¹ This is the most general, and crudest, way of drawing such a distinction. It is, of course, not just the ordinary contrast between what is mental and what is physical, but a philosophical gloss on that distinction, and it should not be thought that I accept its terms. McDowell himself invokes also related, but different, distinctions between the natural and the intentional, and the natural and the normative, in order to canvass more subtle contrasts.

² MW p viii. McDowell actually says: “I am not sure that Strawson's Kant is really Kant .”, but I think his reservations echo the belief that what Strawson approves in Kant is not adequately expressed by Kant, while my claim is that what Strawson disapproves in Kant is not present in Kant's text.

³ See my “Recent Interpretations of the Transcendental Deduction”, Kant-Studien 1974, Akten des 4-ten Kant-Kongresses, Teil I pp 1–14.Google Scholar; “Kant's Transcendental Idealism” in Idealism - Past and Present ed Vesey, G, CUP, 1982 Google Scholar; and “Tradition and Revolution in Kant”, forthcoming in Kant-Studien.

⁴ In “The Problem of Realism and the A Priori”, in Parrini, P (ed), Kant and Contemporary Epistemology, Kluwer, Dordrecht, 1994 CrossRefGoogle Scholar.

⁵ It is difficult to overestimate the marvel. Later I suggest that it is so striking as to count against the attibution of the dark story to Kant.

⁶ In Kant's, Theory of Knowledge, Routledge, 1962, Preface p ix Google Scholar, I had also indicated that historical parallel.

⁷ Kants revolutionary aspirations and his Copernican experiment indicate generally a rejection of traditional dualisms, but that motivation is evident also throughout the Dialectic. Even the distinction between sense and understanding, as McDowell rightly sees, is not a rigid dualist separation, as I had also indicated in Kant's, Theory of Knowledge pp 58ff.Google Scholar

⁸ Later I indicate how important is the distinction between the epistemological and moral contexts in understanding Kant's appeal to the supersensible. McDowell's case, however, rests solely on the former epistemological context and I restrict my own comments to that.

⁹ It is important to separate the attractiveness of the diagnosis for some putative error from the establishment of the error itself. Bennett's ingenious diagnosis of Kant's alleged error in canvassing the synthetic a priori classification presupposed, and did not establish, that there was any error. Kant's Analytic, Cambridge, 1966, pp 6-8, 39–44 Google Scholar).

¹⁰ I have considered that former issue about Kant's commitment to noumenal reality elsewhere in Kant's Theory of Knowledge Ch 2, “Kant's Transcendental Idealism” and “Tradition and Revolution in Kant”.

¹¹ B xix-xxiii, the Preface to the Second Critique and the Introduction to the Third.

¹² For example B 8-10; B 737-742. The explicit distinction between the transcendental and the transcendent is made at B 352.

¹³ Aquila, Richard “Things in Themselves and Appearances: Intentionality and Reality in Kant”, Archiv fur Geschichte der Philosophie 61 (1979) pp 293–308 CrossRefGoogle Scholar. Robinson, Hoke, “Two Perspectives on Kant's Appearances and Things in Themselves”, Journal of the History of Philosophy Vol XXXII, 07 1994 Google Scholar.

¹⁴ Carnap, R, Logical Syntax of Language Kegan Paul, London, 1937 Google Scholar. Carnap's account is more technical than Kant's, but has also evident weaknesses. One is his insistence that formal and material sentences are equivalent, or equipollent, even though for ‘quasi-syntactical’ or ‘pseudo-object’ sentences the material mode of expression is misleading. Another arises from some unplausible analyses, as when he claims that “Yesterday's lecture was about Babylon” is equivalent to “In yesterday's lecture either the word ‘Babylon’ or a synonym occurred”. Kant's account is less formally developed, and rests on a recognition of an intentionality which enables us to speak of apparent objects under different descriptions and without commitment to existence, and of philosophical theories as providing such re-descriptions of familiar objects in experience.

¹⁵ McDowell's correct acknowledgment of Kant's aim to supersede traditional philosophy enables me to express my disagreement with him in a sharper way. Since for McDowell Kant's failure rests on his acceptance of the dark story, and since I deny that Kant accepted that story, it is open to me to claim that Kant was more successful in his aim than McDowell allows. This has implications for the more positive project that Kant pursues, but I cannot deal with them here.

¹⁶ See also, for the moral context, the way in which Kant outlines his procedure in the Grundlegung. But that procedure is standard throughout the Critical philosophy.

¹⁷ Most commentators on Kant nowadays do not think that Kant's priorities should be understood in any such temporal way. A recent example of that general view is Neiman, Susan in The Unity of Reason, OUP 1994, p 50 CrossRefGoogle Scholar.

¹⁸ On Kant's rejection of phenomenalism see Kant's Theory of Knowledge Ch 1, and “Kant's Transcendental Idealism”. McDowell complains of Allison that he wrongly ascribes a ‘psychologistic phenomenalism’ to Strawson's account of Kant, but there is no doubt that Strawson does, wrongly, ascribe a ‘phenomenalistic idealism’ to Kant throughout Part V of The Bounds of Sense.

¹⁹ Austin, J L, Sense and Sensibilia, OUP 1956 Google Scholar makes these points in comparing the philosophical sense-datum tradition with our colloquial use of such terms as ‘appear’. It has sometimes been thought that Austin's interest in the minutiae of ordinary language had no relevance to the philosophy of perception (See Bennett, J, Mind Vol LXXV, 1966, pp 501–515 CrossRefGoogle Scholar) but this under-rates Austin's achievement in that work. Bennett wrote: “What has this to do with the old questions about appearance and reality? Almost nothing.”, and referred to these old questions as “the great tradition of modern epistemology”, op cit p 513. But Austin, like Rorty much later, was precisely aiming to reject that ‘great tradition’. Both Kant and Austin understood how damaging unmarked philosophical distortions of colloquial use could be. See my entry “J L Austin” in A Companion to Epistemology ed by Dancy, J and Sosa, E, Blackwell, 1992, pp 34–36 Google Scholar.

²⁰ The terminology is too complex to be considered here in detail. But sensations, or sometimes ‘impressions’, are clearly mental states with an a posteriori content; intuitions have a direct relation to represented objects and can have a priori content; while appearances are the, indeterminate, objects of empirical intuitions. Sensations can be assimilated to crude ‘subjective’ mental states, but neither ‘intuition’ nor ‘appearance’ can be understood in that simple way. Kant invokes something like an ‘intentional’ account of the content of intuitions as a way of underlining the immediate relation he claims between those representations and the objects represented Although the notions of ‘intentionality’ and ‘externalism’ are both unclear, and associated with later philosophers, nevertheless Kant's views can be associated with those ideas. I discussed these issues extensively in Kant's Theory of Knowledge eg pp 6, 15-16, 35, 54ff, 63, 78.

²¹ See Kant's Theory of Knowledge pp 47ff. for a discussion of Kant's efforts and their relation to the distinction between primary and secondary qualities. Of course Kant wished to distinguish ‘appearance’ and ‘phenomenon’ (see Kant's Theory of Knowledge pp 53-54) by associating the former strictly with the senses and the latter with the understanding's grasp of what appears to the senses. Kant's way of drawing this distinction has more in common with the view of Evans's ‘informational system’, which McDowell criticises, than with McDowell's own position (MW pp 47-65).

²² Erich Adickes provided an exhaustive catalogue of Kant's apparent commitments to noumena in the Critique in his Kant und das Ding an Sich I considered, and rejected, some of those commitments in Kant's Theory of Knowledge and “Tradition and Revolution in Kant”. As a result of examining Adickes's catalogue in detail before 1962 I came to the view that his conclusions were unjustified. I hope to publish the results of that survey.

²³ Quine's form of semantic holism has been influential, and Wittgenstein's later philosophy can be read as a gesture towards some form of holistic view of language. But these notions remain obscure and sometimes controversial. See, for example, Fodor, , Psycho-Semantics MIT Press, 1987 Google Scholar; Clark, Andy, Microcognition MIT Press, 1989 Google Scholar; and Fodor, and Lepore, , Holism; A Shopper's Guide, Blackwell, 1991 Google Scholar.

²⁴ The idea of providing a descriptive map of experience is not confined to Strawson, although he coined the term ‘descriptive metaphysics’. Some version of such a project is present also in Ryle's philosophy and in Quine's. Kant's version differs importantly from Strawson's in a number of ways. It relies on a synthetic a priori classification which Strawson rejects; it is opposed to a ‘justificatory’ rather than a ‘revisionary’ metaphysics; and it produces a different response to scepticism.

²⁵ Kant's Theory of Knowledge Ch 1.

²⁶ See my ‘Kant's Transcendental Arguments” in Reading Kant ed Schaper, E and Vossenkuhl, W, Blackwell, 1989, pp 21–39 Google Scholar.

²⁷ Kant also uses the ‘map’ imagery, and even speaks of Hume as “one of the great geographers of human reason” (B 295 and B 788).

²⁸ Haack, Susan “Reflections on Relativism”, in Philosophical Perspectives 11: Metaphysics ed Tomberlin, James, Ridgeview Press (forthcoming)Google Scholar. Her distinctions between weak and strong mind-(in)dependence, though more complex than the traditional dualism are not quite the same as Kant's. Like Austin, however, I believe that although traditional dualism needs to be rejected, it cannot be adequately replaced with just a three-, four-, or five-fold classification. Experience is more complex than such limited classifications allow.