to be a description of the objects he perceives, or takes himself to perceive. For what else could this feature be, if not an aspect of how the wall appears to one to be when one focuses on it now one way, now another? What does this tell us about how we know what our experiences are like, and what we thereby know? First, the passage articulates much of what the experience is like, while at the same time leaving much unsaid, and perhaps unsayable. So it would be a mistake to suppose that the character of experience is entirely ineffable. Second, Baxandall indicates that he learns things about what it is like for him to view his study by paying careful attention in the way that he does to various features, and we the readers can certainly learn things not only about his inner life, but about our own, through reading the passage, and by following similar procedures. Even if there is a sense in which the character of our own experience is somehow obvious to us, that should not be taken to preclude the possibility that we can make discoveries about what experience is like. Third, and related to the above, learning about one's experience can involve active exploration, primarily of the experienced world around one, but in doing so of one's experience as well. Finally, correlative with the last, attending to what one's experience is like cannot be separated from exploring and attending to features of the world as perceived. This suggests that the way in which we learn what our experiences are like is by attending first to the objects and features which are presented to us in perception. But there is an obvious problem with this suggestion: we can have perceptual experiences even when we are not perceiving anything in the physical world at all. One 170

Setting Things before the Mind might have induced a perfect visual hallucination of a red tomato, rather than simply having the pleasure of seeing one all by itself. Furthermore, one might know full well that that is the position one is in. In such a case, one would not be in a position to scan the elements of the physical scene before one, nor would one take oneself to be in that position. Even in cases of hallucination, there is a way that one's experience is for one, and one can come to know what one's experience is like, yet there are no objects of perception for one to attend to. Nevertheless, the basic model can still be applied even to this kind of case. For, in as much as an hallucination may be indistinguishable for one from a genuine perception, it will still seem to one as if there is an array of objects there for one to scan and explore. This will not necessarily be banished simply by the knowledge that one is suffering an hallucination, any more than the knowledge one is staring at a Miiller-Lyer illusion is liable to make one see the lines as entirely equal in length. So in such a situation, one can still be interested in aspects of one's experience, and proceed to explore it by attending to the putative objects of awareness. Note that the way we attend to our experiences when we reflect on them involves two distinct ways of attending. One can attend to something simply in thinking about it, as when I attend to the average rainfall in August in thinking that it is less than !/2 inch. When one reflects on one's own state of mind, one attends to it much as one attends to any object of thought. In addition, we can attend to objects that we perceive in ways not present when merely thinking about them. As you read along this line, you may note that there are words ahead of the one your eye rests on at the moment, and that there are lines above, and below this one. Your eyes and your attention shift in turn from one word to the next. Now, as a whim, you might be inclined simply to turn your head away from the page to see what is going on in the world behind you. In that case, you shift your attention to a feature of your environment of which you are not currently aware. But, if you do not turn your head, but simply keep reading along the line, it may seem to you as if your attention is guided from the words that you now focus on, to the next set of words, by shifting among the features of which you are already aware. To the extent that you shift your attention, as a matter of voluntary control, rather than having your attention shifted, as when some distraction occurs at the periphery of vision, you seem to have the choice of moving your attention among the range of things of which you are already aware. So in perception, focal attention seems to range over objects which are already objects of awareness, and a motive for directing your attention to something is to find out more. 171

M. G. F. Martin Now in the case of reflecting on one's own experience, one attends to one's state of mind through directing one's attention over the actual or putative objects of awareness. Whether one is perceiving or merely hallucinating, there is an apparent array of objects for one to direct one's attention across. How things are as presented to one is surely one aspect of one's current state of mind: indeed, in a case of hallucination, directing one's attention to what is present will tell one nothing about what is present in one's environment in a case of hallucination. So, for this reason at least, exploiting perceptual attention is a way of coming to know about and attending to one's own experience. When one does so, one can't conceive of what one directs one's attention at as merely a property of one's experience, the way one is affected. For in directing one's attention across a visual scene, one may chose to direct one's attention to the feature on the left, rather than the one to the right. What one selects among are the putative objects presented at various apparent locations. But we do not think of our own experiences or their properties as spatially arrayed in this way. So the only sense that we can make of what one intends to do in attending to one's experience is that one does so through attending to things not taken to be merely properties of the experience. As the Baxandall passage indicates, just such perceptual attention j is exploited in coming to know about one's visual experience. So one j cannot in so attending take what one attends to simply to be a way j of being modified, as the adverbialist conception of experience i claims. In as much as one exploits selective attention in learning ' about experience, such attention must range over the actual or puta- ; tive objects of perception, and so attention to experience is not S entirely distinct from attending to the objects of sense. To this extent at least, we should side with Price and the sense-datum theorists and not their adverbialist opponents. Of course, to attend to one's own state of mind is not the same thing as attending to some aspect of the world one is interested in, but given that one's state of mind has a certain subject-matter, one can attend to the state of mind only by attending to that subject-matter. In the case of sensory experience, that requires that one direct one's attention at what is presented to one. This point is revealed most clearly in the case of visual experience and other experiences where the subject-matter is presented as spatially arrayed. For we clearly do not take entities arrayed spatially to be merely the properties of mind. But it also holds more generally. We have here two contrasting conceptions of experience. On the adverbialist conception, we are to think of experience as simply being a state of the subject, a way of being modified. We are not to 172

Setting Things before the Mind think of this event as intrinsically involving the presentation of anything to the subject, for that would be to import an 'act-object' conception of experience. Instead, experience is to be a modification in the way that being 13 stone is a way of being modified. What marks the former out from the latter is just that this way of being is a way of being conscious. The alternative conception of experience places much more weight on the subject of experience, and the subject's viewpoint. On that conception, to have an experience is to have a viewpoint on something: experiences intrinsically possess some subject-matter which is presented to that viewpoint. To understand such experience and what it is like, one has to understand the viewpoint on that subject-matter, and hence also to attend to the subjectmatter as presented to the viewpoint.12 So, if we could really just think of our experiences as ways of being affected, where the awareness of a subject-matter was not intrinsic to being in such a state, then we would have no reason to reject an adverbialist conception of such states of mind. However, when we think about sensory states such as visual experience, and more generally experiences of audition, smell, taste, even most bodily sensation, we cannot separate our knowledge of what it is like to be in that state from knowledge of the subject-matter presented to one in being in such a state of mind. But that suggests for all such experience that our awareness of what the experience is like is inextricably bound up with knowledge of what is presented to one in having such experience. To know what such experience is like is in part to know how things are presented to one as being. Indeed, I would suggest, all of this can seem so obvious, once one thinks about it, that it should raise a problem of interpretation: how could anyone have plausibly put forward the adverbialist conception of experience as a serious option, given what we know of our experience? There are, I suggest, two aspects to the explanation of this: on the one hand, adverbialists were driven by a desire to reject the metaphysical commitments of sense-datum theories of perception; if taking seriously what we introspect of our experience would com12 One can see Nagel's famous discussion of consciousness and physicalism, T. Nagel, 'What is it like to be a bat?' in Mortal Questions, (Cambridge: Cambridge University Press, 1979), as principally employing the second conception of experience - it is the role of a subject's point of view within experience which explains why one must adopt a subject's point of view to understand what his experience is like; cf. pp. 166, 172, 173-4. In contrast, much of the discussion of the so-called 'Knowledge Argument' against physicalism tends to focus on the adverbialist conception of experience, where the focus on a subject's own perspective comes in only at the level of thinking about one's experience, and not in having the experience itself. 173

M. G. F. Martin mit one to the existence of non-physical objects, then they were prepared to reject the apparently obvious. More insidious than this, though, the equivocation inherent in talk of 'qualia', which collapses the distinction between properties of being appeared to and properties apparent to one, simply obscures the inadequacy of the account. Ill We are now in a position to return to our initial task of laying out a common framework for the debate about the nature of experience and perceptual appearances. To know what one's experience is like is to know what properties, aspects or features are presented to one in having the experience. There seems to be no way to pick out the what-it-is-like properties of the experiences without also picking out corresponding properties which objects may appear to have. It is no surprise, then, to find that the term 'qualia' is happy to migrate between the two. Our first step should then be to replace such ambiguous terminology with an explicitly defined terminology which allows of no such slippage. We need to keep track of two distinct things and pose the question how they are to be related. On the one hand, we are concerned with states of mind, experiences, and how they can be the same or different from each other, in particular how they can be the same or different for the subject of such states: how it is for a perceiver when they are in one of these states rather than another. When talking about this aspect of perceptual situations, we might talk of the phenomenal character, or phenomenal properties of the experience. We shall use these terms strictly to apply only to experiences and their properties and not to the objects of experience and the properties they appear to possess. When we need to talk of the latter, as the above discussion indicates we need to in understanding the phenomenal properties of experience, we shall instead talk about the presented elements or presented aspects of an experience. With these terms in hand we can then state the conclusions of the last section as the keystone for our framework to the debate: reflection on sensory experience should lead one to accept that there are at least some phenomenal properties of experience which have corresponding presented elements, and our understanding of the phenomenal properties is dependent on our understanding of their presented elements. On this view, difference in presented elements between two experiences will be sufficient for difference in their phenomenal properties. Note, incidentally, that Price commits him174

Setting Things before the Mind self to something much stronger in insisting on the diaphanous nature of experience: namely that sameness and difference of phenomenal properties just are sameness and difference is presented elements. It is doubtful that this claim is true: why cannot the ways in which things are presented in experience make a difference to what the experience is like, in addition to what is presented? We should at this point address the principal worry which motivates an adverbialist conception of experience: when we introduce talk of the presented elements of experience, and make differences between phenomenal properties of experience turn on them, are we not simply re-introducing sense-data into our account of experience? In seeing why not, we shall see how we are in a position to gain an overview of the whole debate. If we are to do justice to a subject's own point of view in having such experience, we need to fix on such presented elements; otherwise our account of experience will not be an account of what it is like for the subject of such experience to be so. From the subject's point of view, in both cases of perception and in cases of illusion and hallucination it certainly is as if there is something presented to her. So we can't do justice to that perspective without mentioning such a presented element in saying what the phenomenal character of her experience is. If we fail to mention such things then, as we saw, we end up with a view of experience on which it is not intrinsically a way of being aware of things. But in doing this we need not take ourselves necessarily to be committed to the actual existence of these elements. For one might take a relaxed view of what the mention of a presented element in expressing the subject's point of view in having experience should commit one to. After all, we might think, in order to fix on young James's state of mind we have to mention Santa Claus, saying that James has asked his aunt for a Buzz Lightyear doll, but Santa Claus for a playhouse. At the very same time, we might simply add that James is more likely to be satisfied by his aunt than Santa Claus, since at least the former but not the latter exists. So too we might think that in occupying the point of view in having an experience, we must act as if the elements presented or given are there. If we are to attend to what our experience is like, we need to attend to the various aspects of the presented array, and to do so is to treat them as if they really do exist. But in taking a certain distance from someone's experience, or even in a moment of disbelief from our own, we may not suppose that there really is anything answerable to what is presented to that point of view. The key here is to realize that the thesis endorsed concerning the relation between phenomenal properties and presented elements is 175

M. G. F. Martin principally a claim about how we are to understand what experience is like for a subject, from the subject's point of view. To fix on what we are attempting to explain, what it is for one to have experience, we need to take seriously the first person point of view both in and on experience. It is then a further move to explain the metaphysical commitments of such experience, and to ask what it takes for there both to be points of view and to be things, presented elements, on which such points of view are points of view. We can understand the fundamental debate about the nature of experience as a debate about these metaphysical commitments and the relation between phenomenal properties and presented elements. Consider first the kind of intentional approach to perception which Dretske clearly favours. One will think that it is clear that the kind of experiences we have are intrinsically states of awareness of mind-independent objects and properties. So, one will identify the presented elements of such experiences with things that can exist independently of whether one has such experience.13 At the same time, in insisting on the representational nature of experience, Dretske allows for the possibility that such experiences may be illusory or hallucinatory. On this approach, one's experience may have the relevant phenomenal property without its corresponding presented element actually being there. We have here a two-way independence of presented elements and phenomenal properties. On the other hand, we can interpret those philosophers who insist that there is a subjective aspect to perceptual experience as claiming that there are presented aspects of experience which could not exist independent of one's awareness of them, but which at the same time are guaranteed to be instantiated just in case one does have an experience with the appropriate phenomenal properties. Here we have the mutual dependence of presented elements and phenomenal properties. We can see these different views, then, as disputing two questions. On the one hand, they are concerned with what can be present to the mind: can the presented elements in experience exist independently of our awareness of them? On the other hand, they are concerned with the manner or mode in which objects are presented to one in having experience: can the presented elements of experience be so presented as not to require their actual existence for one's experience to be so? Indeed, we can think of these two questions as defining for us a complete set of options for the kinds of phenomenal property in question, depending on the mutual dependence of presented elements on phenomenal properties: 13 And in doing so, the theorist may claim to show how physical objects can be the direct objects of perception. 176

Intentional Dependent Setting Things before the Mind Is it possible to have: Phenomenal Aspect & Not Presented Property? Is it possible to have: Presented Aspect & Not Phenomenal Property? Note that this generates four possible kinds of phenomenal property. For, one might agree with a defender of the intentional theory of perception that the presented elements of experience include the very mind-independent objects in the world around us which we take ourselves to perceive, and in that case that such presented elements can exist without one having the relevant experience. On the other hand, one might suppose, consonant with a sense-datum theory, that such experience really can only occur if its object really does exist, and hence that one can have an instance of the relevant phenomenal property only if its presented element exists. This possibility is marked in the matrix by the top right hand box: naive phenomenal properties, as might call them.14 Likewise, one might think that while the subject-matter of experience could not exist independent of one's experience, and so its presented elements could not be instantiated without corresponding phenomenal properties, nevertheless the experience itself would not be sufficient to guarantee the existence of that subject-matter. This possibility is reflected in the bottom left box, labelled Dependent Phenomenal Properties. The debate about experience has tended to focus simply on the intentionality and subjectivity of experience, and hence on only two of these four properties, intentional phenomenal properties and subjective phenomenal properties in the terms of our matrix.15 14 Of course, one might think that the existence of illusions and hallucinations are enough to show that there cannot actually be any experiences with such phenomenal properties. Whether such arguments from illusion really establish that conclusion turns in part on how one assesses so-called disjunctive theories of perception as presented in P. F. Snowdon, 'Perception, Vision and Causation', Proceedings of the Aristotelian Society (1980-81), J. McDowell, 'Criteria, Defeasibility and Knowledge', Proceedings of the British Academy (1982), H. Putnam, 'The Dewey Lectures', Journal of Philosophy (1994). 15 Cf. C. Peacocke, Sense and Content (Oxford: Clarendon Press, 1983),