OXFORD UNIVERSITY PRESS
Great Clarendon Street, Oxford OX2 6DP Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Bangkok Buenos Aires Cape Town Chennai Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata Kuala Lumpur Madrid Melbourne 1vlexico City Mumbai Nairobi Sao Paulo Shanghai Taipei Tokyo Toronto Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York
© the several contributors,
2003
The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published
2003
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose this same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Data available ISBN I
0-19-9252823
3 5 7 9
10
8 6 4
2
Typeset by Newgen Imaging Systems (P) Ltd., Chennai, India Printed in Great Britain on acid-free paper by BiddIes Ltd., Guildford & King's Lynn
Introduction •••
The United Kingdom is very fortunate to have an active Kant Society. It is also fortunate to have in Peter Strawson not just one of the greatest living philosophers, but also the leading proponent of analytic Kantianism. Strawson's seminal Individuals rehabilitated metaphysics as a respectable enterprise within analytic philosophy. It also inaugurated a distinctly Kantian project-descriptive metaphysics-and placed the idea of transcendental arguments at the centre of epistemological, metaphysical, and methodological debate. It was followed by The Bounds of Sense, a brilliant and provocative discussion of the first Critique, which continues to influence I(ant scholarship by way of inspiration and opposition alike. It was only natural, therefore, for the UK Kant Society to devote one of its annual conferences to Strawson. The conference was hosted by the Departn1ent of Philosophy at the University of Reading, and took place on 17-19 September 1999. It was the first conference on Strawson in Britain for a long time, and the very first to concentrate on his relation to Kant. In this latter respect, the proceedings of the conference complement three other collections of essays on Strawson, in which Kantian themes are mentioned only in passing. 1 Furthern10re, the date was particularly appropriate in that Sir Peter turned 80 in 1999. I was fortunate to secure the collaboration not just of Sir Peter himself, but also of some of his eminent pupils, admirers, and critics. The papers divide loosely into three kinds. Some of them, namely those by Strawson, Hacker, Bird, Cassam, Stroud, and n1yself, deal with general questions concerning the nature of Strawson's Kantianism and of his rehabilitation of metaphysics. Others, by Westphal, Rosefeldt, de Gaynesford, Allison, and Forster, are devoted to more specific topics in Kant. In the remainder, by Grundn1ann and Misselhorn, Stern, and Hyman, the focus is more on Strawson than on Kant. Taken as a whole, the collection ranges from Kant interpretation and the history of analytic philosophy through philosophical logic, metaphysics, and epistemology to the philosophy of mind and
1 Z. van Straaten (ed.), Philosophical Subjects: Essays Presented to P. F. Strawson (Oxford: Clarendon Press, 1980); P. K. Sen and R. R. Verma (eds.), The Philosophy of P. F. Strawson (New Delhi: Indian Council of Philosophical Research, 1995); and L. E. Hahn (ed.), The Philosophy of P. F. Strawson, The Library of Living Philosophers, xxvi (Peru, Ill.: Open Court, 1998).
2
Introduction
aesthetics. In this, it reflects the range of Peter Strawson's own philosophical interests and achievements. The following abstracts, provided by the authors themselves, give a more detailed picture of their contents. Peter Strawson's opening essay falls into three parts. The first discusses his relationship to Kant, and in what respect Kant's influence on him is a special one. The second part features a (partly appreciative, partly critical) discussion of Rae Langton's recent interpretation of Kant in Kantian Humility. The third part returns to the topic of intellectual autobiography. It turns to some other influences on Strawson's work, especially that of Wittgenstein. Among other things, it mentions points of contrast, such as Wittgenstein's disregard for the constructive and systematic aspects of philosophy, and his more sceptical view of subjective experience and, in particular, of abstract objects. My own contribution first discusses the role of Strawson's Individuals and Bounds of Sense in the rise of what I call 'analytic Kantianism', the distinctly analytic interpretation, defence, and elaboration of Kant's ideas. In the sequel I defend Strawson's particular branch of analytic Kantianism against some widely accepted criticisms: that it is unfaithful to the general idea of transcendental philosophy; that it wrongly dismisses transcendental idealism and transcendental psychology; and that transcendental arguments could only ever establish that we must believe certain things to be the case, not that they are the case. I end by arguing that Strawson has provided us with a special kind of conceptual analysis, one that con1bines certain methods of the analytic tradition with important Kantian ideas. Peter Hacker's essay places Strawson's rehabilitation of metaphysics within the history of metaphysics. Periods of metaphysical system building tend to be followed by brief periods of anti-metaphysical reaction. In this vein, Strawson's Individuals marks a return to metaphysics following the attacks on it by the logical positivists. The paper starts with a sketch of Strawson's distinction between descriptive and revisionary metaphysics. In the second section it argues that descriptive metaphysics preserves only the letter but not the spirit of traditional metaphysics. Instead of purporting to delineate the ultimate structure of the world, descriptive metaphysics investigates the connections between the fundamental concepts we use to describe the world. The final section discusses whether revisionary metaphysics as Strawson describes it conforms to the intentions behind the metaphysical systems of the past, and whether it constitutes a coherent enterprise. Graham Bird discusses the relation between Kant's descriptive metaphysics and that of Strawson. In Individuals Strawson outlined what he called a 'descriptive metaphysics', and it is at least natural to suppose that the views of Kant that Strawson approved in The Bounds of Sense fall under the same heading. Bird takes it that both Kant and Strawson share such a project of descriptive metaphysics; but he argues that their projects
Introduction
3
are nevertheless not the san1e. He distinguishes them under three headings: relations to traditional scepticism and the appeal to transcendental arguments; the two projects' methods; linguistic analysis and transcendental psychology; the nature of necessary, a priori, features of experience. In his essay on a priori concepts Quassim Cassam distinguishes between the view that a priori concepts are justificationally a priori and the view that they are derivationally a priori. He discusses various ways of understanding the notion of justificational apriority, and questions the derivational apriority of at least some of the Kantian categories. Barry Stroud deals with the synthetic a priori in Strawson's Kantianism. Kant's question of how synthetic a priori judgements are possible was in part a question of how philosophical results with the distinctive status of those he reached in the Critique of Pure Reason could be reliably arrived at. Stroud asks whether there is a parallel question about the results of the more 'austere' Kantian project Strawson pursues, while repudiating transcendental idealism and even, apparently, any appeal to a priori knowledge. Stroud argues that conclusions with the special, distinctive status Strawson has in mind can be reached if necessary connections can be discovered between the possession of certain conceptual capacities and others, and that no reliance on the analytic/synthetic distinction or on the idea that we know son1e things a priori is required for the success of that project. The main topic of Kenneth Westphal's piece is Kant's Refutation of Idealism. Mainstream analytic Kant commentary has sought a purely conceptual, broadly 'analytic' argument in Kant's Refutation of Idealism, and then has despaired and criticized Kant when no such plausible argument can be reconstructed from his text. According to Westphal, these disappointments overlook two key features of Kant's response to scepticism: his non-Cartesian philosophy of mind and his non-Cartesian method of 'transcendental reflection'. His paper highlights the nature and role of transcendental reflection in four key thought-experiments through which Kant purports to show that we sense, and do not merely imagine, objects and events in the spatio-temporal world around us. The contribution by Tobias Rosefeldt is concerned with the problem of the self. Kant would accept Strawson's claim that we can have genuine knowledge about ourselves only if we refer to ourselves as persons, i.e. beings whose bodies provide empirically applicable criteria for their identity through time. But he also holds that beyond empirical selfknowledge we have a priori self-consciousness whose object is not the 'real subject of inherence' or the self as a real entity but the 'logical subject of thought' or 'the logical I', which has only 'logical identity'. In his paper Rosefeldt tries to elucidate these notoriously obscure remarks by giving a detailed account of what Kant means by characterizing something as 'logical'.
Introduction Max de Gaynesford is also concerned with Kant and Strawson on the first person. One influential explanation of the divergence between the two is as follows: Kant's 'criterionless self-ascription' thesis (that the immediate self-ascription of thoughts and experiences involves no application of empirical criteria of personal identity) was an unparalleled insight; but, because of residual Cartesianism, Kant failed to press it home. The paper expresses certain reservations about this diagnosis; in particular about whether, for all Strawson shows, Kant held the thesis, and whether it would have been correct, or even consistent, for him to have done so. With Henry Allison's piece we leave the first for the third Critique. Allison analyses the principle of the purposiveness of nature and the deduction that Kant provides for it in the introduction to the Critique of Judgement. He argues that, in spite of its merely subjective nature as a principle of reflective judgement, this principle is a genuine transcendental condition of empirical knowledge qua empirical, and that Kant's justification of it constitutes his definitive answer to Hun1e regarding the vindication of induction broadly construed. Eckart Forster is also concerned with the third Critique, but with the nature of aesthetic judgement. In his recent 'Intellectual Autobiography' that opens the Library of Living Philosophers volume in his honour, Strawson reviews, among other things, his various publications on Kant subsequent to The Bounds of Sense. In this context he writes: 'More recently I paid tribute to his [Kant's] insight into the nature of aesthetic judgement.' It is this tribute, or rather its two central claims with regard to Kant's aesthetics, that Forster discusses in his paper. Thomas Grundmann and Catrin Misselhorn consider the relation between transcendental arguments and realism. Transcendental arguments are supposed to show on a purely a priori basis that the necessary conditions of thought and experience are not only psychological conditions of our thinking and experiencing objects, but also conditions that are true of these objects. Realists protest that psychological facts cannot entail any conclusions whatsoever about non-psychological reality. For this reason Stroud and Strawson have recently argued that transcendental arguments can establish at most psychological truths about what we must believe. In their paper Grundmann and Misselhorn discuss the prospects of a more ambitious strategy for realists, namely the attempt of vindicating our basic procedures of justification in general by means of semantic externalism. Robert Stern's paper concerns Strawson's appeal to a certain kind of Humean naturalism, particularly in his response to scepticism. First, it argues that Strawson's naturalistic turn is in tension with his earlier positions. Second, it argues that the naturalism Strawson appeals to is not adequate as a response to scepticism, and that many of Strawson's earlier arguments can be better understood and defended on their own terms, 4
Introduction
5
without any such appeal, so that it was misguided of Strawson to take this naturalistic path. John Hyman, finally, discusses the modern causal theory of perception, of which Strawson is a leading advocate. The causal theory combines two claims: first, that it is a conceptual truth that our perceptions are caused by the material objects we perceive; and, secondly, that we are immediately aware of these objects themselves, rather than their mental proxies. Since this theory is not committed to the doctrine that the immediate objects of perception are mental entities, it is generally thought to escape the difficulty faced by the classical causal theory in explaining how the ordinary beliefs we acquire when we perceive material objects can be justified. Hyman argues, first, that it faces the same difficulty; and, secondly, that the theory depends on a false view about the nature of perceptual experience.
I
A Bit of Intellectual Autobiography P. F. STRAWSON
• ••
Most of what I have to say under the heading of intellectual autobiography has already appeared in the Library of Living Philosophers volume published in 1998.1 But perhaps I can add something bearing mainly, though not exclusively, on my attitude to the work of Kant. Instead of coming at this directly, I would like to begin by recalling Kantrelated episodes in the lives of two other English philosophers of this century. In a well-known passage in his autobiography2 R. G. Collingwood relates that at the age of 8 he read Kant's Groundwork of the Metaphysics of Morals, presumably in an English translation. He did not, he says, understand it; but he knew at once that this was for him; that the climate of this kind of thinking was to be his climate, the air of philosophical thought the air he must breathe; as he did (though not exclusively, since he was also an eminent historian). The other episode concerns a younger philosopher; namely, A. J. Ayer. His biographer 3 reports that while sailing to Africa in 1943 to undertake a special-operations exercise Ayer undertook to reread Kant's Critique of Pure Reason, and, in the early stages of sunstroke, underwent a remarkable epiphany during which he understood for the first time the full force of Kant's argument. Unfortunately, once he had recovered from his fever he was unable to regain the insight. Sympathetic though one may find both these Kant-inspired experiences, I cannot n1yself report any close parallel to either. Nevertheless, Kant, or more exactly Kant's first Critique, does have a distinctive place in my own intellectual history, such as it is, in a way I will try to make clear. For some years after my first .academic appointment just after the war the questions I was
1 L. E. Hahn (ed.), The Library of Living Philosophers, xxvi, The Philosophy of P. R Strawson (Peru, Ill.: Open Court, 1998). 2 R. G. Collingwood, An Autobiography (Oxford: Oxford University Press, 1938). 3 B. Rogers, A.]. Ayer (London: Chatto and Windus, 1999).
Strawson
8
mainly concerned with fell in the general area of philosophy of language and logic: questions about reference, truth, entailment, the constants of formal logic and their natural-language counterparts, analyticity, etc. Wrestling with these problems, one had, of course, to wrestle with the work of those philosophers whose views on the questions concerned were at the time, and sometimes still are, influential or even dominant-most notably Russell, Quine, and Austin. Indeed, it was sometimes precisely the views that one or another of these had expressed that fired my concern with the question. Nevertheless, closely as one might study the relevant passages in the writings of the philosopher concerned, it was precisely and only because of their relevance to the question at issue that those passages demanded and received such close attention. It was not because those passages were, or seemed to be, an integral part of some wider system of thought associated specifically with the name of that philosopher, perhaps because initiated by him: And this is where the difference with my relation to Kant or, to be more exact, to Kant's Critique of Pure Reason comes in. It was that complete work itself, rather than any of the many particular issues with which it deals, that became the focus of my concern. Indeed, it is the only work, and Kant the only author of such a work, of which, and of whom, I can say this. The reasons for it are, of course, largely internal to the work itself; but also, I must confess, partly historical-to do, in fact, with the structure of the PPE school in Oxford before the war. Anyone reading for that school at that time who wanted to specialize in philosophy was offered no choice of philosophical special subjects; there were just two on offer, and no more: Logic and Kant, the latter to be studied in just two works, the first Critique and the Groundwork. The Groundwork, though like Collingwood I found it deeply impressive, conceived its subject, as I thought then and still think, altogether too narrowly, whereas in the Critique of Pure Reason I found a depth, a range, a boldness, and a power unlike anything I had previously encountered. So I struggled with parts of it as an undergraduate, and later as a college tutor teaching those few pupils intrepid enough to take it on, until finally, having been subtly and in part consciously influenced by it in my own independent thinking about metaphysics and epistemology (in Individuals 4 ), I decided I must try to get to grips with the work as a whole. So I began to give a regular series of lectures on it, a series that ultimately issued in the publication of The Bounds of Sense. 5 In that book I tried to preserve and present systematically what I took to be the major insights of Kant's work, while detaching them from those parts of the total doctrine that, if they had any substantial import at all, I took to
4 5
P. F. Strawson, Individuals (London: Methuen, 1959). P. F. Strawson, The Bounds of Sense (London: Methuen, 1966).
A Bit of Intellectual Autobiography
9
be at best false, at worst mysterious to the point of being barely comprehensible. My book was, you n1ight say, a somewhat ahistorical attempt to recruit Kant to the ranks of the analytical metaphysicians, while discarding those metaphysical elements that refused any such absorption. My position on all this I have subsequently sought to elaborate or clarify a little, particularly in the first two of the four Kantian studies included at the end of the collection Entity and Identity.6 Of course I am not foolish enough to suppose that I have got all or any of these things quite right; and I am sure that there are plenty of philosophers willing to show me ,,,here I have gone wrong. But I can take sonle comfort in the thought that, when I have erred, I have done so in the company of most, if not all, of those who have been brave enough to undertake the interpretation and criticism of Kant's critical philosophy. I shall not here and now undertake anything by way of further elaboration, modification, or defence of the views advanced in my book or the subsequent articles. Instead I should like to consider briefly a recent and, I think, novel attempt to elucidate and defend a central Kantian thesis: the thesis, namely, that we are and must remain ignorant of the nature of things as they are in themselves. I refer to a book published in 1997 by Rae Langton, which is called Kantian Humility? and which is certainly a most interesting, impressive, and scholarly exercise in Kantian interpretation. Early on in the work she refers, effectively by way of comparison and contrast with her own, to another philosopher's solution of the problem posed by the Kantian doctrine of our necessary ignorance of things as they are in themselves. The view in question is Professor Allison's, and, as she rightly remarks, his solution is both elegant and ingenious. It also has what in her view are distinctive merits. It preserves the objective reality of the natural world as studied by the physical sciences; and it disposes completely of the picture of two distinct realms of being: the one the realm of supersensible things in thenlselves, the other the realm of phenomena, however conceived. But also-and this is where her approval ends-it completely draws the sting of the doctrine of necessary ignorance, rendering it harmless, anodyne, even trivial. For it does not have the consequence that there must be anything real at all of which we are necessarily ignorant, though of course there may be much of which we are and may remain contingently ignorant. And this is where Professor Langton jibs. For in her view it is an essential part of Kant's doctrine that there really is something substantial of which we are necessarily ignorant and of which our necessary ignorance is a source of necessarily vain, but humanly natural, regret. Things in then1selves affect our sensibility and thereby make knowledge possible; but they affect us in virtue
6 7
P. F. Strawson, Entity and Identity (Oxford: Oxford University Press, 1997). R. Langton, Kantian Humility (Oxford: Oxford University Press, 1997).
Strawson
10
of their extrinsic, relational, causal properties, which are essentially forces constituting the natural world, phenomenal substance, the subject matter of physical science. But these forces, phenomenal substances with which we are acquainted and of which we can have knowledge, though real enough are but extrinsic, relational properties of things in themselves; and as subjects of these relational properties-substances in the pure sense-things in themselves must also have intrinsic properties; and these intrinsic properties are necessarily unknown to us, since it is only the matter-constituting forces of which we can become sensibly aware. So, though we have knowledge of their relational properties that constitute nature, of things as they are in themselves or intrinsically we remain necessarily ignorant. Of course these few sentences of mine are only a sketch-possibly, though I hope not, a travesty-of what is a very subtly and carefully worked-out position. It is a position, moreover, that Professor Langton skilfully supports with an impressive array of references, not only to the Critique itself and Kant's other writings, but also, and often in a critical vein, to the work of his philosophical predecessors, most notably Leibniz; and to that of many commentators. At the end of her book Professor Langton acknowledges one prima-facie difficulty for her position. This is Kant's clear and repeated assertion of the ideality of space, its subjective source; for this may seem to bring into question her firm belief that the objective reality of the material world, the subject matter of the physical sciences, is an integral part of the critical doctrine. It may seem to threaten us (and Kant) with commitment to a kind of phenomenalistic, or even to the Berkeleian, idealism that Kant himself emphatically repudiates. Professor Langton is convinced that the threat is only apparent, and considers briefly a number of ways of circumventing it. The solution that she finds most satisfactory consists in drawing a distinction: the dynamical forces that constitute bodies are genuinely objective properties, but relational not intrinsic properties, of things as they are in themselves; space, though its source is subjective and hence spatial relations are ideal, is simply the form in which we have intuitive awareness of real dynamical relations; spatial relations are ideal, but they make experience of real dynamical relations possible. Professor Langton is aware that more work would need to be done on this solution. She says: 'the connection Kant sees between dynamical and spatial relations must be regarded as unfinished business.'8 But she seems to have no doubt that a solution on these lines must be correct. It seems to n1e, however, that there is another and quite different difficulty for Professor Langton's interpretation, a difficulty of which she takes no account at all. This difficulty relates not to the objects of outer sense, of
8
R. Langton, Kantian Humility (Oxford: Oxford University Press, 1997), 217.
A Bit of Intellectual Autobiography
I I
which space is the form, but to the contents of inner sense, of which time is the form: in other words, the contents of empirical self-consciousness, which Kant, somewhat like Hume, represents as a succession of constantly changing subjective states, a flux (his own word) of thoughts, perceptions, feelings. How are these to be accommodated in Professor Langton's scheme of interpretation? They are certainly not intrinsic properties of any thing (presumably, in this case, a self) as it is in itself. They are firmly. declared, like the objects of outer sense, to be appearances. But again they cannot have the reality of those real, but extrinsic, relational, causal, dynamic properties of things in themselves that constitute the objects of outer sense, the subject matter of the physical sciences. Yet they cannot just be left in the air, as it were; they must be found a place in the scheme of things, since without them no experience, and hence no knowledge of the objective world, the subject matter of the physical sciences, would be possible at all. They are indeed recognized by Kant as a fit subject for what he called empirical (as opposed to rational) psychology and picturesquely describes as a kind of physiology of inner sense. If Professor Langton is to find a place for them, then, it looks as if she must find besides those real but extrinsic dynamic properties of things in themselves that constitute bodies son1e analogous real but extrinsic properties of things in themselves that are capable of constituting minds or, perhaps better, empirical consciousness. No such account is forthcoming, however; and, even if it were, she would face a problem parallel to that apparently created for the objective reality of bodies by the ideality of space; for time also, the form of inner sense, is declared to be ideal. For these reasons, though not for these alone, I am unconvinced by Professor Langton's work, interesting, impressive, and scholarly as it is. Yet I recommend it for these, its own, certainly intrinsic, properties. After that critical interlude, perhaps I should say a little more to justify the title of this chapter. It might reasonably be thought that in order to do that I should at least say, first, whether any other philosopher has had an influence upon me at all comparable with that of Kant, and, second, whether any particular view I have come to hold seems to me of outstanding importance. For reasons I have already made clear, no single other philosopher and no single work of any other philosopher has had in my philosophical history the position that Kant and the first Critique have had. But I can mention other more diffuse influences. First, then: Russell and Moore, the founding fathers, at least as far as England is concerned, of analytical philosophy in our period. Their influence related to the questions and problems they discussed rather than the answers and solutions they gave. Second: the brightest lights that shone on the Oxford philosophical scene in the 1950s-those of Ryle, Austin, and Grice-though here too it was more ------------------------------
12
Strawson
a matter of style of thought than any particular doctrines to which I responded. And, finally, I must mention Wittgenstein; for, if I share anyone's conception of what our general philosophical aim or objective should be, it is, if I have understood him correctly, that of Wittgenstein, at least in his later period. That is, our essential, if not our only, business is to get a clear view of our most general working concepts or types of concept and of their place in our lives. We should, in short, be aiming at general human conceptual self-understanding. Wittgenstein saw that a necessary condition of achieving this was to liberate ourselves from false understanding; to tear away the veil of simple seductive illusions or pictures that pervaded or constituted much existing philosophical theory and that prevented us from seeing clearly, from getting the clear view we needed. To this task Wittgenstein devoted much of his formidable powers and did so with the unique effectiveness of genius. But I must add, as I think, that his almost obsessive anxiety to liberate us from false pictures, from the myths and fictions of philosophical theory, led to a certain loss of balance in his thinking. It did so in two ways. First, it led to a distrust of systematic theorizing in general-and hence to a disregard of the possibility, indeed, to my mind, the fact, that the most general concepts and categories of human thought do forn1 in their connections and interdependencies an articulated structure that it is possible to describe without falsification. Indeed, what I tried to show in my work on Kant is that the first Critique contains, besides much else that is more questionable, the general outline of many essential features of just such a description. Second, this same anxiety to Ii berate us from false theory led Wittgenstein, as I think, to minimize or dismiss, or at least give too little acknowledgement to, some pervasive features of our experience and of our ordinary non-philosophical thought. It is true of these features that they can, in philosophical thinking, lend themselves to gratuitous inflation, to mythologizing, to false imaginary pictures-all of these proper targets of Wittgenstein's hostility and scorn, the 'houses of cards' it was part of his mission to destroy. But that is no reason for failing to acknowledge them fully as the harmless, inescapable features that they are. So what are these features? I have in mind two things: the first is the reality of subjective experience in all its richness and complexity or, as one of our most distinguished contemporaries expressed it, in all its 'heady luxuriance'the phrase is Quine's; the other is the inescapable presence in our thought of abstract intensional objects. Both, as I remarked just now, are easily misunderstood, prime sources of the generation of 'pictures to hold us captive'. But neither should for that reason be downplayed or denied the character it actually has in our experience or our thought. Another thing I suggested I should do in order to justify my chapter title is to answer the question whether there is any particular view that I have
A Bit of Intellectual Autobiography
13
come to hold that I regard as of outstanding importance. Well, there is such a view: it is by no means new and I do not think I am alone in holding it. It is not exciting: it is even, I think, a truism. But it has been overshadowed and regarded with suspicion in recent times. It is not a view that I myself have con1e to merely recently. Indeed, I had already grasped it in an incomplete and inchoate form before 1950. But a sense of its importance and ramifications has steadily grown with me since. It is this: that the fundamental bearers of the properties of truth or falsity, the fundamental subjects of the predicates 'true' and 'false', are not linguistic items, neither sentences nor utterances of sentences. It is not, when we speak or write, the words we then use, but what we use then1 to say, that is in question. It is whatever may be believed, doubted, hypothesized, suspected, supposed, affirmed, stated, denied, declared, alleged, etc. that is or may be true. Any of these verbs may be followed by a noun clause of the form 'that p', and it is precisely the items designated or referred to by these noun clauses, as used on this or that occasion, that are the bearers of the properties of truth or falsity. We do not have, in common use, a general word for these items. We do not have such a word because we do not in practice need it; in practice, we always use a nominalization of one of the verbs in question as the subject of the predicate (for example, 'your belief', 'his allegation', 'that statement', etc.) or a noun phrase such as 'what she has just said' or even the form 'that p' itself. Philosophers have, at various times, made various attempts to supply this deficiency. Frege's 'thought' is one; Austin groped towards it when he distinguished the 'locutionary' act (in terms of sense and reference) from the 'phatic' on the one hand and the 'illotutionary' on the other;9 G. E. Moore and others have happily used the term 'proposition', which, more recently, has shown a tendency to be replaced by 'propositional content' or merely 'content'; an older term still is 'judgement'. Whatever term we use for items of this kind-and I perhaps date myself by being content with old-fashioned 'proposition'-the essential point is that such an item is not to be identified with an inscription or an utterance or a type of inscription or utterance; it is an abstract, intensional entity, but nonetheless an item of a kind such as we constantly think of and refer to whenever we think of, or comment on, what someone has said or written (in the declarative mode) or indeed on a thought that has, as we say, just entered our own heads. It is objected that there is no clear general criterion of identity for such items. Never mind: we get on well enough, and communicate well enough, without one. With the admission of propositions or judgements or thoughts as abstract intensional entities there goes along of course the admission of
9 J. L. Austin, How to Do Things with Words, ed. Press, 1962).
J. o. Urmson
(Oxford: Oxford University
Strawson others: of senses, of concepts, of properties and universals in general. It is here, most obviously, that the risk of inflation comes in: the risk of seductive images, pictures to hold us captive, myths and fantasies that are often fathered, justly or not, on Plato. But in order to acknowledge the items in question as the harmless necessary things they are, regularly recognized in ordinary thought and talk, there is no need to be thus seduced, no need to be taken captive by such pictures. So I have spoken up for subjective experience on the one hand (the contents of inner sense, as Kant would say) and for abstract intensional entities on the other. And this prompts me to remark, in conclusion, on one mildly ironical feature of our subject in the early twenty-first century. If anyone is entitled to be called the founder of our subject, it is generally acknowledged to be Plato: and if anyone could be called the father of its modern development, most of us would nominate Descartes. The irony is that to accuse a philosopher of Platonism or Cartesianism is currently felt to be a seriously damaging charge. But if, and in so far as, I have exposed myself to it, I an1 unrepentant. Of course both these great men were guilty of exaggerations and more or less grave Inistakes. But each had a grasp, however uncertain, of features of our thought and experience that it would be a ll1uch graver mistake to overlook, to deny, or to minimize. 14
2
Strawson and Analytic Kantianism HANS-JOHANN GLOCK
••
It is a commonplace that the reputation of, and interest in, philosophers of the past waxes and wanes from decade to decade. But while even the greatest members of the philosophical pantheon can become unfashionable, some of them have never been neglected entirely. Plato and Aristotle belong to that select group, and so do the founder of modern philosophy, Descartes, and its most eminent representative, Kant. Still, there was a time when interest in Kant was mainly historical in nature, roughly between the 1920S and the 1960s. After the First World War the neo-Kantianism that had dominated academic philosophy on the Continent for fifty years finally ran out of steam. As a dynamic motor of philosophical development neoKantianism was replaced by phenomenology and its hermeneutic offspring on the one hand, by analytic philosophy on the other. The rise of analytic philosophy is often described as a sustained revolt against Kant. There is some truth in this idea. After flirtations with Kant and Hegel, Moore and Russell rebelled against idealism and initiated the complementary programmes of conceptual and logical analysis. Subsequently, the credo of the most influential school of analytic philosophers, the logical positivists, was the rejection of Kant's idea that there are synthetic judgements a priori. Next, proponents of Oxford conceptual analysis frowned upon the system building that characterized both Kant and neo-Kantianism, and replaced it by piecemeal investigations into the use of philosophically relevant expressions. Finally, in the wake of Quine, analytic philosophy has increasingly been dominated by naturalism, and hence by the anti-Kantian idea that philosophy is identical or at least continuous with empirical science. Nevertheless, the received contrast between Kant and analytic philosophy is untenable. For one thing, there is a distinctive anti-naturalist tradition within analytic philosophy, which insists that philosophy-especially logic, epistemology, and semantics-differs from natural science not just quantitatively but qualitatively. Among its godfathers are not just proclaimed adversaries of
16
Glock
Kant, like Bolzano and Moore, but also Frege and Wittgenstein. Both of these thinkers developed Kant's anti-naturalism, albeit in strikingly different ways.l For another, Kant's account of metaphysics and a priori knowledge set the agenda even for those who rejected the synthetic a priori. More importantly, in spite of their anti-Kantian rhetoric, many logical positivists accepted the Kantian idea that philosophy is a second-order discipline. Unlike science or common sense, philosophy is a priori not because it describes objects of a peculiar kind, such as the abstract entities or essences postulated by Platonism and Aristotelianism, but because it reflects on the conceptual scheme that science and common sense employ in their empirical descriptions and explanations of reality. This Kantian undercurrent is no coincidence. The Tractatus, arguably the most important text in the rise of analytic philosophy, sets philosophy the Kantian task of drawing 'the limit of thought', rather than that of adding to our scientific knowledge of the world. Schlick and Carnap accepted the division of labour suggested by Wittgenstein, presun1ably because they were steeped in neo-Kantian ideas through their philosophical apprenticeship in Germany. Indeed, there is only a single step from the claim of the Marburg school that philosophy is the meta-theory of science to Carnap's slogan that philosophy is the 'logic of science' ,2 that step being the linguistic turn of the Tractatus, according to which the logical limits of thought are to be drawn in language. Accordingly, the mainstream of analytic philosophy from Frege to Quine is not just decisively shaped by Kantian problems, it also includes important Kantian strands. At the same time, none of these strands amounts to anything one might call analytic Kantianism; namely, a distinctly analytic interpretation, defence, and elaboration of Kant's ideas. 3 It is hardly surprising that the 1 See, respectively, my 'Vorsprung durch Logik: The German Analytic Tradition', in A. O'Hear (ed.), German Philosophy since Kant, Lectures of the Royal Institute of Philosophy (Cambridge: Cambridge University Press, 1999) and 'Kant and Wittgenstein: Philosophy, Necessity and Representation', International Journal of Philosophical Studies, 5 (1997), 285-305. References to Kant's Critique of Pure Reason are to the first (A) and second (B) edition, and to his other works according to the Akademie Ausgabe (Berlin: Walter de Gruyter, 1902- ), volume number followed by page number. 2 The Logical Syntax of Language (London: Routledge & Kegan Paul, 1937), 279. 3 In the German literature one often encounters the term Analytische Transzendentalphilosophie (e.g. R. Aschenberg, Sprachanalyse und Transzendentalphilosophie (Stuttgart: Klett-Cotta, 1982), 28-34; T. Grundmann, Analytische Transzendentalphilosophie (Paderborn: Schoningh, 1994). But 'analytic Kantianism' is superior to 'analytic transcendental philosophy', and not just for reasons of elegance. In Kant himself we find conflicting accounts of what Transzendentalphilosophie amounts to. For example, he often seems to equate transcendental philosophy with the critique of pure reason (explicitly in Reflections §4897), while officially regarding it as the complete critical metaphysics for which the critique provides the foundations (A 10-161B 24-30). In the same passage he unequivocally confines transcendental philosophy to theoretical reason, which implies that the label is unsuitable for the important attempts to develop Kant's moral philosophy in an analytic vein (see below).
Strawson and Analytic Kantianism 17 initial pioneers of analytic philosophy-Frege, Moore, Russell, Wittgenstein, and the logical positivists-were not interested in this kind of endeavour, even if they were indebted to Kant. To be sure, there were soon philosophers who combined an acquaintance with analytic philosophy with a sympathetic interest in Kant. C. D. Broad, for example, regularly lectured on Kant in Cambridge both before and after the Second World War. But these lectures were published only in 1978. Stephan Korner's Kant of 1955 was far more influential. But, although it has deservedly been popular in courses on Kant, it did not spark a flurry of publications by analytic philosophers. 4 A major breakthrough came in 1959 with Peter Strawson's masterwork Individuals. s Together with Ryle and Austin, Strawson was the leading representative of conceptual analysis, a loose movement inspired by Moore and Wittgenstein that flourished mainly though not exclusively in Oxford between the 1940S and the 1970s. Ideal-language philosophers like Frege, Russell, and the logical positivists held that natural languages engender philosophical confusion because they suffer from various logical defects, and that they must therefore be replaced by an ideal language-an interpreted logical calculus. By contrast, conceptual analysis tries to resolve philosophical problems by clarifying rather than replacing the concepts that give rise to them. And this analysis or clarification proceeds by describing the use of those words in which philosophically troublesome concepts are expressed. That analytic Kantianism should receive its main impetus from conceptual analysis rather than from ideal-language philosophy is unsurprising. While conceptual analysts tended to be suspicious of metaphysics, they did not display the anti-metaphysical fervour of the logical positivists. They were far less obsessed with denouncing the synthetic a priori, and showed a fair degree of syn1pathy towards Kant. In Ryle this sympathy may have been reinforced by reading the Tractatus and by conversations with its author. In any event, in articles from the 1930S and 1950S Ryle applauded Kant's separation of philosophy from science. He also commended his programme of identifying the categories by looking at forms of judgement while sharply condemning its execution, setting a precedent that later analytic commentators on the Metaphysical Deduction were to follow. 6 There was also an institutional reason for the association between Kant and Oxford conceptual analysis. As Strawson informs us in Chapter I, students specializing in philosophy as part of the PPE course at Oxford
4 See c. D. Broad, Kant: An Introduction (Cambridge: Cambridge University Press, 1978) and S. Korner, Kant (London: Pelican, 1955). 5 Individuals: An Essay in Descriptive Metaphysics (London: Methuen, 1959). 6 See G. Ryle, Collected Papers, ii (London: Hutchinson, 1976), 366, 176-9, and J. Bennett, Kant~s Analytic (Cambridge: Cambridge University Press, 1966), ch. 6.
18
Glock
were forced to study Kant. As a result, Strawson struggled with the Critique of Pure Reason both as an undergraduate and as a college tutor. But there is no direct sign of this struggle in his early writings. Rather, Strawson came to fame by criticizing orthodoxies of logical analysis. Natural languages, he maintained, are distorted by being forced into the Procrustean bed of formal logic, and hence the latter is not a sufficient instrument for revealing all the logically and philosophically relevant features of our language. For my current topic, the most interesting case in point is Strawson's attack on Russell's theory of descriptions. According to Strawson, a sentence like 'The present king of France is bald' is neither true nor false rather than simply false. Furthermore, it presupposes rather than entails the existence of the present king of France; i.e., that existence is a necessary precondition of the statement being either true or false. Finally, by trying to paraphrase away singular referring expressions of the form 'the so-and-so', Russell ignores the distinctive and indispensable role that these expressions play within our language. The tenor of Individuals is more constructive than that of Strawson's previous work. The focus shifts from the description of ordinary use to what Strawson calls descriptive metaphysics. Descriptive metaphysics differs from the revisionary metaphysics one finds in Descartes, Leibniz, or Berkeley, among others, in that it 'is content to describe the actual structure of our thought about the world', rather than attempting 'to produce a better structure'. It differs from previous Oxford analysis in its greater scope and generality, since it seeks to 'lay bare the most general features of our conceptual structure'. These are visible not in the motley of ordinary use, but in fundamental functions of thought and discourse, notably those of reference-picking out an individual item-and predication-saying something about it (pp. 9-10). In spite of the shift marked by Individuals, therefore, there is an abiding concern in Strawson's work; namely, with describing the most general and pervasive features of human thought about the world, in particular 'the operation of reference and predication', and with the presuppositions of such operations. 7 Alongside Aristotle, Individuals lists Kant as the most eminent representative of descriptive metaphysics. Strawson's conception of metaphysics also owes a more specific debt to Kant. As Peter Hacker points out in Chapter 3, by contrast to traditional metaphysics, descriptive metaphysics yields insights not into the necessary structure of reality, but into our 'conceptual schen1e', the connections between the fundamental concepts we use to think about and describe the world. This shift of focus from reality
7 See 'My Philosophy', in P. K. Sen and R. R. Verma (eds.), The Philosophy of P. R Strawson (New Delhi: Indian Council of Philosophical ~~e~~~~~~5J,_I~ - -
Strawson and Analytic Kantianism
19
to our thought or discourse is familiar from the linguistic turn of analytic philosophy, yet it is also a Kantian legacy (see Sect. I below). At an even more specific level, part I of Individuals elaborates a Kantian idea; namely, that our reference to objects depends on our capacity to identify and reidentify them, which in turn depends on the possibility of locating them within a single public and unified framework, the framework of the spatiotemporal world (pp. 62-3, 119). Finally, Individuals maintains that philosophical scepticism distorts or ignores the essential structure of our conceptual scheme. Making his debt to Kant explicit, Strawson used the label 'transcendental argument' for a type of argument that rebuts scepticism on the grounds that these distortions are self-refuting. The Kantian themes in Individuals are unmistakable, though diverse and combined with distinctly Strawsonian ideas in philosophical logic. It is no coincidence, therefore, that among the results of the book was a new kind of debate about Kant. On the one hand, this debate was less historical and deferential than previous Kant scholarship, including anglophone commentaries like those of Paton or Kemp Smith. On the other hand, it was more exegetical and scrupulous in its treatment of Kant than the passing animadversions and commendations of previous analytic philosophers. 8 One important early instance of this new style was Graham Bird's Kant~s Theory of Knowledge of 1962. Its main positive aim was to clarify the relation between appearances and things as they are in themselves, a topic that does not feature in Individuals. But the book explicitly sets out to provide an exegetical basis for the kind of analytic discussion of Kant exemplified by Individuals, and it includes a sustained comparison of Kant and Strawson on the self and personhood (p. ix; ch. II). A slightly later example of analytic Kantianism is Bennett's Kant~s Analytic of 1966. It sets out to fight Kant 'tooth and nail' (p. viii), and treats him as a contemporary analytic philosopher to be compared and contrasted with other contemporaries, Strawson pre-en1inent among them. In the san1e year Strawson himself entered the fray once more. Having been 'subtly and in part consciously influenced' by the first Critique in his independent work on metaphysics and epistemology, he decided to get to grips with the work as a whole, and for its own sake. He started offering lecture courses on the Critique in 1959, and these lectures eventually led to the publication of The Bounds of Sense in 1966.9 The book is not a straightforward commentary on Kant's masterpiece, but an essay that
8 In some respects, Rawls's A Theory of Justice had a similar impact in the sphere of moral philosophy. It is a highly original work, yet subtly influenced by Kant. And although it did not itself purport to interpret Kant's moral philosophy, it spawned numerous such attempts. 9 The Bounds of Sense: An Essay on Kant's Critique of Pure Reason (London: Methuen, 1966). Unless otherwise specified, page references in the text are to this book.
20
Glock
provides a reconstruction of some of its central ideas in the style of analytic philosophy. As Strawson puts it in this volume, it was a 'somewhat ahistorical attempt to recruit Kant to the ranks of the analytical metaphysicians, while discarding those metaphysical elements that refused any such absorption'. The basic interpretative idea of Bounds of Sense is ingeniously epitomized by the title. There are three strands to the Critique. On the one hand, against empiricism Kant maintains that 'a certain minimal structure is essential to any conception of experience which we can make truly intelligible to ourselves' (p. II, see pp. 24,44). On the other hand, against rationalism he insists that concepts-including the categorial concepts that define this n1inimal structure-cannot be applied beyond the limit of possible experience. In these two regards, Kant seeks to draw, respectively, the lower and the upper bounds of sense. But he does so from within a framework that itself transgresses the bounds of sense, a framework that consists of the untenable metaphysics of transcendental idealism and the 'imaginary subject of transcendental psychology' (p. 32). The first two strands constitute the fruitful side of the Critique, the third constitutes its 'dark side', which is 'no longer acceptable, or even promising'. The central task of the interpreter is that of 'disentangling' an 'analytical argument' that 'proceeds by analysis of the concept of experience in general' from its idealist and psychologistic surroundings (pp. 16, 3 I). Strawson has done more than anyone else to stimulate interest in Kant among analytic philosophers, and to show how the Critique can be approached in an analytic spirit. To this extent he is the n10st important source of analytic Kantianism in a wide sense of the term. Furthermore, his own approach amounts to an analytic I
Strawson and Analytic Kantianism
21
1. THE GENERAL NATURE OF TRANSCENDENTAL PHILOSOPHY
The first readers treated the Critique primarily as a contribution to metaphysics, both positive and negative. On the one hand, there was the 'alldestroying' Kant who had swept away the 'pre-critical' or 'transcendent' metaphysics of post-Cartesian rationalism and of traditional philosophy more generally. On the other hand, the German Idealists soon treated Kant's own 'critical' or 'transcendental' metaphysics as a mere stepping .stone towards metaphysical systems that were even grander and more pretentious than those of pre-critical metaphysics. With the collapse of German idealism in the middle of the nineteenth century, however, the focus of Kant interpretation shifted from metaphysics to epistemology. For the neoKantians, Kant's lasting legacy was to establish the theory of knowledge as the fundamental discipline of philosophy.lO Similar views are evident among analytic philosophers. Russell regarded it as one of Kant's few achievements to have 'made evident the philosophical importance of the theory of knowledge' .11 Admittedly, some of the analytic foes of n1etaphysics realized their debt to Kant's attack on transcendent metaphysics, but this was far outweighed by their reservations about transcendental metaphysics and the idea of synthetic judgen1ents a priori. Strawson has generally been read as adopting a thoroughly epistemological approach to Kant. In fact, many critics treat it as a defining feature of his 'analytic interpretation' that it seeks to refute scepticism by way of transcendental arguments. 12 This picture is n1isleading. The prime concern in Individuals is with sketching a new type of metaphysics. Scepticisn1 is mentioned only a couple of times and transcendental arguments only once. The sceptic features primarily not as someone who doubts the possibility of knowledge but as someone who distorts our conceptual scheme. And that certain forms of scepticism are self-refuting is a mere corollary of delineating the structure of our conceptual scheme. 10 'All representatives of "Neo-Kantianism" have been agreed on one point: that the heart of Kant's system is to be sought in his theory of knowledge, that the "fact of science" and its "possibility" constitute the beginning and aim of Kant's putting of the problem' (E. Cassirer, 'Kant und das Problem der Metaphysik', Kant-Studien, 36 (1932), 2 (my trans.). 11 Problems of Philosophy (Oxford: Oxford University Press, 1980; 1st edn. 1912), 46. 12 E.g. S. Gardner, Kant and the Critique of Pure Reason (London: Routledge, 1999), 32. Gardner's characterization of the difference between analytic and idealist interpretations is particularly puzzling, since he presents the latter as preoccupied with the 'problem of reality', the Cartesian problem of explaining how our representations can agree with their objects in reality. According to Gardner, Kant is concerned with a general problem about our representations of reality, i.e. a problen1 that includes synthetic judgements a posteriori. As we shall see instantly, however, at least in his critical writings, Kant explicitly confines himself to a specific problem about our a priori representations of reality, the problem of the synthetic a priori.
22
Glock
By contrast, in The Bounds of Sense the emphasis is on epistemology in one respect. It reconstructs the Critique as an attempt to delineate the essential structure of experience. Even here, however, the prinle target is not the sceptical denial of the possibility of knowledge, but rather the contrasting excesses of rationalists on the one hand, sense-data empiricists on the other-excesses that distort epistemic concepts like experience and selfconsciousness (p. 12). To be sure, much of analytic Kantianism has been preoccupied with the question of whether transcendental arguments are capable of refuting the sceptic. But this preoccupation originated not with Strawson but with Stroud's famous attack on a few seminal pages in Individuals. Strawson himself moved from the idea of a descriptive llletaphysics that delineates our conceptual scheme to the idea that a central part of that scheme, namely the apparatus of reference and predication, of singling out and characterizing objects, is in turn connected to the concept of self-conscious experience. 13 Whether consciously or not, this move from meta-philosophy and metaphysics to epistemology parallels Kant's own transcendental philosophy. The master problem of the Critique of Pure Reason is not, as is often supposed, to vindicate the possibility of empirical knowledge (notably of Newtonian physics) against various forms of epistemological scepticism, especially concerning the external world and inductive reasoning. Although, as we shall see, epistemological issues do play an important role, I(ant is arguing primarily not against the Cartesian lunatic who thinks that he might be dreaming, or the Humean neurotic who wonders whether the sun will rise tomorrow. He confronts the more sensible sceptic who challenges the possibility of metaphysical knowledge, of a priori insights into the essence of reality.14 This challenge necessitates a 'science of the limits of human reason', a 'metaphysics of metaphysics' (Dreams of A Spirit-Seer,
13 Other differences between Individuals and Bounds of Sense are discussed in Sect. III. One common view is that the former is content to describe our conceptual scheme, whereas the latter takes on the truly Kantian or transcendental task of validating that scheme against the sceptic. I am sceptical about this characterization, since Individuals does suggest, if only in passing, that sceptical attacks on our conceptual scheme are self-refuting. 14 'The Critique was intended to discuss the very possibility of metaphysics', and it is this problem 'upon which my Critique entirely hinges' (Prolegomena, app.). By contrast, neither mathematics nor the natural sciences require a vindication of their possibility, since they have already proven that they are capable of obtaining knowledge. ' "How are synthetic propositions a priori possible?" ... We have to do here only with metaphysics' (Prolegomena §S). 'There is no need for a· critique of reason in its empirical employment ... nor is it needed in mathematics' (A 7ro-r rIB 738-9; see A xi, A 738-S7IB 766-85). 'The extension of scepticism to the principles of our knowledge of the sensible and to experience itself cannot reasonably be considered as a serious opinion held at any period of philosophy, but is perhaps a challenge to the dogmatists to prove those principles a priori, on which the possibility of experience itself is based. And since they were incapable of doing so, it presents to them these_E~~~~ipl:~!_s_~g~_al1Y_c!1!1:>iQl!~'
_
Strawson and Analytic Kantianism
23
ii. 368; letter to Herz, II May 1781, x. 269). The master problem of this discipline, otherwise known as the critique of pure reason, is a metaphilosophical one; namely: is metaphysics possible as a science? (B 22, A 75 8- 69/B 7 86-97; Prolegomena §§5, 57, 59)· Kant recognized that this problem can be fruitfully pursued only given a proper understanding of the status of metaphysics. To this end he introduced two famous dichotomies. 15 The first contrasts a posteriori knowledge, which is based on sensory experience, with a priori knowledge, which is independent of experience, not as regards its origin, but as regards its validity (B 1-3). The second distinguishes between analytic and synthetic judgements. Initially, Kant defines analytic judgements as those in which the predicate is implicitly contained in the concept of the subject, while in synthetic judgements the predicate is not part of the subject-concept, but connected with it through a third thing (A 8). The notorious flaws of this definition are avoided by Kant's second explanation: analytical judgements are true solely in virtue of the 'principle of contradiction', which is to say that their negation is self-contradictory. They can be verified solely by examining their constituent concepts, whereas synthetic judgements must be verified by examining the objects they refer to (B 12, A 151-2/B 190-1, A 598/B 626; Prolegomena §2). A priori judgements of an analytic kind (as in formal logic) are unproblematic, since their truth is guaranteed by the interrelation of their constituent concepts (B I I). SO are synthetic judgements a posteriori (as in empirical science), since the connection of subject and predicate is borne
(What Real Progress has Metaphysics Made in Germany?, xx. 263; my trans.). This crucial point is missed not just by those who regard transcendental philosophy as an attempt to vindicate the possibility of representation in general, but also by those who regard it as a mere meta-theory of the scientific and mathematical knowledge of the day. The first to deny that the Critique is primarily a treatise on knowledge (whether everyday or scientific) was Heidegger, Kant and the Problem of Metaphysics (Indianapolis, Ind.: University of Indiana Press, 1971), 21. The denial is substantiated by D. P. Dryer, Kant~s Solution to the Problem of Verification in Metaphysics (London: Allen & Unwin, 1966). 15 For the critique of pure reason the 'first step ... is the distinguishing in general of analytic from synthetic judgements ... The second step is merely to have put forth the question: How are synthetic judgements a priori possible?' (What Real Progress, xx. 265-6). It is therefore misguided when Patricia Kitcher criticizes Strawson's approach to Kant on the grounds that it assumes the possibility of distinguishing between analytic or conceptual and synthetic or factual propositions (Kant~s Transcendental Psychology (Oxford: Oxford University Press), 26-7). Kant opposes the rationalist ideal according to which philosophy simply draws conclusions from analytic definitions, as does Strawson, of course. But this in no way entails that Kant anticipated Quine in rejecting the distinction between the analytic and the synthetic or the a priori and the a posteriori. If Quine is right, then Kant's transcendental philosophy is simply a non-starter. Fortunately he isn't, as Strawson himself did much to show. (See Grice and Strawson 'In Defense of a Dogma', Philosophical Review, 65 (1956) and Glock, Quine and Davidson on Language;, Thought and Reality (Cambridge: Cambridge University Press, 2002), ch. 3.)
Glock out by the experience of the object to which the subject-concept refers. Such 'help is entirely lacking' (A 9/B 12), however, for synthetic judgements a priori, since they hold true of objects, but independently of experienceour only way of getting in touch with objects (see A IO-I liB 14-1 5, A 501 B 74, A 13 liB 17 0 ; Groundwork of the Metaphysics of Morals, iv. 387-9). Nevertheless, Kant claims that the judgements of n1athematics and metaphysics fall in this class. And he devotes transcendental philosophy to the explanation of how synthetic knowledge a priori is possible: 24
I entitle transcendental all knowledge which is occupied not so much with objects as with the mode of our knowledge of objects in so far as this mode of knowledge is to be possible a priori. (B 25, see A S6/B 8 I)
Contrary to many commentators, Strawson included (p. 18), transcendental philosophy is concerned in the first instance not with experience, whether scientific or everyday, or with representation and knowledge in general, but exclusively with synthetic ('of objects') knowledge a priori. 16 Nevertheless, in showing that and how synthetic knowledge a priori is possible, transcendental philosophy does turn into a theory about the necessary preconditions of the possibility of experiencing empirical objects. For, Kant's resolution to the riddle of synthetic knowledge a priori runs as follows: the conditions of the possibility of experience in general are likewise conditions of the possibility of the objects of experience, and ... for this reason ... have objective validity in a synthetic a priori judgement. (A IS 81B I9 6, see A IS s-6/B I94-S)
This famous passage links the meta-philosophical master problem of the Critique-the possibility of metaphysics-to its epistemological concern with the preconditions of experience. Metaphysical judgements hold true of the objects of experience (i.e. are synthetic) independently of experience (i.e. are a priori) because they express necessary preconditions for the possibility of experiencing objects. For example, we experience objects as located in space and time, and as centres of qualitative changes that are subject to causal laws. According to Kant, these are not empirical facts about human nature, but necessary features of experience, which at the same time define what it is to be an object of experience. There is a difference between experiences and their objects, and the content of experience is contingent. But, according to Kant, there are also necessary or structural features of experience, and these determine the necessary or essential features of the objects of experience.
16 Certain passages define 'transcendental' in this way (e.g. Prolegomena app. fn.), but they already presuppose Kant's solution to the problem of how synthetic a priori knowledge is possible; namely, that it expresses preconditions of experience.
Strawson and Analytic Kantianism
25
The result is what I regard as the defensible core of Kant's Copernican revolution. Kant maintained that there could be no a priori truths about reality, because experience is our only way of finding out about reality. Even though there are synthetic a priori truths, these are not de reo Instead of describing mind-independent essences of objects, they articulate 'necessary preconditions for the experience of objects'; that is, the essential features of the way we experience them. The ontological search for essences is thus transformed into a second-order reflection on our conceptual scheme. Kant accepts the Aristotelian idea that metaphysics (more precisely, metaphysica generalis) is the study of being qua being, or, in his terms, of 'whatever is insofar as it is' (A 845/B 873). But for him the investigation of being qua being is not an investigation of an abstract entity or of the most pervasive features of reality. It is an investigation into the essential features of objects, and hence into what it is to be an object. These essential features of objects are determined by the essential ways in which we experience or represent objects. 'Metaphysics deals not with objects, but with cognitions (Erkenntnissen)'; it turns from a 'dogmatic enquiry concerning things (objects)' into a 'critical enquiry concerning the limits of my possible knowledge' of objects (A 75 8/B 7 86; Reflections §4 853, see §§395 2, 4457). The descriptive metaphysics of Individuals is likewise a second-order discipline that reflects on our conceptual scheme. Furthern10re, just like Kant's 'transcendental logic', which investigates preconditions of thinking about objects (A 50-64/B 74-88, A 13 liB 170), Strawson's descriptive metaphysics investigates presuppositions of thought that go beyond formal logic. In both cases these presuppositions concern reference to objects, a feature that is presupposed not just by true thoughts or statements, but by any thoughts or staten1ents about reality that are even in the running for truth. Finally, in The Bounds of Sense descriptive metaphysics turns to the concept of experience, and in particular to the presuppositions of self-conscious experIence. One question remains, however. Is Strawson right in thinking that Kant is concerned with spelling out 'the limiting features ... of any conception of experience which we can make intelligible to ourselves' (p. 24)? Or is Kant's theory a much more specific meta-discipline; namely, one that spells out the preconditions of the science of his day-Euclidean geometry and Newtonian physics? This second option was adopted by Cohen, and later by Collingwood and Korner. Strawson avers that it runs counter to Kant's intentions, and would make the Critique 'a less interesting work than we hoped' (p. 120). For this he has been reprimanded by Friedman, who maintains that the Critique is a meta-theory specifically of Euclidean geometry and Newtonian physics. This reproach is unwarranted on three counts. As regards Kant's system, it ignores the fact that the Metaphysical Foundations rather than the
26
Glock
Critique is supposed to provide a meta-theory of physics. Euclidean geometry and the a priori elements of science playa role in the Critique, but only because, according to Kant, they show that synthetic a priori knowledge is possible in principle, thereby easing the way for his vindication of transcendental n1etaphysics (see B 14-17). Furthermore, given the fallible and dynamic character of natural science, as its meta-theory transcendental philosophy would be relative to a scientific epoch and subject to change. This runs counter to Kant's ambition of providing by way of a priori reasoning a metaphysical system that is apodeictically certain and con1plete once and for all (A xii ff., xx, B xxiii). Of course, Kant may be wrong to think that such a metaphysical system is feasible. On the other hand, Kant and Strawson may be right to think that 'there are categories and concepts which, in their most fundamental character, change not at all' (Individuals, 10). Certainly, those features of experience they focus on-self-consciousness, the distinction between experiences and their objects, space and time, causation, etc.-are not the prerogative of science, let alone of a specific scientific theory. If they can be shown to be necessary preconditions of any intelligible experience whatever, this will be an even more interesting philosophical result than showing that certain concepts or principles are presupposed by specific scientific paradigms.
II. TRANSCENDENTAL IDEALISM AND TRANSCENDENTAL PSYCHOLOGY
The next question is how to conceive of such transcendental preconditions of experience, of 'the conceptual structure which is presupposed in all empirical enquiries' (p. 18). It is at this point that Strawson detects both a light and a dark side of the Critique. That there is such a dark side should not come as a shock. After all, there is the book's manner of presentation, which can only in part be excused on account of its complex genesis. The style is forbidding and the architectonics has rightly been described as 'baroque' or even 'grotesque'.17 Add to this Kant's disregard for terminological coherence-surprising in a certified Prussian pedant-and his propensity for contradicting himself,18 and the result contrasts unfavourably with most
Bounds of Sense, 24; Bennett, Kan(s Analytic, 89. Three examples in lieu of many others. First, B 3 calls the principle 'every alteration has its cause' a non-pure synthetic judgement a priori, while according to B S it is a pure synthetic judgement a priori. Next, according to A 341 and 400-1 I is a concept, according to A 34S-6/ B 403-4 it is not. Finally, according to the latter passage it is a representation 'wholly bereft of content', while according to A 400 being a substance and being simple are features of the concept I. 17
18
Strawson and Analytic Kantianism
27
texts analytic philosophers have taken seriously, not to mention an elegant and stylish book like The Bounds of Sense. What has been harder to swallow, at least for Kant scholars, is Strawson's claim that central doctrines of the book are fundamentally flawed. In this context it is important to distinguish two questions that have not always been kept apart. The first is whether Strawson is right to attribute certain doctrines to Kant, the second is whether these doctrines are indeed untenable and unfruitful. The verdict of recent critics has tended to run as follows: as regards transcendental idealism Strawson has committed an exegetical mistake, but he is right to repudiate the kind of idealism that he finds in Kant; as regards transcendental psychology, Strawson is right to ascribe it to Kant, but he has committed a substantive mistake in repudiating it. I shall defend Strawson on both counts, though I am painfully aware that I cannot do them full justice in this essay. Both transcendental idealism and transcendental psychology are ways of spelling out Kant's Copernican revolution: 'we can know a priori of things only what we ourselves have put into them' (B xviii). The traditional-and most straightforward-interpretation treats this as a genetic claim. Necessary preconditions of experience are features to which the objects of experience have to conform because they are imposed on them by our cognitive apparatus (the passive faculty of sensibility and the active faculty of understanding) in the course of processing 'sensations' (Empfindungen), the material component of our experience. Hence they hold only for appearances, but not for the things in themselves that cause our experiences by 'affecting' (affizieren) our cognitive apparatus (see e.g. A 19-20/B 34). According to Strawson, transcendental idealism is the model of 'the mind producing Nature as we know it out of the unknowable reality of things as they are in themselves' (p. 16). He realized, of course, that Kant tried hard to set apart his 'transcendental idealism' from Berkeley's 'empirical idealism'. Nevertheless, Strawson occasionally suggests that the only difference is that for the former our own states of consciousness enjoy no superiority over physical objects, since even they are mere appearances of an underlying reality (pp. 21-2). By contrast, his critics (e.g. Bird, Allison, Gardner) have welcomed transcendental idealism as a 'salubrious recommendation of epistemic modesty' .19 In my view, they are right to insist that transcendental idealism does not boil down to Berkeley's idealism. Kant distinguishes transcendentally ideal features like space and time from secondary qualities, which, following Locke, he regards as subjective. Only the former are 'necessary conditions
19
P. Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press,
19 8 7),333.
28
Glock
under which alone things [die Gegenstande] can become objects of the senses for us' (A 28-9, see B 45). Transcendental idealism insists that the formal aspects of appearances, those that they must possess to be objects of experience, are due to our cognitive apparatus. But it allows, and indeed insists, that the material aspects of appearances are completely non-mental, a matter of brute fact rather than the product of any human or divine spirit. In fact, Strawson himself acknowledges this point in his discussion of transcendental idealism (pp. 236-7). However, this acknowledgement leaves intact his main complaint about transcendental idealism; namely, that it rests on the idea that appearances are mental items that are produced by non-mental things in themselves causally interacting with our form-imposing cognitive apparatus. In response, idealist interpreters insist that the distinction between appearances and things as they are in themselves is not a distinction between two worlds, a world of mental representations and a non-mental yet unknowable world. Rather, it is a distinction between two perspectives, two standpoints that highlight different aspects of one and the same world. From an empirical perspective, appearances are real, i.e. mind-independent. From a transcendental perspective, however, i.e. considered in relation to our mode of cognition, they are transcendentally ideal in the sense that we can experience them only in so far as they conform to the requirements of our cognitive apparatus. This defence is already implicit in Bird, and was later elaborated powerfully by Allison. 2o According to Allison, there are 'episten1ic conditions' to which all possible objects of experience must conform, and this constitutes their transcendental ideality. The double-aspect interpretation has many virtues. It has a foundation in the text, and it does render Kant's position more tenable. But it does not rehabilitate it, for two reasons. The first is Kant's demonstrable commitment to the idea that appearances are mental representations. Kant defines transcendental idealism as 'the doctrine that appearances are to be regarded as being, one and all, mere representations [blof5e Vorstellungen]' (A 369, see also A 104: appearances are 'nothing but sensible representations'). But representations, and by implication appearances, are 'inner modifications or our mind', which are capable of going proxy for objects if they are subject to the categories (A 197/B 24 2, see also A 320/B 376-7). This suggests that Kant adopted an obscure version not so much of Berkeleian idealism as of Lockean representationalism. He is committed to the idea that what we experience, namely appearances, are mere mental representations of non-mental things in themselves that are unknowable. But among the things we experience are material objects such as chairs,
20
Kant's Transcendental Idealism (New Haven, Conn.: Yale Universi~~r_e!~~_I~~21"-
Strawson and Analytic Kantianism 29 books, and stones, and hence Kant is committed to reducing these to mental representations. It will not help to plead that they are mental 'only from a transcendental perspective'.21 On an innocuous gloss, to consider an object from a transcendental perspective is to consider it as an object of experience (knowledge, thought), and hence as subject to certain conditions. Let us grant, for the sake of argument, that there is such a perspective and even that it is the proper perspective of philosophy. It does not follow that the object of my current experience, the computer screen in front of me, is a representation, .a mere modification of the mind. At most, if we accept a representationalist theory of perception, it follows that it is the intentional object of such a mental representation, i.e. what the representation represents. There is no coherent perspective, however labelled, from which computer screens are mere mental representations. This is where the second problem comes in. The current defence of transcendental idealism ultimately requires an understanding of the idea of a transcendental perspective that is far from innocuous, because it relies on a genetic theory, transcendental psychology. This reliance is often concealed by the platitudinous formulations of transcendental idealism by some of its recent defenders. A statement like 'If there are formal or epistemic conditions objects must satisfy in order to become objects of human knowledge, then all objects of human knowledge must conform to these formal or epistemic conditions' is a truism. But the whole problem shifts, not just to finding out whether there are such conditions, but to clarifying what such conditions could amount to, given that they are supposed to be neither physiological, nor psychological, nor ontological, nor logical. .l\n analytic interpretation treats then1 as conditions imposed by epistemic concepts such as that of experience, or of an object of experience. But the idealist interpretation is forced to treat them as genetic conditions of some kind. It considers objects from a transcendental perspective in the sense of regarding them as constituted by the mind, at least in their formal aspects. In an exegetical respect this is not a problem. It should be uncontroversial that the Critique, especially the first edition, involves such a genetic tale. There are numerous genetic and quasi-causal claims about the relation between things as they are in themselves, our cognitive apparatus, and
21 Tome it seems that the only way of salvaging Kant's position is to invoke his claim that appearances are merely the undetermined objects of empirical intuition (A 20/B 34), and hence not the objects of a full-blown 'en1pirical judgement' but only of a 'judgement of perception' (Prolegomena § I 8). This defence is promising in that, unlike the latter, the former are judgements which involve the categories, and are hence not about how things appear, but about how they are, and hence capable of being objectively true or false. However, it remains difficult to see how conceptual determination by the categories can turn an appearance-a mental item-into a mind-independent object of experience.
30 Glock appearances. Sensibility, our passive cognitive faculty, imposes its formsspace and time-on the sensations that result from it being affected by things as they are in themselves. The resulting manifold of empirical intuitions is then unified ('synthesized') according to the rules specified by the categories. This is done by our active cognitive faculties, the understanding with the help of its ever ready lieutenant the imagination. What is controversial is whether this tale is sound, either in itself or in the context of Kant's theory. I should like to mention a few reasons for answering this question in the negative. The first is an old chestnut going back to Jacobi, but to my mind it remains unresolved. Kant maintains that things as they are in then1selves 'affect' us, that is, cause sensations in us, and thereby give rise to the material aspects of the phenomenal world. But, according to Kant, causal and temporal relations are structural features imposed on the world by the human mind; hence they can obtain only between appearances, not between things considered in themselves. The problem is not resolved simply by switching to a double-aspect account of transcendental idealism. Even if things as they are in themselves and appearances are the same things considered from different perspectives, by Kant's lights one cannot in the same breath consider a thing in itself and yet apply to it a concept that is confined to the perspective of considering things in relation to our cognitive apparatus. As Strawson shows (pp. 38-9, 248-9), a similar problem arises in I(ant's doctrine of self-awareness. We are aware of ourselves, yet not as we are in ourselves, but only as we appear to ourselves, i.e. in a temporal guise. But what kind of truth about ourselves is it that we appear to ourselves in a temporal guise? If we only appear to ourselves to so appear to ourselves, a vicious regress looms. If we really appear to ourselves in temporal guise, however, this implies that it is a fact about what happens in time, that we temporally appear to ourselves. This in turn would mean, however, that we as we are in ourselves can feature in a temporal fact, contrary to the provisions of transcendental idealism, according to which everything that occurs in tin1e belongs to appearances. Thirdly, there are the various acts of synthesis that are supposed to create the synthetic unity of representations that makes them ascribable to a single consciousness. Kant is explicit that the synthesis that makes experience possible is not an empirical act, but 'a prior synthesis' of a 'transcendental' kind. It is a an act of 'spontaneity' that cannot be performed by me as I appear to n1yself, but only by me as I am in myself (see B 129-3°, 151, 158 n.). But this raises not just the question of how I can know myself to engage in such acts. How can I as I am in myself, conceived as something outside of space land time, even meaningfully be said to engage in an act, something that is temporal by definition. 22 To say that the agent of synthesis 'stands at the 22
See pp. 93-7; Bennett, Kant's Analytic,
111-13.
Strawson and Analytic Kantianism
3I
edge of experience rather than its far side'23 has a suggestive Tractarian ring to it, but hardly solves the problem. According to transcendental idealism, only the empirical, not its edge, whatever it may be, is temporal. Fourthly, according to the transcendental unity of apperception, I can become conscious of a representation only if the latter has been subjected to a 'synthesis according to concepts' (A 104). How then can the epistemic subject relate to the pre-conceptual manifold of empirical intuitions in order to operate on it?24 The alternative is to treat synthesis as an unconscious process that takes place at a sub-personal level, whether it be neurophysiological or psychological in the sense of contemporary cognitive science (A 78/B 103). But then the question arises of whether such processes could be of epistemic rather than merely causal relevance to our experience and knowledge. Kant himself certainly suggested that this question must be answered in the negative. 25 A final problem concerns the status of transcendental psychology. Kant's master problem is that of vindicating and explaining the possibility of synthetic a priori knowledge. It is difficult to see what solution to that problem transcendental idealism and psychology offer, or could offer, other than the genetic tale of mind-imposed structures. But that genetic tale, whatever its other merits or flaws, cannot solve Kant's master problem. It violates his own anti-genetic distinction between the question of how we acquire a certain kind of knowledge or experience (quaestio facti) and the question of what the logical and epistemological status of that knowledge is (quaestio iuris) (B 116-17; see Reflections §§4900-1).26 In line with that distinction, Kant also distinguishes the 'objective' side of the Transcendental Deduction,
Gardner, Kant, 299. 24 Bennett, Kant's Analytic, 112. In a letter to Herz of 26 May 1789 he writes of intuitions that are not subject to the categories: 'I should not be able to know that I have them, and they would therefore be for me, as being possessed of knowledge, absolutely nothing. They might still (if I conceive of myself as an animal) be able to carryon with regularity their play in me (a being unconscious of my own existence) as ideas bound together by empirical laws of association and so having influence upon feeling and desire-it being assumed that I should be conscious of each individual idea but not of its relation to the unity of the representation of the object by meaning of the synthetic unity of its apperception-but I should not thereby be able to know anything, not even to know this state of myself' (xi. 52). 26 Kitcher, Kant's Transcendental Psychology, 15, points out that Kant stresses the importance of origins even to a quaestio iuris, e.g. in maintaining that a deduction must provide 'a certificate of birth quite different from that of descent from experience' (A 86-71B 119). But Kant conceives of origins not in causal or genetic terms, but in terms of justification. This is evident from his conception of the a priori: 'There can be no doubt that all our knowledge begins with experience ... In the order of time, therefore, we have no knowledge antecedent to experience, and with experience all our knowledge begins. But though all our knowledge begins with experience, it does not follow that it all arises [entspringt] out of experience ...' (B I). What makes a belief a priori is precisely not the way we acquire it but the justification we have for it, which is why a priori beliefs need not be innate. 23 25
32 Glock which establishes the objective validity of the categories, from the 'subjective' side, which explains how they have that validity, namely by dissecting our cognitive capacities. And he declares that only the former is essential to his main aim (A xvi-xvii). Furthermore, statements about the factors operative in bringing about certain knowledge claims or features of experience do not follow logically or analytically (conceptually) from a description of the result of that genetic process. Take a central case, the transcendental unity of apperception: the clain1 that something counts as my representation only if I can ascribe it to n1yself (B 131-5). As Kant emphasizes, this is an analytical judgement based on the concept of a representation possessed by me. He also claims, less truistically, that I can ascribe a representation only if it has a certain (objective) unity. But that claim does not entail anything about how that unity came about, that is, about how experiences came to possess the unity that, according to Kant, allows them to be self-ascribable. It does not entail the doctrine of synthesis according to which, in Rorty's words, 'diversity is found [by the sensibility] and unity made [by the understanding]' .27 This point also applies to Patricia Kitcher's idea that 'transcendental psychology analyzes cognitive tasks to determine the general specifications of a mind capable of performing those tasks' .28 Unless such analysis is a conceptual analysis of what the task amounts to-for example, an analysis of what it is for experience to be of enduring objects rather than of how experience comes to be of enduring objects-it must be synthetic. Therefore the genetic claims of transcendental psychology must be synthetic rather than analytic. However, if they were a posteriori, they would belong to 'empirical psychology', a 'physiology of the human understanding' a la Locke (A ix; Prolegomena §21a), not to an a priori discipline like transcendental philosophy. It is true that Kant seems to make transcendental philosophy dependent on some kind of fundamentally contingent fact about human cognitive faculties when he writes: 'This peculiarity of our understanding, that it can produce a priori unity of apperception solely by means of the categories, and only by such and so many, is as little capa.bie of further explanation as why we have just these and no other functions of judgement or why space and time are the only forms of our possible intuition' (B 145-6). To this extent he seems to treat certain features as a priori or necessary if they hold in all possible worlds that we can experience constituted as we are. On the other hand, this claim runs counter to his attempt to provide a metaphysical deduction of the categories from the forms of judgen1ent,
27
28
Philosophy and the Mirror of Nature (Oxford: Blackwell, 1979), IS s-6. Kant's Transcendental Psychology, 13.
Strawson and Analytic Kantianism
33
and to the analytic argument that we can regard objects as independent of our experiences of them only by locating them within a spatio-temporal framework. At any rate, what worlds Kant regards as possible, given our forms of intuition and of judgement does not depend on empirical considerations of any kind. The claims of Kitcher's transcendental psychology, by contrast, are empirical claims about the mechanisms required to bring about a certain ability, however general and abstract they may be. Consequently, a truly Kantian transcendental psychology must itself consist of synthetic knowledge a priori, which means that it cannot without circularity establish the possibility of such knowledge. Unless there is an independent argument for the possibility of a discipline consisting of synthetic a priori knowledge, transcendental psychology at best boils down to a highly theoretical yet empirical psychological theory; at worst, it is simply a fairy tale. 29
III. TRANSCENDENTAL ARGUMENTS
Between the late 1960s and the early 1980s, few methodological or epistemological issues exercised analytical philosophers more than that of transcendental argun1ents. Individuals had first brought the label to prominence, but Strawson was not the first to use it. In 'Are there A Priori Concepts?' of 1939,]. L. Austin characterized an argument for the existence of universals as: a transcendental argument: if there were not in existence something other than sensa, we should not be able to do what we are able to do (viz. name things).3o
And in his lectures of 1950-1 Broad maintained: Kant makes great use of transcendental arguments, and considers that he introduced this kind of argument into philosophy. But so far as I am aware he nowhere explicitly discusses the notion of transcendental arguments. 31
This would indeed be very naughty of Kant, except that he nowhere speaks of transcendental arguments, and therefore might not have felt such a desperate need explicitly to discuss that notion. Kant does refer to four other kinds of argument, which are in some way or other linked to the transcendental arguments of contemporary analytic philosophy. These argun1ents,
29 Strawson points out that it may also contain kernels of truth in the philosophy of mind (p. II); see also his 'Imagination and Perception', in Freedom and Resentment (London: Methuen, 1974). 30 Philosophical Papers (Oxford: Oxford University Press, 1970), 34-5. 31 Kant: An Introduction (Cambridge: Cambridge University Press, 1978), 13.
34 Glock however, he discusses explicitly and at considerable length.
I. In the Transcendental Aesthetics he speaks of 'transcendental expositions' (B 40-I, 48-9), a type of argument designed to show that it is a necessary precondition of our having synthetic a priori knowledge in mathematics that space and time are pure intuitions. II. In the Analytic of Concepts he explains that the Transcendental Deduction, the central argument of the Critique, aims to establish the 'objective validity' of the categories, i.e. that certain formal or structural concepts hold necessarily of all objects of experience (any object of a possible experience) because they are not derived from experience, but are conditions of the possibility of experience (B 116-23). III. The Refutation of Idealism proceeds from the premiss that I am aware of my own inner states and their temporal order, and maintains that it is a condition for the possibility of such awareness that I an1 also aware of the existence of objects located outside me in space. In this way, 'the game played by idealism has been turned against itself' (B 276). IV. The Doctrine of Method provides an elaborate theory of 'transcendental proofs'. These are proofs of synthetic a priori judgements, which is to say they are a priori proofs of synthetic judgements (A I55-7/B 194-6, A I84/B 227-8, A 2I6-I7/B 263-4, A 221-21 B 269, A 238-9/B 29 8, A 259/B 3 1 5, A 7I9-37/B 747- 6 5, A 782-3, B 810-II). Any synthetic judgement requires a 'third thing (X)', which allows the synthesis of subject and predicate. In the case of a priori judgements, this is an empirical intuition. In the case of mathematical judgements, it is a pure intuition. In the case of metaphysical judgements, however, it is a concept; namely, the concept of the 'possibility of experience' (Moglichkeit der Erfahrung) or of 'possible experience' (mogliche Erfahrung). Transcendental expositions have not featured prominently in the analytic debate, since their starting-point, the existence of synthetic a priori knowledge in mathematics, is even more contentious today than it was at the time of the Critique. It also seems that when analytic philosophers fervently discussed transcendental arguments in the late I960s and early I970S few of them had ploughed through the whole of the Critique to reach the Doctrine of Method. 32 As a result, this early debate tended to ignore Kant's own elaborate discussion of transcendental proofs. By the time everyone had made it to the end of the book, the debate had waned, and transcendental arguments had lost much of their appeal. 32 The first to link the discussion of transcendental arguments with an exegetical discussion of the Doctrine of Method seems to have been M. S. Gram.
Strawson and Analytic Kantianism 35 In any event, Strawson's conception of transcendental arguments is connected to (II), (III), and (IV). Among the arguments advanced and discussed in the wake of Strawson one can distinguish two different types. Both of them employ the idea of a necessary precondition. But they differ at least in their manner of presentation. The first type is deductive in that it can be presented in the following form:
Pi We have experience (knowledge) of type K (or the ability to
33
Fundamental Questions in Philosophy (Harmondsworth: Penguin, 1969), ch. 12.
Glock conceptual framework, while at the same time rejecting that framework or one of its preconditions: 36
He [the sceptic] pretends to accept a conceptual scheme, but at the same time quietly rejects one of the conditions of its employn1ent. Thus his doubts are unreal, not simply because they are logically irresoluble doubts, but because they amount to the rejection of the whole conceptual scheme within which alone such doubts make sense. (Individuals, 35, see 106, 1°9)
In line with other analytic critics of scepticism, this line of argument accuses sceptical doubts about the possibility of various kinds of knowledge not of having a false conclusion, or of resting on unsound arguments, but of failing in an even more fundamental way, namely by being senseless or nonsensical. In this respect, it is precisely not an attempt to refute the sceptic directly, as many of Strawson's critics have n1aintained, but an indirect response: the sceptical problems are insoluble because they tacitly reject the preconditions of their own sense. Transcendental arguments of the deductive type are prominent in Bounds of Sense, while Individuals tends to present transcendental argun1ents as elenctic. However, there are various connections between these two types of argument. For instance, elenctic arguments might be used to show that premisses of form P1 are indeed unassailable, because their sceptical denial is self-defeating. Conversely, elenctic arguments seem to trade on the kind of conceptual connections claimed by premisses of form P2' Furthermore, it is possible, at least prima facie, to reformulate elenctic transcendental arguments in a deductive fashion. This is what Stroud has done in his celebrated attack on transcendental arguments. According to Stroud, Strawson's argument in Individuals against scepticism about the existence of persistent particulars starts out from (I) We think of the world as containing objective particulars in a single spatiotemporal system. (1) characterizes the way we think of the world, our conceptual scheme. But among the necessary preconditions of that conceptual scheme are things that the sceptic rejects. One of them is (6) Objects continue to exist unperceived.
'The truth of (6) is a precondition of the sceptical doubts making sense. But how can that be, Stroud queries, since (6), a statement about the world, would have to follow fron1 (1), a statement about how we think about the world? According to Stroud, Strawson engineers the transition through the following steps: (2) If we think of the world as containing objective particulars in a single spatiotemporal framework, then we are able to identify and reidentify particulars.
Strawson and Analytic Kantianism
37
and (3) If we can reidentify particulars, then we have satisfiable criteria on the basis of which we can make reidentifications.
But this does not establish (6), unless we add: (4) If we know that the best criteria we have for the reidentification of particulars are satisfied, then we know that objects continue to exist unperceived.
and ('5) We sometimes know that our best criteria for the reidentification of particulars are satisfied.
Stroud confronts this type of argument with three objections. The first concerns the idea that 'the truth of what the sceptic doubts or denies is a necessary condition of the meaningfulness of that doubt or denial'. 34 According to Stroud, the sceptic can accept the premiss of this argument, namely that the meaningfulness of the notions he uses in order to express his doubts presupposes their validity, which he denies. But, instead of withdrawing doubt, he could conclude that these notions are indeed meaningless. This is an example of a sceptical doubt being propped up by arguments that are much more problematic than the knowledge claims it impugns. It is not up to the sceptic to pronounce that our terms for particulars or mental phenomena might simply be meaningless. These ternlS have an established use, they are understood and can be explained by competent speakers, sentences in which they occur have determinate truth-conditions, etc. There are no grounds for suspecting that they are meaningless, short of a general semantic scepticism that Stroud himself seems to regard as self-refuting. In any event, Stroud concedes that this first objection carries weight only against transcendental arguments that invoke preconditions of the sense of a particular class of terms or propositions. They do not work with regard to preconditions of meaningful discourse in general, presumably because even the sceptic is supposed to stay clear of blatantly self-refuting statements like 'There is no meaningful discourse'. Stroud also concedes that there might indeed be a 'privileged class' of propositions that state 'necessary conditions of language in general, or of anything's making sense to anyone'. Still, even such a transcendental argument would not refute the sceptic: for any candidate S, proposed to be a member of the privileged class, the sceptic can always very plausibly insist that it is enough to make language possible if we believe that S is true, or if it looks for all the world as if it is, but that S needn't actually be true. ('Transcendental Arguments', 128)
34 'Transcendental Arguments', repro in R. C. S. Walker (ed.), Kant on Pure Reason (Oxford: Oxford University Press, 1982), 125.
Glock This objection has been repeated countless times since. It was also anticipated in Broad's lectures of I9 5 I, which Stroud could not have known at the time of writing 'Transcendental Arguments': 38
What Kant claims to prove by his transcendental argun1.ents is that certain propositions, such as the law of causation and the persistence of substance, are true with the interpretation and within the range of application which he gives them. But it is doubtful whether his arguments could prove more than that all human beings must believe them to be true, or must act as if they believed them to be true. (Kant, IS)
This convergence suggests that the objection has some foundation in the texts. Nevertheless, as a general criticism of transcendental arguments it is very peculiar. It amounts to the claim that the premiss P2 of a deductive transcendental argument must always be of the form p~
It is a precondition of experience (knowledge) of type K that we believe that p
Some formulations in Strawson may indeed invite such a construal. But I can see no reason to suppose that all transcendental arguments are in fact based on such premisses, let alone that they have to be based on such premisses. Wittgenstein's argument against the possibility of a private language is often treated as a transcendental one. According to Wittgenstein, there must be standards for distinguishing between correct and incorrect applications of a word, if the latter is to be meaningful. Wittgenstein's opponent, the so-called private linguist, believes that there are such standards in the case of 'private' words, words that cannot be explained to others, even in principle. As Wittgenstein argues explicitly, however, this belief-the mere impression that there are such standards-avails him of nothing. There must be such standards if the private linguist's words are to be meaningful, and this in turn presupposes that the word can be explained to others, not that we believe that it can. Of course, Wittgenstein's argument might fail. But this would be due to a failure in establishing a premiss of form P2 rather than a success in establishing a premiss of form P~. In any event, there are sound premisses of form p 2' even though they may not be terribly exciting. For actions and experiences to be possible, there must be, respectively, agents and subjects of experience. Such preconditions can even be found for the very possibility of meaningful discourse that Stroud mentions. For meaningful discourse to be possible, we must not only believe that there is a language, there must be a language. Indeed, it is arguable that in this case the weaker option is not even coherent, since one cannot believe that there is a language without being a linguistic creature. 35 35 Transcendental arguments need not be what Stroud calls 'modest' ('Kantian Argument, Conceptual Capacities, and Invulnerability', in P. Parrini (ed.), Kant and Contemporary
Strawson and Analytic Kantianism
39
This leaves Stroud's third objection. In effect, it is that statements like (4) presuppose that the fulfilment of our best criteria for particulars persisting between encounters guarantees that they in fact persist between encounters (pp. 120-3). Strawson indeed seems to presuppose something of this kind. Thus he maintains that recognizing the warrant we have for attributing mental states to others is a precondition for attributing mental states to oneself. Scepticism about other minds has to employ our concepts of mental states. But those concepts make sense only if one can distinguish between one's own mental states and the mental states of others. And this in turn presupposes that our normal ways of telling that someone else is in a certain mental state must be 'logically adequate kinds of criteria' (Individuals, 105-6). Logically adequate criteria are such that their fulfilment warrants the attribution of the mental state. But the sceptic is wont to insist that this does not prove that the attribution is indeed correct. Put differently, we can ascribe mental states to ourselves only if we also ascribe them to others. But this only shows that we must have the concept of other minds, and that we must believe that concept to be satisfied when our criteria are met, not that the concept is actually satisfied by anything in reality. More generally, it seems that transcendental arguments invoking logically adequate criteria can prove only that our conceptual scheme must possess certain featuresfor example, that the notion of an objective particular is linked to criteria of reidentification and thus to the concept of unperceived existence-not that these features must be correct (the concepts instantiated).
IV. CONCLUDING REFLECTIONS
Strawson has subsequently accepted this criticism. In Scepticism and Naturalism he concedes that transcendental arguments only establish connections within our conceptual scheme, not anti-sceptical conclusions about the existence of things. Strawson's new response to scepticism is a naturalism inspired by Hume and Wittgenstein, according to which sceptical arguments are idle: they cannot persuade us since we cannot help believing-for example, in material bodies or other minds. What unites Strawson's transcendental and naturalist arguments is that they eschew scepticism and foundationalism alike. The sceptical challenge is not refuted by reference to allegedly indubitable beliefs, but rejected on the grounds that it implies abandoning categories that are indispensable to human
Epistemology (Dordrecht: Kluwer, 1994)), if that means that they only establish conclusions about what we must believe to be the case. Whether what must be the case if such an argument is sound can be a claim about the world depends on the nature of its premiss P l' as I shall argue below.
40 Glock thought: 'Having given up the project of wholesale validation, the naturalist philosopher will embrace the real project of investigating the connections between the major structural elements of our conceptual scheme.'36 In n1Y view, the contrast to Strawson's previous work is less pronounced than commonly assumed. The naturalistic project, as described by Strawson, does not differ from the non-reductive 'connective analysis' that he has always regarded as the main task of his philosophizing. 37 And if the sceptic abandons the preconditions of human thought, he suffers the kind of selfrefutation that transcendental arguments were supposed to reveal. It would therefore be precipitate to rest content with a Humean naturalism, according to which the sceptical doubt is correct though impotent. Even if transcendental arguments cannot establish any 'ontological' conclusions about reality, they may be able to silence the sceptic. If a transcendental argument can show that the sceptic employs concepts that are incompatible with his own doubts, then it prevents him from making a coherent contribution to the debate. That is not the same as refuting the sceptic by proving that we have knowledge, but nor is it a second best. To silence the sceptical doubt by means of argument is to resolve the philosophical problem that it poses. Furthermore, deductive transcendental arguments can establish conclusions about reality, provided that they have suitable major premisses. The premiss that there is meaningful discourse cannot be denied without selfrefutation; yet because of the conceptual connection between meaningful discourse and language it also yields the conclusion that there is language. Other conclusions about reality can be established if the possibility of knowledge is granted. This is something we should do, not on pain of selfrefutation but on pain of being irrational. Moore, in particular, has given us reason for holding that sceptical doubts rest on presuppositions that are more contentious than the knowledge claims they are used to attack. The sceptic is irrational because, in defending his doubts, he is willing to repudiate even the most plausible assumptions. This makes it almost impossible to refute him by a knock-down argument. But it also means that it is more rational to assume that we have knowledge of various kinds than to accept the sceptic's conclusions. This is of course precisely the sort of assumption that deductive transcendental arguments make in their first pren1iss. Given such a premiss, moreover, there is no inherent difficulty in drawing conclusions about reality. As the so-called Benaceraff problem shows, if we have knowledge of nun1bers (as we surely do), then numbers cannot be denizens of a Platonic realm that lies forever beyond the reach of the cognitive capacities of
36
37
Scepticism and Naturalism: Some Varieties (London: Methuen, 1985), 19. See Analysis and Metaphysics (Oxford: Oxford University Press, 1992), chs. 1-2.
Strawson and Analytic Kantianism
4I
human beings. More generally, once we shed the currently popular prejudice that epistemology can have no relevance for ontology, the following point becomes obvious: If we have knowledge about a particular kind of objects, then those objects, must be such that we are capable of knowing them. Transcendental arguments can lead from epistemological premisses to ontological conclusions, without appeal to any kind of transcendental idealism:
°
Pi'"
We have knowledge of objects of type If we have knowledge about objects of type 0, then objects of type have such-and-such properties c~:· Objects of type have such-and-such properties
p~.
°
°
Transcendental arguments of this sort are perfectly legitimate. They is not God, or presuppose knowledge about 0, of course. However, if the end of the universe, but numbers, material objects, and other minds, let's say, that assumption is licensed. 38 Finally, transcendental arguments remain an instrument of descriptive n1etaphysics and connective conceptual analysis, quite independently of scepticisn1 and ontology. They even contribute to a conception of the synthetic a priori that is more sympathetic than Strawson's own dismissal of this issue (Bounds of Sense, 42-4). Kant may be right to deny that all a priori knowledge is analytic in the sense that all a priori propositions can be reduced to logical truths by substituting synonyms for synonyms. But if the genetic claims of transcendental idealism fail, as I have argued, he has not made out a case for the idea that there is necessity that is not conceptual. We can know a priori of things not 'what we ourselves have put into them', as Kant's genetic story had it, but only what we ourselves have put into the concept of a thing, or of an object of experience. Unlike most material objects, concepts are creatures of human thought and action. It is the purpose of the premisses P2 of transcendental arguments to unpack these concepts. Indeed, such a procedure is not just compatible with but strongly suggested by Kant's own claim that transcendental proofs rely on the concept of a possible experience, provided that the latter is cleansed of psychologistic elements. In return, Kant can teach connective analysis an important lesson. Many proponents of such analysis have assumed that a priori propositions have their special status because they provide definitions or explanations of at least some of their constituent terms. Wittgenstein suggests, for instance, that the principle of causality-'Every event must have a cause'-is not
°
38 The ideas of the last two paragraphs are defended at greater length in my 'Wie wichtig ist Erkenntnistheorie?', Zeitschrift fur Philosophische Forschung, 56 (2002).
42 Glock a synthetic a priori truth, as Kant thought, but a 'grammatical proposition' that partly determines what we mean by an event. 39 He seems to be committed to the claim that the concept of an event simply rules out as nonsensical the expression 'uncaused event'. This is not the case. Let's assume that one morning we find dinosaur footprints on the ceiling. Let's also assume that we have a reason to abandon the search for an explanation of the footprints, for example, because the lav/s of nature not only fail to provide one but suggest that none is to be had-the example of quantum mechanics shows that this is at any rate a possibility. Even in that case, we would not cease to call the appearance of the footprints an event. A physical change would be an event, even if a causal explanation of it could be ruled out ab initio. Consequently, being caused is not part of our explanation of the term 'event', or of the linguistic rules governing its use. Kant is right, therefore, to deny that 'Every event has a cause' is analytic, or, more generally, that its status can be explained by simple reference to the concept of an event. Wittgenstein and Strawson claim, plausibly, that 'Every event has a cause' can have an a priori status only because of its constitutive role in our conceptual scheme. 4o Kant claims, plausibly, that it does not simply explicate the concept of an event. A possible resolution of this quandary lies in the idea that the connection between the concept of an event and the concept of causation is provided by a third concept, one that does not itself occur in the judgement, namely that of experience. Events must be caused not because random and chaotic changes do not qualify as events, but because· persistently chaotic events are not possible objects of self-conscious experience. Whether this claim can actually be sustained in the case of 'Every event has a cause' is notoriously problematic. But Strawson has shown that descriptive metaphysics holds the promise of establishing this kind of connection. The transcendental preconditions of experience are conceptual rather than psychological; at the same time, important conceptual connections that involve metaphysical concepts might be mediated by the concept of experience, or, more generally, by the whole network of our epistemic concepts.
Wittgenstein's Lectures 1932-35, ed. A. Ambrose, (Oxford: Blackwell, 1982), 16. I presume that Strawson would underwrite some version of this claim, even though, as Hacker deplores in Ch. 3, he has been reluctant to provide an explicit account of the kind of necessity and a priority that attaches to statements expressing necessary preconditions of experience. 39
40
3 On
Strawson~s Rehabilitation
of Metaphysics P. M. S. HACKER
• ••
1. THE ELIMINATION OF METAPHYSICS
There appears to be a rhythm in the history of philosophy. Periods of great metaphysical system building are followed by brief periods of antimetaphysical reaction. The great constructive systems of seventeenth- and early eighteenth-century metaphysics were followed by Hume's damnation and Kant's drastic restriction of metaphysics (even though we rightly deem both to have elaborated metaphysical systems). The early nineteenth century saw the revival of grand systematic metaphysics among the postKantian German Idealists, only to be followed by the destructive reaction of nineteenth-century positivism. But this too was a passing phase, followed by such metaphysical system-builders as Bergson, Bradley, McTaggart, Alexander, and Heidegger. And this again bred a reaction in the inter-war years. The manifesto of the Vienna Circle, a pamphlet entitled 'The Scientific World Conception', published in 1929, declared that metaphysical assertions are not false propositions, but nonsensical pseudo-propositions. All true or false propositions are either analytic or empirical. The former yield no knowledge of matters of fact. The latter are the totality of cognitively significant propositions. These have a meaning inasmuch as they are verifiable in experience. Metaphysical assertions are not analytic, for they are not true in virtue of the meanings of their constituent terms, are not derivable from the laws of logic and explicit definitions alone, and do not express conventions for the use of terms (these alternative formulations surprisingly being taken to be roughly equivalent). But they are not empirical propositions either, for they are not verifiable in experience, they have no determinable truth-conditions, and make no difference to any possible experience (these formulations being similarly taken to be equivalent).
Hacker The argument was essentially a rerun of I-Iume's condemnation of divinity and school metaphysics as sophistry and illusion. The logical positivists' elimination of what they called 'metaphysics' differed from I-Iume's primarily in the manner in which they harnessed the new logic and the techniques of logical analysis derived from Russell and Wittgenstein to their rationale for the condemnation of metaphysics, and secondarily in their perfunctory willingness to assimilate metaphysical utterances to works of poetry or music. For they were willing to concede that metaphysical utterances might be taken to express an attitude towards life, a Lebensgefuhl. 1 Their criticisms of metaphysics differed from Kant's inasmuch as they rejected the intelligibility of Kant's category of synthetic a priori propositions and hence of his conception of transcendental metaphysics as a description of synthetic a priori conditions of possible experience, and repudiated his clain1 that certain propositions of transcendent metaphysics, though not knowable, are nevertheless intelligible as ideas of reason. It is noteworthy that their primary inspiration was Wittgenstein's Tractatus. For they took the Tractatus to have established that logical propositions in general, and analytic propositions in particular (including propositions of mathematics), are vacuous tautologies which say nothing about reality.2 All cognitively significant propositions are empirical. All empirical propositions, i.e. all propositions with a sense, are reducible to truth-functional combinations of elementary propositions. Elementary propositions were equated, by members of the Vienna Circle, with descriptions of the given, i.e. of immediate experience, which they called 'protocol sentences'. Partly as a consequence of conversations with Wittgenstein in 1929, they took verifiability to be the criterion of significance for empirical propositions, and took the meaning of any empirical proposition to be given by its method of verification. Without warrant, they read this principle of significance and the principle of verification back into the Tractatus. 3 44
1 R. Carnap, 'The Elimination of Metaphysics through the Logical Analysis of Language', repro and trans. in A.]. Ayer (ed.), Logical Positivism (Glencoe, Ill.: Free Press, 1959),78-80. 2 They were quite mistaken in ascribing to Wittgenstein the view that propositions of arithmetic are tautologies. 3 There was reason to interpret Wittgenstein's elementary propositions as what they called 'protocol sentences', immediately verifiable in experience. So there was some warrant for reading the verificationist criterion of meaningfulness into the Tractatus account of elementary propositions, although none for the principle of verification as giving the meaning of even elementary propositions. For, the sense of an elementary proposition was conceived to be its agreement and disagreement with the existence and non-existence of states of affairs, and not its method of verification. Moreover, there was no ground for giving a verificationist interpretation to the account of the meaning of open generalizations. It was only in 1929 that Wittgenstein, perhaps under the influence of Weyl's 'Uber die Neue Grundlagenkrise der Mathematik' (Mathematische Zeitschrift, 10 (1921), 39-79), came to view such sentences as expressions of 'hypotheses', which are not genuine propositions but rather rules for the construction of propositions. He then extended this conception of hypotheses to material-object statements an~s~a!e~~n~s_a~oJl~oJlle~ JD.iQ.d~. _
Strawson~s
Rehabilitation of Metaphysics
45
Interestingly, they disregarded the fact that Wittgenstein, in the Tractatus, had committed himself to the ineffability of metaphysical truths. For, although he had argued that metaphysical truths cannot coherently be stated in language, he claimed that they are shown (but cannot be said) by well-formed propositions of natural language. So his Weltanschauung was profoundly different from theirs. He drew the bounds of sense not in order to eliminate metaphysics tout court, but rather in order to make room for ineffable metaphysics. The main proponent of the anti-metaphysical doctrines of the Circle was Carnapo Already in his 1928 paper 'Pseudo-Problems in Philosophy: The Heteropsychological and the Realism Controversy' he had argued that disputes between realism and idealism regarding the external world or between realism and solipsism regarding other minds are bogus, since no possible experience can decide between the respective alternatives. In 'The Elimination of Metaphysics through the Logical Analysis of Language', published in 1931-2, Carnap argued that n1etaphysical utterances are mere pseudo-propositions which fail to meet the requirement of verifiability that is a condition for possession of cognitive significance. And in The Logical Syntax of Language, published in 1934, he tried to show that the Tractatus~s constraints upon what can be said do not, pace Wittgenstein, make room for ineffable metaphysics and that the pseudo-propositions of the Tractatus~s metaphysics are not attempts to say what can only be shown. Rather, what can be salvaged from such pseudo-propositions are syntactical statements about words. The pseudo-propositions of the Tractatus are pseudo-object sentences in the material n10de that can be recast as genuine syntactical sentences in the forn1al mode. 4 Although Carnap was among the most fervent anti-metaphysicians in the Vienna Circle, others, especially Schlick and Neurath, participated eagerly in the crusade. These doctrines of the Circle were eloquently transmitted to the English-speaking world by A. J. Ayer's pugnacious Language~ Truth and Logic, published in 1936, which became the classic English statement of the central tenets of logical positivism. Like Carnap, Ayer argued that putatively metaphysical propositions are neither tautologies nor empirical propositions. No possible sense-experience is relevant to determining their truth or falsehood. Consequently, they are neither true nor false but rather are literally senseless. Transcendent metaphysics is nonsense, for no statement that refers to a putative reality that transcends any possible sense-experience can have cognitive meaning. In short, metaphysicians produce sentences that fail to conform to the conditions under which alone a sentence can be literally significant. 4 For a detailed examination of Carnap's views see P. M. S. Hacker, 'On Carnap's Elimination of Metaphysics through Logical Analysis of Language', in Hacker, Wittgenstein: Connections and Controversies (Oxford: Oxford University Press, 200r), 324-44.
Hacker The primary respect in which Ayer differed from Carnap was over the relatively trivial concession Carnap made in holding that the utterances of metaphysicians should be treated as on a par with the writings of poets. Ayer observed that the sentences of poetry are not normally forms of tacit nonsense-they may be false, but they typically have a (cognitive) meaning. By contrast, the sentences of the metaphysician are, inadvertently, sheer nonsense. Apart from the logical positivists, there was in the 193 os another powerful source of animus towards traditional rnetaphysics. While the positivists assailed metaphysics with the blunt tool of the principle of verification, Wittgenstein himself was lecturing at Cambridge. By 1932 he had altogether abandoned his Tractatus doctrines concerning the ineffability of metaphysics and the impossibility of making metaphysical statements with any sense, rejected verifiability as a criterion of meaningfulness, and repudiated the principle of verification. Nevertheless, he launched a new onslaught on metaphysics. He now claimed that what appear to be metaphysical statements are at best expressions of rules for the use of the constituent terms of the putative statement in the misleading guise of descriptions of reality, and that traditional metaphysical doctrines, such as solipsism, idealism, and realisll1, are nonsense. His views were not published until after his death, and were known only to the select circle of his pupils and friends. Nevertheless, they undoubtedly contributed to the general anin10sity of analytic philosophers towards metaphysics in the 193 os and even more markedly in the post-war years. Throughout the 1930S in Britain, analytic philosophy, originating in the pre-war work of Moore and Russell, made progress. The earlier style of logical atomist analysis gave way in Cambridge to a new form of analytic philosophy, inspired largely by Wittgenstein's teachings, which spread from Cambridge to other philosophy departments. Logical positivism made inroads among the young Turks of academic philosophy, and Oxford awoke from its slun1bers, guided by Ryle and his circle. The metaphysical systems of the previous generation of Bradley, Alexander, and McTaggart were out of fashion. Surveying the scene in 1933, the young Braithwaite, influenced by Russell, Moore, and Wittgenstein, held that the new style of analytic philosophy showed the futility of supposing that it is possible to construct a deductive metaphysical system to which experience must conform. Hence, he wrote, 'we can be certain beforehand that a system professing to derive by logically necessary implications from logically necessary premisses interesting empirical propositions is wrong somewhere. We in Cambridge have been fortunate in having The Nature of Existence of J. E. McTaggart as an awful example.'s Similarly, in Oxford, as Ryle later 46
5 R. B. Braithwaite, 'Philosophy', in H. Wright (ed.), University Studies (Cambridge: Nicholson and Watson, 1933),23.
Strawson:Js Rehabilitation of Metaphysics
47
wrote, 'In the 1930S the Vienna Circle made a big impact on my generation and the next generation of philosophers. Most of us took fairly untragically its demolition of Metaphysics. After all we never met anyone engaged in comn1itting any metaphysics; our copies of Appearance and Reality were dusty; and most of us had never seen a copy of Sein und Zeit.'6 Collingwood accepted the positivists' contention that metaphysical utterances, such as 'Every event has a cause' or 'There is a God', are neither true nor false. In An Essay on Metaphysics, published in 1940, he argued that such assertions are indeed not genuine propositions, but expressions of the absolute presuppositions of the thought of a given historical epoch. On his view, a proposition is essentially an answer to a question, and every question rests on a presupposition from which it arises. Presuppositions may be relative or absolute. Relative presuppositions are both presuppositions of one question and answers to another. Absolute presuppositions are presuppositions of questions but never answers to one. So they do not express propositions, can be neither verifiable nor falsifiable, and are neither true nor false. The study of n1etaphysics, however, is not the argumentative defence of absolute presuppositions. It is a historical discipline that articulates the absolute presuppositions of the science or history of an epoch, and it results in true or false historical statements, which are indeed metaphysical propositions, characterizing the absolute presuppositions of, for example, Greek science or history, Newtonian mechanics or modern physics. The idea of identifying something that can be called 'the presuppositions of thought' and that can be said to characterize the science or history of an epoch is appealing to the historian of ideas. But Collingwood's conception's of a proposition and hence too of truth and falsehood were idiosyncratic, open to numerous objections, and won no adherents. His conception of absolute presuppositions, which are not propositions, are not supportable by argument, and which lack truth-values, provided no explanation of the fact that traditional metaphysicians supported their assertions by extensive argument, and endeavoured to prove the falsity of their denials. His account of metaphysics does not correctly represent either the intentions or the productions of traditional metaphysicians, who conceived of their claims as true or false descriptions of how things are, and not of how things are transiently presupposed to be. A different defence of metaphysics, which likewise accepted the positivist claim that .metaphysical utterances are neither true nor false, but denied that they are nonsense, was given by H. H. Price. In his presidential address
6 G. Ryle, 'Autobiographical', in O. P. Wood and G. Pitcher (eds.), Ryle: A Collection of Critical Essays (New York: Doubleday, 1970), 10. Heidegger's remarks about 'the Nothing' in Sein und Zeit had been a major target of Carnap's criticism of metaphysical nonsense in 'The Elimination of Metaphysics through the Logical Analysis of Language'.
48
Hacker
to the Mind Association in 1945, in which he surveyed the current state of philosophy in Britain, he argued against the trend of analytic philosophy of the previous decade, insisting that 'clarity is not enough'. Under the influence of Wittgenstein and of the logical positivists, analytic philosophers had argued that clarification is the fundamental task of philosophy, that philosophy can yield no new knowledge but only make clear what we already know. The sole task of philosophy is analysis of the statements of science, history, common sense, and ethics-the clarification of the meanings of statements that generate philosophical puzzlement. A corollary of the analytic task, it was argued, is the therapeutic task of eliminating perplexity and conceptual confusion and of disclosing the latent nonsense in much philosophical thought. Price defended the new-style philosophy against many of the accusations that were being directed at it. He was willing to concede that analysis, in some generous sense of the term-i.e. conceptual clarification-is the large part of philosophy of logic, epistemology, and even moral philosophy. But, he demurred, it had been a mistake to condemn metaphysics as nonsense and to consign the venerable subject to the rubbish heaps of intellectual history. Echoing Braithwaite, he agreed that it is illegitimate to argue to a conclusion concerning matters of fact if there are no matters of fact among one's premisses. But, he claimed, the only metaphysical argument that would be eliminated by this principle is the ontological argument for the existence of God. And it would be parochial to accept the verdict that all the great systematic speculative metaphysical systen1s of past philosophers are nothing but nonsense. How then should we view such systems? Price, like Ayer, rejected the Carnapian suggestion that they are akin to works of art and poetry, expressing attitudes towards the world. Rather, we should think of them as 'alternative modes of conceptual arrangement by which the body of empirical data is systematically ordered', like alternative maps of the same territory with different methods of projection. 7 In this sense they are no more true or false, right or wrong, than Mercator's projection is true or false, right or wrong. They are possible or not possible, adequate or inadequate to the task of representing what we wish to represent by their use, illuminating or unilluminating in highlighting relationships that are important to us for the purposes we have. Hence, the choice between different metaphysical systems is not between the true and the false but between the less good and the better, or between the good in such-and-such respects and the good in suchand-such different respects. The task of the speculative metaphysician is indeed not to add to our knowledge of facts. But it is not analysis either. It is
7 H. H. Price, 'Clarity is Not Enough', repro in H. D. Lewis (ed.), Clarity is Not Enough (London: Allen & Unwin, 19 63), 37.
Strawson's Rehabilitation of Metaphysics
49
rather 'to produce a unified conceptual scheme under which all known types of empirical fact may be systematically arranged'. 8 Price's plea for such metaphysics went unheeded. As far as I know, there was no debate over whether this conception of what he had called 'speculative metaphysics' was correct. 9
II. STRAWSON'S REHABILITATION OF METAPHYSICS
From 1945 until the end of the 195 as analytic philosophy evolved in Britain and elsewhere without any metaphysical pretensions, and, on the whole, without much attempt to aspire to the degree of generality characteristic of the ontological and metaphysical pronouncements of the great systembuilders of the past. In 1959, however, Strawson published his rightly renowned book Individuals. It operated at dizzying heights of generality hitherto unknown among post-war British analytic philosophers, and it professed unashamedly to be an exercise in metaphysics-it was, as its subtitle announced, 'an essay in descriptive metaphysics'. In the elegant but all-too-brief introduction to the book Strawson distinguished between revisionary and descriptive metaphysics. Descriptive metaphysics, he explained, 'is content to describe the actual structure of our thought about the world', while revisionary metaphysics 'is concerned to produce a better structure' .10 Descriptive metaphysics does not differ from conceptual analysis in intent. Like conceptual analysis Of, as Strawson was later to characterize it, 'connective analysis', it is concerned with describing and clarifying the concepts we employ in discourse about ourselves and about the world, and in elucidating their relationships-their forms of relative priority, dependency, and interdependency. Descriptive metaphysics differs from connective analysis in general primarily in that the concepts that it investigates are characteristically highly general, irreducible, basic, and, in a special sense, non-contingent. The generality of the concepts it studies is manifest in any list of concepts that have attracted the attention of metaphysiciansconcepts of material object, of property and relation, of causation, and of space and time. Some of these, such as material object or property, are quasi-technical-regimentations of concepts available in ordinary speech. Price, 'Clarity is Not Enough', 39. It is, however, interesting that Waismann, in 'How I see Philosophy', in H. D. Lewis (ed.), Contemporary British Philosophy, 3rd sere (London: Allen & Unwin, 1956), argued, primarily against Wittgenstein but also against the view of the Vienna Circle, that to say that metaphysics is nonsense is nonsense. For the great systems of philosophy of the past apprehend and articulate new ways of looking at the facts. Metaphysicians experience something akin to a shift in aspect perception, enjoy and advocate a new vision, a different way of conceiving of experience and its objects. 10 P. F. Strawson, Individuals (London: Methuen, 1959), 9. Subsequent page references in the text are to this volume. 8
9
50
Hacker
They are general in being categorial, or at least in subsuming numerous more specific concepts under them, in the sense in which the concept of a material object subsumes numerous material-object concepts such as the concept of a chair, of a lump of sugar, or of a mountain. 11 They are irreducible, not in being simple and unanalysable, but rather in not being eliminable without circularity in favour of other concepts that wholly define them. They are basic inasmuch as they are members of a set of general, pervasive, and irreducible concepts or concept types that together fortn .a structure that constitutes the framework of our ordinary thought. 12 And they are non-contingent in the sense that they are necessary constituents of any conception of experience that we can make intelligible to ourselvesessential to our conception of the experience of self-conscious beings. Descriptive metaphysics, Strawson declared, also differs from conceptual analysis in general in its method. Although examination of the use of words is the only sure way in philosophy (p. 9), the discriminations we can make and the connections we can thus establish are neither sufficiently general nor far-reaching enough to meet 'the full metaphysical demand for understanding'. For the uses of expressions of natural language 'are apt to assume, and not to expose, those general elen1ents of structure which the metaphysician wants revealed. The structure he seeks does not readily display itself on the surface of language, but lies submerged. He must abandon his only sure guide when that guide cannot take him as far as he wishes to go' (p. 10). Strawson took pains to distinguish the kind of investigation he was advocating from Collingwood's conception of metaphysics as a historical science. Descriptive metaphysics is not the description of the changing absolute presuppositions of thought. Nor is it an instrument of conceptual change. For 'there is a massive central core of human thinking which has no history-or none recorded in histories of thought; there are categories and concepts which, in their most fundamental character, change not at all' (ibid.). These form the indispensable core of the conceptual equipment of all human thought. Strawson took as his point of departure the correlative pair of linguistic functions of reference and predication. Picking out sonle individual item and saying or thinking something about it are fundamental functions of thought and speech. The categories of expressions that are employed in order to fulfil these basic functions are singular referring expressions-names, singular
11 Here it is evident that the term 'material object' is a quasi-technical one. One would not, I fancy, ordinarily say that a mountain is a material object. For, that term is ordinarily reserved for movable, middle-sized dry goods. Whether generalization of this notion is illuminating or merely the source of further unclarity is debatable. Certainly the boundaries of the technical term are even vaguer than of the ordinary, untechnical one. If a mountain is to be counted as a material object, what of a valley? Or a continent? Is a puddle a material object? A river or an ocean? A cloud? 12 P. F. Strawson, Analysis and Metaphysics (Oxford: Oxford University Press, 1992), 24.
Strawson~s
Rehabilitation of Metaphysics
5I
definite descriptions, or indexicals-and predicates. The question he addressed is what are the most general conditions for identifying reference to and reidentification of particulars, and, in the light of the requirements of identifying reference and reidentification, what are and must be the most fundamental, primitive, or basic objects of reference and subjects of predication. He argued that the basic particulars in any conceptual scheme capable of describing experience and its objects are necessarily of two general types of individuals of a relatively substantial and enduring sort located in a unified spatio-temporal framework-material objects and persons. Material objects ~re the subjects of a class of predicates that he denominated 'M-predicates', and persons are the subjects both of M-predicates and 'P-predicates'-that is, predicates that presuppose their subject's possession of states of consciousness. Given that these are the basic particulars of any such conceptual scheme, he proceeded to investigate various forms of non-basic or dependent particulars, identifying reference to which depends upon the more fundamental reference to and identification of independent, basic particulars. An10ng these are such items as individual experiences, which he claimed to be identifiabilitydependent upon the subject whose experiences they are, and individual events, which are identifiability-dependent upon the particular undergoing change. Furthermore, although the basic particulars are and must be material objects and persons, universals, abstract and intensional objects can likewise be objects of reference and subjects of predication. With a characteristically relaxed attitude to ontological commitments, Strawson pleaded that we should abandon whatever natural but ill-founded nominalist qualms we might suffer from and recognize the existence of such objects of thought. Descriptive metaphysics is to be contrasted with revisionary metaphysics. Revisionary metaphysics is concerned not with describing the actual structure of our thought about the world, but with producing a better structure. The productions of revisionary metaphysics, at its best, are 'both intrinsically admirable and of enduring philosophical utility' (p. 9). But they are so only in so far as revisionary metaphysics is 'at the service of descriptive metaphysics' (ibid.). Strawson conceded in Individuals that perhaps no philosopher has ever been, both in intention and in effect, wholly a descriptive or wholly a revisionary n1etaphysician. But one can allocate Descartes, Leibniz, and Berkeley to the class of revisionists, and Aristotle and Kant to the descriptivists. In Individuals Strawson did not elaborate further on what he meant by 'revisionary metaphysics', and did not explain in what sense revisionary metaphysics is 'at the service of descriptive metaphysics' or what enduring philosophical utility it has. But he elaborated somewhat further in his paper 'Analysis, Science and Metaphysics' .13 Some metaphysics, 13 P. F. Strawson, 'Analysis, Science and Metaphysics', repro in R. Rorty (ed.), The Linguistic Turn (Chicago, Ill., and London: The University of Chicago Press, 1967).
52 Hacker he there wrote cautiously (p. 318), is 'best, or most charitably' seen as consisting in imagining how things might be viewed through the medium of a different conceptual scheme. And he conceded that, even when such metaphysics can be so interpreted, it is not presented thus, but rather as a picture of how things really are, as opposed to how they delusively seem to us to be. This presentation, Strawson admitted, with its contrast between esoteric reality and our daily delusion, involves and is a consequence of the unconscious distortion of ordinary concepts-that is, of the ordinary use of linguistic expressions. So revisionary metaphysics, 'though it can sometimes be charitably interpreted [thus], in fact always involves paradox and perplexities ... and sometimes involves no rudin1entary vision, but merely rudimentary mistakes'.
III. IS DESCRIPTIVE METAPHYSICS A FORM OF METAPHYSICS?
It is clear enough, and should have been evident at the time, that the positivists' characterization of what are to be deemed 'metaphysical concepts' was unsatisfactory. If it is deemed a requirement of any genuine concept that it have determinate criteria of application, and if n1etaphysical concepts are characterized exclusively by reference to the fact that they lack such criteria, then it is hardly an astounding discovery that they are pseudo-concepts. Strawson's proposal that we should view as metaphysical only such concepts as are, in the manner described, very general, basic, irreducible, and non-contingent gives us at least a rough-and-ready, although vague, criterion. Moreover, it is a criterion the application of which produces a set of concepts or types of concept that partially coincides with the preoccupations of metaphysics as traditionally conceived. There can be no licit positivist objection to the task of providing connective analyses of such concepts and of such categorial concepts as space, time, material object, property, relation, cause, person, and so forth. Nor indeed can there be any objection to delineating the complex forms of dependency and interdependency an10ng these concepts or types of concept, which conjunctively form the framework of our thought about the world and ourselves in it. It is not wholly clear whether very general concepts of ethics and aesthetics, such as moral goodness and beauty, or very general concepts of the philosophy of action, such as act and omission, ability, intention, reason for action, and purpose, should be counted as metaphysical or not. Perhaps they should count as concepts of the metaphysics of morals, of aesthetics, and of agency. Or perhaps they should be excluded, on the ground that only concepts presupposed by any langua e in which
Strawson~s
Rehabilitation of Metaphysics
53
a distinction can be drawn between experience and its objects are to be deemed metaphysical. 14 Either way, descriptive metaphysics-in so far as it is concerned merely with the analysis of such concepts-is just more conceptual analysis, albeit at a very high level of generality. What might make it distinctive in a further and deeper sense would be the arguments that purport to show that these concepts are, in some deep sense, necessary concepts. If, in addition, such arguments were to result in propositions that purport to describe necessary features of reality, descriptive metaphysics would make contact with one salient feature of traditional metaphysics. For, it was the hallmark of traditional metaphysics that, like physics, it investigated the nature of reality, but, unlike physics, it did not investigate the contingent features of the world but its allegedly necessary features. Physics tells us how things are in reality, what contingent laws of nature describe the behaviour of things in space and time. Metaphysics aspires to tell us how things must (unconditionally) be. Metaphysical truths, if there are any, will be non-logical, necessary truths. So the principle that every event has a cause was held to be a truth of metaphysics, for it is allegedly inconceivable that there be uncaused events in nature. The principle that every attribute inheres in some substance was a truth of metaphysics inasmuch as it was held to be unthinkable that there be attributes that are not essential or accidental properties of some thing. And, descending to a much lower level of generality, it was held to be a metaphysical truth that nothing is simultaneously red and green all over, for it is unimaginable that there be any such thing. Propositions of metaphysics, so conceived, are quite different from the propositions of physics. They must be truths that can be discovered by the use of pure reason alone, independently of experience. But they are neither logical or analytic truths, nor empirical truths. Kant famously characterized some of them as a subclass of synthetic a priori propositions. Are the concepts to be analysed by the descriptive metaphysician necessary concepts in any deep sense? Perhaps not in the sense in which seventeenthcentury metaphysicians thought that we are born equipped with either actually or virtually innate ideas, or in the sense in which Kant thought of the categories as a priori concepts, unifying a manifold of intuition in judgement to yield experience. The general, basic, and irreducible concepts that concern the descriptive metaphysician are necessary in the sense that they are an essential structural element in any conception of experience that we can render intelligible to ourselves. Since the salient concepts that Strawson
14 On the other hand, if concepts of experience, and with them the concept of a person or subject of experience, presuppose concepts of active agency, as they arguably do, then such concepts or categories as act and omission, ability, voluntariness and intention, reason for action, and purpose must presumably count as metaphysical.
54 Fiacker discusses are formal or categorial concepts, such as substance, property, cause, experience, etc., it seems evident that it is not necessary that any language in which we can distinguish between experience and its objects Inust contain these formal concepts-rather, it is necessary that any such language contain concepts that belong to these types, i.e. substance concepts, concepts of properties of substances, causal concepts, etc. Moreover, the necessity for such concepts is relative-relative to the possibility, in such a language, of distinguishing in general between experience and its objects. It seems evident that we can entertain the idea of a language without a word equivalent to 'substance'. But, more importantly, we can surely also entertain the idea of a language without substance names. For there could be a language consisting only of orders 'Come!', 'Go!', 'Eat!', 'Drink!', etc., together with words expressing willingness or unwillingness, approval and disapproval. It would be an impoverished affair, to be sure, but I doubt whether we would have qualms in characterizing it as a rudimentary language. And it is at least arguable that we can imagine a language without any causal expressions, which consists only of substance names, and colour and number words-as Wittgenstein suggested in the evolved second language-game in the Investigations. The necessity of the select concepts or concept types that are of concern to the descriptive metaphysician is relative to the concepts of experience and its objects, or, more precisely, relative to any language and thought in which a distinction can be drawn between experience and what it is experience of What that means is that such concepts are partly constitutive of what we call 'experience' and 'objects of experience'. We would not denominate anything 'an experience' unless it could be ascribed to a sentient being, and would not characterize anything as 'conceptualized experience' (i.e. 'experience' in the weighty sense that is Strawson's concern) unless it were ascribable to a person. And we cannot intelligibly characterize anything as a person unless it is a space-occupying sentient being tracing an autobiographical route through a unified spatio-temporal world consisting of relatively enduring material objects. In short, the weighty concept of experience is located at the centre of a vast web of concepts of space and time, of substance and causation, of cognition and volition. The investigation of these conceptual involvements and commitments is of great philosophical interest, in part because so n1uch philosophical reflection failed lamentably to grasp precisely how weighty and how extensive are these involvements and commitments, and how they are interwoven. But it is not obvious that the elucidation of this web of connections, dependencies, and interdependencies is anything more than connective analysis of a select range of general concepts partly constitutive of what we call 'a description of an objective world'. It may indeed be called 'metaphysics'-but only in this attenuated sense. Are the propositions of descriptive metaphysics as presented by Strawson a kind of metaphysical proposition as traditionally understood? One difference
Strawson's Rehabilitation of Metaphysics
55
is evident. Where traditional metaphysicians conceived of themselves as limning the ultimate structure of the world, the descriptive metaphysician will conceive of himself as sketching the basic structure of our conceptual scheme-of the language we use to describe the world and our experience of it. Or, more ambitiously, of delineating the structure of any conceptual scheme which can be employed to describe a world and a subject's experience of it. Hence descriptive metaphysics does not aim to describe the necessary, superphysical, structure of reality-about which it may well remain altogether sceptical. Rather, as Strawson writes, it aspires 'to establish the connections between the major structural features or elements of our conceptual schemeto exhibit it, not as a rigorous deductive system, but as a coherent whole whose parts are mutually supportive and mutually dependent, interlocking in an intelligible way' .15 So conceived, descriptive metaphysics breaks with the metaphysical tradition, which purported to give us insights into the necessary structure of reality. So far so-soberly-good. The conception of a form of necessity that is not logical, but no less adamantine than logical necessity, that is an objective, language-independent form of necessity that can nevertheless be apprehended a priori by reason alone is, surely rightly, dismissed as a fiction. 16 Nevertheless, descriptive metaphysics results in an array of propositions that are held to be necessary truths constitutive of our conceptual scheme. So, for example, it is argued that it is a conceptual truth that places are defined by the relations of material bodies, that material bodies provide the framework for spatial location in general, that they are basic from the point of view of referential identification and reidentification of all other particulars of different categories, that persons have bodies, that the experiences of a person are identifiability-dependent on the identity of the person whose experiences they are, that a condition for the intelligibility of self-ascription of experience is the legitimacy of other-ascription of experience on the basis of logically adequate behavioural criteria, and so on. Again, the interest of such propositions is not in dispute. They endeavour to articulate fundanlental structural features of our conceptual scheme and arguably of any conceptual scheme in which the distinction between experience and its objects can be drawn. But the moot question is: what is the status of such propositions? Obviously, they are not and are not intended to be empirical propositions. They purport to be necessary truths. They are not obviously analytic propositions, assuming that we have a tolerably clear grasp of that problematic category. Are they then synthetic a priori?
Strawson, Skepticism and Naturalism: Some Varieties (London: Methuen, 1985), 23. To be sure, in the wake of Kripke and Putnam, there has been a revival of conceptions of essentialism and de re necessity. This regression cannot be discussed here. 15
16
Hacker Strawson did not address the question in Individuals. But in his later book The Bounds of Sense he repudiated the idea of the synthetic a priori. He argued that 'Kant really has no clear and general conception of the synthetic a priori at all' .17 For, once the apparatus of transcendental idealism is abandoned, as Strawson argued it should be, the category of synthetic a priori propositions is, he claimed, merely a residuum of propositions that are neither analytic nor empirical. The kinds of proposition that Kant deemed to be metaphysical are such propositions as: experience essentially exhibits temporal succession; there must be such unity among the menlbers of a diachronic series of experiences as is required for the possibility of selfconsciousness or self-ascription of experience; experience must include awareness of objects that are distinguishable fronl the experiences of them; there must be one unified spatio-temporal framework embracing all experience and its objects; and the Principles of the Analogies of Experience, i.e. of the permanence of substance, of causality, and of causal reciprocity. Strawson saw no explanatory value in characterizing such propositions as 'synthetic a priori'. What then are they? And what is the nature of the apparent necessity that is associated with them? At this point, it seems to me, Strawson's account falters. The Kantian propositions do, he concedes, 'have a distinctive character or status'. Nevertheless, he sees 'no reason why any high doctrine at all should be necessary here'. The conceptual scheme we employ reflects our nature, our needs, and our situation. It is subject to change as our needs and situation change. But such changes are conceivable only as variations within a fundamental general framework of ideas. With regard to some of the propositions of descriptive metaphysics that he elicits in Individuals, he observes that 'it does not seem to be a contingent matter about empirical reality that it forms a single spatia-temporal system'. For, if someone told of a kind of thing and of events that occurred to it, but insisted that that object was not located at any distance fronl here, and that those events stood in no temporal relation to now, since they did not belong to our spatio-temporal system, we should take him to be saying that the events had not really occurred and that the thing in question did not really exist. In so saying, we show how we operate with the concept of reality. 'We are dealing here', he concludes, 'with something that conditions our whole way of talking and thinking, and it is for this reason that we feel it to be non-contingent' (p. 29). This does not seem to me to be altogether satisfactory. The fact that something conditions our whole way of talking does not obviously suffice to explain why we should think of it as non-contingent. Our size conditions at least much of our way of talking and thinking too, but there is nothing 56
17
P. F. Strawson, The Bounds of Sense (London: Methuen, 1966), 43.
Strawson~s
Rehabilitation of Metaphysics
57
non-contingent about it. The fact that we are sexual beings with a determinate gender conditions our social existence and relations and our thought about it, but it is easy enough to imagine the very different lives of asexual beings otherwise akin to us. Although we may feel certain features of the world to be non-contingent, such as the pervasiveness in it of causal regularities, the relative permanence in it of three-dimensional material objects, it does not follow from our so feeling that there is anything non-contingent about such features. On the other hand, non-contingency is surely rightly associated with the thoughts that any event is either earlier, later, or simultaneous with any other event, that every material object is spatio-temporally . related to every other material 0 bject, that every experience is the experience of some sentient being, and that subjects of experience have bodies. Even if no 'high doctrine' is necessary here, some modestly low doctrine is surely needed to satisfy the requirements not of metaphysical understanding but of philosophical understanding. For, if the necessity of such propositions is not merely a misguided projection of our feelings of necessity (as Hume thought that our ascription of necessary connection to any causal relation was), then we crave some explanation of the nature of such non-logical, yet nonempirical, necessity that is not evidently analytic. 'In order to set limits to coherent thinking', Strawson concluded his discussion of Kant's synthetic a priori propositions, 'it is not necessary, as Kant, in spite of his disclaimers, attempted to do, to think both sides of those limits. It is enough to think up to them.'18 This nicely echoes Wittgenstein's similar observations in the preface to the Tractatus. But it is noteworthy that the propositions of the Tractatus were themselves ultimately condemned as nonsense, since they did not meet the criteria of significance of that book. That was, as Wittgenstein later realized, an unhappy solution to the problem of the status of propositions delineating the bounds of sense. It is evident that Strawson would not wish to venture down that cul-de-sac. But then some explanation of their status is required. The thought that any event must be earlier, later than, or simultaneous with any other event does not look like a statement about our conceptual scheme. Nor does the proposition that space and time form a unity. They look like insights into the essential nature of reality. So, do such propositions vindicate the project of traditional metaphysics? Or are appearances misleading? Surely n10re needs to be said. The way out of the quandary was signposted by Wittgenstein in his later reflections on the same kinds of proposition as those he had condemned as nonsensical attempts to describe the bounds of sense in the Tractatus. We should treat such propositions, which overtly or covertly describe the
18
Ibid. 44.
Hacker conceptual connections between the major structural features of our conceptual scheme, as expressions of norms of representation. The proposition that every event is spatio-temporally related to every other event is in effect a rule for the use of the word 'event'. And we manifest our adherence to this rule precisely in our agreeing with Strawson's verdict that a description of an object and of the changes that it underwent, which were not spatioten1porally related to whatever is here and now, would and should be taken to be a description of a non-existent object and a non-occurrent event. The proposition that every event has a cause, if taken to be a metaphysical proposition rather than an empirical hypothesis, and hence licensing us to claim of any observed event that it must have a cause even if none can be discerned, is in effect a norm of representation which entitles us to infer from any event-identification that there is a cause of that event, which may or may not be known (and mayor may not be discovered). The proposition that every experience is necessarily someone's experience, that every thought is necessarily someone's thought, is an expression of one's commitment to a norm of representation, namely that, pace Hume and James,19 no sense is attached to such forms of words as 'There is a pain in the room, although no one has it' or 'There is a thought in the quad, although no one is thinking it'. The necessity we associate with such propositions is fully explained by their normative status. For they are what Wittgenstein, a little provocatively, called 'grammatical propositions'-expressions of rules of representation in the misleading guise of staten1ents about reality. Any putative reference to the occurrence of a pain that denied its ascribability to some sentient being is not to be counted as a genuine reference to the occurrence of a pain. For, in so far as we accept these propositions we rule out, a priori, any use for the relevant terms, e.g. 'uncaused event', 'subjectless pain', as we rule out any use for the term 'checkn1ate' in draughts-there is no such thing in this game. It is important to realize that in ruling out the intelligibility of uncaused events, of events spatio-temporally unrelated to other events, of actual experiences unascribable to subjects of experience, of colour coinstantiation, we are not delimiting nature. We are not laying down possibilities that nature cannot realize. We are laying down the limits of description, not describing possibilities that are impossible. We are characterizing the bounds of sense. For we have given no use to the forms of words 'simultaneously red and green all over', 'an occurrence of a pain that no one felt', 58
19 Hume thought it made sense for perceptions or ideas to exist unperceived, although there are empirical reasons for thinking that as a matter of fact they don't (see A Treatise on Human Nature, bk. I, pt. IV, sect. II). Similarly, James remarks that 'Whether anywhere in the room there be a mere thought, which is nobody's thought, we have no means of ascertaining, for we have no experience of its like' (The Principles of Psychology, i (New York: Dover, 1950),226).
Strawson>s Rehabilitation of Metaphysics
59
'an event that was neither earlier than, later than, nor simultaneous with such-and-such present event'. We are not expressing insights of pure reason into what nature can or cannot do, but reminding ourselves that we attach no significance to such expressions. To be sure, we could attach significance to them-after all, nothing is stopping us. All we have to do is lay down additional rules for the use of the relevant terms. But if we do, we are thereby changing their meanings and speaking of something different. It was a confusion of traditional metaphysics to project insights into the structure of our conceptual scheme on to the objects described by our employment of it. In so doing, it confused rules determining the correct use of words and the licit inferences that their application licenses, which conjunctively define the essences of things, with objective, language-independent necessities and necessary connections in the world. The putative necessities in reality are merely the shadows cast by rules for the use of words in our language that are partly constitutive of their meanings. The question of whether the propositions in question are analytic or synthetic (with all the unclarities associated with these categories) can be sidestepped. For the status of such propositions is clarified in recognizing that they are rules of representation, not descriptions of reality. They can be said to be true only in the sense in which it is true that the chess king n10ves one square at a time. The possibility of knowledge of such propositions is relatively unproblematic, since it is knowledge of, or recognition of, the rules we follow in using the relevant words of our language correctly. Why then is it not trivially easy to attain? Largely because of the generality at which we operate in this domain and because of the ramifications and interrelations of the rules, which are anything but easy to survey. Hence to the extent to which we wish to speak of attaining knowledge here (as opposed to attaining understanding), it takes the form of realization rather than discovery. Metaphysics thus construed yields no insight into reality, but only into our forms of description of reality. So it is just more grammar, in Wittgenstein's extended sense of the term. The generality at which we operate in the domain of Strawsonian descriptive metaphysics is manifest not only in the generality of the concepts and concept types that are the focal point of the investigation but also in the fact that they constitute, as he surely rightly points out, structural elements of our conceptual scheme. Precisely because such concepts as space and time, substance and property, cause and effect are, roughly speaking, categorial, their rule-governed connectedness ramifies throughout our conceptual scheme. The formal or categorial concepts of substance and attribute, or of cause and effect, subsume thousands of material concepts that are in constant employment in our daily discourse. And their forms of connectedness determine our thinking and inferring in all our description, reflection, and action. Small wonder that at least some of these categories
IIacker
60
seem non-contingent; for without thenl we would not engage in the thought and intentional action of the kind that characterizes our form of life. We could no more decide to abandon these categories of concepts than we could decide to cease to be human beings. What then of ontology, which Strawson, in a monlent of exuberance, dignifies by the name of 'the general theory of being'?2o It seems to me that ontology is no more than an investigation of what is meant by saying of itenlS belonging to a certain general category that they exist, obtain, occur, or go on. We say that material things come into existence and pass away; we refer, unashamedly, to general characteristics of things or to the occurrence of events and the obtaining of states of affairs; some philosophers insist-perfectly reasonably under some interpretation-that there are objective values; and we assert without qualms that there are so-and-so many primes between m and n. Ontology, if it is anything, is surely the elucidation of what is meant, from case to case, by such existence claims. Its task is not to draw up inventories of the contents of the universe, nor can it be illuminatingly described as the study of being qua being. So it is not so nluch a general theory of being, but rather a matter of connective analysis, concerned in particular with existence claims and with the elucidation of conceptual dependencies involved in such claims. It might be thought, although I do not think that Strawson ever suggested as much, that Strawson's descriptive metaphysics yields transcendental arguments that prove the existence of the external world or of other minds. But that is, I think, mistaken. It would be absurd to argue fronl conceptual connections in thought to existential truths about the world, or, in Wittgensteinian idiom, fronl gramnlatical propositions to empirical ones. Strawson argued that a condition for the intelligibility of criterionless selfascription of experience is the adequacy of the behavioural grounds for other-ascription of experience. This conceptual connection does not prove that there are other experience-enjoying beings-what it proves is the incoherence of scepticism about other minds that adnlits self-ascription of experience and simultaneously denies the adequacy of the criteria for other-ascription. And it can be argued, although I shall not attempt to argue it here, both that the demand for a proof of the existence of the 'external world' or of other minds is itself incoherent and that we have a vast hoard of genuine knowledge about objects in the world around us and about our fellow human beings and the experiences they enjoy, without per impossibile possessing any such proof. It should be noted that, although some of Strawson's conclusions echo Kantian synthetic a priori propositions, e.g. concerning causal regularities
20
Strawson, Analysis and Metaphysics, 35.
Strawson~s
Rehabilitation of Metaphysics
61
and the existence of substance, these are not transcendental deductions of the necessity that every event have a cause or of the existence of an external world. They are rather observations upon the background conditions for the exercise of such concepts that enable a subject of experience to ascribe experience to hin1self. For, Strawson argues persuasively, the concept of subjective experience only gets a grip to the extent that the concepts of independently existing objects of experience get a grip, and both obtain a purchase only to the extent that such objects of experience are generally connected by causal regularities. But this is no proof of the existence of an external world or of the principle of sufficient reason. Scepticism is to be refuted by showing that it makes no sense-not by producing a proof of the existence of objects. We do indeed know of the existence of multitudinous objects around us, as we know of innumerable causal connections between substances and the events they make happen-but not on the grounds that the existence of substances and of widespread causal regularity is a condition for the employment of concepts of experience and its objects, nor on the grounds that we do enjoy subjective experiences. Does descriptive metaphysics differ in its method from connective analysis in general? The examination of the use of words, Strawson averred, is the only sure way in philosophy, but the structures that the descriptive n1etaphysician wishes to reveal are not displayed on the surface of language and the connections he wishes to establish are too far-reaching to be discernible by scrutiny of the use of words. So he must abandon his only sure guide when that guide cannot take him to the peaks of abstraction that he aims to scale. What method should he then use? Strawson offers us disappointingly little-'I know of no procedure or recipe for getting at the answers except to think about those ideas and questions as hard as you can' .21 It is true that there is a sense in which ordinary usage offers few hints or clues to philosophical insight when it comes to such concepts as space and time, substance and accident, subject and object of experience. The terms do not offer that variegated field of subtle distinctions that is to be found, as Austin noted, in such domains of discourse as excuses. Nevertheless, it seems to me that the connections Strawson seeks-and finds-are not in any sense submerged beneath the surface of ordinary usage. The use of the term of art 'substance concept' is not likely to offer the philosopher much help. The use of the term 'cause' may be positively misleading, since 'cause' and 'reason' are, over a range of contexts, interchangeable, while the insightful philosopher interested in causation will wish to differentiate causes from reasons-and will indeed find ample
21 P. F. Strawson, 'My Philosophy', in P. K. Sen and R. R. Verma (eds.), The Philosophy of P. F. Strawson (New Delhi: Indian Council of Philosophical Research, 1995), 17.
62 Iiacker reason to do so fronl a more careful examination of the use of these two terms. The clues the descriptive metaphysician seeks, and the only tribunal before which his claims can be adjudicated, are the general pattern or patterns of use of the multitudinous substance concepts and causal concepts that abound in natural language. Those patterns are in full view, even if it takes uncommon skill to discern them. Strawson, it therefore seems to me, saved the letter of traditional metaphysics, but abandoned its spirit. Descriptive metaphysics is distinctive, and unlike other philosophical endeavours, in so far as it strives to disclose the most general forms of connectedness that permeate our conceptual scheme and to reveal the conceptual involvements of the most general kinds of speech functions that characterize our use of language-indeed, not only of our language and our conceptual scheme, but of any language and any conceptual scheme in which certain kinds of distinction are drawn and certain fundamental kinds of speech acts performed. But it is also like other philosophical endeavours within the field of connective analysis and unlike the aspirations of traditional nletaphysics. It yields no knowledge of reality, let alone insight into the necessary structure of reality-but only insight into the forms and structures of our thought about reality. It might indeed be said to be the legitimate heir to what used to be conceived of as metaphysics, but a dethroned heir, deprived of the ancestral crown and orb. It is metaphysics without its nimbus.
IV. IS THERE ANY SUCH THING AS REVISIONARY METAPHYSICS?
Price pleaded for the preservation of metaphysics understood as 'alternative modes of conceptual arrangement', and recommended that philosophers should continue to be engaged upon devising different unified conceptual schemes. And he suggested that the works of the great metaphysicians of the past should be viewed as directed at such a goal. This seems akin· to Strawson's conception of revisionary metaphysics in Individuals. The first question to address is: was this the project of the great system-building philosophers of the past? Were Descartes, Leibniz, and Berkeley-to mention only those whom Strawson characterized as revisionary metaphysicianstrying to construct a different language, with a fundamentally different grammar, from the natural languages of mankind? I can see no trace of such an intent in their works. Descartes, for example, was not recommending that we adopt a new form of language. He was trying to describe the world and the fundamental kinds of substance that exist in it, to characterize their essential natures and their modes of interaction. Leibniz's monadology did not advocate a change in notation, but elaborated a putative insight into the
Strawson~s
Rehabilitation of Metaphysics
63
constitution of reality. Berkeley was not recommending that we adopt a language without names of material substances, which would merely be better for certain needs we have than our existing language. He too was trying to specify the nature of the vvorld and of what exists in it. And he thought that any talk of material substance, understood as he took Locke to understand the term, was not an inoffensive part of our ordinary conceptual scheme, but incoherent nonsense lying at the heart of Locke's mistaken metaphysics of the scientific revolution of the seventeenth century. Indeed, the putative proof that it was nonsense, and not just 'an alternative conceptual scheme', was a crucial step in Berkeley's endeavour to confound sceptics and atheists. If that was not what traditional metaphysicians were trying to do, was that in effect what they actually did? Did Descartes or Berkeley or Leibniz construct a language, different from our natural languages but equally fitted for describing the world and our experience of it? Surely not. Cartesian, Leibnizian, or Berkeleian metaphysics does not provide a novel grammar, constitutive of a novel conceptual scheme. Rather, these metaphysical systems, thus interpreted, would result in an incoherent grammar that at crucial points specifies kinds of particular without any associated criteria of identity, introduces psychological predicates that would only have a sense if they were, per impossibile, definable by private ostensive definition, and retains many of our concepts while simultaneously severing them from the web of connections to other concepts that alone makes them intelligible. Of course, it does not follow that the attempt to reconstruct these grand systems in this fashion may not be instructive. If one cannot reasonably interpret the metaphysical systems of the past as recomn1endations to adopt a new conceptual scheme, can it be argued, as Price suggested and Strawson, in Individuals, intimated, that devising alternative conceptual schemes is a task that metaphysics might reasonably undertake? Is there any sense in which such an endeavour might, as Strawson suggested, be at the service of descriptive n1etaphysics? It is noteworthy that the only philosopher who comes to n1ind in association with the idea of a philosophical programme of constructing alternative languages that could be deemed to constitute alternative conceptual schemes was the most fervent anti-metaphysical crusader of the twentieth century, namely Rudolf Carnap. He would rightly have been surprised to learn that doing so was a form of metaphysics. 22 It is difficult to see the point of such endeavours within philosophy, unless they are aimed at constructing fragments of a conceptual scheme, which can then be held up as a useful
22 It is also noteworthy that his idea that we have a choice between a sense-datum language and a n1aterial-object language is wholly incoherent. His endeavour to construct a sense-datum language necessarily fails inasmuch as the concept of a sense-datum is parasitic on our general concepts of objects of which the sense-data are data.
Ilacker object of comparison highlighting features, uses, and forms of contextual dependencies of the corresponding fragment of our own. It is only in this sense that something akin to revisionary metaphysics could be thought to be part of the task of philosophy and indeed at the service of descriptive metaphysics, i.e. general connective analysis. But part of the usefulness of revisionary metaphysics, thus conceived, runs counter to the Strawsonian vision, at least to a degree. For part of the point of devising an alternative gran1mar for, say, colour description (e.g. adoption of colour adverbs rather than adjectives), or ascription of experience (elimination of the first-person pronoun, as in 'There is pain' instead of 'I have a pain'), is precisely to show that our concepts and their articulations in these domains are not the only possible ones. For we are prone to think that our modes of conceptualization are uniquely correct or true to the facts, or that our concepts are necessary ones inasmuch as they uniquely match the logicometaphysical forms of reality. But over a wide range of concepts we can envisage different grammars to fulfil analogous tasks. And we can readily imagine such changes in us or in the world as would render such-and-such concepts useless and such-and-such other novel concepts more useful for our purposes in the novel contexts envisaged. The necessity we imagine associated with our conceptual scheme is a necessity internal to our conceptual apparatus-not a form of objective, language-independent necessity. In this manner, the invention of fragments of a different conceptual scheme, which fulfil roles akin to fragn1ents of our own, can be useful in disabusing us of some of the illusions of traditional metaphysics that incline us to think that our conceptual scheme pays homage to the objective metaphysical nature of the world. Nevertheless, it does not follow that, for us, with the kinds of conceptually moulded interests and purposes we have, there are serious alternatives to those major structural features of our conceptual schen1e that lie at the heart of Strawson's investigations. For, they constitute, as he says, 'the massive central core of human thinking' (p. 10). It is not merely that they are 'the indispensable core of the conceptual equipment' (ibid.) we deploy. Rather, they are partly constitutive of our nature as selfconscious human beings, involving concepts and categories that we could not abandon without ceasing to be human. Philosophy is in general a matter not of concept formation, but of concept description. Hence there is little role in it for the invention of (fragments of) conceptual schemes. Mathematics is concept formation, and the mathematician does indeed invent new forms of description that may be put to use by physicists in describing spatial relations and in the description and transformation of propositions about magnitudes, quantities, velocities, etc. and their relations-as Riemannian geometry proved fruitful for relativity theory, and as the calculus proved indispensable for Newtonian physics. However, the task of philosophy is not to devise alternative 64
Strawson~s
Rehabilitation of Metaphysics
65
conceptual schemes, but to describe and elucidate our own. Part of that task is to elucidate the most general forms of connectedness that permeate our conceptual scheme and that are partly constitutive of our very general conceptions of substance, causation, person and personal experience, space and time, reference and predication, and so on-which Strawson has pursued with his characteristic elegance, economy, and profundity. With regard to revisionary metaphysics, Strawson's second thoughts were, 1 think, more accurate than the view so briefly sketched in Individuals. Traditional metaphysics, though it can sometimes be charitably interpreted as involving a recon1mendation to adopt a new conceptual scheme, 'in fact always involves paradox and perplexities ... and sometimes involves no rudimentary vision, but merely rudimentary mistakes'. So there is not really any such subject as revisionary metaphysics-although scientists are free to devise fragments of alternative conceptual schemes for their purposes, if the new scheme is more fruitful in the generation of explanatory and predictive theories than the existing one. Nevertheless, there is a connection between the idea of devising a fragn1ent of a novel conceptual schen1e and the products of the metaphysical tradition. For traditional metaphysics, e.g. representational realism, idealism, or solipsism, presents its doctrines as if they were correct descriptions of reality, as if it were truer to the facts to say 'There is pain' rather than '1 am in pain', since the self is not a constituent of the experience of pain, or to say 'Grass looks green (presents an idea or representation of green) to normal observers in normal conditions' rather than 'Grass is green', since objects are not in themselves coloured, but only have a power to produce a representation of colour in human observers. But, as Wittgenstein pointed out, this is to suppose that a form of representation could say something false even when the proposition expressed says something true. The only way in which '1 am in pain' can be false is by '1 am not in pain' being true, and the only way in which 'Grass is green' can be false is by grass not being green but some other colour. So 'the one party attack the normal form of expression as if they were attacking a statement; the others defend it as if they were stating facts recognized by every reasonable human being'.23 In this sense, one might interpret some of the writings of metaphysicians as recommendations to adopt a different conceptual scheme. But then it is noteworthy that these metaphysicians fail to carry through the idea, and conflate elements of the new notation with elements of the existing one, and conclude that other people don't really have pain, or that the objects around us are not really coloured. 24
Wittgenstein, Philosophical Investigations, §4 02 . For more detailed discussion see P. M. S. Hacker, Wittgenstein~s Place in Twentieth-Century Analytic Philosophy (Oxford: Blackwell, 1996), 117-23. 23
24
66
Hacker
The conception of metaphysics characteristic of modern (post-Cartesian) philosophy, the idea of attaining a priori insights into the objective language-independent essences of things, was deeply rooted in the culture of seventeenth- and eighteenth-century Europe. The uprooting of this fiction required the labour of many thinkers, and took many decades. It is Strawson's signal achievement to have salvaged from the wreck of the traditional enterprise a form of general connective analysis, which he has called 'descriptive metaphysics'. What it aspires to is not knowledge of the essential nature of the world, but understanding of the general structure of our thought about the world. And that can be achieved. Since Strawson wrote Individuals, however, a new form of the old disease has broken out, and the mythology of metaphysics has been revived. Its roots lie deep in our contemporary culture-in the science and scientisrn of the late twentieth century. To eradicate it, will, I fear, be as difficult as it was to eradicate its more august ancestor. 25
25 I am grateful to Dr H.-J. Glock, Dr J. Hyman, and Sir Anthony Kenny for their comments on an earlier draft of this paper.
4 Kant's and Strawson's Descriptive Metaphysics GRAHAM BIRD
The project of descriptive metaphysics in Strawson's Individuals l is that of identifying fundamental features, such as the role of external objects and persons, in our ordinary experience, and it is natural to associate that project with the defensible part of Kant's epistemology identified in The Bounds of Sense. 2 Subsequently Strawson modified some of the views expressed in both early books, and I shall note some of those changes; but my main aim is to contrast Kant's position with those early accounts. Such an account of Kant contrasts with that of other commentators who, like Manfred Baum, regard Kant's first Critique as essentially ontological rather than epistemologicaP or who, like Michael Friedman, emphasize Kant's interest in the exact sciences rather than in ordinary experience. 4 It is not my intention here to defend Strawson's view against these alternative positions, for they need not exclude each other, but I accept that at least part of Kant's project is a descriptive metaphysics. My intention is, however, to claim that Kant's conception of such a descriptive metaphysics is not the same as Strawson's, and I identify three salient differences under the following headings: I. Relations to traditional scepticism (the appeal to transcendental arguments) II. The projects' methods (linguistic analysis and transcendental psychology) III. The nature of necessary, a priori, features of experience.
I shall try to show how these differences converge on a problem about understanding the modalities involved in descriptive metaphysics, but P. F. Strawson, Individuals (London: Methuen, r959). P. F. Strawson, The Bounds of Sense (London: Methuen, 1966). 3 Manfred Baum, Deduktion und Beweis in Kant's Transzendentalphilosophie (Konigstein: Athenaum Verlag, 1986) 43-4, 180-1. 4 Michael Friedman, Kant and the Exact Sciences (Cambridge, Mass.: Harvard University Press, 1992). 1
2
B~d
68
I want initially to make two related preliminary points about the comparison. The first is to note that I take the term 'descriptive metaphysics' seriously as a label for the project of accepting our experience as given without philosophical preconceptions in order to identify the fundan1ental principles that govern it. The second is to note the corollary that I am setting aside the claim that Kant's project is already invested with a commitment to traditional idealism. Henry Allison wrote a paper in 19695 that distinguished I(ant's project from Strawson's descriptive metaphysics, but he was primarily interested in correcting Strawson's account of Kant's commitment to traditional idealism. For Strawson, a major difference between the two projects was that while Kant's made such a commitment his own did not. Allison argued that the claimed commitment resulted from confusing transcendental idealism with what I(ant called empirical idealism, and I had also noted the same confusion, although I had not associated it with Strawson. 6 I therefore accept Allison's correction, and that is why a commitment to idealism does not appear among the three differences noted above between the two projects. But I know that there are still commentators who insist on ascribing a traditional idealist commitment to Kant, and some who do so by ascribing to Kant an empirical idealism that he plainly rejects. Since I assume, and take seriously, the philosophically non-committal starting-point of descriptive metaphysics, those commentators will have to read my argument as hypothetical. It will claim: or, better, even if, Kant's descriptive metaphysics is not traditionally idealist, nevertheless it still differs from Strawson's descriptive metaphysics in those three ways. 1. RELATIONS TO TRADITIONAL SCEPTICISM (THE APPEAL TO TRANSCENDENTAL ARGUMENTS)
In Individuals (p. 9) Strawson contrasted 'descriptive' and 'revisionary' metaphysics in the following way: Descriptive metaphysics is content to describe the actual structure of our thought about the world, revisionary metaphysics is concerned to produce a better structure.
That contrast is open to the objection that it provides a cross-classification of the relevant philosophical positions, and so distorts the relation of descriptive metaphysics to scepticism. For revisionary metaphysics presupposes a question of justification, raised in the philosophical context by the traditional sceptic. In opposing descriptive to revisionary metaphysics Strawson implied 5
Henry Allison, 'Transcendental Idealism and Descriptive Metaphysics', Kant-Studien, 60
(19 69), 216-33' 6 Graham Bird, Kant)s Theory of Knowledge (London: Routledge & Kegan Paul, 1962) chs. I and 3. I did not associate Strawson with the ascription of empirical idealism to Kant because The Bounds of Sense appeared only in 1966.
Kant's and Strawson's Descriptive Metaphysics
69
that the former arose within the same context of justification, and differed from the latter only in its conclusion that the actual structure of our thought is justified. Descriptive metaphysics is thus represented as already containing an answer to traditional scepticism. The subsequent discussion of scepticism in Individuals (pp. 32-6, 106-9) and the later debates with Stroud7 over scepticism and transcendental arguments reinforce that account. They have given rise to a familiar discussion 8 of Kant's transcendental arguments the principal ingredients of which are: (I) Kant's transcendental arguments are specifically designed, like Strawson's, to refute scepticism directly; (2) they attempt this, as Strawson does, by outlining principles necessary for thought or meaning, so that the sceptic's claims become meaningless or inexpressible; (3) they focus, like Strawson's, on a global Cartesian scepticism about external objects and persons; (4) they appeal, like StraV\Tson's explicitly or implicitly, to a semantic verificationism. Later I underline how irrelevant this account is to the picture of Kant I shall give. One consequence of Strawson's taxonomy is that this sceptical background conflicts with his original claim that 'descriptive metaphysics is content to describe the actual structure of our thought'. That divergence, between mere description and philosophical justification, will arise unless it is assun1ed, as it was in some versions of common-sense or ordinary language philosophy, that merely to describe the actual structure is to justify it. In this way common-sense or ordinary-language philosophy in the 1960s was similarly ambiguous; some of its practitioners took it to offer a refutation of traditional scepticism, but others simply used it to disregard such traditional issues. 9 Without that dubious assumption Strawson's taxonomy remains ambiguous between the two accounts shown in Fig. I. The opposition between descriptive and revisionary metaphysics corresponds to scheme A, but the account of descriptive metaphysics as merely describing, and not justifying, our thought corresponds to schen1e B.10 It is 7 B. Stroud, 'Transcendental Arguments', Journal of Philosophy, 65 (1968); B. Stroud, 'Kantian Argument, Conceptual Capacities, and Invulnerability' and P. F. Strawson, 'The Problem of Realism and the A Priori', both in P. Parrini (ed.), Kant and Contemporary Epistemology (Dordrecht: Kluwer, 1994). 8 Examples of this account of Kant's and Strawson's transcendental arguments can be found in Ross Harrison's and Ralph Walker's contributions in Schaper and Vossenkuhl (eds.), Reading Kant (Oxford: Blackwell, 1989). 9 Some 'ordinary language' philosophers, like Antony Flew, regarded appeal to a 'paradigmcase argument' as a refutation of traditional scepticism; others, like]. L. Austin in 'A Plea for Excuses', simply disregarded such a scepticism. 10 It is perhaps worth noting that even now many analytic philosophers still take for granted that scheme A is the only proper taxonomy for philosophy. It is an attitude which takes traditional scepticism as central to any genuine philosophy. Other philosophers adopt either the attitude of ]. L. Austin, mentioned in n. 9, or else that of]. McDowell, whose Mind and World (Cambridge, Mass.: Harvard University Press, 1994) has as its aim (p. 113) 'not to answer sceptical questions, but to begin to see how it might be intellectually respectable to ignore them, to treat them as unreal, in the way that common sense has always wanted to'. Barry Stroud's The Quest for
7°
Bird
A. Issues of Philosophical Justification (scepticism)
Descriptive metaphysics (Rejects scepticism)
B. Metaphysics
Revisionary metaphysics (Accepts scepticism)
Descriptive
Justificatory (scepticism)
------
------
No revision Revisionary (Rejects scepticism) (Accepts scepticism)
Fig.
I
not difficult to see in The Bounds of Sense, and in the debates with Stroud, the belief that Kant's and Strawson's transcendental arguments have the same role of refuting traditional scepticism about external objects and persons. But I shall argue that the first difference between Strawson's and Kant's descriptive metaphysics is that while the former adopts scheme A the latter adopts B; while Strawson's transcendental arguments are directly aimed at such a traditional scepticism Kant's are not. Later, in Skepticism and Naturalism (p. 23), Strawson gave an account of descriptive metaphysics and its opposition to revisionary metaphysics that is closer to scheme B. For he there opposes descriptive metaphysics not just to 'revisionary' but to 'validatory', or what in scheme B is called 'justificatory', metaphysics. This change is, moreover, represented as a response to Stroud's criticisms of the anti-sceptical force of Strawsonian transcendental arguments, and in the context of a doubt (Skepticism and Naturalism, 10) whether transcendental arguments have any anti-sceptical point at all. It will be evident from what I say later that, as an account of Kant, I think this goes too far in the opposite direction. I(ant's transcendental arguments must have some anti-sceptical point; the problem is to say exactly what it is. Even if it is accepted that Strawson's descriptive metaphysics in Individuals was avowedly anti-sceptical, it may seem just as obvious that Kant's project is also anti-sceptical and so matches scheme A and not B. It may seem absurd to deny that Kant rejects scepticism. It is, however, not my intention to deny that; the question is not whether but how his descriptive metaphysics relates to sceptical issues. The point can be reinforced by noting that traditional scepticism comes in different forms, e.g. as Cartesian or Humean, global or local, not all of which may be explicitly targeted by Kant, and which may be variously the targets of different arguments, even transcendental arguments, in the Critique. Scheme B does not positively exclude any reference to these varieties of traditional scepticism; it offers initially a neutral description of our experience which may then be deployed Reality: Subjectivism and the Metaphysics of Colour (Oxford: Oxford University Press, 2000) provides a recent example of the belief that scheme A is the fundamental taxonomy for philosophy.
Kant~s
and
Strawson~s Descriptive
Metaphysics
71
subsequently to resolve sceptical issues. I want to identify two such indirect strategies 11 that are closely integrated in the Critique. In the first Kant uses the project to address a global scepticism about experience in general in the challenge of the Copernican experiment and, for example, in the Transcendental Deduction. In the second he uses his descriptive project to address specific local scepticisms as he does in the Second Analogy against Hume's account of cause, and in the Refutation of Idealism against Berkeley's and Descartes's account of external objects. I have represented that second strategy elsewhere 12 as one in which Kant aims to correct sceptical misallocations, or misdescriptions, of priority between elements in experience. In the Refutation of Idealism the correction is to reverse the traditional idealist priority claimed for inner experience over outer; in the Second Analogy it is to reverse the empiricist priority claimed for particular causal connections over the general causal principle. Here I concentrate on the need to clarify the first, more general, strategy. The initial descriptive project of scheme B corresponds to Kant's expressed intention to provide an inventory of the a priori elen1ents in our experience (A xx), or to map the complex web of our beliefs (B 117: 'das sehr gemischte Gewebe'), so that we may successfully navigate a way around experience (B 128: 'to find for human reason safe conduct between these two rocks' (of dogmatism and scepticism) A 235-6). The use of such a map, as a navigational aid, is indirectly to correct errors in the sceptics' maps, as in the two examples given above; Kant expresses this strategy very clearly and well in the Prolegomena (iv. 262 13 ): We have become accustomed to see old, threadbare ideas newly refurbished and given new clothing; this is what most readers will now expect from the Critique. The Prolegomena will bring them to see that it provides a quite new science, never before conceived, and anticipated only in a hint in Hume's doubts, which, however ... led him to beach his ship on the strand of scepticism. My intention is to provide a pilot, who, with secure principles of navigation ... a complete map of the sea and a compass, can (as Hume wished) lead the ship to safety.
11 Sally Sedgwick suggested to me that the 'direct/indirect' distinction was not wholly clear. The primary point is that Kant's descriptive metaphysics of experience, as I understand it, provides initially a neutral survey of our experience, whose relevance to scepticism remains open and indirect. It turns out that for Kant the survey is relevant to a Hun1ean scepticism, or empiricism, which denies the existence of a priori concepts or intuitions. Strawson's anti-scepticism in Individuals is plainly very different from this in many ways. The taxonomy of scheme A, and the explicit rejection of scepticism in Individuals (pp. 33-6), suggest that for Strawson descriptive metaphysics already contains directly an anti-sceptical position. 12 Graham Bird, 'Kant's Transcendental Arguments', in Reading Kant, and 'A Reply to Ralph Walker' in Robert Stern (ed.), Transcendental Arguments: Problems and Prospects (Oxford: Clarendon Press, 1999). 13 References to Kant's works, other than the Critique of Pure Reason, are given to the Akademie edition (Berlin: Walter de Gruyter, 1902- ) by volume and page number.
Bird Even where Kant's descriptive metaphysics is used, as in those cases, against a local scepticism, for example, about external objects in the Refutation of Idealism, it does not already contain a direct, and potentially questionbegging, answer to such scepticism. It takes more seriously than Strawson's the neutral attitude of 'being content merely to describe experience', and so matches scheme B rather than A. Once that neutral description is available, however, it can be used as an indirect argument against a sceptical idealism to correct an erroneous priority in that doctrine; and then it meets the criticism I made earlier of Strawson's later doubts whether transcendental arguments have any anti-sceptical point at alL I have, however, also suggested that Kant uses his descriptive metaphysics in a more general context, namely that of the Copernican experin1ent; but I shall claim that not even that more general use of such a descriptive metaphysics is a direct refutation of traditional global scepticism. It would, of course, be merely question begging to assume that Kant, as a traditional philosopher, must have intended to use his descriptive n1etaphysics directly to refute traditional scepticism of whatever type. For the issue is precisely whether Kant should be treated as a standard traditional philosopher, when there are so many explicit indications, from the 'Copernican experiment' on (B xvi-xix), that he wished to reject the previous tradition. 14 Significant here is Kant's attitude of accepting science and experience but querying philosophy; in that way he reverses the posture of a traditional scepticism that accepts the authority of philosophy and questions aspects of science and ordinary experience. Kant's so-called Copernican revolution is not only a radically new direction for philosophy, but also a reversal of previous philosophical orthodoxy. That attitude is clearly expressed at B 119-20 and Prolegomena 40, but perhaps its most forceful statement is in the Logic (ix. 83-4), where Kant considers, and dismisses, traditional scepticism and replaces it with a legitimate 'sceptical method'. The central points in the passage are as follows: 72
scepticism destroys all our efforts to achieve certainty by casting doubt on all knowledge clain1s. The n10re damaging scepticism is, the more useful and fruitful is sceptical method, by which I understand a way of treating something as uncertain, testing it to the highest point of uncertainty, with the prospect of getting on the track of truth ... It is a method which holds out the prospect of achieving certainty. There is no place for scepticism in mathematics or physics. Only that
14 The language of the Copernican experiment gives one such indication. Another is in the opening page of the Prolegomena: 'it is absolutely necessary ... to set aside all previous work (in metaphysics), to regard its past as never having happened, and as a priority to ask whether metaphysics is even possible.' Kant's reply to Garve in the appendix to the Prolegomena (iv. 372-82) also offers a vehement expression of Kant's revolutionary intentions, in which he complains that Garve has simply failed to understand the radical change in metaphysics canvassed in the Critique.
Kant's and Strawson's Descriptive Metaphysics
73
knowledge is vulnerable which is neither mathematical nor empirical; that is, pure philosophy. Absolute scepticism represents everything as illusory. It thus distinguishes illusion from truth, presupposes a criterion for distinguishing them, and so contradicts itself.
In the passage Kant reverses a traditional belief in the authority of a first philosophy over science, anticipates a Popperian hypothetico-deductive method for science, and indicates a sensitivity to the different kinds of 'certainty' appropriate for different disciplines, even a priori disciplines. It is true that Kant plainly intends to reject a general Humean scepticism and not just the local issues about cause, external objects, or personal identity, but we need to look more closely at the kind of general Humean scepticism that Kant rejects. Kant, after all, accepts Hume's scepticism about transcendent claims, and accepts many of Hume's analyses, for example of specific causal relations (B 793-4) and of the kaleidoscopic character of an empiricist self (B 132-4). The central point at which Kant rejects a general Humean scepticism concerns the question of a priori features of experience, to which both the Copernican experiment and the descriptive inventory of experience are immediately relevant. Kant's general rejection of scepticism is primarily directed at a Hun1ean en1piricist conception of the a priori as only analytic truth (B 20; B 793) .15 It is primarily a rejection of that empiricist conception, and indeed of empiricism generally. It is not a direct attempt to refute local scepticism about external objects and personal identity, but a wholesale rejection of an empiricism held to be responsible for scepticism. Such a response is indirect because it rests on the descriptive inventory of experience, and because that inventory yields a correction to a sceptical empiricism. Just as the treatment of local scepticisms corrects the n1isdescription of priorities in experience, so the treatment of this general scepticisn1 corrects the empiricist misdescription of the a priori in experience. The problem of scepticism is not resolved, in either strategy, directly by accepting the sceptics' framework; it is resolved indirectly by rejecting the frameworks, of traditional idealism and traditional empiricism, in which the problem arises. Since Strawson's conception of the a priori is closer to an empiricist Hume than to an anti-empiricist Kant,16 that issue underlines this fundamental difference between the two kinds of descriptive metaphysics.
15 Kant makes the point at B 20 and Prolegomena (iv. 272-3) that Hume would not have persisted with his empiricist scepticism if he had recognized that mathematical propositions were not all analytic truths, but that some were synthetic a priori. (See also B 127.) 16 Bounds of Sense, 43: 'it must be concluded that Kant really has no clear and genera] conception of the synthetic a priori at all.' The passage is compatible with a belief that there is a clear conception of the synthetic a priori even though Kant lacked it, but even in the later Skepticism and Naturalism (p. 91) Strawson focuses on the analytic a priori in his reference to Humean 'relations of ideas'.
Bird It may be immediately objected that Kant appeals to his non-empiricist conception of a priori principles precisely as a means of guaranteeing knowledge and so attempting to refute traditional scepticism within its own terms; but two initial responses can be made to this. First, the objection concedes the point that the initial, direct, response to a general Hun1ean scepticism arises over the status of the a priori principles disclosed in the project of a descriptive inventory. It concedes the central point represented by scheme B. Second, it pren1aturely answers the further question of the indi.rect relation between Kant's descriptive project and traditional scepticism when that response needs to be left open. It aSSUll1es, prematurely, that Kant's appeal to a priori principles is offered as a way of satisfying the sceptics' standard for 'knowledge'; that is, of providing a guarantee that sceptics would accept. Kant's appeal to scheme B as I have represented it, with its identification of a priori principles, certainly raises the question of its further implications for traditional scepticism, but does not yet answer it. In order to take that further step, more has to be said about Kant's conception of the a priori, and this will be considered in Section III. 74
II. THE PROJECTS' METHODS (LINGUISTIC ANALYSIS AND TRANSCENDENTAL PSYCHOLOGY)
Strawson's apparatus in his descriptive metaphysics is derived from semantic/ pragmatic aspects of language. The central notions of 'identification' and 'reidentification' are defined in terms of speaker/hearer con1munication. The apparatus reflects in that way the analytic philosophy of the 1960s, in which epistemological issues were rephrased in linguistic terms partly in order to avoid the supposed errors of an ambiguous 'psychologism'. Kant also makes important references to language and logic, but his central thesis concerns our cognitive powers. References in the Metaphysical Deduction to logical forms, and in the Transcendental Deduction (B) to judgement-conjunction (Verbindung) are at the service of that exploration of our cognitive powers. This marks the second major difference between the two projects: that within a philosophical framework Strawson's is primarily linguistic and Kant's primarily psychological. It is unsurprising in the light of a background hostility to psychologism that Strawson should have criticized Kant's project as the 'imaginary subject of transcendental psychology' (Bounds ofSense, 32). Later commentators have rightly rejected that criticism of Kant i ? and so long as
17 See Graham Bird, 'Kant's Transcendental Idealism', in G. Vesey (ed.), Idealism-Past and Present (Cambridge: Cambridge University Press, 1982); P. Kitcher, Kant's Transcendental Psychology (Oxford: Oxford University Press, 1990); A. Brook, Kant and the Mind (Cambridge: Cambridge University Press, 1994).
Kanfs and Strawson's Descriptive Metaphysics
75
there is an intelligible role for Kant's transcendental psychology it is difficult to see any reason for preferring Strawson's approach to Kant's. Both aim to identify conditions that are constitutive of our experience, and differ only in their diverse appeals to language or to cognition. Kant would certainly have claimed that his cognitive conditions are prior to Strawson's linguistic conditions, but I shall not defend that view here. Both projects have been accused of a failure to acknowledge the 'relativity' of their constitutive conditions. Strawson was criticized, for example, by Susan Haack, for treating a local feature of Indo-European languages, namely their use of subject-predicate forms of judgement, as if it were universal. 18 !(ant was criticized for treating the historical dominance of Euclid and Newton in the eighteenth century as if it were universal. Whether these criticisms are sound or not, the two projects face a related problem about their methods of enquiry, for the issue about 'relativity' is the obverse side of an issue about 'necessity'. At its most general the problem is to explain how a descriptive survey of contingent experience can be used to yield necessary a priori principles. A potential difficulty for Strawson's later views is expressed clearly by Michael Friedman,19 when he notes Strawson's 'un-Kantian' appeal to a 'sui generis rational intuition of such necessary, a priori, truths'. That appeal is not only un-Kantian but also, more seriously, leaves the relevant modalities in need of elucidation. As the comments about 'relativity' suggested, it is at best dangerous to infer necessary, universal, truths from the local character of experience, and unclear how such an inference can be made. These difficulties arise for Kant as well as for Strawson. It might even be said that they arise for Kant in a more intractable form, since he is committed to the controversial characterization of the fundamental principles as synthetic a priori. Kant has the advantage of adopting a quite definite classification, but the disadvantage that that classification is open to innumerable queries. Beyond that, however, !(ant has also an appeal to a method through which the a priori principles are to be identified, namely his 'abstraction' or 'isolation' procedure. The procedure, formally presented in the introduction (B 5-6) as a way of distinguishing a priori and a posteriori elements in experience, is evidently a primary resource for the descriptive inventory. The procedure is not extensively explained in the introduction, although many references are made to it throughout the Critique (e.g. A 116, A 119, B 144, B 162-3, etc.), and its importance is often overlooked or 18 See Susan Haack, 'Descriptive and Revisionary Metaphysics', Philosophical Studies, 35 (1979), 3 61 -7I. 19 In Proceedings of the Aristotelian Society, suppl. vol. lxxii (1998), 116-17, Friedman refers to Strawson's Scepticism and Naturalism (London: Methuen, 1985). In my reply to Friedman in the same volume, pp. 131-5 I, I disagree with Friedman's naturalism but accept his criticism of Strawson.
Bird 20 misunderstood. The outcome in those references is of an abstract, schematic, constitutive structure for experience whose principles are a priori or necessary. The abstraction procedure provides for Kant a transition from the survey of our contingent experience to items that are not themselves contingent but are constitutive of that experience. Nothing in such an account ensures that the transitional procedure is successful, or that its operation is even fully understood, but I suggest that Kant provides a more substantial account of it in his treatment of Inathematics and formal science,21 especially Euclidean geometry. For Kant's conception of Euclidean geometry provides a model in terms of which to understand the procedure in his metaphysics, despite the differences between the two disciplines which Kant repeatedly emphasizes. 22 Just as, for Kant, Euclid provides a schematic, abstract, representation of spatial experience, so descriptive metaphysics is to provide an abstract, schematic, representation of our experience as a whole. Just as Euclid provides such a schematic account of the structure of spatial experience, so Kant's metaphysical principles are to provide such an account of the structure of the whole of experience. Some provisos need to be made here. Kant's conception of Euclidean geometry is not that of a formal system in a contemporary sense, as Michael Friedman has persuasively argued. 23 I
20 For a recent discussion that turns on Kant's use of the 'abstraction' procedure in his appeal to transcendental apperception and the self see Susan Hurley, 'Kant on Spontaneity and the Myth of the Giving', Proceedings of the Aristotelian Society, xciv/2 (1994), 137-64, Graham Bird, 'Kantian Myths', and Susan Hurley 'Myth upon Myth', Proceedings of the Aristotelian Society, xcvi/2 (1996), 245-51; 253-60. For another recent reference to this procedure see Lorne Falkenstein, Kant~s Intuitionism (Toronto: University of Toronto Press, 1995). 21 Henry Allison, Kant's Transcendental Idealism (New Haven, Conn.: Yale University Press, 1983), 241-2, makes a similar point about Kant's understanding of Newton's mechanics as providing a model for his account of abstraction. But Newton's theory is also regarded by Kant as requiring an appeal to a posteriori matter, and so is not as useful a model for an a priori metaphysics as Euclidean geometry, or any other pure a priori discipline. 22 See e.g. ii. 168 and B 712-38. 23 Michael Friedman, Kant and the Exact Sciences, ch. I, esp. pp. 63-6. 24 Kant often speaks of geometry in this way: e.g. ii. 168, ii. 278, iv. 283, B 207-8, B 217-18.
Kant's and Strawson's Descriptive Metaphysics
77
That account brings to light other significant differences between Strawson's and Kant's projects. The metaphysical structure of experience that Kant identifies is at once both more wide-ranging and more narrowly focused than Strawson's. Whereas Strawson was concerned in Individuals primarily with the sceptically contested notions of external objects and persons, Kant's interest ranges more widely over the categorial notions of intensive and extensive magnitude, substance, cause, and modality, all directed towards the associated time discriminations we make in experience. It involves our notions of size, measurement, events, and the modalities of actuality, necessity, and possibility as well as those of external objects and persons. It is, moreover, concerned just as much with those notions as they are used in science and mathematics as with their use in ordinary experience. Kant's descriptive metaphysics is a descriptive metaphysics of experience that includes science; it is a descriptive metaphysics of science, including psychology, and of ordinary experience. 25 Those points make it clear that Kant's project, even in the first Critique,26 ranges more widely than Strawson's in Individuals. On the other hand, Kant's most fundamental concepts, those of conceptual and personal unity incorporated in transcendental apperception, are so schematic and abstract as to be, as they stand, unidentifiable in experience. They are, as both Strawson himself and Henry Allison note,27 conditions of experience rather than identifiable elements in it. Just as Euclid's points, lines, and planes cannot be directly experienced, so the fundamental structural concepts of experience cannot be directly identified through discriminable items within it. Just as the abstraction procedure makes this paradox intelligible for Euclidean geometry, so it serves to explain the apparent paradox of descriptive
25 In our joint symposium 'Kantian Themes in Contemporary Philosophy', Aristotelian Society, suppl. vol. 72 (1998), I disagreed with Michael Friedman's claim that Kant's synthetic a priori principles should be seen simply as scientific principles, and with an underlying 'naturalism' which encouraged such a view. Kant explicitly regards his principles as metaphysical and not merely scientific; I take the view that they make both science and ordinary non-scientific experience possible. It is worth noting that the inclusion of psychology within Kant's descriptive metaphysics allows there to be a priori principles of psychology without any commitment to an 'in1aginary subject' (see n. 16). 26 Beyond the first Critique, into the second and third Critiques and the applied philosophies of law and science, Kant's appeal to an a priori structure for experience has, of course, an even wider scope. 27 Strawson makes such a claim in 'The Problem of Realism and the A Priori', in Parrini (ed.), Kant and Contemporary Epistemology, 169: 'what is a necessary condition of any empirical judgement is not itself the product of any such judgement, i.e. is a priori.' Henry Allison refers to a sin1ilar point, but without wholly endorsing it, in Kanfs Transcendental Idealism, 143, when he says: 'it is generally assumed that, as conditions of experience, transcendental activities cannot in principle become objects of awareness.' The point I am making through the Euclidean model is different from these, and relies on the idea that an abstract representation of experience should be distinguished from its empirical realizations in experience.
78
Bird
metaphysics that necessary, a priori, elements can be identified from an exploration of our contingent experience. Kant's project takes seriously the Euclidean model, in which the fundamental elements are abstract and schematic and not directly to be identified with their empirical realizations . . In experIence. Strawson's project has no such model; it identifies fundamental aspects of experience just as they empirically occur in that experience. The fundamental roles outlined for external objects and persons are fundamental roles for our conception of those items just as they figure in ordinary experience. This is why, for example, Strawson's account of persons places such weight on embodiment, and, by the same token, why Kant's does not. 28 Kant's abstract appeal to personal identity in transcendental apperception has no more need to acknowledge our embodiment than Euclid's abstract appeal to points and lines in plane geometry needs to acknowledge the thickness of their empirical realizations in experience. This difference between the two projects need not produce any conflict; it may be that both are viable and yet quite different from each other. Strawson's project is an empiricist project, rather like Locke's, which Kant thought inappropriate for his purposes; Kant's is explicitly anti-empiricist, and more like Leibniz's, even though both Kant and Strawson, but especially Kant, would want to differentiate their views from those of their predecessors. Whichever of these procedures, Strawson's empiricist or Kant's anti-empiricist, is acceptable or preferable, at least Kant offers such an explanation of his method and of the relation between our contingent experience and its necessary structure. Even so, such an account of the transitional procedure from contingent experience to a priori structure says nothing about how we are to understand the a priori necessities it discloses, and I finally consider that in Section III.
III. THE NATURE OF NECESSARY A PRIORI FEATURES OF EXPERIENCE
The differences so far outlined between Strawson's and Kant's descriptive metaphysics converge on two residual issues. The different taxonomies (in I) raise the question for Kant of the way in which a priori principles respond, directly or indirectly, to traditional scepticisn1. The diverse procedures (in II) leave open the question of Kant's understanding of the a priori. 28 For more on the relation between Kant's account of the self and his 'abstraction' procedure see McDowell, Mind and World (Cambridge, Mass.: Harvard University Press, 1994), 103-4 and Graham Bird, 'McDowell's Kant: Mind and World', Philosophy, 71 (1996), 219-43. A similar point can be made about Strawson's understanding of the First Analogy, which addresses the issue of the empirical criterion, or empirical realization, for 'substantiality' despite the fact that Kant's argument explicitly sets that aside (B 232).
Kant~s
and
Strawson~s Descriptive
Metaphysics
79
Strawson's response to these issues is clear but vulnerable. His descriptive metaphysics in Individuals already contains an answer to traditional scepticism, although it is one that he now admits to be unsuccessful; and he understands the a priori in an empiricist way as the analytic, despite Quine's attack on that empiricist dogma. Of the two residual issues for Kant the second has priority, since we cannot understand how the a priori relates to scepticism until we know what the a priori is. I shall therefore consider that issue first, but only in a limited way. Ideally what is needed is a properly formal account of Kant's 'synthetic a priori' classification and of his conception of modal necessity, but although I don't simply assume that these requirements cannot be met, I cannot meet them here. 29 Instead I shall consider some of the apparent an1biguities in Kant's account that indicate, broadly, two distinct conceptions of the a priori. I then consider the different implications these two conceptions have for traditional scepticism. Summarily I(ant's account of his synthetic a priori principles might be understood in the following terms: Class I:
(i) Principles that are true in all possible worlds. (ii) Principles that are unrevisable, and indefeasible, in any possible experience. (iii) Principles that are true in some salient subset of all possible worlds.
Class II:
(iv) Principles that are true in some fundamental way in experIence. (v) Principles that are fundamentally prior to, and constitutive of, empirical experience. 3o
I want to class (i) and (ii) together as conception I, and to class (iv) and (v) together as conception II. (iii) cannot immediately fall into either I or II until the 'salient subset' is adequately identified, although it seems naturally to fall closer to II than to I, as it does, for exan1ple, in Gordon Brittan's account. 31 It may seenl a mistake to regard class-IT principles as a priori at all, on the ground that the a priori is unrevisable, indefeasible, necessarily
29 i hope to provide such a more fundamental account in a forthcoming commentary on the first Critique. 30 In the discussion it was put to me that (iv) and (v) differ in that the former, but not the latter, ascribes truth to the principles, with the suggestion that the formulation of (v) is preferable. Clearly if there are any such principles they cannot be merely empirical truths, but Kant seems committed to their being true in some other way and I cannot at present see any strong objection to such a claim. 31 Gordon Brittan canvassed such an account in his Kanfs Theory of Science (Princeton, N]: Princeton University Press, 1978).
80
Bird
true, and holds in all possible worlds, but many recent accounts of the a priori reject that claim. 32 One relevant difference between the two classes is that I seems to offer an immediate, traditional, answer to scepticism, whereas II does not. For both conceptions in I claim that there are certain principles whose fixed, eternal, necessary truth can be guaranteed for all experience and in any possible world. The evident price to pay for such a guarantee is the extreme difficulty of establishing that there are any such principles, or identifying which they are. That issue cannot be sidestepped in the Kantian context by associating I with only analytic, or logical, truths; for these are not at issue among empiricist sceptics, and do not in any case correspond to Kant's synthetic a priori principles. That handicap in the anti-sceptical response marks also a difficulty in ascribing that conception of the synthetic a priori to Kant. The analytic a priori may, but synthetic a priori principles cannot, be understood in those terms. This may be concealed by classifying both class-I principles and Kant's synthetic a priori as necessary truths and, as I shall suggest, by failing to distinguish Kripke's notion of necessity from Kant's. Class-II conceptions plainly respond to traditional scepticism in a different way. Principles that are fundamental to, constitutive of, and prior to, empirical experience are not necessarily true in all possible worlds, and may be both revisable and defeasible. Whereas I emphasizes a strict Kripkean necessity, II focuses more on the contingency of our experience; but if I seems too strict to cover Kant's synthetic a priori, II may seem not strict enough. If II focuses on the contingency of our experience how can its principles deserve to be called 'necessary'? This is to repeat the connected problems about 'relativity' and 'necessity' noted earlier. That difficulty will seem all the more intractable if Kripke's account of necessity is taken for granted, and if it is not recognized that Kant's conception of necessity differs from Kripke's. Kripke understands necessary truth as truth in all possible worlds, and includes logical and other, e.g. metaphysical, truths in this class; Kant officially recognizes necessity primarily as a mark of the a priori and may therefore treat only logical or analytic truth as truth in all possible worlds; that is, as necessary in Kripke's sense. This divergence in their respective taxonomies explains why, even though Kant's own synthetic a priori truths reflect a substantial similarity with Kripke's class of contingent a priori truths, he could not have accepted Kripke's label of 'contingent a priori'
32 Paul Moser in his introduction to A Priori Knowledge (Oxford: Oxford University Press, 1987) and Christopher Peacocke in his 'The Origins of the A Priori', in Parrini (ed.), Kant and Contemporary Epistemology, both allow the a priori to be revisable and defeasible. The point is sometimes assumed, sometimes questioned, in the issue of Philosophical Studies (92/1-2 (Oct. 1998)) devoted to discussion of the a priori.
Kan('s and
Strawson~s Descriptive
Metaphysics
81
truths. Since for Kant all a priori truths are necessary truths, and what is necessary is not contingent, the contingent a priori would be for him an inconsistent classification of what is both contingent and necessary. Class-IT conceptions of the synthetic a priori are anchored firmly in many central aspects of Kant's position. Those central aspects include, for example, his evident anthropocentricity, in which synthetic a priori principles reflect what is necessary for our human experience; his strong appeal to immanent priorities and dependence relations within that experience; his general rejection of absolute necessities; and his characterization of the a priori principles as 'synthetic'. These aspects reinforce the admitted relativity of synthetic a priori principles, which fits class-II conceptions more naturally than class I. Two inadequate reasons for ascribing I to Kant are a current association of the a priori with knowledge 33 and a connection between the a priori principles and certainty in our experience. It has become customary to treat the a priori/a posteriori distinction as epistemic rather than semantic, and as having to do with what we can, or cannot, know independently of experience. It may seem inevitable, therefore, to assume that to characterize a principle as a priori is to guarantee it against a sceptic's challenge to knowledge. Such a claim, however, requires two additional, and questionable, premisses; first, that the appeal to knowledge is essential, and, second, that any such appeal addresses the sceptics' challenge. The first premiss would be rejected if, following Kripke,34 the a priori were characterized in terms of 'belief' rather than knowledge. At one point in Naming and Necessity (p. 260) Kripke makes the following proviso: 'It might be best, therefore, instead of using the phrase "a priori truth" ... to stick to the question whether a particular person or knower knows something a priori, or believes it true on the basis of a priori evidence.' The second assumption would be rejected if the appeal to knowledge, or to belief, was neutral with respect to traditional scepticism, or concerned a justified but defeasible standard rather than some guarantee that would meet a sceptic's requirements. The same difficulty attends the reference to Kant's evident thought that a priori principles transmit certainty to the empirical circumstances that realize them in experience. In the passage where Kant outlines his 'abstraction' procedure (B 5-6) he also says: For whence could experience derive its certainty, if all the rules according to which it proceeds were always themselves empirical and so contingent?
33 In Chapter 5 Quassitp Cassam distinguishes between 'derivational' and 'justificatory' accounts of a priori concepts. The account I have offered does not fit neatly into either alternative, but is closer to the former. 34 S. Kripke, 'Naming and Necessity', in D. Davidson and G. Harman (eds.), Semantics of Natural Language (Dordrecht: D. Reidel, 1972), 260.
Bird However natural it may be to treat the reference to certainty as an antisceptical gesture, such a view depends on additional assumptions about the appropriate standard of certainty. If it is assumed that the appropriate standard is that of an anti-sceptical guarantee, then the passage points towards a conception of the a priori falling under I. But, since this is just what is in question here, to make that assumption is to beg the question. 35 Neither of these arguments is compelling. They conflict moreover with the passages already noted from B 120, Prolegomena 40, and Ak. ix. 83-4, where Kant rejects that traditional approach to scepticism. On the other side Kant evidently is concerned to outline a necessity that functions immanently in, and relative to, our experience rather than one that claims to hold in every possible world. Kant limits the extent of our understanding of other possible worlds where our a priori principles do not hold, and so marks the difference between those principles and pure contingencies in experience. Whatever is purely contingent in our experience should place no limits on the intelligibility of possible worlds without those features. Kant's references to what I called many years ago 'deviant experiences',36 such as an 'intuitive understanding' (B 135, B 138-9) or a 'non-categorial intuition' (B 122-3), show on one side that our cognitive arrangements may not be duplicated in other possible worlds and on the other that certain of our arrangements have an in1manent necessity in our experience. They are constitutive of our experience and consequently make it increasingly difficult to understand alternatives that progressively deviate from that experience. The idea implicit in such a picture is that our experience provides a core structure, constituted by its a priori principles, departures from which become progressively harder for us to understand. 37 At the limit such deviations become strictly unintelligible, but before that stage is reached it is possible to conceive of worlds where our a priori
35 In Chapter 6 Barry Stroud makes exactly this assumption and quotes Kant's belief that the results of metaphysics should be 'apodeictically certain'. In this context, however, the question is whether this marks a standard of certainty, or truth, or knowledge, designed to satisfy a traditional sceptic, and Kant indicates at B 120 and B 56 and in the Logic (ix. 83-4) that that issue is irrelevant: 'Geometry proceeds with security in knowledge that is completely a priori and has no need to beseech philosophy for any certificate of the pure and legitimate descent of its fundamental concept of space' (B 120); 'This ideality of space and tin1e leaves the certainty of empirical knowledge unaffected, for we are equally sure of it whether these forms inhere in things in themselves or only in our intuition of them' (B 56). The passage from the Logic is quoted and commented on in the text. 36 I considered these cases in Kanfs Theory of Knowledge, ch. 9. 37 The model of core cases with deviations from them will seem foreign to commentators who regard Kant's distinctions as always rigid dichotomies. My contention is that the 'core/deviant' model is a consequence of many of Kant's claims, such as his account of the three-fold synthesis, and that it is something of a prejudice to deny this. Recent commentators on Kant's moral philosophy have found it easier to reject that prejudice than those who consider his metaphysics or epistemology.
Kant's and Strawson's Descriptive Metaphysics
83
principles do not hold. Kant's a priori principles would not then be necessary in Kripke's sense, would not hold in all possible worlds, and would not offer an answer to traditional scepticism in virtue of their guaranteed truth in all possible worlds. My claim is not that we can easily explain the Kantian requirements that this picture represents, but only that they are Kant's requirements. They point towards II rather than I, and provide no direct answer to scepticism within its traditional framework. Kant's motive is instead to abandon that traditional framework, and to focus philosophical attention on the more progressive enquiry, into the structure of our experience, of his descriptive metaphysics.
SUMMARY
I suggested earlier that the picture of Kant I proposed was radically different from Strawson's project in Individuals, in relation to traditional scepticism, to the methods of the two enquiries, and to the status of the modalities involved in the variously disclosed fundamental principles. I hope that the extent of those divergences will already be apparent, but I summarize and underline some of the consequences of that divergence. Kant's attitude to traditional scepticism is neither that of Strawson's Individuals nor that of the later Skepticism and Naturalism. In the former, descriptive metaphysics already contains a claimed refutation of scepticism; in the latter, it is consciously distanced fronl any such goal. I have argued that Kant's descriptive metaphysics is not directed against scepticism in Strawson's former way, but that it is not irrelevant to scepticism either. It provides a descriptive account, a map or inventory, of the structure of our experience, which identifies fundamental principles, more wide-ranging than Strawson's, and the relations of priority and dependence they have to each other and to specific features in experience. Kant's map can then be unrolled, as he claims, to show that traditional sceptics had, in certain cases, simply misdescribed our experience, and had carelessly assigned priorities to certain features of experience that are not only erroneous, but need to be reversed. Kant's Copernican 'revolution' is, in this and in many other ways, not merely a radically new path for philosophy but also a reversal, a turning round, of earlier traditional assumptions. In such an account Kant is not accepting the traditional assumptions that generate scepticism and providing an answer, a refutation, that meets the sceptics' requirements; he is questioning, and rejecting, those assumptions and the whole traditional framework in which they figure. This is nowhere more apparent than in the ill.ost general application of descriptive metaphysics to scepticism about the a priori status of the fundamental principles. I have suggested that in that context Kant's anti-scepticism is
Bird a rejection of empiricism. Either Kant thought that empiricism was necessary for scepticism in the context of the previous tradition from which he wished to escape, or he thought that empiricism was at least sufficient for scepticism and by rejecting empiricism he blocked that route to sceptical doubt. In neither this global nor other local contexts does Kant's position rest on the empirical idealism with which transcendental idealism is sometimes identified; nor does it rest on any appeal to a semantic verificationism. If Strawson's anti-scepticism rests on that latter verificationism, then it differs significantly from Kant's in that respect too. That difference, of course, points also to the more general difference that Strawson is essentially an empiricist while Kant is decidedly not. Semantic verificationism goes naturally with empiricism and would sit uncomfortably with Kant's anti-empiricism; but it is also by now well understood that Kant did not accept such a verificationist doctrine. It seems to me important to stress the Euclidean model in Kant, which explains some of the apparent conflicts between Strawson's descriptions and Kant's, and allows both projects to function compatibly in some cases as descriptions of experience. Strawson adopts a robust Anglo-Saxon empirical approach; Kant's is more rationalist, more scientific, more continental. Strawson's method is that of the empirical surveyor, Kant's is that of the geometrician. Both are equally viable in general, both have their own limitations, but their results are not necessarily in conflict. It is in the area of the two, empiricist or non-empiricist, conceptions of the a priori that the two projects are most diverse and most vulnerable. Strawson's empiricist appeal to an analytic a priori is at least open to wellknown objections, but clearly has little relevance to Kant's conception of synthetic a priori principles. The later appeals, in Skepticism and Naturalism, to 'clear and distinct perception' or 'rational intuition' of necessary truths seem to be less resolutions of the central issue about modalities than indicators of the need for further enquiry and clarification. I(ant may seem to be no better off. The synthetic a priori classification has been widely rejected, at least by empiricists, Kant's conception of a formal system such as Euclidean geometry is thought to be superseded by more recent logico-en1piricist treatments, and any scepticism generated by empiricism is probably thought to be resolvable in principle only within an empiricist framework. If Kant's position queries or denies these clain1s, as I have suggested, then its prospects may seem unpromising. I comment only (I) that an assumed commitment to empiricism is surprising in the face of its n1any known failures; (2) that many of the attempts to reject the synthetic a priori classification provide no adequate argument for that conclusion,38 84
38 I include in these discussions A. J. Ayer's account of the a priori in Language, Truth and Logic (London: Gollancz, 1936), ch. 4, and Jonathan Bennett's account of the 'analytic/synthetic' distinction in his Kant's Analytic (Cambridge: Cambridge University Pr~s~ ~~61
Kant~s
and
Strawson~s Descriptive
Metaphysics
85
and seem to constitute one more unempirical dogn1a of empiricism; and (3) although Kant does not provide an adequate account of the modalities involved in his a priori principles, any more than Strawson does in the passages cited, he nevertheless points in directions that seem to me worth pursuing. To see that pursuit as worthwhile, however, requires the abandonment of the idealist, verificationist, account of Kant's transcendental arguments in favour of the descriptive metaphysics of experience that he actually provides.
5 A Priori Concepts QUASSIM CAS SAM
••
Are there any a priori concepts? On one reading, a concept is a priori if and only if it cannot be derived or acquired from experience. In Kant's terminology, a priori concepts 'must be in a position to show a certificate of birth quite other than that of descent from experiences' (A 86-7/B II9).1 The contrast is with empirical concepts, which can be derived from experience. 2 The concepts whose apriority Kant is most concerned to emphasize in the first Critique are the categories, but he also applies the label 'a priori' to many other concepts, including those of space and time, mathematical concepts, moral concepts, and the transcendental ideas. When a concept is deemed to be a priori in virtue of the fact that it cannot be derived from experience I will say that it is derivationally a priori. On another reading, the distinction between a priori and empirical concepts has to do with the different ways in which it is possible to legitimate or justify a concept. One suggestion would be that a concept is a priori only if its employment cannot be justified by experience. Concepts whose use cannot be justified at all are not a priori, but concepts which can be justified by experience are empirical rather than a priori. On a different reading, empirical concepts are ones that can only be justified by experience. On this interpretation, the distinguishing mark of a priori concepts is not that they cannot be justified by experience. It is that their use can be justified independently of experience. When a concept is deemed to be a priori in 1 All references in this form are to I. Kant, Critique of Pure Reason, trans. Norman Kemp Smith (London: Macmillan, 1933). 2 In Kant's words, an empirical concept 'springs from the senses through comparison of the objects of experience, and receives, through the understanding, merely the form of generality. The reality of these concepts rests on actual experience, from which they have been extracted as to their content' (Logic, trans. R. Hartmann and R. Schwarz (Indianapolis, Ind., and New York: Babbs-Merrill, 1974),97.
88
Cassam
virtue of how it is possible to justify its employment I will say that it is justificationally a priori. According to Kant, derivational and justificational apriority are closely related. 3 At least in the case of the categories, he seems to think that they are only justificationally a priori because they are derivationally a priori. He also thinks that the derivational apriority of the categories implies that they are wholly subjective in origin. Once it is agreed that a concept does not have its origin in experience, the only alternative is, as Allison puts it on Kant's behalf, 'to view it as part of the cognitive machinery of the human mind, and in this sense as ideal'.4 The cognitive faculty in which the categories are alleged to originate is the understanding. In contrast, the transcendental ideas 'to which no corresponding object can be given in sense-experience' (A 327/B 383) are a priori concepts of reason rather than of the understanding. Nevertheless, as with the categories, their derivational apriority is held to imply their subjectivity. Like Kant, Strawson is ready to accept that there are derivationally a priori concepts. He writes that 'as conditions of the possibility of empirical knowledge of objects' categorial concepts such as substance and cause 'will certainly not themselves be empirical, i.e. is derived from within experience'. 5 If the categories are conditions of the possibility of empirical knowledge of objects, this accounts, at least in part, for the fact that they are justificationally a priori. Strawson's view is therefore that the justificational apriority of the categories implies that they are derivationally a priori. 6 Unlike Kant, however, Strawson does not accept that the derivational apriority of the categories implies that they are wholly subjective in origin. He clain1s that it is not just false but incoherent to suppose that our own cognitive constitution is the source of necessary general features or conditions of experience or empirical knowledge of objects. 7 Strawson's position is, on the face of it, difficult to understand. If it is appropriate for a transcendental epistemologist to ask where the categories come from, and to argue that it is not possible for concepts like substance and cause to come from experience, it would be reasonable to expect someone who argues in this way to provide a positive account of the source of such
3 As Dieter Henrich has emphasized in his important paper 'Kant's Notion of a Deduction and the Methodological Background of the First Critique', in E. Forster (ed.), Kant~s Transcendental Deductions (Stanford, Calif.: Stanford University Press, 1989). 4 H. Allison, The Kant-Eberhard Controversy (Baltimore, Md.: Johns Hopkins University Press, 1973), 80. s P. F. Strawson, 'Kant's New Foundations of Metaphysics', in his Entity and Identity and Other Essays (Oxford: Clarendon Press, 1997), 238. 6 For a closely related line of thought see Patricia Kitcher, 'Revisiting Kant's Epistemology: Skepticism, A Priority and Psychologism', Nous, 30 (1995), 288. 7 See P. F. Strawson, The Bounds of Sense (London: Methuen, 1966), 15-16.
A Priori Concepts 89 concepts. As indicated above, Kant's view is that there are only two possibilities: either categorial concepts come to us from the world, via experience, or we supply them ourselves. If he is right about this, then there is no room for a position that endorses the derivational apriority of the categories while denying their subjectivity. Yet this is precisely the position that Strawson attempts to occupy. In my view, the lesson is not that the categories are, in Kant's sense, wholly subjective in origin. The lesson is that one should not grant the derivational apriority of the categories, or concede that their derivational ap'riority follows from their justificational apriority. One nlay refuse to grant the derivational apriority of the categories either because one rejects the very idea of derivational apriority, together with the distinction between derivationally a priori and empirical concepts, or because one is not persuaded that the categories cannot be derived from within experience. Either way, the transcendental epistemologist should drop the claim that the categories are derivationally a priori. To the extent that it is legitimate to ask about the 'source' of the categories, this is not a question for transcendental episten10logy. The important question is whether the categories, or any of the other concepts that Kant describes as a priori, are justificationally a priori. Unfortunately, this question is, for reasons that will soon become apparent, extremely difficult to answer.
II
What would it be for the employment of a concept to be justified? According to what I will call the instantiation account, one is justified in employing a concept C only if the belief that C is instantiated is justified. 8 Setting aside the special questions raised by fictional contexts, this account implies that concepts such as horse and neutrino are legitimate, whereas mermaid and centaur are not. In the case of some concepts, the belief that they are instantiated can be justified by experience. On one reading, these are empirical concepts. A concept has 'objective reality' as long as it 'stands in relation to an object' (A 109), and, as Kant remarks, 'many empirical concepts are employed without question from anyone' since 'experience is always available for proof of their objective reality' (A 84/B 116). This leads to the suggestion that for C to be justificationally a priori it must be such that experience is not available for proof of its objective reality. In other
8 For a critical response to the instantiation account see Philip Kitcher, 'How Kant Almost Wrote Two Dogmas of Empiricism (And Why He Didn't)', in J. Mohanty and R. Shahan (eds.), Essays on Kant's Critique of Pure Reason (Norman, Okla.: University of Oklahoma Press, 1982),
229-3°·
Cassam words, C is justificationally a priori only if experience cannot justify the belief that it is instantiated. This is a necessary condition for what I will refer to as strong justificational apriority. The proposed condition is not sufficient because, despite the fact that the belief that there are round squares is not capable of being justified by experience, one does not want to allow that the concept round square is strongly justificationally a priori. Experience can justify the belief that an empirical concept C is instantiated in at least two ways. On the one hand, it can justify this belief directly, by presenting us with instances of C at the level of observation. On the other hand, experience can also justify the belief that C is instantiated indirectly, on broadly theoretical grounds. For example, what legitimates the concept of a neutrino is not the observability of neutrinos but the fact that the best scientific description and explanation of the world warrants the belief that this concept is instantiated. Broadly speaking, this kind of indirect or theoretical justification is still justification 'by experience', so neutrino is not strongly justificationally a priori. The present proposal is that for a concept to be strongly justificationally a priori it must be such that experience cannot justify the belief that it is instantiated either directly or indirectly. Are the Kantian categories strongly justificationally a priori? On one view, substances and causes are given in experience. Suppose, next, that perceiving a substance can justify the belief that the concept substance is instantiated. As long as experience can justify the belief that the concept of substance is instantiated, it follows that this concept is not strongly justificationally a priori. Sin1ilarly, cause is not strongly justificationally a priori if, as Anscombe claims, a lot of causality is perceivable. 9 On the present reading of justificational apriority, therefore, the most important of the Kantian categories appear not to be a priori concepts in this sense. An obvious objection to this line of argument would be to deny that the categories are or can be instantiated in experience. For example, Kant claims that the concept of cause 'manifestly contains the concept of a necessity of connection with an effect' (B 5), and that experience only tells us 'what is so but not that it must necessarily be so, and not otherwise' (A I). To this extent, the concept of cause is not instantiated in experience. The impression that this is Kant's view is strengthened by his remark in the 'Schematism' that 'no one will say that a category, such as that of causality, can be intuited through sense and is itself contained in appearance' (A 137-8/B 176-7). On the other hand, when Kant explains the distinction 90
9 See G. E. M. Anscombe, 'Causality and Determination', in her Metaphysics and the Philosophy of Mind: Collected Papers, ii (Oxford: Basil Blackwell, 1981), 133-47. In Analysis and Metaphysics (Oxford: Clarendon Press, 1992) Strawson endorses the closely related view that 'the notion of causation in general does find a footing or, rather, a foundation, and a secure foundation, in the observation vocabulary' (I I 5).
A Priori Concepts
9I
between a category and a transcendental idea, he does so by saying that a category 'can be given in experience',10 whereas an idea is a concept that cannot be so given. This distinction hardly supports the view that the categories are incapable of being instantiated in experience, or that they are strongly justificationally a priori. 11 In order to make sense of Kant's position it needs to be ren1embered that when he derives the categories from the table of judgements they are unschematized. The schemata of the categories are 'the true and sole conditions under which these concepts obtain relation to objects and so possess significance' (A 146/B 185). Without schemata, the categories represent no object and have a 'purely logical' (A 147/B 186) meaning. For example, the unschematized category of substance is the concept of 'something which can exist as subject and never as mere predicate' (B 149). The pure concept of causality is the concept of the relation of ground and consequent. The schema of substance is permanence of the real in time. In the 'General Note on the System of Principles' Kant adds that in order to demonstrate the objective reality of the concept of substance 'we require an intuition of space (of matter)' (B 29 I). The intuition that corresponds to the category of causality is that of 'the movement of a point in space' (B 292). As Michael Friedman puts it, Kant's thesis is that 'substances, causes, and so on are indeed given in experience-in virtue precisely of the fact that material substances interacting in accordance with the laws of motion through the fundamental forces of repulsion and attraction are given in experience'.12 Kant's considered view, then, is that the categories are indirectly instantiated in experience, via the instantiation in experience of less formal concepts such as that of material substance. In this sense, experience can justify the belief that the categories are instantiated. There is a parallel with Strawson's position. He remarks that the categorial notion of an individual substance is 'highly abstract' and does not belong to 'the vocabulary of particular observation'.13 Concepts that do not belong to the vocabulary of particular observation are concepts that are not instantiated in experience in their own right. There is, however, 'a host of expressions for specific kinds or varieties of individual substance which do belong to the vocabulary of particular observation' .14 The concepts (e.g. horse, table) that correspond to these expressions are directly instantiated
10 1. Kant, Prolegomena to Any Future Metaphysics, original trans. Paul Carus, rev. James W. Ellington (Indianapolis, Ind.: Hackett, I977), 3 2 9. 11 It is also worth remarking that when Kant asks in his 'Vienna Logic' whether it is possible to encounter in experience 'objects' (i.e. instances) of the concept of causality, his unequivocal answer is 'Yes. This happens through examples. An example of causality is: fire destroys wood.' See Kant's Lectures on Logic, trans. J. Michael Young (Cambridge: Cambridge University Press, I993), 349· 12 M. Friedman, 'Regulative and Constitutive', Southern Journal of Philosophy, 30 (I99I), 83. 13 Strawson, Analysis and Metaphysics, I I 5. 14 Ibid.
Cassam in experience and are realizations of the abstract categorial notion of substance, so it would be a mistake to insist that there is no sense in which the latter is instantiated in experience. In so far as the categories of substance and cause are at least indirectly instantiated in experience, this weakens the case for regarding them as strongly justificationally a priori. Substance and cause are, however, not the only categories. Another Kantian category is the concept of negation. The case for regarding this concept as strongly justificationally a priori is straightforward: the belief that it is instantiated cannot be justified by experience, either directly or indirectly. Experience cannot present us with instances of the concept of negation at the level of observation, and it would be just as implausible to suppose that this concept can be justified on broadly theoretical grounds, in the way that concepts like neutrino can be justified on broadly theoretical grounds. Yet, far fron1 being usurpatory or illegitimate, the concept of negation is one whose employment is indispensable for anything recognizable as rational thought, and therefore for experience. This argument calls into question the instantiation account of conceptual legitimacy. According to this account, a concept is legitimate only if the belief that it is instantiated is justified. The problem with applying this account to logical concepts like negation is that, on the face of it, they cannot be said to have instances, at least in the sense in which non-logical concepts can be said to have instances. In general, it is plausible that the generality of concepts implies that there can be numerically distinguishable individual objects falling under one and the same concept,15 but it is difficult to attach any sense to the idea of there being numerically distinguishable individual objects falling under the concept of negation. Unless one is prepared to respond to this point by denying that negation is a concept, it follows that legitimating a concept cannot always be a matter of justifying the belief that it is instantiated. It is one thing to legitimate a concept by pointing out that it is indispensable for thought or experience, but this does not commit one to regarding it as having instances. 16 The instantiation account applies, at best, to a subclass of concepts. The case against classifying substance and cause as strongly justificationally a priori relies on the assun1ption that a concept is not a priori in this sense if the belief that it is instantiated can be justified by experience. The very fact that this assumption calls into question the justificational apriority of the categories of substance and causality suggests that it does not accord 92
15 This is Strawson's view of what the generality of concepts involves. See his 'The Problem of Realism and the A Priori', in Entity and Identity and Other Essays, 246. 16 For further discussion of the view that concepts can be legitimated by showing that they are indispensable for thought or experience see B. Stroud, 'Transcendental Arguments', in R. C. S. Walker (ed.), Kant on Pure Reason (Oxford: Oxford University Press, 1982), 117-31, and Q. Cassan1, 'Self-Directed Transcendental Arguments', in R. Stern (ed.), Transcendental Arguments: Problems and Prospects (Oxford: Clarendon Press, 1999), 83-110.
A Priori Concepts
93
with Kant's conception of justificational apriority. On a different reading of Kant's position, what n1akes a concept justificationally a priori is the possibility of justifying the belief that it is instantiated independently of experience. This is a necessary condition for what I will refer to as weak justificational apriority. In the case of empirical concepts, the belief that they are instantiated can only be justified by experience. While it is plausible that experience can justify the belief that there are substances and causes, it does not follow that the corresponding categories are empirical unless the belief that they are instantiated cannot be justified other than on the basis of experience. The present proposal, then, is that the real case for regarding substance and cause as justificationally a priori is the fact that the belief that there are substances and causes is capable of being justified without the aid of experience. Since this revised reading of Kant's position continues to represent justificationally apriority as having to do with the grounds for believing that putatively a priori concepts are instantiated, it still faces the problem of accounting for the (weak) justificational apriority of the concept of negation. It is also needs to be explained what it would be for the belief that there are substances and causes to be justifiable 'independently of experience'. Taking these issues in reverse order, Kant claims that 'the objective validity of the categories as a priori concepts' rests on the fact that 'through them alone does experience become possible' (A 93/B 126). Suppose, then, that the belief that the concepts of substance and cause are instantiated can be justified independently of experience if and only if these concepts are indispensable for experience. Leaving aside logical concepts like negation, concepts that are indispensable for experience are 'necessarily realized or instantiated in experience'.1 7 This n1arks an important distinction between empirical and weakly justificationally a priori concepts. Empirical concepts need not be instantiated, and experience can fail to justify the belief that any particular empirical concept is instantiated. In contrast, if C is weakly justificationally a priori, then C is necessarily instantiated in experience, and experience cannot fail to justify the belief that C is instantiated. In experience, one cannot fail to come across substances and causes. The concept of negation is a problem for this account because it does not follow from the fact that it is indispensable for experience that it is necessarily realized in experience. The best response to this difficulty would be to revise the definition of weak justificational apriority, so that a concept counts as being a priori in this sense just if it is indispensable for experience. This account can be applied to concepts like substance and cause, as well as a priori logical concepts such as that of negation. Instead of representing the weak justificational apriority of substance and cause as primarily
17
Friedman, 'Regulative and Constitutive', 73.
94
Cassam
consisting in the fact that the belief that they are instantiated can be justified independently of experience, it would be better to say that what makes any concept weakly justificationally a priori is, at least in the first instance, its indispensability for experience. This proposal avoids giving the impression that the justificational apriority of substance and cause cannot be explained in the same way as the justificational apriority of the concept of negation. Some concepts that are weakly justificationally a priori are also such that the belief that they are instantiated can be justified independently of experience, but one of the advantages of not characterizing weak justificational apriority directly in these terms is that one is thereby in a position to make sense of the justificational apriority of concepts for which the instantiation account of conceptual legitimacy seems inappropriate. An objection to this approach is that there are concepts that Kant classifies as a priori but that cannot plausibly be regarded as indispensable for experience. Experience, in Kant's sense, is 'a knowledge which determines an object through perceptions' (A 176/B 218). The present approach therefore implies that a concept is weakly justificationally a priori just if it is required for empirical knowledge of objects. It is not uncontroversial that the concepts of substance and causality are indispensable for empirical knowledge of objects, but it is at least possible to see how an argument for their weak justificational apriority, together with that of the concept of negation, might go. The same cannot be said for, say, moral concepts. For example, without thought there would be no knowledge, so a way of showing that the concept of negation is indispensable for experience would be to show that it is indispensable for thought. 18 As for the categories of substance and cause, the best case for regarding them as weakly justificationally a priori relies on the principle that conditions of knowledge must reflect the nature of the objects of knowledge, as well as the cognitive faculties of the knower. 19 The categories of substance and causality are indispensable for experience because a fundamental identification of the objects of empirical knowledge would identify them as causally interacting material substances. 2o In contrast, moral 18 The concept of negation that is indispensable for experience is, roughly speaking, the logical concept. For Kant, however, negation is 'that the concept of which represents not-being (in time)' (A I43/B I82). This is not the concept of negation as modern logic understands it. This might help explain why, when it con1es to the issue of whether the categories can be realized in experience, Kant fails to recognize a fundamental asymmetry between substance and negation. 19 Cf. P. Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, I987), 340, and the opening paragraph of P. F. Strawson, 'Echoes of Kant', TLS, 4657 (I992), I2- I 3· 20 This is essentially the argument sketched in Strawson, The Bounds of Sense, 82-5, and again in Strawson, 'Kant's New Foundations of Metaphysics'. For further discussion of this argument see Q. Cassam, 'Mind, Knowledge and Reality: Themes from Kant', in A. O'Hear (ed.), Current Issues in Philosophy of Mind (Cambridge: Cambridge University Press, I998), 32I-48. My use of the expression 'fundamental identification' draws on G. Evans, The Varieties of Reference (Oxford: Clarendon Press, I982), I05-I2.
A Priori Concepts 95 concepts such as right and equity do not appear to be indispensable for rational thought or for empirical knowledge of material substances. Yet Kant lists right and equity, along with substance and cause, as concepts that are 'given a priori' (A 728/B 756).21 A way of accommodating the weak justificational apriority of moral concepts would be to broaden the definition of experience to include what might be called 'moral experience'. A way of accommodating the weak justificational apriority of the transcendental ideas would be to point out that although they are not constitutive of the possibility of experience, they have an 'indispensably necessary, regulative employment' (A 644/B 672) in relation to experience. 22 In this sense at least, they are indispensable for experience. These proposals illustrate the elasticity of the notion of 'indispensability for experience'. In doing so, they raise a more general question of principle. The question of principle is whether, given the range of concepts to which Kant applies the label 'a priori', it makes sense to try to account for the apriority of concepts as diverse as those of causality, God, and right in the same way. A better approach would be to limit the class of weakly justificationally a priori concepts to those that are strictly indispensable for empirical knowledge of objects, narrowly construed, while keeping open the possibility that concepts that are not justificationally a priori might be a priori in some other sense. For Kant, the core notion of apriority is not justificational but derivational apriority. His view is that derivational apriority is a necessary condition of justificational apriority, and that there are also derivationally a priori concepts that are not justificationally a priori in the sense in which the categories are justificationally a priori. In this way, he can allow that there are a priori concepts that are not constitutive of the possibility of experience. Thus, it would be a serious objection to I(ant's position if the notion of derivational apriority turns out to be dubious. The consequences for Strawson would be less serious, since he is primarily concerned with the justificational apriority of the categories rather than with their derivational apriority. Nevertheless, he does allow that the (weak) justificational apriority of a concept implies its derivational apriority. It would therefore be appropriate to take a closer look at the notion of derivational apriority, and at the relationship between derivational and justificational apriority.
21 The contrast is not just with a posteriori concepts but also with a priori concepts which are 'made' rather than 'given'. A concept is given 'in so far as it does not arise from my faculty of choice' (Kant, 'Vienna Logic', 356). A concept is 'made a priori when it is made through pure reflection, without the objects being given through experience' (ibid.). 22 For further discussion of the distinction between 'constitutive' and 'regulative' concepts see Friedman, 'Regulative and Constitutive'.
Cassam III
A concept is derivationally a priori if and only if it cannot be derived or acquired from experience. Concepts that can be derived from experience are derivationally empirical. What would it be for a concept to be 'derived' from experience? The answer to this question depends, in part, on what one thinks a concept is. On one view, to possess a concept is to posess a cluster of skills or capacities. This appears to be Strawson's view. 23 Kant's conception of concepts is more difficult to pin down. He characterizes concepts as 'perceptions'; that is, as conscious representations. This suggests that having a concept in Kant's sense is not just a matter of having certain capacities. As Bennett remarks, Kant's considered view is probably that 'having a concept is being in a mental state which endows one with certain abilities' .24 For present purposes, however, I will ignore the idea that concepts are perceptions in Kant's sense and will continue to assume that possession of the appropriate abilities is both necessary and sufficient for possession of a concept. Some concepts are recognitiona1. 25 To possess a recognitional concept C one must have the ability to recognize instances of C as instances of C. Not all concepts are recognitional in this sense. On the assumption that recognitional capacities are perceptual, neutrino is not a recognitional concept. Assuming that neutrinos are unobservable, possession of the concept of a neutrino cannot require the ability to recognize neutrinos as neutrinos at the level of observation. Neutrino is theoretical rather than recognitional, and grasp of this concept requires an understanding of the appropriate portion of physical theory. Recognitional concepts also have a theoretical component. Someone who has the concept square will typically have the ability to recognize square things as square, and grasp its relation to other shape concepts. In addition, mastery of shape concepts requires mastery of what Evans calls a 'prin1itive mechanics'.26 Although shapes are perceivable, this means that
23 See Strawson, Bounds of Sense, 20. In saying that to possess a concept is to possess a cluster of skills or capacities, Strawson wisely does not commit himself to the thesis that concepts are skills or capacities. For a defence of this thesis see P. Geach, Mental Acts (London: Routledge & Kegan Paul, 1957), 11-17, and ]. Bennett, Kant~s Analytic (Cambridge: Cambridge University Press, 1966), 53-6. For an entirely different approach to the nature of concepts see J. Fodor, Concepts: Where Cognitive Science Went Wrong (Oxford: Clarendon Press, 1998). According to Fodor, a concept is a kind of mental particular. Kant's position lies somewhere in between Fodor's position and Strawson's. 24 Bennett, Kanfs Analytic, 54. 25 For a contrary view see]. Fodor, 'There are No Recognitional Concepts-Not Even RED', in his In Critical Condition: Polemical Essays on Cognitive Science and the Philosophy of Mind (Cambridge, Mass.: MIT Press, 1998). 26 G. Evans, 'Things Without the Mind: A Commentary upon Chapter Two of Strawson's Individuals', in Z. van Straaten (ed.), Philosophical Subjects: Essays Presented to P. F. Strawson (Oxford: Clarendon Press, 19 80), 95.
A Priori Concepts 97 there is an analogy between the way in which our grasp of shape concepts rests upon 'implicit knowledge of a set of interconnected principles' and the way in which our grasp of a concept such as neutrino 'rests upon explicit knowledge of a set of propositions more familiarly regarded as a theory'.27 Once granted that to possess a concept is to possess a cluster of skills or capacities, it follows that acquiring a concept is a matter of acquiring a cluster of skills or capacities. 28 To acquire a recognitional concept from experience is to acquire a cluster of recognitional, inferential, and intellectual capacities from experience. On the present reading, a concept is derivationally a priori just if the capacities in which it consists cannot be acquired from experience. In the case of the categories, Strawson's view is that the capacities in which they consist cannot be derived from experience precisely because these concepts are necessary conditions of the possibility of empirical knowledge of objects. The point at issue between Strawson and Kant is whether, as Kant clain1s, the derivational apriority of the categories implies that they are wholly subjective in origin, or whether derivationally a priori concepts can reflect the structure of reality as it is in itself. What would it be to derive or acquire a concept fron1 experience? Suppose that C is a recognitional concept, and that to possess C one must therefore have the ability to recognize instances of C as instances of C. One way in which one might come to acquire this ability is by being shown ostensive samples of C. This will lead to the acquisition of C only if one extrapolates in the right way from the ostensive samples, and there is no guarantee that one will do this. On the other hand, if one does extrapolate in the right way from the ostensive san1ples, and thereby comes to possess C, it would not be implausible to describe this as a case of acquiring or deriving C from experience. The point of saying, in such a case, that C is acquired from experience is that the recognitional and other capacities in which possession of C consists are acquired as a result of experiential encounters with instances of this concept. A more difficult case is this: suppose that one acquires the ability to recognize and reason about tigers without ever having encountered a tiger. Is acquiring the concept tiger as a result of, say, hearing a zoologist speak on the subject of tigers a case of acquiring it 'from experience'? On the face of it, it would be difficult to deny that many concepts can be acquired in this way. Suppose, next, that acquiring a concept from someone else is, Evans, 'Things Without the Mind', 95-6. Jonathan Bennett claims that 'the question of where a given skill "comes from" is just not well-formed: we cannot answer Yes or No to the question "Does experience supply us with our ability to walk?", for example' (Kanfs Dialectic (Cambridge: Cambridge University Press, 1974), 35). For the reasons given below, this claim does not appear to be correct. Even if concepts are skills, there is nothing wrong with the idea that some concepts 'come from', that is 'are acquired from', experience. 27
28
Cassam broadly speaking, a case of acquiring it from experience. An immediate consequence of this supposition is that, in general, deriving concepts from experience does not require one actually to have encountered instances of them at the level of observation. This point applies to concepts that can be instantiated in experience and even more clearly to theoretical concepts that cannot be so instantiated. In the light of this proposal, why should anyone think that concepts such as those of substance, causality, and God cannot be derived from experience and are therefore derivationally a priori? A familiar argument is this: unlike the concept tiger, these concepts are incapable of being instantiated in experience and are therefore incapable of being acquired from experience. I will call this the simple argument for derivational apriority. Kant sometimes employs this argument in support of the derivational apriority of the category of causality. On the assumption that causation involves necessitation, and that experience cannot tell us what must necessarily be so, the concept of cause would be 'altogether lost' (B 5) if we attempted to derive it from experience. Since it cannot be derived from experience, it is derivationally a priori. Analagously, the derivational apriority of the concept of God is arguably a reflection of the fact that 'experience cannot teach us whether all possible perfections may be united somewhere'.29 What cannot be taught by ex:perience cannot be derived from experience. An objection to the simple argument is that it overlooks the possibility that some supposedly a priori concepts are complex concepts made up of simpler concepts that are instantiated in 'outer' or at least 'inner' experience. This is how Hume tries to account for the idea of God. For Hume, the idea of God is the idea of an 'infinitely intelligent, wise, and good being'. It 'arises from reflecting on the operations of our own mind, and augmenting, without limit, those qualities of goodness and wisdom'. 30 A further point is that, as argued above, Kant is prepared to allow that there is a sense in which the concepts of substance and cause are instantiated in experience. Indeed, they are necessarily instantiated in experience. To this extent, he is not entitled to the simple argument for their derivational apriority. Finally, even if a particular categorial concept is incapable of being instantiated in experience and is not made up of simpler concepts that are instantiated in experience, it still does not follow immediately that it cannot be derived from experience. At the very least, one would need to consider the possibility that the concept in question can be acquired from experience in the sense in which theoretical concepts can be acquired from experience. 98
Kant, 'Vienna Logic', 350. D. Hume, An Enquiry Concerning Human Understanding, in Enquiries Concerning Human Understanding and Concerning the Principles of Morals, ed. L. A. Selby-Bigge, 3rd edn. (Oxford: Clarendon Press, 1975), 19. 29
30
A Priori Concepts
99
It was suggested above that one sense in which theoretical concepts such as neutrino can be acquired from experience is that they can be acquired from someone else. Evidently, this cannot be the only route to the acquisition of such concepts, since the first person to have framed the concept of a neutrino did not acquire it from someone else. In any case, the Kantian categories are not like the concept of a neutrino. Apart from the fact that at least some of the categories are instantiated in experience, it is also open to question whether they can be acquired from someone else. Indeed, if it were . possible to acquire a category from someone else, then the stipulation that acquiring a concept in this way is acquiring it 'from experience' would force one to conclude that the categories can be acquired from experience and are therefore not derivationally a priori. One basis for thinking that the categories cannot be acquired from someone else would be a prior commitment to the view that they cannot be acquired at all. This is not how Kant sees things. Concepts that are possessed without having been acquired are innate, but Kant denies that the categories are innate. He claims that they are acquired concepts, but that their acquisition is an 'original acquisition'. The original acquisition of the categories is a matter of abstracting them from 'the laws inherent in the mind (by attending to its actions on the occasion of an experience)'.31 Abstracting a concept in this way is very different from acquiring it from someone else, but is not very different from Locke's conception of the means by which 'ideas of reflection' are acquired by the understanding when it 'turns its view inward upon itself, reflects on its own operations, and makes them the Object of its own Contemplation'.32 Yet, for Locke, allowing that sonle of our concepts or ideas can be acquired in this way is not inconsistent with his insistence that experience is the source of 'all the materials of Reason and Knowledge'. 33 For this reason, Kant's insistence that the acquisition of the categories is an original acquisition does not, on its own, make it plausible that these concepts are not, and cannot be, acquired from experience. In the light of these considerations, together with the failure of the simple argument, what is needed is an argument for the derivational apriority of the categories that does not ignore the point that, as Kant himself recognizes, concepts like substance and cause are at least indirectly instantiated in experience. Consider, once again, the point that substances are given in experience in virtue of the fact that material substances or material objects
31 I. Kant, 'On the Form and Principles of the Sensible and Intelligible World (1770)', [Inaugural Dissertation] in D. Walford and R. Meerbote (eds.), Immanuel Kant: Theoretical Philosophy 1755-70 (Cambridge: Cambridge University Press, 1992),387-8. 32 ]. Locke, An Essay Concerning Human Understanding, ed. P. H. Nidditch (Oxford: Clarendon Press, 1975), 105. 33 Ibid. 104.
100
Cassam
are given in experience. The question that this raises is whether the concept of a ll1aterial object call be derived from experience. If so, then one would need to consider the possibility that the pure concept of substance can be derived from experience via the derivation froll1 experience of the concept of a material object. If, on the other hand, the concept of a material object cannot be derived from experience, even though material objects are given in experience, this would strengthen the case for thinking that the category of substance cannot be derived from experience. The initial challenge, then, is to explain why one il1ight think that the concept of a material object cannot be derived from experience, despite the fact that material objects are .. . gIven In experIence. To say that material objects are given in experience, or that the concept material object is instantiated in experience, is not just to n1ake the point that we frequently experience what are in fact material objects. It is to draw attention to the fact that in experience we are typically aware of material objects as material objects. 34 What makes an object a material object is its possession of primary qualities or primary properties. To be perceptually aware of an object as a material object is therefore to be perceptually aware of it as having primary properties. Primary properties are 'theoretical' in so far as our grasp of them relies upon implicit knowledge of the propositions of a primitive or 'intuitive' mechanics. By the same token, experiencing an object as having primary properties is, at least in part, a matter of experiencing it as something that is subject to certain mechanical principles or laws. 35 The question that now arises is whether these principles can be derived from experience. In his paper 'Things Without the Mind' Gareth Evans puts forward an argument that apparently supports the conclusion that the primitive mechanics that provides the indispensable surrounding for the concept of a material object cannot be derived from experience. In arguing for this conclusion, Evans does not deny that primary properties are sensible or observable, for 'we are obviously able, after the appropriate training, to perceive the shape, motion, and hardness of things'. 36 In this sense, concepts of primary properties are instantiated in experience. Why, then, is it not possible to derive such concepts from experiential encounters with their instances? Evans's answer to this question turns, in part, on his conception of the nature of the interconnected principles that form the basis of
34 As C. Peacocke, puts it, in 'Intuitive Mechanics, Psychological Reality and the Idea of a Material Object', in N. Eilan, R. McCarthy, and B. Brewer (eds.), Spatial Representation: Problems in Philosophy and Psychology (Oxford: Basil Blackwell, 1993), 169: 'we experience objects specifically as material objects'. 35 This is one sense in which there is a 'constitutive link between the capacity to perceive objects as material objects and possession of an intuitive mechanics' (ibid. 174). 36 Evans, 'Things Without the Mind', 96.
A Priori Concepts
101
concepts of primary properties. In order to grasp these properties, one nlust grasp the idea of a unitary spatial framework in which both oneself and the bodies of which one has experience have a place, and through which they move continuously. One must learn of the conservation of matter in different shapes, of the identity of matter perceived from different points of view and through different modalities, and of the persistence of matter through gaps in observation. One must learn how bodies compete for the occupancy of positions in space, and of the resistance one body may afford to the motion of another. And so on. 37
Suppose, against this background, that hardness is taken to be a primary property. Evans claims that it is a consequence of his characterization of the primitive mechanics that the concept of hardness cannot be distilled 'solely out of the experiences produced by deformation of the skin which is brought into contact with a hard object, for it is not possible to distil out of such an experience the theory into which the concept fits'. 38 More generally, the lesson is that our conception of the material world is 'neither directly nor exclusively, woven out of materials given in experience'.39 Although Evans represents his argument as lending support to the view that the primitive mechanics is innate or natively given,4o it can easily be seen as an argument for the derivational apriority of the primitive nlechanics, and therefore for the derivational apriority of the concept of a material object. From the thesis that the concept of a material object cannot be distilled out of n1aterials given in experience, it is only a short step to the conclusion that the category of substance cannot be derived fronl experience, and is therefore derivationally a priori. To distinguish it from the simple argument, I will refer to this argument from the derivational apriority of the primitive mechanics to the derivational apriority of the concept of a material object, and from there to the derivational apriority of the category of substance, as the complex argument. An objection to the complex argument is that it fails to attach sufficient weight to its own insistence that concepts of primary qualities are instantiated in experience. If it were true that our conception of the material world as a world of space-occupying continuants goes beyond what is literally given in experience, then it would be plausible that the primitive mechanics is incapable of being directly or exclusively woven out of what is given in this way. Once it is agreed, however, that in experience we are typically aware of objects as nlaterial objects, then it might be held that there is no longer any reason to suppose that the principles that make up our elementary theory of bodies cannot be read off from experience. The complex Ibid. 95. 38 Ibid. 96. 39 Ibid. 97. It is Norman Daniels who describes our elementary theory of bodies as a 'natively given primitive mechanics', but the passage in which this characterization appears is quoted approvingly by Evans. See 'Things Without the Mind', 97-8. 37
40
102
Cassam
argument's defence of the view that the primitive mechanics is derivationally a priori only makes sense on the assumption that it is not experience itself that teaches us that bodies persist through gaps in observation, compete for the occupancy of positions in space, and so on, but this assumption is more in keeping with the simple than with the complex argument. A further consideration is this: despite the fact that Evans apparently endorses some form of nativism with respect to the primitive mechanics, he also writes of the need to learn of the identity of matter perceived from different points of view and of the resistance one body may afford to the motion of another. How does one learn these things? The natural answer to this question is that one learns such things by observing and interacting with material objects. Principles that are learned are also principles that are acquired, and acquired principles are not innate. 41 What is more, learnirig how bodies behave by interacting with them and observing their behaviour is surely a case of acquiring knowledge from experience. If the primitive mechanics is acquired from experience, then it is not derivationally a priori. The complex argument can defend itself against these objections in the following way: it can agree that in experience we are typically aware of objects as material objects, but deny that it follows from this that our elementary theory of bodies can be read off or extracted from experience. This does not follow for two reasons. First, our experience can only present itself to us as experience of material objects because we already have implicit knowledge of the interconnected principles of a primitive mechanics. Grasp of such principles, together with the concept of a material object, is a precondition or presupposition of experience of objects as persisting spaceoccupiers. I will call this the presupposition thesis. Secondly, a concept or principle can only be said to derive from experience of a certain type if experience of that type is possible apart from that concept or principle. I will call this the independence thesis. Since, according to the presupposition thesis, experience of objects as material presupposes our possession of the concept of a material object, the independence thesis implies that this concept cannot be derived from experience. Given the presupposition and independence theses, there is no need for the complex argument to backtrack on its initial insistence that material objects are given as such in experience. It can therefore avoid having to say that we infer that bodies persist through gaps in observation, or that the concept of a material object is a theoretical concept that cannot be directly instantiated in experience. It is experience itself that teaches us that bodies persist through gaps in observation and compete for the occupancy of positions in space, but the experience that teaches us these things is already 41 Following Bennett, I take it that innate means 'possessed but not acquired' (see Kant's Analytic, 98).
A Priori Concepts
103
objects. 42
thoroughly permeated with concepts of material It is unconceptualized or uninterpreted experience in which the concept of a spaceoccupying continuant cannot be instantiated as such. It is no use responding to this point by trying to represent this concept as originating in conceptualized experience, since conceptualized experience is only possible because the concept of a space-occupying continuant is already in place. What Kant says about the category of causality can be given an explanation along similar lines. The question is this: how can Kant maintain both that this category is necessarily instantiated in experience and that experience cannot tell us what must necessarily be so? In order to answer this question it needs to be acknowledged that the concept of necessity is closely related to that of force, and that there is such a thing as the experience of force. 43 Experiences of force might be described as experiences of necessity, but experiences of this type are possible only because we already have the concept of force. To have the concept of force is to have the capacity to conceive of the occurrence of mechanical transactions-pushings and pullings-in whi'ch we ourselves are not involved as agents or patients. 44 We can only conceive of the occurrence of such transactions because our experiences of the mechanical transactions in which we ourselves are involved are informed by the conception of ourselves, as well as the bodies with which we interact, as subject to 'an intuitive mechanics mentioning forces' .45 This intuitive mechanics is closely related to the intuitive mechanics that enables us to think of the objects of perception as material substances. Since our experiences of force presuppose our possession of an intuitive mechanics mentioning forces, neither the intuitive mechanics nor the concept of force that it makes intelligible can correctly be said to be derived from these experiences. These considerations suggest that the acceptability, or otherwise, of the complex argument is largely a matter of whether the presupposition and independence theses are correct. An objection to the presupposition thesis is that infants and some non-human animals experience objects as material objects but do not have the concept of a material object, so it is false that this concept is a presupposition of the experience of objects as material, or of what I will refer to from now on as 'M-experience'. A familiar response to this objection is to agree that the 'looking preferences' and 'reaching patterns' of infants indicate that they are capable of 'representing' the continuity and
42 For a defence of the idea that sensible experience is permeated with concepts of realistically conceived objects see P. F. Strawson, 'Perception and its Objects', in G. F. Macdonald (ed.), Perception and Identity: Essays Presented to A. J. Ayer (London: Macmillan, 1979). 43 Cf. Strawson, Analysis and Metaphysics, 117, and Peacocke, 'Intuitive Mechanics', 172. 44 Strawson makes this point in his discussion of the idea of force, Analysis and Metaphysics, 118. 45 Peacocke, 'Intuitive Mechanics', 172.
Cassam solidity of objects in their environment, but to question the claim that the representations that can correctly be attributed to infants amount to M -experiences. 46 Why should it be supposed that infants are incapable of M-experience? According to what I will call the extreme view, the reason is that the content of experience is conceptual. 47 In other words, conceptual capacities are operative in anything that deserves to be called 'experience'. On the assumption that infants lack such capacities it follows that they do not have experience, and so do not have M-experience. According to what I will call the moderate view, the content of conscious experience is not itself conceptual. One can experience something as being a particular shade of red even if one lacks the concept of that particular shade, but it is required for conscious experience that 'the subject exercise some concepts,.48 Like the extreme view, the moderate view implies that non-concept-exercising creatures cannot properly be said to have experience, and so cannot properly be said to have M-experience. The moderate view is not committed to denying that a creature with some concepts can have M-experience even though it lacks the concept of a material object. For this reason, the moderate view does not, on its own, lend support to the presupposition thesis. The extent to which the extreme view supports the presupposition thesis depends on whether the conceptual capacities that are operative in M-experience necessarily include the capacity to think of objects as material objects. On the face of it, the thesis that M-experience is thoroughly conceptual leaves it open whether the concepts with which M-experience is permeated must include the concept of a material object. I will come back to this issue. With regard to the independence thesis, suppose that one were to claim that in order to experience something as having a particular shade of colour, the concept of that particular shade must be operative in the experience itself. How, in that case, can colour concepts be acquired from colour experience, given that colour experience presupposes one's possession of colour concepts? One proposal is this: we 'do not have to have ready, in advance of the course our colour experience actually takes, as many colour concepts as there are shades of colour that we can sensibly discriminate'.49 Rather, the capacity to embrace a particular shade of colour within one's conceptual thinking 'sets in with the experience'. 50 In this way, one can agree that 104
46 For a defence of the view that infants are capable of representing objects see E. S. Spe1ke and G. A. Van der Valle, 'Perceiving and Reasoning About Objects: Insights from Infants', in Eilan, McCarthy, and Brewer (eds.), Spatial Representation. 47 For a defence of this view see]. McDowell, Mind and World (Cambridge, Mass.: Harvard University Press, 1994). 48 Evans, The Varieties of Reference, 157. 49 McDowell, Mind and World, 58. 50 Ibid. 57.
A Priori Concepts
105
it is possible to acquire the concept of a particular shade of colour from the appropriate colour experience while continuing to maintain that the concept is operative in the very experience from which it is acquired. Why, by the same token, should it not be possible to acquire the concept of a material object from experiences in which that very concept is operative, and which therefore presuppose one's possession of a primitive mechanics? The suggestion, in other words, is that even if the presupposition thesis is correct, and M-experience would not be possible without the concept of a material object, it still does not follow that this concept cannot be acquired from M-experience. In effect, this suggestion grants the presupposition thesis but disputes the independence thesis. This suggestion is difficult to assess since I have so far not said what the positive argument in favour of the independence thesis is supposed to be. One influential proposal is this: if C is a concept that is acquired from experience of a certain type, then it follows that experience of that type is possible prior to one's acquisition of C. If experience of the type in question is possible prior to one's acquisition of C, then experience of that type is possible apart from, or independently of, C. That is why it is only possible to acquire a concept or principle from experience if experience does not presuppose that concept or principle. 51 The problem with this argument is that concepts that are derived from experience need not be temporally posterior to experience, any more than, by Kant's lights, a priori concepts need to be temporally prior to experience. Concepts that derive from experience are concepts that are made available by experience, but, to return to an earlier example, the fact that colour concepts are made available by colour experience does not imply that colour experience was possible before we had any colour concepts. 52 A conceptual capacity that sets in at the same time as experience of a certain type can quite properly be said to originate in experience of that type, despite the fact that neither the experience nor the conceptual capacity is temporally prior. Just as a metal can be extracted from an ore that already contains it, so a concept can be extracted from a type of experience in which it is already operative. In the light of these considerations, it can be seen that the complex argun1ent for the derivational apriority of the concept of a material object is inconclusive, since there are serious objections to the independence and presupposition theses. Crucially, in claiming that experience is the source of the concept of a material object, one is not committed to the view that conscious experience is possible before one acquires this concept, and so is
51 This argument for the independence thesis is essentially the argument given in Allison, The Kant-Eberhard Controversy, 79-80. 52 There might, of course, be other reasons for allowing that colour experience was possible before we had any colour concepts.
Cassam not con1n1itted to disputing the claim that grasp of this concept is necessary for conscious experience. Without the independence thesis, the justificational apriority of a concept does not entail its derivational apriority, and it is doubtful whether the independence thesis is correct. Before leaving the complex argument behind, there is one more issue to consider. Suppose that the concept of a material object is not derivationally a priori. Would it be right to conclude that the pure concept of substance is not derivationally a priori? On the face of it, grasp of the concept of that which can exist as subject and never as mere predicate has very little to do with one's grasp of the interconnected principles of a primitive mechanics. Consequently, it might be argued that the logical idea of substance can be derivationally a priori even if the concept of a material object is derived from within experience. For it would surely be no more plausible to suppose that the logical idea of substance can be acquired from experience than it would be to suppose that the concept of negation can be acquired from experience. A way of meeting this objection would be to show that the unschematized category of substance is an abstraction from the notion of n1aterial substance. If one concept C 1 is only an abstraction from another concept C2 , then the possibility of deriving C2 from experience weakens the case for classifying C 1 as derivationally a priori. To this extent, the fact that the concept of a material object can be derived from experience cannot fail to have some bearing on the question of whether the pure concept of substance is derivationally a priori. Nor, given the connections between the concepts of substance and cause, can it fail to have a bearing on the derivational apriority, or otherwise, of the concept of cause, though it would take much more argument than I have given here to establish that this concept is not derivationally a priori. Unlike the concept of substance, the concept of negation is not an abstraction froln any other concept, and it cannot be acquired from experience, either directly or indirectly. One cannot be shown ostensive samples of negation and it is arguable that one cannot acquire this concept from somebody else. If it is appropriate to speak of the concept of negation, then it is plausible that this concept is both justificationally and derivationally a priori. Since the thesis that the concept of negation is derivationally a priori is neither unintelligible nor implausible, the objection to the view that the concepts of substance and cause are derivationally a priori is not that there is something wrong with the very idea of derivational apriority. The objection is simply that the arguments for the derivational apriority of these concepts are not good ones. 53 106
53 The argument of this paragraph assumes that the concept of negation is to be understood in the way that modern logic understands it. As remarked above, Kant understands the concept of negation somewhat differently.
A Priori Concepts
1°7
IV
Kant argues from the derivational apriority of the categories to their subjectivity. Strawson agrees that the categories are derivationally a priori but denies that they are wholly subjective in origin. I began by saying that the transcendental epistemologist should drop the claim that the categories are derivationally a priori. It now appears that this is not quite right, since negation is one of Kant's categories and is derivationally a priori. Does it follow that the concept of negation is wholly subjective in origin? That depends on what it means to describe a concept as part of the cognitive machinery of the mind, and whether regarding a concept in this way is the only alternative to regarding it as one that derives from experience. These are interesting questions, but they do not affect the concepts of substance and cause. These are the a priori concepts with which Kant is primarily concerned, and I have suggested the thesis that they are derivationally a priori is far more problematic than is commonly supposed. The question of whether their derivational apriority implies their subjectivity therefore does not arise. For the purposes of transcendental epistemology, it ought to be enough if one can show that they are weakly justificationally a pnon. Would one be right in thinking that substance and cause are weakly justificationally a priori concepts? The weak justificational apriority of a concept consists in its indispensability for experience, where experience is understood to be, or to involve, empirical knowledge of objects. What was described above as the best case for the weak justificational apriority of the concept of substance relies on two principles or assumptions, both of which are questionable from the standpoint of transcendental idealism. The first is the principle that conditions of knowledge must reflect the nature of the objects of knowledge, as well as the cognitive faculties of the knower. In Strawson's words, 'knowledge must be subject to its objects, epistemology to metaphysics'. 54 The second is the assumption that the objects to which our empirical knowledge is subject are, in themselves, material substances. Given that empirical knowledge requires both intuitions and concepts, it follows that the intuition required for empirical knowledge of material substances is spatio-temporal intuition, and that the concepts that are a presupposition of such knowledge must include the concept of material substance. 55 In other words, the concepts with which M-experience is permeated must include the The objects that are in question here are objects as they really are or are in themselves. For Kant, empirical knowledge requires the making of judgements, but, as Strawson remarks, if our judgements 'are to be in general true, then the concepts employed in those judgements must in general be concepts of kinds of things which actually are in the world and of properties which those things actually have' (Analysis and Metaphysics, 63). 54
55
108
Cassam
concept of a material object because without this concept M-experience would not constitute knowledge of material reality. For Kant, the fact that the objects to which our empirical knowledge pertains are material is a reflection of the fact that space and time are the forms of our sensibility. If 'objects must conform to our knowledge' (B xvi), it makes no sense to suppose that the conditions of knowledge or M-experience reflect anything other than our own cognitive constitution. 56 This shows that what I have been calling the best case for the weak justificational apriority of the category of substance would not be acceptable by Kant's lights. It 'Yould also be unacceptable by the lights of someone who has no prior commitn1ent to transcendental idealism, but who wonders why conscious experience must be M-experience. As far as establishing the weak justificational apriority of the concept of substance is concerned, therefore, there is clearly more work to be done. To base an argun1ent for the weak justificational apriority of a concept on the transcendental-realist principle that knowledge must be subject to its objects is already to be committed to rejecting the thesis that the putatively a priori concepts are wholly subjective in origin. However, once it is agreed that weakly justificationally a priori concepts are not subjective, it is difficult to see why a transcendental epistemologist with realist leanings should still want to insist that they are not derived from within experience. Quite apart from the issue of whether there is any alternative to regarding derivationally a priori concepts as originating in our own cognitive faculties, weak justificational apriority does not entail derivational apriority. It is one thing to respond to Kant, as Strawson does, by attempting to construct transcendental arguments for the justificational apriority of the categories that are independent of any commitment to transcendental idealism, but if one is really interested in disentangling transcendental arguments from transcendental idealism, what needs to be given up is not just the idea that a priori concepts are wholly subjective in origin but also the idea that experience cannot be the source of such concepts. 57
56 In Kant's Transcendental Idealism: An Interpretation and Defense (New Haven, Conn.: Yale University Press, 1983), Henry Allison claims that Kant's idealism relies on the principle that 'whatever is required for the recognition or picking out of what is "objective" in our experience, must reflect the cognitive structure of the mind (its manner of representing) rather than the nature of the object as it is in itself'. (p. 27). 57 I thank Hagit Benbaji, John Campbell, Beatrice Longuenesse, and Wayne Waxman for helpful comments on an earlier version of this paper. This paper was written during a period of specialresearch leave funded by the Arts and Humanities Research Board. I am grateful to the AHRB for its support.
6 The Synthetic A Priori in Strawson~s Kantianism BARRY STROUD
••
Kant's 'general problem of pure reason' (B 19)1 was to be solved by following 'a special science' called 'the critique of pure reason' (A I liB 24). It would examine the 'sources and limits' (A I2/B 25) of reason's power, and 'lay down the ... plan' of what he called a 'transcendental philosophy' (A 13/B 27). Ideally, such a philosophy would account for all of human knowledge in so far as it is possible a priori, including even our knowledge of what is already contained in our concepts. But Kant's Critique of Pure Reason-the book by that name-would not investigate such merely 'analytic' knowledge. It would go only as far as is needed for the examination of 'knowledge which is a priori and synthetic' (A 14/B 28). So the whole project, and the whole question to which Kant's masterpiece is devoted, is 'How are a priori synthetic judgements possible?' (B 19). Peter Strawson says that 'Kant really has no clear and general conception of the synthetic a priori at all,.2 If that is true, what becomes of the general problem of pure reason? What then of transcendental philosophy? Does this mean that Strawson thinks Kant has no clear and general conception of the task of his Critique of Pure Reason, or of the problem he is trying to solve there? Not necessarily, of course. It might mean only that Strawson thinks the real problems Kant deals with so masterfully are not best expressed, or perhaps even fully intelligible, in that form. Readers of The Bounds of Sense are familiar with this general idea. We are urged again and again to look through the surface features of Kant's often confused and sometimes grandiose presentation of the issues to what is really at stake behind them. It is like being taken round a baroque cathedral by a wise cicerone whose practised eye and seasoned intelligence see right
1 2
I. Kant, Critique of Pure Reason, trans. N. K. Smith (London: Macmillan, 1953). P. F. Strawson, The Bounds of Sense (London: Methuen, 1966),43.
Stroud
110
through all those impressive twisted columns and grimacing giants to discern what is really carrying the weight and keeping the whole improbable thing upright. What then is the austere, unembellished truth about what is going on in the Critique of Pure Reason? Strawson sees its programme as that of 'determining the fundamental general structure of any conception of experience such as we can make intelligible to ourselves'. 3 That too was the project of the first half of his own earlier Individuals, before he turned his attention more explicitly to the work of Kant. Objective particulars in space and time, and human persons with both physical and psychological characteristics, were found to playa central role in our conception of the world, just as they do in Strawson's account of Kant's achievement. But the task of the special Strawsonian or Kantian enterprise is not captured simply by observing that we think of the world in terms of relatively permanent spatia-temporal particulars and embodied persons as subjects of experience. That is true, but it is just a very general remark about us-about how we think of the world. Nor does it get any closer to say that there are relatively permanent particulars and en1bodied persons. That too is true, but it is just a very general remark about the way the world is. This special investigation of our conception of the world achieves its goal not sin1ply by reaching such results, but only by our coming to see that those propositions have a certain special and 'distinctive character or status,4 in our thought. Kant uses the term 'synthetic a priori' to capture what he thinks is that distinctive character. But if the point of the enterprise is to discover the distinctive character or status of certain components of our experience of the world, and Kant calls that distinctive status 'synthetic a priori', what is the point of his asking how synthetic a priori judgements-or propositions that have that distinctive character-are possible? If such propositions must be present in any thought we can form, that amounts to asking how thought and experience are possible. But for Kant that distinctive character involves those propositions' being known, or knowable, in a certain way. So his question is also a question of how it is possible for an investigation into our conception of the world to yield conclusions that can be known to have such a distinctive character or status. It is not enough for something to have that distinctive status; it must be knowable or discoverable that certain things have it. Kant thought that if there are discoverable results with the distinctive status he had in mind, then metaphysics-the attempt to discover general truths about the world-would no longer be 'a merely random groping' (B xv). It could be set on 'the secure path of a science' (B xix). It would be
3
P. F. Strawson, The Bounds of Sense (London: Methuen:
19 66),
44.
4
Ibid.
Synthetic A Priori in
Strawson~s
Kantianism
I I I
secure because there would be an explanation of how undeniable metaphysical results can be reached. Kant's explanation of the possibility of that knowledge-what he thought was the only possible explanation-was transcendental idealism. We know that was his explanation, even if we don't know what transcendental idealisn1 actually is, or says. To the extent to which we can understand it, it seems to leave us with the sinking feeling that the knowledge it credits us with is not really knowledge of the kind of world we thought we knew before we took up the critical enterprise. But, to put it most gently-more gently than Strawson sometimes does-transcendental idealism can add nothing positive to our understanding of whatever position we are in. It is partly for these reasons that Strawson thinks there is 'nothing to demand, or permit, an explanation such as Kant's's in any results that n1ight be reached in an investigation of the fundamental general structure of our conception of the world of experience. There are two questions here: whether the possibility of discovering the distinctive character or status of certain components of our conception of the world demands or permits an explanation such as Kant's, and whether it demands or permits an explanation at all. What the explanation might be depends on what the distinctive character or status of those special propositions is. Strawson does not deny that they have some special or distinctive character. That is crucial to his own understanding of the project. 6 So the question is what austere, unembellished description he would give of propositions of the kind that Kant called 'synthetic a priori', and whether the possibility of propositions with that character demands, or permits, explanation. Strawson at times appears to doubt or deny that any explanation at all is needed. He thinks that in exploring the fundamental general structure of our conception of the world of experience 'it is no matter for wonder' if we come up against certain 'necessary limits' to any such conception we can make intelligible to ourselves. 7 But: In order to set limits to coherent thinking, it is not necessary, as Kant, in spite of his disclain1ers, attempted to do, to think both sides of those limits. It is enough to think up to them. 8
Is that really enough? It is of course not possible to think beyond the limits of coherent thinking. Even when we think right up to those limits, or think something that expresses a necessary condition of thinking anything, we are Strawson, Bounds of Sense, 44. 'If there is such a structure [of any conception of experience which we could make truly intelligible to ourselves], if there is a set of ideas which enter indispensably into such a structure, then the members of this set will surely have a distinctive status' (Strawson, Bounds of Sense, 49-50). 7 Strawson, Bounds of Sense, 44. 8 Ibid. 5
6
S~oud
112
still within those limits. To think up to what is in fact a limit of thought is, in a way, easy to do, in fact in1possible to fail at. But the point of the Kantian or 5trawsonian philosophical enterprise is not sin1ply to think up to those limits, but to recognize those limits, or to identify some of the necessary conditions of the possibility of thought and experience. We must come to see, of some of the things we think, that they are necessary conditions of the possibility of our having any thoughts or experiences at all. And how such discoveries are possible, if they are, is something that might call for explanation even if we are never tempted by the 'doctrinal fantasies'9 of transcendental idealism. Kant does not simply start off with his transcendental explanation. A number of steps lead up to the felt need for it. First, he seeks some necessary conditions of the possibility of thought and experience in general. Next, he thinks he finds something which holds necessarily, and which he sees to hold necessarily. Third, he infers that his knowledge of such necessities is a priori. By 'a priori' here he can be taken to mean nothing more than non-empirical. As he says in the introduction: 'we shall understand by a priori knowledge, not knowledge independent of this or that experience, but knowledge absolutely independent of all experience' (B 2). And 'independent' means that the knowledge does not 'arise out of' experience, not that it does not 'begin with' experience (B I). Fourth, Kant makes an inference to the a priori on the basis of the general principle that necessity is what he calls a 'sure criterion' of the a priori. 'Experience teaches us that a thing is so and so, but not that it cannot be otherwise', he says. So 'if we have a proposition which in being thought is thought as necessary, it is an a priori judgment' (B 3). 50 any discoveries we make of the necessary conditions of the possibility of thought and experience in general must be made and so known a priori, independently of all experience, according to Kant. So far, there appears to be no appeal to, or even mention of, transcendental idealism. Even the next, and admittedly n10re problematic, step seems free of tran.;. scendental idealism. Kant holds that the necessities he discovers, or the way he discovers them-at least those that are central to the Critique of Pure Reason-cannot be accounted for as merely 'analytic'. They do not serve simply to express what is somehow covertly contained within the very concepts they make use of. They are in that sense 'ampliative' or 'synthetic'. This is of course obscure. 'Synthetic' just means 'non-analytic', so our understanding of the idea of the synthetic a priori can be no better than our understanding of the notion of analyticity. And our understanding of the notion of analyticity is not good. It seems to rest on the notion of one
9
Strawson, Bounds of Sense, 51.
Synthetic A Priori in Strawson's Kantianism
I
13
concept or meaning somehow 'containing' or not 'containing' another. Kant's suggestion that an 'analytic' judgement is one in which one concept is 'extracted' from another 'in accordance with the principle of contradiction' (B 12) would help if we had some independent grasp of the notion of a contradiction. But it tells us at least that the distinctive, non-analytic propositions that he is concerned with are not such that their negations are contradictory. For now, it is enough to observe that the admitted obscurities surrounding all these notions appear to have nothing to do with transcendental idealism. That doctrine begins to n1ake its appearance only at the final step, when Kant asks how it is possible for anyone to know a priori that such nonanalytic necessities hold-how synthetic a priori judgements are possible. This is where the secret operations of 'our faculties', and what is 'in us' and not in the world, and all the other incoherences of what Strawson calls 'transcendental subjectivism' are appealed to. A priori knowledge of the fundamental structure of the world of experience is possible only because that world must conform to the constitution of our minds. If that were not so, we could at most have a posteriori knowledge of what its structure is, and not what Kant thought we do have-a priori knowledge of what its structure must be. But it is possible to have such knowledge, he thought, only if it is somehow knowledge of us, or of what 'we' bring to the 'raw material' of 'sensible impressions' to make up 'that knowledge of objects which is entitled experience' (B 1).10 A priori knowledge is knowledge that arises out of something called 'reason', and 'reason has insight only into that which it produces after a plan of its own' (B xiii). These 'fantasies' and excrescences of transcendental idealism are put forward by Kant as the only possible explanation of something he thinks he has independently identified. When Strawson says, as I quoted him at the beginning as saying, that 'Kant really has no clear and general conception of the synthetic a priori at all', he is making an observation about what he takes to be the incoherence and explanatory emptiness of transcendental idealism. It is because transcendental idealism can explain nothing that he finds Kant's question about the synthetic a priori unrewardingly unclear, and so to be set aside. But how closely are the two connected? Kant's answer to the question might be unclear or incoherent even though there is
10 The connection between synthetic a priori knowledge and the subjective source of that knowledge is well expressed by Kant in a single rich sentence, also appropriately quoted by Strawson (Bounds of Sense, I 15 n.): 'For this unity of nature has to be a necessary one, that is, has to be an a priori certain unity of the connection of appearances; and such synthetic unity could not be established a priori if there were not subjective grounds of such unity contained a priori in the original cognitive powers of our mind, and if these subjective conditions, inasmuch as they are the grounds of the possibility of knowing any object whatsover in experience, were not at the same time objectively valid' (A 125-6).
S~oud
114
something real, and puzzling, in the Kantian project that calls for explanation. What of the earlier steps leading up to the Kantian explanation? First, it is clear that Strawson seeks and thinks he finds, in Kant and in his own investigations, 'statable necessary conditions of the possibility of experience in general',11 or 'what is necessarily involved'12 in any conception of experience that we can make intelligible to ourselves. Without asking by what specific arguments conclusions like this are supported, we can ask whether for Strawson, as for Kant, our knowledge of such necessities is a priori, or independent of all experience. The question is difficult to answer, since Strawson gives a very austere interpretation of the expression 'a priori'. In fact, in his revulsion from any transcendental-idealist understanding of it as standing for knowledge that has a subjective source in the nature of our cognitive constitution, he seems to go so far as to understand it as not really an epistemic term for a kind of knowledge at all. An element of our conception of the world can be called 'a priori' for Strawson if it is what he calls 'an essential structural element in any conception of experience which we could make intelligible to ourselves'.1 3 It can be contrasted with those less general, and so dispensable, features of our experience that at least in principle could be, or could have been, abandoned 'without imperilling the entire structure of the conception'.14 Those features could be called, in a correspondingly austere sense, merely 'empirical'. With the so-called a priori features, 'we can form no coherent' conception of how experience might have been without them. 1s This virtually identifies the 'a priori' features of our conception of the world with the necessary or absolutely indispensable features of it. It is not as if the necessity of what is known serves as a 'criterion' of its being known in a certain way, as in Kant. Something's being necessary or indispensable to any conception of experience implies nothing about how it is known, or knowable, or how it is known to have that status. This would leave untouched Kant's question of how a 'scientific', or generally reliable, metaphysics is possible. There might be felt to be no problem-or no need for what Strawson calls 'any high doctrine'16-here at all, beyond the completely general question of how we know the things we think we know. Since necessary truths, or necessary connections between things, are among the things we know, no doubt our knowledege of them must be accounted for in some way. But the notion of 'independence from all experience', which is essential to an epistemic reading of 'a priori', is too difficult to apply with any confidence, or too laden with dubious epistemological assumptions, to be of much help. But if we abandon an epistemic notion of the a priori altogether, and with it any general distinction between 'priori' and 'empirical' ways of knowing 11
14
Strawson, Bounds of Sense, 120. Ibid. 50. 15 Ibid. I I 5.
16
12 Ibid. 121. Ibid. 44.
13
Ibid. 68.
Synthetic A Priori in
Strawson~s
Kantianism
I I
5
things, the features of the Strawsonian or Kantian conclusions that might seem to call for explanation would concern only what Kant called their 'synthetic', rather than their 'a priori', character. In a fan1iliar but surprising passage, Kant says that of all earlier philosophers it was Hume who came closest to seeing the problem of how synthetic a priori judgements are possible. Closest, but no cigar, even for Hume. He did not see the generality of the problem. Or, to put Kant's point in another way, if we generalized to all propositions what Hume said in particular about the 'synthetic' proposition that every event has a cause, 'then all that we call metaphysics [would be] a mere delusion whereby we fancy ourselves to have rational insight into what, in actual fact, is borrowed solely from experience, and under the influence of custom has taken the illusory semblance of necessity' (B 20). Metaphysical results of the kind Kant thought could be reached would be unavailable, since the propositions in question would not really be necessary after all. That is why he thinks that if Hume had seen the universality of the problen1 'his good sense' would have saved him from the conclusion he reached in the case of causality. Kant's reasons for this generous assessment of Hume show that what he thinks is really important is the synthetic character of the necessary propositions he seeks. Of course, for Kant the necessity carries with it the implication that the propositions are known a priori, but it is not on the question of the a priori that he differs at this point from Hume. He is confident that I-Iume would have seen that he could not say the same thing about the necessary propositions of pure mathematics as he said about the proposition that every event has a cause. That would make pure mathematics impossible. I take !(ant to mean by this that the propositions of pure mathematics would not really be necessary after all, but would have only 'the illusory semblance of necessity', borrowed from experience under the influence of custom. That is more or less the view from which John Stuart Mill was not saved by whatever good sense he had. But it is not a view by which Hume was even tempted. 17 He thought the propositions of pure mathematics are necessarily true because they express no n10re than 'relations of ideas', so their negations are contradictory. He thought that was enough to explain how we know them too, but it cannot be said that he ever gave such an explanation. So he would not disagree with Kant about their being necessary, but about their being what Kant called 'synthetic'. Neither Hume nor Kant
17 Or not quite. There is one passage in which he shows that a similar thought has occurred to him. 'Thus as the necessity, which makes two times two equal to four, or three angles of a triangle equal to two right ones, lies only in the act of the understanding, by which we consider and compare these ideas; in like manner the necessity or power, which unites causes and effects, lies in the determination of the mind to pass from the one to the other' (D. Hume, A Treatise of Human Nature, ed. L. A. Selby-Bigge (Oxford: Oxford University Press, 1958), 166).
S~oud
116
doubts that the proposition 'Every event has a cause' is synthetic. But Hume held that the propositions of pure mathen1atics are not. So the challenge represented by Hume is to explain how it is to be known that something really does hold with necessity, and not just with 'the illusory semblance of necessity', when neither its truth nor its necessity is discoverable 'analytically', by inspection of the contents of concepts alone. This is Kant's question of how non-'analytic' and therefore synthetic necessities can be known. Is there a similar challenge facing the more austere interpretation that repudiates transcendental idealism con1pletely and even abandons a distinction between a priori and a posteriori knowledge? Strawson recognizes and discusses a certain kind of challenge along these lines. He sees that it is important to distinguish the Kantian project as he understands it fron1 a certain 'historical'-what he sometimes calls a 'n1erely historica1'18-conception of an investigation of the fundamental framework of ideas in terms of which we make sense of our experience. On that 'historical' view of metaphysics, the general structure we can uncover by philosophical reflection might be fundamental and unquestioned in a particular scientific epoch or culture, but it could change or be gradually and silently abandoned when scientific or social thinking enters a new historical phase. There could be great illumination in revealing the basic categories and presuppositions of what with a fairly determinate reference could be called our thought and experience of the world, but even the most secure results would not necessarily hold for everyone, or for all thought and experience whatever. Strawson thinks this conception is 'very far from Kant's intentions', and if true would make the Critique of Pure Reason 'a less interesting work than we had hoped' .19 He wants to preserve a project that he thinks 'corresponds more closely to Kant's own'.20 The 'historical' view of the metaphysical enterprise would restrict us to the fundamental ideas and presuppositions of a particular time and culture. But it would not have to deny that the results achieved by metaphysical reflection have a certain kind of necessity. It could find that anyone who thinks of the world in certain ways must also think of it, or at least acknowledge it as being, this and that other way as well. Within the very concepts in which a society grasps the world, there might be certain other concepts necessarily connected with them. In short, the 'historical' conception of metaphysics could endorse what Kant would call the 'analytic' character of the necessities it discovers. This would be metaphysics as the study of what Hume called 'relations of ideas'. That does not mean that it would be as easy as Hume sometimes suggests knowledge of the 'relations of ideas'
18
Strawson, Bounds of Sense,
121.
19
Ibid.
20
Ibid.
Synthetic A Priori in
Strawson~s
Kantianism
I
17
can be: merely holding up an idea for inspection and gazing 'inside' it to see whether a certain other idea is included in it or not. It might require a deep enquiry of great complexity to lay bare some of the most fundamental and far-reaching ways of thinking involved in our experience and thought of the world. But any outcome reached in that way could still be seen as an 'analysis' or a bringing to light of the contents of those experiences and thoughts. The more Kantian conception of the project that Strawson favours involves the discovery of some necessities that are different from that. On the 'historical' view there is no necessity in anyone's having any particular conception of the world, even a conception with a certain fundamental structure. For all that view says, we could, or someone could, have had radically different kinds of experience, and so a radically different conception of the world. The necessities discoverable on that view would be in that sense conditional, concerned only with what is necessarily involved in a certain way of thinking and experiencing things, without implying that those are the only ways in which thought and experience of a world are possible. Other ways of thinking would contain different necessities of their own. This is not the conception of their project shared by Strawson and Kant. They seek stronger conclusions, and so different necessities. That Strawson aspires to something more can be seen in another way when he is struck with a possible way of taking Kant's claim to have discovered what is 'necessarily involved in any coherent conception of experience'21 that would rob it of most of its interest for him. Experience, for Kant, is possible only if it is experience of objects in a fairly weighty sense, involving the application to items in experience of certain identifiable general concepts or 'categories' that carry certain implications of objectivity. But if this were simply a matter of the definition. of the word 'experience', or if the claim were simply laid down as a pren1iss from which the rest of Kant's investigation proceeds, Strawson thinks our hopes for the most fruitful outcon1e of the critical enterprise would be considerably deflated. Fortunately, it turns out that this is not so. When Kant is understood correctly, his claim that experience necessarily involves experience of objects that are independent of particular experiences of them is seen to have 'all the depth and interest which we were threatened with the loss of,.22 The threat came from the now-rejected idea that what Kant establishes are at most 'analytic' or 'definitional' necessities. But for Strawson Kant does not simply 'analyse' or seek a definition of the concept of a unity of experience; he asks what it would take, what must be so, in order for a set of experiences to be unified in the relevant way.
21
Bounds of Sense, 72.
22
Ibid. 74.
118
S~oud
This leads to a series of necessity claims that can be briefly summarized as follows. The set of experiences must satisfy the conditions of being united in a single consciousness. That requires that they satisfy the conditions that n1ake possible the self-ascription of those experiences to the subject whose experiences they are. That in turn requires the possibility of consciousness on the part of the subject of the identity of that to which the different experiences belong and could be ascribed. And that requires further the possibility of some of those experiences being experiences of objects that are distinct from and independent of experiences of them. Judgements about such objects are true or false no matter what the state of any experiencing subject might be. So 'unity of diverse experiences in a single consciousness requires experience of objects'. 23 This is a necessity claim; one thing is said to be required for another. It is supported by all those intermediate claims leading up to it, and they in turn are to be further supported by looking for deeper or more general reasons why they must hold. Perhaps they can be derived, as Strawson tries to derive them, from something as basic to the possibility of thought and experience as the idea of particular items being thought of as falling under concepts. The strength of such claims can be tested, as Strawson tries to test it, only by seeing how they stand up to attack. There is no foolproof method here. What is special about the project is not the method but the kind of necessities that are to be established. The question is what must be so in order for a person to be capable of certain thoughts or experiences. The possibility of having certain kinds of thoughts or experiences must be in a person's repertoire in order for experience to be possible at all. And further conditions n1ust be fulfilled in order for those things to be in his repertoire. What other things must he be able to think or experience, what other distinctions must he be able to draw, or in what other ways must he be able to make sense of things? These are questions about what Strawson elsewhere has called 'conceptual capacities', and the necessary linkages between them. 24 They are not questions about the necessary connections between the concepts or meanings employed by a thinker. They do not seek simply to 'analyse' or lay bare the contents of a person's concepts or thoughts or experiences. They are concerned with the conditions of anyone's having thoughts or experiences with certain contents. Concept possession, or conceptual and experiential capacity, is what is in question, not just the concepts themselves. One concept's being known to be 'covertly contained' in another would contribute nothing to an understanding of how thought and experience are possible if possession and employment of those concepts were not required for thought and experience. 23 24
Bounds of Sense, 98. See his Skepticism and Naturalism: Some Varieties (London: Methuen, 1985), 21 ff.
Synthetic A Priori in
Strawson~s
Kantianism
119
To establish such necessary connections between some of our conceptual and experiential capacities would reveal a certain distinctive character or status of some of the elements of our conception of the world. If a person had to be able to think in certain specific ways-e.g. to think of himself as a subject of experiences-in order to think or experience anything at all, and if thinking of the world as containing objective particulars that are independent of thought and experience was required in order to think of oneself as a subject of experiences, then those parts of our conception of how things are would have a special status. They would not have been shown to hold necessarily, or even to be something we must think holds necessarily. It would mean only that thinking of the world as contingently containing subjects of experiences would require thinking of it as contingently containing objective particulars independent of experience as well. Nor would that imply that there are subjects of experiences or objective particulars. The truth of what we think does not follow from our thinking it, or even from our being required to think it. 25 Even the link between the contents of the things we think might not be necessary. If thinking that p requires thinking that q, it does not follow that it is necessarily true that if p then q. Of course, it is not ruled out either. But if certain ways of thinking are required for the possibility of any thought and experience at all, then what we think in thinking in those ways has a special, and a specially invulnerable, standing. It must be accepted or acknowledged by anyone who thinks or experiences anything. It could not consistently be exposed as a mere metaphysical illusion, as nothing more than a widely held belief that for all its familiarity fails to capture the way things really are. That there are independently existing objective particulars, for instance, would be a belief that must be held even by someone who claims to conceive of, or to believe in, a world in which that belief is held by many people but is not really true. No one could so detach himself from that indispensable belief as to find that the world is that way. All indispensable beliefs are invulnerable to that kind of 'unmasking'. This is actually one of the sources of incoherence or inconsistency in transcendental idealism. That doctrine wants to say that objective particulars independent of experience are indispensable to any world we can find intelligible, and also that, transcendentally' speaking, what we take to be independent objects are not really independent of experience after all; nothing is. That seems to take back with one hand what it offers with the other.
25 This is the substance of my objection to what I have called 'ambitious transcendental arguments' which would draw conclusions about how things are from premisses about how we think (see my 'Transcendental Arguments' and 'Kantian Argument, Conceptual Capacities, and Invulnerabilty' in my Understanding Human Knowledge: Philosophical Essays (Oxford: Oxford University Press, 2000)).
120
S~oud
Transcendental idealism tries to escape the charge by claiming that it is not the same thing in each of the hands. Establishing that some elements of our conception of the world have the kind of standing that guarantees invulnerability to unmasking requires an 'absolute' or non-conditional understanding of the necessities discovered by metaphysical reflection. The indispensability of certain ways of thinking must be indispensability for any possible thought or experience at all, not merely for this or that conception of the world that we, or some other culture or epoch, happen to have. The 'historical' view challenges the assumption of this conception of the metaphysical task. So, in another way, does the Humean view. The questions they raise about the possibility of establishing such absolute necessities are perhaps as close as we can come to austere analogues of Kant's question of how synthetic a priori judgements are possible. For any necessity claim the Strawsonian Kantian project conles up with, the 'historical' view will find it true only 'for us', or true only 'for' those whose conception of the world is in question; for thought and experience with such-and-such particular form or structure, but not necessarily for all thought and experience. Even the apparently stronger Strawsonian claim to have identified elements that are necessary to any conception that we can make intelligible to ourselves is something the 'historical' view might accept, by taking it strictly. It nlight see our inability to conceive of other possibilities as a result of our being so immersed in our period and culture that we cannot even make sense of things being otherwise in certain ways. But still, it says, they could be. There could be forms of thought and experience of which we, for reasons having to do only with us, can form no coherent conception at all. But the fact that we can form no coherent thought of any such alternatives does not imply that there simply are no such possibilities. The wisest response to this last point, it seems to me, is to grant it. Fronl our failure to make sense of something we can recognize as a possibility, it does not follow that there is no such possibility. There is nothing in the austerely understood Strawsonian or Kantian project to guarantee without possibility of failure that the necessity claims the philosopher comes up with nlust be true. Nor is there anything in our ways of trying to find out things in general that guarantees that we will succeed. That is a fact of life we just have to accept; and it applies as much to philosophy as to anything else. This is something that I think Kant's own richly transcendental, nonausterely-understood project denies, at least with respect to philosophy. It was meant to give the philosopher such a guarantee. The knowledge sought by the project was to be a priori, therefore 'in us', even in some sense 'contributed by us' or 'by reason', and therefore directly available to us by the operation of 'pure reason' with complete certainty. Since the critique of
Synthetic A Priori in Strawson's Kantianism
121
pure reason 'has to deal ... only with itself and the problems which arise entirely from within itself, and which are imposed by its own nature', 'it should easily be able to determine [its limits] with con1pleteness and certainty' (B 23). 'Judgments [of reason] are never opinions', Kant says; 'it must affirm with apodeictic certainty' (A 775/B 803). 'Otherwise we should have no guidance as to truth' (A 823/B 851). This is how metaphysics can be put on the path of a science that we can know in advance is secure. There is no reassuring epistemological theory of this or any other kind behind the austerely understood Strawsonian project. The metaphysical investigator can do no better than to reflect on whether or how it would be possible for one thing to be present without the other. He can tryout necessity claims, and try to defend them against potential attack. To understand more fully why something that looks like a possibility is not really possible after all would help explain the basis of the necessity claim, and so would add further support to it. This is the most we can ever do in establishing necessities, when it is not a matter of explicit demonstrative proof. To the 'historical' metaphysician's caution that there could be forms of thought and experience of which we so far have been unable to forn1 any coherent conception, it is best to concede that, yes, perhaps there could be. But, yet again, perhaps there could not be, either. What we have now found to be the limits of the possible could well be the limits of the possible. Anyway, that is what we now believe. To pursue the question further the only thing we can do is keep trying our best to find out what is necessary and what is not. To refuse to draw conclusions about the necessary features of any possible experience simply because, after all, it is only we who are drawing the conclusions, only we who live and think and experience in a particular culture at a particular time and place, would sin1ply be to give in to a different abstract and general theory of thought and experience. And there is much in what we have discovered so far that does not support that subjectivist or relativist theory. One challenge represented by the 'historical' view was that it would absorb all the necessities discoverable by metaphysical reflection into merely conditional or 'analytic' necessities, and so rob the project of its universality and absolute necessity. The interesting austere project was to avoid that deflation by concerning itself with necessities among the conditions of the possession and employment of certain concepts, not simply among the concepts themselves. But it is just possible, I suppose, to see even that enterprise as yielding at best only 'analytic' or 'definitional' necessities. They would express what is 'covertly contained' not in the concept of, say, 'experience', or the concept of 'subject of experiences', but in the concept 'possesses the concept of experience', or the concept 'possesses the concept of a subject of experience'. 'Analysis' might reveal that the quite different concept 'thinks of the world as containing objective particulars' is 'contained' in one or both of those concepts.
122
S~oud
As I say, it is perhaps just possible to think of an examination of the necessary conditions of the possession of certain concepts in this way. Devotion to the hopeless idea of analyticity apparently dies hard. I think that is because analyticity is widely thought to be the only possible explanation of necessity, and of our knowledge of it. That double requirement on the idea of analyticity is what I think is the source of its inveterate obscurity. It is expected to account for too many different things all at once, and accounts for none of them. So it adds nothing to a necessity claim to the effect that having one concept or capacity requires having certain others. The 'historical' view of metaphysics does not present a sharp, identifiable challenge when it n1akes essential use of the idea of analyticity in this way. The Humean view is not 'historical' or social or potentially relativistic. It is universalistic, about all human beings. But it says that any necessities discoverable by a priori reflection are merely 'analytic' connections, or 'relations of ideas' alone. Its differences from the austere Strawsonian or Kantian project can therefore also be measured only in terms of the repudiated notions of analyticity and the a priori. Hume says that any nonanalytic necessities we might discover about the ways we think and experience the world are merely 'natural' or causal necessities. If it is true that no one can avoid believing that there are objective particulars, or that every event has a cause, as he thinks it is, that is simply because of the way human beings are. Given the way the world is, and the way it affects them, they cannot avoid getting those beliefs. That is for Hume a causal fact of the world. But causal necessities are not absolute necessities. It is possible for things that are in fact causally related not to have been so causally related. Does the idea that we are all caused to believe, and in that sense cannot help believing, in objective particulars and causal connections pose a threat to the idea that those beliefs have the distinctive character or status in our thought that I have identified? I don't think so. For one thing, if it is true that we must believe in such things, then we do believe in them. So we can see our being unavoidably caused to believe in them as an indication that the way' we are unavoidably caused to believe things are is, in certain respects, just the way things are. Hume's more troubling view is not just the causal thesis that we inevitably get those beliefs, but that the beliefs in question are not true; we only think there are objective particulars and causal connections. But making that negative claim seen1S to overlook the positive part of the story that says that everyone has to believe such things. It appears to be claiming for the philosopher who discovers such bad news an exemption from the operation of laws that are said to apply to all human beings. But then the positive part of the story would not be universally true. So it is questionable whether Hume could consistently arrive at the negative part of his view if the positive causal part about human beings and the world is also true. That again is the question of whether beliefs that really
Synthetic A Priori in Strawson's Kantianism
123
are indispensable can consistently be unmasked as illusions. It does not seem to matter whether the indispensability is only causal or something stronger, even absolute. But the fact that causal or natural necessities are not absolute-that it is possible for two things that are causally related not to be so related-does not itself threaten the absolute necessity of the claims a defender of the austere Strawsonian or Kantian project would make. Suppose that Hume is right that we are all inevitably caused to believe in a world of objective particulars; that is a natural necessity. And suppose Strawson and Kant are right that believing in objective particulars is a necessary condition of the possibility of anyone's having any experience of a world at all. Then any world in which that natural necessity does not hold, and human beings in interaction with that world get no thoughts or beliefs about objective particulars, would be a world in which those human beings do not have experience of that world, or of anything else. They would not fulfil one of its necessary conditions. Maybe the idea of possible human beings with no thought or experience of anything at all is too much to accept. But, looking on the bright side, if Hume is right about the natural necessities, then human beings cannot fail to fulfil that necessary condition of the possibility of experience. It is inevitable, given the way human beings are and the way the world is. So any conflict between the Humean view and the austere Strawsonian or Kantian conception turns again only on the negative part of the Humean view: the idea that no necessities are discoverable a priori except 'analytic' connections or 'relations of ideas'. I have admitted a way in which even that might be compatible with the search for conditions of the possession and employment of concepts. But the negative Humean view represents a challenge to the austere Strawsonian project only if some knowledge is possible 'absolutely independently of all experience', and if all necessities are knowable only in that a priori way. Even then the force of the challenge would be directly proportional to the degree to which some clear light can be thrown on the idea of purely 'analytic' connections, or 'relations of ideas' alone. So the austere project appears to be left only with what looks like a standing challenge to all knowledge, or to all philosophical theory. Strawson takes up the question briefly at the end of The Bounds of Sense. 'How is it, after all, possible to establish that experience must exhibit such-and-such general features?' he asks. 26 He gives what seems like the only possible general answer to the question. We do our best to find out what must be so if something else is, and we test what we find against the best potential counter examples we can muster. Of course, any particular necessity claim can be
26
Strawson, Bounds of Sense,
271.
Stroud challenged, and perhaps overthrown. But that will be done, as Strawson says, 'only by making us able to understand the possibility of an alternative'. 27 But for many philosophers, and here I think we corne to the crux of the matter, there is thought to be something special, and especially problematic, about necessity. It is thought to be in some way mysterious. That is why our knowledge of it is felt to need a special kind of explanation. The idea itself apparently cannot be explained or defined, except in terms like 'possibility', 'contradiction', and so on, which in turn can be explained only in terms of necessity. I think that is true, but if it creates a special problem about our knowledge of necessities, then there is a special problem about our knowledge of everything. What we know are truths, and the idea of truth cannot be defined in terms that can be understood independently of the notion of truth either. Reductionism is out of the question for truth, just as I believe it is for necessity. But necessity has been thought to be especially problematic because, roughly speaking, it has been believed that there is and can be no such thing as necessity in the way things really are. It is said to be only something we impose on our thinking about the way things are. Necessary propositions have been said to be 'empty' or 'devoid of factual content'. That is harmless if it means only that necessary truths do not state contingent facts. If it means more than that, it expresses a conception of necessity in general that is analogous to Hume's deflating account of causal necessity: the idea that there is no such thing in the world as it really is, we only think there is because of the ways our minds work. It is precisely this conception of necessity, and our knowledge of it, that I think the idea of analyticity has been brought in to explain. That is the real source of its continued appeal, despite the dismal record of attempts to explain what analyticity actually is. It is believed that some idea of analyticity must be accepted, however difficult it is to formulate it, because it is the only way we could explain how there can be necessary truths among the things we know, given that necessity is really only something we impose on the world by the ways we think about it. The source of whatever necessity we recognize is therefore thought to be ultimately only 'in us', in the meanings we build in to the concepts we use to understand what exists independently of our thought. But if necessity has a merely human or conventional source, that source must somehow be found in what can only be contingent facts about human beings. That is why I think no such notion could do all the jobs demanded of it. It is this broadly Humean picture of necessity as somehow coming 'from us' that I think generates the idea that there is a special difficulty about 124
27
Strawson, Bounds of Sense,
271.
Synthetic A Priori in Strawson"s Kantianism
125
necessity and our knowledge of it. The picture is not far from the kind of 'transcendental subjectivism' that the austere Strawsonian conception repudiates in Kant. And to this version of the picture I think we should give the same response. We do think of some things as necessarily true; and perhaps we do so because of the ways our minds work. But if what we believe is that some things are necessarily true, and the idea of necessity is irreducible to any other notions that do not presuppose it, then no one could consistently reach the view that necessity is only something subjective and not part of the way things are. That some things hold necessarily would be invulnerable to philosophical unmasking. Without the shrunken picture of an impoverished independent reality in which there are no necessities, and so without the special problem that picture gives rise to, perhaps we are left after all only with Strawson's very general question of how it is possible to establish that experience must exhibit such-and-such features. Nothing very interesting can be said with complete generality; we can only look at the details of particular proposals, and try them out. None of this suggests that it is easy. It is often very difficult to find out what is necessarily true. But then it is also often difficult to find out what is true.
7 Epistemic Reflection and Transcendental Proof KENNETH R. WESTPHAL
• ••
1. INTRODUCTION
After initial enthusiasm sparked by Professor Strawson's The Bounds of Sense, Kant's transcendental arguments have been sharply criticized by analytic commentators. Barry Stroud has observed that 'it is not easy to incorporate the depth and power of Kant's transcendental deduction into present-day philosophical attitudes and preconceptions'. 1 However, this may be less a problem with Kant's transcendental proofs than with our understanding of them. Instead of incorporating Kant's transcendental proofs into present-day philosophical attitudes, it may be more illuminating to reconsider some of our current philosophical attitudes in order to understand and benefit from Kant's transcendental proofs. Kant's Refutation of Idealism has an anti-Cartesian conclusion: 'inner experience in general is only possible through outer experience in general' (B 278).2 Owing to preoccupation with Cartesian scepticism, and to the antinaturalism of early analytic philosophy, most of Kant's recent anglophone commentators have sought to find a purely conceptual, broadly 'analytic' argument in Kant's Refutation of Idealism-and then have despaired when no such plausible argument can be reconstructed from Kant's text. Allegedly, Kant must argue by elimination, though his transcendental arguments do not
1 Barry Stroud, 'Transcendental Arguments and "Epistemological Naturalism'" Philosophical Studies, 31 (1977),105-15, at 105. 2 Unless noted otherwise, translations are mine. Kant's first Critique is designated by its two editions, A and B. Other works, designated by their German initials, are cited by volume/page numbers of Koniglich PreuBische (now Deutsche) Akademie der Wissenschaften, Kants Gesammelte Schriften (Berlin: G. Reimer (now De Gruyter), 1902-), usually referred to as Akademie-Ausgabe. 'Guyer and Wood' designates Critique of Pure Reason, trans. P. Guyer and A. Wood, (Cambridge: Cambridge University Press, 1998).
128
Westphal
eliminate the possibility of Descartes's evil deceiver, or alternative forms of cognition, or the possibility that the mere (individually subjective) appearances of things suffice for the possibility of self-consciousness. Such disappointments overlook two key features of Kant's response to scepticism. One is Kant's decidedly non-Cartesian philosophy of mind. The other is Kant's semantics of cognitive reference (which requires separate discussion). Kant's epistemology highlights four integrated ways in which we depend cognitively on a common-sense spatio-temporal world. This is crucial to Kant's proof that we sense, not merely imagine, perceptible objects distinct from ourselves (B xl-xli n.). My remarks will be selective and suggestive, though I hope to show that this approach merits further consideration.
II. KANT'S NON-CARTESIAN PHILOSOPHY OF MIND
11.1. There is some historical irony in the fact that in the mid-196os Kant should be found to have provided a purely (if broadly) analytic argument against scepticism. The dismissal of Kant's transcendental psychology was of a piece with the anti-naturalism of analytic epistenl010gy, according to which our actual cognitive processes are the province of empirical psychology, and are philosophically irrelevant. Ironically, these claims about Kant's views were advanced just at the time Gettier began convincing epistemologists that our actual cognitive processes must, at some level, be taken into philosophical account. 3 Kant's transcendental psychology is his way of taking our actual cognitive processes into philosophical account. Kant based his epistemological arguments on an inventory of our basic cognitive capacities to employ our forms of intuition and our fornls of judgement (A 66/B 91, B 145-6).4 Ultimately, Kant supposes that determining the correct inventory of human cognitive capacities is a collective undertaking. 5 Now, is Kant's view, for example, that identifying one's inner states requires making certain kinds of judgement an insight? Or is it just another philosophical view? It's easy to propose a non-Cartesian philosophy of mind. Kant does so by maintaining that our understanding can 'only think', that 3 Edmund Gettier, 'Is Justified True Belief Knowledge?', Analysis, 23/6 (1963), 121-3; cf. Philip Kitcher, 'The Naturalists Return', Philosophical Review, 101/1 (1992), 53-114, at 59. 4 This point has been stressed by Strawson in 'Imagination and Perception', in L. Foster and J. W. Swanson (eds.), Experience and Theory (Amherst, Mass.: University of Massachusetts Press/London: Duckworth, 1970), 3 I-54, and 'Perception and its Objects', in G. F. MacDonald (ed.), Perception and Identity (Ithaca, NY: Cornell University Press, 1979, 41-60). The Kantian pedigree of Strawson's analysis is made plain by Bela Milmed's explication of Kant's views in "'Possible Experience" and Recent Interpretations of Kant', in L. W. Beck (ed.), Kant Studies Today (LaSalle, Ill.: Open Court, 1969), 301-2I. 5 Onora O'Neill, 'Vindicating Reason', in P. Guyer (ed.), The Cambridge Companion to Kant (Cambridge: Cambridge University Press, 1992), 280-308.
Epistemic Reflection/Transcendental Proof
129
is, connect representations, whereas our sensibility can only provide, but not connect, representations (B 130, 134-5, 145). If so, and if our outer sense is only receptive, a mere susceptibility to stimulation by things distinct from us (A 19-21/B 33-5, A 50-2/B 74-6), then 'inner experience' cannot have Cartesian primacy over 'outer experience'. Kant's philosophy of n1ind maintains that: (I) our outer sense is passive; (2) our outer sensations are caused by something other than us, but this causal relation does not suffice for sensations to represent their causes ('sensationism');6 (3) our senses cannot combine or represent any combination among our sensations (B 13 0); 7 (4) our understanding cannot provide representations (sensory or otherwise), but can only combine representations judgementally (B 135, 145);
and (5) judgemental combination of sensations is required to form percepts of objects and to ground cognitive perceptual judgements. If these theses are true of us, then any self-conscious experience we can have requires (though it may not consist solely in) experience of mind-independent real objects, which are our source of sensations ((I), (2)). These real objects would be spatio-temporal if it were further maintained that (6) our outer sense only represents spatio-temporal phenomena. Such provocative, unconventional theses reinforce the question, why believe Kant's claims, rather than their Cartesian (or empiricist) alternatives? More pointedly, Kant appears to assume them without proof. We should not, however, hasten to reject Kant's enterprise.
11.2. How can or should consensus about our cognItIve capacities-the building materials of human knowledge (A 707/B 735)-be reached? Who are we exactly, and what are our cognitive capacities? Descartes wrote meditations rather than disputations in part in order to encourage his readers to identify and reflect on their own, God-given cognitive nature and stock of innate ideas. This was wise in an important regard: although it can be reported to others, some kinds of evidence can only be had first-person. 8 Rolf George, 'Kant's Sensationism', Synthese 47/2 (1981), 229-55. See Robert Howell, Kant's Transcendental Deduction (Dordrecht: Kluwer, 1992), 70-89, I05-1I. 8 Kenneth R. Westphal, 'Sextus Empiricus Contra Rene Descartes', Philosophy Research Archives, 13 (1987-8), 91-128, at 108-9. 6
7
130
Westphal
Kant provides a variety of considerations and arguments to identify our cognitive capacities. They involve a kind of philosophical reflection on our own cognitive capabilities and their preconditions that is significantly richer than has been noticed by recent commentators, though it falls fortunately short of the hoary details of Kant's transcendental psychology. Below (Sect. III) I sketch four thought-experiments Kant proposes to enable us to recognize some of our basic, characteristic cognitive capacities, and some limits and requirements they entail for the nature and objects of human knowledge. These thought-experiments involve a kind of reflection on who we as cognizant subjects are, on what our basic cognitive capabilities and circumstances are. This kind of reflection is a more general, preliminary form of what Kant calls 'transcendental reflection', defined as the action through which I make the comparison of representations in general with the cognitive power in which they are situated, and through which I distinguish whether they are to be compared to one another as belonging to the pure understanding or to pure intuition. (A 261/B 317)
Transcendental reflection belongs to Kant's transcendental logic. Kant insists that 'transcendental reflection is a duty [sic] from which no one can escape if he would judge anything about things a priori' (A 263/B 319). Kant thus warns us that disregarding transcendental reflection would thwart anyone's search for a sound argument in the first Critique. The lack of discussion of 'transcendental reflection' in recent anglophone Kant con1mentary may indicate that we have missed something very basic and important about Kant's arguments and analyses, and may suggest that the lack of satisfactory results stems from too selective a reading of I
9 On 'transcendental reflection' see Herbert Schnadelbach, Reflexion und Diskurs. Fragen einer Logik der Philosophie (Frankfurt/Main: Suhrkamp, 1977), 87-133; Dieter Henrich, 'Kant's Notion of a Deduction and the Methodological Background of the First Critique', in E. Forster (ed.), Kant's Transcendental Deductions (Stanford, Calif.: Stanford University Press, 1989), 29-46; Beatrice Longuenesse, Kant and the Capacity to Judge (Princeton, NJ: Princeton University Press, 1998), 113, 123-7, 156 n.; cf. 203 and n. 13; and Marcus Willaschek, 'PhaenomenalNoumena und die Amphibolie der Reflexionsbegriffe', in G. Mohr and M. Willaschek (eds.), Immanuel Kant: Kritik der reinen Vernunft (Berlin: Akademie Verlag, 1998), 325-51.
Epistemic Reflection/Transcendental Proof
r 3r
to inventory (though not to analyse nor to explain) our most basic cognitive capacities and incapacities through a series of thought-experiments.
11.3. The most familiar (and historically least persuasive) of these thoughtexperiments concerns Kant's claim that (r) we cannot represent to ourselves the absence of space (A 24/B 38). More familiar and historically more persuasive is Kant's claim that (2) each of us must be able to identify our representations as our own, 'for otherwise I would have as multicolored, diverse a self as I have representations of which I am conscious' (B r 34, trans. Guyer and Wood). Less familiar are two further thought-experin1cnts. One is that (3) 'only under the presupposition of diversity in nature, just as it is only under the condition that its objects have homogeneity among themselves', are we at all capable of self-conscious experience. 10 The last concerns the fact that the three Analogies form a tightly integrated set of mutually supporting principles, so that (4) we can only identify coexistence or succession by discriminating each from the other. I discuss each example in turn (Sects. 1Il.r-4), and then discuss some of their joint inlplications (Sect. III. 5).
III. FOUR THOUGHT-EXPERIMENTS
III. I. Kant begins his discussion of space (or its concept; B-edition) by stating: 'Through outer sense (a property of our mind) we represent to ourselves objects as outside us, and all of them in space' (A 22/B 37), and he notes that '[t]ime cannot be intuited externally, any more than space can be intuited as something in us' (A 23/B 37). To prove that space is a necessary a priori representation, basic to all our outer intuitions, Kant claims that 'one cannot make oneself a representation that there be no space' (A 24/B 38), though we can of course think of void space. To represent and to recognize something intuited as distinct from us, as outer, or as having determinate spatial characteristics (A 22/B 37) requires that we have both a way to represent space and a concept of space by which to identify, to 'determine', these parameters. Kant argues analogously about time and its concept (A 3 riB 46). Kant's 'second expositions' of space and time raise many questions. 11 Kant contrasts what we can think (which is limited only by the law of noncontradiction and the wilds of our imagination (B xxvi n.)) and what we can represent to ourselves. His main point is that we cannot experience objects distinct fronl ourselves unless we represent them as, and recognize them as
KdrV, A 657/B 685, cf. A 90-IIB 122-3, A 653-4/B 68r-2; A rOO-I, A 108, A 121-3. Lorne Falkenstein, Kant's Intuitionism: A Commentary on the Transcendental Aesthetic (Toronto: University of Toronto Press, 1995), 186-216. 10
11
Westphal being, spatial and in space (A 22-3/B 37) and as enduring through at least son1e period of time (A 3riB 46). Our sensory experiences are inextricably temporal, and a prominent class of them is inextricably spatial. 12 The inextricably spatial character of a key set of our representations is crucial, for this spatiality is necessary both to identify and to reidentify particulars distinct from ourselves. 13 Though logically contingent, Kant's point is that in fact our way of representing objects distinct frorn ourselves is and, so far as we are human, n1ust be spatial. Concepts or representations of space and time are necessary, ineluctable grounds of our possible representation of what we experience. Kant's claim is not introspective; he does not inventory 'in the mind's eye', as it were, the contents of either his or our consciousness. Kant's claim is based on our inability to perform a certain kind of act of 'making ourselves a representation' of the absence of space, or analogously of time (A 3 liB 46).14 Kant's claim is about our cognitive capacities, not the contents or even the capaciousness of our imagination; his claim is based on epistemic reflection, not introspection. Kant takes himself as a representative human cognizer (A l2-13/B 26), but he need not and does not take his own psychological states as a representative sample. Instead, Kant notes these facts about our capacities to represent and identify items in space and time, and uses them to help us determine what we can n1ake intelligible sense of in cognitive contexts. 1S Distinguishing between philosophical or, in the present case, epistemic reflection and introspection is in1portant, for analytic philosophers have uniformly followed Mill in equating reflection with introspection in a Humean sense, as the finding of particular mental contents, 'sense impressions' as it were. 16 Searching only for these will leave us as bereft of Kantian insights as Hume was of an idea of hin1self. Conflating Kant's epistemic reflection with introspection guarantees misunderstanding, if not missing, many of Kant's most important invitations and provocations in the first Critique to reflect philosophically in order to determine our basic cognitive capacities. Kant grants that other beings may have different forms r 32
Falkenstein, Kant's Intuitionism: A Commentary on the Transcendental Aesthetic, 195-9. See Strawson, Subject and Predicate in Logic and Grammar (London: Methuen, 1974), 15-16, 'Sensibility, Understanding, and the Doctrine of Synthesis: Comments on Henrich and Guyer', in Forster (ed.), Kant's Transcendental Deductions, 69-77, 72, and Analysis and Metaphysics (Oxford: Oxford University Press, 1992), 54-6; and Onora O'Neill 'Space and Objects', Journal of Philosophy, 78/2 (1976), 29-45; cf. Jay Rosenberg, 'On Strawson: Sounds, Skepticism, and Necessity', Philosophia, 8 (1978),4°5-19. 14 Cf. Arthur Melnick, Space, Time, and Thought in Kant (Dordrecht: Kluwer, 1989), 28-9; Longuenesse, Capacity, 229. 15 Cf. Schnadelbach, Reflexion, I 14, I 18. 16 Robert Scharff, Comte after Positivism (Cambridge: Cambridge University Press, 1995); Richard Rorty, 'Strawson's Objectivity Argument', Review of Metaphysics, 24 (1970), 207-44, at 216. 12
13
Epistemic Reflection/Transcendental Proof
133
of understanding or sensibility than ours (B 7 2, 13 8-9, 145, 159, A 286-8/B 342-4). Such beings may not ineluctably require representations of space, time, or of the 'I think' to represent what they experience. However, Kant neither offers nor needs to offer an epistemology on their behalf; his concern is with the scope and limits of human knowledge. Of course epistemology will seem bankrupt, and scepticism unanswerable, if our basic cognitive capacities are disregarded. Hence Kant attends, much more carefully than he has been credited with, to determining what they are.
111.2. Philosophers have been more sympathetic to Kant's thesis that each of us must be able to identify our representations as our own, 'for otherwise I would have as multicolored, diverse a self as I have representations of which 1 am conscious' (B 134, my emphasis; cf. A III, A 112).1 7 Minin1ally, Kant's point is that because sensory representations are fleeting, their n1ere occurrence is insufficient for us to recall them as having been our own. Such recollection is necessary for us to gain any stable knowledge-or even stable beliefs-about what we have experienced. Moreover, such recollection requires more than that some current state be caused by some prior, putatively recollected state. It requires that our present recollection be, and manifestly be, of a prior state of our own. This is required both for the recognition of any stable object or of any process (whether motion or transformation) over any period of time, however brief, and for the recognition of any personal history of experiences, however haphazard or extensive. Kant's point is that the mere occurrence of a recollection-impression within a bundle, or the n1ere inherence of a representational state, the object of which happens to be past, within a Cartesian mental substance (or within a Humean bundle of perceptions) does not suffice-not for human beings-to identify those states as our own and so to be able to base cognitive judgements on them. III. 3. Kant bases a third important claim on our reflecting on a wildly counterfactual circumstance. Kant claims: 'only under the presupposition of diversity in nature, just as it is only under the condition that its objects have hon10geneity among themselves, do we have understanding [Verstand]'or self-conscious experience-at all. 18 If the synthetic transcendental unity of apperception is the key necessary noetic condition for the possibility of self-conscious experience (Sect. 111.2), this 'transcendental affinity of the manifold of intuition' is one key necessary material condition. 17 See Manfred Baum, Deduktion und Beweis in Kants Transzendentalphilosophie (KonigsteinfTs.: Hain bei Athenaum, 1986), 93-118; cf. Paul Guyer, Kant and the Claims of Knowledge (Cambridge: Cambridge University Press, 1987), 136-49, 'The Transcendental Deduction of the Categories', in Cambridge Companion to Kant, 123-60, at 150-2; and Howell, Transcendental Deduction, 155-9°. 18 KdrV A 657/B 685 (quoted below), cf. A 90-liB 122-3, A 653-4/B 681-2; A 100-I, A 108, A 121-3.
Westphal Kant contends that our making cognitive judgements requires that appearances be associable. If self-conscious experience is possible at all, then this associability of appearances must have an objective ground, which Kant calls the 'transcendental affinity of the manifold of intuition' (A 121-3). This requirement follows from the fact that a complete human sensibility and understanding, capable of associating perceptions, does not of itself entail that any appearances or perceptions it has are in fact associable. If they were not, we might have fleeting, random sensations, but no integrated self-conscious experience. In part this is because irregular perceptions do not afford reproductive synthesis; they would afford no psychological association, and so would disallow our developing empirical concepts and our using categorial concepts to identify objects. To make this point, Kant describes a radically counterfactual circumstance, on which we ought to reflect carefully: 134
If among the appearances offering themselves to us there were such a great a variety-I will not say of forn1 (for they might be similar to one another in that) but of content, i.e., regarding the manifoldness of existing beings-that even the most acute human understanding, through comparison of one with another, could not detect the least similarity (a case which can of course be thought) ... [then] no concept of a genus even, nor any other universal concept, indeed no understanding at all would obtain, since it has to do solely with such [concepts] ... [Hence] homogeneity is necessarily presupposed in the manifold of a possible experience (although we cannot determine its degree a priori), because without it no empirical concepts and hence no experience would be possible. 19
Significantly, Kant identifies the relevant kind of similarity as a matter of the content ('dem Inhalte'; alternatively 'the object' (A 112-13)), rather than the form, of appearances. Kant thus indicates that the regularity of the objects of (or in) experience is a material condition. The fact that it concerns relations among the characteristics of the contents of experience indicates that this is also a 'formal' condition, in Kant's sense; it concerns orderability. (However, this formal condition is neither intuitive nor conceptual.) The consequences of the lack of such similarities are of transcendental import: without recognizable regularities we \voulJ have no empirical concepts, no use of our schematized categories, indeed no functioning understanding, and hence no self-conscious experience. Consequently, this principle is constitutive of the possibility of integrated self-conscious experience. In the extreme case suggested by Kant, where there are no detectable regularities or variety within the contents of our 19 KdrV A 65 3-4/B 681-2, trans. Guyer and Wood; emended, emphasis added. The last sentence indicates that this issue is constitutive, not merely regulative. Elided is Kant's n1ention of the 'logical law of genera'; at this constitutive level, this logical law coincides with transcendental affinity of the manifold of intuition (see K. R. Westphal, 'Affinity, Idealism, and Naturalism: The Stability of Cinnabar and the Possibility of Experience', Kant-Studien, 88 (1997), 139-89).
Epistemic Reflection/Transcendental Proof 135 sensory experience-eall it 'transcendental chaos'-there could be no human thought, and so no human self-conscious, at all. This transcendental proof establishes a conditionally necessary constraint on the sensory contents provided to us by the objects we experience. 2o III.4. Kant next introduces a further constraint on the contents of human experience involving 'objectivity concepts'. Kant's response to Hume's analysis of causal relations has been widely misunderstood, most importantly because it has been supposed that the principles of the three Analogies of Experience are mutually independent, so that Kant's 'answer' is to be found in the Second Analogy, which requires only selective appeal to the First, and permits disregard of the Third, Analogy.21 Instead, Kant's three Analogies form a tightly integrated set of mutually supporting principles, each of which can be used only conjointly with the other two. 22 The First Analogy treats the persistence of substance through changes of state (transformations). The Second Analogy only treats rule-governed causal processes within anyone substance. Only the Third Analogy treats causal interaction between any two (or more) substances. Kant is express about this (B III; cf. KdU v. 181); hence only in the Third Analogy do we find a direct response to Hume's scepticism about our knowledge of causal powers, because only there does Kant defend a transeunt account of causality, namely that something in a causally active substance goes out beyond that substance to influence or causally affect something else; that is, to effect a change in a distinct substance. This is Kant's thesis that all physical events have external causes. These issues are complex, both exegetically and philosophically. Fortunately, I(ant's main point about the necessarily joint use of the three principles of the Analogies can be put briefly. Determining that we witness either coexistence or succession requires discriminating the one from the other, and both determinations require that we identify objects that persist through both the real and the apparent changes involved in the sequence of appearances we witness. We cannot directly perceive or ascertain either time or (according to the Third Analogy) space, and the mere order in which we apprehend appearances does not determine an objective order of
20 Thus transcendental proofs can justify conclusions much stronger than Rorty ('Strawson's Objectivity Argument', 236; 'Verificationism and Transcendental Arguments', Noils, 5 (1971), 3-14) recognizes. 21 For discussion of Kant's unjustly neglected Third Analogy of Experience see B. Jeffrey Edwards, Substance, Force, and the Possibility of Knowledge: On Kant's Philosophy of Material Nature (Berkeley, Calif.: University of California Press, 2000). 22 Guyer, Kant, 168, 212-14, 224-5, 228, 239, 246, 274-5; K. R. Westphal, 'Does Kant's Metaphysical Foundations of Natural Science Fill a Gap in the Critique of Pure Reason?', Synthese, 103 (1995),43-86, at 56- 62.
136 Westphal objects or events. Consequently, given our cognitive capabilities, we can determine which states of affairs precede, and which coexist with, which others only under the condition that we identify enduring substances that produce changes of state in one another by causal interaction. Identifying enduring substances is necessary for us to determine the variety of spatial locations objects or events occupy, to determine changes of place, and to determine non-spatial changes (transformations) objects undergo. Anyone such identification or discrimination requires joint use of all three principles defended in the Analogies. Failing to employ these principles successfully would leave us, Kant indicates, with 'nothing but a blind play of representations, i.e., less than a dream' (A 112). Kant's thesis that the principles of the three Analogies are an integrated set can be ratified by epistemic reflection, if we bear in mind certain of Hume's perplexities in 'Of Scepticism with Regard to the Senses' and certain facts Kant notes about the requirements for our distinguishing our subjective order of apprehension from the objective order of events. Apprehending the manifold features of a house is successive, though no one concedes that the features of the house are themselves successive (A 190/ B 236). Prominent among those who admit that the manifold features of perduring objects exist concurrently is Hume, who when a porter delivered him a letter recognized that the porter climbed stairs that must still exist beyond the bounds of Hume's study, and that the door to his study must still exist behind his back if he heard the porter's knock on the door and the door's squeaky hinge as the porter entered. 23 Hume's observations have manifold implications. 24 Note first that Hume's observations acknowledge that we ascribe both perduring existence and causal properties to ordinary physical objects. Second, these ascriptions require concepts that cannot be defined in accord with Hume's own concept-empiricism, namely the concepts 'cause' and 'physical object'. 'Cause' cannot be defined in accord with concept-empiricism because, as Hume notes (T 193, 196-8, cf. 632), we so often experience only a (supposed) effect (or a cause), without experiencing its cause (or effect). If Hume's official principles of psychological association were true, each time we observed only an effect but not its supposed cause, or only a cause without its supposed effect, this would weaken our belief in that supposed causal relation. Such observations are not merely common, but prevalent. Hence we could have at best only very few, very weak, beliefs about 23 David Hume, A Treatise of Human Nature, (ed.) L. A. Selby-Bigge and P. H. Nidditch, (Oxford: Clarendon Press, 1978; hereafter 'T'), 196-7. 24 See Norman Kemp Smith, The Philosophy of David Hume (London: Macmillan, 1941), 443-94; Robert Paul Wolff, 'Hume's Theory of Mental Activity', in V. Chappell (ed.), Hume: A Collection of Critical Essays (New York: Anchor, 1966), 99-128; and Barry Stroud, Hume (London: Routledge & Kegan Paul, 1977), 96 117.
Epistemic Reflection/Transcendental Proof
137
particular causal relations. This would disable our forming any general concept of cause (according to which every event has a cause). That is why Kant confidently rejects concept-empiricism and psychological principles of association and maintains that the general concept of cause is a priori (B 240-1).25 Analogous points pertain to the concept of 'physical object' (Hume's 'body'). Hume recognized that this concept cannot be accounted for by his concept-empiricism and official principles of psychological association. For that reason alone he introduced three 'propensities' to respond to certain patterns of impressions with certain beliefs about the continued and independent existence of physical objects (T 199-208). However, these propensities can only explain the occasion of the belief in 'body', they neither explain the origin of nor define the very concept of 'physical object'. Hume's propensities in fact smuggle into his account a priori concepts, concepts that cannot be defined in accord with concept-empiricism. 26 Third, Hume notes that ascribing continued existence and causal properties to physical objects outstrips our sensory observations, as Hume understands them (T 197, 198, cf. 217). Nevertheless, ascribing these characteristics to physical objects is necessary in order to preserve the coherence of our beliefs about the world (T 195-6, 198). Hume found such 'coherence' too weak to warrant trusting his senses (T 217). Hume overlooked what Kant saw; namely, the coherence of our beliefs about our surroundings is not the basic issue. At stake is their very existence, their very possibility. Without the capacity to make causal judgements we could never 'derive' the subjective order of apprehension from the objective order of the world (A 193/B 238), nor could we distinguish between our subjective order of apprehension and any objective order of things and the events in which they participate (A 193-5/B 238-9)-including those events called 'perceiving' thenl. We could not identify sensed objects at all, not even putatively; we could not identify the door on the basis of its squeak. When he received a letter by a porter Hunle clearly distinguished the subjective order in which his experiences occurred from the objective causal order of objects and events, though his epistemology cannot account for this ability. Kant's transcendental proofs concern not merely the possession of certain concepts, but their legitimate use in cognitive judgements of just these sorts. Kant recognized a crucial point of his analysis only when he wrote the preface to the Metaphysical Foundations of Natural Science (1786): we can make causal judgements only in the case of three-dimensional, spatial objects and events; we cannot identify substances contained solely in the
25
Lewis White Beck, Essays on Kant and Hume (New Haven, Conn.: Yale University Press,
197 8), 121-5· 26 I discuss these points of Hume's analysis in detail in Hegel, Hume und die Identitat wahrnehmbarer Dinge (Frankfurt/Main: Klostermann, 199 8), §4.
138
Westphal
unidimensional form of inner sense, time. Hence we can only use the categories of cause and substance with regard to spatial objects and events. 27 Therefore, if we can identify a ten1poral order of events only by using the concept of cause, and if only in that way can we distinguish the subjective order of apprehension from the objective order of events, then the objective order of objects and events we identify must be a causal order of perceptible spatia-temporal substances. Identifying an object or an event requires being able to track it through some span of time during which we observe it, where 'it' may exemplify any one of three possibilities. (I) The object may be stable spatially and undergo no transformation, though we may notice or perceive various of its properties sequentially (Kant's house; A I90/B 235). (2) The object may undergo no transforn1ation, though it moves in space (Kant's ship; A I92/B 237). Or (3) the object may undergo a transformation of one or more of its characteristics (Kant's room heated by a stove; A 202/B 247-8), regardless of whether it moves. (It may, of course, both move and transform.) Identifying anyone of these scenarios as the instance at hand requires determining that the present case is neither of the other two. To identify a spatially and transformationally stable object \ve must be able to determine that it is not moving, and not undergoing transformation. To determine that an object undergoes transformation we must determine that what appears to be a transformation is not simply a local motion revealing a previously occluded aspect of the object; we must also determine that the object does not simply move away or vanish, only to be replaced by a different object that slips into the place it formerly occupied. Each of these determinations requires that we can identify and reidentify something that perdures through the changes we perceive, whether they are real (transformations) or merely apparent (local or relative motions with respect to ourselves). Reidentification of particulars distinct from ourselves requires their spatiality and discriminating their causal interactions; hence reidentification is required even to identify particulars distinct from ourselves. 28 Here one may desire more analysis, but even then the question would remain whether this analysis holds true of us. To determine this we must reflect carefully and honestly-we must reflect epistemically-on our own capacities to identify and discriminate such objects and events. Recent analyses of the possibility of a 'pure sense-datum experience' have not matched Hume's, much less Kant's, acuity in this regard.
27 K. R. Westphal, 'Gap,' 47-9, 'Kant's Dynamic Constructions', Journal of Philosophical Research, 20 (1995), 381-429, at 386-8, and 'Kant's Critique of Determinism in Empirical Psychology', in H. Robinson (ed.), Proceedings of the Eighth International Kant Congress (Milwaukee, Wise.: Marquette University Press, 1995),2/1: 357-70. 28 On Kant's spatiality requirement see the ref. given above in n. 13.
Epistemic Reflection/Transcendental Proof
139
III. 5. These four points are intimately related. Kant claims, on the basis of epistemic reflection, that our kind of outer sense is passive; it is merely a form of receptivity to stimulation by things distinct from ourselves. 29 Our sensory experiences are inextricably temporal, and a major class of them-those of things distinct from ourselves-are inextricably spatial as well. In order to gain knowledge through sensory experience we must be able to integrate various aspects of our experience self-consciously into a whole. This is true both in the local case of identifying various manifest features by which we discriminate anyone object or event, and in the global case of identifying a (perhaps incomplete) history of our own experience of the world. Note that this thesis does not require 'unsynthesized sensations'. It is sufficiently supported by the fact that recognizing anyone object or event requires identifying several of its (synchronic and diachronic) features; e.g. Hume recognized a sound as a squeak of a door only by identifying several features of his door, not all of which he perceived concurrently. Identifying particular objects or events in our environs requires that they present us with a sufficient minimum degree of regularity and variety among sensory contents such that our understanding can function, both by developing empirical concepts and by using our a priori categories, both of which are necessary for us to identify common-sense objects or events. Finally, identifying objects or events requires that we can successfully discriminate among spatially and transformationally stable objects, moving objects (whether local or translational motion), and transforming objects (regardless of whether they also move). This requires that we can track objects and events in space through some periods of time, and across occasions of perceiving them; identification requires the possibility of reidentification. Successful discriminations of these sorts require that there are perduring, perceptible, causally interacting physical objects. Otherwise we could not 'derive' the subjective order of our perceptions from any objective order of objects and events (A 193/B 238). In such a case we would have at most a flood of sensation, affording us even less than a dream. In this way Kant aims to justify a point succinctly put by Will: Thinking is an activity which we engage in not only in the world of things, but by means of things in this world, supported and sustained by them. 3o
Our cognitive dependence on our circumstances, on the world in which we live, runs deep, so deep we can all too easily overlook it. Kant did not, and he pointed out that these dependencies are vastly more complex than modern empiricists (or their recent avatars) assumed.
29 K. R. Westphal, 'Noumenal Causality Reconsidered: Affection, Agency, and Meaning in Kant', Canadian Journal of Philosophy, 27 (1997), 209-45, at 224, 226-7. 30 Frederick L. Will, Pragmatism and Realism (Lanham, Md: Rowman & Littlefield,
1997), 14·
Westphal IV. CONCLUSION
I must concur with Strawson's observation of some thirty years ago that 'nearly two hundred years after they were made [Kant's key insights] have still not been fully absorbed into the philosophical consciousness'.31 Kant's epistemic reflections on human knowledge, and our cognitive capacities, are designed to enable us to realize that perceptual seemings (or, analogously, pieces of linguistic behaviour) are not self-sufficient conscious episodes. We cannot address this issue second-person, and we cannot address it for our own case from a third-person standpoint. We can address it only by reflecting very carefully on Kant's analysis of the conditions of possible self-conscious hun1an experience, and on our own manifold cognitive dependencies on the world in which we live. The real power of Kant's transcendental proofs is found through Kant's epistemic reflection, not in the senlantic assent characteristic of analytic reconstructions of Kant's alleged arguments. Rorty is right that this approach to transcendental arguments can only yield much weaker 'parasitism arguments' that can at most point out interconnections between certain pieces of linguistic behaviour, but cannot establish epistemic conclusions about the truth-value of any basic beliefs that may be linguistically expressed. 32 By taking as basic beliefs about ourselves or our world, or bits of linguistic behaviour, recent discussions of Kant's epistemology too often have bypassed Kant's radical break with the Cartesian tradition, because they have failed to address not only Kant's answer but his key question: 'On what ground rests the relation of that in us which is called representation to the object?' (Letter to Herz, x. 130; cf. A 197/B 242). Only careful epistemic reflection can give us a witting, rather than unwitting,33 understanding of our capacity for empirical knowledge. 34
31 Bounds of Sense, 29. In personal correspondence (I May 1999) he confirmed that his statement remains true. Elsewhere Strawson (Subject and Predicate, 14-15) identifies a tenet of empiricism that, he claims, is 'a strand in a tradition which, since Kant, embraces us all'; namely, that our most basic concepts 'enter most intimately and immediately into our common experience of the world. They are what ... we experience the world as exemplifying ... experience is awareness of the world as exemplifying them'. (Why this counts as a tenet of empiricism, he does not explain.) In Entity and Identity and Other Essays (Oxford: Clarendon Press, 1997; 232-43) Strawson also sketches a number of Kantian elements in contemporary philosophy. However, none of these Kantian elements concern's Kant's core issues presently under discussion. 32 Rorty, 'Verificationism and Transcendental Arguments,' 4-5, 9; cf. Christopher Hookway, 'Modest Transcendental Arguments and Sceptical Doubts: A Reply to Stroud', in R. Stern (ed.), Transcendental Arguments: Problems and Prospects (Oxford: Oxford University Press, 1999), 173-8 7. 33 Cf. Stroud, 'Understanding Human Knowledge in General', in M. Clay (ed.), Knowledge and Scepticism (Boulder, Colo.: Westview, 1989), 31-50, at 47. 34 I thank Robert Howell for prompt and helpful comments on a preliminary draft of these remarks, and Robert Greenberg for some further thought.
8 Kanfs Self: Real Entity and Logical Identity TOBIAS ROSEFELDT
To a certain extent Kant is a pretty good Strawsonian. In much of what he says about self-consciousness he seems to agree with Strawson's claim that to be self-conscious one must conceive of oneself as a corporeal object among corporeal objects.! At least he is saying that we cannot have knowledge of ourselves as individual thinking beings in so far as we regard ourselves as merely thinking beings, but rather only if we refer to ourselves as thinking human beings, i.e. beings whose bodies provide empirically applicable criteria for their identity through time. 2 Unfortunately, Kant destroys this harmonious picture by making a couple of rather obscure remarks about something he calls 'the constant and abiding I (of pure apperception)' (A 123) or 'the identical self' to which I ascribe all my representations (A 129/B 138), and of which I have an 'a priori consciousness' (A II6). These remarks are obscure because the phrase 'a priori consciousness' makes clear that the object of such a consciousness ('the identical self') cannot be identified with the object of empirical self-knowledge, i.e. the thinking human being. On the other hand, Kant denies the claim, maintained in rational psychology, that self-consciousness includes a priori knowledge of ourselves as immaterial, persisting, and non-composite thinking substances. So the object of a priori self-consciousness cannot be identified with such a thinking substance either. When Kant himself tries to explain what 'the self' as the object of a priori self-consciousness is, and what it is not, he frequently makes use of the
1 P. F. Strawson, The Bounds of Sense: An Essay on Kant's Critique of Pure Reason (London: Methuen, I966), I02. 2 Cf. Kritik der reinen Vernunft, B 415/A 361 ff. and Prolegomena zu einer jeden kunftigen Metaphysik, die als Wissenschaft wird auftreten konnen, § 48, in I., Kant, Gesammelte Schriften (GS), herausgegeben von der Koniglich PreuRischen Akademie der Wissenschaften (Berlin, G. Reimer [now de Gruyter] I902-), iv. 335; my translation.
142
Rosefeldt
distinction between logical and real features of the self, or, as he sometimes puts it, between features of the self as a real entity and the self as a n1erely logical entity, the so-called 'logical 1'.3 He claims that we can only know that the self is the 'permanent logical subject of thinking', not that it is a 'real subject of inherence' (A 350); we can know that it is a 'logically simple subject', but not that it is a simple substance (B 407-8). He speaks of the 'logical identity of the l' in opposition to the numerical identity of a real thinking being (A 363), and he says that there is only a 'logical' not a 'real' possibility for us to exist as thinking non-corporeal beings. 4 In this chapter I shall present an interpretation of the role of the term 'logical' in Kant's theory of self-consciousness. 5 I will limit myself to an explanation of Kant's use of the expressions 'logical subject' and 'logical identity' and to some remarks on his general distinction between real and logical entities. I will also address Kant's theory of concepts and singular judgements, because I think that only within the framework of this theory can Kant's claims be made intelligible. To conclude my paper I shall propose that nothing Kant says contradicts Strawson's insights into the essential role that the reference to our own bodies plays for self-consciousness, but rather that in some things that Kant says we can find insights that even go beyond those of Strawson's own theory.
I AS AN ABSOLUTE LOGICAL SUBJECT OF JUDGEMENTS
The Paralogism of Substantiality in the first edition of the Critique of Pure Reason reads as follows: That, the representation of which is the absolute subject of our judgen1ents and cannot therefore be employed as determination of another thing, is substance. I, as a thinking being, am the absolute subject of all my possible judgements, and this representation of myself cannot be employed as predicate of any other thing. Therefore I, as a thinking being (soul), am substance. (A 348)
Later Kant says that though I, as a thinking being, am the 'permanent logical subject of thinking', I do not therefore know that I am a 'real subject of inherence' (A 350). These remarks leave us with three questions. What is an absolute subject of judgements? Why am I, as a thinking being, an absolute subject of judgements? And why am I only an absolute logical subject of judgements not a real subject of inherence? 3 The term 'logical I' is used by Kant in his late essay Welches sind die wirklichen Fortschritte~ die die Metaphysik seit Leibnizens und Wolff's Zeiten in Deutschland gemacht hat? (GS xx. 270). 4 Ibid. (GS xx. 3 2 5-6). 5 I give an extensive interpretation of Kant's conception of the 'logical I' in my book Das logische Ich. Kant aber den Gehalt des Begriffes von sich selbst (Berlin: Philo, 2000).
Kant's Self: Real Entity and Logical Identity
143
To help answer the first question I will cite a passage from Kant's Prolegomena: Pure reason requires us to seek for every predicate of a thing its proper subject, and for this subject, which is itself necessarily nothing but a predicate, its subject, and so on indefinitely (or as far as we can reach). But hence it follows that we must not hold anything, at which we can arrive, to be an ultimate subject ... for the specific nature of our understanding consists in thinking everything discursively, that is by concepts, and so by mere predicates, to which therefore the absolute subject must always be missing. 6
From this passage it should be clear that by 'subject of a judgement' Kant means a thing of which we can say something in a judgement. The subject of a certain judgement is the thing that is represented by the concept used as the subject-concept in this judgement. Hence, an absolute subject of judgements would be something that is represented by a concept that only can be used as a subject-concept of a judgement but not as a predicate. Because, according to Kant, all concepts can be employed as predicates of possible judgements (A .69/B 94), we cannot represent absolute subjects of judgements by means of concepts. In the first premiss of the Paralogism of Substantiality I(ant makes another important remark on the absolute subjects of our judgements. He says that substance is son1ething the concept of which cannot be 'employed as determination of another thing'. In order to understand this ren1ark one has to take into consideration the logical theory delineated by Kant's predecessors. According to pre-I(antian logic, a concept that cannot be employed as determination of another thing would be a 'singular concept'. A singular concept is the kind of concept Leibniz calls 'a complete concept of an individuum', i.e. a concept with a content so rich that there is only one possible object to which it can be applied. Singular concepts refer to individual monads, and these monads are substances, for they are the ultimate and simple parts of which everything else is composed. In the proof of the 'thesis' of the second Antinomy Kant calls these monads 'elementary substances', which are the 'first subjects of all composition' (A 436/B 464).7 Prolegomena § 46 (GS iv. 333). Kant's use of the term 'substance' is ambiguous, for he uses it as a mass noun (in which case it has a similar meaning to 'matter'), as well as an articulative and pluralizable noun (in which case it refers either to an elementary substance or to a thing which is composed of such elementary substances, i.e. a composite substance; cf. A 434/B 462). Although Strawson may be right in claiming that for Kant's discussion of substance in the Analogies of Experience the first meaning is the relevant one (cf. 'Kant on Substance', in P. F. Strawson, Entity and Identity and Other Essays (Oxford: OUP, 1997), 268-79, at 279), I think it is obvious that in the chapter on the 'paralogisms' by 'substance' Kant means an elementary substance. The judgement that I, as a thinking being, am a simple substance, for example, would not make any sense if 'substance' were taken as a mass noun in this judgement. This interpretation is also supported by A 441-21 B 469-70, where Kant says that by the expression 'monad' Leibniz means something 'that is immediately given as a simple substance (e.g. in self-consciousness)' (cf. also A 443/B 471). 6 7
144
Rosefeldt
So Kant's claim that we do not have concepts that we can only use as subject-concepts in our judgements and never as 'determination of another thing' is equivalent to the claim that we do not have singular concepts, a claim he makes in many of his remarks on logic. 8 If we want to make a judgement about a certain individual thing, we cannot do so by employing a singular concept, for all concepts we have are general concepts. We must rather make what Kant calls 'a singular use' of a general concept. 9 The singular use of general concepts is possible, because the use of concepts is not the only means we have to refer to objects. Objects are given to us in intuition and an intuition is always a representation of a singular object (A 320/ B 376-7). The important point is, then, that applying a concept to an object given in intuition does not make this concept a singular concept, i.e. a concept we cannot apply to 'another thing'. If I point to a rose and say 'This rose is red' and then point to another and say 'This rose is white' I am making two judgements about two numerically distinct roses, although the subjectconcept of my judgement is the same in both cases. I can do this because, with regard to objects of intuition, Kant says, 'we do not concern ourselves with comparing the concepts, but, although they may be in this respect perfectly the same, the difference of their places at the same time is a sufficient ground for asserting the numerical difference of these objects (of senses)' (A 263/B 319). This means that I can distinguish between two numerically distinct objects by their position in space, although I might not be able to distinguish between them by means of the concepts I apply to them. Hence, although I can refer to a particular rose (the red one, say) using the concept 'rose' as the subject-concept of my judgement, the concept 'rose' is not the concept of an absolute subject, for it can also be predicated of 'another thing' (e.g. the white rose). This situation would not change if I were able to give a more detailed description of the rose about which I intend to speak. For even in the case of a concept like 'the pale red rose with the sweet smell, which has such-and-such a shape, etc.' it is always possible that there is another rose at another place to which I could apply this concept as well. That is to say that no concept-however rich its content may be--ean be such that it cannot be 'employed as determination of another thing'. The second question was why I, as a thinking being, am an absolute subject of judgements, the representation of which cannot be employed as predicate of any other thing. The passage from the Prolegomena cited above continues: Now we appear to have this substance in the consciousness of ourselves ... for all the predicates of inner sense relate to the I as subject, and this cannot further be thought as the predicate of any other subject.
8 Cf. e.g. Wiener Logik (GS xxiv. 910-11), Logik Politz (GS xxiv. 569),jasche-Logik, § II n. (GS ix. 97); and M. Thompson, 'Singular Terms and Intuitions in Kant's Epistemology', Review of Metaphysics, 26/2 (197 2 ), 314-43. 9 Cf. Logik Politz (GS xxiv. 567) and Wiener Logik (GS xxiv. 908).
Kant~s
Self: Real Entity and Logical Identity
145
Kant's point here seems to be the following: the representation we have of ourselves as thinking beings, i.e. the concept 'I, as a thinking being', fulfils the criterion for concepts of absolute subjects of judgements in so far as we cannot apply it to different possible objects. 'I' is a concept that I can apply only to one possible object, i.e. myself. This interpretation is supported by a passage from Kant's Lectures on Metaphysics, in which he formulates the rational psychologists' argument for the substantiality of the soul as follows: I am a substance. The I refers to the subject in so far as it is not the predicate of another thing. That which is not the predicate of another thing is a substance. The I is the common subject of all predicates, all thinking, all actions, all possible judgements which we can make about ourselves as a thinking being. I can only say: I am, I think, I act.... but I cannot predicate the I of something else, I cannot say: another being is the 1. 10
Although these remarks are not formulated very clearly, it should be apparent that the reason why we regard ourselves as thinking substances is a peculiarity of the representation we have of ourselves as thinking beings. I cannot apply the concept'!' to any being other than myself in the same manner as 1can apply the concept 'rose' to different roses. Does this mean that'!' is a singular concept in the pre-Kantian logicians' understanding of this term?11 Kant points out two fundamental differences. The first is that a singular concept is a concept with a content so rich that there is only one possible object to which it can be applied. Now it is true that I can only apply'!' to one possible object, i.e. myself, but this is not because the content of 'I' consists of a bundle of qualities that only I happen to have. On the contrary: 'I' is an expression that is 'completely empty with regard to its content' (A 355), i.e. its content is not richer than that of a concept like 'a thinking being = x' (A 346/B 404). The second difference between 'I' and a singular concept is that the description 'can only be applied to one possible object' is only true of the concept'!' if it is completed by the restriction 'by n1yself'. For though it is true that I can only apply the concept 'I' to one possible object, i.e. myself, it can nevertheless be applied to other possible objects by other thinking beings, if they apply it to themselves. Kant hints towards this characteristic of the concept'!' by formulating the second premiss of the Paralogism of Substantiality in a slightly different way from the first. The first premiss states that a substance is an absolute subject
Vorlesungen uber Metaphysik (Ll nach Politz) (GS xxviii. 266). Kant seems not to be sure whether that which is expressed by the first-person pronoun 'I' should be called a concept at all. Although there are some passages in which he calls'!' a concept (cf. e.g. A 4°1, A 342/B 400, A 341/B 399), there are others in which he explicitly refuses to do so (cf. e.g. A 345-6/B 403-4, A 382); for a detailed discussion of this issue see Rosefeldt, Das logische Ich, 15-22. 10 11
146
Rosefeldt
'of our judgements', while the second only claims that I, as a thinking being, 'am the absolute subject of all my possible judgements'. Therefore, Kant says, it is only true that 'everybody has to conceive of himself as a substance' (A 349), for'!' is a concept that he can apply only to one thinking being, i.e. himself, but not to 'another thing'. This last remark obviously relates to Kant's claim that although'!' is in one respect similar to a singular concept, which cannot 'be employed as determination of another thing', I nevertheless do not, by representing myself by this concept, know that I, as a thinking being, am a 'real subject of inherence', i.e. a substance. So let us now come to our third question: what does Kant mean by saying that I, as a thinking being, am only an absolute logical subject of judgements and not a real subject of inherence? Before I answer this question I would like to make some general remarks about Kant's distinction between logical and real features of an object, or, as one could also put it, between features of merely logical and those of real entities.
LOGICAL PREDICATE AND LOGICAL SOMETHING
The chapter on the Paralogism of Pure Reason is not the only text in which Kant makes use of the distinction between a concept understood in its 'logical' and in its 'real' meaning. Aside from making a distinction between a logical and a real subject, between" logical and real simplicity, logical and real identity, and logical and real possibility, he also distinguishes between logical and real essence,12 logical and real negation,13 a logical and a real reason (Grund),14 as well as logical and real necessity.1 5 The best known of these distinctions is that made between a logical and a real predicate in Kant's refutation of the ontological proof for the existence of God. Since this distinction seems to be the counterpart to the distinction between a logical and a real subject, I shall give a short reminder of what a logical opposed to a real predicate is.1 6 Existence, says Kant, is only a logical, not a real, predicate. It is a logical predicate because it can be predicated of a thing, i.e. the concept 'existence' can be used as a predicate-concept in a judgement, e.g. in a judgement like 'God exists'. It is not a real predicate because it is not the property of a thing, but rather concerns the relation between a concept and an object.
Cf. e.g. Vorlesungen uber Metaphysik (L2 nach Politz) (GS xxviii. 552-3). Cf. e.g. Versuch den Begriff der negativen Grof5en in die Weltweisheit einzufuhren 14 Cf. e.g. ibid. (GS ii. 203). (GS ii. 127). 15 Cf. e.g. Vorlesungen uber Metaphysik (L2 nach Politz) (GS xxviii. 557-8). 16 For the following cf. A 598-9/B 626-7' 12 13
Kan{'s Self: Real Entity and Logical Identity
147
In a judgement like 'God exists' we claim that there is an object to which the concept 'God' corresponds, but not that there is an object, God, which has the property of existence. Because existence is not the property of a thing, the concept 'existence' cannot be part of the concept of a thing, and, hence, a judgement like 'God does not exist' can never entail a conceptual contradiction. From this we can conclude that a logical predicate is everything that can be predicated of a thing, i.e. everything that can be represented by the predicate-concept of a judgement, while a real predicate represents a property that things can have or not have. Since not everything that can be . predicated of a thing in a judgement is a property of that thing, not every logical predicate is also a real one. How does this help us with Kant's distinction between a real subject of inherence and an absolute logical subject of judgements? If we ignore the attribute 'absolute' for a while, we can transfer what Kant says about a real and a n1erely logical predicate in the following way: a logical subject is everything of which we can predicate something, i.e. everything the concept of which we can use as the subject-concept in a judgement, while a real subject is something in which properties inhere, a substance. I(ant's distinction between logical and real subjects would then result from the fact that not everything of which we can predicate something in a judgement is a thing in which properties inhere. This seems like a rather sensible thing to say, for we can even make judgements about properties, such as 'Red is my favourite colour', although properties like the colour red are by definition not substances. In one of his Lectures on Metaphysics Kant calls something about which we can make judgements although it is not a real subject a 'logical something'. 'By "something" " he says, 'we mean every object of thinking; this is the logical something. The concept of an object in general is called the highest concept of all knowledge. Such an object is also called a sOlnething, but not a metaphysical, but only a logical, something.,17 The distinction between a logical and a metaphysical something obviously corresponds to that between a logical and a real subject. At the end of the Transcendental Analytic of the Critique of Pure Reason Kant makes a similar claim. Among the things that fall under the concept of an 0 bject in general he distinguishes between objects that are 'something' (in the narrow, or 'metaphysical' sense of 'something') and those that are 'nothing', and hence, one could conclude, only something in the logical sense of 'something' (A 29 0 / B 346-7). Exaruples for objects of the latter kind are, among others, the so-called
17 Vorlesungen aber Metaphysik (L2 nach Politz) (GS xxviii. 544). In his essay 'Entity and Identity' Strawson makes a son1ewhat similar claim by distinguishing between entities which are 'substantial objects' and those which are objects only in the sense that they are 'the subjects of indispensable first-order predication' (in Strawson, Entity and Identity, 21-5 I, at 34).
148
Rosefeldt
'objects of mere thought' (A 292/B 348), i.e. 'object[s] of concept[s] to which no corresponding intuition can be given' (A 290/B 347). Kant's most common example for such an 'object of mere thought' is a spirit, i.e. a thinking non-corporeal being. 18 The concept of a thinking non-corporeal being ('spirit') is not contradictory, and hence we can make judgements about spirits such as 'The n10st intelligent spirit is always just'. But, since we cannot give examples of thinking non-corporeal beings from experience, we do not know whether the concept 'spirit' refers 'to a corresponding object, or is empty' .19 This example of a merely 'logical something' suggests a way in which one could understand Kant's claim that I, as a thinking being, am only a logical subject of judgements. In his Lectures on Metaphysics Kant distinguishes two ways in which we can understand the concept '1'. With '1' understood in sensu latiori, he says, I refer to 'myself as the whole human being with a soul and a body', while with '1' taken in sensu stricto I refer to myself solely as a sou1. 20 Now one could argue that as far as I use the concept '1' in sensu stricto (as rational psychology does) this concept represents only a 'logical son1ething'. I, who use this concept, am not a non-corporeal soul, but a thinking human being. So I have no intuition of myself that will be that of a non-corporeal soul, and hence the concept'!', understood in sensu stricto, remains as empty as the concept 'spirit'. This interpretation provokes a serious objection. According to it, Kant's argument for the claim that I, understood as a thinking non-corporeal being, am only an absolute logical subject of judgement, not a real subject of inherence, would start from the premiss that I am not a thinking noncorporeal being, rather than have this claim as its conclusion. The rational psychologist would probably deny this premiss. Through inner sense, he could argue, I have an intuition of myself as a thinking non-corporeal being. Hence, Kant is in need of an argument for the claim that, although through inner sense we have intuition only of our mental properties, this intuition nevertheless cannot constitute knowledge of ourselves as real subjects of inherence, i.e. as substances. In my opinion, Kant presents such an argument in his discussion of the Third Paralogism. There he argues that a necessary condition for knowing of an object of intuition that it is a substance is the capacity to reidentify this object at different moments in time. But, as long as we regard ourselves solely as objects of our inner sense, we do not have any criteria by which to decide whether two properties of which we have intuition through inner sense belong to one and the same or to two
18
19 20
Cf. A 96 and Welches sind die wirklichen Fortschritte (GS xx. 325). Ibid. (GS xx. 3 26). Vorlesungen aber Metaphysik (L1 nach Politz), GS xxviii. 265; see also ibid. 224-5.
Kant~s
Self: Real Entity and Logical Identity
149
numerically distinct thinking beings. Again, Kant makes use of the 'logicaV real' distinction in order to explain why it is so easy for the rational psychologist to miss this point. He says that if I represent n1yself by the concept'!', understood in sensu stricto, I am only aware of the 'logical identity of the l' (A 363), but not of the numerical identity of myself as a real thinking being.
LOGICAL IDENTITY
Before I turn to Kant's notion of logical identity, I want to make some short remarks on how pre-Kantian logic would have accounted for our ability to ascribe different properties to one and the same thing. As I said before, preKantian logic explained singular judgements, i.e. judgements about a particular individual, as the use of singular concepts, i.e. concepts that can be applied to only one possible object. So if in two judgements I combine one and the san1e singular concept with two different predicates, I ascribe the two properties that are represented by the predicates to one and the same object. I claimed earlier that Kant denies the existence of singular concepts. This claim was not completely correct, for there is one concept that Kant does treat as a singular concept, namely the concept 'God', as the concept of the 'ens realissin1um'. 'In this case alone', he says, 'the concept of a thing is completely determined by and through itself, and known as the representation of an individuu1l1' (A 576/B 604). This means that the content of the concept 'God' is so rich that it is logically impossible for there to be two objects to which this concept could be applied. Kant also makes clear that the concept 'God' cannot be used in experience but that it is only a 'transcendental ideal', a 'concept of pure reason', which has no constitutive but rather only regulative use (A 619-20/B 647-8). Since we cannot have any intuition of the object represented by the concept 'God', one could say that, for us, GOd is only a 'mere object of thought' and hence a merely 'logical something'.21 Since the content of the concept 'God' is so rich that it can only be applied to one possible object, and it is a concept to which no corresponding intuition can be given, God is a good candidate for what Kant calls an 'absolute logical subject of judgements'. I do not want to go into any further details regarding Kant's remarks on the concept 'God', but am only using this concept here as an example for a singular concept, which may help to explain how the use of singular concepts could account for our
21 According to Kant all objects of transcendental ideas are mere 'objects of thought' (cf. A 337/B 394).
Rosefeldt ability to ascribe different properties to one and the san1e thing. Suppose I make the following two judgements: 150
(I )
(1~l-)
God is sometimes strict. God is always just.
Because 'God' is a concept that can be applied to only one possible object, it is clear that I ascribe the property of being sometimes strict and the property of being always just to one and the same object. Since 'God' is a concept to which no corresponding intuition can be given, one could say that I ascribe two properties to one and the same 'logical son1ething'. This means that, if I hold the judgements (I) and (1~:') to be true, I will also hold the judgement (1~:'~l-)
God is sometimes strict, but always just.
to be true. In Kant's opinion, concepts that we can apply to objects that are given to us in intuition (i.e. concepts of objects that are not only 'logical somethings') are not singular, but rather general concepts. If I want to use a general concept to make a singular judgement, I have to apply this concept to an object that I can discriminate from other objects by its position in space. This has consequences with regard to our ability to ascribe different properties to one and the same object of experience. Suppose I walk through my garden and make the following two judgements: (2) This rose bloomed yesterday. ( 2~l-) This rose is withered today.
In this case, the question whether I ascribe the property of having bloomed yesterday and the property of being withered today to one and the same or to two numerically distinct objects, cannot be decided by means of an analysis of the concept 'rose' alone. Since 'rose' is a concept which can be applied to more than one possible object, it is possible that I hold (2) and (2~l-) to be true, but do not hold the judgement (2~:'~:')
This rose bloomed yesterday, but is withered today.
to be true. This would be the case if by 'this rose' in (2) and in (2~l-) I meant two numerically distinct objects of my intuition, two objects that I had discrin1inated by their position in space. Hence, in order to ascribe the property of having bloomed yesterday and the property of being withered today to one and the same rose, I have to make sure that by 'this rose' in (2) I refer to a rose that is at the same place at the same time as the rose I refer to by 'this rose' in (2~l-).
Kant's Self: Real Entity and Logical Identity
151
Let us now turn to the judgements I make about myself as a thinking being. In a way these judgements are similar to those I make about God. Suppose I make the following two judgements: (3) I saw a blooming rose yesterday. (3 ~:.) I see a withered rose today.
The concept'!' is a concept that I can apply only to one possible object, i.e. myself. For that reason it is clear that I ascribe the property of having seen a blooming rose yesterday and seeing a withered rose today to one and the same object, i.e. myself. If I am aware of this feature of the concept'!', it is impossible that I hold (3) and (3 ~:.) to be true, but do not hold the judgement: (3~:'~l-)
I saw a blooming rose yesterday, and see a withered rose today.
to be true. I think that exactly this aspect of judgements in which I use the concept 'I' as a subject-concept is what Kant is referring to when he says that in these judgements I am aware of the 'logical identity of the 1'. In his discussion of the Paralogism of Personality he claims that in such judgements I 'relate each and all of my successive determinations to the numerically identical self' (A 362). Therefore, he continues, the claim that I am the identical thing to which all of these determinations belong 'has to be regarded as a completely identical proposition of self-consciousness in time, and this is also the reason why it is valid a priori' (ibid.). Kant uses the expression 'identical proposition' as a synonym for the term 'analytical proposition' .22 So his claim is that solely by the content of the concept'!' it is guaranteed that if I combine this concept with two different predicates I ascribe the properties that are represented by these predicates to one and the same thing. In this respect those judgements about myself are akin to those about God and different from those about roses. Why does Kant say, then, that by means of the content of the concept'!' I am aware only of the 'logical identity of the 1', and not of the numerical identity of myself as a real thinking being? In order to answer this question one has to take into consideration the differences between the concept'!' and real singular concepts, which I mentioned in the first section of this paper. According to pre-Kantian logic, a singular concept is a complete concept of an individuum. It entails every predicate that could be truly ascribed to that individuum. Hence, knowledge of this complete concept would enable us to decide whether or not a given judgement about that individuum is true. In this respect, the concept'!' is entirely different from a singular concept, for,
22 Cf. jasche-Logik § 37 (GS ix. (GS xx. 322).
I I I)
und Welches sind die wirklichen Fortschritte
152
Rosefeldt
as I have already mentioned, it is 'completely empty with regard to its content' (A 355). So judgements in which I ascribe a certain property to myself are never analytical judgements, (B 135), and hence I cannot decide whether a judgement like (3~·~:·) is true by means of an analysis of the concept'!'. In this respect, then, the concept'!' is different from a concept like 'God' and akin to a concept like 'rose'. If I want to make verifiable judgements about properties that one and the same object of intuition has at different tin1es, I can do so only if this object is a corporeal object, for only in such a case do I have the means to reidentify the object at different times. 23 So if I want to make verifiable judgements about n1yself as a thinking being, I can do so only if I regard myself as a thinking human being, i.e. if I use the concept 'I' not in sensu stricto, but rather in sensu latiori. If in judgements like (3) and (3~:-) I use the concept'!' in sensu stricto it is still true that in these judgements I ascribe two properties to one and the same thing, i.e. it is still true that if I hold (3) and (3 ~:.) to be true I will also hold (3 ~:-~.) to be true. But, since in this case I have no means to verify this judgement, for I do not have any criteria for the identity of non-corporeal objects of intuition, the thing to which I ascribe the two properties is only a 'logical something', and my knowledge that it is one and the same logical something to which I ascribe different properties is only an awareness of the 'logical identity of the 1'. I will have this awareness of the 'logical identity of the I' in the case of every property that I ascribe to myself as a thinking being, for by the concept'!' I will always represent myself as an 'absolute logical subject of judgements', i.e. something the concept of which I cannot apply to any other being than myself. If we are aware of this feature of the concept'!', we have, as Kant claims in the Transcendental Deduction of the Categories, 'an a priori consciousness of the complete identity of ourselves in respect of all representations which can ever belong to our knowledge' (A 116). He also says that this consciousness a priori of my own identity can only be found 'in my own consciousness', but need not be accepted 'from the point of view of an outside observer' (A 362), for'!' is a concept that every thinking being can only apply to itself. Hence, in order to refer to me, an outside observer could not do so by applying the concept'!' to me, but would have to use some normal general concept like 'thinking being' or 'person'. Since he could apply these concepts to different thinking beings, his use of these concepts is not connected to any awareness of the 'logical identity' of the objects of these concepts.
23 According to Kant I can find out whether a judgement like (2**) is true because there are criteria for the identity of corporeal objects through time, which are given by their 'permanence' (cf. A 361-2). If nobody changed the position of the roses in my garden overnight, I can conclude from the fact that the rose that bloomed yesterday is at the same place as the rose that is withered today that it is one and the same rose that bloomed yesterday and is withered today.
Kant~s
Self: Real Entity and Logical Identity
153
CONCLUSION
At the beginning of this chapter I claimed that Kant is a good Strawsonian since nothing that he says contradicts Strawson's insights into the essential role that the reference to our own bodies plays with regard to selfconsciousness. Kant would agree with Strawson that we can only make verifiable judgements about ourselves as thinking beings if these judgements are judgements about ourselves as thinking hunlan beings, i.e.-in Kant's terminology-if in those judgements the concept'!' is used in sensu latiori not in sensu stricto. However, in a way Kant's remarks on a priori self. consciousness and the 'logical identity of the I' go beyond Strawson's insights, for they provide us with arguments supporting the claim that verifiable judgements about ourselves are not the only meaningful judgements we can make about ourselves. Even if I use the concept'!' in sensu stricto, I can ascribe different properties to one and the same thing, i.e. myself, although in this case I have no means to decide whether this ascription is correct or not. In this case-to use Kant's terminology-I ascribe different properties to one and the same 'logical something', i.e. myself as a logical entity, and by doing so I am aware of the 'logical identity of the 1'. To make clear what this could n1ean, let n1e give you an example: Suppose you have a serious car accident and you are brought to hospital. Initially, you fall into a deep coma, but after a while you regain some of your cognitive capacities: you are still stripped of every sensory awareness of corporeal objects, i.e. you neither see, hear, feel, taste, or smell anything, nor do you have any sensory awareness of your own body, but you can remember things now and do have thoughts. After a while you arrive at the conclusion that you must have died in the accident and that you now have started your afterlife as a soul in heaven. Suppose you have the following two thoughts, then: (4) I didn't know what was going on at first. (4~:-)
Later I found out that I was in heaven.
The first thing to say about such thoughts is that their truth-conditions still depend on the question of whether you, the injured person who lies in hospital, did not know what was going on at first, but later found out that you were in heaven. Therefore (4~:-) is wrong, for you are not in heaven but in a hospital, and hence you cannot find out that you are in heaven. The second thing is that you do not have any means to verify your judgen1ents, because they entail claims about your past and there are no means to verify statements about one's own past unless one regards oneself as a corporeal and perduring object. However, the third and crucial point about the given example is that in the cited judgements you still ascribe two properties to one and the same object, i.e. yourself as a 'logical something'. You do so
154
Rosefeldt
because-if you are still aware of the fact that you can only apply the concept'!' to one possible object, i.e. yourself, then-if you hold (4) and (4~·) to be true, you will also hold the judgement (4~:-~:·)
1 didn't know what was going on at first, but later found out that 1 was in heaven.
to be true. You would also still know that a judgement like '1 know that I am in heaven and 1 doubt that 1 am in heaven' is contradictory (which it would not be if you did not ascribe the property of knowing to be in heaven and the property of doubting to be in heaven to one and the same 'logical something'). Hence, you are still aware of what Kant calls the 'logical identity of the 1'.24
24 In my opinion, the case is also a counter example to what Quassim Cassam calls the 'identity argument for materialism about self-consciousness' (cf. his Self and World (Oxford: Clarendon Press, 1997), 117 ff.). According to Cassam, in order to ascribe different representations to oneself as their identical subject one must either conceive of oneself as a physical object or have an intuitive awareness of oneself as a physical object, for otherwise one would not fulfil the so-called discrimination requirement, i.e. could not distinguish the object of one's self-ascriptions from all other things (cf. ibid. 123). Kant would agree with Cassam as far as consciousness of oneself as the real thinking being is concerned, for if I use the concept'!' in sensu stricto it is 'not connected with the slightest intuition by which I could distinguish between myself and other objects of intuition' (A 350). However, Kant would say, in such a case I still could distinguish between myself regarded as a 'logical something', and other 'logical somethings', in so far as I would still know that a judgement like 'I am a finite being, and God is an infinite being' is not contradictory (which it would be if I could not distinguish between myself and God). And I also still could ascribe different properties to one and the same logical something, for I would still know that a judgement like 'I am a finite being and 1 am an infinite being' is contradictory.
9 Kant and Strawson on the First Person MAXIMILIAN DE GAYNESFORD
1. THE 'RESIDUAL CARTESIANISM' DIAGNOSIS
Peter Strawson thinks it the 'veriest truism' and an answer to a question that no one with 'ordinary linguistic competence' and ordinary philosophical innocence would even think of asking, that the very understanding of I requires the understanding of it as referring to a subject which is also a corporeal being among others in an objective world, namely, a human being. 1 Call this the Corporeality thesis. There are answers to three philosophically controversial questions here. (a) Does I refer? (b) If it refers, what does it refer to? (c) If it refers to corporeal objects, of what kind are those objects? Since Strawson argues for this view in several celebrated passages of his commentary on Kant's first Critique, The Bounds of Sense,2 we are encouraged to ask quite what support can be found for it in Kant, in Strawson's interpretation of Kant, and in what Strawson exploits of Kant. One way to make such questions sufficiently precise as to be useful is to ask why Kant's own answer diverges. My concern is to raise certain doubts about one influential diagnosis that has deep roots in The Bounds of Sense and is fully articulated by John McDowell, with fulsome acknowledgement to Strawson, in Lecture V of his Mind and World. 3 That Kant does diverge radically from the opinion of Strawson's ordinary person is clear. As regards (a), Kant distinguishes between various uses of I, particularly those expressive of the transcendental unity of apperception and those used to frame empirical statements, but is wary of claiming outright that any use at all refers. And as regards (b) and (c) Kant claims that 1 'Reply to McDowell', in L. E. Hahn (ed.), The Philosophy of P. F. Strawson (Chicago and 2 (London: Methuen, 1966). LaSalle, Ill.: Open Court, 1998), 147. 3 (Harvatd: Harvard University Press, 1994).
156
Gaynesford
the use of I made significant by the Transcendental Deduction-i.e. the '1 think' by which representations are, interdependently, both self-ascribed and given objective reference-neither expresses nor refers to a corporeal object or substantial identity but is merely 'a formal condition of my thoughts and their coherence'.4 Moreover, as Strawson acknowledges, Kant failed to affirm that one's empirical uses of I refer to the human being one is. 5 One influential explanation of why Kant diverges, lent credence by Strawson's own interpretation, is residual Cartesianism: Kant failed quite to dig out the roots of Cartesian error in the Paralogisn1s and fell victim to certain illusions fostered by that position. Strawson explains Kant's deviation as follows: in the Paralogisms Kant fails to 'press home' that 'unparalleled insight' that makes his treatment of the subject 'so greatly superior to Hume's':6 namely, his adoption of a thesis Strawson. labels 'criterionless self-ascription' and formulates precisely as follows: When a man (a subject of experience) ascribes a current or directly remembered state of consciousness to himself, no use whatever of any criteria of personal identity is required to justify his use of the pronoun I to refer to the subject of that experience. 7
Strawson uses a stronger forn1ulation when repeating the point: the immediate self-ascription of thoughts and experiences involves no application of empirical criteria of personal identity. 8
Strawson's position, put crudely, is that Kant saw the negative point: criterionless self-ascription gives no comfort to the rational psychologist, who takes the view that I must refer to an in1material substance. What he failed to see was the positive point: that criterionless self-ascription is quite compatible with the view that I refers to a corporeal object. For the links with empirical identity criteria are not severed: the referent is identifiable for others, and recognizes his identifiability, precisely by the application of empirical criteria. 9 Call this Strawson's non-severance solution to the problem set by criterionless self-ascription. Strawson himself appears hesitant in The Bounds of Sense to say quite why Kant did not take up the non-severance solution. But John McDowell, explicitly pursuing Strawson's interpretation, may be providing an answer on his behalf in Lecture V of Mind and World when he claims that Kant was committed to a Cartesian position rendering this option quite unavailable.
4 Critique of Pure Reason, trans. Norman Kemp Smith, 2nd edn. (London: Macmillan, 1933), A 363. 5 'Reply to McDowell', 147. 6 Bounds of Sense, 169; 166; 16 5. 7 Ibid. 16 5. 8 Ibid. 166. 9 Ibid. 16 5.
Kant and Strawson on the First Person
157
While spotting certain Cartesian illusions in the Paralogisms, Kant fell for the Cartesian illusion: the assumption that in providing for the content of [the] thought of a persisting self, we must confine ourselves within the flow of [self-]consciousness itself.l°
Call this 'the narrow assumption'. McDowell conceives of it as a rejection of the view he has received from Strawson and adopted himself: that selfconsciousness or 'the subjective take' is to be 'understood as situated in a wider context'; that 'the continuing referent of the I in the "1 think" ... is also a third person' .11 Of course, if Kant did adopt this assumption, it might explain why he did (or could) not avail himself of Strawson's non-severance solution: that option oversteps the bounds set by the flow of selfconsciousness. The fact that others require satisfaction of empirical personalidentity criteria to grasp my uses of I can be of no concern to me if all I am entitled to be concerned with is what lies inside my 'subjective take'. Some confirmation of the view that the narrow assun1ption is what Strawson himself has in mind is provided by his description of 'the Cartesian illusion' as 'the illusion of a purely inner and yet subject-referring use for I' .12 Strawson's point seems to be that Kant wrested himself from the wrong part of the illusion: acknowledging the possibility of a purely inner use for I, he denied that use subject-referential force. Thus he avoided the fully Cartesian position (making the referent of I an immaterial substance) at the cost of denying that this use of I does more than express 'the bare form of reference' .13 If he had appreciated that there is no 'purely inner' use of I, he would have been free to adopt the non-severance solution, leaving him able to acknowledge that I has full subject-referring force without fear of making its referent a Cartesian ego. And what prevented him fron1 appreciating that there is no purely inner use of I is precisely his adoption of the Cartesian narrow assumption. Call this the 'residual-Cartesianism' explanation of Kant's deviation: Kant adopted the narrow assumption, while acknowledging that there is no appeal to criteria of personal identity in the use of I in the '1 think'. The combination makes the application of the thought of a persisting self £ron1 within the flow of self-consciousness effortless; thus, as McDowell puts it, it can easily seenl that we had better draw Kant's conclusion: the idea of persistence that figures in the flow of [self-]consciousness had better be only fornla!. If we allowed that it is an idea of substantial persistence, the continuing to exist of an objective item, we would be committed to understanding self-consciousness as awareness of a Cartesian ego. 14
10 12 14
Mind and World, lOr. 11 Ibid. 101-2. Bounds of Sense, 165-6. My emphasis. 13 Ibid. 166. Mind and World, lOr.
158 Gaynesford More precisely, this explanation of why Kant diverges has hin1 reasoning as follows: (I) The I of the '1 think' refers. 1S (2) There is no Cartesian ego (rational psychologism's in1material substance) for the I of the '1 think' to refer to. (3) The narrow assumption holds for the '1 think'. (4) Criterionless self-ascription holds for the '1 think'. (5) So the referent of the I of the 'I think' must be merely formal (if it were a substantial object, then it could only be an immaterial one, pace (2); for only so could the idea of its persistence be provided for in accord with (3) and (4)-i.e. effortlessly and entirely from within the flow of self-consciousness).
II. RESERVATIONS ABOUT THE 'RESIDUAL-CARTESIANISM' DIAGNOSIS
But there is some reason to be suspicious of the 'residual-Cartesianisn1' explanation. It is not clear that (5) represents Kant's conclusion, nor that any of the premisses (1)-(4) occurs in support of the conclusions he did draw in the Paralogisn1s. There are several problems here, the first few of which I shall deal with only briefly in order to concentrate on those that go deeper. (i) (1) and (5): Kant is careful not to commit himself to the claim that the I of the 'I think' refers. Kant goes no further, in either edition, than saying the I of the 'I think' 'represents' [vorstellen] (A 381) or 'designates' [bezeichnen] (B 407) and then only 'a thing of indeterminate signification' (B 1°3), 'the mere forn1 of consciousness' (A 382). This I is neither an intuition nor a concept (ibid.), does not 'distinguish a particular object' (A 346), and is 'one and the same in all consciousness' (B 132). This evidently is not to go so far as saying that the I of the 'I think' is either blank (having no function at all) or mere vocalization (a function outside structured language, like a grunt); nor to assert that it is an expletive (like the semantically redundant 'it' in 'it is raining') or eliminable (contributing nothing that could not be expressed fully by other terms). But it is to claim that this I could not fulfil the task that Strawson regards most plausibly as definitional of referring-of forestalling the question 'who or which one are you talking about?' .16 Strawson, for
15 16
This is clear from Mind and World, 'On Referring' (1950), 71.
102-3.
Kant and Strawson on the First Person
159
one, evidently recognizes this: he describes the I of the 'I think' as having a 'purely nondenotative significance' for Kant. 17 Since Strawson clearly distinguishes reference from ascription, this is, of course, consistent with his claim that the I of the 'I think' nevertheless has an ascriptive function for Kant. (ii) (2): Kant does not reject the rational psychologist's conception of the ego as one of his premisses in the Paralogisms. This is clear if we consider what a (transcendental) paralogism is: a syllogism whose falsity is due to its forn1 'whatever its content may otherwise be~, and whose ground is the nature of human reason (A 34 1 / B 399). If Kant had reasoned as McDowell suggests, employing (2) as a premiss, then he would merely have shown that rational psychologism employs syllogisms whose falsity is due to content and whose ground is the inclination to regard the self as the rational psychologist does: an immaterial substance. In fact, of course, Kant finds rational psychologism guilty of employing an invalid syllogistic form whose adoption is encouraged by the nature of human reason-namely, combining analytic judgen1ents in the major premisses with synthetic judgements in the minor. (iii) (3): The narrow assumption is Kant's target in the Paralogisms, not his tenet. Kant's expressed aim in the Paralogisms is to reveal the inadequacy of rational psychology, whose defining strategy is precisely adoption of the narrow assun1ption: treating the 'I think' as 'the sole text ... from which to develop its entire wisdon1' (A 243/B 401). As Wilfred Sellars showed in his paper 'this 1 or he or it (the thing) which thinks', each Paralogism arises from the attempt to derive metaphysical claims merely by reflection on what is apparent from within the act of thinking-i.e. the attempt to impose the narrow assumption as a constraint on substantive metaphysical investigation. 18 All that can be provided from within this constraint is what Henry Allison has aptly called 'an entirely vacuous characterization of the nature of the mind: "a thing which thinks"'.19 (iv) (4): The thesis may be correct, but Strawson fails to show that Kant held it. First, Strawson is almost alone among Kant's commentators in even mentioning the thesis of criterionless self-ascription in relation to Kant. 2o 'Reply to McDowell', 147. Proceedings of the American Philosophical Association, 44 (197 0 ), 5-3 I, esp. 9. 19 Kanf's Transcendental Idealism (New Haven, Conn.: Yale University Press, 1983), 286. 20 The exception being Andrew Brook, Kant and the Mind (Cambridge: Cambridge University Press, 1994). 17
18
160
Gaynesford
He does not cite texts where the thesis is adopted by Kant, either explicitly or implicitly. Second, Strawson admits himself that Kant's exposition is 'obscure' here, that it takes a 'short cut', that 'bits of the diagnosis have to be supplied'. In particular, Strawson grants that Kant makes only 'minimal reference' to what would be crucial if he did indeed subscribe to criterionless self-ascription: namely, (empirical) criteria of personal identity.21 Third, Strawson's argument appears to be that we need to supply Kant with a developed notion of criterionless self-ascription if we are to understand his diagnosis of the ills of rational psychology. For he clainls that Kant found criterionless self-ascription to be 'at the root of the Cartesian illusion'.22 But this is difficult to square with the text. Admittedly, Kant complicates the issue somewhat by offering subtly different diagnoses. But, crucially, none seem's to depend on criterionless self-ascription-either explicitly or implicitly. For in the A text Kant thinks the rational psychologist's basic error is equivocation over terms (e.g. confusing the logical sense of 'subject' with its non-logical 'real' sense of an underlying substratum (A 349-5 I)). Later, he suggests that the basic error lies in combining the transcendental and empirical employment of categories in different premisses (A 402-3). In the B text Kant fixes first on another equivocation: 'thought' this time, rather than 'subject' (B 411-12 n.). Then he claims that the rational psychologist's mistake is to conflate analytic with synthetic propositions in the premisses of the relevant syllogisms (e.g. taking what is true of the logical identity of subjects of thought to be true of the numerical identity of persons (B 417-20)). It is this final diagnosis that commentators usually centre on. 23 But, in each case, Kant's diagnosis may be appreciated without supplying criterionless self-ascription. Indeed, so it seems to me, Kant's case against the rational psychologist would remain unchanged whether criterionless self-ascription were adopted or rejected by both parties. (v) (4): The thesis may be correct, but it would be oddly inconsistent for Kant to hold it. If Kant held the thesis (discovered it even) then he must have regarded it as applicable to the 'I think' of Transcendental Apperception; indeed, it is entirely on those passages of Kant describing this use of I that Strawson develops his case. Now to think of the I of the 'I think' as criterionlessly self-ascriptive is at least to think of it as ascriptive. And, according to Strawson, it is definitional of the ascriptive use of terms that they forestall the question 'What are you saying about it (him, her)?'.24 But Kant is clear Bounds of Sense, 166-7. 22 Ibid. 16 5. e.g. Allison, Kant's Transcendental Idealism, 282; Sebastian Gardner, Kant and the Critique of Pure Reason (London: Routledge, 1999), 227. 24 'On Referring', 71. 21
23
Kant and Strawson on the First Person
161
that the I of the 'I think' does not forestall this question, that it does not in any way determine the manner in which I exist (B 420); indeed, that in 'attaching I to our thoughts we designate the subject ... without noting in it any quality whatsoever-in fact, without knowing anything of it either by direct awareness or by reasoning' (A 355). Thus Kant's point in notoriously describing this I as 'a completely empty expression' (A 436) seen1S to be precisely to deny that it has what Strawson would regard as ascriptive (or descriptive, or classificatory) significance. In attaching this I, one may be aware of oneself, and of oneself as oneself, but certainly without being aware of any property of oneself that would answer the question 'What is being said about me?'. Now if this I has no ascriptive significance, then the thesis of criterionless self-ascription does not apply to it. And if Kant did not think that the thesis applied to the I of the 'I think', he surely did not think it applied at all. (vi) The thesis may be correct, but Strawson's argument for it seems to be indebted not to Kant but to Wittgenstein. Strawson's argument for criterionless self-ascription, which immediately follows his formulation of the thesis, is as follows: It would make no sense to think or say: This inner experience is occurring, but is it occurring to me? (This feeling is anger; but is it I who am feeling it?) Again, it would make no sense to think or say: I distinctly remember that inner experience occurring, but did it occur to me? (I remember that terrible feeling of loss; but was it I who felt it? )25
The thought seems to be this: There is a certain phenomenon that we might call the 'nonsense-question phenomenon'-(it would make no sense to ask certain questions); and we require the thesis of criterionless self-ascription to explain it. I cannot find a trace of such a thought in Kant; but precisely the same manreuvre is executed by Wittgenstein for a subtly variant but complementary purpose in his Blue Book: There is no question of recognising a person when I say I have toothache. To ask 'are you sure that it's you who have pains?' would be nonsensica1. 26
The residual-Cartesianism interpretation of Kant ascribes the narrow assumption to him in an attempt to explain why he failed to take up Strawson's non-severance solution to the problems set by the criterionless self-ascription thesis. But, for all Strawson shows, Kant did not subscribe to that thesis. So there is apparently no need to accuse him of residual Cartesianism in the way suggested. If he did not avail himself of the solution, that may simply be because he was not saddled with the problem.
25
26
Bounds of Sense, I65' The Blue and Brown Books, 2nd edn. (Oxford: Blackwell, I969), 67.
Gaynesford
162
III. RESERVATIONS ABOUT THE 'CRITERIONLESS SELF-ASCRIPTION' THESIS
Of course, it may be objected that this is hardly a relief for Kant. If he did not espouse the criterionless self-ascription thesis nor avail himself of the non-severance solution, so much the worse for him: for the truth of the matter lies there. So we must now ask whether, for all Strawson shows, the truth does indeed lie with the criterionless self-ascription thesis. (My intention here is to express certain reservations about that thesis as it is formulated and defended by Strawson; it is not my purpose to offer reasons that would necessarily count against any formulation or defence of it.) (i) The criterionless self-ascription thesis fails to take account of Strawson's own communication point. Before discussing whether criterionless self-ascription is supported by Strawson, I wish to express an objection to its formulation by him. To recall: When a nlan (a subject of experience) ascribes a current or directly remembered state of consciousness to himself, no use whatever of any criteria of personal identity is required to justify his use of the pronoun J to refer to the subject of that experience. 27
If this is what is meant by the thesis it is problematic, because there seems to be a counter example along lines given by Strawson himself. In Individuals Strawson encouraged us to remember that I is a device for giving information to others (Strawson stressed the priority of this role, but it is sufficient for my point that it is just one amongst others): In one sense, indeed, there is no question of my having to tell who it is who is in pain, when I am. In another sense, however, I may have to tell who it is, i.e. to let others know who it is. 28
Criteria are ways of telling. If, in using I, there is a sense in which I must tell who it is who is F, when I am, then there is a use of criteria. These criteria are of the relevant kind-i.e. criteria of personal identity. For the point is that, in using I, I have to let others know, provide them with ways of telling, which person is thus referred to. And criteria of personal identity are precisely ways of telling whether someone-in this case, the referent of a use. of I-is identical with some person a. Thus, pace Strawson's formulation of criterionless self-ascription, when a subject ascribes a current or
27
Bounds of Sense, 165.
28
Individuals,
100.
Kant and Strawson on the First Person
163
directly remembered state of consciousness to himself-e.g. 'I would like to go swinlming'; 'I remember there is a good beach in that direction'-some use of criteria of personal identity is required, the satisfaction of which justifies that use of the pronoun I to refer to the subject of that experience. Ordinary features of use, combined with the meaning rule for I, are usually sufficient to satisfy these criteria and thus justify the use-these features provide others with ways of telling that the token of I in question proceeded from person a's mouth; the rule tells us that any token of I refers to its user; hence others can tell that its referent is identical with person a. The whole point is, I believe, nicely illustrated by the slightly unusual case where the mere act of saying I is insufficient for a person to have himself picked out by others as the subject of some state of consciousness. In a crowd of others, for example, I-sayers will automatically augment ordinary features of use, perhaps by sticking their hands in the air. As the I-sayer, the gesture is precisely my acknoV\Tledgement that, recalling the Strawson of Individuals, there is a sense in which I must tell who it is who is F, when I am. Recalling Strawson's 'On Referring', the gesture is nlY acknowledgement that I is a referring expression; as such, my use of it must forestall the question 'Who is being talked about?' (ii) The criterionless self-ascription thesis is unsupported by Strawson. Taking such criticism on board, we may now ask whether, in The Bounds of Sense, Strawson supports the criterionless self-ascription thesis in its slightly weaker formulation-i.e. that, at least as far as the I-user himself is concerned, the immediate self-ascription of thoughts and experiences involves no application of criteria of personal identity.29 I have already quoted the argument in its entirety. It comes down to the following: there is a certain phenomenon, the 'nonsense-question phenomenon', and we require the thesis of criterionless self-ascription to explain it. The conclusion scarcely seems to follow from the evidence, however. What is the agreed datum? Not that the question itself ('This inner experience is occurring, but is it occurring to me?') literally makes no sense, but rather that it would make no sense for the subject to ask it. This is, of course, true, at least in certain circumstances. But all that seems to follow is that the subject is entitled to assume a certain answer. There is undoubtedly some feature of the examples Strawson raises that licenses the assumption. But what is that feature? We are precisely trying to determine what that feature is, and no reason has yet been given for believing it is the one Strawson proffers: that criteria of personal identity have no application here. So the argument in favour of the criterionless self-ascription thesis
29
Cf. Bounds of Sense, r66.
Gaynesford certainly needs supplementing. This might be done by arguing that there is no conceivable licence for the assumption that personal identity criteria have no application here. But this seems evidently incorrect. The assumption might be licensed pragmatically-by a resolution on our part, conventionally entered into, to ask whether one has got the reference right only when there is, and because there is, an evident possibility or likelihood of one's having got the reference wrong. This explanation is, of course, consistent with the claim that personal-identity criteria nevertheless have application here. For in thinking 'a is F' (where 'a' is a singular term) one may be licensed to assume that one has got the reference right even though one recognizes that criteria are applicable and may be invoked to justify one's reference. Nothing so far has been said to rule out this possibility. So Strawson's argument for the criterionless self-ascription thesis seems to lapse; it presents nothing to rule out an alternative account of the nonsense-question phenomenon and its licensed assumption. 164
(iii) The criterionless self-ascription thesis may not be the appropriate explanation of the nonsense-question phenomenon. One n1ight go further and question Strawson's own rationale for the criterionless self-ascription thesis-to consider, in other words, whether it is not the wrong explanation of the nonsense-question phenomenon. For there are reasons to think that criterionless self-ascription is altogether too weighty a hammer to crack the nonsense-question nut-indeed, perhaps even the wrong type of tool altogether. The possibility of a pragmaticconventional account illustrates that the thesis may be too weighty. I shall give another reason for thinking so in a moment. What suggests that the thesis may be the wrong tool is that it has so restricted an application-to presenttense self-ascriptions of inner experience-when the nonsense-question phenomenon itself is apparently so very broad, occurring in cases that do not involve mental predicates, uses of I, nor even attempts at self-ascription. Thus a subject might say 'I should have recanted', 'Now is the time for repentance', 'It is hot here', or 'This tastes smoky' in circumstances where it would make no sense for him to ask 'Somebody should have recanted, but was it me?', 'Some time is the time for repentance, but is that time now?', 'It is hot somewhere, but is it hot here?', or 'Something tastes smoky, but is it this that tastes smoky?'. A further reason for doubt concerning Strawson's argument is that it appeals exclusively to ascriptions of what he terms 'inner experience~ '1 feel pain, loss, anger'. Such evidence might justifiably be regarded as of dubious relevance. For these ascriptions belong to a more peculiar class than those that manifest the nonsense-question phenomenon-one in which the subject is guarded not simply from mistaken reference but from mistaken predication. The subject is not simply right about who is feeling
Kant and Strawson on the First Person 165 pain but that it is pain that is felt. Compare this with the ordinary cases given above: the nonsense-question phenomenon holds true even though I may be mistaken that anyone should recant, that any time is the time for repentance, that anywhere is hot, or that anything tastes smoky. Intoxication with inner experience therefore seems wrongheaded in a way that Strawson himself has elsewhere warned us about. (iv) The criterionless self-ascription thesis may not be consistent with Strawson's claim that self-ascription requires knowing what kind of thing one is. One might go further still and consider whether Strawson himself can afford to embrace the criterionless self-ascription thesis. Strawson himself considered one such consistency problem in The Bounds of Sense: if 1 can be used without criteria of personal identity, how can it be regarded as referring at al1?3o The non-severance solution was his answer. Recently, he has himself cut the grounds from underneath this answer somewhat by claiming that criterionless self-ascription is 'strictly beside the point' and 'has nothing at all to do with the case' of whether the subject refers using 1. 31 Whether or not this is the correct response, I am concerned here not with the question of whether 1 refers but with what it refers to; and so will set the matter aside. I shall raise two further consistency problems, though somewhat tentatively since considerations of space prevent me developing them sufficiently to bear much weight. It is plausible to suppose that a subject might consistently misconceive what kind of thing he is, but not so uncomprehendingly that he would be judged unable to entertain thoughts of the form 'I am F'. Now suppose that the criterionless self-ascription thesis is correct for the reason Strawson gives. Since asking 'Someone is F, but is it me?' would be nonsensical to the subject, it must be the case that he is able to self-ascribe such thoughts without application of personal-identity criteria. But claiming that he is able to self-ascribe such thoughts at all seems to run counter to Strawson's own position that being able to self-ascribe thoughts and experiences requires being able to pick out from all other things that to which the thought or experience is ascribed; and that in turn requires knowing what kind of thing one is. For this subject precisely does not know what kind of thing he is. What has gone wrong? Strawson cannot reject the sortal claim without losing the crucial premiss of his central argument at this point in The Bounds of Sense-that self-consciousness requires conceiving of the subject of one's thoughts and experiences as a corporeal object amongst others. It might be objected that the type of case I have offered all too briefly is not in fact
30
Cf. ibid.,
165.
31
'Reply to McDowell',
149.
166
Gaynesford
conceivable. Misconception problems certainly require more stage-setting than I can offer here. 32 So I shall move on, after simply registering the thought that surely the weakest link here is precisely the claim that the nonsensequestion phenomenon is sufficient to ground the thesis of criterionless selfascription. (v) The criterionless self-ascription thesis may sever the links between uses of I and empirical criteria of personal identity in a way that threatens the corporeality thesis. Consider another consistency problem. The corporeality thesis would be false if, so far as our understanding of I is concerned, the term might refer to something less than a corporeal object-for example, a subject of experience, capable of making objectively valid judgements and of conceiving what it intuits as representations of an objective world, which, though spatio-temporally located, is neither impenetrable nor force-exerting (something with the physical presence of a hologram, perhaps). For such an object would not be corporeal, at least in Kant's view, since both properties of inlpenetrability and force-exertingness are allegedly 'thought in the concept' (A 8/B 12). Of course, if there were sonle requirement that personalidentity criteria be applied in using I, then that would be a safeguard. For 5trawson thinks that persons are in fact corporeal objects. 50 the call for empirical criteria of personal identity is a call for criteria of the identity of corporeal objects. But suppose there were no such requirement on certain uses of I, as the criterionless self-ascription thesis asserts. What is to prevent there being, pace the corporeality thesis, bona fide referents of I that are something less than corporeal objects? We can ignore the problem of less than full-blooded uses of I here-the seemingly first-personal output of computers, answerphones, speak-yourweight machines, etc. may be classed alongside quotation, fiction, and other conniving uses of I as limiting cases of no pressing interest. The question at issue here is whether the corporeality thesis is proof against the possibility of less than full-blooded beings-sub-personal, non-corporeal items-being considered bona fide referents of I. It might be thought that the non-severance solution could be pressed into service here. The idea would be that, although personal-identity criteria need not be invoked for these uses of I, links between those criteria and the uses remain and are sufficient to safeguard the corporeality thesis. But what are these links? The first is that the token of I issues publicly from the mouth of a man who is recognisable and identifiable as the person he is by the application of empirical criteria of personal identity. 33 32 33
Cf. Quassim Cassam, Self and World (Oxford: Oxford University Press, r997), ch. 4. Bounds of Sense, r65.
Kant and Strawson on the First Person
I67
But a token of I may issue publicly from the mouth of a man without referring to him-if, for example, the man is merely used as a mouthpiece, his vocal organs manipulated by another. Given the meaning rule for I, this other, being the user of I, is its referent. But, for all Strawson shows, this other might not be a hun1an being or corporeal object at all. Thus the corporeality thesis is still under threat. Strawson's second link is that the token of I is used by a person who would acknowledge the applicability of those criteria in settling questions as to whether he, the very man who now ascribes to himself this experience, was or was not the person who, say, performed such-and-such an action in the past. 34
But, by fixing on an I-user who is a person, this evidently begs the question against the spectre of a sub-personal I-user. Of course, if the I-user is a person, then the token of I refers to a corporeal object and the links between I and empirical criteria of personal identity are unsevered; for any token of I refers to its user, and a person is a corporeal object. But perhaps the I-user is not a person. To conclude. One explanation of why Kant and Strawson diverge on the corporeality thesis is residual Cartesianism. But there is some reason to doubt both whether Kant held the various premisses he is thereby claimed to have held, and whether he connected them in the way alleged. In particular, reservations have been expressed as to whether the criterionless selfascription thesis is either correctly attributable to Kant or sufficiently supported by Strawson. 35
Ibid. For very helpful written comments, I am especially grateful to Eugene Mills. I am also grateful to participants at the Kant-Strawson conference at Reading, and to subsequent audiences at Oxford, Pittsburgh, Richmond (VCU), and Williamsburg (CWM). 34 35
10
Reflective Judgement and the Application of Logic to Nature: Kanfs Deduction of the Principle of Purposiveness as an Answer to Hume HENRY E. ALLISON
• ••
In his analysis of judgement in the two versions of the introduction to the Critique ofJudgment Kant introduces the purposiveness of nature as a new transcendental principle pertaining to judgement in its reflective capacity; and in the second or published Introduction he provides it with a transcendental deduction. Although this principle has been the topic of considerable discussion in recent years, neither its role as a general condition of empirical knowledge nor the nature and force of its deduction have been adequately explored. Such an exploration is the concern of this chapter. I shall argue that this principle lies at the centre rather than the periphery of Kantian epistemology; and that, even though Kant does not present it as such, its deduction may be viewed as his definitive answer to Hume regarding the rational grounding of induction broadly construed.! The discussion is divided into three parts. The first examines the principle of purposiveness, its source in reflective judgement, and its multifaceted epistemic function. The second analyses the deduction of this principle in section V of the published Introduction. The third evaluates the adequacy of this deduction as an answer to Hume. 1 Actually, on my reading, the response to Hume is only one half of the story, since the same line of argument may also be taken as Kant's attempt to mediate the dispute between Locke and Leibniz regarding natural kinds. In short, here, as elsewhere, Kant should be seen as attempting to kill two birds with one stone. For my discussion of the latter side of Kant's argument see 'The Critique of Judgment as a "True Apology" for Leibniz', in V. Gerhardt, R.-P. Horstmann, and R. Schumacher (eds.), Kant und die Berliner Aufkliirung, Akten des ix. Internationalen KantKongress, i (BerlinlNew York: Walter de Gruyter, 2001), 286-99. I also provide a more systematic account of the whole issue in ch. I of Kant's Theory of Taste: A Reading of the Critique of Aesthetic Judgment (Cambridge: Cambridge University Press, 2001).
Allison I
In both versions of the introduction Kant describes judgement's principle as that of the purposiveness of nature. In the first version this purposiveness is characterized more specifically as 'logical' (FI xx. 216-17,4°4-5) and in the second as 'formal' (KU v. 18o-I, 20);2 but in both cases it signifies the contingent agreement of the order of nature with our cognitive needs and capacities. Moreover, in both versions Kant explicitly links this principle with familiar formulas or maxims such as 'nature takes the shortest way', 'nature makes no leap in the diversity of its forms', and 'principles must not be multiplied beyond necessity' (KU v. 182, 21-2; see also FI xx. 210, 399).3 As these formulas suggest, the basic idea is that we look upon nature as if it had been designed with our cognitive interests in mind; though, of course, we have no basis for asserting that it was in fact so designed. In the formulation of the Second Introduction Kant describes the principle thus: [S]ince universal natural laws have their ground in our understanding ... the particular empirical laws must be considered, with respect to that which the universal laws left undetermined in them, according to a unity such as they would have if they had been given by an understanding (though not ours) for the benefit of our cognitive faculties, by making possible a system of experience in terms of particular natural laws. (KU v. 180, 19)
This formulation in terms of a system of empirical laws (or, as it is often referred to in the literature, of 'systematicity'4) is prevalent in both introductions. It is not, however, the only way in which this principle and its function are characterized. For example, in the First Introduction it is presented as the
2 Apart from the Critique of Pure Reason, all references to Kant are by volun1e and page number of Kant's Gesammelte Schriften, herausgegeben von der Deutschen (formerly Koniglichen Preussischen) Akademie der Wissenschaften, 29 vols. (Berlin: Walter de Gruyter, I902.- ). References to the Critique of Pure Reason are to the standard A and B pagination of the first and second editions and are given in the text. In addition, I shall use the following abbreviations for the titles of Kant's works: Erste Einleitung in die Kritik der Urteilskraft (FI); Grundlegung zur Metaphysik der Sitten (Gr);jasche Logik (jL); Kritik der Urteilskraft (KU); Prolegomena zu einer jeden kunftigen Metaphysik (Pro). Citations from the Critique of judgment and the First Introduction are based on the translation by Werner Pluhar, Critique of judgment (Indianapolis, Ind.: Hackett, I987). However, I have frequently modified Pluhar's translation of particular passages. 3 As Reinhard Brandt points out, these are equivalent to the 'principles of convenience' to which Kant refers in §30 of the Inaugural Dissertation (see 'The Deductions in the Critique of Judgment: Comments on Hampshire and Horstmann', in Kant's Transcendental Deductions: The Three 'Critiques' and the 'Opus Postumum' (ed.), by Eckart Forster (Stanford, Calif.: Stanford University Press, I989), I8I-2). These same maxims are also analysed at greater length in the appendix to the Transcendental Dialectic of the first Critique (A 652/B 680-A 660/B 688). 4 See Paul Guyer, 'Reason and Reflective Judgment: Kant on the Significance of Systematicity', Nous, 24 (I990), I7-43, 'Kant's Conception of Empirical Law', Proceedings of the Aristotelian Society, suppl. vol. 64 (I990), 220-42.
Reflective Judgement
171
principle that 'for all natural things concepts can be found that are determined empirically', which is then glossed as 'we can always presuppose nature's products to have a form that is possible in terms of universal laws which we can cognize' (FI xx. 211-12, 400). In the Second Introduction, however, Kant appears to argue that its main function is not simply to systematize empirical laws but to ground their very necessity, that is their claim to nomological status (see KU v. 183, 22). In fact, in various places in the introductions Kant suggests that the principle of the purposiveness of nature is necessary for the formation of empirical concepts, the classification of 'natural forms' into genera and species, the unification of empirical laws into a systen1 (theory construction), the formulation of empirical laws in the first place, and the attribution of necessity to such laws. 5 Nevertheless, it is possible to find some coherence in this variety of formulations, if we keep in mind the essential function of reflective judgement, namely to find universals for given particulars. First of all, this search for universals can take the form either of finding empirical concepts under which particulars can be subsumed for the sake of classification or of finding empirical laws in terms of which their behaviour can be explained. Moreover, as Hannah Ginsborg has pointed out, these two types of universal are themselves closely connected, as are a taxanomic classification of 'natural forms' in terms of genera of species and a systematic organization of empirical laws. For one thing, without assuming something like natural kinds we could not even begin to look for empirical laws or hope to distinguish such laws fron1 contingent regularities. For another, determinate empirical concepts presuppose known causal laws, since the inner properties in terms of which we conceptualize and classify things must include causal properties. Finally, the necessity and therefore the nomological character of relatively specific laws, such as that of the solubility of gold in aqua regia, are a function of their derivability from higher-level laws, such as those that hold at the molecular and atomic levels. 6 Perhaps of greater immediate relevance, the same connections can also be spelled out in terms of Kant's conception of judgement. To begin with, concepts for Kant serve as predicates of possible judgements (A 69/B 94), which means that the whole purpose of bringing intuitions under concepts is to make possible determinate judgements about their corresponding objects. The judgements in which Kant is interested are, however, those termed in
5 The diversity of formulations and functions is emphasized by Guyer, who claims that Kant affirms both a 'taxanomic' and an 'explanatory' version of logical purposiveness (or systematicity) and that they are unrelated (see Guyer, Kant and the Claims of Taste (Can1bridge, Mass.: Harvard University Press, 1979) 44-5. 6 Hannah Ginsborg, The Role of Taste in Kant's Theory of Cognition (New York/London: Garland, 1990), 190.
172
Allison
the Prolegomena 'judgen1ents of experience'; that is, objectively valid, grounded claims about objects of possible experience. Moreover, it seems clear that in order to qualify as such, a judgement must either be itself a statement of empirical law or be derivable from such a law. Accordingly, the search for empirical concepts that can serve as predicates in judgements of experience is inseparable from the search for empirical laws; and since, as suggested above, the latter is inseparable from a hierarchical organization of such laws, it follows that the quest for the conditions of the possibility of empirical concepts and for the systematic organization of empirical laws are best seen as two poles of a quest for the conditions of the empirical knowledge of nature qua empirical, or, equivalently, for judgements of experience. 7 In the more expansive and helpful account in the First Introduction Kant insists that even though the principle of purposiveness is transcendental, it is 'merely a principle for the logical use of judgement', which, as such, allows us to 'regard nature a priori as having in its diversity the quality of a logical system under empirical laws' (FI xx. 214, 402). The 'logical use of judgement' is to be distinguished from its transcendental use, which according to the first Critique is to provide the schemata that are the sensible conditions for the application of the categories. It consists in the formation of empirical concepts and their organization into genera and species, which makes possible the subordination of these concepts in judgements and the connection of the judgements in syllogisms. 8 In so far as our concepts are orderable in a single set of genera and species they have the form of a logical system, and in so far as this order reflects the actual order of nature it too may be thought of as a 'logical system under empirical laws'. Such a view of nature has, of course, merely the status of a regulative idea; but, as Kant points out, in light of it we can proceed to investigate nature either from the bottom up or from the top down. The former procedure begins with the classification of diverse particulars as n1embers of a single species, then distinct species are unified on the basis of common properties into a genus, and different genera into a higher genus, etc. Ideally, the process culminates in the unification of all these higher-order genera into a single highest genus. Conversely, the movement from the top down is one of increasing specification, wherein differentiations are continually introduced between items that were initially taken to be members of a single species. 9 7 Although I do not accept all the details of her analysis, I am here following the suggestion of Beatrice Longuenesse that the concern with judgements of experience and their distinction from judgements of perception is not simply an aberration of the Prolegomena but an important feature of Kant's thought. For Longuenesse's view see Kant and the Capacity to Judge (Princeton, N]: Princeton University Press, 1998), ch. 7. 8 For the connection between judgement in its logical use and syllogistic reasoning see Longuenesse, Kant and the Capacity to Judge, 90-5. 9 Although Kant does not make the point here, in his discussion of the same issue in the first Critique he points out that this process of specification is ideally infinite, since it is always appropriate to search for further subspecies (see A 65 5-7/B 683-6).
Reflective Judgement
173
Kant's view of the significance of such an ideal scheme for empirical knowledge is best expressed in two footnotes in the First Introduction. The first of these is attached to the previously cited formulation of the principle as 'for all natural things concepts can be found that are determined empirically' (FI xx. 211, 400). In the note Kant attempts to argue that, even though this principle may seem to be merely logical and tautologous, it is actually synthetic and transcendental because it expresses nothing less than 'the condition under which it is possible to apply logic to nature' (FI xx. 211-12, 400; my emphasis). By 'logic' Kant does not, of course, mean forrnallogic, but rather our discursive, conceptual abilities. The initial claim is that some such systematic structure (approaching the ideal of a logical system) is necessary for the successful exercise of this capacity because this exercise is based on comparison (the logical act), and comparison requires something to compare. And in light of this Kant concludes that, as a condition of the possibility of its own logical activity, reflective judgement must assume that nature, with its boundless diversity, has hit upon a division of this diversity into genera and species that enables our judgement to find accordance among the natural forn1s it compares, and to arrive at empirical concepts, as well as at coherence among these by ascending to concepts that are more general though still empirical, i.e. judgement presupposes a system of nature even in terms of empirical laws, and it does so a priori and hence by means of a transcendental principle. (FI xx. 212 n., 400)
Although at the end of the note Kant repeats the unexplained transition from empirical concepts to a system of empirical laws made in the text to which it is attached, the main focus is on the conditions of forming a set of empirical concepts that cohere with one anothero Some degree of coherence is clearly necessary if the concepts obtained through comparison are to be connectable with one another in judgement-that is, if they are to function as concepts at all; and this is what is provided by their systematic ordering in terms of the relation of genera and species. It would be a mistake, however, to regard such an ordering merely as a kind of supplemental requirement or desideratum rather than as a necessary condition of the possibility of the concepts themselves. On the contrary, the necessity for a hierarchical ordering in terms of genera and species follows from the very nature of a concept. Consider, for example, the concept 'gold', understood as a yellow metal, soluble in aqua regia, etc. It is composed (in part) of these distinct concepts, which constitute its intension, and it stands to each of them in the relation of species to genus (or logical form to matter). Thus, 'gold' designates a species of yellow objects, of metal, of things soluble in aqua regia, etc., while at the san1e time the concept also functions as a genus under which different types of gold (or of things composed of gold) are to be distinguished as species. And this, of course, is not
174 Allison something unique to the concept of gold, but is a feature of every empirical concept. In short, every such concept (except for that of the highest genus 10 ) is itself both a species of the concepts contained in it and a genus for the concepts falling under it, so that the very possibility of concepts as general representations presupposes a system of concepts subordinate to one another in terms of the relation of genera and species. It follows from this that such a system of concepts is a necessary condition for the application of logic to nature; that is, for empirical judgement. It does not, however, follow that such a systen1 is, as such, also sufficient to account for the kind of judgement in which Kant was really interested; namely, judgements of experience, which, as I have argued, either themselves state or are derivable from empirical laws. Accordingly, it is worth exploring a second note in the First Introduction, in which Kant seems to go further. In this frequently discussed note he writes: One may wonder whether Linnaeus could have hoped to design a systen1 of nature if he had to worry that a stone which he found, and which he called granite, n1ight differ in its inner character from any other stone even if it had looked the same, so that all he could ever hope to find would be single things-isolated, as it were, for the understanding-but never a class of them that could be brought under concepts of genus and species. (PI xx. 215-16 n., 403 )11
This goes beyond the preceding note by making explicit the requirement that a classificatory system reflect an underlying order of nature. Thus, whereas any number of such systems might be possible, the assumption is that there is one (and only one) that, as it were, 'carves nature at its joints'. And the goal of a systematizer such as Linnaeus is to provide the system that reflects this order (or at least comes as close as possible to doing so). Moreover, since the classification of phenomena has to be based on observed uniformities and differences, the operative assumption must be that outer similarities and differences correspond to inner or intrinsic ones. To use Kant's own example, objects with the observable features of granite must also be similar in their inner character; for otherwise there would be
10 In the Jasche Logic Kant defines the highest genus as that which is not a species and the lowest or infima species as that which is not a genus. But, whereas it is necessary to assume a highest genus, the possibility of a lowest species is denied on the grounds of the generality of every concept (JL ix. 97). In the Critique of Pure Reason (A 290/B 346) Kant identifies the highest genus as the 'concept of an object in general', which is subdivided into the concepts of something and nothing. There, however, he is concerned with transcendental rather than empirical concepts. 11 This may be seen as the Kantian equivalent to Leibniz's principle that 'every outer appearance is grounded in the inner constitution', or that 'whatever we truthfully distinguish and compare [on the basis of outer appearances] is also distinguished or made alike by nature' (New Essays on Human Understanding, bk. III, ch. VI, trans. and ed. Peter Remnant and Jonathan Bennett (Cambridge: Cambridge University Press, 19 8r ), 309.
Reflective Judgement
175
no basis for inferring from the fact that an object has granite-like features that it will behave similarly to other objects with these features. In light of this it is instructive to consider Kant's cryptic account of the 'inferences' of reflective judgement in the Jasche Logic. According to this account, there are two species of such inference-that is, two ways of inferring (empirical) universals from particulars, namely induction and analogy. The former moves from the particular to the universal according to what Kant terms the 'principle of universalization' (Princip der Allgemeinmachung): 'What belongs to many things of a genus belongs to the remaining ones too.' The latter moves from a similarity between two things with respect to a particular property to .a total similarity according to the corresponding 'principle of specification: Things of one genus, which we know to agree in much, also agree in what remains, with which we are familiar in some things of this genus but which we do not perceive in others' (JL ix. 133). Moreover, these principles are themselves specifications of the higher-order principle governing the inferences of reflective judgen1ent as a whole; namely, 'that the many will not agree in one without a common ground, but rather that which belongs to the many in this way will be necessary due to a common ground' (JL ix. 132). Although Kant does not make the point, it seems clear that induction and analogy are the inference forms through which judgements of experience are grounded. 12 In Hunlean terms, they describe the inferential processes through which we move froin something observed (a present impression) to something unobserved. Or, in more contemporary language, they are the vehicles through which predicates are 'projected': either from some instances of x to all xs (induction) or of a given x on the basis of other predicates already known to pertain to that x (analogy). It is also clear that the principle on which they are based, and that therefore licenses such inferences or 'projection' of predicates, is itself an application of the principle of purposiveness as described above. For to claim that 'the many will not agree in one without a conlmon ground, etc.' is just to claim that observable regularities are not merely incidental, but reflect an underlying ground or order of nature. Finally, it follows fron1 this that the stakes involved in a deduction of the principle of purposiveness are high indeed, amounting to nothing less than what is now termed the 'vindication of induction'. More specifically, the issue in its Kantian forn1 is whether inductive procedures (construed in a broad sense to include reasoning by analogy as well as induction proper) can be given a rational justification within the framework of reflective judgement. 13 12 Kant does, however, refer to the distinction between judgements of perception and judgements of experience in the Jasche Logic, with the latter characterized as 'an empirical judgment through which I get a concept of an object' (JL ix. 114). For a discussion of the distinction in this text and its relation to that of the Prolegomena see Longuenesse, Kant and the Capacity to Judge, 188-95· 13 It is noteworthy that in the Jasche Logic Kant characterizes induction and analogy as merely 'logical presumptions', since they lack the true necessity possessed by inferences of reason (JL ix. 133).
176 Allison For, such a justification is required in order to provide an answer to Hume's sceptical doubts regarding the rational credentials of inferences from the observed to the unobserved.
II The 'official' deduction is contained in section V of the Second Introduction, which is given the heading: 'The Principle of the Formal Purposiveness of Nature Is a Transcendental Principle of Judgement' (KU v. 181, 20). As a prelude to the actual deduction, Kant tries to show both that this principle is genuinely transcendental and that it is required because the task that it performs is not already accomplished by the transcendental principles of the understanding established in the Transcendental Analytic of the first Critique. In the First Introduction, where Kant likewise insisted on the transcendental nature of the principle, the alleged problem was that, because of its connection with empirical concept formation, it seemed to be merely logical and tautological rather than synthetic and transcendental. And, as we have already seen, Kant there argues for its transcendental status by trying to show that it is the condition of the possibility of applying logic to nature. By contrast, in the Second Introduction this latter claim is dropped, together with any reference to a worry that it might be merely a tautologous, logical principle. Instead, Kant now insists that the principle cannot be merely empirical or psychological, since it makes a normative claim about how we ought to judge rather than simply describing how we do, in fact, judge (KU v. 182, 22). And in support of this he appeals to the previously mentioned maxims of judgement, which serve as an a priori basis of the investigation of nature (KU v. 182, 21). By claiming that the principle is transcendental Kant also commits himself to the task of providing it with some kind of deduction or justification. But, even granting this, one might still argue that a separate deduction would be redundant on the grounds that the basic task was already accomplished in the first Critique. In order to deflect any such possible objection and to underscore the unique nature of formal purposiveness as a principle of judgement, Kant takes great pains to argue that the need for such a deduction is not obviated by the Transcendental Deduction of the first Critique. The basic point is that the transcendental laws laid down in the Analytic of the first Critique do not themselves guarantee the existence of a cognizable order at the en1piricallevel. Since these laws concern merely the 'formal' conditions under which objects can be cognized together in a single spatiotemporal framework (the unity of experience), they are compatible with
Reflective Judgement 177 any number of different empirical orderings. Or, simply put, they underdetermine the ordering of the particulars falling under them. Thus, even though these laws ensure the existence of some order in nature, it need not be one discernible in appearances by the human mind. 14 And it certainly need not take the form of an arrangement according to natural kinds. For example, the most important of these transcendental laws (the principle of causality) states that for any event b, there must be some antecedent condition a, such that given a, b necessarily follows. This licenses the search for causes, but it hardly ensures that it will be possible to find theIn; that is, to distinguish between a n1erely accidental succession a-b and one that is .genuinely causal. 15 For, as far as this transcendental law is concerned, experience might present few (if any) discernible lawlike regularities capable of supporting induction. Moreover, if this were the case, then, a fortiori, we could neither discover empirical laws in terms of which particular phenomena can be predicted or explained nor connect these laws in overarching theories. As Kant puts it in a passage that constitutes part of the elaboration of the deduction: For it is quite conceivable that, regardless of all the uniformity of natural things in terms of the universal laws, without which the form of an empirical cognition in general would not occur at all, the specific differences in the empirical laws of nature, along with their effects, might still be so great that it would be impossible for our understanding to discover in nature an order it could grasp, i.e. impossible for it to divide nature's products into genera and species, so as to use the principIes by which we explain and understand one product in order to explain and comprehend another as well, thereby making coherent experience out of material that to us is so full of confusion (though actually it is only infinitely diverse and beyond our ability to grasp). (KU v. 185,25)16
This passage raises a spectre that is both reminiscent of and significantly different from the more famous spectre that Kant raises in connection with the Transcendental Deduction in the first Critique. In introducing the problematic of that deduction, Kant suggested that (for all that had been shown so far): 'Appearances ll1ight very well be so constituted that the understanding should not find them to be in accordance with the conditions 14 Admittedly, there is a tension in Kant's thought at this point, since, as Guyer notes, he appears to waver between the view that there might be no laws at all and that there might be laws that are not discoverable by the human mind (see Guyer, 'Reason and Reflective Judgment,' 36-7, 'Kant's Conception of Empirical Law', 233-4). I am assuming, however, that the latter reflects Kant's considered opinion (or at least what he ought to have maintained), since the Second Analogy of itself entails that there must be causal laws of some sort (albeit not necessarily ones that can be recognized as such). 15 I discuss this issue in more detail in connection with what I term the 'weak' interpretation of the Second Analogy in 'Causality and Causal Law in Kant: A Critique of Michael Friedman', in Idealism and Freedom (Cambridge: Cambridge University Press, 1996), 80-91. 16 For a corresponding passage in the First Introduction see FI xx. 2°9, 397-8.
178
Allison
of its unity' (A 90/B 123). This may be termed 'transcendental chaos' (disorder at the transcendental level). Clearly, one of the major concerns of that deduction is to exorcize this spectre, which Kant atten1pts to do by showing that the possibility that appearances are 'so constituted' is ruled out on the grounds of its incompatibility with the conditions of the unity of apperception. By contrast, the present spectre may be termed that of 'empirical chaos' (disorder at the empirical level), and it can be characterized as a scenario in which something like Hume's 'uniformity principle' does not hold. 17 Or, more precisely, it is one in which the uniformity that nature necessarily exhibits in virtue of its conformity to the transcendental laws imposed by the very nature of the understanding does not translate into an empirically accessible uniformity, understood as one which could support induction and analogy. Since the operative assumption is that this possibility is left open by the Transcendental Deduction of the first Critique, even though the latter succeeded in its appointed task of establishing the necessary conformity to law of experience at the transcendental level, the spectre obviously cannot be exorcized by appealing to the unity of apperception and the transcendental laws derived therefrom. Indeed, the problem arises precisely because the possibility of empirical chaos or lack of sufficient uniformity is not precluded by these laws, which ensure, for example, that nothing happens without a cause, but not that these causes are discoverable on the basis of empirical regularities. In both introductions Kant expresses this point by noting that an empirically cognizable order of nature is contingent with respect to these transcendental conditions, which entails that it cannot be deduced as a consequence thereof. l'vloreover, as Kant suggests in the Dialectic of Teleological ]udgen1ent, this contingency may be seen as a consequence of an even more fundamental one that is endemic to our discursive understanding; nan1ely, the contingency that the 'particular, as such' (als ein solches) has with respect to the universal supplied by the understanding (KU v. 404, 287). For since, as discursive, our understanding proceeds from universals (concepts and laws) to the particulars that are to be subsumed under then1, and since these particulars, as sensibly given, are not themselves products of the act of understanding,
17 In the Treatise, bk. I, pt. III, sect. VI, p. 89 (with which, of course, Kant was not familiar) Hume formulates the principles as holding 'that instances, of which we have had no experience, must resemble those, of which we have had experience, and that the course of nature continues always uniformly the same'. I have said 'something like' this principle because, as formulated, it is hopelessly vague and I am not here concerned with the details of a correct formulation. For a discussion of some of the problems involved see Barry Stroud, Hume (London/BostonlHenley: Routledge & Kegan Paul, 1977) 54 H.
Reflective Judgement
179
it follows that there is an unavoidable element of contingency in the fit between universal and particular. And if this is the case then the same contingency must also apply to judgement in its reflective activity of seeking universals under which to subsume sensible particulars. Nor, once again, can it be objected that the assertion of an ineliminable contingency of fit between universal and particular conflicts with the results of the Transcendental Deduction of the first Critique. For that deduction was not concerned with the 'particular as such', but merely with it qua spatiotemporal entity or event; and taken under that description particulars remain fully subject to the categories. But the question of empirical lawfulness does concern precisely the 'particular as such'. Consequently, such lawfulness is contingent with respect to the universal and the spectre of empirical chaos ren1ains in place. It follows from this that if this latter spectre is to be dealt with a distinct transcendental principle is required. And it likewise follows, given Kant's definition of purposiveness as the 'lawfulness of the contingent as such' (PI xx. 217, 4°5), that it must take the form of a principle of purposiveness. 18 This principle cannot be used, however, to deny the very possibility of empirical chaos in anything like the n1anner in which the transcendental unity of apperception denies the possibility of the transcendental variety; that is, by somehow proving that nature, in its empirical diversity, necessarily conforms to our cognitive needs. For any such objective deduction, even one subject to the standard 'critical' limitation to objects of possible experience or phenomena, is precluded by virtue of the above-mentioned ineliminable contingency of fit between the (empirically) universal and the particular. This does not, however, rule out the possibility of a subjective deduction, which would leave the results of the Transcendental Analytic of the first Critique in place but go beyond them. 19 The goal of such a deduction would not be to remove the spectre by showing it to be incompatible with the transcendental conditions of experience (since that is impossible), but merely to render it idle. Moreover, this is precisely what Kant's actual deduction attempts to accomplish by establishing the subjective necessity of
See also KU v. 4°4,287. The thesis that Kant's new transcendental principle does involve an abandonment of original 'critical' principles has been explicitly affirmed by Burkhard Tuschling, 'The System of Transcendental Idealism: Questions Raised and Left Open in the Kritik der Urteilskraft', System and Teleology in Kant's Critique of Judgment (Spindel Conference 1991), The Southern Journal of Philosophy: (ed.) Hoke Robinson supp!. vol. 30 (1992), 109-27, and 'Intuitiver Verstand, absolute Identitat, Idee. Thesen zu Hegels friiher Rezeption der Kritik der Urteilskraft,' in H. F. Fulda and R. P. Horstmann (eds.), Hegel und die Kritik der Urteilskraft (Stuttgart, Klett-Cotta, 1990), 174-88. I criticize Tuschling's analysis in 'Is the Critique of Judgment "PostCritical" '?, in Sally Sedgwick (ed.), The Reception of Kant's Critical Philosophy: Fichte, Schelling, Hegel, (Cambridge: Cambridge University Press, 2000), 78-92. 18
19
180
Allison
presupposing the purposiveness of nature in the process of empirical enquiry. In other words, the claim is not that nature is purposive, i.e. that we have some sort of a priori guarantee that it is ordered in a manner commensurate with our cognitive capacities and needs; nor is it even that we must believe it to be purposive in this sense (which is basically Hume's position). The claim is rather that we are rationally constrained to approach nature as if it were so ordered. Or, in Kant's own terms, at the basis of all reflection on nature (the search for empirical laws) lies the a priori principle that 'a cognizable order of nature in terms of these [empirical] laws is possible' (KU v. 185,24), Since Kant's reasoning here is akin to that underlying the well-known claim in the Groundwork that we can act only under the idea of freedom, it may prove useful to exan1ine the latter briefly.2o There the 'spectre' is that in spite of the resolution of the Third Antinomy, which showed only that transcendental freedom is compatible with causality according to laws of nature, our apparent practical rationality and agency might be ultimately tropistic; that, even though we take ourselves to be rational self-determiners, we are really moved by underlying causes (e.g. instinct). Kant's ultimate response to this problem, given in the Critique of Practical Reason, is that our consciousness of standing under the moral law (the 'fact of reason') assures us of our freedom from the practical point of view. 21 Kant does not take this route in the Groundwork, however, arguing instead that freedom is a necessary presupposition of reason in so far as it regards itself as practical. And here the point is not that we must believe ourselves to be free in order to believe that we are agents rather than automata; it is rather that we must act as if we were free, which is just what it means to act under the idea of freedom. In other words, the idea of freedom has an essentially normative force. To act under this idea is to place oneself in the 'space of [practical] reasons', and therefore to take oneself as subject to rational norms (of both a moral and prudential sort) rather than merely to causal conditions. Thus, even though it remains alive as a metaphysical possibility, from the practical point of view (that of agency) the spectre that we might be merely automata is perfectly idle. The suggestion, then, is that the presupposition of the logical or formal purposiveness of nature be understood in essentially the same way; that is, as having normative or prescriptive force. In investigating nature we ought to treat it as if it were purposive because this is just what is involved in
20 What Kant actually claims is that 'To every rational being possessed of a will we must also lend the idea of freedom as the only one under which he can act' (Gr iv. 448). What follows is based on my analysis given in 'We can Act only Under the Idea of Freedom', Proceedings and Addresses of the American Philosophical Association, 7 1 (1997), 39-50. 21 I analyse this in detail in Kant's Theory of Freedom (Cambridge: Cambridge University Press, 1990), ch. 13.
Reflective Judgement
181
'applying logic' to it. For reasons that should now be clear, there is simply no other procedure possible for judgement in its reflection on nature, at least not if the goal is to form empirical concepts that can be combined in something like judgements of experience. Accordingly, one might say that the principle of purposiveness characterizes the 'space of judgement', since it defines the framework in which alone rational reflection on nature is possible. And this serves to explain Kant's en1phasis on the a priori nature of the principle. There is nothing optional about approaching nature in this way, just as there is nothing optional about presupposing freedom in so far as we take ourselves to be rational agents. But, of course, this no more proves that nature really is purposive than the latter proves that we really are free.
III Does Kant's deduction, so construed, constitute a viable response to Humean scepticism regarding the rational basis of induction broadly conceived? Indeed, can we even regard it as a deduction? Setting aside the historical issue of whether Kant actually intended his account as an answer to Hume, with which I am not here concerned, the answer to both questions appears at first glance to be clearly negative. In fact, one might think that by appealing to a merely subjective necessity to presuppose logical or formal purposiveness Kant is really conceding Hume's point regarding the uniformity principle rather than answering him. Moreover, if it is taken as an attempted answer to Hume, then I(ant might seem to be guilty of the very error of which he famously accused Hume's Scottish common-sense critics in the Prolegomena, namely of taking for granted what Hume doubted and of demonstrating what he never thought of doubting (Prolegomena iv. 258). Nevertheless, I think that such a dismissal of Kant's argument and its antiHumean thrust would be mistaken. For, as in the case of causality and the other a priori concepts and principles that are at stake in the first Critique, Kant saw clearly that the basic question between himself and Hume was one of norn1ativity or right, not simply indispensability or pragmatic necessity, which Hume certainly did not deny in the case of the uniformity principle. Hume's question concerns the grounds for our belief in this principle; the belief itself and its indispensability are never called into question. And after arguing, quite correctly, that it is neither a demonstrable truth (since its negation is conceivable) nor a generalization fron1 experience (since it is a condition of the possibility of any such generalization), he concludes that this belief can be understood only as the product of custom or habit. 22
22 See Hume, Treatise, bk. Understanding, sect. V, pt. 1.
I,
pt. III, sect. XII, and Enquiry Concerning the Human
182
Allison
For Kant, by contrast, the necessity, though merely subjective, is nonetheless rational rather than causal or psychological. Moreover, it is not a matter of belief. As we have already seen, what is required is not the belief that nature is purposive, which would be inseparable from the belief that it has an intelligent author, but rather that we approach it as if it were. The principle thus has prescriptive force; it dictates how we are rationally constrained to approach nature, if our concern is to 'apply logic' to it; that is, to bring the data of experience under empirical concepts and laws. And if this is correct, as I have tried to show that it is, then Kant can claim to have grounded the right to make such a presupposition. But to do this is to have resolved the quid iuris, which is the proper concern of any Kantian transcendental deduction. In insisting on the genuineness of this deduction, however, I do not wish to minimize its differences from the transcendental deduction of the first Critique, including the argument for the Principles (particularly causality). There, it was a matter of understanding legislating to nature, of laying out the formal constraints on any thing that could become an object of empirical knowledge (what Kant termed the 'formal conditions of empirical truth' (A 19 liB 236). Accordingly, the principles that specify these constraints or conditions are in this sense constitutive of possible experience. Here, by contrast, the claim is not that nature or objects of possible experience must conform to certain conditions stemming from the nature of our cognitive faculties (what I have elsewhere called 'epistemic conditions'23), but merely that we must investigate nature on the basis of the assumption that it accords with the conditions of our cognition of it; namely, that it constitutes a 'logical system'. In this respect, the principle of purposiveness really functions as an imperative that judgement imposes on itself, since it expresses the condition of the possibility of the successful application of this faculty. Kant indicates the true nature and function of this principle when he claims that through it 'judgment prescribes not to nature (which would be autonon1Y) but to itself [my emphasis] (which is heautonomy), a law for the specification of nature' (KU v. 185-6, 25). Thus, even though the principle concerns nature as the object of investigation, its prescriptive force is directed back to judgement itself. 24 The term 'heautonomy' is introduced by Kant in order to en1phasize the purely reflexive, self-referential nature of
23 See Allison, Kant~s Transcendental Idealism (New Haven, Conn. London: Yale University Press, 1983), chs. I and 2. 24 At this point my analysis is very close to that of juliet Floyd, who likewise sees Kant's appeal to heautonomy as at the heart of his answer to Hume (See 'Heautonomy: Kant on Reflective judgment and Systematicity', in Herman Parret (ed.) (BerlinJNew York: Walter de Gruyter, 1998), Kant's Asthetik~ Kant~s Aesthetics, L~esthetique de Kant, esp. 206-14.
Reflective Judgement
this principle. 25 To claim that judgement is 'heautonomous' in its reflection is just to say that it is both source and referent of its own normativity. And this is what distinguishes judgement's a priori principle from those of the understanding, which legislates transcendental laws to nature, and of (practical) reason, which prescribes the objectively necessary laws of a free will. 25 The term 'heautonomy' derives from attaching the Greek definite article 'he' to the pronoun 'auto', which stands for either 'self' or 'itself'. (On this issue see Juliet Floyd, 'Heautonomy: Kant on Reflective Judgment and Systematicity', 205.)
II
Strawson on Aesthetic Judgement in Kant ECKART FORSTER
In his recent 'Intellectual Autobiography' that opens the Library of Living Philosophers volume in his honour Sir Peter Strawson reviews, among other things, his various publications on Kant subsequent to The Bounds of Sense. In this context he writes: 'More recently I paid tribute to his [i.e. Kant's] insight into the nature of aesthetic judgement.'l This attention to Kant's theory of beauty must be most welcome to anyone who, like myself, has learned greatly from Strawson's brilliant reading of the first Critique and who shares his high esteem for the Critique of Aesthetic Judgement. Unfortunately, the mentioned tribute is exceedingly short and condensed, occurring as it does in the context of a book review. Nevertheless, it is this tribute, or rather its two central claims with regard to Kant's aesthetics, that I want to discuss in what follows. Let me begin by citing both claims. First: One of the distinctive features of Kant's genius is his power of unifying his thought, of weaving together the many diverse strands of theory into a single fabric, or-to vary the image-of building a single complex structure out of many prima facie heterogeneous parts which are nevertheless exhibited as interlocking and mutually supportive. Nowhere is this power more strikingly manifested than in the third Critique, where the theory of aesthetic judgement is brilliantly integrated with the epistemology of the Critique of Pure Reason. 2 The second claim concerns the explanation of the role of the cognitive faculties in a judgement of taste, the alleged 'free play' of understanding and imagination: As Kant implies, and as any sensitive person appreciates, no general concept could conceivably capture or encapsulate the essential source of one's delight in the 1 The Philosophy of P. F. Strawson, Lewis Edwin Hahn (ed.), The Library of Living Philosophers, xxvi (Chicago and Lasalle, Ill.: Open Court, I998), 13. 2 P. F. Strawson, review of Paul Guyer (ed.), The Cambridge Companion to Kant, in European Journal of Philosophy, I (I993), 226.
186
Forster
beautiful object, whether natural scene or work of art ... One could say that while no general concept can capture the unique aesthetic essence of the beautiful thing, the thing, as object of beauty, is the necessary unique instance of its own necessarily individual concept, incapable of being expounded in general terms; or even that it embodies, or is, that concept itself (cf. some idealists' talk of the "concrete universal"); so that the very faculties, whose normal and mundane function is fulfilled when they reach for and find an already existing general concept, are here engaged in free and harmonious play around, or with, the unexponible "concept" embodied in the beautiful object. 3
I want to address both claims in turn. Whereas I have certain reservations about the second claim that I will spell out in Section II of this paper, I want to begin with some observations in support of Strawson's view regarding Kant's ability to integrate his theory of aesthetic judgement into the overall epistemological framework of the first Critique. This view, I take it, is shared by a number of COlnmentators. However, I want to suggest that Kant's remarkable ability to unify his thought can be made out even at those places where traditionally this ability has been questioned or challenged. And there are at least two aspects of the third Critique where this unifying power is not easily detectable-indeed, where the prima facie heterogeneous parts from which the structure is built seem heterogeneous even after two hundred years of scholarship on the text, or where the connection with the first Critique is anything but clear. The first aspect concerns the justification for treating the two heterogeneous parts of the third Critique-aesthetics and biology, or taste and teleology-in a single volume. Schopenhauer already called their connection a 'baroque' union, and this view has prevailed to the present day. For the traditional connection between beauty and objective purposiveness that from Plotinus to Leibniz had formed the basis of all physico-theology (cf. A 622/B 650)4 had been dismantled precisely by Kant himself with his critique of this position in the Dialectic of the Critique of Pure Reason. So why their renewed union in the third Critique? The second aspect concerns the relation between the regulative employment of the ideas of reason in the Appendix to the Dialectic in the first Critique and the principle of reflective judgement, i.e. the principle of a formal purposiveness of nature, in the third. Kant claims that the latter principle is an altogether new principle, yet virtually all commentators agree that this principle is just a condensed version of reason's three regulative principles
3 P. F. Strawson, review of Paul Guyer (ed.), The Cambridge Companion to Kant, in European Journal of Philosophy, I (1993), 227- 8. 4 References to Kant's Critique of Pure Reason are given in the text with the usual A and B numbering for the first and second editions respectively. References to Kant's other writings are given by volume and page of the Academy edition of Kant's Gesammelte Schriften (Berlin/Leipzig: de Gruyter, 1900 ).
Strawson on Aesthetic Judgement in Kant
187
of homogeneity, variety, and affinity. Moreover, they agree that Kant has given us no reason or word of explanation why such a condensation-with its associated reassignment of the principle from reason to judgement-is justified. One example from the literature can here stand for many: In the Critique of Pure Reason Kant assigns the origin as well as the employment of the regulative ideal of systematicity in empirical knowledge to the faculty of pure theoretical reason although, to be sure, to reason in its hypothetical rather than apodictic employment. In the Critique ofJudgement, however, published only three years after the revised second edition of the Critique of Pure Reason, the regulative ideal of systen1aticity is reassigned to the newly introduced faculty of reflective judgement. Kant offers some explanation of what he means by reflective judgement but he does not mention that the assignment of the regulative ideal of systematicity to this new faculty represents a revision of his previous view. Indeed, he does not even mention that he had a previous view about systematicity. 5
I
What, then, is the relation between Kant's two Critiques? The first thing we must notice is that, although published in 1790, Kant's plans for a critique of taste coincide with his work on the second edition of the Critique of Pure Reason. 6 In a letter to Schutz of 25 June 1787 !(ant reports that he is now about to begin work on a critique of taste (x. 490), and the catalogue for the Leipzig book fair of the same year already announces, together with the revised edition of the first Critique, a new book by I(ant under the title Grundlegung zur Critik des Geschmacks (x. 488). The most detailed statement of Kant's new plan, however, can be found in his letter to Carl Leonhard
5 Paul Guyer, 'Reason and Reflective Judgement: Kant on the Significance of Systematicity', in Nous, 24 (1998), 17. In the same vein Michael Friedman writes: 'What the principle of reflective judgen1ent actually generates here are merely the heuristic or methodological principles presented in the first Critique as products of the regulative use of reason at A 652-63/B 680-91 ... [I]t in no ways goes beyond the methodological regulative maxims already enumerated in the first Critique' (Kant and the Exact Sciences (Cambridge, Mass.: Harvard University Press, 1992), 251-3) ). For similar views see e.g. Henry Allison, 'Is the Critique ofJudgement "Post-Critical"?', in Sally Sedgwick (ed.), The Reception of Kant's Philosophy: Fichte, Schelling and Hegel, (Cambridge: Cambridge University Press, 2000); Rolf-Peter Horstmann, Bausteine kritischer Philosophie, (Bodenheim: Philo Verlagsgesellschaft, 1997), 149'-5°, 178-9. 6 A trace of a new development can be found in this text itself. In the first edition, in a footnote to the Transcendental Aesthetic, Kant had still insisted that any attempts to bring the treatment of the beautiful under rational principles are fruitless because its rules or criteria are merely empirical and can never serve as a priori laws by which our judgements of taste must be directed (A 21 n.). In the second edition of 1787, however, this passage is revised and Kant now claims that aesthetics is a part of speculative philosophy and must be viewed 'partly in the transcendental and partly in the psychological sense' (B 35 n.).
188
Forster
Reinhold of December 1787. Kant writes: I am now at work on the critique of taste, and I have discovered a kind of a priori principle different from those heretofore observed. For there are three faculties of the mind: the faculty of cognition, the faculty of feeling pleasure and displeasure, and the faculty of desire. In the Critique of [Theoretical] Reason I found a priori principles for the first of these, and in the Critique of Practical Reason a priori principles for the third. I tried to find them for the second as well, and, though I thought it impossible to find such principles, the systematic nature of the analysis of the previously mentioned faculties of the human mind allowed me to discover them, giving me ample material for the rest of my life, material at which to marvel and if possible to explore. So now I recognize three parts of philosophy, each of which has its own a priori principles, which can be enumerated and for which one can delimit precisely that knowledge that n1ay be based on them: theoretical philosophy, teleology, and practical philosophy ... I hope to have a manuscript on this completed though not in print by Easter; it will be entitled Critique of Taste. (x. 514-15)
Three things are worth stressing here. First, Kant claims to have discovered a new principle, not just condensed three heuristic maxims of reason into a single principle of judgement. Second, at this time Kant plans to write only a critique of taste. The Critique of Teleological Judgement clearly is of a later origin. Third, in this letter to Reinhold Kant explicitly identifies this critique of taste, or the analysis of the aesthetic judgement, with teleology. How is this to be understood? An answer to this question emerges if we take a closer look at Kant's analysis of this type of judgement. Like every good philosopher, Kant begins his investigation with a definition of what is to be examined. 'The definition of taste on which the following is based', he writes, 'is that it is the ability to judge the beautiful. The analysis of judgements of taste must discover what is required for calling an object beautiful' (v. 2°3). As the steps of Kant's analysis are well known, I can be brief in summarizing them. A judgement of the form 'x is beautiful', according to Kant, is not a cognitive judgement that determines an object. There are no rules of verification for judgements of taste as there are for cognitive judgements. Yet although the judgement 'x is beautiful' is subjective, it is not psychological. It differs from all subjective judgements concerning what is agreeable to a particular individual in that it is accompanied by a claim to, or expectation of, universal agreement. We expect, according to Kant, that other human perceivers of the object we judge as beautiful will agree with us-an expectation we do not associate with an expression of agreeableness. Expressions of agreeableness are tied to particular subjects in such a way that we can always ask, 'Agreeable to whom?', and we usually feel free in principle to be indifferent to, or even find disagreeable, what is agreeable to another. With regard to beauty, however, we do not ask 'Beautiful for whom?' any more than we would ask, 'Blue for
Strawson on Aesthetic Judgement in Kant
189
whom?' when the colour of the clear sky was pointed out to us. In a judgement of taste, Kant says, the pleasure felt by us is exacted from everyone else as 'necessary' just as if it were an objective judgement (v. 218). And it is precisely this feature of the aesthetic judgement, that it is subjective yet claims universal validity like an objective or cognitive judgement, that according to Kant demands the 'effort of the transcendental philosopher' (v. 213, cf. 266) to explain its origin. His explanation, very briefly, is the following: in the cognition of any object, imagination and understanding co-operate, the former apprehending and exhibiting the manifold that judgement then subsumes under a concept supplied by the latter. If the object is also beautiful, however, judgement realizes that this subsumption does not exhaust the object; rather, it finds in its reflection that understanding and imagination now stand in a 'free play' in which the imagination's ability to exhibit and the understanding's ability to conceptualize vivify and further each other. To this free play we respond with the feeling of pleasure. We express this pleasure by saying that the object that gives rise to this free play and its subsequent delight is beautiful. Since imagination and understanding are assun1ed to exist in all human beings, we also assume that others will feel the same pleasure in view of the object and hence agree with our judgement. 'The quickening of both faculties (imagination and understanding) to an indefinite, but yet, thanks to the given representation' harmonious activity, such as belongs to cognition in general, is the sensation whose universal communicability is postulated by the judgement of taste' (v. 219). If this is the explanation of why we can expect agreement in matters of beauty, there is a consequence to this explanation that must not be overlooked. If beauty does not determine an object but signifies the sensation of the harmonious play of the two faculties involved in all cognition, it follows that only beings with these two faculties can experience their free play and hence enjoy beauty. Kant emphasizes this point as early as § 5 of the Critique of Judgement: 'Agreeableness holds for non-rational animals too; beauty only for human beings, i.e. beings who are animal and yet rational, though it is not enough that they be rational (e.g. spirits) but they must be animals as well; the good, however, holds for every rational being as such' (v. 210). Thus only human beings can experience beauty. Moreover, on Kant's explanation of aesthetic judgement, apart from any reference to the sensation of the human observer, beauty is 'nothing' (v. 218) for and in itself. It must seem, therefore, as if certain products of nature-namely those we judge to be beautiful-were there for the sole purpose of human enjoyment. Natural beauty exhibits a purposiveness in its form that, because entirely contingent from the viewpoint of the transcendental concept of nature, makes the object appear as it were predetermined for our power of judgement (v. 245). And it
190
Forster
is precisely this fact, I suggest, that underlies Kant's claim that he has discovered a new a priori principle. Because of the actual experience of natural beauty, and only because of it, judgement is compelled in its reflection' to adopt as its own principle the view that nature specifies its empirical laws for the purpose of judgement. 7 What those commentators who allege that this principle can already be found in the first Critique seem to overlook is Kant's repeated insistence that the principle of a formal purposiveness of nature is disclosed not by reason's own systematic tendencies, nor by any teleological reflections, but solely by aesthetic judgements concerning natural beauty. Thus Kant writes for example: So it is only in taste, more precisely only in taste concerning objects of nature, that judgement reveals itself as a faculty that has its own principle and hence is justified in claiming a place in the general critique of the higher cognitive faculties-a place one might otherwise not have expected for it. (xx. 244; my italics)
Or: [A ] esthetic judgement alone contains a principle that judgement lays completely a priori at the basis of its reflection on nature: the principle of a formal purposiveness of nature, in terms of its particular (en1pirical) laws, for our cognitive powers. (v. 193; my italics)
Or: Independent natural beauty reveals [entdeckt] to us a technic of nature that allows us to represent nature as a system in terms of laws whose principle we do not find anywhere in our understanding. (v. 246; my italics)
This is why Kant can speak of 'teleology' when describing the principle of taste in his letter to Reinhold. The experience of independent natural beauty, and only it, 'expands' (erweitert) our concept of nature (v. 246). It expands our concept of nature from that of a blind mechanism-the concept of nature of the first Critique and the Metaphysical Foundation of Natural Science-to nature as art, hence to a nature that is in itself purposive and systelnatic. And such expansion of the concept of nature is of course not at all operative in the Appendix to the Dialectic of the first Critique, where reason seeks to bring greatest possible unity into the various blindly mechanistic cognitions supplied to it by the operations of the understanding. So the expanded view of nature gives rise to a new a priori principle of
7 'There can be purposiveness without a purpose, in so far as we do not posit the cause of this form in a will, and yet can make the explanation of its possibility intelligible to ourselves only by deriving it from a will. Now what we observe we do not always need to have insight into by reason (as to how it is possible). Hence we may at least observe a purposiveness as to form and take note of it in objects-even if only by reflection-without basing it on a purpose (as the matter of the nexus (ina/is)' (v. 220; my italics).
Strawson on Aesthetic Judgement in Kant
191
nature's own systematicity with regard to our cognitive powers, a principle quite different from reason's own heuristic n1axims. Kant's remarkable way of unifying different elen1ents of his philosophy can now be seen in a clearer light. Initially, at the outset of his transcendental enterprise, he had sought only to investigate the possibility of a metaphysics of nature. The fundamental principle of morals, he thought he had learned from Rousseau, is well known even to the populace. Transcendental philosophy is thus, in its initial conception, only an investigation into the possibility of a priori knowledge of objects. More precisely, it is an invest.igation into the problem of how a priori concepts can determine objects without the aid of experience. Whence the alleged 'agreement' of pure concepts of the understanding with their objects, when no experience guides the reference? This is the question that Kant first formulated in the famous letter to Marcus Herz of 1772 (cf. x. 131), and that gave rise to the definition of transcendental knowledge at the beginning of the Critique of Pure Reason (at A 11-12). Transcendental knowledge, thus understood as addressing the problem of veridical yet non-empirical reference, excludes morals, for. in n10rals the reference of a priori concepts to their 0bj ects is unproblematic: the concept or representation of what ought to be the case is supposed to bring about its corresponding object or state of affairs (cf. x. 130). Morals is consequently entirely 'foreign to transcendental philosophy' (as Kant writes at A 801), since the latter 'is exclusively concerned with pure a priori modes of knowledge'. Yet the very first review of the Critique already forced Kant to the realization that (a) the fundamental principle of morals, i.e. the categorial imperative, is not as clearly known and unquestionably acknowledged as he had thought; and that (b) the 'question of how a categorial imperative is possible' has a 'striking similarity with the problem of transcendental philosophy' (xxiii. 60); namely, how synthetic knowledge is to be possible a priori. The immediate result of this realization is the Groundwork to a Metaphysics of Morals of 1785,8 but when in the following year a second edition of the Critique had to be prepared, Kant at first decided to incorporate a critique of practical reason into the text (cf. iii. 556, x. 4 69, 471) This plan was soon abandoned. Lewis White Beck suggested that it was abandoned because the inclusion of a critique of practical reason into the corpus of the first Critique would have destroyed its architectonic structure, and would have forced the already long text to grow out of proportion. 9
Cf. Eckart Forster, Kant's Final Synthesis (Cambridge, Mass.: Harvard University Press, ch. 5. 9 Cf. Lewis White Beck, A Commentary on Kant's Critique of Practical Reason (Chicago, Ill./London: The University of Chicago Press, 19 84), 15. 8
2000),
Forster This is no doubt correct, but I think the real reason for Kant's change of mind can be detected in the letter to Reinhold cited above. Once the general question of how synthetic knowledge is possible a priori is placed at the centre of Kant's critical philosophy, it is only natural to ask whether the 'third faculty of the mind', the faculty of feeling pleasure and displeasure, also has an a priori principle of its own. And, given the epistemological apparatus he had developed in the first Critique,10 Kant now had to answer this question in the affirmative (cf. v. 289). The critique of taste would thus also have to be included in the new edition of the first Critique, alongside its practical counterpart, and it cannot come as a surprise that Kant opted instead for three separate Critiques. Prior to the first Critique, in the I770s, Kant had thought that taste has no a priori principle of its o\vn. To be sure, he assumed all along that what is agreeable is merely subjective whereas judgements of beauty also exact agreement. Yet he could only explain this fact empirically, in terms of a shared subjectivity, or a sensus communis: 'Taste is the power of judgement of the senses, through which we recognize what agrees with the sense of others; it is thus a pleasure or displeasure in community with others' (xxviii. 25 I). More precisely, what agrees with the laws of sensation in general, such as the perception of order and harmony in an object, or of an idea of the whole, must please generally and is called beautiful. But it cannot be subject to a priori principles; the rules of taste (order, harmony, unity, etc.) must always be based on what is given in experience (cf. xxviii. 251). After the first Critique, however, Kant realized that he could give an explanation of aesthetic judgement entirely in terms of those conditions involved in the cognition of objects. Moreover, he realized that this explanation would now also allow him to give an a priori justification-a deductionof aesthetic judgement's claim to universal agreement that would link it with the results of his investigations into moral philosophy. Thus, by integrating the claims of taste with the conditions of knowledge in general, there opened up the previously unforeseen possibility of unifying the areas of cognition, morals, and taste into one mutually supporting whole. (I shall return to this point later, in Section III.) But why add to the 'Critique of Taste' a critique of teleological judgement? On the one hand, both types of judgement, aesthetic as well as teleological, belong to the same faculty; nan1ely, the power of reflective judgement. And since this power has been shown to have its own a priori principle, it is incumbent upon the transcendental philosopher to examine the origin, limit, and extent (cf. A xii) of this principle, and to subject its faculty to a critique 192
10 'The systematic nature of the analysis of the previously mentioned faculties of the human mind allowed me to discover' the a priori principle of taste, Kant explained in the letter to Reinhold.
Strawson on Aesthetic Judgement in Kant
193
in its entirety, not just in its employment in matters of taste. (This is the argument of the First Introduction, xx. 244.) Moreover, just as we may reflect upon natural beauty as exhibiting the concept of the formal (merely subjective) purposiveness of nature, we may reflect on living organisms as exhibiting the concept of a real or objective (not merely subjective) purposiveness of nature. (This is the argument of the Second Introduction, v. 193). From this perspective, too, the two parts of the third Critique may be said to have an inner unity. Nevertheless, there is also an important difference between aesthetic and teleological judgements. The experience of beauty is such that an otherwise . unknown harmonious play of imagination and understanding can be sensed in the form of pleasure. And it is this subjective yet generalizable sensation, according to Kant, that gives rise to the principle of a subjective purposiveness of nature in its forms, with regard to our powers of cognition. Yet there is, Kant thought, no corresponding experience of purposiveness in products of nature. Whereas the aesthetic estimation of natural beauty discovers a purposiveness of nature for the subject, any purposiveness in the object is completely 'beyond' (xx. 233) the power of judgement: 'what is more, even experience cannot prove the actuality of such purposes' (v. 359). We can never know from experience whether an object is indeed purposive or solely the product of mechanical causes. It is for this reason that Kant said that teleological judgement, unlike aesthetic judgement, could also have been treated as an appendix to the theoretical part of philosophy (cf. v. 170, 194). And here we may see a strong reason for denying a genuine internal unity to the third Critique. Perhaps. But we have not yet heard Kant's last word on this matter. Because in the third Critique he had contrasted natural purposes (which organize themselves) with human artefacts (whose designer is always external to them), he concluded that the self-organization of nature has nothing analogous to any causality known to us. But soon after Kant began to realize that it is indeed through experience that we have the concept of a natural purpose, but it is not the human artefact and the realization of practical purposes that originally permits the formation of this concept. Rather, it is the experience of our own bodily organization, of the harmonious play of our physical and mental powers in the exercise of intentionally moving forces: 'Because man is conscious of himself as a self-moving machine, without being able to further understand such a possibility, he can, and is entitled to, introduce a priori organic-moving forces of bodies into the classification of bodies in general' (xxi. 213, Opus postumum, 66). Our own bodily experience functions as the paradigm for the estimation of other bodies as organic; it is the primary example by which we judge all others. If this is correct, new possibilities open up for a further systematization of Kant's philosophy-possibilities he explored especially in the Selbstsetzungslehre of the Opus postumum. But this
194 Forster is a topic I do not want to explore further here. It is tin1e to return to the notion of a free play of our cognitive powers in the third Critique, and the explanation Strawson provides of it.
II The background of Strawson's explication of aesthetic judgement in Kant, if I understand it correctly, is marked by two general observations: first, that the object of aesthetic appraisal is individual and unique (inimitable), and, second, that the aesthetic appraisal is lacking in general descriptive criteria. 'When you draw attention to some feature on account of which terms of aesthetic evaluation may be bestowed', he wrote some years earlier in 'Aesthetic Appraisal and Works of Art', 'you draw attention, not to a property which different individual works of art may share, but to a part or aspect of an individual work of art ... If this is true, then the impossibility of general descriptive criteria of aesthetic excellence follows as a consequence.'ll And again: 'When we have a class of objects of which the name, "works of art" , marks them out as primarily to be assessed in this way, then there cannot be numerically distinct members of the class, or parts of these members, which yet share all the features relevant to this kind of assessment.'12 The view that seems to emerge, then, is that, since to recognize something as something is to see it as falling under a concept, what makes us recognize an object as beautiful can only be captured by a concept of which the object is a necessarily unique instance, i.e. by an individual concept. Our normal tendency to look for a general concept fails in the case of the aesthetic appraisal, because no generally shared properties can account for the fact that one object nlay be beautiful, while a qualitatively indistinguishable one is not. In order to explain the delight the one object arouses in nle while the other fails to do the same, I perforce must look for an individual concept of which the beautiful object is the sale instance. Or, put differently, if two qualitatively indistinguishable objects (in the relevant sense) are both said to be beautiful, it is not because they have a general property in cornman (whatever properties they in fact share are irrelevant in the aesthetic case I3 ), but because the one object falls under an individual concept under which the other could not possibly fall, and the other object falls under a different individual concept that is also, and necessarily so, inapplicable to any other object. The free play of which Kant speaks would thus consists in the fact that the individual concept is inexponible, that consequently the essence of the beautiful is inexpressible although we always try to express 11 P. F. Strawson, 'Aesthetic Appraisal and Works of Art', in Freedom and Resentment and Other Essays (London: Methuen, 1974), 186. 12 Ibid. 187. 13 See ibid. 188.
Strawson on Aesthetic Judgement in Kant
195
it. The freedom consists in a failure, albeit a necessary one, to express with an individual concept what only a general one could express, i.e. something of general validity. Or, in Strawson's words: 'the very faculties, whose normal and mundane function is fulfilled when they reach for and find an already existing general concept, are here engaged in free and harmonious play around, or with, the inexponible "concept" embodied in the beautiful object'.14 How well does this account capture the spirit of Kant's analysis of aesthetic judgements? A first response might be to object that Kant does not know 'individual' concepts in the sense suggested. When he defines the concept 'concept', as for instance in his logic lectures, he defines it as a general representation (representatio per notas communes); that is, as a representation of what is common to several objects (cf. ix. 91). Hence concepts are essentially general in nature, whereas intuitions, by contrast, are always singular representations, or particular. This is also the view familiar from the first Critique. Nevertheless, this objection, I think, would be misguided. For there are also passages in the Reflexionen where Kant speaks of conceptus singulares (cf. xvi. 342). As examples of these he lists such concepts as 'the earth', or the proper name 'Julius Caesar'. Hence, although not explicitly mentioned by Kant in the Critique of Judgement, one might nevertheless feel justified in bringing individual concepts to bear on his analysis of aesthetic judgements. More problematic, however, seen1S Kant's repeated insistence that no concepts at all are of relevance in such judgements. For example: The judgement of taste is differentiated from logical judgement by the fact that, whereas the latter subsumes a representation under a concept of the object, the judgement of taste does not subsume under a concept at all. (v. 286)
Or: Taste ... contains a principle of subsumption, not of intuitions under concepts, but of the faculty of intuition or exhibition [Darstellung], i.e. of the imagination, under the faculty of concepts, i.e. the understanding, so far as the former in its freedom accords with the latter in its conformity to law. (v. 287)
Kant does not speak in these passages of a subsumption of an intuition under a concept-general or individual-but of two faculties, in1agination and understanding, that are said to enter into a free or harmonious relation when confronted with an object of beauty. Now I suspect that it is Strawson's well-known dislike for 'faculty talk' that makes him look for an account of aesthetic judgement in Kant that bypasses essential reference to faculties. We will thus have to see if an explanation of I
14
P. F. Strawson, 'Review',
227- 8.
196
Forster
possible that can do without individual concepts yet that does not run into some of Strawson's fundamental objections. It is noteworthy, I take it, that Kant (although perhaps not consistently so) characterizes what is relevant about both faculties in terms of their ability to exhibit (or darstellen). We have just heard that taste contains a principle of subsumption 'not of intuitions under concepts, but of the faculty of intuition or exhibition [Darstellung], i.e. of the imagination, under the faculty of concepts, i.e. the understanding' (v. 287). And in the First Introduction we read that the judgement of taste compares imagination (regarded merely in the Auffassung of a given individual object) with the understanding . (regarded merely in the Darstellung of a concept in general) (cf. xx. 223). In this passage the understanding's ability to exhibit and the imagination's ability, not to exhibit but to auffassen, are said to further each other harmoniously, yet this is no cause for irritation, for Kant states explicitly that 'the power of Darstellung and that power of Auffassung are one and the same' (v. 279). The power of Darstellung or exhibition, however, 'is the imagination' (v. 232). Thus the functions of imagination and understanding that are relevant for Kant's analysis are characterized with the same ternl: Darstellung, or exhibition. is What does this nlean? In the first Critique Kant had defined 'exhibition' in general as referring a representation to experience (possible or actual) (cf. A 156/B 195). And in the third Critique we read that to exhibit a concept means 'to place beside the concept an intuition corresponding to it' (v. 192). Thus it is easy to see what it means to view the understanding in its ability to exhibit a concept in general: it is to view the understanding simply as applying concepts. But what does it mean to say that imagination is the faculty of exhibition? To answer this question we must turn to the first Critique and the deduction of the categories. In the cognition of any object of sense, Kant argued there, three things must come together. First, the manifold data of sense must be run through and held together in order to be represented as a unity. This Kant calls the synthesis of apprehension in intuition. Second, what has been thus run through and held together must also be reproduced, Kant insists-for mere apprehension does not by itself yield a connection of perceptions. For the latter to arise, I must also reinstate a previous perception alongside the subsequent perception to which it has passed, and so form whole series of perceptions. It is in this context that I(ant introduces the term 'imagination' (A 100 ff.), which he also defines as the power of representing in intuition an object that is not itself present (B 15 I). But this reproduction of the imagination, 15 The importance of exhibition in Kant's account of the free play of the faculties is also emphasized by Dieter Henrich in Aesthetic Judgment and the Moral Image of the World (Stanford, Calif.: Stanford University Press, 1992),47 ff.
Strawson on Aesthetic Judgement in Kant 197 thirdly, if knowledge of an object is to arise from it, must be not arbitrary but rule-governed. The imagination must connect representations not just in any order they happened to come together, but in the order in which they belong together. The rule that guides the imagination in its synthesis is the concept of the object in question. For a concept, in Kant's terminology, is that which 'combines the n1anifold, successively intuited, and thereupon also reproduced, into one representation' (A 103). It is precisely the second component of "\vhich Kant here speaks, that of imagination in the cognition of an object, that Strawson subjected to a highly instructive analysis in his essay 'Imagination and Perception'. Using "the perception of a dog as an example, he argued that the thought of past perceptions is necessarily alive, although not necessarily consciously so, in the present perception of the animal. For we could not count any transient perception as a perception of an enduring object of some kind, so he argued, unless we also counted some different transient perceptions as perceptions of the same object. And this sort of combination is of course dependent on the possession and application of a concept of the object in question. 16 But to combine and conceptualize our perceptions in this way is, under normal circumstances, to see the object as being of a particular kind, and thus to see the present perception as also related to, or anin1ated by, other possible perceptions perhaps not yet encountered. In Strawson's example, to see the object as a dog, silent and stationary, is to see it as a possible mover and barker, even when we have only perceived this dog in its stationary position. 'It seems, then, not too much to say that the actual occurrent perception of an enduring object as an object of a certain kind, or as a particular object of that kind, is, as it were, soaked with or animated by, or infused with-the metaphors are a choix-the thought of other past or possible perceptions of the same object.'17 Kant's remark that imagination is the faculty of exhibition can now be given a more precise sense. It is the mind's ability to associate with a given perception past or possible perceptions of the same object. To say that a present perception of an object is 'soaked' with past or possible perceptions of the same object would be another way of saying that in any knowledge of an object, sense, imagination, and concept must be intimately joined. Let us now, with this Strawsonian result in mind, return to Kant's analysis of the aesthetic judgen1ent. A judgement of this kind always presupposes a perceptual object to which it is applied. Sense, imagination, and understanding must jointly have yielded an object of sense before the judgement 'This is beautiful' can be made. And this latter judgement, according to
16 Cf. P. F. Strawson, 'Imagination in Perception', in Freedom and Resentment and Other Essays (London: Methuen, 1974), 51-2. 17 Ibid. 53.
198
Forster
Kant's additional claim, is the expression of the feeling that the given object of perception sets both imagination and understanding in a free and mutually vivifying relationship: 'The cognitive powers brought into play by this representation are here engaged in a free play, because no definite concept restricts them to a particular rule of cognition' (v. 2 I 7). This was not the case in the example discussed above, the case of a sleeping dog. The past and possible perceptions with which the actual perception of the dog could be said to be soaked had to be past and possible perceptions of a dog. The imagination was not free in its activity of animating the actual observation with non-actual perceptions but restricted to the particular rule of cognition supplied by the understanding, namely, the concept 'dog'. If now the imagination can be said to be free in its exercise, it must exhibit representations or non-actual perceptions of the object unrestricted by the concept. But, as we have noted, it must not do this arbitrarily. Freedom, for Kant, never means without order. That is to say, the connection of representations by the imagination must not be 'groundless' (v. 342), but in a certain order in which they could be said to belong together, in virtue of the given perception of the object to be labelled beautiful. Moreover, if the free play is mutual, these must be representations to which the understanding could apply concepts on the basis of the given perception. Let us try to unpack these metaphors some more. In its employment on behalf of cognition, the imagination was subjected to the constraint of the understanding and had to conform in its activity to a specific concept of the object supplied by the latter. In its aesthetic employment, Kant writes, in1agination 'is free to furnish of its own accord, over and above that agreement with the concept, a wealth of undeveloped material for the understanding, to which the latter paid no regard in its concept' (v. 316-17; my italics). The wealth of undeveloped material mentioned here Kant also characterizes as 'such a multiplicity of supplementary representations [Nebenvorstellungen] bound up with the representation of the imagination, that no expression standing for a determinate concept can be found for it, [but that] allows a concept to be supplemented in thought by much that is ineffable' (v. 316). What is judged to be beautiful gives rise to representations of the imagination that induce much thought to which no determinate concept can be adequate but that is neither arbitrary nor groundless. On the contrary, what is judged to be beautiful allows the imagination to show itself in its 'purposiveness' (v. 317) for exhibiting concepts in general and thus its 'harmony' with the understanding's conformity to law. The imagination vivifies the understanding and aesthetically expands its concepts, Kant writes, when it supplements (unterlegen) these concepts with representations of its own that, even though they belong to the exhibition of the concepts, prompt thoughts concerning implications and kinship with other concepts such as could never be comprehended within any determinate concept as such. They open up,
Strawson on Aesthetic Judgement in Kant
199
in Kant's phrase, 'a view into an in1mense realm of kindred representations' (cf. v. 314-15). If an example is desired to illustrate Kant's general point, take the apple the snake puts into Eve's hand as she reaches out to Adam in Durer's famous 15°4 copperplate engraving Der Sundenfall. Immediately the viewer sees the fruit, certain connotations or Nebenvorstellungen will enter the mind, of ten1ptation and fertility, of sin and release from innocence. Or it may induce the thought of a felix culpa and the birth of human consciousness, as for instance in Hegel and other German Idealists. In a different context, the apple may symbolize immortality, like the golden apples of the Hesperides in Greek mythology, or worldly power, when depicted as an imperial orb, as for instance in Durer's Nuremberg painting of Charlemagne, or again, it may be associated with thoughts of spiritual knowledge, as in the epics of the Celtic traditions. And so on. What Kant means by the feeling of a free play of the faculties, then, is the realization of the essential inexhaustibility of the aesthetic object, of its permitting a wealth of interpretations of which none can be the definite or final one. Beauty not only permits but encourages an imaginative activity of the mind that exceeds any restricted conceptual determination. This encouragement we experience as pleasant, as stimulating and vivifying, and we give expression to it by judging the object to be beautiful. The range of possible interpretations in this case is unlimited and changes with time. New concepts will induce the imagination to exhibit new and unforseen representations; the imagination's exhibition in turn will yield new but associated conceptual relations that we may never have thought of previously. Kant's analysis of aesthetic judgement thus not only confirms the fundamental principle of hermeneutics that different eras are bound to interpret works of art differently, but also provides us with an explanation of why this is so. This feature of Kant's theory is not, or at least not clearly, brought out by its explanation in terms of individual concepts. But there is another aspect of Kant's account of aesthetic judgement, it seems to me, that such explanation fails to do justice to, and it is an aspect to which I want to turn in the last section of my paper. It will also allow us to reconnect with the argument of the first section.
III A satisfactory account of aesthetic judgement in Kant must somehow address the fact that it is not so much the ineffability of its object but the justification of its claim to universal validity that prompted Kant's 'Critique of Taste' (cf. v. 266). As we saw previously, in his pre-critical period Kant had always listed this claim among the characteristics of aesthetic judgements,
Forster but had regarded it as empirical, as quid factum, not quid iuris. This changed with the integration of his theory of taste into the wider epistemological framework of the critical philosophy. He now regarded the expectation of agreement as justifiable a priori, the agreement itself as a 'kind of duty' (v. 296). How is a deduction of the demand to agreement to be understood? First, such a deduction cannot be expected in the context of a mere analysis of a given type of judgement. The mere analysis of something given, in this case the occurrence of a particular kind of judgement, cannot justify a normative claim. The Analytic of Aesthetic Judgement can in this regard be compared with the procedure of the Prolegomena. There Kant had listed certain synthetic a priori judgements in mathematics and the natural sciences that he himself regarded as unproblematic, and proceeded to analyse them regarding their possibility. But the investigation only yielded a result that was valid relative to the premise; that is, relative to the assumption that there are indeed valid synthetic a priori judgements in mathematics and the sciences. That is why an analytic procedure like that of the Prolegomena was legitimate only after the Critique of Pure Reason had synthetically established the possibility in principle of such a type of judgement (cf. iv. 274-5.), and was adopted only to endow with popularity an otherwise exceedingly dry and complicated philosophical argument. Similarly, the reference to certain cognitive faculties that all human beings have in common can explain why, as a nlatter of fact, we expect agreement in aesthetic judgements if indeed we do so, but not that we are justified in exacting such agreement from others as a 'kind of duty'. An analytic argument can only explicate what is implied by such judgements; it cannot itself justify their claim to universal validity. First appearances to the contrary notwithstanding, Kant does indeed not give a deduction of the aesthetic judgement's claim to universality in the Analytic of the Critique of Aesthetic Judgement. Although §§30-8 are entitled 'Deduction of pure Aesthetic Judgements', Kant here does little more than repeat the results of the previous analysis of the judgement of taste. That no deduction has taken place in §§30-8 is also clear from the end of §40, where he writes: 'Supposing, now, that we could assume that the mere universal communicability of our feeling nlust of itself carry with it an interest for us ... we should then be in a position to explain how the feeling in the judgement of taste comes to be exacted from everyone as a kind of duty' (v. 296). In other words, we are not yet in this position and cannot make the said assumption. And at § 57, almost at the end of the Dialectic, Kant makes explicit that the deduction is still far from being completed when he remarks: 'if it be granted that our deduction is at least on the right track, even if not yet sufficiently clarified in all details' (v. 346). As a matter of fact, the deduction is not completed until the very last section of the Dialectic (§59), the last but one of the 'Critique of Taste'. 200
Strawson on Aesthetic Judgement in Kant
201
Thus the mere explication of aesthetic judgement was only a first step in Kant's account,18 and it is for a good reason that he moves from the Analytic to a dialectic, from natural beauty to fine art, and from an analysis of aesthetic judgement to a discussion of genius. How does the argument proceed? We just noted that the aesthetic claim to agreement could be given a deduction if 'the mere ... communicability of our feeling must of itself carry with it an interest for us'. It could then be exacted from everyone as a kind of duty. Already in §4 Kant had pointed out that delight in the good is always conjoined with an interest, for it is the object of a pure will. To will something and to be interested in its existence are identical representations for Kant. As moral agents we are also inevitably interested in the consequences of our deeds, for the realization of the highest good in this world must be the ultimate end of all moral actions. Since we can only postulate the objective reality of the concept of the highest good, we are bound to be interested in any 'trace' or 'hint" nature might give us that she permits the realization of our moral ends in the physical realm. This is why natural beauty must arouse the interest of anyone with a moral sensibility. It cannot be a matter of indifference to such a person that nature produces beautiful forms and thus indicates a formal purposiveness for our human powers and abilities. Although the aesthetic judgement itself is essentially disinterested, a subsequent or associated interest in natural beauty must go hand in hand with a moral sensitivity: 'the mind cannot reflect on the beauty of nature without at the same time finding its interest engaged' (v. 300). If we accept this argument, we may say that an explanation is forthcoming of why we demand agreenlent fronl other human beings with regard to natural beauty. Since our ultimate end is the realization of a moral world in which all rational beings act in accordance with the moral law, we cannot be indifferent to the moral sensibility of other human agents. Yet the limitation of this kind of argument is also obvious. It does not seem to extend easily to artistic beauty and fine art. That is why, beginning with §43, Kant shifts his focus from natural beauty to a discussion of fine art. What is fine art? It is, first and foremost, the product of a creative artist, of what the eighteenth century called 'genius'. Genius presupposes, in addition to an extensive technical mastery, a talent, a creative originality that cannot be copied or taught in a set of rules. Many skills can and must be acquired before a work of art becomes possible, but the creative talent itself is a gift of nature that cannot be passed on to pupils. Genius is a natural endowment, Kant emphasizes, an 'innate mental predisposition [ingenium] through which nature gives the rule to art' (v. 307). This natural endowment, 18 'Although the empirical exposition of aesthetic judgements may be the first step towards accumulating the material for a higher investigation, yet a transcendental examination of this faculty is possible, and forms an essential part of the critique of taste' (v. 278).
202
Forster
more precisely, is an animating principle of the mind that enables the artist to exhibit aesthetic ideas in such a way that, as it were, a second nature is created out of the material of the first, the real nature. And here we see an important point of contact with the previous analysis of judgements of taste, for the 'aesthetic ideas' the genius exhibits in her work are nothing other than those representations of the imagination 'which pron1pt much thought, but to which no determinate thought, i.e. concept, can ever be adequate' (v. 314). These are the representations of the aesthetic imagination that we encountered earlier, so that Kant can define genius also as 'the exemplary originality of the natural endown1ents of the individual in the free employment of his cognitive faculties' (v. 318).19 With these characterizations of artistic creativity the argument of the Analytic comes to a close, and Kant shifts to the Dialectic in order to engender an antinomy of taste that ,;yill force upon the reader the distinction between two different concepts of nature: the physical or empirical realm of nature, and the nature that can be said to underlie physical nature as its ground and to which Kant gives the name of supersensible nature. For the nature that gives the rule to art by endowing the genius with a special talent cannot be the physical nature known from the first Critique, the nature constituted by the transcendental principles of cognition. The natural gift that permits the genius to use her faculties to create as it were a second nature from the n1aterial of the first is as independent of the causal mechanisms of the physical realm as our cognitive powers are in the constitution of the objects of sense (cf. v. 344). With this much established Kant only needs to add one further elaboration of the artistic activity to reach the goal of his deduction. Genius, we said, produces a second nature with the material of the first-that means, n10re precisely, that the artist exhibits in his work representations of the imagination that are Nebenvorstellungen or aesthetic ideas of given concepts. In this way, the concepts acquire corresponding intuitions, but not directly, as when an instantiation of a given concept is pointed to, or when a pure concept is schematized, but indirectly, and by way of an analogy. Kant calls this a 'symbolic' exhibition and illustrates its general point with the help of the following example: a monarchical state, he says, is symbolically represented as a living body when it is governed by constitutional laws, but as a mere machine (like a hand-mill) when it is governed by an individual absolute will. Here the hand-mill, which crushes what it contains, is used to symbolically represent what the despotic head does to the people contained in his state. Judgement here performs a double function: first, it applies 19 Genius consists in 'the happy relation ... first, of finding ideas for a given concept and, second, of hitting upon a way of expressing these ideas such that the subjective mental attunement enduced by them can be communicated to others, as the concomitant of a concept' (v. 317).
Strawson on Aesthetic Judgement in Kant
203
a concept to the object of a sensible intuition, and then, second, it applies the n1ere rule by which it reflects on that intuition to a second object, of which the former is only the symbol (v. 352). The way we reflect on a hand-mill is transferred to, or used in, our reflection about a despot. Now Kant is in a position to state his explanation of why we rightly demand universal agreement for judgements of taste: a deduction of this demand is possible because we can say that beauty is the symbol of the morally good: 'Now, I say, the beautiful is the symbol of the morally good, and only in this light . .. does it give us pleasure with an attendant claim to the agreement of everyone else' (v. 353; my italics). Why? We have two objects in the wider sense: beauty, on the one hand, and the morally good, i.e. 'the ultimate end of humanity' (v. 298),20 on the other. If the forn1er, beauty, is the symbol of the latter, the morally good, then there must be an analogy in the way in which we reflect upon both objects. Now with regard to the former Kant points out that in respect of such objects of pure delight judgement 'gives a law to itself, just as reason does in respect of the faculty of desire' (v. 353). How does reason do that? This is stated most clearly in the Critique of Practical Reason: 'This [moral] law gives to the sensible world, as sensuous nature (as this concerns rational beings), the form of an intelligible world, i.e. the form of a supersensuous nature, without interferring with the rnechanisn1 of the former' (v. 43).21 The moral reflection consist in grasping a non-empirical idea or law, the moral law, and in prescribing a course of action in the empirical world that realizes in this world what ought to be the case. That is, as moral agents we ought to change existing nature in accordance with the idea of the form of another world. Analogously, the artist creates a second nature, by reshaping the physical world in accordance with non-en1pirical or aesthetic ideas. When reflecting on beauty and the morally good we realize that in both cases a form is being imposed on the physical realm such that the result must be regarded as the exhibition of ideas-aesthetic or moral. For this reason, it is incumbent upon everyone to take an interest in artistic beauty as much as in natural beauty, and I am justified in demanding agreement with my judgn1ents of taste from everyone else. In conclusion: the full extent of Kant's remarkable ability to unify the diverse elements of his philosophy comes to the fore only if we look beyond
20 In this way § 59 is reconnected with §42 and the discussion of the moral interest in natural beauty. 21 'Through reason we are conscious of a law to which all our maxims are subject as though through our will a natural order must arise. Therefore, this law must be the idea of a supersensible nature, a nature not empirically given yet possible through freedom; to this nature we give objective reality, at least in a practical context, because we regard it as the object of our will as pure rational beings' (v. 44).
204
Forster
the inexponibility of the aesthetic appraisal and also attend to his insistence that judgements of taste essentially make a claim to universal validity, and that this claim is capable of being justified in a deduction of its rightfulness. Only from this angle can we see the full extent to which for Kant epistemology, morals, aesthetics, and teleology mutually support one another and form one integrated whole that might itself, it seems to me, be termed beautiful.
12
Transcendental Arguments and Realism THOMAS GRUNDMANN AND CATRIN MISSELHORN
It almost amounts to a platitude that realism and transcendental arguments do not go well together. To see why, let us briefly rehearse the characteristics of transcendental arguments. 1 (i) Their premisses have to be justified independently of any experience about the external world, i.e. they must be justified introspectively or a priori. (ii) The premisses cannot be rejected by the sceptic. (iii) Transcendental arguments uncover necessary conditions of thought and experience. (iv) The conclusion of the argument entails truths about the external world.2 But, as Barry Stroud has emphasized since the late 1960s,3 the premisses of transcendental arguments seem to have only psychological status. They, therefore, do not warrant any inferences about the external world. In this manner Stroud forces the proponent of transcendental arguments to answer the following question: 'How can we proceed "deductively", or by necessity, from premisses only about our thinking in certain ways to conclusions about the independent world which we thereby think about?,4 In order to meet this challenge a defender of transcendental arguments has in principle three strategies to choose from. s First, one could opt for idealism.
1 This is basically the way Barry Stroud characterizes transcendental arguments in B. Stroud, 'Kantian Argument, Conceptual Capacities, and Invulnerability', in P. Parrini (ed.), Kant and Contemporary Epistemology (Dordrecht: Kluwer, 1994), 231; see also Stroud, 'The Goal of Transcendental Arguments', in R. Stern (ed.), Transcendental Arguments: Problems and Prospects (Oxford: Oxford University Press, 1999). For a more comprehensive discussion of transcendental arguments compare T. Grundmann, Analytische Transzendentalphilosophie. Eine Kritik (Paderborn: Schiiningh, 1994). 2 Transcendental arguments meeting these conditions are called 'world-directed' by Q. Cassam, 'Self-Directed Transcendental Arguments', in Stern, Transcendental Arguments, 82. He distinguishes them from 'self-directed' transcendental arguments, which allow only inferences from given knowledge to the nature of our cognitive faculties. 3 See Stroud, 'Transcendental Arguments', Journal of Philosophy, 65 (1968). 4 Stroud, 'Kantian Argument', 236. 5 Ibid. 235 ff.
206
Grundmann and Misselhorn
This is, basically, the Kantian strategy. It validates the inference from 'inner' evidence to the external world, by supposing that the external world is somehow dependent on epistemic conditions. However, this strategy compromises the objectivity of our knowledge. The second option is to weaken the conclusion. Instead of arguing that the external world has to be structured in a certain way, one may just strive to establish certain conceptual connections. In his book Skepticism and Naturalism (1985) Strawson defends this kind of transcendental argument. Following Stroud, let us call it a modest transcendental argument. 6 The aim of such an argument, as Strawson describes it, is to establish 'a certain sort of interdependence of conceptual capacities and beliefs: e.g.... in order for self-conscious thought and experience to be possible, we must take it, or believe, that we have knowledge of external physical objects'.7 Accordingly, certain beliefs are invulnerable because they are conceptually necessary conditions of the sceptical challenge. In contrast to Strawson, who no longer believes that transcendental arguments can refute the sceptic, Stroud shows more optimism about the anti-sceptical impact of this argument. 8 Yet, it leaves hard-core realists dissatisfied, since the invulnerability of certain beliefs implies neither their truth nor their justifiability. It is hard to see how this could answer the sceptic. 9 The third possibility is to introduce a stronger premiss. Only if (at least) one premiss functions as a bridge principle that establishes a relation between the 'inner' evidence and the independent world may we deduce a truth about the world. Although Strawson no longer approves this kind of argument, he characterizes it accurately: A philosopher who advances such an argun1ent may begin with a premise which the skeptic does not challenge, viz. the occurrence of self-conscious thought and experience; and then proceed to argue that a necessary condition of the possibility of such experience is, say, knowledge of the existence of external objects ... 10
Stroud is rather pessimistic about the prospects of such an ambitious transcendental argument, 11 and currently the great majority of philosophers is with him. The aim of this chapter is to find out whether, given realism, ambitious transcendental arguments can still be defended. The argument we propose concerns perceptual knowledge about the external world. It is supposed to establish that beliefs about the external world that rely on
Stroud, 'Kantian Argument', 24 2 . P. F. Strawson, Skepticism and Naturalism (London: Methuen, 1985), 21. 8 Stroud, 'Kantian Argument', 24 2 . 9 See also C. Hookway, 'Modest Transcendental Arguments and Sceptical Doubts: A Reply to Stroud', in Stern, Transcendental Arguments, 177 £f. 10 Strawson, Skepticism and Naturalism, 8-9. 11 Stroud, 'Kantian Argument', 241. 6
7
Transcendental Arguments and Realism
207
perception-perceptual beliefs-are justified. As will emerge, there is a bridge principle connecting mind and world. It relies on content-externalism, i.e. the clain1 that the content of our perceptual beliefs depends somehow on the objects in the external world, which these beliefs represent. In this way, the link between our having certain perceptual beliefs and the external world can be established. Yet, how is content-externalism supposed to be justified? The challenge is to show that the sceptic cannot consistently deny the justification of semantic externalism. Our argument for this claim will proceed as follow-so First, we will spell out the structure of our anti-sceptical argument, and discuss the problem of the bridge principle's justification. The second step is to show that the sceptic has to accept modal intuitions as a reliable method for the justification of modal claims, otherwise his position becomes epistemically inconsistent. If this is correct, the sceptic cannot undermine the justification of our perceptual beliefs any more, as will be argued. Thirdly, we want to set out in some more detail how content-externalism is supported by modal intuitions, and how the bridge principle follows fron1 it. In addition, we will discuss some sceptical objections. Finally, we must consider whether our line of argument satisfies the conditions of an ambitious transcendental argument.
I
Let us now turn towards the proposed argument for perceptual knowledge about the external world. It proceeds as following: (I) I am having perceptual beliefs about the external world. (2) Necessarily, perceptual beliefs about the external world
are largely true (bridge principle). (3) Necessarily, my perceptual beliefs about the
premIss
premIss fronl (I) and (2)
external world are largely true. (4) Iff perceptual beliefs about the external world are by and premIss large true, then perception is a reliable epistemic method. premIss (5) Reliable epistemic methods prima facie justify the beliefs supported by them. (6) Necessarily, my perceptual beliefs about the external conclusion world are prima facie justified. Some of these premisses have to be accepted even by the sceptic, since they are justified in a manner that cannot be challenged by him. This is true of premisses (I), (4), and (5). They are justified by two methods that are not called into question by the sceptic. Premiss (I) is justified introspectively, (4) and (5) are analytic. Premiss (4) is analytic because the reliability of a method
208
Grundmann and Misselhorn
is defined by its producing mostly true beliefs. 12 Similarly, (5) is analytic, since the reliability of a method defines justification. The aim of justification just is to get at the truth, or, technically put, justification has to be truth-conducive. As we said, a reliable method of justification produces mostly true beliefs. Therefore, reliability is a defining characteristic of justification. 13 Why are these premisses immune from the sceptical challenge? The sceptic utilizes sceptical hypotheses in order to undermine our methods of justifying our beliefs in light of certain evidence. Sceptical hypotheses offer some causal explanation of our evidence that renders the beliefs supported by the evidence predominantly false. Yet, it is not enough for such an explanation just to say that most of our beliefs could be false. Rather, the sceptic has to provide a scenario that makes this claim plausible. The examples of such scenarios are well known: we could be brains in a vat hooked up to a computer by a mad scientist, or we could be deceived by an evil demon. In both cases, most of our beliefs about the external world would be false, and, for this reason, our justification procedures would be unreliable. Sceptical hypotheses, however, do not apply to analytic truths and introspective knowledge. In contrast to (I), (4), and (5), the justification of (2) is challenged by sceptical hypotheses. The attempt to face this challenge even leads to a dilemma: either (2) is analytic or it is synthetic. Suppose, first, that premiss (2) is analytic, then the justification of this premiss would be unproblematic. However, there are good reasons not to follow this line of reasoning. First, in case of the analyticity of (2), sceptical hypotheses would be logically inconsistent. However, it would be rather surprising if all those who found sceptical hypotheses plausible had not noticed this contradiction. In this respect we agree with McGinn's analysis of the situation: We should not, I think, rush into accepting any philosophical theory that makes scepticism go quickly away, counting it a merit of the theory that it shows the sceptic as a sophist and a dim wit. Rather, we should view with considerable reserve any theory that has such immediately anti-sceptical consequences, for it is liable to have underestimated the problen1atic nature of our true epistemological predicament. And especially so if the theory offers no convincing story about how scepticism could have seemed so convincing in the first place. 14
Moreover, is it not counter-intuitive to think that we are able to learn something about the external world just by considering introspective evidence and its conceptual entailn1ents? 15 12 A. Goldman, 'What is Justified Belief?', in G. S. Pappas (ed.), Justification and Knowledge (Dordrecht: Kluwer, 1979). 13 Goldman, 'What is Justified Belief?'; W. Alston, The Reliability of Sense Perception (Ithaca, NY/London: Cornell University Press, 1993), 3. 14 C. McGinn, Mental Content (Oxford: Oxford University Press, 1989), 114. 15 McGinn, Mental Content, 1 13.
Transcendental Arguments and Realism
209
Yet, there is another more important objection: if (2) were analytic, it would not be compatible with a realist conception of truth, or a fallibilist conception of justification. According to realism truth is a non-epistemic notion; as Michael Williams puts it 'thinking does not make it so' .16 Understood this way, realism entails logical fallibilism, i.e. the claim that there is no logical connection between evidence and truth. 17 In other words, if truth is not an epistenlic notion, our evidence cannot logically guarantee the truth of the beliefs it supports. Therefore, premiss (2), which concerns the relation between evidence and truth, cannot be analytic. This is the first horn of the dilemma. Let us now turn to the second horn, which arises if (2) is synthetic. In this case it can be justified in two ways: empirically or a priori. Yet, both kinds of justification seem to be undermined by the possibility of sceptical hypotheses. If (2) is empirically justified, the argument becomes epistemically circular. An argument is epistemically circular iff the justification of the premisses presupposes the truth of the conclusion. 18 Suppose we try to justify empirically the claim that most of our perceptual beliefs are true. Since empirical justification depends on perception, this attempt could be successful only if perception were a reliable method of justification. The empirical justification of premiss (2) would, therefore, presuppose the truth of the argument's conclusion. But this is just what sceptical hypotheses put into question. Still worse, sceptical hypotheses not only undermine empirical justification; they destroy a priori justification as well. Even if it were possible to justify the reliability of perception a priori, the sceptic could still attack the reliability of a priori justification itself. It is (at least prima facie) possible that all our a priori evidence about necessary truths might be false. (Imagine the evil demon makes us hold a host of things necessarily true which in fact are not. 19 ) For this reason, the prospects of the ambitious transcendental argument do not seem to be very promising. There is no way of justifying the crucial premiss (2) without falling prey to scepticism. If (2) were analytic, its justification would be unproblematic. However, such an understanding does not correspond to our realist conception of the objectivity of knowledge. If (2) were, on the other hand, synthetic, no justification seems possible. Nevertheless, we are going to argue, the situation is not as hopeless as it appears.
16
M. Williams, Unnatural Doubts, 2nd edn. (Princeton, N]: Princeton University Press,
199 6 ),228. 17 See J. Heil, 'Skepticism and Realism', American Philosophical Quarterly, 35 (199 8), 70; McGinn, Mental Content, 107. 18 Alston, The Reliability of Sense Perception, IS. 19 Actually, Descartes considers this case in his First Meditation, where even mathematical beliefs are affected by the sceptical hypothesis.
210
Grundmann and Misselhorn II
Our argument will amount to the following: despite appearances, the sceptic cannot consistently deny the justification of premiss (2), which we will propose. Rather, he himself has to justify his sceptical challenge in the same manner, because he can maintain the metaphysical possibility of sceptical hypotheses only with the help of modal intuitions. Our modal intuitions, however, support premiss (2) rather than the sceptical hypotheses, as will be shown. Let us, thus, first understand why the sceptic has to fall back on Inodal intuitions. Modal intuitions tell us which worlds are metaphysically possible, where metaphysically possible worlds are worlds that could be actual. The crucial question is: why should the sceptic claim that his hypotheses are metaphysically possible? Would their logical consistency not be sufficient to undermine the justification of (2)? The logical consistency of sceptical hypotheses, however, can hardly be denied, if one wants to hold on to a realist conception of truth and a fallibilist conception of justification, as we have argued above. Yet, this is not sufficient for raising the sceptical challenge. The logical possibility of sceptical hypotheses implies only that the evidence cannot logically guarantee the truth of the beliefs in question. This claim alone is not capable of undermining the justification of (2). It is just a conceptual consequence of a fallibilist conception of justification. To put it the other way round: if the logical consistency of sceptical hypotheses were sufficient to undermine justification, fallibilism about justification would be a logically inconsistent position. There could not be any fallible justification. This amounts to a reductio ad absurdum. It is generally accepted that fallibilism is not sufficient to generate scepticism. Here is Stroud as a representative voice on the matter: It is often said that traditional epistemology is generated by nothing more than a misguided 'quest for certainty' ... and that once we abandon such a will-o'the-wisp we will no longer be threatened by skepticism .... But that diagnosis seems wrong to me ... 20
Therefore, the sceptic has to claim more than that sceptical hypotheses are logically possible. In order to undermine the justification of (2) he has to be justified in believing that sceptical hypotheses are metaphysically possible. The justification of metaphysical possibilities, however, has to rely on modal intuitions. Finally, the sceptic cannot attack the justification of (2) in
20 Stroud, 'Understanding Human Knowledge in General', in M. Clay and K. Lehrer (eds.), Knowledge and Skepticism (Boulder, Colo.: Westview, I9 89), 35.
Transcendental Arguments and Realism
211
an epistemically consistent way. The epistemic inconsistency of his position arises because the truth of the sceptical hypotheses with respect to modal intuitions would undermine their own justification. 21 If the sceptic claims that modal intuitions are unreliable, sceptical hypotheses could not be justified. For this reason, the sceptic must grant the reliability of modal intuitions as a method of justification. 22 Yet, as will emerge below, modal intuitions speak in favour of (2) rather than supporting sceptical hypotheses. Once more we agree with Strawson on the diagnosis: 'It is only because the solution is possible that the problem exists. So with all transcendental arguments. '23 III
So far, we have learnt that the sceptic cannot deny the reliability of modal intuitions without undermining his own justification. We will now go on to argue that modal intuitions are appropriate to justify the bridge principle in our argument for perceptual knowledge about the external world. We will establish this claiITI in two steps. First, we will argue that content-externalism can be defended as a metaphysically necessary truth by our modal intuitions. Second, we will demonstrate that the necessary truth of content-externalism entails the truth of our bridge principle; namely that, necessarily, perceptual beliefs about the external world are by and large true. 24 Let us start with some reflections on the epistemology of modality. In our claims about metaphysical possibilities or necessities we rely on modal intuitions. Or, to speak n10re accurately, our evidence for the modal belief that p is possible consists in finding p conceivable; and our evidence for the modal belief that p is necessary consists in finding not-p inconceivable. Whenever we find p conceivable, we thereby represent p as being possible. 25 Since we are realists, we do not believe that metaphysical modality is the same as conceivability. However, we do believe that conceivability is a reliable guide to
21 Thomas Grundmann and Frank Hofmann: '1st der radikale Empirismus epistemisch selbstwiderspriichlich?', in J. MittelstraR (ed.), Die Zukunft des Wissens (Konstanz: Universitatsverlag Konstanz, 1999), 684-91. 22 See also C. Misselhorn, 'The Turn Towards A Priori Justification', in J. L. Falguera et al. (eds.), Analytic Philosophy at the Turn of the Millennium (Santiago de Campostela, 1999). 23 Strawson, Individuals (London: Methuen, 1959), 40. 24 Sometimes content-externalism is understood as a conceptual truth. See e.g. D. Davidson, 'A Coherence Theory of Truth and Knowledge', in E. LePore (ed.), Truth and Interpretation (Oxford: Basil Blackwell, 1986), 319. We do not embrace this view, since it would not be consistent with realism, provided that content-externalism entails the veridicality of our beliefs about the external world. For a metaphysical understanding of externalism see A. Brueckner, 'What an Anti-Individualist Knows A Priori', Analysis, 52 (1992), 116. 25 See S. Yablo, 'Is Conceivability a Guide to Possibility?', Philosophy and Phenomenological Research, 53 (1993),7·
212
Grundmann and Misselhorn
modality. And, as we have seen before, this cannot even be disputed by the sceptic. Now, our main argument relies heavily on the difference between being a logically consistent set of propositions and being a metaphysically possible world. For this reason we should better be able to give a list of additional methodological constraints on conceivability over and above logical consistency. This is to give flesh to the intuitive notion of conceivability. Various suggestions have been made to constrain conceivability.26 Among them two constraints are especially significant for us. First, we can conceive of propositions only if we can imagine at least one fully determinate world that makes this proposition true. As long as sets of logically consistent propositions describe worlds that are in some sense or other indeterminate they do not pass this test. Consider the following example of Stephen Yablo's. Initially, it seems as if we can conceive of the denial of Goldbach's conjecture, since it is still unproved that every even number is obtainable as the sum of two primes. But this appearance proves to be illusory according to the above constraint. In order to imagine a world in which there is a counter example to Goldbach's conjecture I have to imagine all the details of the proof of this counter example; and this seems to be impossible as long as we do not actually possess such a proof. 27 Second, we can conceive of propositions only if we can imagine worlds making them true that are explanatorily coherent. This constraint is not completely unrelated to the first one. Think of it this way: if a world contains properties that are unintelligible, mysterious, or unexplainable within this world, then this world is to a certain extent indeterminate. Explanations within other imaginable worlds need not follow the laws of our actual world, but there must be some laws that do the explanatory work within these worlds. Otherwise, we have no right to think of these worlds as metaphysically possible. 28 Let us apply these epistemological reflections to the question as to whether content-externalism can be justified as necessary truth by means of our modal intuitions. A very prominent version of content-externalism claims that the representational content of n1ental states is determined by their distal causal history.29 Hence, the representational content depends on the environn1ent in which the cognitive system has been brought up. A wellknown conceivability argument for the necessary truth of content-externalism is based on Putnam's and Burge's twin-earth thought-experiments about
26 See S. Kripke, Naming and Necessity (Oxford: Basil Blackwell, 1980); Yablo, 'Is Conceivability a Guide to Possibility?'; C. Peacocke, 'Metaphysical Necessity: Understanding, Truth and Epistemology', Mind, 106 (1994). 27 Yablo, 'Is Conceivability a Guide to Possibility?', 31-2. 28 This is not meant to be an inference to the best explanation. Rather, it is sufficient for the conceivability of a possible world that there be some explanatory nexus or other among its 29 This view is defended by Putnam, Davidson, Burge, etc. characteristics.
Transcendental Arguments and Realism
213
twater-thoughts. 3o
water and Although this type of argument is obviously based on modal intuitions, it is not completely convincing. First of all, it does not establish externalism for mental states in general, but only for a limited class of concepts such as indexicals or natural-kind terms. Secondly, it cannot be ruled out that the modal intuitions in this case are simply generated by our semantic-background theory rather than being evidence for the truth of this theory. We think that there is a different conceivability argument for contentexternalism that fares much better. It seems inconceivable that the representational content of mental states is not determined by their distal causal history, i.e. that content-externalism is wrong. We cannot imagine a world in which representational content about objects in the external world is not determined by external relations between the representations, on the one hand, i.e. the mental states that possess representational content, and the objects represented by them on the other. Furthermore, we cannot imagine a world in which these external relations do not have a causal nature. For consider a world in which this is not true. In this world there is representational content without there being any causal relations whatsoever between representations and represented objects. Let us assume instead that representations and their objects are only similar to each other. This world, though logically consistent, is not a conceivable world, for the sin1ple reason that it is not explanatorily coherent. We do not understand how representational content in this world comes about. It remains a mysterious property. The similarity relation between representations and objects represented by them is neither sufficient nor is it necessary to explain representational content. In Reason, Truth and History Putnam presents a nice analogy that makes it obvious that similarity is not sufficient for representational content. Suppose you see an ant creeping on the sand. The line it involuntarily draws looks like a caricature of Winston Churchill. It happens to be similar to Churchill, but nobody would ever claim that this line represents Churchil1. 31 Nor is similarity a necessary condition of representational content. There is no doubt that our ideas and thoughts represent external objects. However, to assume that these ideas and thoughts are similar to their objects seems absurd. They are certainly not little images in our brain. The mental state referring to the property of blueness is probably some neural state that is anything but blue. Hence, similarity cannot explain representational content. But can we not say the following: even if similarity does not give us the required explanation, we can imagine a world in which representational
30 H. 'Putnam, 'The Meaning of "Meaning"', in Putnam, Mind, Language and Reality (Cambridge, Mass.: Harvard University Press, I975); T. Burge, 'Other Bodies', in A. Woodfield (ed.), Thought and Object (Oxford: Clarendon Press, I982). 31 Putnam, Reason, Truth and History (Cambridge, Mass.: Harvard University Press, I98I), ch. 1.
214
Grundmann and Misselhorn
content is explained by some non-causal property-although we have no idea which property it is? Now, this may well be a logically consistent description of a world, but it does not specify an imaginable world, since the description violates the constraint of determinateness. Remember that we can imagine only worlds that are completely determinate. So it proves to be true that we cannot conceive of representational content that is completely independent of any causal relations whatsoever bevNeen representations and their objects. Of course, this dependence relation needs some qualification. The connection between the representational content of mental states and their causes must not become too tight. If an actual causal relation were required for a representation to be about an object, no case of misrepresentation could occur. However, the possibility of misrepresentation seems to be the mark of representational content. Hence, externalism is only acceptable if it makes error possible. For misrepresentation to be possible, the content of tokens of a certain type of representation must be robustly determined by a causal correlation with distal events of a certain type during the learning period. It is only the historical relation to distal causes that can explain representational content about the external world. We can conclude that the necessary truth of content-externalism is justified by modal intuitions. Since perceptual beliefs belong to the class of mental states with representational content about the world, we can also conclude that the content of perceptual beliefs is determined by their distal causal history. We are now ready to take the second step in our argument for the bridge principle. Assume that content-externalism has been established for perceptual beliefs by n1eans of modal intuitions. Does content-externalism imply that our perceptual beliefs about the world are by and large true? If it is true, the content of our current perceptual beliefs is determined by the distal causes of their predecessors. Thus, there must have been some robust correspondence between our former perceptual beliefs and their causes. This amounts to saying that our former perceptual beliefs must have been true. So externalism implies a historic connection of our perceptual beliefs with truth. Since, however, it does not make the content of these beliefs dependent on our current environment, massive error n1ay occur in our current perceptual beliefs. Suppose a person has acquired her representations in a normal environment. Now, if we put her brain into a nutrient solution keeping it stimulated by a con1puter, the content of her representations does not change. It continues to be determined by the former norn1al causes, at least for some time. Thus, the vatted brain has mostly false perceptual beliefs about living in a normal environment. Externalism cannot rule out this kind of error. Our perceptual beliefs can be massively flawed at any time. 32 But from this it does not follow
32
See R. Foley, 'Can Metaphysics Solve the Problem of Skepticism?', Philosophical Issues,
2 (1992), 139; Burge, 'Cartesian Error and the Objectivity of Perception', in P. Pettit et at. (eds.), Subject, Thought and Context (Oxford: Clarendon, 1986), 122, 130-1.
Transcendental Arguments and Realism 215 that our perceptual beliefs can be by and large false in the long run. We can massively err at any time only if, in the past, we have had mostly correct beliefs. As soon as massive error extends beyond a limited period of time, the content of mental states changes, and, finally, readapts to the world. This seems to be sufficient to establish the truth-ratio of perceptual beliefs required by the bridge principle. To SUITI up: both steps have been taken successfully. We were able to establish the necessary truth of content-externalism for perceptual beliefs; and we have demonstrated that this implies the necessary veridicality of these beliefs. If we furthermore assume for conceptual reasons that beliefs of a single epistemological kind are justified if they are veridical in general, then we can conclude that we have perceptual knowledge about the external world, provided we introspectively know of our perceptual beliefs. lIence, our argument for perceptual knowledge about the world seen1S to be sound. At this point in the argument the sceptic's patience is wearing thin. 'Sure', he will admit, 'I cannot deny the reliability of modal intuitions in general. But', he will continue, 'there are competing modal intuitions. Maybe there is a modal intuition speaking for content-externalism and the veridicality of perceptual beliefs entailed by it. But there are certainly other modal intuitions n1aking the case against this necessary veridicality of perceptual beliefs. Among them are the sceptical hypotheses and the intuition that semantic internalism is correct. If some sceptical hypothesis is conceivable or some version of content-internalism, as sonle semanticists believe, then we have modal evidence speaking against the necessary truth of contentexternalism. Maybe it beats the evidence in favour of the necessary veridicality of perceptual beliefs. At the least, it neutralizes this evidence.' Or so the sceptic might argue. We cannot respond to the sceptic by simply ignoring his evidence. If we want to hold on to our claim that externalism is justified by modal intuitions, we have to explain away the sceptic's apparent counter-evidence. To begin with, we do not believe that internalism and sceptical hypotheses are independent intuitions against the necessary veridicality of our beliefs. It is only the Cartesian sceptical scenario that makes internalism about representational content plausible. So in fact we have only to deal with the sceptical hypothesis as counter-evidence. Does it really rely on modal intuitions, as the sceptic suggests? Or is it simply a consequence of realism and logical fallibilism, i.e. the idea that our evidence does not logically imply the truth of the beliefs supported by it? The sceptic might argue that his scenario does not only consist in the proposition that the beliefs supported by our evidence are massively mistaken. The sceptical hypothesis is a concrete and vivid illustration of this point, dressed up and equipped with a lot of details about how the massive error is to be explained. So why shouldn't the sceptic claim that the sceptical hypothesis is conceivable as part of an imaginable world-simply because it is highly determinate? Sure, the sceptic
Grundmann and Misselhorn pretends to give us an example of an imaginable world with his scenario. But it isn't. Something essential is missing. After all, the sceptical hypothesis does not describe a world that is imaginable for us, since it is explanatorily incoherent and, hence, not completely determinate. The sceptical hypothesis does not explain the representational content of beliefs that it presupposes. So we succeed in explaining away the sceptic's apparent modal counter-evidence. His intuition is not a case of proper modal intuitions, but a consequence of a purely logical truth. Prima facie there seems to be another objection against our transcendental argument concerning the relationship between content-externalism and selfknowledge. The sceptic might argue as follows: if content-externalism were correct, the content of our beliefs would be determined by the environment. Therefore, it seems as if I would need to find out perceptually whether there exists an external world in the ordinary sense or one of the sceptical scenarios in order to justify that my perceptual beliefs are, indeed, about the external world in the ordinary sense. How else could we justify premiss (I) given content-externalism? But thereby we are presupposing that perceptual beliefs about the external world are justified, i.e. we must presuppose the truth of the conclusion we wanted to establish. Yet, this objection is based upon the assumption that content-externalism requires a perceptual theory of self-knowledge-a claim that has been discarded by major proponents of externalism, who argued that externalism is compatible with privileged and non-empirical access to mental content. 33 Nevertheless, the sceptic could insist that, for the externalist, self-knowledge does not come for free as within the Cartesian framework. The problem of scepticism seems to recur at one level up. Yet, we can answer this problem straightforwardly. As soon as the sceptic accepts the justification of externalism on the basis of modal intuitions, he can no longer question the justification of premiss (I). Externalism implies the justification of introspective beliefs, since the same external factors determining the content of the firstorder beliefs also determine the content of the introspective beliefs about them. The reliability of introspective beliefs follows directly from externalism. Therefore, a justification of premiss (I) is available. For the sceptic there remains the desperate move to deny the existence of beliefs in general. But then he would fall prey to a performative contradiction. Provided the sceptic claims anything at all, he cannot consistently deny the existence of content in generaL This would amount to committing cognitive suicide. Apart fronl these epistemological objections, the sceptic might also put forward a nletaphysical challenge. In our argument we pretend to accept 216
33 Compare e.g. Davidson, 'Knowing One's Own Mind', Proceedings and Addresses of the American Philosophical Association, 60 (1987); Burge, 'Individualism and Self-Knowledge', Journal of Philosophy, 85 (19 88 ).
Transcendental Arguments and Realism
217
realism and, at the same time, rely on a bridge principle that our having perceptual beliefs implies certain external facts in the world; namely, the existence of facts that make them mostly true. To the sceptic the latter sounds like a version of idealism, since, according to it, external facts seem to depend on the mind. But this is understanding it the wrong way.34 As we have just seen, the truth of the bridge principle is implied by content-externalism. This doctrine explains the metaphysical connection between beliefs and truth by making beliefs dependent on the world and not the other way round, as idealism would have it. Externalism, therefore, does not imply the idealistic view that the external world would not exist without any psychological reality. As it turns out, there is no real tension between our bridge principle and realism. This brings out the weakness of the sceptic's position: in order to challenge the justification of our perceptual beliefs about the world, he needs more than purely logical evidence. He must rely on modal intuitions. But there are no modal intuitions supporting his argument. Properly understood his sceptical intuitions are nothing but logical insights in disguise. In this sense they can be explained away. Modal intuition reveals the metaphysical truth-connection of our perceptual beliefs; and this is perfectly compatible with a realist understanding of truth.
IV Does this mean that our argument for perceptual knowledge succeeds as a transcendental argument of the ambitious kind? To answer this question let us recall the four conditions for ambitious transcendental arguments. Such an argument must rely on premisses that are (i) justified independently of experience and (ii) resistant to the sceptical challenge. (iii) Such an argument must proceed by establishing necessary conditions of thought or experience. Finally, (iv) its conclusion must entail truths about the external world. It is easy to see that our argument satisfies the conditions (iii) and (iv). Its bridge principle (2) claims something about necessary conditions of beliefs; and its conclusion entails that our perceptual beliefs are reliable, i.e. something about the beliefs' relation to the world. But what about the first condition? Are the pren1isses justified without relying on experience? If we assume that both introspective knowledge and conceptual knowledge are non-experiential, then only the justification of our bridge principle (2) seems problematic. Are modal intuitions a priori rather than empirical? This seems to be a matter of controversy. In the post-Kripkean era of modal theory many philosophers 34 For a similar response see Brueckner, 'Modest Transcendental Arguments', Philosophical Perspectives, IO (I99 6), 273-
218
Grundmann and Misselhorn
calling themselves 'actualists' think that modalities depend in some way or other on the actual world. If this is true, the right way of finding out modal truths is to gain a posteriori knowledge of the actual world. 35 Kripke's example of water and H 2 0 may illustrate this point. If by means of empirical investigation we find out that water and H 2 0 are identical in the actual world, then we are justified in believing that they are necessarily identical. This is an a posteriori necessary truth. Now, if the actualist is right in claiming that modal intuitions rely on empirical knowledge of the actual world, then our bridge principle cannot be justified non-empirically. This would also have the effect of making our argument epistemically circular. The justification of the bridge principle would presuppose the truth of the argument's conclusion; namely, that the perceptual beliefs supporting our modal intuitions are justified. However, the argument would not lose its anti-sceptical force, since the sceptic still could not deny the reliability of modal intuitions. On the other hand, there are philosophers who defend the view that at least some modal intuitions are purely a priori. 36 We will remain neutral on this issue here. But we have to admit that the transcendental status of our argument decisively depends on this question. We also hesitate to claim that our argument's premisses are completely resistant to sceptical challenges, i.e. that the argument satisfies condition (ii). Most clearly, this is a potential problem for our defence of the bridge principle. We showed neither that the sceptic cannot consistently deny this principle nor that the sceptic cannot justifiably deny it. What we did establish is that the sceptic cannot justifiably deny the adequacy of modal intuitions as a method of justification. This permits us to clain1 that the sceptic cannot justifiably deny the bridge principle unless he has proper modal intuitions speaking against it, and, as we argued, the sceptic has not provided convincing modal evidence for his claim, so far. However, we cannot rule out the case that he might come up with evidence against the principle some day. For the above two reasons one might say that it would be premature to claim that our argument can be classified as an ambitious transcendental argument. However this may be, it still seems fair to say that our argument is an ambitious anti-sceptical one-a powerful weapon against scepticism about our knowledge of the external world. Any stronger claim is negotiable. 37
35 See e.g. Kripke, Naming and Neccessity; Peacocke, 'Metaphysical Necessity'; K. Fine, 'Essence and Modality', Philosophical Perspectives, 8 (1994). 36 E.g. L. Bonjour, In Defense of Pure Reason (Cambridge: Cambridge University Press 1998); Misselhorn, 'The Turn Towards A Priori Justification'. 37 For helpful comments and discussion we are especially grateful to Graham Bird, Hanjo Glock, Frank Hofmann, Jurgen Pafel, Silvio Pinto, Robert Stern, Karsten Stuber, and Kenneth Westphal.
On
Strawson~s Naturalistic
Turn
ROBERT STERN
I. STRAWSON, KANT, AND HUME
It is customary when considering Strawson's philosophical outlook to categorize him as a Kantian, and understandably so: Strawson's The Bounds of Sense is one of the most influential studies of Kant's first Critique; his use of transcendental arguments in that work and in his earlier book Individuals is clearly inspired by Kant; and Strawson's opposition to certain kinds of empiricism and anti-philosophical scientism (the former embodied for him by Ayer and the latter by Quine) appears to reflect Kant's own hopes for his transcendental philosophy, as offering a 'third way' between these and a rationalistic traditional metaphysics. Above all, perhaps, it has been assumed that it is in relation to the problems posed by scepticism that Strawson is close to Kant, in attempting to undercut the sceptical position by showing that it rests on a questionable underlying assumption (for example, that inner sense is possible without outer sense, or that first-person experience is possible without experience of others), and that once the fallaciousness of this assumption is exposed, the sceptical position can be rebutted. However, it is a notable feature of Strawson's work that, while the initial presentation of his views and arguments appears to have this Kantian flavour, in subsequent writings this has been played down, and Strawson himself has cast those views and arguments in a rather different light, where it is Hume and not Kant who is acknowledged as his central inspiration. This shift is particularly noticeable in relation to Strawson's response to scepticism, where a Humean dimension has been written into that response, which was not apparent the first time round. This Humean dimension is characterized by Strawson as naturalism, because it is based on Hume's appeal to the powerlessness of sceptical reasoning in the face of nature, which makes it impossible for us genuinely to give up the beliefs or
Stern
220 methods the sceptic
questions. 1
As Strawson himself puts it:
According to Hun1e the naturalist, skeptical doubts are not to be met by argument. They are simply to be neglected (except, perhaps, in so far as they supply a harmless amusement, a mild diversion to the intellect). They are to be neglected because they are idle; powerless against the force of nature, of our naturally implanted disposition to belief. 2
It is this Humean naturalism, Strawson now insists, that forms the overarching context for his response to scepticism, under which its more Kantian aspects (such as the use of transcendental arguments)3 should be subordinated and adapted. Two central cases in which Strawson has tried to give this Humean naturalism a role when at first sight it did not appear to be operative in his anti-sceptical argument is in relation to the problem of induction and to the problem of the external world. Thus, in the first case Strawson followed his discussion of the problem of induction at the end of his Introduction to Logical Theory (1952), where Hume might be assumed to be a target rather than an ally, with an appeal to Humean naturalism in his response to Wesley Salmon in his slightly later paper 'On Justifying Induction' (1958) and in his treatment of this issue in Skepticism and Naturalism (1985). And in the second case he supplemented his seemingly Kantian attempt to refute external-world scepticisn1 in The Bounds of Sense (1966) with the more Humean response in Skepticism and Naturalism, in a way that casts the arguments from the earlier work in a new light. Now, it is a matter of Strawson's intellectual biography whether his later writings on these matters represent a change of n1ind on his part (a shift of allegiance from I(ant to Hume), or simply a clarification of his intentions (which always were to bring together Kantian and Humean considerations in some way). But, even if the latter is true, I think we can still speak here of a naturalistic turn on Strawson's part, in so far as Strawson's earlier arguments are in fact (at best) independent of his naturalistic reworking of them, or (at worst) at odds with it, so that his first presentation of these arguments
1 Cf. David Hume, A Treatise of Human Nature, L. A. Se1by-Bigge (ed.), 2nd edn., rev. P. H. Nidditch (Oxford: Oxford University Press, 1978), 187: 'Thus the sceptic still continues to reason and believe, even tho' he asserts, that he cannot defend his reason by reason; and by the same rule he must assent to the principle concerning the existence of body, tho' he cannot pretend to any arguments of philosophy to maintain its veracity. Nature has not left this to his choice, and has doubtless esteem'd it an affair of too great importance to be trusted to our uncertain reasonings and speculations. We may well ask, What causes induce us to believe in the existence of body? but 'tis vain to ask, Whether there be body or not? That is a point, which we must take for granted in all our reasonings.' 2 P. F. Strawson, Skepticism and Naturalism: Some Varieties (London: Methuen, 1985), 13. 3 Cf. ibid. 21-3.
On Strawson's Naturalistic Turn
221
and his subsequent recasting of them in a naturalistic guise are in clear tension (whether Strawson only took up naturalism after he had first presented these arguments, or whether he held to it all along). I will spell out this tension in the next section, and try to show that Strawson's initial arguments against inductive scepticism and the external-world sceptic do not fit at all easily into the framework of Humean naturalism, where Strawson subsequently tries to locate them. This then gives rise to a second question, which I will consider in the last two sections of the chapter: namely, even if it is accepted that Strawson's arguments can be understood without casting them in a naturalistic light, would it nonetheless be better to see them that way, as this reworking is required in order to make them philosophically effective? I will argue that this is not so: not only is Humean naturalism flawed as an anti-sceptical strategy, but also Strawson's initial arguments can be satisfactorily defended on their own terms, without bringing in this naturalism at all. I will therefore conclude that Strawson's naturalism is wrong-headed, and that many of his anti-sceptical arguments can be more profitably exploited without it.
II. STRAWSON'S NATURALISTIC TURN
In this section my aim is to argue that we should see Strawson's later attempts to construe his anti-sceptical arguments naturalistically as a 'turn', in so far as no such con1mitn1ent plays a role in those arguments as originally presented (whether or not such a commitment was present in Strawson's thinking all along). As I have said, I will here focus on two cases: namely, Strawson's response to the inductive sceptic in his Introduction to Logical Theory and his response to the external-world sceptic in The Bounds of Sense. The problem the inductive sceptic poses, I take it, is one of epistemic circularity: namely, how can we claim that we have justified beliefs on the grounds that they are inductively formed, when we have nothing but inductive evidence to tell us that induction is a reliable belief-forming method? At the heart of the problem of induction, therefore, is a reliabilist account of justification, whereby a belief-forming method can only give a believer a justified belief if it is reliable ('externalist reliabilism') or the believer has independent grounds for believing it is reliable ('internalist reliabilism'). Now, as I understand Strawson's response to the sceptic in his Introduction to Logical Theory,4 he simply rejects any such reliabilist account of justification, and so undercuts the ground on which this inductive scepticism is
4
See P. F. Strawson, Introduction to Logical Theory (London: Methuen, 1952), 248-63.
222
Stern
built: that is, he argues that forming a belief using a method like induction just is to form a justified belief, whether or not the method actually is reliable (contra externalist reliabilism) and whether or not one has independent reason to believe it is (contra internalist reliabilism). The mistake the reliabilist makes is to make his conception too consequentialist, and insufficiently deontological: that is, he takes it that inductive grounds only justify a belief if believing in that way leads to certain epistemic results, whereas on Strawson's view inductive grounds justify a belief simply as such, regardless of how effective induction is in the search for truth. Now, this seems to me a legitimate way of understanding Strawson's position in his Introduction to Logical Theory, and in certain respects it is a standard view of that position. 5 It also seems to me that it can be differentiated from Strawson's later presentation of his position in a way that means it does not require the naturalistic commitments of the latter. As I have outlined it, Strawson's original argument provides a fairly direct answer to the inductive sceptic, by switching from a reliabilist account of justification (which is prey to the sceptic's circularity objection) to a nonreliabilist account (which is not). However, in his later treatments of this issue naturalism comes in, as this kind of direct answer is dropped in favour of a different approach. In 'On Justifying Induction' Strawson claims that Hume not only posed the problem of induction, but he also answered it, by showing that we have no choice but to reason inductively, as 'it is forced on us by Nature'.6 This contrasts with the approach taken in An Introduction to Logical Theory. There it was claimed we can resolve the sceptical issue
5 Two respects in which my presentation of Strawson's position differs from that put forward by others is in downplaying the 'ordinary-language' aspect (this is what we mean by justified belief) and the 'doxastic-practice' aspect (induction forms a fundamental part of our doxastic practice). On my reading, Strawson's position is stronger than either of these two approaches: inductive evidence is justification-conferring because inductively formed beliefs simply are rational (not because we call them rational, or because we take them to be so as part of our practice). Cf. John L. Pollock, Contemporary Theories ofKnowledge (London: Hutchinson, 1986), 22, where Pollock labels the position I am attributing to Strawson here 'internalism' and emphasizes what he calls its cognitive essentialism: 'The idea behind internalism is that the justifiedness of a belief is determined by whether it was arrived at or is currently sustained by "correct cognitive processes". The view is that being justified in holding a belief consists of conforming to epistemic norms, where the latter tell you "how to" acquire new beliefs and reject old ones. In other words, epistemic norms describe which cognitive processes are correct and which are incorrect, and being justified consists of "making the right moves". The internalist makes the further assumption that the correctness of an epistemic move (a cognitive process) is an inherent feature of it. For example, it might be claimed that reasoning in accordance with modus ponens is always correct, whereas arriving at beliefs through wishful thinking is always incorrect ... In particular, varying contingent properties of the cognitive processes themselves will not affect whether a belief is justified. This might be called cognitive essentialism. According to cognitive essentialism, the epistemic correctness of a cognitive process is an essential feature of that process and is not affected by contingent facts such as the reliability of the process in the actual world.' 6 P. F. Strawson, 'On Justifying Induction', Philosophical Studies, 7 (1958), 20-1, at 21.
On
Strawson~s Naturalistic
Turn
223
by showing it to be based on a mistaken account of justification, while here it is claimed that we can avoid it by following Hume in holding that reason is subordinate to nature, where this means there is no longer any requirement on us to give a justification for induction in general terms. We are therefore here committed to Humean naturalism as part of our anti-sceptical strategy, whereas previously we were not. Then, in Skepticism and Naturalism, Strawson takes an equally naturalistic approach, albeit one that is directed slightly differently: for, now what he claims to be inescapable is not inductive reasoning as such, but the belief that induction is reliable, 7 so that the latter is not 'open to serious doubt'. 8 Here, then, Strawson appears to pe reverting to a reliabilist account of justification, but where the circularity objection is blocked using the naturalistic argument that because we must take it for granted that induction is reliable no argument for this belief is required. In so far as his earlier position on this issue can be distinguished from his later ones in this way, it appears that Strawson's switch from the former to the latter gives naturalism a role which it did not play before. Turning now to Strawson's argument against the external-world sceptic, a similar pattern emerges. Thus, in The Bounds of Sense we are presented with a transcendental argument modelled on Kant's Transcendental Deduction and Refutation of Idealism, in which, again, naturalistic considerations appear to play no part. In outline, this argument appears to take the following form: (I) Being self-conscious is a matter of being able to ascribe diverse experiences to oneself, whilst being conscious of the unity of that to which they are ascribed. (2) To be in a position to think of experiences as one's own, one must be able to think of them as experiences. (3) For experiences to be such as to provide room for thought of experience itself, they must provide room for a distinction between 'This is how things are' and 'This is how things are experienced as being'. (4) Only experience of objects that are subject-independent could provide room for this is/seems distinction.
Therefore: (5) Subject-independent objects exist.
Strawson summarizes his argument as follows: we Inust perceive some objects as enduring objects, even if our perceptions of them do not endure, n1ust see them as falling under concepts of persisting objects,
7
Cf. Strawson, Skepticisrn and Naturalism, 18-19.
8
Ibid. 19.
224
Stern
even though objects of non-persistent perceptions. The idea of a subjective experiential route through an objective world depends on the idea of the identity of that world through and in spite of the changes in our experience; and this idea in turn depends on our perceiving objects as having a permanence independent of our perceptions of them, and hence being able to identify objects as numerically the same in different perceptual situations. 9
Strawson's anti-sceptical strategy in The Bounds of Sense therefore appears to take a fairly standard Kantian form, of attempting to show that the existence of the external world is a necessary condition of the kind of subjective experience the sceptic takes as his starting-point, where none of this appears related to anything naturalistic. 1o However, when Strawson comes back to the problem of the external world in Skepticism and Naturalism, he again eschews this kind of direct approach, commenting as follows: To attempt to confront the professional skeptical doubt with arguments in support of these beliefs, with rational justifications, is sin1ply to show a total misunderstanding of the role they actually play in our belief-systems. The correct way with the professional skeptical doubt is not to attempt to rebut it with argument, but to point out it is idle, unreal, a pretense; and then the rebutting arguments will appear as equally idle; the reasons produced in those arguments to justify induction or belief in the existence of body are not, and do not become, our reasons for these beliefs; there is no such thing as the reasons for which we hold these beliefs. We simply cannot help accepting them as defining the areas within which the questions come up of what beliefs we should rationally hold on such-and-such a matter. 11
Here, again, we have evidence of a naturalistic turn: that is, whereas previously Strawson's anti-sceptical argument could be understood without any commitment to Humean naturalism, here we are told that this is essential, as a new way of handling the sceptic is introduced.
III. PROBLEMS WITH TAKING THE NATURALISTIC TURN
We have seen, then, that there has been a change in the stance Strawson has taken vis-a-vis scepticism, with a shift from a non-naturalistic to a naturalistic 9 P. F. Strawson, The Bounds of Sense: An Essay on Kant's 'Critique of Pure Reason' (London: Methuen, 1966), 125-6. 10 Cf. Karl Ameriks, 'Recent Work on Kant's Theoretical Philosophy', American Philosophical Quarterly, 19 (1982), 1-24, at II: 'something must be said about the distinctive idea which, since at least [Strawson's] The Bounds of Sense, has widely been assumed to define what Kant was trying to do. This idea is that the transcendental deduction is to be read as a direct response to Humean skepticism and that, very roughly speaking, starting from a weak premise of something like the fact that we are conscious beings, Kant's main aim is to establish that there is an objective realm (what Strawson calls "the objectivity thesis").' 11 Strawson, Skepticism and Naturalism, 19- 20 .
On Strawson!'s Naturalistic Turn
225
approach. However, although this may show that Strawson's arguments as originally presented actually involved no commitment to a naturalistic strategy, it still may be that such a commitment is required if they are to be defended against certain obvious objections, and so be made to work. It is certainly the case that when in his subsequent presentations of his position Strawson has turned towards naturalism, this has invariably been as a response to some critique of his earlier argument. Thus, on the problem of induction Strawson was responding to a critique of his position by Wesley Salmon, who argued from a reliabilist perspective that no answer to the inductive sceptic could be sufficient that did not show that our inductive methods are truth-conducive. 12 And on the problem of the external world Strawson was responding to Stroud's well-known attack on transcendental arguments in general and Strawson's use of them in particular, where Stroud suggested that they invariably involve an overambitious modal premiss, where it is claimed that something non-psychological is required to make experience or thought or language possible, but where in fact the sceptic can 'very plausibly insist' that the world must just appear to us to be a certain way, or that we must just believe it to be a certain way to make experience, etc. possible, without the truth of anything non-subjective having been established thereby.!3 This objection of Stroud's has been widely taken up by others; for example, T. E. Wilkerson has argued against Strawson's objectivity argull1ent as follows: On the one hand I must regard my experiences as mine, and on the other I must regard at least some of them as perceptions of external things. But consider the word 'regard'. It may be true that I must apply concepts of objects, that I must regard some of my experiences as perceptions of external things, that I must believe in the existence of tables and chairs. But it does not follow that I successfully or correctly apply such concepts, that I am right to regard some experiences as perceptions of external things, that my belief in the existence of the external world is true. Clearly there is a difference between applying a concept and successfully applying it. [Strawson's argument] shows at most that we must suppose that there is an external world, not that there is an external world. 14
12 See Wesley C. Salmon, 'Should We Attempt to Justify Induction?', Philosophical Studies, 8 (1957), 33-44, repro in Herbert Feigl, Wilfrid Sellars, and Keith Lehrer (eds.), New Readings in Philosophical Analysis (New York: Appleton-Century-Crofts, 1972). References will be to this reprinted version. 13 Barry Stroud, 'Transcendental Arguments', Journal of Philosophy, 65 (1968), 241-56, at 255. 14 T. E. Wilkerson, Kant's Critique of Pure Reason (Oxford: Oxford University Press, 1976), 57. Cf. also Ralph C. S. Walker, Kant (London: Routledge & Kegan Paul, 1978), 117. Comparable criticisms have been made against variants on Strawson's argument: see e.g. Anthony Brueckner, 'Another Failed Transcendental Argument', NOMs, 23 (1989), 525-30, esp. 528, where he is replying to Morris Lipson, 'Objective Experience', Nous, 21 (1987),319-43.
226
Stern
Thus, in view of these objections to Strawson's position, it might be said that even if his arguments can be construed in a non-naturalistic way they are nonetheless left extremely vulnerable, and that Strawson's naturalistic turn is required in order to salvage them. Now, for this way of motivating Strawson's naturalistic turn to carry weight, we must be convinced that this appeal to naturalism can provide support for an argument that might otherwise be problematic. There are, however, good reasons to think naturalism is unsatisfactory in its own terms, so it is unlikely to provide any support that might be required. 1s Let me outline those reasons, which will vary depending on how the naturalistic position is developed and understood. On one understanding of Strawson's position it could be taken purely pragmatically, as showing that the question of justification that the sceptic raises can have no real practical implications in actually getting us to give up induction, or our belief that induction is reliable, or in the existence of the external world, and thus that this can be set aside as an 'unreal' problem. 16 This seems unconvincing, however: for surely how 'real' the sceptical problem is should here be thought of in philosophical not practical terms, and depends on how plausible his arguments are, not on whether those arguments can actually get us to alter our beliefs? On a second understanding, it might be said that Strawson's naturalism is not intended to block the sceptic's question of justification on pragmatic grounds, but rather to answer it, albeit indirectly. Thus, one suggestion, that is perhaps implicit in what is said regarding Hume's appeal to nature as an 'absolute and uncontrollable necessity', 17 is that naturalism provides a response to the sceptic by showing not just that his justificatory questions
15 Strawson's naturalism has been widely criticized in the literature: see Michael Williams, Unnatural Doubts (Princeton, NJ: Princeton University Press, 1996), 10-17; Ernest Sosa, 'Beyond Scepticism, to the Best of Our Knowledge', Mind, 97 (1988),153-88, at 160-3, 'P.F. Strawson's Epistemological Naturalism', in L. E. Hahn (ed.), The Philosophy of P. F. Strawson (Chicago and LaSalle, Ill.: Open Court, 1998), 361-9; J. J. Valberg, The Puzzle of Experience (Oxford: Oxford University Press, 1992), 168-96; and Pranab Kumar Sen, 'On a Gentle Naturalist's Response to Skepticism', in Pranab Kumar Sen and Roop Rekha Verma (eds.), The Philosophy ofP. F. Strawson (New Delhi: Indian Council of Philosophical Research, 1995), 266-304- For a more sympathetic treatment of Strawson's naturalism see A. C. Grayling, The Refutation of Scepticism (London: Duckworth, 1985). 16 Cf. Grayling, The Refutation of Scepticism, 92-3: 'In short, then, sceptical doubt is shown to be idle or pointless because the beliefs the sceptic asks us to justify turn out to be necessary to our thought and talk of the world, and nothing counts as thought and talk unless it is recognisable as such fron1 the standpoint of the thought and talk we enjoy; so that beliefs to which we are comn1itted, and to which essential reference must be made for any explanation or description of experience in general, are simply not negotiable, that is, are not open to doubt. One might extend the legal metaphor suggested by Kant's "deduction" and say that the terminus of the argument is the defeasance of scepticism; one renders the sceptical doubt null and void by a demonstration of the fact that we are bound to hold the beliefs the sceptic asks us to justify.' 17 Hume, Treatise, 183; cf. Strawson, Skepticism and Naturalism, 10-11.
On
Strawson~s Naturalistic
Turn
227
are pointless, but that they fail to bite, because we cannot be criticized in this way for what we cannot control or change. Here the claim would be that what the naturalistic position shows is that we are immune from the kind of considerations of irresponsibility, irrationality, or dogmatism the sceptic raises, because they are shown to violate the principle that 'ought implies can'. In this way it could be argued that, because these beliefs are ones we cannot give up, owing to their centrality in our belief system, the sceptic cannot possibly criticize us for believing them, making his stance wholly inappropriate. This response can be represented as more deeply Humean in outlook than the purely pragmatic version just discussed, in its insistence that the sceptic raises normative issues where they are inapplicable, by taking too much of what we think and do to be within our control, when in fact we are constrained here in the options that are available to us. It is not clear, however, that this kind of involuntarist naturalism is any more satisfactory than the purely pragmatic kind. A central difficulty is how far this approach really succeeds in undercutting the normative issue. Certainly at one level it does: for, if it can be shown that the structure of our belief system makes certain propositions impossible for us to doubt, then we cannot coherently be criticized or blamed for believing then1; to this extent, the principle of 'ought implies can' can be used to do some work. However, this in itself does not block the sceptic from asking a deeper question: namely, whether if we could (per impossibile) feel some doubt regarding this proposition, we would thereby be being more rational in doing so? The general difficulty is this: although the principle of 'ought implies can' means that nobody can be blamed for thinking or doing what they could not have thought or done differently, it does not mean that questions of right and wrong, rationality and irrationality, stop with what they are thereby capable of controlling. For example, it may still be argued without incoherence that people ought morally to be totally impartial in their treatment of others, even whilst accepting that as a matter of human disposition such impartiality is unachievable; likewise, the sceptic could claim that, regarding some beliefs, even while we are not to be blamed for our lack of doubt, this does not show that we should not properly feel such doubt, given the inadequacy of our reasons. Thus, it can be said that action x is morally better than action y, even if people are such that they can never perform x, and so can never be blan1ed for doing what is (despite this) morally wrong in this case; likewise, it can be said that having some doubt that p would be more rational than believing p without any doubt, even if our belief system makes p indubitable, and so we can never be blamed for doing what is (despite this) irrational. The involuntarist response therefore still leaves our belief vulnerable to the sceptic's normative claims. 18
18 I think this helps explain the felt non sequitur in the following presentation of the naturalistic response by Grayling: 'Showing that we must have such a belief [in the continued unperceived
228
Stern
A third approach might be to argue that the naturalistic strategy is more direct than this, in that it tries to forstall the sceptic's question of justification, by showing that this question is in fact senseless or meaningless. 19 There are two difficulties with this approach, however. The first is that it is now unclear to what extent this is a naturalistic strategy in any sense: for it is hard to see why the sceptic should have overstepped the bounds of meaningfulness simply by questioning a belief which nature has made it impossible for us to doubt. The second is that this forn1 of semantic attack on scepticism is extremely ambitious and notoriously hard to bring off: as Michael Williams has put it, '[the sceptic's questions] do not seem defective in point of intelligibility and no one has so far succeeded in showing they are,.20 A fourth and final approach might be to use the naturalistic strategy in a direct way, but to supplement this with an appeal to coherentist considerations, according to which a belief is justified in so far as it forn18 a central part of a wider network of beliefs, so that giving it up would undermine the belief system as a whole (or large parts of it). However, even if the coherentist account of justification being used here can be ll1ade plausible, if this is required to make the naturalistic strategy go through, it is again no longer clear how much work the naturalistic argument itself is doing, as it appears that the real thrust of this response as an answer to scepticism comes from its coherentism rather than its naturalism. It appears, therefore, that Strawson was misguided, if he thought that in taking his naturalistic turn he would avoid the objections urged against him by critics of his earlier arguments; for it appears that naturalism raises as many difficulties as it solves. Finally, therefore, we need to return to those argulnents, to see if they can be made to work, while admitting that Strawson's naturalistic turn cannot be used to support them.
existence of objects] as a condition of experience is not the same as proving that such objects exist. One is stating what we must believe, not how things are; but since the sceptic wishes us to justify the belief, doing so-the argument goes-is enough to put an end to scepticisn1' (A. C. Grayling, 'Transcendental Arguments', in Jonathan Dancy and Ernest Sosa (eds.), A Companion to Epistemology (Oxford: Blackwell, 1992), 506-9, at 508). 19 Strawson himself at one point signals that this is how his position should be understood: cf. 'Reply to Ernest Sosa', in Hahn (ed.), The Philosophy of P. F. Strawson, 370-2, at 371: 'It is not a matter of dismissing the demand for a justification of one's belief in a proposition on the ground that one can't help believing it. That would be weak indeed. The position is, rather, that the demand for justification is really senseless.' 20 Williams, Unnatural Doubts, 15. Confusingly, Strawson at other points seems to acknowledge this: cf. 'Reply to Hilary Putnam', in Hahn (ed.), The Philosophy of P. F. Strawson, 288-92, at 291: 'Putnam's announced position is that the skeptical challenge on the matter [of the existence of the external world] is unintelligible, that it makes no sense; that skeptical doubt on it cannot be coherently entertained. The position is a strong one ... [b]ut perhaps a thought too strong.'
On
Strawson~s Naturalistic
Turn
229
IV. DISPENSING WITH THE NATURALISTIC TURN
Beginning with the problem of induction, it seems to me that Strawson can offer a defence of his original position against Salmon's subsequent critique and those critiques it has inspired, without needing to take any naturalistic stance. As I have presented it, Strawson's original argument was a response to the sceptic who raises the circularity objection: nan1ely, we have no independent grounds for thinking that induction is reliable, so we cannot claim that the inductively supported belief that p is justified. On my reading, . Strawson's response is simply to argue that we can claim that the belief that p is justified without having any such independent grounds, as all that is required for the belief that p to be justified is for it to be supported by inductive evidence, whether or not induction is reliable, or whether or not we have independent grounds for thinking it is. On this account of justification, the circularity objection therefore drops out. Now, it might be said at this point that Strawson's deontological conception of justification is too strong: for, it might be said that Strawson appears to treat inductive evidence as sufficient grounds for justified belief, but this is not so, as it is also a necessary condition for justified belief that the believer have no grounds for taking induction to be an unreliable method. There is certainly considerable intuitive plausibility in this view: thus, to take the parallel case of perception, if s has visual experience as of a house, his belief that a house is in front of hin1 is presumably not justified if he has previously had visual experiences that have frequently turned out to be hallucinations, or if he is aware that he is in unusual viewing conditions, or knows he is surrounded by house replicas, etc. Thus, even if induction, like perception, provides a prima-facie justification for belief, this justification can be defeated by reliabilist considerations of this sort, and so is not unconditional. However, Strawson can accept this point and still claim to have avoided the circularity objection: for this negative reliabilism merely requires us to have no grounds for believing that our belief-forming method is unreliable, where this allows that this method gives us justified beliefs until it is shown to be guilty in this way. Thus, Strawson can accept the claim being made here, while still rejecting what might be called positive internalist reliabilism, which demands that we have independent grounds for believing that the method is reliable, where it is this that leaves room for the circularity objection. But, the reliabilist could argue that criticisms can be made of Strawson's position that are more fundamental than this, and in ways that make it less easy to see how the circularity problem can be avoided using this approach: for (on this view) what is fundamentally wrong with Strawson's position is that it takes it we can answer inductive scepticism without showing that
230 Stern n1ethods like induction are truth-conducive. Common to these criticisms is the claim that Strawson's position is flawed by what I have been calling its deontologism: that is, its assumption that certain kinds of methods are rational in themselves and hence make beliefs justified that are formed in conformity to them, regardless of whether those methods are reliable (just as in ethics the deontologist holds that certain acts are right in themselves, regardless of whether these kinds of actions promote son1e further end like happiness or desire-satisfaction). Two criticisms of this sort are raised against Strawson by Salmon. The first criticism is that Strawson's account of justification is open to the charge of conventionalism: for Strawson's deontologism means he cannot say why one method of belief-formation is rational and one not, as this is simply 'brute', so that if someone picks out a method as being rational that is radically different from ours, we cannot show he is mistaken by appealing to anything extrinsic to our circle of norms, like reliability; hence, Salmon argues, Strawson must concede that he is a conventionalist, who holds that '[w]e choose, quite arbitrarily it would seem, some basic canons of induction; there is no possibility of justifying the choice'.21 Now, it was in response to Salmon's objection here that Strawson made appeal to Humean naturalism in 'On Justifying Induction', where he argues that because (as Hun1e himself put it) 'it is forced upon us by Nature' to employ inductive reasoning at some level, it is nonsense to claim that induction is a convention, for a convention must be something over which we can exercise some choice. As Strawson puts it: 'Suppose I am convinced that there is nothing to choose, as far as Reason goes, between the "basic canons" of induction, and a consistent counter-inductive policy. Is an "arbitrary choice" then really open to us? Is it? (Just try to make it.)'22 It is in response to Salmon's challenge, therefore, that Strawson's new-found or latent Humeanism is put to work. However, it appears to me that Strawson could have responded to Salmon's charge here quite adequately, without making any appeal to Hun1e in this way. For a better response would have been to argue that the charge is based on a non sequitur. According to Salmon, it can be shown that for Strawson it is a matter of 'arbitrary choice' that we reason inductively, because Strawson thinks that we can only appeal to our intuitive conviction that induction is rational to explain why we take inductively grounded beliefs to be justified: but if Strawson holds that our practice of inductive reasoning is based on our intuitive convictions in this way, Salmon surely cannot claim that this makes it a matter of 'arbitrary choice' or 'mere convention', any more than it is a matter of 'arbitrary choice' or
21 22
Salmon, 'Should We Attempt to Justify Induction?', 503-4. Strawson, 'On Justifying Induction', 21.
On Strawson's Naturalistic Turn
23 I
'mere convention' when we do not go in for killing people because we have the moral conviction that murder is wrong (when in both cases the conviction is not based on anything more fundamental or extrinsic). This response may give rise to a second criticism, however, which is also put clearly by Salmon: namely, because Strawson here makes being reasonable simply a matter of using certain methods as such, it is not clear why the sceptic should care about being reasonable; it is therefore difficult to see how (on Strawson's account) we could go about persuading the sceptic to adopt those methods, when they are said to be rational on purely intrinsic grounds, unrelated to the attainment of any further end. This consequentialist criticism of Strawson's deontologism is put clearly by Salmon as follows: If we regard beliefs as reasonable simply because they are arrived at inductively, we still have the problem of showing that reasonable beliefs are valuable. This is the problem of induction stated in new words. If we regard beliefs as reasonable simply because they are arrived at inductively and we hold that reasonable beliefs are valuable for their own sake, it appears that we have elevated inductive method to the place of an intrinsic good ... The resulting justification of induction amounts to this: If you use inductive procedures you can call yourself 'reasonable'-and isn't that nice! 23 We have here a clash of outlooks that is familiar from ethics: whereas the deontologist claims that there is ground enough not to do action a (or to believe proposition p), if doing a (or believing p) violates the moral (or rational) law, the consequentialist claims that this is insufficient, as it leaves it mysterious why obeying the moral law (or the law of reason) really matters. Now, it seems to me that the best way for Strawson to respond to Salmon here is to show that, if he takes this line, it will be hard for him to hang on to his reliabilism, as it is not clear that reliabilism is in any better position than deontologism on this issue. The difficulty is that Salmon assumes that, if he can show the sceptic that induction is truth-conducive, this can provide the sceptic with a reason to adopt this method, in so far as induction is now related to the attainment of an end in a way that previously it was not. But, as Stich has argued,24 the question now arises as to what makes the formation of true beliefs desirable, and whether this has intrinsic or instrumental value. If the reliabilist claims it has intrinsic value, the reliabilist may find it hard to convince the
Salmon, 'Should We Attempt to Justify Induction?', 506. See Stephen Stich, The Fragmentation of Reason (Cambridge, Mass.: MIT Press, 1990), 99-100 (for comments on Salmon) and chs 5 and 6 (more generally). Cf. also David Papineau, ~Normativity and Judgement', Proceedings of the Aristotelian Society, supp. vol. 73 (1999), 17-43. Papineau summarizes his view as follows: 'I shall contend that the most significant norms of judgement can be viewed as prescriptions to the effect that, in order to achieve the truth, you ought to judge in such-and-such ways. In my view, there is nothing constitutively normative about the end of truth itself. So I will take the force of these prescriptions to derive from independent moral or personal reasons for attaching value to truth' (ibid. 18). 23
24
Stern sceptic to care about it in the same way as Strawson found it hard to convince the sceptic to care about having rational beliefs; but, if he allows that it has instrumental value, then it is no longer clear how far he is a reliabilist, for justification would now no longer come from truth-conducivity as such. In view of this, the reliabilist may come to see that there is sonlething misguided in his original criticism of Strawson; for it turns out that the notion of intrinsic value is something on which his account of justification also depends, making it hard to see how he can object to it when it plays a role in Strawson's account. We may now turn to the problem of the external world, where again Strawson introduced a naturalistic element in his response to this problem, but where I would argue it is not required. Here the difficulty arises because Strawson accepts the point made by Stroud, that the modal claim made in transcendental arguments is invariably too strong, but that, once it is weakened, they can merely establish w·hat we must believe or how things must appear to us. In response, Strawson endorses Stroud's point, but nonetheless claims that transcendental arguments can be of value from a naturalistic standpoint, in so far as they establish what we must believe, and so reinforce Humean claims regarding the pointlessness of sceptical doubt on questions like the existence of the external world. Now, one option here would obviously be to refuse to accept the Stroudian point, and to try and make good on the kind of strong conclusion apparently aimed at by Strawson in his earlier objectivity argument. However, I would argue that, even if we accept Stroud's point, and thus accept that transcendental arguments must be given a weaker conclusion, we can do so without being obliged to use them in a naturalistic way. For, even if we allow that transcendental arguments can merely establish what we must believe or how things must appear to us, there are still sceptical issues they can be used to resolve, in a way that does not bring in any naturalistic considerations. This can be shown, I believe, in relation to the problem of the external world, once this is conceived of in justificatory terms: for, the problem the sceptic then presses is how we can be justified in believing in the existence of such a world, when this belief lacks proper inferential support (e.g. either as a causal inference or as an inference to the best explanation). Now here we can use a modest transcendental argument, which shows merely that we have experience as of an external world, to show that no such inference is required, in so far as our belief is then shown to conform to a perceptual (non-inferential) norm, and hence is justified. Here, then, we can go along with Strawson's willingness to accept the Stroudian point, and seek to 'absorb' it, but in a way that once again dispenses with Strawson's problematic neo-Humeanism. 25 232
25 This point is argued at greater length in my Transcendental Arguments and Scepticism: Answering the Question ofJustification (Oxford: Oxford University Press, 2000).
On Strawson's Naturalistic Turn 233 Thus, in relation to the problem of the external world, as previously in relation to the problem of induction, it appears that there is a way of n1aking good on some of Strawson's anti-sceptical arguments, whilst refusing to go along with his naturalistic turn; and if that turn is motivated by a loss of faith in those arguments, I have suggested that this motivation falls away once we see how Strawson's original insights can be preserved without it.
The Evidence of our Senses JOHN HYMAN
••
The modern causal theory of perception-the theory defended by Grice and Strawson-differs from the classical theory advanced by Descartes and Locke in two ways. First, the modern theory is an exercise in conceptual analysis. Secondly, it is a version of what is sometimes called direct realism. I shall comment on these points in turn. First, in Grice's words, the modern causal theory is meant 'to elucidate or characterize the ordinary notion of perceiving a material object'.1 So, on the one hand, its object is not to define a concept better suited to scientific enquiry than the familiar concept that we learn to use as children. On the contrary, its object is to elucidate or characterize that very concept. And, on the other hand, it is not sufficient, in order to count as accepting the causal theory in some form, merely to hold that the perception of a material object can always be causally explained in a way which involves the object perceived. The minimum claim made by the theory is (in Strawson's words this time) as follows: (PI) It is a conceptual truth that when a subject S perceives an external object
0, 0 is causally responsible for S's perceptual experience. 2 Secondly, the modern theory's advocates are committed to the proposition that we perceive material objects immediately. The original, classical theory is generally thought to share a central doctrine with phenomenalism: namely, that the only immediate objects of perception are mental entities-thoughts
1 H. P. Grice, 'The Causal Theory of Perception', Proceedings of the Aristotelian Society, suppl. vol. 35 (1961), 121-2. 2 In fact these are not quite Strawson's words. I have substituted 'perceives' for 'sees' and 'perceptual' for 'visual' P. F. Strawson, 'Reply to Paul Snowdon', in L. E. Hahn (ed.), The Philosophy of P. F. Strawson (Chicago and LaSalle, Ill.: Open Court, 1998), 3 I 1.
Hyman or images or ideas. And, because of this doctrine, the classical theory is thought to imply that the existence of objects beyond the mind cannot be known, except perhaps as a result of making an inference fron1 ideas or images that it is hard to justify. There has been a debate about whether this view of the classical theory is historically accurate. But, whatever the truth is about that, Grice's article 'The Causal Theory of Perception' convinced many philosophers that it is possible to formulate a version of the causal theory which discards this doctrine. Grice argues that, if a person perceives a material object, the object is causally responsible for his having a sense-impression or sense-datum, or (in a rIlore cautious form of words) for 'a state of affairs' that is reported by 'some present-tense sense-datum statement' about him. 3 But, instead of introducing the tern1S 'sense-impression' and 'sense-datum' as names of a kind of object that is present in the mind of a person who is perceiving something, Grice suggests that they be introduced contextually, by introducing (e.g.) the sentence, 'S has a sense-impression of 0' as a paraphrase of some such sentence as 'It seems to S as if he can perceive 0', adding, 'I shall myself ... often for brevity's sake talk of sense-data or sense-impressions; but I shall hope that a more rigorous, if more cumbrous, mode of expression will always be readily available.'4 He considers at length the objection that it is not true to say that it seems to S as if he can perceive 0 unless there is some reason for doubting whether S can perceive 0; and, in a discussion whose ramifications extend far beyond the philosophy of perception, he rejects this objection, on the grounds that it confuses semantic and pragmatic aspects of the use of language. The advantage of introducing the tern1 'sense-datum' or 'sense-impression' contextually is twofold. First, as Grice explains, it means that there is nb need to prove 'the existence of objects of a special sort for which the term ... is offered as a class-name'.5 For the term is not introduced as the name of a special sort of object, but (roughly) as part of an abbreviation. Secondly, if a term is not the name of a special sort of object, a fortiori it is not the name of a mental object that represents or stands proxy for a material object. Hence, the n1inimum claim made by the causal theory can be combined with the proposition that we are immediately aware of material objects, rather than their mental proxies. As Strawson puts it: we take ourselves to be immediately aware of real, enduring physical things in space ... The immediacy which common sense attributes to perceptual awareness
Grice, 'Causal Theory of Perception', 152. Ibid. 12 3-4. I shall follow Grice here, and assume that the free use he makes of the ternlS 'impression' and 'sense-impression' is acceptable. 5 Ibid. 123. 3
4
The Evidence of our Senses
237
is in no way inconsistent ... with the causal dependence of [perceptual experience] on [the things we perceive]. 6
Now since the modern theory is not committed to the doctrine that the immediate objects of perception are n1ental entities, it is generally thought that it escapes the difficulty faced by the classical theory in explaining how the ordinary beliefs we acquire when we perceive objects can be justified. But I think this is a mistake. I think it faces the same difficulty, at least in respect of an important class of these ordinary beliefs, namely perceptual beliefs, such as my present belief that I can see a table in front of me. And I think it faces this difficulty precisely because it combines the two things I have mentioned: conceptual analysis and direct realism. The principal object of this chapter is to explain and defend this claim. In Sections II-VI I shall argue that, if the minimal claim made by the modern causal theory is correct, a perceptual belief (such as my present belief that I can see a table in front of me) cannot be reasonable unless it is supported by evidence. In Sections VII-VIII I shall argue that the modern causal theory cannot provide a satisfactory account of how such a judgement can be supported by evidence. In Section IX I shall argue that the theory is based on a false view about the nature of perceptual experience, and propose an alternative view, which does not appear to lead to the same difficulties. Finally, in Section X, I shall comment briefly on the epistemology of sense perception. I should add that, like Grice and Strawson, I shall talk mainly about visual perception.
II The first question I want to consider is this: (QI) Does PI imply that a perceptual belief cannot be reasonable unless it is supported by evidence?
By a perceptual belief I.shall mean a belief that would be expressed by uttering a sentence whose main verb is a perceptual verb in the first person and the present tense: for exan1ple, 'I see a table in front of me', or 'I can see a table in front of me'. Before I address QI I want to make four preliminary points. First, a terminological point. The term 'perceptual belief' and similar terms have been used in various different ways. For example, Strawson calls his belief that he perceives a table in front of him an 'M-perception-belief', whereas Grice calls his belief that there is a table in front of him a 'perceptual belief'.
6 P. F. Strawson, 'Perception and its Objects', in G. F. Macdonald (ed.), Perception and Identity (London: Macmillan, 1979), 53·
Hyman This belief is supposed to be the result of a perception; but it is not about a perception. The point that I wish to draw attention to is that, in the sense in which I have defined the term, my present belief that there is a table in front of me is not a perceptual belief. My use of the term 'perceptual belief' is like Strawson's use of the term 'M-perception-belief', but it is unlike Grice's use of the term 'perceptual belief'. And my topic is the justification of perceptual beliefs, in this restricted sense. It is not the justification of all of the beliefs that perceptions lead us to acquire. Secondly, a point about the interpretation of PI. When S perceives 0, S's perceptual experience is a perception of O. Hence, PI says that it is a conceptual truth that when S perceives 0, 0 is causally responsible for S's perception of O. Now PI is Strawson's formulation of the modern causal theory's minimum claim. And, as we have seen, Grice holds that it is a conceptual truth that when S perceives 0, 0 is causally responsible for S's impression of O. So, what is the relationship between these two doctrines? The simplest answer is that they are equivalent, because it is a conceptual truth that a perception of 0 is an impression. It is, to be sure, an impression of a particular kind; namely, one that was caused (in the right sort of way) by O. But, since a perception of 0 is an in1pression, when S perceives 0, S's perception of 0 and S's impression are one and the same experience. This simple answer is suggested by several passages in Strawson's writings. For example, in 'Causation in Perception' Strawson remarks that only those impressions that are in a certain sense dependable 'are to count as the perceptions they seem to be'. 7 And this clearly implies that some impressions count as perceptions, i.e. that some impressions are perceptions. (Of course it transpires that an impression is a perception of 0 only if it was caused by 0. 8) I shall discuss the claim that a perception is a kind of impression towards the end of this chapter. As we shall see, there are grave objections to it; and 238
7 Strawson, 'Causation in Perception', in Freedom and Resentment and other Essays (London: Methnen, 1974), 71. I shall discuss this remark and the passage in which it occurs in detail towards the end of this chapter. 8 In his later article 'Perception and its Objects' Strawson expounds the modern causal theory's minimum claim as follows: 'The thought of my fleeting perception as a perception of a continuously and independently existing thing implicitly contains the thought that if the thing had not been there, I should not even have seemed to perceive it. It really should be obvious that with the distinction between independently existing objects and perceptual awareness of objects we already have the general notion of causal dependence of the latter on the former' (51). The first sentence is obviously meant to impress upon the reader that if S believes that he can see 0, S believes implicitly that his impression was caused by 0: had the thing not been there, S would not even have seemed to perceive it. But in the second sentence Strawson insists that if S perceives 0 his perceptual awareness of 0 is causally dependent on o. And by 'perceptual awareness' Strawson presumably means perception. Hence, a plausible interpretation of the passage, which comfortably avoids the charge of non sequitur, will depend on the thought which was expressed explicitly in the earlier passage, that some impressions 'count as the perceptions they seem to be'.
The Evidence of our Senses
239
Strawson has recently attempted to defend the causal theory without assuming that it is true. So, in the meantime, I shall interpret PI as saying only that when S perceives 0, 0 is causally responsible for S's perception of O. This is consistent with holding that when S perceives 0, S's perception is an impression, and hence 0 is causally responsible for S's impression. But it does not imply it. The third point is that neither Grice nor Strawson addresses QI. But they could accept an affirmative answer to QI without discomfort, because they both hold that our impressions provide the evidence that supports our ordin= ary perceptual beliefs, and makes them reasonable. So, for example, my present belief that I can see a table in front of me is reasonable because it seems to me as if I can see a table in front of me. They deny that an ordinary perceptual belief is the result of making an inference, or of what Grice calls a 'passage of thought'.9 And they deny that we postulate material objects, in order to explain the occurrence of our impressions-because we do not postulate what we perceive immediately. But Grice says that in ordinary circumstances S accepts the existence of 0 'on the evidence of certain sense-impressions', and presumably he holds that S believes that he perceives o on the same evidence. lo For his part, Strawson says that an impression 'presumptively makes reasonable' the corresponding perceptual belief, which is presumably the same thing as providing the evidence needed to support it. ll Finally, the doctrine that our impressions provide the evidence that makes our ordinary perceptual beliefs reasonable is consistent both with Grice's denial that perceptual beliefs result from inferences, and with Strawson's vigorous criticism of the doctrine that we postulate objects in order to explain in1pressions, as the following analogy confirms. Suppose I believe that a friend intends to take a holiday in France next year. If I believe this because, and only because, he told me so, then my evidence consists in the fact that he told me so, and my belief depends upon this evidence. But it does not follow that I inferred from his telling me so that he intends to take a holiday in France next year, or that I attributed the intention to him in order to explain his utterance. The doctrine that intentions are states of mind that we postulate in order to explain behaviour is a popular one. But it does not follow simply fron1 the fact that the evidence for a person's state of mind lies in his saying and doing things.
9 The phrase 'passage of thought, or inferential passage' occurs in H. P. Grice, Studies in the Way of Words (Cambridge, Mass.: Harvard University Press, 1989), but I am unable to give a page reference. 10 Grice, 'Causal Theory of Perception', IS0. 11 Strawson, 'Causation in Perception', 68. Strawson also holds that S's having an impression of a 'presumptively makes reasonable' S's belief that a exists. For example, if it sensibly seems to S just as if he were seeing a book on a desk in front of him, the occurrence of this experience presumptively makes reasonable S's belief that there is a book on a desk in front of him. (Both the ternlinology and the example are Strawson's own: ibid. 68.)
Hyman III Now for QI. I believe that PI does imply that a perceptual belief cannot be reasonable unless it is supported by evidence. The argument is as follows. PI implies that a perceptual belief is implicitly a causal belief. Specifically, it implies that a perceptual belief is implicitly about the cause of an experience. By this I do not mean merely that PI implies (for example) that I can see a table in front of me only if a table in front of me is causally responsible for my visual experience, although of course this is true. When I say that PI implies that a perceptual belief is implicitly a causal belief, I mean that it implies that a perceptual belief is a causal belief; specifically, that it is a belief about the cause of an experience. The word 'implicitly' is only there to signal the fact that a cursory look at the sentence used to express a perceptual belief may not disclose this fact. It may not be obvious until the content of the belief, or the meaning of the sentence used to express it, has been explained correctly. A couple of comparisons may help to make this clear. If I believe that P, where P is the conjunction of Euclid's axioms, I do not believe implicitly that Q, where Q is Pythagoras's theorem, although of course P implies Q. But if I believe that Candy spilt her milk, I do believe implicitly that Candy caused her milk to spill, and this will be obvious if the content of my belief, or the meaning of the sentence 'Candy spilt her milk', is explained correctly. If PI is true, my belief that I can see a table in front of me is like my belief that Candy spilt her milk. Neither the word 'cause' nor any other word that has a similar meaning needs to occur in a sentence that is used to express one of these beliefs. But when these sentences are fully understood, or when the content of each belief has been spelled out correctly, it will be obvious that they are both causal beliefs. Strawson claims more than this. He write as follows: this notion of the causal dependence of the experience enjoyed in sense-perception on features of the spatio-temporal world ... is not something we discover with the advance of science, or even by refined common observation. Neither does it require any refined philosophical argument. It is conceptually inherent in a gross and obvious way in the very notion of sense perception as yielding true judgements about an objective spatio-temporal world. 12
I shall be more cautious. Perhaps the concept of causation is inherent in a gross and obvious way in the belief that Candy spilt her milk. But what is obvious to some is unobvious to others, and, if PI is true, my belief that I can see a table in front of me may be a case in point.
12
Analysis and Metaphysics (Oxford: Oxford University Press, 1992), 61.
The Evidence of our Senses
241
Be that as it may, PI implies that a perceptual belief is-in the sense explained-implicitly a causal belief. Conceptual inherence, whether or not it is gross and obvious, is not just a matter of implication. It is not even just a matter of entailment, although we may miss the difference if the \vord 'inherent' reminds us of the idea that the conclusion of a valid argument is 'contained' in the premisses, and if we lazily accept this idea.!3 If PI is true, a perceptual belief just is a causal belief. For example, my belief that I can see a table in front of me just is the belief that a table in front of me is causally responsible (in a special sort of way that is still obscure) for my current visual experience. 14
IV Now, depending on one's views about causation, one may think that for this reason alone it follows from PI that a perceptual belief cannot be reasonable unless it is supported by evidence-because one nlay think that every causal belief needs the support of evidence, in order to be reasonable. Certainly, Hume, Mill, Russell in 1914, and Wittgenstein in 1933 all held that view. But one may hold, on the contrary, that (P2) Causation is directly observable in the particular case;
and (P3) What is directly observed is believed without evidence.
Strawson holds that P3 is true. He writes as follows: there is an enormous variety ... of kinds of action and transaction which are directly observable in the particular case and which are properly to be described 13 The metaphor of containment is in any case a seriously misleading one-except perhaps for syllogistic logic, where there are effective decision procedures for validity. (See P. Geach, God and the Soul (London: Routledge & Kegan Paul, 1969), 79-80.) 14 The parenthetical phrase is there because neither Grice nor Strawson holds that if 0 is causally responsible for S's perceptual experience then S perceives 0, because they disagree on the right way to state necessary and sufficient conditions for the truth of the proposition that S sees 0, and because no consensus has been reached by philosophers who accept the theory on this question. The main point stated in this paragraph, namely that if PI is true, a perceptual belief just is a causal belief, needs to be underlined for the following reason. I shall argue that since PI implies that a perceptual belief is implicitly a causal belief, it also implies that a perceptual belief cannot be reasonable unless it is supported by evidence, because (except in the case where causation is directly observed) causal beliefs need to be supported by evidence in order to be reasonable. But if PI only implied that the proposition that S perceives 0 entails that 0 is causally responsible for S's perception, this argument would be obviously fallacious. For it would be absurd to maintain that if P entails Q, and Q is implausible without evidence, it follows that P is also implausible without evidence. For example, it would be absurd to maintain that if a theorem follows from some axioms and the theorem is implausible unless it is supported by evidence, the same must be true of the axioms.
Hyman
242
as causal in so far as they are varieties of bringing something about, of producing some effect or some new state of affairs. 1s
For example, one can directly observe a boulder flattening a hut, or a man lifting a suitcase on to a rack. As we have seen, Grice and Strawson do not hold that P3 is true. 16 But Kenny, for example, does: A proposition n1ay itself be evident [he writes], or it may be something for which other propositions provide evidence. Propositions which are themselves evident may either be self-evident (as that 2 + 2 = 4) or be evident to the senses (as that it is snowing as I write this) ... Things which are evident may be said to be believed without evidence. Certainly there is no other proposition put forward as evidence for them. 17
Kenny does not say explicitly that his belief that it is snowing is reasonable, but this is certainly his view. In fact the passage is part of a classification of rational beliefs. If the views expressed in both of these passages are correct, PI may be consistent with the view that a perceptual belief can be reasonable without being supported by evidence. Because it may be possible to hold, first, that (for example) the production of my visual experience of a table in front of me is one of the 'enormous variety ... of kinds of action and transaction which are directly observable in the particular case' by me; and hence, secondly, that my belief that a table in front of me is causally responsible for nlY visual experience is not supported by evidence, and does not need to be.
v So the next question is this: (Q2) Can I directly observe the production of my perceptual experience? For example, if there is a table in front of me, and I can see it, can I also see its production of my visual experience? If the answer to Q2 is no, then the answer to QI is yes. Two arguments may be thought to show that the answer to Q2 is no, at least in the case of visual experiences. I shall not consider whether these arguments apply in the case of every other sense. First, plausible examples of directly observable causation, such as the boulder flattening the hut, are unlike the production of a visual experience in two relevant ways. First, flattening-along with lifting, breaking, pulling, pushing, etc.-is one of the many basic sorts of mechanical transaction on which the concept of a cause is founded, whereas the production of a visual
15
17
Strawson, Analysis and Metaphysics, I 15. 16 See above p. 239 and n. I I. A. J. P. Kenny, Faith and Reason (New York: Columbia University Press, 1983), 10.
The Evidence of our Senses
243
experience is not. Secondly, a boulder's flattening a hut does not look like action at a distance, whereas the production of a visual experience presumably would. Hence, it isn't plausible to hold that I can directly observe the production of my visual experience. This argument is not inept, but it is not conclusive, because there may be causal transactions that are directly observable, but that differ from a boulder's flattening a hut in both of these respects. For example, perhaps I can directly observe a sudden noise startle someone, although making someone start is not a mechanical transaction like a push or a pull; and the question of whether the noise and its effect are or appear to be contiguous does not have a straightforward answer, if it has an answer at all. The second argument is that visible objects do not generally look as if they are producing our visual experiences; and so we cannot generally see that they do. Sometimes visible objects look active and sometimes they look inert. It simply isn't plausible to n1aintain that they never look inert because they always look as if they are acting on our senses. There is a reply to this argument, but I do not think it succeeds. The reply is similar to something Wittgenstein is supposed to have said to Elizabeth Anscombe. Apparently, Anscombe remarked that it wasn't surprising that people used to believe the sun goes round the earth, because that is how it looks. And apparentlyWittgenstein asked in reply: 'How would it look if it looked as if the earth goes round the sun?' So perhaps someone who believed that visible objects do look as if they're producing our visual experiences could challenge someone who denied this to say how he thinks visible objects would look if they did look as if they were producing our visual experiences. The remark attributed to Wittgenstein here is on target, but the parallel does not succeed, because the point of the remark is that the sun would look the same either way. But if visible objects would look the same whether they looked as if they were producing our visual experiences or not, then we could not see that they were producing our visual experiences. It follows that the answer to Q2 is negative: at least in the visual case, I cannot observe the production of my perceptual experience. And hence the answer to QI is affirmative: at least in the visual case, PI does imply that a perceptual belief cannot be reasonable unless it is supported by evidence. 18 VI
There is another route to the same conclusion. For, if PI is true, the proposition that S's belief that he can see 0 does not depend on evidence entails a regress. 18 I shall not add the qualification 'at least in the visual case' again. The reader is invited to add it where appropriate.
244 Hyman There are two prelin1inary points that need to be borne in n1ind. First, as we have seen, PI implies that a perceptual belief is implicitly a causal belief. For example, if S believes that he can see 0, he believes implicitly that 0 is causally responsible for his visual experience. For the most part, Grice and Strawson restrict their discussion to the perception of bodies. But-this is the first point-if this is true of beliefs about bodies it must also be true of beliefs about events and actions. For one does not use the verb 'see' equivocally if one says, 'I believe I can see a man; now I can see the expression on his face changing; and now I can see him opening the door and leaving the room.' Secondly, the idea we are considering is that S's belief that 0 is causally responsible for his visual experience does not need to be supported by evidence in order to be reasonable, because the production of S's experience is one of the 'enormous variety ... of kinds of action and transaction which are directly observable in the particular case' by S. But what exactly is involved in the thought that S observes the production of his experience? Clearly, n10re must be involved in this thought than that S perceives the production of his experience. If S perceives the production of his experience, but does not realize that this is what he is perceiving, then we should not be willing to say that S's belief that his experience was produced by 0 does not need the support of any evidence in order to be reasonable because he observed the production of his experience. 19 Compare the following case. Suppose I believe that Candy spilt her milk at breakfast this morning. My belief may be reasonable, without being supported by any evidence, if spilling milk is one of the 'enormous variety ... of kinds of action and transaction ... etc.', and if I observed the act. But not if I saw her spill it but failed to realize that this is what she was spilling. If I saw Candy spill her milk but did not believe that it was milk I saw her spill (e.g. because I thought that it was paint), the fact that I saw it happen cannot make my belief a reasonable one. So, if S's belief that 0 is responsible for his experience does not need to be supported by evidence in order to be reasonable because S observes the production of his experience, then observation (in the relevant sense) hasto imply belief: i.e. if S observes the production of his experience then S believes that he perceives the production of his experience. That is the second point. If these two points are accepted, the regress can be simply den10nstrated. Thus, suppose S believes that he can see O. If PI is true, it follows that S believes implicitly that 0 is causally responsible for his visual experience. Let us call this experience E.
19
Cf. Grice, 'Causal Theory of Perception', 147.
The Evidence of our Senses
°
245
Suppose now that S's belief that is responsible for E does not depend on evidence, because S observes the production of E. It follows (given the second point above) that S believes that he perceives the production of E. And, if PI is true, it also follows (given the first point above) that S believes implicitly that the production of E is causally responsible for his perception of the production of E. Suppose now that this belief does not depend on evidence, because S observes the production of his perception of the production of E. It follows (given the second point above) that S believes that he perceives the production of his perception of the production of E. And, if PI is true, it also follows (given the first point above) that S believes that the production of his perception of the production of E is causally responsible for his perception of the production of his perception of the production of E. Obviously, the same step can be repeated indefinitely. If it is conceded at any point that S's belief does depend on evidence, then it must be conceded that his original belief, i.e. his belief that he can see 0, depends on the same evidence. But the result of repeatedly refusing to make the concession is absurd. For it could not be seriously maintained that, in ordinary circumstances, when S believes that he can see 0, he also believes that the production of his perception of the production of his perception ... (ad infinitum) of E is causally responsible for his perception of the production of his perception ... (ad infinitum) of E. Indeed, it could not be seriously maintained that, in ordinary circumstances, when S does see 0, and is aware of doing so, the production of his perception of the production of his perception ... (ad infinitum) of E is causally responsible for his perception of the production of his perception ... (ad infinitum) of E, regardless of what S believes about the matter.
VII For the reasons set out in the last three sections the answer to QI is affirmative. Hence, Grice and Strawson are bound to hold that a perceptual belief cannot be reasonable unless it is supported by evidence. And, as we have seen, they do hold this. They hold that ordinary perceptual beliefs are supported by the evidence provided by in1pressions. For example, n1Y present belief that I can see a table in front of me is reasonable because it seems to me as if I can see a table in front of me. So the next question is this: (Q3) How can an impression make the corresponding perceptual belief reasonable?
What sort of evidence can it provide?
246
Hyman
It cannot generally be inductive evidence. 2o There may be circumstances in which the occurrence of an impression of a table could provide me with inductive evidence that I am actually seeing a table in front of me. For example, if I suffered from a severe agnosia, and found it impossible to recognize many things by sight, I might discover that tables are among the things which I am able to recognize, by confirming-e.g. by testimony or by touch-that when it seems to me as if I can see a table in front of me that is usually what I am seeing. If this happened, the fact that it seemed to me as if I could see a table in front of me could be a good inductive reason for me to believe that I was seeing a table in front of me. But the example shows why the evidence which makes our perceptual beliefs reasonable cannot in general be provided by in1pressions. For, if it were, nothing could play the role that touch or testimony plays in the example. In other words, having an impression of a kind of object could not be correlated with having a perception of that kind of object, because the perceptions could not be identified until the correlation had been established. Hence, the kinds of premisses on which inductive arguments depend would never be available. So the evidence provided by impressions for the truth of our perceptual beliefs cannot generally be inductive evidence. But it cannot generally be theoretical evidence either, if the modern causal theory is right in holding that we perceive material objects immediately. The doctrine that our ordinary perceptual beliefs are supported by theoretical evidence is consistent with the classical version of the causal theory, because proponents of classical theory can hold that our ordinary perceptual beliefs embody a conception of the physical world that (as Ayer puts it) '[has the character of] a theory with respect to the immediate data of perception';21 or (as Quine puts it) that 'our talk of external things ... is just a conceptual apparatus that helps us to foresee and control the triggering of our sensory receptors'.22 But this is the principal point that distinguishes the modern theory from the classical one. Hence, proponents of the modern causal theory are bound to reject the view that the evidence provided by impressions for the truth of our perceptual beliefs is theoretical evidence. And so they do. Grice, for example, dismisses it in a parenthesis. If the inductive model is rejected, he writes, recourse may be had to an assimilation of material objects to such entities as electrons, the acceptability of which is regarded as being (roughly) a matter of their utility for the purposes of explanation and prediction; but this assimilation is repugnant for the reason that material objects, after having been first contrasted, 20 My use of the phrase 'inductive evidence' is not n1eant to imply that a premiss of a deductively valid argument cannot also be a premiss of an invalid but inductively strong one. 21 A.]. Ayer, 'Replies', in Macdonald (ed.), Perception and Identity, 289. 22 W. V. Quine, 'Things and their Place in Theories', in Theories and Things (Cambridge, Mass.: Harvard University Press, 1981), 1.
The Evidence of our Senses
247
as a paradigm case of uninvented entities, with the theoretical constructs or entia rationis of the scientist, are then treated as being themselves entia rationis. 23
So, what sort of evidence can it be? Grice and Strawson give very similar answers to this question, and I shall only comment on Strawson's answer. VIII
Strawson claims that 'a certain relation weaker than entailment', which he calls 'presumptive implication', holds between (a) the proposition that it seems to S as if he can see a table in front of him and (b) the proposition that S can see a table in front of him; and of course (b) is supposed to entail (c) that S's impression is caused by a table in front of him. 24 Strawson explains that 'p presumptively implies q' is to be read 'if p is true, then normally or generally q is also true' .25 Hence, his claim is that if (a) is true, then normally or generally (b) and hence (c) are also true. In essence, the proposal is that S's impression of a table in front of him can make his belief that he sees a table in front of him reasonable, because (P 4) It seems to S as if he perceives 0 ~ S perceives O.
('p ~q' is to be read: 'If p then normally or generally q'.) But does this proposal provide a satisfactory answer to Q3? I do not believe that it does. In the first place, a satisfactory answer to Q3 will explain how an impression provides non-inductive and non-theoretical evidence that the corresponding perceptual belief is true, and so makes it reasonable. But, if one proposition presumptively implies another, it does not follow that the evidence provided by the first proposition is neither inductive nor theoretical evidence. For example, if someone drinks a pint of gin in the space of five minutes, then normally or generally he will shortly become unconscious. But the fact that someone has just drunk a pint of gin does not provide non-inductive and non-theoretical evidence that he will shortly become unconscious. Hence, the claim that (a) presumptively implies (b) does not provide a satisfactory answer to Q3. This objection can be met quite easily, by claiming that the presumptive implication from (a) to (b) is non-contingent. In other words, P4 can be n10dified to read as follows: (Ps) D (It seems to S as if he perceives O~S perceives 0.) 23 Grice, 'Causal Theory of Perception', 149-50. Grice seems to imply that unobservable entities such as electrons are postulated for convenience-to smooth the laws of physics, in Quine's phrase-with no pretence that they are real. But the point that our impressions cannot generally provide theoretical evidence for our perceptual beliefs if we perceive material objects immediately is independent of this doctrine. 24 Strawson, 'Causation in Perception', 67. 25 Ibid. 67.
Hyman ('D(p)' is to be read: 'It is a non-contingent truth that p.') As a matter of fact, I am fairly sure that this is what Strawson intended. A parenthetical comment hints at it. And Grice claims explicitly that the connection between an impression of something red and the presence of something red is 'a non-contingent connection'. 26 So this is probably a clarification rather than a modification of the proposal. And, if PSis true, then S's impression evidently can provide S with non-inductive and non-theoretical evidence that he sees a table in front of him. The second objection is that Ps is false. It could only be a contingent truth that, if it seen1S to S as if he perceives 0, then normally or generally S does perceive 0, since it is possible that S recently became addicted to LSD or that his visual cortex recently went haywire for some other reason. But this objection can also be met quite easily, by modifying Ps to read as follows: 248
(P6) If S has normal sight (or hearing, etc.) and circumstances are normal, then D(it seems to S as if he perceives OQ-:?S perceives 0). The third and final objection is that P6 is also false. Indeed, I think it will be obvious that P6 is false if we compare it with P7: (P7) D (If S has normal sight (or hearing, etc.) and circumstances are normal then (it seems to S as if he perceives OQ-:?S perceives 0).) (P6) implies that in normal circumstances a normally sighted person's impression provides him with non-inductive and non-theoretical evidence for the truth of a perceptual belief, but (P7) doesn't imply this. But whereas (P7) is true-since this is what makes a person's sight or a set of circumstances qualify as normal-(P6) is false. Compare P8, which is false, and P9, which is true: (P8) If there are nine planets then 0 (there are more than seven planets). (P9) D (If there are nine planets then there are more than seven planets.) I do not see how this objection can be met. And if it cannot be met then the thought that the notion of presumptive implication can provide an answer to Q3 ultimately depends upon a well-known modal fallacy. (It has been known at least since Boethius.) And if there isn:Jt a plausible answer to Q3 it follows that the modern theory cannot avoid the principal difficulty faced by the classical theory, because it cannot explain how our ordinary perceptual beliefs can be justified. Now, as I said that it would be, towards the beginning of this chapter, my topic has been the justification of perceptual beliefs; that is, beliefs that
26
Grice, 'Causal Theory of Perception', 15 I.
The Evidence of our Senses
249
would be expressed by uttering a sentence whose main verb is a perceptual verb in the first person and the present tense-my present belief that I can see a table in front of me, for example. But could an advocate of the causal theory seize the chance to cut his losses here? Could he concede that my belief that I can see a table in front of me cannot be justified, while insisting that my belief that there is a table in front of me is unaffected by this argument, since it is not a causal belief? I doubt it. If my belief that I can see a table in front of me is unreasonable, it does not follow that the same is true of my belief that there is a table in front of me, for an eye-witness can have independent grounds for accepting the presence of an object that he believes he can see. But if I have no independent grounds for believing that there is a table in front of me, then either both of these beliefs are reasonable or neither is. Besides, drawing this division between the two beliefs is not a remedy that will save the causal theory, since a theory of perception that implies that our ordinary perceptual beliefs are uniformly unjustified must be untenable, regardless of whether it entails a more extensive scepticism.
IX The final question I want to consider is this: (Q4) What reason is there to hold that PI is true?
But first I shall briefly recapitulate. The minimum claim made by the modern causal theory is this: (PI) It is a conceptual truth that when a subject S perceives an external object 0, 0 is causally responsible for S's perceptual experience.
Hence, the theory implies that a perceptual belief is implicitly a belief about the source or cause of a perceptual experience. It follows that the theory also implies that a perceptual belief cannot be reasonable unless it is supported by evidence of an appropriate sort. This would follow immediately if every reasonable causal belief must be supported by evidence. But even if this is false, because causation is sometimes directly observable in the particular case, it still follows, because we do not directly observe the production of our perceptual experiences. Latter-day proponents of the classical causal theory have argued that the evidence that supports our ordinary perceptual beliefs is theoretical evidence, because postulating bodies is the best way of explaining the occurrence of our impressions. But the modern theory is opposed to this view, as it is to any view that denies 'the immediacy that common sense attributes to perceptual awareness', and so it cannot have recourse to this solution. And that leaves it with no room for manreuvre-unless, as Descartes believed, it is possible to devise an a priori proof of God's existence and benevolence.
Hyman
250
So, what reason is there to hold that PI is true? Strawson argues for PI in several places, and a thorough answer to Q4 would look at every argument in turn. I shall not do that. But I shall comment on the most recent (1998) and the earliest (1974) texts. It is a relatively straightforward matter to apply the comments I shall make about these passages to the other ones. Here is the 1998 passage: in order for an experience to amount to a genuine perception of an object (and hence a way of gaining knowledge about it) there must be such a relation between object and experience as to rule out the case of a subject's being n1erely flukishly or accidentally right in taking it that there is just the object before him that he takes himself to be perceiving ... the relation of causal dependence ... remains the only plausible candidate. 27
And here is the 1974 passage: The concept of perception is too closely linked to that of knowledge for us to tolerate the idea of someone's being merely flukishly right in taking his M-experience [sc. impression] to be the M-perception that it seems to be. Only those M -experiences which are in a certain sense dependable are to count as the M -perceptions they seem to be; and dependability in this sense entails dependence, causal ... dependence on appropriate M-facts. 28
These two passages are similar, but not as similar as they may look at first glance, and I shall comment on them separately, starting with the 1998 one. My objection to the argument in this passage is that what the first part says is false. It is true that the senses are cognitive faculties, i.e. capacities for acquiring knowledge; but it does not follow that in order for an experience to amount to a genuine perception of an object it has to be related to the object in a way that prevents the subject's belief from being merely flukishly right. And in fact this is false. The argument is simple. If it were true, then every perception would be related in the fluke-excluding way to the object that it is a perception of; and hence a subject could never be merely flukishly right in taking it that there was just the object before him that he took himself to be perceiving. But this is false. For exan1ple, if I did not know that the famous sprinter Sally Fleetfoot has an identical twin sister who is also a sprinter, I could quite easily take myself to be seeing Sally Fleetfoot before me, at a race somewhere, and be flukishly or accidentally right. (The literature on the Gettier problem contains many examples that prove the same thing, i.e. that one can be merely flukishly right in taking it that there is just the object before one that one takes oneself to be perceiving: for instance, Goldman's well-known story about papier mache barn fa~ades.29)
27 29
Strawson, 'Reply to Paul Snowdon', 314. 28 Strawson, 'Causation in Perception', 71. A. I. Goldman, 'Discrimination and Perceptual Knowledge', Journal of Philosophy, 73
(197 6 ), 77 1-91.
The Evidence of our Senses
251
The argument in the 1974 passage is more interesting, for two reasons. First, the flaw in the argument is less obvious. Secondly, the argument reveals more about the relationship between the modern causal theory and the classical causal theory advanced by Locke. My objection to this argument is that it involves an assumption which is false; namely: (PIO) A perception is a kind of impression.
viz., one that is the perception that it seems to be. The assumption is important, because, if we accept that it is true, then we shall naturally want to ask what kind of impression is the perception that it seems to be. And since a perception does not carry a hallmark of veracity, the answer will be bound to refer to a relation in which an impression can (but need not) stand to something else-presumably the object of which it seems to be a perception. And this takes us a good part of the way towards PI. In fact the only step which remains is to claim that 'the relation of causal dependence [is] ... the only plausible candidate'. There are two principal reasons for denying that a perception is a kind of in1pression. First, a subject may (for example) see something without its seeming to him as if he can see anything, because the perception is subliminal. The literature is well known, and the examples are legion. But if one can have a perception without having an impression, a perception cannot be a kind of impression. Secondly, the term '5's impression of 0' was introduced as an abbreviation of the term 'its seeming to S as if he perceives 0'. (It was precisely by introducing the term in this way that PI was shown to be consistent with direct realisn1.) But it involves a basic gramn1atical n1uddle to suppose that its seeming to S as if he perceives 0 seems to S to be a perception of o. The reason is this. The term 'its seeming to S as if he perceives 0' is a non1inalization of the sentence 'It seems to S as if he perceives 0'. The nominalization is produced by a perfectly standard transformation; namely, by replacing the main verb of the sentence with a gerund and putting the subject in the genitive caseeven if it is a so-called prop subject, as it is in 'It seems to S as if he perceives 0'. 'It is warm in Aberdeen' yields 'its being warm in Aberdeen' in the same way. (The 'it' is a prop subject in this sentence too.) And 'Brutus killed Caesar' yields the so-called imperfect nominal 'Brutus's killing Caesar' and, with the insertion of the word 'of', the so-called perfect nominal 'Brutus's killing of Caesar'. Now the nominal 'its being warm in Aberdeen' does not designate a thing that is warm and 'Brutus's killing of Caesar' does not designate a thing or person that killed Caesar. If we imagine that it does, we are probably confusing it with the definite description 'the x such that x killed Caesar'. But
Hyman there is no more truth in the idea that its seeming to S as if he perceives 0 is a thing that seems, than there is in the idea that its being warm in Aberdeen is a thing that is warm or that Brutus's killing of Caesar is a thing that kills. Hence, S's impression does not seem to be a perception. It is a seeming, not a thing that seems. If impressions were experiences that seem to be perceptions, it would be plausible to hold (as Strawson did in 1974) that some impressions do not merely seem to be, but actually are, perceptions-namely, the perceptions that they seem to be. But, since an impression is not an experience that seems to be a perception, this view is untenable. S's perception of 0, for example, is not an impression-an instance of its seeming to S as if he perceives something. It is (I suggest) simply an instance of the relation expressed by the verb 'perceives' in the sentence'S perceives 0'. This follows from the fact that'S's perception of 0' is a nominalization of the sentence'S perceives 0'.30 But it contradicts the idea that a perception is an in1pression, or a 'slice of sensible experience'-the phrase is Strawson's31-if, that is, a slice of sensible experience is meant to be an episode inside the brain, or in the soul, or confined within the subject's skin. For, if S's perception is a relation between Sand 0 it cannot be confined n10re narrowly than Sand 0. 32 252
x I shall add a few remarks in conclusion. In the final analysis, the modern causal theory is rooted in two empiricist doctrines: (i) that the foundations of empirical knowledge are propositions about our impressions; and (ii) that
30 I discuss these constructions at some length in '-ings and -ers', in E. Borg (ed.), Meaning and Representation-Ratio, 14 (December 2001), special issue, 298-317. 31 Strawson, 'Perception and its Objects', in Macdonald (ed.), Perception and Identity, 43. 32 Philosophers have sometimes argued that (e.g.) my perception of a tree cannot be an instance of a relation between me and a tree on the grounds that I can have exactly the same experience when there is no tree present for me to be related to. For example, D. M. Armstrong writes as follows: 'There is a most serious objection to [the] attempt to construe mental states as relations to things in the world. Suppose, as is perfectly imaginable, that I have exactly the same perceptual experience as I had when I looked at the tree, but suppose that this time there is no tree there ... [In this case] there is, by hypothesis, nothing in the world for me to be 'mentally related' to. So no unique, irreducible, relation can be involved. Yet, also by hypothesis, the mental state is no different from the n1ental state in [the case where I looked at the tree]. So no mental relation of ourselves to things in the world is ever involved' (A Materialist Theory of the Mind, rev. edn. (London: Routledge, 1993) 39). This argument is fallacious. For, what is perfectly imaginable is that I should have exactly the same impression when there is no tree present. By hypothesis, the impression is not different in the two cases. But the mental states which I hold are relations (or, better, instances of relations) are not impressions, but perceptions. Armstrong simply assumes that a perception is a kind of impression, and therefore finds it easy to infer from the fact that impressions are not 'relations to things in the world' that perceptions are not either.
The Evidence of our Senses 253 a perception is a species of idea or impression. Strawson has recently attempted to defend PI without assuming that (ii) is true. But, as we have seen, the attempt fails for a reason that has been a commonplace in the epistemological literature for several decades. And so it is not surprising that the unexpurgated theory continues to predominate. In any event, if the modern causal theory is incapable of explaining how our ordinary perceptual beliefs can be reasonable, as I have argued that it is, there are two obvious ways out of the quandary. One is to revert to a form of indirect realism, such as that defended by Ayer or Quine. The other is to accept Kenny's proposal (contra (i) ) that, for example, the proposition that there is a table in front of me is evident to me, and therefore I believe it without evidence; and to devise a theory of perception that is consistent with it. In the last paragraph of XI, I indicated what I believe the core doctrine of this theory ought to be. I shall conclude by commenting briefly on Kenny's proposal. If it is true, the propositions that we believe without evidence, and that provide the evidence that n1akes our body of empirical beliefs largely reasonable, include propositions that are not about experiences, but about their objects. This is not to deny that propositions about our sensations or impressions can provide us with evidence. They certainly can. For example, my evidence that the stuff in a bottle is intoxicating may partly consist in my feeling tipsy, and my evidence that the water in a pool is chlorinated may partly consist in my eyes stinging. But although becoming tipsy is a pleasant way of finding out whether the contents of a bottle are intoxicating, it is not an example of sense perception; and although it may be certain that the contents of the bottle are intoxicating, it cannot be evident. So, I may accept that the contents of a bottle are intoxicating, or that the water in a pool is chlorinated, on the evidence of my sensations. But, if Kenny is right, I do not accept that there is a table in front of me on the evidence of my impressions. In ordinary circumstances, I neither have any evidence that there is a table in front of me, nor need any, because seeing is a way of knowing-and looking is a way of finding out-without evidence. Hence, my judgement that there is a table in front of me is made-borrowing a phrase of Wittgenstein's-without a justification, but with a right. And if I am asked how I know that there is a table in front of me, and I reply that I can see it, my answer does not indicate my evidence-namely, the impression of a table. On the contrary, it indicates why evidence is unnecessary. Obviously, this view of the matter represents a break with the empiricist tradition. But the question of whether this is a merit or a defect in the view is one which I shall not attempt to decide.
Index abstract entity/object 12-13, 51; see also universals abstraction 75-6, 106 aesthetic judgement ch. I I not merely subjective 188 aesthetic idea 202-3 Allison, H. 9, 28, 68, 77, 88, 159 Analogies of Experience 135-6 analogy 175 analysis (conceptual/logical) 15, 17, 44, 117-18, 121-2, 235 connective 40-2, 49-50, 61 analytic/synthetic 23,32,41-2,59,79, 112-13, 122, 151-2; see also analysis and ideas, relations of analytic Kantianism ch. 2 wide vs. narrow sense 19-2 I analytic philosophy 9, 19-20, 33-4, 4 6 and Kantianism 15-16 see also linguistic turn Anscombe, G. E. M. 90, 243 Antinomies 18o appearances 27-30, 177-8 see also things as they are in themselves and transcendental idealism a priori/a posteriori 23,25-6,73-4,78-81, 114-15 derivational vs. justificational a priori 87-9,92-5,96-8,101,1°7-8 Aristotle 16,18,25,35,51 Austin, J. L. 8, II, 13, 33 Ayer, A. J. 7, 45-6, 253 Baum, M. 67 beauty 188, 197-8, 204 natural 189-90, 193, 201-3 Beck, L. W. 191 Benaceraff, P. 40-1 Bennett, J. 19, 20, 9 6 Berkeley, G. 18,27-8,51,63 Bird, G. 19, 28 body see object and I1self Bolzano, B. 16 Bounds of Sense 8-9, 19-20, 22, 25, 27, 36, 56, 67, 12 7, 155-6 , 16 3, 16 5 Bradley, F. H. 47 Braithwaite, R. 46 Brittan, G. 79 bridge principles 206,2°7,211,214,215, 21 7,218
Burge, T. 212 Broad, C. D. 17,33, 38 Carnap, R. 16, 45-6, 63 Cartesianism 14,127-9,133,140, ch. 9 categories 17, 20, 26, 30, 49-5 0, 5 2-4, 87-9, 9 1, 9 8 , 139 see also concepts, conceptual scheme, Metaphysical Deduction causal theory of perception ch. 14 modern vs. classical 235-7,246,249,251 causation 24, 30, 88-9, 90, 99, 103, 107, 122-3, 135-7, 17 1, 177,241-4, principle of 41-2,116 see also natural laws cognitive processes/capacities/faculties 32-3, 88,94,118- 19,128-33,17°,195 see also psychology and transcendental psychology Cohen, H. 25 Collingwood, R. G. 7, 25, 47 common-sense 181 content see propositions concepts 14,41-2, II8,ch. 8, 171-4 and abilities 96 a priori vs. empirical ch. 5 instantiation of 89-92 moral 52, 94-5 . presupposed/independent of experience 102-5 recognitional 96-7 singular/individual (complete concept of an individuum) 143, 145, 149-51, 186, 194-5, 199 conceptual schemelframework 12, 18-19, 25, 39-4 2 , 55, 62 alternative 48, 62-5, 120 Copernican Revolution 25, 27, 72-3, 83 Critique of Judgement ch. 10, ch. II Critique of Practical Reason 180, 203 Critique of Pure Reason 7-8, 10, I I, 12, 18, 21-2,25-6,67,185-6,19°-1 dark side of 20, 26-7, 109-10 First Edition 29 criteria 37-9 of identity 13, 141, 148-9 Descartes, R. 14, 18, 5I, 62-3, 128-9, 235 Doctrine of Method 34 Durer, A. 199
Index empiricism 20, 22, 73-4, 7 8, 84-5, 129, 139, 25 2 concept- 136-7 epistemology 21-2,4°-1,67,1°7,128,133 Evans, G. 96-7, 100-1 events 4 1-2, 5 I evidence 239,240-1,245-6,253 exhibition (Darstellung) 195-6, 199 existence 146-7 experience 12, 20, 51, 117, 164 conceptual 1°3 inner vs. outer 129 transcendental preconditions/limiting features of 24-5, 26-9, 34, 67, 7 1, 77-8,81-3,94-5,111-14 of material objects 103-5, 107-8 vs. objects of experience 53-4 see also judgements of experience; perception externalism 207, 211, 212, 213, 214, 215, 216, 217 external world 223,225,232-33 extrinsic vs. intrinsic properties 10-1 I first-person pronoun see I/self foundationalism 39 free play (of understanding and imagination) 18 5, 18 9, 194, 19 8-9 Frege, G. 13, 16, 17 Friedman, M. 25, 67, 75, 76, 9 1 Gardner, S. 30-1 genius 201-2 German idealism 21, 43, 199 Gettier, E. 128, 250 Ginsborg, H. 171 God 98, 149-5 I; see also ontological argument Goldman, A. I. 250 Grice, P. 11,235-9,242,247,248 Groundwork of the Metaphysics of Morals 7-8, 180, 191 Guyer, P. 27, 187 Haack, S. 75 Hacker, P. M. S. 18 Hegel, J. W. F. 15, 199 Heidegger, M. 47 hermeneutics 15 Herz, M. 191 Hume, D. II, 39-40, 43, 73-4, 9 8 , 115-16, 122,124,132,133,135-9,156,175-6, 178, 180-3, 21 9-26, 24 1 I/self ch. 8, ch. 9 as a concept 145-6, 158; see also concept, singular/individual
embodiment of 14 1- 2, 155-6, 165-7 in sensu latiori/in sensu stricto 148, 15 2, 153 reference of 155-8, 161 see also persons, self-consciousness and transcendental unity of apperception ideal language philosophy see ordinary language philosophy idealism 10, 15 see also transcendental idealism ideas regulative 95, 17 2 relations of 117-18, 123 transcendental 88, 91 identification/re-identification 19, 36-7, 39, 55, 13 6- 8 , 15 2 identity logical vs. real 142, 149-5 2, 153-4 personal 156,160,162 imagination 30, 195-7; see also exhibition and free play innateness 102 Individuals 8, 17-19, 21-2, 25, 36, 49-56, 66, 67,83,110,162 induction 169, 175, 177, 181-3, 220-3, 225, 24 6 intention 239 introspection 130, 132 intuition 30-I, 144, 150 forms of 128, 139 pure 34 transcendental affinity of the manifold of 133-4 Jacobi, F. H. 30 judgement 13, 129, 143, 171- 2 forms of 32-3, 128 heautonomy of 182-3 of experience 172-5, 179-81 reflective 174 see also concepts justification 68-70, 207-9, 218, 237-8, 24 8-9,253 Kemp Smith, N. 19 Kenny, A. J. P. 24 2, 253 Kitcher, Patricia 32-3 Kitcher, Philip knowledge 35, 3 8,4°-1, 81, 133 tacit/unconscious 3 I Korner, St. 17, 25, 3 5 Kripke, S. A. 80-I, 83, 218 Langton, R. 9-11 language 8, 18, 37-8, 49-5 0, 53-4, 62, 73-4; see also conceptual scheme and meaning Lectures on Metaphysics 145, 148
Index Leibniz, G. W. F. 10, 18, 51, 62-3, 143, 186 linguistic rules 57-9 linguistic turn 16, 19, 140 Linnaeus, C. von 174 Locke,]. 27,32,78,99,235,251 logic 8, 18, 92 pre-Kantian 143, 15 1 transcendental 25,13°,172-3 Logic (J aesche) 72, 175 logical positivism 15,16,17,43-6 McDowell, ]. 104, 155-9 McTaggart, J. E. 46 mathematics 22, 24, 25, 34, 75-6 meaning 37-8, 40, 46, 161-3 mechanics (primitive/intuitive) 96-7, 102, 105 Metaphysical Deduction 17 Metaphysical Foundations of Natural Science 25,137, 190 metaphysics 17,21-6, ch. 4,107, III, 114-15 descriptive vs. revisionary 18-20, 22, 25, 4 1, 49-66, 67-85; see also conceptual scheme, alternative historical view of I 16, 120-2; see also Collingwood traditional 46-7,53-7, 62-5 transcendent vs. transcendental 21 see also ontology Mill, J. S. 115, 13 2,241 mind 51,129, 159; see also cognitive processes, experience and self-consciousness modal intuitions 207, 210-18 monads 143 Moore, G. E. II, 13, 15, 16, 17, 40, 46 natural laws/natural kinds 17°-7; see also purposiveness of nature naturalism and anti-naturalism 15, 40-I, 12 7- 8 , 21 9-33 necessity 67, 75, 7 8- 83, 84, 112, 114 rooted in language or reality 4 1- 2, 55-6o, 117- 2 5 see also modal intuitions and synthetic a priori neo-Kantianism 15, 16, 21 Neurath, O. 45 nominalism see abstract objects and universals nominalization 251-2 nonsense see meaning object intentional 29 logical (entity/something) 146-50, 152, 153-4 material 50-I, 67, 100-6, 129, 13 6, 144
257
of experience 24-5,41 see also particulars ontological argument 146 ontology 40-I, 60 Opus Postumum 193 ordinary language philosophy 17, 69, see conceptual analysis and ideal language philosophy Oxford philosophy 8, I I, 17-18, 46 Parologisms 142-3,145-6,148,151, 15 6-9 particulars 19, 36-9, 110, 132, 138, 171- 2 basic 51 Paton, H.]. 19 perception 214, ch. 14 directvs. indirect 235-7,246,253 based on inference 239, 243 see also sensibility and evidence persons 19, 51, 54, 67, 78, 110 phenomenalism 10,235-6 phenomenology 15 philosophy as conceptual self-understanding 12 as different from science 15-17 pictures in 12, 14 systematic 12 physicalism physics 9, II, 26, 75 Plato 14 Platonism 14, 16 Plotinus 186 pre-critical writings 199 predicate/predication 18, 22, 50, 17 1- 2 , 175 logical vs. real 146-9 presumptive implication 247-9 presupposition 18,25,47,5°,102-7,116, 13 1, 133 Price, H. H. 47-9 primary vs. secondary qualities 27, 100-1 private language 38 Prokgomena 71,143,145,172,200 properties 14 intrinsic vs. extrinsic 10-1 I propositional attitudes propositions 13 psychologism see transcendental psychology psychology 32-3, 128 rational 141,145,148-9,156; see also Parologisms purposiveness of nature ch. 10, 193 formal vs. logical 170, 176, 186, 190 Putnam, H. 212-13 quaestio facti vs. quaestio iuris 31-2, 200 Quine, W. V. 8, 12, 15-16, 79, 24 6, 253
Index rationalism 20-2 realism ch. 12; see also idealism and transcendental idealism reason 88, 113, 120-I, 131, 159; see also idea and transcendental philosophy reference 18-19, 22, 25, 50-1 reflection transcendental vs. empiricaVepistemic 130, 14° Refutation of Idealism 34, 71- 2, 127 Reinhold, C. L. 187,19°-1 relations dynamic vs. spatial 10 representations/representationalism 28-9, 3 1 ,235- 6 supplementary (Nebenvorstellungen) 198-9, 202 Rorty, R. 32, 140 Rousseau, J. J. 191 Russell, B. 8, II, 15, 17, 18, 21, 44, 46, 24 1 Ryle, G. II, 17, 46-7 Salmon, W. 220, 225, 229-32 sceptical method 71-3 scepticism 22, 35-40, 67-74, 79, 82-3, 84, 133, 208, 209, 210, 216, 218, 219-33 Cartesian 22, 127-8, 215 Humean (induction) 22, 73,176 other minds 39 self-refuting 21, 35, 210 direct vs. indirect responses to 36 Schematism 91 Schlick, M. 45 Schopenhaue~A. 186 Schutz, C. G. 187 science 22, 25-~ 73, 77 Sellars, W. 159 self ch. 8; see also self-consciousness self-consciousness/self-ascription I I, 22, 30, 35,133, ch. 8, ch. 9 criterionless 156, 160-7 empirical vs. a priori 14 I sensations 27, 129, 139 sense outer/inner 10-II, 129, 139, 148 sensibility 30, 32, 13 2 sense-datum/sense-impression 113,236,238, 25°-1 singular term see reference Skepticism and Naturalism 39, 70, 83, 206 space 10-11,24, 34, 35, 101-2, 13 1, 144 Stroud, B. 22, 36-9, 69, 127, 205-6, 210, 23 2
subject logical 142-8; see also object substance 88-9, 91-4, 99-103, 106, 107, 14 1-3, 145-7 synthesis 30-I, 133, 139 synthetic a priori 15, 17, 24-6, 33, 34, 4 1- 2, 44,56-7,79-80, 84-5,ch. 6 systematicity 171-3 teleology 188, 190, 192; see also purposiveness of nature things in themselves 9-1 I time I I, 24, 34, 13 I transcendental arguments 20, 21-2, 33-9, 69-7 0 , 85, 12 7, 20 5-18, 220, 23 2 ambitious vs. modest 38, 60, 119, 206, 207, 209, 217, 218 deductive vs. elenctic 35-6 Transcendental Aesthetics 34 Transcendental Analytic 176-9 Transcendental Decuction 34, 15 2, 156, 177-9, 182 objective vs. subjective 31-2 in the Critique of Judgement 169, 176-82, 19 2,200-2 transcendental idealism 20, 26-33, 68, 107, 111-13, 119-20 double aspect interpretation of 10, 28-30 transcendental philosophy 20, 24, 109 as a meta-theory 25-6 transcendental proofs 34, 135, 140 transcendental psychology 20, 26, 29-33, 74-5, 128-9 transcendental unity of apperception 3 I, 133,152,155,178; see also self-consciousness, empirical vs. a priori truth 125 presuppositions of 18 bearers of 13 understanding 30, 32, 129, 132; see also exhibition and free play universals 14, 33, 171, 179 verification 43-4, 85,152,153, Vienna Circle see logical positivism will 139 Williams, M. 209, 228 Wittgenstein, L. 12,16, 17, 38,44-5,59-60, 65, 161, 24 1, 243 Yablo, S. 212
