Philosophical Review

Kant's Intentions in the Refutation of IdealismAuthor(s): Paul GuyerSource: The Philosophical Review, Vol. 92, No. 3 (Jul., 1983), pp. 329-383Published by: Duke University Press on behalf of Philosophical ReviewStable URL: http://www.jstor.org/stable/2184480Accessed: 11/10/2010 00:45

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/action/showPublisher?publisherCode=duke.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact support@jstor.org.

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extendaccess to The Philosophical Review.

http://www.jstor.org

The Philosophical Review, XCII, No. 3 (July 1983)

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM Paul Guyer

I

In the "Refutation of Idealism"' which he added to the second edition of the Critique of Pure Reason, Kant claimed that "the

mere, but empirically determined consciousness of my own existence proves the existence of objects in space outside me." The argument for this thesis turns on the claims that "All time-determination presupposes something 'permanent' in perception,"2 and that "perception of this permanent is possible only through a thing outside me and not through the mere representation of a thing outside me" (B 275; see also B xl). But these confidently asserted claims raise two serious problems. First, in spite of the stress Kant places on the contrast between a "thing outside me" and a "mere representation," it is not obvious what this contrast means. Thus, what the Refutation is sup-

'Printed with initial capitals, the expression "Refutation of Idealism" will refer to the specific text at pages 274-279 of the second edition of the Critique; the expression "refutation of idealism" without capitals will refer to Kant's entire effort to refute idealism as recorded in all of the texts to be discussed in this paper.

Reference to the Critique of Pure Reason will be given by the pagination of' the ("A") and/or second ("B") editions. Translations usually but not always follow Norman Kemp Smith, Immanuel Kant's Critique of Pure Reason, cor- rected impression (London: Macmillan, 1933). References to other Kan- tian passages will be given by volume and page number of the Akademie edition (Kants gesammelte Schriften, edited by the Kiniglichen Preussischen [later Deutschen] Akademie der Wissenschaften, 29 vols., Berlin: Walter de Gruyter and predecessors, 1902- ), however, citations to the numbered Reflexionen included in volume 18 will omit the volume number after foot- note 8 below. Translations of these Rejlexionen are my own.

2Kant's word for permanent is beharrlich. It would probably be prefera- ble to translate this by words like "persisting" or "enduring," which con- note the idea of more than momentary existence without implying indefi- nitely extended or eternal existence as "permanent" may. But since Kemp Smith has canonized "permanent" (though he does at least occasionally use "persisting" himself' -cf. A 185/B 229), I will continue to use that transla- tion when quoting from the Critique, though not necessarily when quoting from other sources.

329

PAUL GUYER

posed to prove is unclear. Second, nothing in the text of' the Re- f'utation itself' actually explains how its conclusions is supposed to be reached. It offers no argument at all fOr the thesis that, something permanent is needed to make temporal determinationls Mnd only a woefully inadequate one f'or the thesis that whatever satisfies this function must be something distinct fromt the self' and any of' it's own representations in whatever way is inl flact intended.

It is easy to explain why the intended colnclusion of' the Refuta- tiOn is obscure in spite of' Kant's strongly worded distinction be- tween a thing outside me and the mere representation of' such a thing. It is obscure because Kant's discussion of' idealism in the./irst edition of' the Critique had claimed that the very "expression outside us is unavvoidably ambiguous." In the f`o6urth Paralogism of 1781 Kant had argued that a reference to a thing outside Us might be understood to signify "what a things it itseltfexists (Iistinct from us, that is, what is ontologically independent of' ourselves, but. could also be taken to mean just, "what belongs solely to outer appearatwe," that is, what has the phenomenological form of' things "which ar)e to be found in spe"p( without being "external in the transcendental sense," that is, ontologically independent of' us (A 373).3Nor di(d Kant merely diagnose this ambiguity. Rather, he argued that the problem of' "skeptical idealism," the incapacity to prove the exis- tence of' external objects (A 377) because of' the necessary inI- conclusiveness of' any attempt to infer to their existence as the cause of' our representations of' then (A 368, 372), could be solved only by his ownl "transcendental idealism" unlderstood prectsely as "the doctrine that appearances are to be regarded as being, one and all, representations only, not things in themselves" (A 369). That is, Kant did not merely contrast the concepts of' ontological and phe- nomienological externality, numerical distinctness from the self'and any of' its states on the one hand and the mere f`orm of' spatiality on the other; he also argued that, objects with spatial f'ormn had to be reduced to what are ontologically merely states of' the self' in order

4-htis, Kemp Smith's claim that the expression "objects in space outside me in the thesis of the Refuttation in B is pleonasticc" (A (Cmmel/tary lo Kanlt's Critique oJ Pure Reasoti, 2nd ed. [Lonc-don: Macmillan, 1 923], p. 309 n . 2) is true only if "outside me" is taken in the merely phenomenological rather than ontological sense distinguished by Kant in 1781. Whether it slhold be so taken is precisely the question to be addressed here.

33()

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

to render them safe from doubt. He held, of course, that there was a phenomnenological difference between representations of inner states and of external objects, but insisted that "External objects (bodies), however, are mere appearances, and are therefore noth- ing but a species of' my representations, the objects of which are something ()nly through these representations" (A 370); "matter and corporeal things" are "merely appearances, that is, mere kinds of representation, which are never to be met with save in US, and the reality of which depends on immediate consciousness, just as does the reality of ourI own thoughts" (A 372). But if in 1781 Kant held that external objects could be ontologically reduced to a vari- ety of our ofwn representations yet still be described as "outside us," then it can hardly be self-evident that when he claims in 1787 that the temporal determinations of self-consciousness require "objects in space outside us" he means to prove the independent existence of things numerically distinct from the self and its states. He could just mean to add atn argument that we must have representations with the form of spatiality as well as ()nes without it. Unless other evidence can be found, it seems it must remain unclear whether or not the Refutation of Idealism is intended to break with the on- tological reductionism of 17811.

4Kant's p1retatory amplification of' the Refutation claims that the new addition is "one affecting the method of proof only? (B xxxix). In spite of its endorsement by H. J. Paton (Kant's Mvielaphysies (4 Experience [London: George Allen and Unwin, 1936], 2:383), however, this can hardly settle the issue of whether or not the second edition Refutation represents a break with the ontology of spatial objects adopted in the first, since Kant's well- atteste(1 desire to appear consistent coul( easily have led himl to use the rhetorical context ofc a prreflce to attempt to persuade his reader (or, fbOr that matter, himself) that there had been no change in his view when in filct there hadl been. And so debate has long continued to rage about what Kant intended, with such figures as Vaihinger, Caird, Adalmson (cf. Kemlp Smith, Caotinmentaty, p. 314) and Glerhard Lehmann (in "Kants Wider- legung (les Idealismus," in Ieitriige zun Ge1shieh/e nnd Inlt pee/a/ion derI P/n/- sophie Kani's [Berlin: Walter de Gruyter, 1969], pp. 171-187) joining Paton in the majority position that the 1787 Refutation is intended to be con- sistent with the ontological reductionism of 178 1, and only a scorned mi- nority party of such figures as Benno Erdmnarin, Henry Sidgwick, Arthur Balfour (Kemp Smith, loc. (it.) and H. A. Prichard (see Paton, 2:378-380) defending the view that the argument of 1787, whether so intended or not, requires a radical departure from the position of 178 1. When the alternative to the position of 1781 is understood to be that in knowing

331

PAUL GUYER

It is also unclear precisely what argument or arguments Kant thought could be used to fulfill the objective of' the Reftutation. Its text simply assumes that something permanent is required for the determination of' time and proffers only the most perfunctory ar- gument that such an object cannot be the self' or anything in it. Even if' we look to what is apparently Kant's only other argument flor the connection between permanence and time-determination, namely the first of' the three "Analogies of' Experience," we still find nothing which makes it clear why determining anything about the temporal structure of' self-consciousness should require objects which are either phenomenologically spatial or "external in the tran- scendental sense." The first Analogy appears to contain at least three distinct arguments for the thesis that time-determination re- quires something permanent, but two of' these are unpersuasive and none of' them seems to contain anything entailing that the permanent object required must be either spatial or numerically distinct from the self'.

Kant's first argument5 is that since time itself "remains and does not change" or is the "permanent f'brn-m of' inner intuition,' but also "cannot by itself' be perceived," its permanence must be repre- sented by something permanent in perception which is its "sub- stratum," and that this substratum must be "substance in ap- pearance" (B 224-5). But the assunmption on which this argument turns, that the representation of' something permanent must itself' be permanent, seems to be ungrounded and indeed to fly in the face of' Kant's own claim that "the representation of' something

things in space we knowv things (as they are itn themselves, the reason for the scorn of the minority party will be obvious; as will become clear later, however, my view is that this view of the alternatives conflates ontological and epistemological conceptions of independence (as did Kant himself in1 1781), and thus misses the force of Kant's position in 1787. (Kemp Smith's own position in all of this is unclear. He does indeed think that the argu- ment of 1787 "proves the direct opposite of what is asserted in the first edition, viz. that though outer appearances are immediately apprehended they must be existences distinct from the subjective states through which the mind represents them "[Commientaryv, p. 312], but since he then goes on to interpret this claim in terms of his unique distinction between "subjec- tivism" and "phenomenalism" it is not evident whether he really has oni- tological distinctness in mind here.)

5First in order of exposition, that is; it was actually written only for the second edition.

332

KANT'S INTIC NTIONS IN THE REFUTATION OF IDEALISM

permanent is not the same as permanent representation" (B xli);'5 and in any case there is no obvious reason why such a substratum, even if required, would have to be a substance which is either spatial in form or ontologically independent from the self. Why should an enduring empirical self, for instance, not be an adequate sub- stratum for the permanence of time?

The second argument of the first Analogy (which Kant does not explicitly distinguish from that just considered) is that the two spe- cit'ic temporal "modes" (cf. A 177/B 219) of'succession and coexis- tence are themselves modes, not of time itself, but of enduring or permanent objects in time, so that if either of these modes of tem- poral relation is ever to be instantiated, there must also be objects to which the property of duration can be ascribed. As he puts it, "we require an underlying ground which exists at all times, that is, some- thing abiditig and permanent, of which all change and coexistence are only so many ways (modes of time) in which the permanent exists" (A 182/B 225-226). But this argument seems to beg the question why could there not be just simultaneous or successive bundles of states of affairs (such as Hume's perceptions),7 without anything in which these inhere except enduring time itself? And even if' it is necessary to posit an enduring object so that simul- taneity and succession can be exemplified, why cannot such an object be the self rather than anything spatial, let alone on- tologically independent of the self?

However, Kant may have intended yet a third argument, a more obscure but apparently epistemological argument that alterations, that is, actual transitions from one objective state of affairs to an- other rather than changes in which two in fact merely coexistent states are being perceived, "can be empirically known only as changing determinations of that which endures" (A 188/B 231). What Kant seems to mean here is that there is Ino way of dis- tinguishing between two successive perceptions which represent two successive states of affairs and two which represent merely a change in which of two coexistent states are being perceived except

'As this comment is part of the second edition Pref'ace's explanation of the Reftutation, there could then even be inconsistency between the Re- futation and the first Analogy on which it is usually thought to depend.

7See A Treatilse of Huma Nature, Bk. I, Pt. IV, Secs. v, vi (in the edition of L.A. Selby-Bigge [Oxford: Clarendon Press, 1888], pp. 233-234, 252).

333

PAUL GUYER

by invoking in the first case an object which has a history of change such that the two perceptions can only represent successive rather than coexistent states of it, and which of course must endure in order to have such a history.8 Yet even on such a favorable recon- struction of Kant's intention there is once again no obvious reason why such an object cannot 1)e an enduring self rather than anything external to the self.

Thus, neither the published Refutation of Idealism nor the first Analogy of Experience makes clear either the position which Kant meant this Refutation to assert or the tactics by which he intended to defend it. And Kant must have recognized almost immediately that his new Refutation failed in at least the latter task, for even before the revisions of' 1787 were ready for the press Kant added two enmendationls to his new Refutation which attempt to provide additional support for its conclusion. Again, however, neither of these passages succeeds in putting beyond doubt the intended tac- tics of the Refutation.

A long footnote added to the second edition Preface contains the first of' these emendations (at least in order of appearance). The initial objective of this note is to improve upon the blunt assertion of' the main Refutation that the enduring object required for the time-determnination even of self'-consciousness "cannot... be something in me, since it is only through this permanent thing that my existence in time can itself be determined" (B 275). For this line, Kant tells us, the following should be substituted:

13iii this petrmaent can no(lt be an ituitiou in me. or0 all grounds o/determiua- lion of my existence which are to be set with in me alre representations; and as rePresentatiotts themselves require a permTha nent distinct Jom them, il ida/ion to which their change, and so my existenlce in the lime wherein i/wy (change, may be determined (B xxxix).

Moreover, at the end of the note, Kant remarks that all of our representations, "not excluding that of matter" itself, are "very transitory and variable" (B xli). These remarks can be taken to add two premises to the argument of the Refutation. First, they lay down the premise that all representation regarded as such in Cartesian terms, in their formal rather than objective reality are to be regarded as themselves enjoying only momentary rather than

8Stich an interpretation is suggested by Arthur Melnick in Kant's Analo-

xiprs ahXperw'ce (Chicago: University of Chicago Press, 1973), pp. 71-77.

334

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

enduring existence' Second, they suggest that the empirical self is presented only by means of such transitory repesentations, or that determinations about its temporal structure can be based only on determinations about their temporal relations, and so is not itself available as a permanent object of' perception by reference to which the temporal features of its own representational states may be made determinate. These premises do seem to imply that i/some- thing permanent is needed in order to make some indispensable determination about the temporal relation of transitory represen- tations and the empirical self' of' which these are our only means of acquaintance, then this permanent thing must indeed be something other than any representation and at least the empirical self. This would appear to require knowledge of something ontologically in- dependent of the self, if' not eo ipso spatial.

Unfortunately, this note still does not explain why something permanent is required for this purpose. It does indeed improve upon the main Refutation and some of the first Analogy by sug- gesting what temporal feature of representations a permanent thing is supposed to be necessary to determine, namely, their change, or their succession one upon another. But it remains un- clear why anything more than mere acquaintance with representa- tions which in fact succeed one another in otherwise uninterpreted experience'I( should be necessary for one to judge that there has

9Kant's exposition may suggest this is simply a brute but contingent tact about representations. It is not obvious that this would ultimately under- mine the force of any argument employing this premise; however, D. P. Dryer's rather obscure remark that, "The absence of continuance in what immediately presents itself to introspection is bound up with what differ- entiates introspection from outer observation," (Kant'S SolutionJor Verifica- lion in Metaphysics [London: GCeorge Allen and Unwin, 1966], p. 364) might be taken to suggest that the momentary existence of a representation is not a contingent feature of an entity which could conceivably last longer than a moment but rather a necessary feature of what should really be under- stood not as an entity at all but rather as an act of alten/iott with the brief' duration such an act must have.

i)In the sense of the "raw material of the sensible impressions" which is invoked in the first occurrence of' the word Erfa hrutng at B 1. See Lewis White Beck, "Did the Sage of K(ingisberg Have No Dreamis?" in Beck, Essays oat Kant and Hume (New Haven, 1978), pp. 38-60. The contrast between the raw materials for judgments and fully determined judgments is, as we shall see below, crucial to Kant's ultimate refutation of idealism, although it is obscured by Kant's ambiguous use of more terms than just Er/ahrung.

335

PAUL GUYER

been such a succession, if' indeed this is what Kant intends. Indeed, Kant's second Analogy of' Experience is ordinarily understood to assume precisely that one is immediately aware "that the manifold of appearances is always generated in the mind successively" (A 190/B 235), and to argue that the interpretation of' one's own rep- resentations by reference to objects external to them is necessary only to determine whether a succession of' representations also rep- resents succession rather than coexistence in the states of such ob- jects. But then why should Kant think that an enduring object, even if' necessarily distinct from any representation, is required in order to determine that there is change in mere representations?

This note to the second edition Preface does not even mention the spatiality of' the external objects alleged to be necessary for time-determination. To complete the argument for spatiality is ap- parently the purpose of Kant's other emendation of' the Refuta- tion, the "General Note on the System of the Principles" (B 288-294) added at the end of the chapter into which that Ref'uta- tion had earlier been inserted. Here, Kant again makes two claims.

First, accepting the conclusion of' the first Analogy that time- determination requires something permanent, he argues that to perceive such an object "we require an intuition in space (of' mat- ter)," because "space alone is determined as permanent, while time, and therefore everything that is in inner sense, is in constant flux" (B 291). This assertion is surprising, however, because Kant else- where suggests that time is permanent even if' not itself perceivable, and because the endurance of space hardly seems entailed by Kant's customary and Leibnizian definition o(f space as the form ct intui- tion by which two or more coexistent objects may be presented (cf. A 23/B 38).''

' IFor the Leibnizian definition of space purely in terms of coexistence, see, e.g., his "Third Paper" in reply to Clarke, Sec. 4: "I hold space to be something merely relative, as time is: ... I hold it to be an order of coexistences, as time is an order of successions. For space denotes, in terms of possibility, an order of things which exist at the same time, considered as existing together, without enquiring into their manner of existing" (The Leibniz-Clarke Correspondence, Ed. H. G. Alexander [Manchester: Manches- ter University Press, 1956], pp. 25-26). See also Alexander Gottlieb Baumgarten, Metaphlysik, tr. GF. F. Meier (Halle: Carl Hermann Hem- merde, 1766), Sec. 160.

336

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

Second, Kant claims that the perception of' space is a necessary condition of' the perception of' alteration itself'. He argues that rea- son alone could not comprehend the "combination of' contradic- torily opposed determinations in the existence of' one and the same thing" (B 291) but instead requires an intuition as an example of' how such "opposed determinations" can constitute a change rather than a contradiction, and he then asserts that

The intuition required is the intuition of' the movement of' a point ill space. The presence of' the point in different locations (as a sequence of' opposite determinations) is what alone first yields to Us al intuition of' alteration. For in order that we may afterwards make inner altera- tions likewise thinkable, we must represent time (the fbrm of' inner- sense) figuratively as a line, and the inner alteration through the drawing of' this line (motion), and so in this manner by means of' outer intuition make comprehensible the successive existence of' ourselves in differentt states (B 292).

But again these claims seem peculiar. First, Kant's claim that it is the motion of' an object in space which gives us the idea of' altera- tion is unargued, as is his further claim (in the second edition Transcendental Deduction and in the second note to the main Refutation itself) that it is only by observing regular motions in space ("for instance, the motion of the sun relatively to objects on the earth" [B 279]) that we can do anything like assign determinate magnitudes to "lengths of time" (B 156). If' true at all, these claims would thus seem empirical and contingent. Second, Kant's present claim seems inconsistent with the suggestion of an earlier passage of the Principles chapter that time is sufficient to avoid the reduc- tion of opposed determinations to an actual contradiction (A 152-153/B 191-192).12 Finally, Kant's claim that the perception of motion in space is necessary even to render thinkable the "successive existence of ourselves in different states," which is to say, any suc- cession in representations at all, seems to contradict not only the already mentioned assumption of' the second Analogy that rule-

'2Kant also adds this point to the second edition of the Transcendental Aesthetic under the rubric of "The Transcendental Exposition of the Concept of Time" (B 48-49), thereby suggesting that it is so fundamental as to be virtually self-sufficient for the proof that time is an a priorti form of intuition.

337

PAUL GUYER

governed objects are necessary only to judge objective successions but also con1fln)lo sense: how can we become conscious of motions in outer objects without at least simultaneously, if' not earlier, being aware of' changes in our own representational states as such?

Thus, Kant's attempts to improve upon the exposition of' the Refutation of' Idealism within the revised Critique of' 1787 do not throw much additional light on what tactics he might really have intended that argument to employ; and they do not even attempt to clarify the conclusion he means it to attain. However, there is evidence which can clarify both the objective Kant wanted to achieve in his new attack on idealismn and the tactics by which he thought he could attain his goal. In fact, Kant's dissatisfaction with the new Ref'utation of' 1 787 did not result in only the two additional passages just discussed; rather, Kant continued to work at the re- futation of idealism f'or at least half a dozen years after 1787, and his surviving literary remains (Nacidass) include at least ten lengthy fragments from the years 1788 to 1793 which, in view of their titles, form, and content, could only have been intended as improved versions of' the argument first attempted in 1787 and thus as efforts to fulfill the intentions of' that argument. These notes

i'-Most, of these friaginents were first published in Rudolf Reicke, Lowe Bitter aas. Kaid'i Nachlaso. (K6nigysbel-g, 1889). I have used the transsclip- tions anc dealing of them provided by Erich Adickes in Volumlle 18 of the Akadenile edition (first published in 1 928), where they are as follows: Rf(eX- otinent 5653 and f5654 (Ak. 18:305-313), written in the period after 13 October 1788; R 5709 (p. 332), dating fromt some less determninate time in the late 1 780's; R 63 11-6317 (pp. 606-629), written in conjunction with the visit of Kant's dlisciple 1. (2. E. Kiesewetter to K6nigsberg in the Ftll of 1790, canld R 6323 (pp. 641 -644), fromt the period of April to August 1 793. (Another catalogue of these fragments may be found in Lehmaunn, "Kants Widerlegunlg dces Idealismus,' p. 176.) That these notes represent a con- tinuous development of the intentions of 1787 cannot of course be estab- lished by any single piece of verbal evidence; nevertheless the similarities between the titles Kant gave to them and the title of the of ticial Refutation 1)rovide strong prtima ftcie evidence for such a conclusion. Some of these titles are: Gegeie den (Matlerialen) Idealisui (R 5653); Wieder dent Idealil'sm1 (R 5654); Widerlegung des Problemitatischele Idealismnos (R 63 1 1); Wieder- deni Ideal- imsxl (R 6313); uber [sic] dett Idealism (R 6314); 7ya.m Idealilsm (R 6315); Wieder dent Idealilsn (R 6316); and (a subheading) Idealli'smt (R 6323).

These materials have received virtually no discLussionI in English- lanLguage Kant commentary. In his initial catalogue of Kant's variOus r guL- mrents against idealism, Kemp Smith refers, without any bibliogrlaphical

3 3 8

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

reveal a number of' important points about the intended aim and tactics of' the campaign against idealism that Kant had begun in 1787. First, both by their arguments and by certain specific turns of' phrase they strongly suggest that in spite of' the ambiguity he ana- lyzed in 1781 Kant did intend his refutation to establish our knowl- edge of' objects ontologically independent of' and not just phe- nomenologically external to ourselves. They thus provide evidence that what Kant was trying to do in 1787 was to break with the reductionism of' the fourth Paralogism of' 1781, in spite of' his own claim merely to be providing a "new method" of' proving it (see note 4 above). Second, they throw light on the ways in which Kant thought such knowledge could be demonstrated. Specifically, they contain arguments considerably amplifying the assumptions of Kant's two emendations of' 1787: that there is a direct connection between the determination of' tiIme and spatial form, and that there is a direct connection between the determination of' time and the positing of' enduring objects other than the self'. Although neither of' these is complete in itself', both instead requiring supplementa- tion by a premise that must be elicited from the Transcendental Deduction of' the Categories, at least one of' them is a significant improvement upon anything published in 1787. In addition, these fragments contain passages, unparalleled in the published Ref'uta- tion, which are explicitly designed to show that these arguments

citation, to one of the "Seven Small atpes-S which originatedi in Kant's cliscus- siOI with Kieswetter -apparently a reference to R 63 11, which according to Adickes is in the halnld of Kieswetter rather than Kant and which was availalble in several nineteenth-century editions of Kant-but he then con- cludes his chapter on the refutation without either quotillg or dliscussing this "paper." Recently, Michael Washburn has referred to these docu- ments in "The Second Edition of the Critique: Toward an Understanding of its Nature and Genesis," Kant-Slndien 66 (1975), 277-290, but he as- cribes them to the period belvtween the two editions of the Csitique (pp. 279, 283) withouIt providing Dany reason for departing from Adickes's dating, and has little to say about their content. Beck makes one reference to R 6311 in "Did the Sage of Ktinigsberg Have No Dreanms?" p. 48. The situation in German commentary is only slightly better: these fra-gments receive more extended (discussion in Lehmann, "Kants Widerlegung des Idealismus," and in Wolfgang MLIller-Luter, "Kants Widerlegung ties Ma- terialen Idealismus," Archizvfiir Geschie(hte det Philsalpiel 46 (1964), 60-82, but neither of these authors arrives at any satisfactory interpretation of them.

339

PAUL GUYER

about time-determination do indeed have the ontological and not just phenomenological consequences Kant intends; the one addi- tional argument, however, which Kant deploys at this stage is unfor- tunate but also unnecessary. Finally, these fragments attempt something else which was left undone in the published Refutation: at least some of them explicitly attempt to show how an argument for the thesis that we can know that there are objects independent of ourselves, a position which Kant himself calls "dualism" (R 5653, p. 310) as well as "realism" (R 6315, p. 620), can be reconciled both with his original analysis that skeptical idealism is due to the in- conclusiveness of any causal inference to external objects and with transcendental idealism, understood of' course not as the thesis that outer objects are representations but as the thesis that space and time are only forms of appearance. In the subsequent sections of' this paper, I will expound these points.

II

Some of the notes in Kant's Nachlass clearly imply that the repre- sentation of an object in space is the representation of something numerically distinct from the self' without employing any explicitly ontological terms of art. Thus this passage from the first new re- futation of 1788 emphasizes that treating the self as itself an object of' knowledge requires relating it to something which is repre- sented as outside the self because it is represented as other than the self:

That if I make myself into an object space is not in me but (yet) is in the formal subjective condition of the empirical consciousness of my- self', that is time, proves that something outside of' me, that is, some- thing which I must represent in a different manner [aif eine andere Art] than myself', is connected with the empirical consciousness of, myself', and the latter (is) at the same time consciousness of an external relation, without which I could not empirically determine my own existence (R 5653, p. 309).

The nonreductionist implications of this passage lie in the fact that an object in space is said to be represented in a different way than the self is represented. This is hardly a natural thing to say of something which is merely a state of the self. Further, such an

34()

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

object is said to stand in an external relation to the self', which is again not a natural thing to say of something which is nothing but a modification of the self (especially for someone writing against the background of the Leibnizian denial of' extrinsic relations). ' The same conclusion seems to be implied a few lines later when Kant says of' space itself', rather than of- an object in space, that it "is not a representation related to the self (as object)," that is, it does not represent the self-, "but is rather immediately related to something as existing distinct from the subject" and is thus "the consciousness of the object as a thing outside me" (loc. cit.). Here the phrase "a thing outside me" is virtually glossed as something which is numer- ically distinct from me.

Other passages emphasize a distinction between an external ob- ject and our representation of' it. This one, from Kant's second fragment from 1788, is not very clear because it tends to confuse the objective content intended by a thought of' something indepen- dent of the self with the mental status of the thought itself', thereby undermining the intended independence: 1 5

... every object signifies something distinct from the representation, which however is only in the understanding; thus the inner sense itself, which makes ourself an object of our own representation, is related to something distinct [etwas . . . verschiedenes] from ourself (as transcendental object of apperception). If we did not therefore relate our representations to something distinct from ourself, they would never yield knowledge of objects (R 5654, p. 312).

I4Lehmann takes precisely this passage to reveal that Kant's Ultimate refutation is still a "Phinomnenologic des RealitatsbewLusstseins . . . die mit einem Beweis der 'Existenz' ausserer Dinge gar nichts ZU tLun hat" ("Kants Widerlegung des Idealismus," p. 183). This conclusion would appear to rest on the assumption that just by virtue of standing in relation to my mind something must be a modification of it rather than an independent existence; but this is belied by Kant's most basic assumption that the very notion of an appearance-a relation in which something stands to me- implies the independent existence of something which appears ("other- wise we should be landed in the absurd conclusion that there can be appearance without anything that appears"; [B xxvi-xxvii; cf. also A 249: "For if the senses represent to us something merely as it appears, this something must also in itself be a thing .

l The absence of language for a clear distinction between intention and intentional object plagues Kant's argUment elsewhere as well; see the con- clusion of Section VII below.

341

PA UL GUYER

Much clearer, however, is the distinction between a representation and its object invoked in this explicit endorsement of' realism from 1790:

My representations cannot be outside ile and an external object of rep resentations cannot be in me, for that would 1)e a contradiction. It could well be, however, that although the representation is in me its ol)ject is yet without contradiction outside me, or else that the relpre- sentation together with its object is in mie. On idealism it is asserted that it is not possible to decide that the object of a representation is not in me along with its representation, even when the latter is repre- sente(I (in intuition) as existing outside me.-The realist, on the con- trarv, asserts of outer intuition that this is possible, aindc indeed cor- rectly . . . (R 6315, p. 620).

The position with which Kant now means to align himself', in other words, holds precisely that we can know that in addition to our representations objects numerically distinct fIrom them exist, al- though this was just what was denied in 1781.

The most persuasive passages ot' all, however, are those in which Kant uses his own ontological terminology to assert explicitly that what we know when we know that something exists outside us is that something other than our representations exists in itself Thus, a different fragment from 179() maintains that the ground of' the possibility of' a temporal determination (in this case that of' simul- taneity) "lies in the relation of' representations to something outside uS, and, indeed, to something which is not inl turn merely inner representation, that is, lormn of' appearance, thus, to something which is something in itself' [sache an sic/i]' (R 6312, p. 612). And Kant's last effort at the refutation contains this explicit rejection of, the reduction of' that which serves as the condition of' time-deter- mination to any representation or succession of representations:

Ihe imlpossil)ility of' determnining [our-] existence in the succession of' timee through the succession of representations inl US, and yet the actuality of' this determination of' [our] existence, [requires] an ininie- (iate consciousness of' something outsi(le ine, which corresponds to these representations,* and this intuition cannot be mrne illusion [Schiei]..

*(and which does not exist merely in my replresei-tation (rather (as thing) in itself) . . .(R 6323, p. 643)).

342

KANVT'S INTENTJONS IN THE REFUTATION OF IDEALISM

If' we read this to contrast "something outside me" with any succes- .siol of' my representations, and take the latter to include possible as well as actual successions of' representations, then this passage pre- cludes the phenomenalist reduction of' the contrast between object and representation in general to that between possible sequences of' perceptions and particular perceptions in the style of' Berkeley or Mill. Kant's new method of' proving outside objects as the condition of' time-determination thus seems intended to secure knowledge of' things ontologically independent of' ourselves.

Just how such an at least relatively robust form of' realism can be reconciled with the retention of' any form of transcendental ideal- ism is a problem on which Kant's Nachlass also throws light, but to which we will return only later; the next item on our agenda is the examination of the tactics by which Kant finally thought he could achieve this strong result.

III

In the first of' the three notes which follow the 1787 Ref'utation Kant suggests that the underlying premise of the argument is that inner experience" or "empirical knowledge" of' the self' "not in-

dleed the consciousness of my own existence, but the determination of' it in time"-requires intuition as well as "the representation 'I am ," or the mere "thought of' something existing" (B 277). But this condition taken by itself', which is merely Kant's requirement that empirical judgment involves intuition as well as a concept, would seem to be satisfiable by inner intuition alone, and so is not itself sufficient to generate an inference to anything independent of' the self. One of' Kant's fragments fromt 1790, however, employs a more complex and suggestive division of' the ways in which the self is presented. Though the rest of' Kant's notes fIollow the text of 1787 in making explicit only a two-fIold distinction between "the tran- scendental consciousness of' the self" and "empirical consciousness of myself' (e.g., R 5653, p. 306),'13 this note implies that the refuta- tion of' idealism actually requires a threefold distinction between:

'iAs Beck points out, the phrase "transcendental consciousness for the mere concept of self-existence is introduced only in Kant's Nacitlass (his reference is to R 63 11, p. 61(); see "D)id the Sage of K6 nigsberg Have No Dreams?" 1). 48). He does not discuss the passage I am about to quote, however, which Could only have buttressed his argument in that paper.

343

PAUL GUYER

[1] The transcendental consciousness of my existence in general. (2) My existence in time, thus only in relation to my own representations, insofar as I determine myself through them. This is the empirical consciousness of myself. (3) The knowledge of myself as a being deter- mined in time. This is empirical knowledge (R 6313, p. 617).'7

The distinction of "empirical consciousness" from both the mere concept of self-existence and any actual knowledge of the self's histo- ry in time suggests that there must be a difference between the mere succession of representations which constitutes the raw mate- rial of' consciousness, on the one hand, and, on the other hand, the recognition of the temporal relations of' such representations which constitutes empirical knowledge of the history of the self consisting of such a succession of' empirical consciousness.18 This in turn would suggest that knowledge of' outer objects is the condition by which the transition is made from undetermined empirical con- sciousness to empirical self-knowledge, from the mere occurrence of successive representations to the recognition of their succession and its history.

With his usual terminological fluidity' Kant makes the same distinction in another passage in the same fragment by contrasting the mere "form of' representation in time" with "empirical con- sciousness," which here means precisely what the paragraph just quoted called "empirical knowledge." Here Kant claims that idealism

I7One of' the f'ew commentators who has noticed this passage is Pierre Lachieze-Rey, L'idWalisme Kantien, 2nd ed. (Paris: J. Vrin, 1950), pp. 64 ff.

18 Gerhard Lehmann, who has also discussed this fragment ("Kants Widerlegung des Idealismus," p. 179), sees it as an anticipation of the doctrine of' ErscheinungsiufI ("levels of appearance") which Kant was at- tempting to develop in the Opus Postumum. If' we think of' an "appearance" as the object of' an empirical intuition, however (A 201/B 34)-"undeter- mined," to be sure, insofar as it is thought of' only in connection with empirical intuition, but determinate insofar as concepts and judgment are brought to it-there would seem to be no need to think of' two kinds or stages of appearance as being involved here. "Empirical consciousness" would simply designate the undetermined intuitions by which the self and its history are presented, and "empirical knowledge" of' the self would designate determinate knowledge of' the "appearance" which is the object of those intuitions.

'9Lachieze-Rey, p. 67n.

344

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

... can be refuted [first] by showing that the representation of exter- nal things must not lie merely in the imagination, but in an outer sense, because thelform of representation in time would make possible no empirical consciousness of one's own existence in time, thus no inner experience, unless supplemented with [the form of representa- tioll] in space (R 6313, p. 613).

If we take the form of' representation in time to consist simply of' the successiveness of' all representations in time (the manifold of' inner sense, of which time is "nothing but the form," "constitutes a series of' one dimension . . . the parts [of' which] are always succes- sive" [A 33/B 49-50]), then this too implies that the mere fact of' the successiveness of representations is not sufficient for empirical knowledge of their relations in time, and that for some reason the representation of space or objects in it is also necessary for this purpose.

But why is the successiveness of consciousness insufficient for its own recognition, and why should spatial, let alone independent objects be necessary for this purpose? Here is where Kant's pub- lished efforts of 1787 left us dissatisfied, but here is where the efforts of the subsequent years do offer illumination. In fact, what Kant tries to do in these notes is to improve upon the two alterna- tive tactics unsatisfactorily attempted in the two emendations to the Refutation already published in 1787. On the one hand, Kant at- tempts to improve upon the bald assertion that the representation of time and of alteration in it requires the representation of space and motion. He does this by developing an actual argument that the determination of temporal relations, even among purely sub- jective states, requires the representation of space, and by then supplying further grounds for inferring from the representation of space to the existence of enduring objects ontologically indepen- dent of the self. On the other hand, Kant also attempts to improve upon the bare suggestion that the change of representations them- selves requires representation of something permanent by devel- oping an argument that goes directly from time-determination to enduring objects, and which only subsequently adds that such ob- jects must be represented spatially.

Both these tactics represent epistemological analyses of the con- ditions under which judgments of time-determination can be

345

PAUL GUYER

made. Some passages in Kant's notes, however, suggest that such epistemological arguments are not sufLfi(ient to imply the ontologi- cal conclusion ultimately intended by his refutation of' idealism. This may not be immediately apparent when Kant writes that the type of'argument called for by the first paragraph of' R 6313 (just quoted) must be followed by a second step, in which it is shown

... that the ma/l/er of representations in space would not possibly oc(cur in the mind Without an Outer sense. For the imagination can create a representation of the external only if it [is] afffect[ed by] the Outer sense . . . and there would be no material for external representations in the imagination if there were not an outer sense (p. 61 3).

But as we shall later see, what this actually describes is a metaphysi- cal argument assuming premises not just about the conditions un- der which representations may be judged but also about the sub- stances which can produce them. Such an argument is deeply problematic; however, as we shall also see, it is not in fact required by both of' the tactics by means of' which Kant attempts to execute his ultimate refutation of' idealism. It is not required by the argu- ment which goes directly f'romn time-determination to the necessity of' enduring objects, and only thence to their spatiality; and in fact it is not even required by all forms of the argument which goes directly from time to space, but only by what is in fact an in- complete form of' this argument which does not go on to show that the existence of' enduring objects is in fact n epistemological condi- tion of' the perception of' space. Rather, all that is required in these cases is a review of' the argument already offered to remind us that an ontological commitment to objects other than the self' is a condi- tion of' the possibility of' time-determination. Thus, the problems that arise when Kant adds a second stage to the refutation need not undermine it.

IV

We saw earlier that neither the Reftutation of' 1787 nor the first Analogy explicitly contains any ground for the conclusion that an enduring object must be spatial, and that Kant's attempt to reach this further conclusion in the second note to the Reftutation (B 277-278) and the "General Note" rests on the undefended asser-

346

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

tion that the empirical intuition of' motion in space is necessary to measure or even to represent succession in time. One of' Kant's tactics in the subsequent years was clearly to attempt a direct argu- ment to the necessity of' space ftr time-determination which would remove the arbitrariness of' these earlier suggestions and indeed obviate the difficulty of' having first to establish the connection between time-determination and endurance.

The first of' Kant's fragments from 1788 begins with the simple assertion that change can be recognized only by comparison to something which endures, and then infers that this something must be spatial both by a simple argument by exclusion as well as by the additional assertion that space is intrinsically permanent. Thus, Kant first writes:

For in space alone do we posit that which endures, in time there is unceasing change. Now, however, the determination of the existence of a thing in time, that is in such a change, is impossible unless its intuition is connected to that which endures. This must therefore be inituited outSide us as the object of outer sense (R 5653, p. 307).

This simply assumes that the perception of' change requires the perception of' something enduring, and since time does not furnish the latter perception it must be space which does. But Kant also attempts to buttress this negative argument by a more direct state- ment, namely:

Permanence is intrinsic to the representation of space, as Newton said. T'he permanence of the forin of our mind is not the same (for the form of'time is just as permanent), rather [the representation of space is] the representation of something permanent, to which we subsume all time-(letermination and therefore represent as permanent, and thus cannot represent as the spontaneity of' self-determination (R 5653, p. 3()8).

None of' this improves upon the ungrounded asse-tiofls of' 1787. However, Kant's next paragraph intimates a possible argument:

TYhe representation of space is the ground of time-determination on account of permanence (insofar as only in space can one acquire a representation of time as a magnitude through a line which I draw [even] while I am Conscious of' my synthesis merely in my subject (R 5653, p. 308-309).

347

PAUL GUYER

This starts by repeating that it is permanence which is the premise for the necessity of' space, which would require that there already be an argument for permanence; however, what follows then sug- gests that there is some connection, in which permanence plays no direct role, between space and the synthesis of' a representation of' the duration of' time. Still, this is not that different from several remarks already made in 1787 (B 154) or even 1781 (A 33/B 50). What new argument is in fact intended by this reference to "synthesis"?

The notes of' 1788 do not explain this, but one from 179() finally suggests what sort of' argument Kant had in mind f'or a direct connection between time and space:

We cannot represent any number to ourselves except through succes- sive enumeration in time and then through collecting this multiplicity together in the unity of a number. This latter however cannot Occur otherwise than by my placing [the units of the multiplicity] beside one another in space; for they must be thought of as given similla'tieo'isly, that is, as collected together in one representation, for otherwise this manifold [Viele] constitutes no magnitude (nluminber); sinmultanleitv, however, cannot possibly be cognized unless if in addition to my act. of placing the manifold together I can apprehend (not merely think) it as given both forwards and backwards. Therefore an intuition must be given in which the manifold items are represented outside of and next to one another, that is, the intuition which makes the representation of space possible must be given (in perception) in order to determine my own existence in time; that is, an existence outside me is the ground of the determination of my own existence, that is, the empiri- cal consciousness of myself. Therefore I must be just as conscious of the existence of outer things as I am of (my) existence in time, though to be sure only as appearances yet as real things (R 6314, p. 616).

Kant's argument appears to be that to determine the magnitude of' any duration of time is to assign a numerical value to this duration, and that such an act of' enumeration, like any other, requires in addition to the successive apprehension of' the individual moments of time a single act which we can count as their collective represen- tation as well. In the language of' the (first edition) Transcendental Deduction, a synthesis of' reproduction as well as of apprehension is required, for while intuition "does indeed offer a manifold," this "manifold can never be represented as a manifold, and as con-

348

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

tained in a single representation, save in virtue of' such" a further act (A 99).20 Such a single act of' representation is interpreted as the simultaneous representation of' the successive moments of, time themselves, and this is then conceived of' as a spatial display of' these items, because it is only in space that things may be represented as numerically distinct yet available for a form of' perception-one in which the coexistent items may be indifferently perceived in either of' two orders-which provides evidence of' their coexistence. Thus, the ability to represent spatial extension is held to be necessary in order to have a single representation of' even a temporal succession.

Flow could this argument be supposed to justify a further in- f'erence from the necessity of spatial objects to the existence of' enduring objects, which have not so f'ar been mentioned explicitly? One possibility would be that which appears to have hovered be- fore Kant in our previous quotation from R 5653, namely that spatiality is necessary for a single representation of temporal suc- cession because it is the only way in which coexistence can be repre- sented, and that endurance is then implied simply because space itself', "as Newton said," is permanent. Yet as we have seen, this just rests the argument on an assumption not otherwise entailed by Kant's more Leibnizian conception of' space. However, an alterna- tive way of' continuing the argument may already be implicit in the quotation just considered, and is explicitly suggested in two other notes fromn 1790. On this alternative, once the epistemological ne- cessity of' space as the form of' a simultaneous representation of' a manifold has been shown, it is then argued that the existence of, enduring objects is itself' the epistemological condition of' the recogni- tion of' such simultaneity.

That there is an epistemological necessity that objects to be rec-

20This quote is actually drawn from the text of "The Synthesis of Ap- prehension in Intuition" (A 98-100) and not that of "The Synthesis of Reproduction in Imagination" (A 100-102). However, most of the latter section is occupied with a fallacious argument that there must be an a priori ground for the (absolute) necessity of the synthesis of reproduction rather than with the nonfallacious argument that such a synthesis is (relatively) necessary for knowledge to occur (see my "Kant's Tactics in the Transcen- dental Deduction," Philosophical Topics 12 [Fall, 1981] 157-199, Section IB (I))

349

IA UL GUYER

ognized as having spatial form must be enduring and therefore not just mere representations is first argued in R 6312:

How do we recognize the simultaneity of things, since all OuI repre- sentatiOInS Succeed one another in apprehension? By means of the fict that we can represent the maniflid in such a case both forwards and backwards. Now since in inner sense everything is successive, thus nothing can be taken backwards, the ground of the possibility of the latter must lie in the relation of' representations to something outside us, and indeed to something which is not in turn just inner representa- tion, that is, form of appearance, but which is something in it- self . . -Thus the representation of something which endures must pertain to that which contains the ground of time-determination, but not in respect of succession, for in that there is no permanence; conse- quently only in that which is simultaneous must that which endures lie, or in the intelligible2 l which contains the grou-Ind of appearance (p. 612).

Spatiality is necessary to represent simultaneity, itself necessary for a single act of reproduction of representations, but the recognition of simultaneity is in fact possible only in the case of objects which endure at least long enough so that we may obtain perceptions of pairs of them in either of two orders. This is made even clearer in a passage (which is in fact part of a larger argument) in Kant's next fragment:

The simultaneity of A and B cannot even be represented without something which endures, f'or all apprehension is really successive. But insofar as the succession can take place not only forwarl-s fromn A to B but also (as often as I want) backwards from B to A, it is necessary that A endure ['oridaurel. The sense-representations A and B must therefore have a ground other than in inner sense, but yet in some sense, therefore in outer sense; therefore there ImuLrst be objects of outer sense (and as far as dreaming is concerned, this object, which causes the illusion of the presence of several outer objects, is the body itself) (R 6313, p. 614).

2 1 "Intelligible" here woLld not seem to carry its usual connotation of accessible to reason alone, but would instead seem to be another clumsy way of referring to an intentional object. However, Kant's use of this term may be an expression in his traditional terminology of' his view in 1 787 and after that, the enduring object of time-determinationl must in fact be con- ceived of as ontologically independent of the self'.

350

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

In order for representations of' two objects, A and B, to be deter- mined to be representations of' coexistent and thus necessarily spa- tial22 objects rather than just successive states of' affairs, one if' not both of' A and B must be supposed to endure long enough to allow the perception of' it in two different sequences. This argument is not sufficient, nor, as Kant's use of' the word foridaure may suggest, is it intended to imply permanence in the sense of' the eternal existence of' any particular object. But it does seem to imply that objects actually perceived to be spatially related must endure longer than a single representation does, and this is enough to imply that they cannot be equivalent to representations, which are fleeting by nature. So the determination of' time requires the representation of space and this in turn requires the positing of' objects which endure and are therefore distinct fromt any representation of' them.

Of course, there are serious problems with this argument. (1) First, it appears to depend on the assumption that we must be able to assign a numerical magnitude to the duration of our own experi- ence, yet there seems to be no obvious reason why a "skeptical idealist" must concede such an ability. However, Kant's argument does not really assume any very refined ability to measure time, but simply the ability to distinguish between a single present experi- ence and a multitude of' experiences, past and present;23 and even prior to his proof of' the divine guarantee of' external reality the Cartesian "skeptic," for instance, does assume that he can know that "I now exist and further recollect that I have in former times existed, and . . remember that I have various thoughts of which I can recognize the number. "24

(2) Second, however, the idea that there is any necessary connec- tion at all between the ability to reproduce representations of' counted items and the spatial representation of them still seems problematic. On the one hand, such a connection seems inade- quately defended: when in the Transcendental Aesthetic Kant ii-

22Necessary given o0-r epistemic limits, that is. ') ht is, unlike his remarks at B 156 Kant's present argument aIssu-Imes

only topological and not metrical alilities with respect to time. This way of pLtting my point was suggested to me by Chris Swoyer.

91Meditation III; in The I-hdlosophical WorkN o/ Descarites, edited and trans- latedl by E. S. Haldane and G. R. T. Ross (Cambridge: Cambridge Univer- sity Press, 191 1), 1: 165.

351

PAUL GUYER

lustrates the reproduction of' intuitions in a synthetic operation of addition by reference to such spatial objects as "our five fingers" or "five points" (B 15), we may be inclined to let it pass; but when he asserts that only spatial separation can represent temporal succes- sion, we may wonder why, for instance, different qualitative f'ea- tures of a present representation, such as feelings of' pain or tem- perature, could not be employed to represent past representations. Yet on the other hand Kant's connection may be too strong: would it not make the succession of inner states determinable only by assigning all of them distinct spatial positions, and would this not then mean that we could prove that we know external objects at all by proving that we know only external objects. Yet surely Kant does not want this conclusion; indeed, he specifically asserts that "Not everything which is in time is also in space, e.g. my representations" (R 5653, p. 309).

(3) Finally, and what would seem most damaging, Kant's argu- ment appears to be circular. It holds that a succession of' time can be known only if' spatial coexistence can be recognized, but then adds that spatial coexistence itself' can be recognized only through a series of' representations "backwards and forwards." Yet surely rec- ognition that one is having such a series presupposes that one can recognize temporal ordering independently of' spatial relation.

Objection (2) seems insuperable, but Kant's first fbOrm of' argu- ment in his mature refutation may nevertheless point to a premise which could be used in a more promising argument and which could be understood in a way that escapes objection (3). This is the fact that even in the temporal determination of' self-consciousness a synthesis of' reproduction as well as apprehension is required. That is, at any given moment in which knowledge of' a succession even of mere representations is claimed it must not merely have been the case that these subjective states successively occurred but also be the case that there is something which presently counts as the represen- tation of them. For "if' I were always to drop out of' thought the preceding representations (the first parts of the line, the anteced- ent parts of' the time period, or the units in the order repre- sented) . . . a complete representation would never be obtained" (A 102). To be sure, Kant does not explicitly mention this premise in the notes we have been discussing, and in originally introducing it in the Transcendental Deduction what he was concerned with was the case in which we must recognize the identity of' something over

352

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

time given the diversity of our successive representations of it. What he argued there was that such reproduction is necessary because "if we were not conscious that what we think"-the object of our thought, that is-"is the same as what we thought a moment be- fore, all reproduction in the series of representation would be use- less" (A 103). But the reason Kant adduced for this claim-that without a concept of the identity it contains such a reproduction of previous representations "would in its present state be a new repre- sentation which would not in any way belong to the act whereby it was to be gradually generated"-is in fact relevant to the present argument as well, for on Kant's own conception of representation a present one always is a new representation which cannot be linked to past representations, to represent either their diversity or their continuity, without some interpretation. In fact, such a condition on the recognition that a succession of representation is a succes- sion is a consequence of Kant's most fundamental assumption con- cerning any recognition of the unity of a manifold:

Every intuition contains in itself a manifold which can be represented as manifold only in so far as the mind distinguishes the time in the sequence of one impression upon another; for each representation, in so far as it is contained in a single moment, can never be anything but absolute unity (A 99).

In its merely formal reality, that is, a single representation signifies (or logically implies) nothing beyond itself. But Kant's premise that all representations are fleeting and transitory means that in fact no more than one representation ever is present to us; the manifold of successive repesentations is not in fact before one's mind at the moment of its recollection in the way in which a dozen eggs can be before one's eyes. And what this all means is that for us to have any knowledge of even a subjective succession of representation-a manifold even apart from any objective significance it may have- some form of interpretation by (extra-logical) rules or concepts must be placed on our present representation in order to allow it to repre- sent such a multiplicity.25 Without such an interpretation the pre-

25That rules or concepts are required for this purpose means that the boundary between "the synthesis of reproduction in imagination" and "the synthesis of recognition in a concept" (A 103) dissolves when Kant's premises are fully understood. I take it that this does not constitute an

353

PA UL GUYER

sent representation is simply a new representation without any connection to a temporally extended-or "gradual"-act of' repre- senting, or a succession of' representations. Yet representation of' such a succession is surely the minimal condition of' any empirical determination of' the self' and its history in time.26

It is of' course absurd to take this act of' interpretation as a mental act by which at a given moment temporal order is first gen'rerated out of' some kind of' diversity (or potential diversity) which is befOre the mind in some kind of' nontemporal way. That is, it is absurd to understand Kant's theory as one by which any psychological sense of' the passage of' time is first produced. For even if' the very idea of' somehow transmuting a nontemporal diversity into a temporal suc- cession is not itself' absurd, the alternatives available fOr the tern- poral location of' the requisite act of' interpretation on such a con- ception of' it would be: either it must itself' be instantaneous, which is crazy, or not be in time at all, which is also crazy,27 or else must

objection to the present interpretation, however, but is rather jUst a reflec- tion of the "preliminary" (A 98) and. indeed superficial nature of the distinctions among the three fbOrms of synthesis. It is of course parallele(l by the fact that the very distinction between imaginatio an(l a understand- ing tends to collapse in Kant's hands (e.g., A 1 19).

2fiAt this point, a comment On Jonathan Bennett's deeply suggestiVe interpretation of the Transcenldental Deduction is reequirel. Bennett's summary of Kant's argument is that "Self-consciousness, then, involves whatever intellectual capacities may be required for the establishment- sometimes by empirical synthesis-of the truth of statements Clbout one's past mental states. To verify such a statement requires more thaln just a group of criteria for mental identity . . . but all that Kant needs at the present stage of his argument is that it does involve thatt mucLlh" (Katit'.s Atialytic [Cambridge: Cambridge University Press, 1966], p. 1 19). He thenr objects, however, that while criteria may be needed to establish the identity of the self over timne ("identifying-rules"), none are needed to establish its identity at a time ("counting-r-ules"), so that Kant cannot establish the applicability of the categories (or "criteria") in every case in which he wants to (including the case of the present uLnity of the mind) ( 120- 122). Indeed, Bennett takes the thesis of A 99 on which I have been commenting to be a tacit (or grudging) recognition of' this restriction ( 123). But it seems to me that Bennett fails to see that Kant's point is that there is no question of the diversity (and thus identity) of the self's representation at all 1111less tem- pora1l succession is recognized, and thLus that (categorically-based) "i(Ienti- fying-rules" are a necessary condition of any meaningful recognition of' the unity of a subjective manifold at all.

I7Though Kant himself seems to have toyed with such all idea-cf. R 5661 (also known as one of the "Seven Small Kiesewetter Papers" under

354

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

succeed the representation which it interprets-in which case an infinite regress of' such interpretative acts will be required. Ob- viously, the interpretation of' our present manifold must not be thought of' as a psychological event in which a sense of' its succession is first generated. Rather, this interpretation must be thought of' as the kind of' argument-which will often of' course remain unstated- by which our judgigmerts about even subjective temporal order could be confirmed orjustified, for there is no reason why such a process of' justification, if' it is in fact ever consciously undertaken, should not itself' be both extended in time and later than the time of' the particular judgment which it is to confirm.

Conceding this point undercuts objection (3), for as long as the correlation with diverse spatial positions is not thought of' as a psychological event which first generates temporal order, but only as a method by which judgments of' temporal order can be con- firmed, there is no problem in the fact that the perception of' space itself' takes time. However, the problem of' objection (2), that there is no convincing direct connection between the reproduction of' counted units and spatial ftorm, remains. Can Kant circumvent this problem by employing the underlying premise just examined to construct a connection between the representation of' subjective temporal succession and eniduri-g objects, and then only subse- quenitly showing that these objects must be spatial as well?

V

This is precisely what is attempted in the alternative tactic for the refutation which Kant tried out after 1787. A number of' passages attempt to link the perception of' even subjective change or succes- sion of representations with the existence of' an enduring object, but many of these contain no notable advance over the text of'

the title "Is it an experience that we thinkW"), where he argues that a secon(1 time-sequence would be generated if' a judgment that I have had some experience were empirical and in time in the same sense as the experience itself', "which is absurd" (Ak. 18:319). Absurditv arises, howev'- er, only when one thinks of' such a judgment as something that both itself' takes time yet must occur at the same time as the experience which is its subject. If' such a judgment can be thought of' as being made and con- firlmed (if' confirmed) in time but subsequent to the experience which is its own subject, there is no absurdity whatever.

3 5 5

PAUL GUYER

1787.28 However, the same R 6313 on which I have already drawn in several places contains these two paragraphs:

Since the imagination (and its product) is itself only an object of inner sense, the empirical consciousness29 (apprehenisio) of this condition can contain only succession. But this itself cannot be represented except by means of something which endures, with which that which is suc- cessive is simultaneous. This enduring thing, with which that which is successive is simultaneous, that is, space, cannot in turn be a represen- tation of the mere imagination, but must be a representation of sense, for otherwise that which lasts would not be in the sensibility at all.

And then, after the intervening paragraph, already quoted, in which it was argued that the endurance of' simultaneously existing objects is a condition not of' their simultaneity itself' but of' the perception of it, Kant continues:

Since we therefore could not perceive succession in ourselves, and thus could not order [anstellen] any inner experience, if we could not also become empirically conscious of simultaneity, but since this latter is possible only by means of an apprehension ordered both forwards and backwards, which does not occur in the case of the objects of inner sense, thus even inner experience can be thought only by means of the

28For instance, R 631 1: "In our inner sense our existence is determined in time and the representation of' time itself' is therefore presupposed; in time, however, the representation of' change is contained; change presup- poses something enduring, against which it changes and which allows the change to be perceived. Time itself is, to be sure, enduring, but it cannot be perceived by itself', consequently there must be something which en- (lures, against which one can perceive change in time. This enduring thing cannot be our self', f'or precisely as object of' inner sense we are determined through time; that which endures therefore can only be posited [gesetzt] in that which is given through outer sense. Thus the possibility of' inner experience presupposes the reality of' outer sense . . . - (R 6311, p. 611).

But according to Adickes this note is in the hand of' Kiesewetter rather than Kant, and while it could represent the former's attempt to transcribe an argument which Kant presented to him orally during their meetings in the fall of' 1790, it is more likely that it simply represents Kiesewetter's attempt to reproduce what haid been published in 1787 as a memorandum for his forthcoming discussion with Kant.

2'Here "empirical consciousness must be meant in the special sense of' R 6313', p. 615, as the raw material f'or empirical self-knowledge, rather than such self-knowledge itself'. Its association with apprehensio, the least interpreted form of' synthesis, is evidence of' this.

356

KANT'S INTENTIONS IAN THE REFU-TTATION OF IDEALISM

relation of our senses to objects external to us. (Inner sense would otherwise have to be represented as outside us, etc.) (R 6313, p. 614).

The starting point of' this argument is clearly that the mere occur- rence of' a succession of' representations or inner states is not suf'f'i- cient for the representation or recognition of this succession. But Kant's further claim that such recognition can be grounded "only on something which endures, with which that which is successive is simultaneous" can then mean only that successive representations in one's own experience can be judged to be successive only if' they are judged to be severally simultaneous with the severally succes- sive states of' some enduring object. That is, in order f'or the pre- sent representation which is all one actually possesses at a given moment to be interpreted as a representation of' several past and present representations which have succeeded each other in some determinate order, a correlation must be posited between such a succession and the successive states of' an enduring object, such that the various representations which now seem to have previously occurred could have occurred only simultaneously with the succes- sive states of that object. For only thus can it be judged that these states must have occurred successively. But that means that it is only if' one's present representation can be taken to include a representa- tion of' such an enduring object and evidence about its history that it can be interpreted to provide evidence ftr a belief' in a succession of past representations.

This point may be made more graphically. (consider the possible objective states of' affairs A and B and the qualitative aspects of, mental states in virtue of which they might represent A and B, which we can designate as being appeared to A-ly and being ap- peared to B-ly (for instance, being appeared to deskishly and chairishly). Kant assumes that time cannot be perceived by itself, and unlike Hume never suggests that there is any phenomenologi- cal feature such as degree of' vivacity3') which could automatically mark one appearance as, for instance, a present impression and another as a mere memory. Following the Baumgartian equation of the "possible" and the "representable""' he must also assume

"?Treacise, Bk. I, Pt. I, sec. iii (Selbv-Bigge, p. 8). 3lBaunigarten, Melap/hysik, sec. 8.

357

PAUL GUYER

that the faculty of' representation itself', precisely because we can presumably represent anything which is logically possible, cannot place any limits on the potential contents or complexity of' a given representation. On these assumptions, there would be no way of' telling whether a present, state signifies just (a) now being appeared to both A-ly and B-ly or else (b) nowv being appeared to A-ly and (now remembering) previously having been appeared to B-ly, uldess the objective states of' affairs A and B themselves place some con- straint on the possible sequence of' representations of' them (ftr instance, the chair was brought into the room only after the desk). That is, it will be possible to determine that one's present represen- tational state is itself' a representation of' a succession of' representa- tions-(b) rather (a)-only if' A and B are themselves successive states of' enduring objects rather than, say, simultaneous states of' affairs.32

32'I'here is obviously an affinity betxseen this argument and the "realismll argument" which Bennett finds in the Refutation of IdlealisIlm (Kal'ls Ana- lyiic, pp. 202-214). My account and Bennett's dliffer inI that I provide the textuIl evidence of R 63 1 3 for what Bennett seems to have arrived at by

insspired guesswork an(l that my (listinction1 between situations (b) and (a) mav be clearer than Bennett's between 'I was . . . ' and 'I re(ollect , (1). 207), since 'recollect' could well be taken as what Bennett (loes not inten(l, namely a success-verb. BUt the main difference between my ac'ouint and Bennett's is that since On1 my reading of A 99 the ability to dist inguish between (a) and (b) becomes anr ability necessary fori atn representatiol of a subjective manifold at all, the argument of the refutation of idealism is revealed as more of'a comn lplement (or comjpletioni) of the argument of'tlhe trans(en(lenntal D)eduction than it is On Bennett's account. I agree with Bennett that the point otf Kant's argument is nlOt to proxidle a mnethol f'or making judgments aboit the past "trtustworthy` but to make them "p)ossi- ble" (p. 209), if' what he means by that is that Kant is not attempting to 1)roxidle a method fkor lending clii, ./illy to particular jdilgments about one's mental history-an objective obviouislv in omilpatible with Kant's repeated assertion (at, B 278 and at manx) paces in the Nachlass) that, his refutatioll (loes not lend certainty to the paltticulat judgments aboit external objects On which judgments aboit the self hamve been shown to dlependl-lbut only a method by which we can lend them mxINy degreee of confirmiation at, aill. '[hit is, I ultillaitelv interpret conditions for the possibility of' experience aIs condlitions f'or the verification0 of jud(lgIments about even subjective experi- ence. (But since-unlike Bennett-I (1o not associate this view with the attribution of a verificationist th//ory f en/coing to Kant, my cIcCouLnt ma1x' be able to avoid some of' the pr-oblemis which have been charged agailist transcendental rgUmients as a form of verificationisin. On the rejection of' Bennett's ascription to Kant of' verification-ist theory of meaing, see my

3 58

KANi'S IAii7NTIONS IA' THE REFUTATION OF IDEALISM

TIis tactic, suggested by the first of Our two paragraphs fro-m R 63 13, does not follow Kant's earlier tactic in threatening to ildeltif'yv inner states with objects in space; it only correlates them. The

qjleStioIn llOW remains, however, how does Kant introduce spJatiality into this argument at ally In f'act, I think that Kant entertained three ways of' adding to the present line of thought the require- ment, that the objects f'to time-determination be represente(1 as ill

space. (A) The first of OUI two paragraphs seems to inf-er thle necessity

of space directly fromt thle necessity' of'poSiting smilething enduring in or-der to to determine timee. This may see arbitrary, but Kant actually holds a plrelmise by which such an inference call be made. Ihe determinatio of' time reqtiires belief' in enduring objects. But,

we Must conceive of' such objects aS nmericallv distinct fvrom any particular representations, since representatiolls are by their very nature nonenduring. F'rtvher, we call now add thlat the enduring object which functions in subjective time-deter-minlation must also be numerically distinct. from the empirical self', f'or in fact the em- pirical self' is nothing but another interpretation of' one's present, representational state as representing a succession of' previous rep- resenitations, of precisely the sort which requires some other en-

during(I object f'or- its construction-. As Kant, puts it,

I cannot kinow time as antecedently determilne(I, in order to determille mlly OWnI existence therein. (TherefOre [I can determine it] oily ill SO tar as I c(onnnect ix' own alterationls according to the law ot Acausality). Now in Odcler to determine that empirically, something which ell(lIu-es must be given, ill the apprehension ot which I can cognize the succes- Sion of my OWn1 representations and through which alone the sim1_ul-

taneitv of da series, of which each p)art dIisapp)ears when another comlies into being, call comee a whole. Wherein I posit my existence (R 6313, j. 615).

In other words, my whole existence as an empirical self' is simply the limiting case of' the kind of' succession of' representations which

re view of his Kau/s Dialectic, The IlhilosoAhiall Review 85 (19716), 274-282, and1 Graham Bird, "Kant's Transcend(ental Idealism," in Godf rey Vesey, e(l., Idea/oSn IPas/ and Iretsen/, Rovl Institteof Philosophy Lectul-e Sel-ies: 13 (Cambridge: Cambridge UniVersity Press, 1982), pp. 71-92.)

I59

PAUL GUYER

is never actually before me but which must be reconstructed out of' my present state, so it is not itself available as the ground f'tr such a determination."' 'Therefore, the enduring object necessary for time-determinationl must be other than the empirical self as well as any of its particular states. But now Kant can infer the conclusion that such an object must be represented spatially simply by adding the further premise that space is the form of intuition by which we represent things other than ourselves and our states.""

This seems to have been a premise to which Kant had long subscribed.35 Only by taking it to mean this, for instance, can we render nontautologotis the Transcendental Aesthetic's opening claim that "By means of' outer sense, a property of' our mind, we represent to ourselves objects as outside us, and all without excep- tion in space" (A 22/B 37) or the claim, made in 1777, that "space . . .is the form of the mind's ability to perceive things as external" (R 4673, Ak. 17:640-641). But it is a claim to which Kant gives special prominence in the refutations of' idealism written after 1787. Thus, R 5653 claims that it is "through space that the repre- sentation of an object as outside me (in intuition) first acquires reality" (p. 307), that if I must place myself in a relation of' genuine distinction to "a correlate to my condition" then "the sensible, but real representation of this external relation is space" (p. 309), and, finally, that the very "proof' of' dualism" is based on the fact that for the determination of' our existence in time we must have "a percep- tion of the relation of' our subject to other things and regard space as the mere form of' this intuition" (p. 310). So if' I must think of' myself as standing in relation to something which is ontologically distinct from my empirical self and my representations, then I

iOnly by thinking of the empirical self as a virtual object reconstituted by the interpretation of one's present representational state (interpreting all of this epistemically and not psychologically, of course), can Kant even consider the possibility of the selft's being imparted trom one short-lived substance to another as momentum can be imparted from one elastic ball to another (A 363-364; John Biro reminded me of this passage).

44Strawson objects that "There is, as usual, no independent argumnent to the effect that the objective order must be a spatial order" (The Bollflds oJ Sense [London: Methuen, 19661, p. 127); he does not seem to see that Kant takes himself already to have provided the ground f'br this step by the premise now to be introduced from the Transcendental Aesthetic.

e5See for instance Inhtattgeral Dissertation, Sec. 3, Ak. 2:392.

36()

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

must represent this thing-or more precisely the fact of its inde- pendence from me-by representing it as separated from me in space.tm

This inference to the spatiality of enduring objects may now seem disappointingly trivial, but in fact it is an appropriate re- sponse to the "skeptical idealist" whose doubt, unlike that of the Berkeleian or "dogmatic idealist," is not in fact addressed to the spatiality of external objects, but ottly to their ontological status of independent existence. If' it is independence that is the target of skepticism, then Kant's argument has already addressed that, and the skeptic who has come this far will not balk at this final step to spatiality as merely the form in which independent existence is represented.

(B) Nevertheless, Kant does suggest two more considerations by which the spatiality of enduring objects may be even further en- trenched in his refutation. The first of these depends on the fact that, as R 6313 says, the method for time-determination just (le- scribed requires that we can "become empirically conscious of si- multaneity" since we must be able to judge that various represenita- tions are simultaneous with various states of the enduring external object. But as R 6313 has also argued (see Section IV above), the empirical consciousness or recognition of simultaneity requires ob- jects coexistent in space; only spatial separation creates the pos- sibility of alternative sequences of perceptions necessary to confirm the judgment of simultaneity.

This point may create a difficulty. If two objects A and B can be confirmed to coexist only if it is possible to go from the perception of A to that of B and then from another of B to another of A, and if' say A is to be an external object and B itself a representation, it might seem as if both A and B must endure, thus undermining the assumption that B, qua representation, is necessarily momentary. Perhaps this is why in the earlier quoted paragraph Kant suggests only that it is necessary to be able to go from A to B and back to A, and thus that only A must endure-this allows for A to be an enduring object and B a nonenduring representation. It is not

36To say that spatiality is that by which we represent the ontological independence of an object is of course not to say that it has its spatial properties independently of us; see Section VII below.

361

PA L GULYER

obviotis, however, that Kant had stifficiently thought through the idea hie is presenting to have seen this problem, and thus not clear that he actually inten(le(l this solution.

Aln apparently more pressing problem is that Kant's con(litions foy the possibility' o cleterminatiois otAsimultaneity seem to re(lUire that both of two items to be juIlge(I to exist simultaneously have spatial location, f'or this would imply that a representation jtu(lged to be siniuiltaneous with at state of atn oh)ject in space must itself' have t spatial location-which, as in Kant's dlirect argument ftroil time to space, would collapse the (listinction between a )repesentittioil o/

something in space and a spatial object itself'. In this context, howN- ever, the problem can be solved by holding not that representations must themselves be spatial, but that they must be given a spatial location by heing assigne(l to an object in space as their hearer- namely, the bo(y of the perceiver. But since this radically anti- (Cartesian conclusion' is also imlplie(d by Kant's final point in the

present argument, its further (liscussion will be deferred for at moment.

(C) 'the last, link in Kant's complete justification of'"objects oit- sicle Us in Space" turns on the fact'that the correlation between at succession ot relpresentations in the self' and of' states in an enclur- ing object requires that both the relations among the state of' exter- nral ob jets nd the relations between those states and the represen- tattions of them be rule-governed; otherwise the inferences bv which judgments ot succession are to l)e confirmedl could not be drawn. But this is just to SaY that tim-e-determination even of, sub- jective successions is possil)le oily if'all representations are pamrt of' a thoroughly causal o-der-and this is precisely what Kant hats said in claiming that to determinee my existence in tile I mutest collectt my own ialterations according to the law of' causality" (R 6313, p. 61 5). Then if' Kant's remark in the Critique that "in order to exhibit. lldealltion as the intuitioln corresponding to the concept of' cauaslit/,

we must take as our example motion, that is, alteration in space" (B 29 1; cf'. B 1 56) were taken to mean not that it is the idea of (altera- tions in outer sense which first makes alterations in inner- sense thinkable (B 292) but rather that it is the fimrmer which first makes the latter deterninable according to rule, there would then be an argu- mient that the necessity of causal laws t'or tilne-determinatiol also

implI)lies the necessity that enduring objects be perceived in space.

i36 2

K.NT' S IXiEXT\7()XS IA' THE REL YrATIOX OF IDEALISM

This list step might seem arbitrary, annd is certainly not explicitly defended by Kant. Nevertheless, there is atn rgumruent for it- namely, the consideration, already mentionled, that precisely )e- cause representations regarded merely as such must 1)e regarded as

p1)tentially capable of representing anything, their oewn nature merely as representations cannot be regarded as the source ot any constraints o)n thei- contents or secjuence. Onlv something external

to the formal reality of representations, such as space, could supply such constraint, ftor instance by precluding "contradictorily op-

posed determinations" (B 29 1) from existing in the same place at the same time. Or, to combine the present consideration with step (A), any constraints On the sequence of representations must be derived1 fromi something other than representations-and f'or us that caln only 1)e represented spatiallx'.

Like the claniml that verifiacl)l simultaneous states of affairs must all have spatial location, the claim that causally connectable states of affairs lmutst have spatial form also suggests that representations as well as thei- objects mUst have spatial location, I(1 must thus be associated with an enmbodied perceiver. Though Kant failed to mention this corollary of his argument in the text of 1 787, he did not fail to note it ill his notes fromln the subsequent years. Thus, in 1788 Kai-it drew the conclusion that "Since time cannot be exter- nally perceived in things, insofar as it is onlv at determination ot

inner sense, so we can determinee ourselves in timee only insofar as we stand in relatioll to outer things and consider ourselves therein" (R 5654, p. 3 1 3)-that is, consicler ourselves as standing in the same netwNrork ot relations which relates outer things to one another, or space. The same conclusion is drawn even more strongly in a note from 1 790: "We are first objed o/outer semse to ourselves, for other- wise we could not p~erceive our place in the world and thus intuit ourselves in relation to other things" (R 63 15, p. 619), Or "I am myself an( oject of outer intuition an(1 without that could not know my pOSitiOll in the world" (1) 62i0). But to sa thc0t I must perceive myself in a place in space is to say' that I mnust have a bodv, and that i-ny representations must be assigned the spatial location ot this body. Kant implies this in concluding fromi the last remarks that "'I'he soul as ob ject of inner sense cannot perceive its place w1,1it11i the

body, but is in the place in which the person [Xleulsch] is" (p. 6 1 9; cf. p-. 620). For the reason why the soul-the bearer of representa-

363G

PAUL (UYER

tions-cannot be a particular part of' the body is that it must be assigne(I the location of the body as a whole t'Or the system of tillle- determination to t't-Iictio11.'"7

This corollary of' Kant's refutation might appear to ulflIerlflifle it, t'or as an enduring object in space OntS own b)odyv could be held to be not merely a necessary but also a sufficient condition t'or subjective time-determination, and then the (oflclusion that there must be an object independent of' the self' would appear to be in danger. But this would be a misguided objection, for at least in modern skepticism"8 one need only think of' Descartes's first Meditation-one's own body is not excepted from the objects of' unreliable perception and doubt, but is rather the paradigmatic example of' n ob ject independent of one's consciousness or inner sense. It is precisely one's own body which is made of flesh though one might think it made of glass, or upstairs asleep while one thinks it downstairs before the fire. By inferring the necessity of' embodi-

'37As Karl Ameriks points out (with reference to passages in Kant's lec- tures [Ak. 28:225, 7561 rather than the Re/lexionen; see his Koal'N Theory of Mitid [Oxf'ord: (larendon Press, 19821, p. 107) this is an anti-Cartesial conclusion dlirectedl against the specific location of the soul in the pineal gland. As such, it is surely stated too broadly: it might well le possil)le to exclude some parts of' the body from the location of' the soul by showing them to be completely dispensable for the time-determination of' the em- pirical self' without thereby narrowing the location of' the soul down as far as Descartes would have liked. (Ameriks suggests that Kant has "not of'- fered conclusive arguments against placing the soul within the brain" with- out suggesting what the objections against Kant's arguments are.) It seems to me, by the way, that Ameriks's conclusion that on Kant's account emrl- bodiment is not "essential" but is "central" to our being (p. 108), besides being vague, misrepresents the more crucially anti-Cartesian aspect of, Kant's account, on which embodiment may well not be "logically neces- sary" to our existence hut is epistemologically necessary. It might also be noted that although Kant did not argue for the epistemicc) necessity of' embodiment in the Critique, he had prefigured this aspect of' the mature refutation in a number of' earlier works-f'or example, Teiaunie eines Geister- sehers, Ak. 2:323-324, and Inattgural Dissertalion, Sec. 30, Ak. 2:419. It might further be noted that Kant's argument follows Baumgarten's Con1- clusion that "the human soul is the faculty of' representation which repre- sents the world according to the situation of' the human body" (Metlphysik, sec. 545).

38That this is not so clearly true of' ancient skepticism is suggested by M. F. Burnyeat in "Idealism and Greek Philosophy: What Descartes Saw and Berkeley Missed," Philosophical Reviezw 91 (1982), esp. pp. 23-32.

364

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

ment fhrno his final refutation of idealismn, Kant has confirmed rather than undermined its anti-skeptical impact.

Rather, the more serious objection to be consi(lere(1 at this point is analogous to objection (3) considered in the previous section. This would be the charge that insofar as Kant's final refutation of idealism turns on the necessity of causal inferences in subjective time-determination it involves a vicious circle, for it requires that empirical consciousness of subjective succession depend upon knowledge of the causal powers of external objects, but also, pre- sumably, assumes that such causal knowledge must be generated by an induction fromi subjective successions, since it is not itself a prior (B 165).

Kant's argument would certainly imply a vicious circle if both his refutation of idealism and his acceptance of an empiricist model for the knowledge of particular causal laws were meant to describe the psychological processes through which the consciousness of subjective time-order and of objective causal laws respectively are actually generated, for they would then jointly describe a psycho- logical process that could never get off the ground. Causal laws would have to be learned from particular experiences, but already be known in order to individuate and order these experiences themselves. If', however, Kant's argument does not describe the generation of particular occurrences of' belief' about, both the self and the external world, but rather the perhaps ideal conditions under which such beliefs, however they actually arise, may be con- firmed or made into justified members of' a system of' empirical knowledge, then it involves no circularity. Even (an abstract justifi- cation, of, course, would be circular if' a particular belief' about, a subjective sequence were used to confirm a causal belief' about a particular external object but were also confirmed only by that same belief itself'. But as long as the f'ornmer belief' is not the only evidence f'()r the latter and as long as any other belief' about subjec- tive successions f'br which that particular belief about external ob- jects has functioned as evidence could then also be revised or sup- plied with additional evidence, and vice versa, there will be no difficultyy for Kant's theory in the supposition that any particular belief' about external objects might be accepted or rejected on the basis of' some belief about a subjective succession of' representa- tions. In other words, as long as Kant's arguments are construed as

365

IPAUL (GUYER

epistemology and not psychology, the f'uiiction of belief's in exter- ncal objects avnd theiv causal relations in the (letermination of' self'- knowledge itself' does not p-esent any vicious circularity. It is f'ar from clear that Kant always recognized this distinction-; indeed, it is clear that he sometimes tailed to make it.'" But it is also clear that he did sometimes see it, 90c and certainly there is much in his philos- ophy beside the ref'utation of idealism that can be provided even a

pnrima fie defense only if it is accepted.1

VI

Kant's efforts in 1788 and the following years thus pro-duced( at least, the outline of one sustaine(d argument. f'romli epistemological considerations alovie f'or the conclusion n that we must. know endutr- ing objects in space in or(lde to detevflilne the succession of our own representations. On this argument these objects ust be vegar(Ie(I as ontologically independent of' these representations annd the em-

pir-icl self represented by them simply because they are encu-ing. As I mentionled earlier, however, these notes also contain a f'ulrther attempt to argue foi such an onitological conclusion fromt patently

metaphysical premlises. This is Fan1 attempt to explain our possession of representations with spatial form by the existence of objects independent of' ourselves. It is deeply problematic, but also seems to be required to supplement only those f'Ormns of' Kant's refuratiovil which carry the argum-lent only f'r-omli time to spatial f'orm but not beyond. A separate inference from spatiality to otitological inde- pendence would be otiose when the primary inference is f'rolll succession in timle directly to enduring ob)jects, and when spatiality is introduced only to represent the already established indepen- dence of' such objects. Fhus, the problems which infect this sepa-

-'}Sticuh a conftisioti I haxe argtued. is the source of Kant's dloctlinle of "transcendental synthesis and "afffinity"; see my "Kant on Apperception and A priori Synthesis A er(own Philosophical Qo rert 17 (1980), 205-212.

'1Ssee, for instance, R 4678, Ak. 17:6(i11, from the so-called Dlisboig N(i dilass of the nid -1 770 's.

41 1For instance, mnLch in his aesthetics. For my most recent dIiscnLssioll of this issue, see "Kant's Distinction between the Beantifnl alld the Snhlini1e,' Review of /IavMl)bysis 35 (1982), 753-783.

366

KAA7,T'S INTTiNTHONS IN THE REFULTATION OF IDEALISM

rate metaphysical stage of one form of Kant's refutation need not destroy the whole. '2

Kant's first fragment fronm 1788 contains passages which ex- pound this additional metaphysical argument as well as others which seeml designed only to clarify the ontological commitments already contained in his more purely epistemological argument. At first, Kant's argument is ambiguous:

Ihe intuition of a thing as outside me p)reSLIj)I))Ses the consciousness of 'a determinability of 'mv subject, wherel)y I am not itself (leterlmli- nant, which theref ore (:loes not being to [mny] sponItaneity, Since the determining object is not in me. An(i in(Iee(I I c Cannot think of SpCce as in e. ITherefore the p)ossibility' of rep)resenti-ng things in space in intuitiol is g-l0oul(ie(d on the consciousnless of' a deterImlinatio0l through other things ... (R 5653, p. 3)7).

This could jUst mean that in inferring the history ot my own repre- sentations front the history oft an independent enduring object, I am of course thinking of that independent object 's responsible for the sequence of my representations rather than vice versa, and that it is because of the need to represent such an object as independent of myself that I m41ust also represent it as ill space. This analysis would clarify the ontological commitments of Kant's second tactic for the retutation without introducing any new premises into it. And both in later passages of R 56153 an(1 in subsequent notes as well, Kant does make remarks which can bearl such an interpret-a- ti(n. lThe imlmeliate continuation ot the discussionn in R 5653,

however, proceeds instead as if this passage introdUced an explana- torN inference from our possession of replresentations with phe-

42It sll otll(l I)e n1ote(d tha^-t the metaphysical argument about to l)e (is- cussed, although first a(lde(l to the reftutation of 1787 by the notes of 1788, is not aIctally an M argument tiew in 1788. Rather, it had b)een (leploye(l by Kant as early as the Ianugurl I)issert/al/io (secs. 1)0-II, Ak. 2:296-'397). 1Pelrhal.ps then its occurllellce in the later f)elio(l should l)e explained awav

as another instance of Kant's tunfortuniate attachment to i(leas which the dleelopment of his owin thought h1iad rendlere(l anachronistic. It so, this matakes infelicitous MWller-LauMtel's negaltive aIssesslllellt of the J)0st-1787 refutation, since it (lepen(ls precisely on seeing this metaphysical argument as its dominantt element ("Kants Widerlegung (les materialen Idlealismus," pp. 7 1-72).

367

PAUL GUYER

nomenologically spatial ftorm to the existence of' ontologically inde- pendent objects. Such an account (:oul( only be based on the metaphysical premise that the resources of' the self' as it, really is- and not just the contents of' its representations-are actually inca- pable of'producing spatial representations unless attffected by some- thing other than itself', with respect to which it is passive. The argument would then contend not that the emnplotnierit of' spatial

representations for the epistemic purpose of' time-determination requires the interpretation of' them as representing something other than mere representation itself', but rather that the explana- tion of' our mere possession of' spatial representations, whatever their use, implies the existence of' objects distinct f'romi a passive self' but capable of affecting it.

The way Kant glosses the last line of' the passage just (uote(d is in filct by explaining the possibility of' spatial representation as due to external action o(l a passive faculty of' the self' rather than to a spontaneous capacity of' the self':

'thus the possibility of representing in intuition things in sp)ace is grounded on the conscio)Isness of a determination through other things, which signifies nothing other than the original passivity of myself, with respect to which I am not active. That dr-eamis produce the illusion of existence outside me proves nothing against this; for there must always have been preceding external perceptions. Origi- nalily to acqUire a representation of something outside me without in Fact being passive is impossible (R 5653, p. 307).

That it is possible to be wrong about the existence of' the objects of' particular representations, "as in dreams and delusions, without that undermining the broader claim that "inner experience in gen- eral is possible only through outer experience in general" was of' course one of' the anti-Cartesian conclusions of' 1787 (B 278-279), and seemed attractive indeed. But its attraction will surely be (Ii- minished if it must be grounded on the assumption that 1)ecause of' the mind's incapacity to produce representations with spatial fborm by itself' the occurrence of' even nonveridical instances of' such a fbrm of' representation can only be explained by a prior action of' independently existing objects on the passive self'. Yet further pas- sages in both R 5653 and elsewhere emphasize precisely such an explanatory claim.

368

KANT'S INT'ENTIONS IN THE REFUTAT10N OF IDEALI SMA

Thus, the next paragraph of' R 5653 reiterates that the passivity of' the self' requires an external agency to explain the mhere posses- sion of' the representation of' space, "'or if' we were affected only by ourselves, yet without noticing this spontaneity, only the form of' time would be f'Ound in our in1tuitiol; and we would not be able to represent any space (an existence outside us)" (p. 308). As a conse- quenitce of' this metaphysically explained lack, of' course, the epis- teniological function of' spatial objects would also be impossible- "empirical consciousness as the determination of' my existence in time would be caught in a circle and presuppose itself' but ob- viously be impossible, since even the representation of' that which endures would be lacking . ." but this is not the premise of' Kant's present argument. The next note From 1788 also miakes the ulti- mate premise of' the refutation the explanatory thesis that

if there were not an external sense, that is, a capacity to become inmmediately COHsCiOHS (without an. inference of reason) of' someltlling AS outside us anI of ourselves, on the contrary, in elation [to it], then the representations of' outer things as such, space itself, Wvoul(I not even possibly belong to our intuition. For inner sense can Contain nothing lbut the temporal relation of' our replresentations (R 5654, pp. 1313-3 14).

Similarly, some passages fronm 1790 claim both that the representa- tion of' the external in general is inexplicable without an actual experience of it and that particular nonveridical representations of' the external cannot be explained except by reference to the on- tologically independent. In particular,

It is possible to take an image of external objects for perception (to dIrecam), but only on the p)resLup)position of (n outer sense, that is, on the presupposition thlat our outer sense relates to objects which are really to be found outside us (R 6315, p. 618).

And in general,

... in the realism of 'Outer sense nothing is asserted except that not even imagination could make any things at all representable as objects of the senses outside Lus as such, unless there really were such a sense, thus we (lo not distinguish the latter as a capacity distinct from the imagination by feeling [Em pJinduntg] alone, but by a certain inference [sicLwren Sdhldus] . . . (R 6316, pp. 622-623),

369

PA UL G UYER

As these last, words make clear, the present attempt to complete the argument o' the reftitation of idealismn deperfl(s on nothing less than an explanatory infel-ence to the Cause ot our possessiol of spatial representatioms i, I gcelcral. "'Yet by so prrocee(ling it would seen to lay itself open to the objection that. Kant made 1o skeptical idealism not only in 1 7X81 bUt, throughout the peio(l stlbse(JUeClt to 1 787 as well the objection that attempting to groulnd the indepen- (lent reality ot external objects on the mere contents ot oulr owvn representations by' means ot acaulsal infel-elnce muLIst lead to skepti- cism because such an inference (can never be conlcltlsive. This objec- tion to idealism is fvequentlv reiterated in these notes, II but no- where more strikinrigly than iII one Which then goes on to offer a argument open to the very satmre criticismll:

Idealism is thle opilliol that we imm1ll1ediately experieliCe onlyl o0t- ownVI existence, 1)Ht (CI onix' inteF thatt of' outer o1)ject s (Which inlf'erence from effect to cauLse is in II Lct nIcertain). However we (An11 only exiperi- ence our own existence inI so t'L1 aIs We determine it inI time, FLr which that which endur yes iS I-eCiL-edi, which representation has 110 object inI oUrselVes. This r-ePrlesentation also cannot 1)e grounded (onl the mere imagination [a11/'dcr blo.sseu LEnbildmigil of' something enl(durillg oLtsi(de us, I or something merely imagined for which nol corresponding oNject (can Ie given is impossible. It is that which gives the object in inltulition, and OUr representatiOn insofLr as it b)elongs merely to the consciouLs- ness of' ouL-Self' hs Io object of' tht sort (R 5709, p). 3132).

Ihe i(lealist is criticize( ftor trying to secure the realitv of' external

objects inl gen-erlal by a causal infteenc. butt Katt then goes onI to

infer such reality from an avg11-gUment which itself in1VolVes a caIlS explanation ot the availability to the imaginaitiol of' certain t,0 fov-

ot( image.

Perhaps Kant thoulLghtl that his association o't the forms of' inner, and outer sense with active and pa-ssiVe POWrel5s of' the self respec-

-1 31HIt is, it (O)es riot arguLe m]lerCely thlat giv('e1 thle peculiarlit ()f inn1lel senIse ouLr (letenlYinlt ions of sUCCessionIs withill it IIImLst he nllade onI the hasis of f'ant Irodr causal inferences f tom in(lep)end(lent o)jec(ts. which 1for US

happrn to b)e representte(l spat ially; rather. it attenipts to give a callsal exl)lanation of the very eroerl fict that Kant elsewhere clailme(l to he simply inexplicahle, nlamely that swe Possess the 8l)h/ial form of selnsil)ilitx at all (see also note 48 helow).

-'jSee R 3653, pj . 306(; R 3634, p. 31 2 2 and R 63 1 1 , ). 1 (I ).

37()

KANT'.-S INTFNY7 IONS IN THE REFUTA1TION OF IDEALISM

timely could allow him to overcome the indeterminacy which inftects the ordinary idealist's attempt at a causal inf'erence to the external. But he pIrovi(les no accoullt ot' how we knowT these powers ot' the self'. In t'act, his argument rests on a metaphysical assumption no more secure than the alleged insight into the essence of'a thinking substance which is supposed to teach the Cartesian meditator that intellect is aln active t'acultv ot the soul but imagination not, so that given the nondeceittulness of' God the clear and distinct, ideas of' covporeal oI jets furnished by the imagination Must be due to the active agency of' such bodies rather than of' the self' (leditltiatsl',

VI).15 Indeed, Kant's argum-ent is essentially the same its that of' Descartes, and rests our knowledge of' external existence on a premise which is in his 0owIn wVords "valid merely as a hypothesis" (R 565it4, P. 312).

But the ontological commitmllent to the indepn(leet. reality of' external objects is already implicit in Kant's second main tactic f't() the refutation, which derives the spatiality ot' objects ol froll the prior proot o' their endltlrance. It is also implicit even in t'hat foblr of' his first tactic which goes directly ft-oi tilmne to space I)ut then adds that endurance is a condition of' the possible perception of' svtce. Theret')re, Kant's attempt at- a metaphysical completion of' the ref'utattion is just as unnecessary as it is unpersutsivse. And, as I earlier mentioned, even R 5653 itself' contains passages Which (0o not try to explain the origin ot our spatial representations but only allalhZe the ontological commitments oft our use oft them ftor time- determnination. This one, f'or- instance, merely describess how we must regard the representation ot an object in space if' we use it f'or this purpose:

That if' I make imnyself' Fai object .spce is not in le bUt (yet) in the

f'0m1Al sUtbjectiVe cofl(Iitionl of' the empirical conscioUslness of, himself' that is, in time, proves that something Outside me, that is, something which I must represent in a manner other than myself, is connected with the emlpirical consciousness of tmnvselt and( this is at the samne time the conSciousnesS ot an external relatiOn WithOUt whiCh I cotl(1 not

emp)ilicallv determinee my own existence (R 5653, p). 309).

45IH 1/i T IhePlosophi(dl l'ork- o/ Do(,scar/les, e(dite(d b), Haldane and Ross, 1: 1 90-9 1.

37 1

PAUL GUYER

Here the representation of' space is significant not because of' its own explanation but only because it is what is used to represent what is already posited as distinct from the self'. Then the most, that is needed to complete Kant's argument is a reminder that it we are to use our representations of' spatial objects t'oy- purposes of time- determination we cannot also t'all back into the ontological attitude in which they are regarded as nierecly representations with no COrH-

tent referring beyond themselves. It is at least plausible to read what Kant says next as just such a reminder:

It comes to this, that I can become conscious of' myself in an external relation through it special sense, which is however re(Iuisite for the determination of' inner sense. Space proves to ice a representation which cannot be related to the subject (as object); for otherwise it would be the representation of' time. 'Ihat it is 1lot, but is rather immediately re/at/ed as existing to something distinctt fromi the SI1)ject, that [is] the consciousness of the object as a thing outside me (R -653, p. 3 0 9).

It is possible to read this to mean only that, wliatezer the explanation of' their origin, if' spatial representations are regarded merely as inner states without independent ref'erents, then they can be as- signed only the same purely temporal status as any inner states regarded merely as such-that is, transitoriness rather than en- durance. And it is even more natural to read this last paragraph of' R 5653 as arguing that the ontological reduction of objects in space to mere representations undermines their usefulness f'tr time- de- termination even while it is still assumed that as inner states they canl themselves be ordered in time:

The proof of dualism is based on the fact that the determination of our existence in time by means of the representation of space would contradict itself it one did not consider the latter as the consciousness of a relation quite other than that of representations ill uIS to the subject, namely, as the perception of the relation of our subject to other things and of space as the mere form of this intuition. For if the perception of space were grounded merely in ourself without an ob- ject outside LuS, it would at least be possible to become conscious of these representations as containing merely a relation to the subject. But since by the latter only the intuition of time ever results, the object which we represent to ourselves as spatial must rest on the representa- tion of something other than our own subject. However, that we canl

372

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

lhe conscious of an external relationship, without being able to cognize the object itself, but only the formn of this relationship of our self' to its presence, Occalsions no diffiCLltN (R 5653, p. 31 0).

We will return to the final claim that the interpretation of space as only the fOrm by which we represent the ontologically independent causes "no difficulty" in Section VII; our concern now is only the argument that it is contradictory for tis to think of spatial represen- tations as just like any other representations. Kant's claim is that in so doing we in fact continue to assume that. we can become con- scious of their determinate temporal relations, but also limit our- selves to the mere "intuition of time," which is to say we regard these representations as merely successive and thus fleeting and transitory. We thus deprive ourselves of the function of spatial representations necessary for the determination as opposed to the mere intuition of time, namely their representation of endurance. We contradict, not some metaphysical explanation of spatial iml1- ages, but simply the ontological commitments of an epistemic pro- cedure to which we have no alternative.

Such an analysis is also contained in some passages from 1790. One note supports its inference that the spatial representation "through the outer sense" of something enduring which allows us to perceive change must be thought of' as ontologically distinct from any mere representation by this argument:

Thus the possibility of inner sense presupposes the reality of outer sense. For Suppose one wanted to say that even the representation of' the enduring object which is given through outer sense is only a per- ception given through inner sense, which is only represented as if given through outer sense by the imagination, it would yet still be possible in general (even if then not for US) to become conlsciou0s of' it as belonging to inner sense; but then the representation of space would be transformed into a representation of' timie (according to one dimension), which contradicts itself'. ITherefore outer sense has reality, since without it inner sense is not possible (R 63 11, ). 611-612).16k

Since even at his most metaphysical the mature Kant never argues that we would not have any representations in inner sense unless we

4-1Keep in mind, however, that this fragment is in the hand of Kiesewet- ter rather than Kant.

373

PAUL GUYLER

possessed representations of' otuter sense, l7 the last line can only

mean that the dletermialltion of' inner sense is not possible wvlitl-lholt the interpretation of' spatially represente(d objets as enrllting ob-

jects in(lependent of' the self', or WithouLt the ascription of' "realitY

to "ouLter sense." But, perhaps even clearer than this passage is another, f'romi the last of' Kant's ret'utLLtions ot' i(lealism fr-onm 1 790, in which Kant argues that the representation ot'an object in space

is a special kind of' representation in1 uLs, WhiCh ca1not rel)resenlt that which is inl LnS, thLus What exists ill the flux of' tillme, f'oi then as a mere representa'tion0 it Woniri lbe calalable of being thought only in relation-s of time, therefore such an in1tuLitionl Imu1lst standl in a real relation to somethillg outside LnS and space i'eclly signifies something, which it is 1)ossit)le to el)ep'esent in this form of' iltuitioii olyi l)v nieans of' a relations to at real thiig ouLtsi(le us-thus the refutation of skepti- cisni, i(lealism, spinozism, even so of' materialism, plredletermiiinismI (R 6'317, pp. 627-628).

All of' the alternatives to Kant's Own realism, in other words, are

finally held to Lill before tan argument, not thdat thinking of' spatial representations as mere representations makes their actual fori'm or occurrence inexplicable, hut rather that so thinking of'them would,

by assigning them t~o the mere flux of' time, make then entirely transitory and thus undermine their fulction in "the dletermilna- ti(n of' our own existence in time" (p. 627). '[his rieqUires no meta- physical premise not already employed in Kant's two tactics f'or tlhe final ref'lutation of' idealism, but is merely the analysis of' the on- tological commlnitments contained in the seconni of' those argtumlellts and even in the more complex version of' the first. "I

7Ith'le vounigei Kant, however, had frist t riedl to refute ilelalism hby airgu-

ing that Io changes inI inntler sense CoUld (Oc(cu unless caused bi t ae it gency of' independent sul)stbances (cf. Noia Del/uid/llio, Ak. 1:4 11-4 1 2). Kant's Imature refutatioll 1ma' l)e r'egar'de(l as the epistemolo<gical parallel of' such a metal)phvsical argumlet.t

"48-lhus Kant's final ref utation of' iledalism (loes not 'oitiadlict his remark tllat "it is nmo mllol-e possible to profile a ground for the )ectuliarity of' the undlelrstandling, that it is only bv means of' the categories . . . tllat it ('all prodltce (a pioi tiunitv of' (A)aerception tian [it is to 1)i'ox\i(le a gi-'otill(ll for0 wxhv we have just these and no other fUnctions f'or jtLidging 01 ' for why timlie

an(l space re the onl\ f 'Ors ofuir possible intulition" (B 145- 146). F'o it (loes not attempt to explain why we possess spaCe aIs at f'or-mII of' inltulitioll

3 74

KAX> IEX1l('ON)SX LV THE REFT7A 7iON OF IDEALISM

VII

Now, however, two questions about the ontological commitments of' Kallt's ultim-nate reftlta0tion of' idealism can no longer be sup-

pressed. First, if' our knowledge of' the independent existence of,

external objects in general is not to be a causal inference fi'-om the f't()rm of' our own representations, as Kant continues to maintain

even when he indulges in such antl inference himself', then what lS the method by which this knowledge is derived? Second, how can Kant argue f'or '(dualism- (R 5653, p. 3 10), or be a "realist of' outer inlttlition' (R 63 15, p. 2 1 0) without surrendering every shred of' his own transcendental idealism'? How can he argue that outer ob- jects really are ontologically distinct from ourselves even while ad- hering to his established line that "if' our knowledge of' external objects had to be knowledge of' them (and of' space) as things in themselves, we would never be able to prove their reality f'ronm our sense-representations of' them (as outside us)" (R 6313, pp. 6 14-615)?

One long passage from 1790 suggests Kant's answers to 1)oth these questions. As if' in answer to the first, Kant writes:

Whether the objects outsidee LS) or their representations affect US . . .

can be (lecidledl thus. 'e needl space iI order to construct [Co0ux/r111iiOC1]

time, and therefore determinee the latter bv means of' the f'0rml1er. Space, which represents the outer, theref'ore precedes the possibility of' time-determination. Now since ill respect of, time we are affected oily by representations, nlot l)y otiter things, In alternatiVe relmllaills btit that ill the representation of space we mtiSt l)e Con1SCioSu of OtIr--

selves as affectedl by external things. We (lo not know this throutih an inference, but it is grouinded in the waxN in which we affect ourselves in

order to construct time as the mere form of the replresentation of otir iller, condition, whereby there vlst vet always be given to us some- thing other, not belonging to this inner condition (that is, something external, the const-Liction of which also contains the intuition of time andl is its grt()11).

either in addition to time or in Contrast to some other conceivable (thoUgh of' course tinitnagineable) form of' otiter iIttlition; it only offers the epis- temological argument that if' we can make determinations (f stibjective time-or(ler then we must possess some form for outer inttiition-otir only can-(lilate f which inexplicably happens to be space.

75

PAUL GUYER

And then, as if' in answer to the second, Kant continues:

In order that something can aippear to I)e outside us, there must re.ilx be something outside us, thouLgh 1not CollstitUte( in the way wve have the representation of it, since other kinds of sense could atfor(1 other Ways of representing the same thing. For otherwise the representatioll of something outside LuS could never come to be thouglit, since wNe would only be Con1SiOLIS of oUr representations as inner (leterllilna- tions and fo- their objet toul(1 have only inner sense, which wve vet carefhli' distillgtlish f rom outer sense (R (63l 12, pp. 6 12-61 I 3S).

Since the second of' these answers can be interpreted morve Straight- f'orwardly than the first, we mav examine it first.

Kant's interpretation of' transcendental iclealism in the FouL-th

Paralogism of' 178 1 as requiring the reduction of' spatial ap- pearances to a species of'our own representations (lepended onl the assumption that the ontological and epistemnological status of'exter- nal objects must be isomorphic. It assumed that if' things numLler- ically distinct ftrom us exist independently of' us then the properties by which we represent them mIUSt be ones they can be known to possess independently of us, an(1, conversely, that if the properties by which objects are represented are known to depend upon oulr-

selves in any way then the objects themselves must also be depen- dent on us for their own existence. Thus the "transcendental real- ist" was defined as one who "hypostasizes outer appearances and comes to regard them not as representations but as t1/ilgls exi Xstilig byl themselves outside Us, with1 the samtie qalital l S 11(itt whiti1 wlwh d1/iy ex ist iil Us" (A 386). Kant's own transcendental idealism I)ecause, ats he always continued to suppose, it had already established "that time and space are . . . only sensible forms of our intuition, not dleter- minations given as existing by themselves nor conditions of objects as things in themselves" was thought to require the ontological assertion that the objects represented by means of these forms, namely objects as appearances, "are to be regardlecl as being, one and all, representations only, not things in themselves" (A 369). But it was only necessary for Kant to drop this assumption (which even in the first edition of the Critique is insisted on only in the Paralogisms and a few passages in the Antinomies that Kant did not revise in 1787), in order for him to conclude that there is "no

376

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

difficulty" in the idea that we could be conscious of a relation to something independent of ourselves "without being able to cognize the object itself, but only the form of [its] relationship [to] ourself" (R 5653, p. 310) or in the conclusion that I am "conscious of (my) existence in time, also of the existence of external things, to be sure as appearances, yet as real things" (R 6314, p. 616). Kant simply dropped the assumption that the epistemological status of being an appearance," or known through one of our own forms of intui-

tion, required the ontological status of being a "representation," an actual state or modification of the self, and thus he could exploit the possibility that we could know that something exists indepen- dently of us without knowing what it is like independently of us.Y1

By this means Kant could retain his arguments for the transcen- dental ideality of space and time even while advancing his argu- ment for ontological realism; thus he could assert that

the idealism of' space and time, which is merely f'Ormal, does not contain real idealism, which asserts that in the perception of' things in space no object at all beside the representation is given, rather [it implies] only that the same form of' space does not pertain in itself' to this object or these external objects (which of' these it is renmalirls un1- decided) by means of' which we intuit it or them, since this forrm belongs solely to the subjective manner of' our faculty of' representa- tion in perception, which can be inferred from the fact that space contains nothing in itself which could be a representation of' a thing in itself or of the relation of' (different) such things to one another, and if' it were considered as such a determination, as an ens ims1(agin tiariuni7 it would be a non ens (R 6316, pp. 621-622).

49As suggested in note 4, not only the Kant of 1781 but also the majority school of his commentators, who see no doctrinal difference between the treatments of idealism in 1781 and 1787, have been plagued by the failure to see that the epistemological dependence of that by which something is represented does not entail the ontological dependence of what is repre- sented. An example is Paton: "On Kant's view, however, the existence of permanent substances in space can never be proved unless these perma- nent substances are phenomenal substances dependent, like space and time, on the constitution of the human mind. Hence he is not departing in the slightest degree from his own doctrine" (Kant's Metaphysics of Experience, 2:380). What Paton fails to see is that Kant's argument in 1787 and after is that we must employ the two admittedly dependent forms of intuition- space and time-precisely because we must represent objects which exist independently of ourselves as well as states which do depend on us.

377

PA L.T GUVER

Precisely because he could now maintain the (lependent status of' the fior by which we represent external objects without implviilg that the existeltlw of' these objects is dependentt on uIS, Kant COUld

rightly call his mature position normalal rather than "veal icleal- ismn"-it is idealismn with respect to the forimis but not the objects of inituitionl.50

Indeed, once Kant dlroppe(1 the simplistic dichotomy underlying the f'Ourth Paralogisml of 178 1 he could even retain the position, unsuccessfully argued there, that our knowledge of' the indepen- dent reality of' objects in space can be secured oily if't we recognize that their spatiality, unlike their existence, cannot 1)e indlepen(lent of' ouL replresentations of' them. Thus the satme R 63 17 which conl- tained Ktant's final argument against skepticism, spinozisim, ideal-

ism and all his other targets could also argue that

reality can be seCUred for outer oI)jects (as things in themselves)

p)recisely insofar as one (loes not aIssIuIme that their intUitioll is one of a thing in itself; for if it were this atnd the form of space were the form of a thing which pertaine(d to it even without the special on1stitution0 of ouL sUbject, then it would b)e possible that we should have the rel)pe- sentation of such a thing without it existing (R 63 17, p. 627).

Even if' the argument underlying this claim that if' the characteris- tics of' space can only be known empirically rather than a prior1i,

then there will be a lack of certainty to this knowledge which will also infect our knowledge of' the existence of' object's in space is dubious, it should now be clear that the ftbrim of' transcendental idealism which it requires is compatible with a realistic interpreta- tion of' the intended conclusion of' the refutation of' idealismi.

5()See also this comment in a letter to J S. Bek of I)eceinhei 4, 1 792: Messrs. Eherhard's and Gari-e's opinion that Berkeley's idealism is

the same as that of the critical philosophy (Which I Conld l)etter call 't he principle of the i(leality of space and time") (toes not deserve the slightest attention. For I speak (of ideality in I ef e ence to thefi far t( r iaa ;

bLnt they interpret this to mean idleaslity wvith respect to the mailer, that is, the idealitv ()or the ()oject (1i(1 its x ei x existence.e (Ak. 1 1:395; translated in Arnnllf I'Zweig, ec. Khaul:. Philosaphical Carrcwsaoidrcwe 1759-99 [Chicago: University of(l Chicago Press, 1 967], ). 198). The (charge of Berkelilanismll had rankled Kant for- teln years; the argument of the matrl-e r-eftlaltion, Unlike the rednctionism of the t'Onrth Paralogismll, had in(deedI made it Ln 'j Lnstified ex paosi f/tO.

378

KANT'S INTLVTOXNS IW THE REFUTATION OF IDEALISAI

So at. this point we mllUst return to our first question, namely, how is our knowledge of' the existence of external objects to be under- stoo(1 it' it is uot too be interpreted as the result o't causal inference ftro-m- some general feature of' our representationls such as their spatial f'tor-n. ' This question is not so easy to answer as the previous one, f'or in both the published text of' 1787 and many of' the f'rag- ments f'romn l 788 and l 790 Kant created considerable difficulty f'or himself by tacitly assuming that there is ante exhaustive distinction between imnedluct(, and infereiiall knowledge. Thus our knowledge that there are independently existing objects must be one or the other of' these, and when he then tried t~o force the result of his actual argument into these two categories he was in a way bound to distort it. Therefore we saw thhat in spite of his own diagnosis of' the source of ''skeptical idealism" Kant was telflpted to argue that the distinction between mere imagination and actual outer sense can be known by means of' a "correct inference" (R 6316, p. 622). But Kant was also drawn to the alternative and asserted that his refutta- tion showed our knowledge of' external objects to be immediate rather than if'erential. Thus the published Refutattion concludes

that "the consciousness of' mny existence is at the same time an immediate consciousness of' the existence of' other things outside

e" (B 276), and this claim is repeated in 1 788:

I'he mind must be imlmfle(diately COnSciOUS of a representation of outer sense as such [a representation], that is, nOt through ain inference from the representation as the effect to something external as at cause . ..if there were not Fanl outer sense, that is, a capacity to be c(OSCiOUs of something as outside us immediately (wNithout an in- ference of reason) and to be conscious of ourselves on the contrary as in relation to it, then the representation of external things as such would not even possibly 1)elong to an intuition (R 55654, pp. 3129-31 3).

The claim of immediacy is also raised in 1 790: the alternative to ascribing to spatial representations "only the form (and dimensiol) of time" is that spatial formn "must not be thought but intUited, that is, immediately related to an object" (R 63 15, p. 61 8).51

But there is also a problem with the claim that owur knowledge ot

,I"EvU i/I; as Kant continutes, "we cIlso do not know zho!t thli is. ins it/l/) bIlt only hNow it appears to us" (pp. 618-619).

379

PA UL (;UYER

external reality is immediate as well as with the alternative. This is not the problem which some 52 have found in the idea that a claim of immediacy should be the conclusion of an argument which Kant himself' describes as requiring a climnb "through ditTicult subtleties to the peak of' principles"; f'or as Kant himself observes, the point of' this argument is not so much to prove something which "the healthy understanding could not attain without all this t'alderal, but rather to deprive entirely of' their force all the sophisticated sub- tleties which are raised against it" (R 5654, p. 313). That is, it can perfectly well be the case that our knowledge of' something is imi- mediate while our further claim that this knowledge is immediate itself' requires considerable defense. Rather, the problem is that Kant's refutation of idealism undermines at least the most obvious candi- date f'or a premise fromn which the immediacy of' our knowledge of' external reality Could be inferred. If an argument is to entail that one torm of knowledge which is the condition of' another is immnie- diate, it must include either a premise that the latter is itself' imme- diate or at least an explanation of' why it must rest on immediate knowledge. But Kant cannot claim that the empirical knowledge of' subjective time-determination tor which knowledge of external ob- jects' serves as a condition is itseq/fimmediate; his distinction between the unconditioned, transcendental consciousness that I exist and empirical knowledge of' how I exist in time implies precisely the opposite. In Kant's own words, "The transcendental consciousness of ourselves, which accompanies the spontaneity of all acts of our understanding, but which consists in the mere 'I' without the deter- mination of our existence in time, is certainly immediate, but the empirical consciousness of' myself', which constitutes the inner sense . . . by no means occurs immediately" (R 5653, p. 306). So the "empirical consciousness of' myself'"3 which is the premise of the refutation as a whole is not itself' a premise for the immediacy of' our knowledge of the external. Yet an argument that individual claims to knowledge of external objects must be immediate because they justify without in turn being justified by mediate claims of' selt'- knowledge will also not work; for as we saw, Kant's refutation can be saved from circularity only by the assumptioii that while a par-

52E.g., Paton, Kant' Metaphysics of ExperieIce, 2:382. 53This phrase is again used in the sense of empIIJiriCll self-knowledge; see

Section III and note 29 above.

38()

KANT'S INTENTIONS IN THE REFUTATION OF IDEALISM

ticular claim of' self-knowledge cannot be used to justify the very same judgment(s) about external objects which are used to confirm it, such a claim could be used to justify some other claim about external objects. On this account, however, judgments about exter-

nal obj are no more imediate than determinations about self'- consciousness. Thus we must conclude that the claim of' immnediacy is not directly supported by the argument of' the refutation itself'. At the most, it can only be successfully maintained if read simply to mean the supposition that knowledge of' the external in general is inferential leads to skeptical idealism.

However, what the first of' the two paragraphs with which we began this section can now be seen to suggest is that Kant at least once realized that the very assumption that our knowledge of ex- ternal objects must be either immediate or inferential was mislead- ing, that neither of these alternatives is suitable ftr describing the status of a presupposition which must be brought to bear on the raw material of experience before any determinate judgment whatever can be made about that data. Instead, what Kant saw was that the concept of' endurance and the belief' in its representation of an independent object had to be brought to the data of sense in order to "construct" time-a term which we may best understand to con- note not a literal (phenomenalist) construction but rather a con- strual, an interpretation of the subjective through a network of judg- ments about. temporal relations tor which there is in turn some method of justification. On this account the presupposition that

there are external objects would be neither immediate nov inferen- tial. For these are terms appropriate for classifying the status of' particular claims within a system of'justified beliefs, yet the supposi- tion that there are some external objects, just like the supposition that there are some subjective states, is part of' the interpretative framework within which individual efforts at justification can proceed.

That neither the existence of' the self' nor the existence of, objects are ilndii(lual items of empirical knowledge )tlut rather presupposi- tions f'or the determination of' any empirical belief to which (lesig- nation as immediate or inferential applies is tlS(O imlplie(I in the

51Paton seems to be making the same point in his arguIent that the permanent is really presupposed rather than perceived on Kant's account. See Kant's Metaphysics of Experience, 2:379 and 2:204-207.

381

PAUL (UYER

words following Kant's express dlental that empirical c)itsciousness

is immediate:

the c)nsciOIusn1ess of other things outside me (which as I initellect u Al must also be presupposed an(l which is thus not at [mere] lepiesenta- tion of them in splace, I)Ut rather can I)e called intelle(tunal int uition, through which we have no knowledge of things) and the (leteriina- tion of their existence in space must l)e simultaneous with the (Ieter- minaition of mx's existence in timre, therefore I [know] mv' own emi1- piricially letermine(l existence 110 moie thailn that of things ouitsi(le mile (which, (als to) whait they are in themselves, I (10 not know) (R 565)3, p). 306).

It is SUIprpisling for- Kant to characterize the knowledge that exter- nial o),ects exist ts iutellectttal inttlition," when that phi-ase or- dinarily means what is of CoUrse not, possible knowledge fr-om concepts alone, anl when lhe has even invoked the impossibility of intellectual inttLition of' the self' in this sense as the xverv basis for the reftttation of' idlealism (B xl). AncI it is clumsy, of' hillm cla relpse into the (lichotomytv of' 178 1 to say that this "intellectual inlttuition0"

plroxides 110 knowledge of' things when even as the last line of' this extract makes clear-, lhe iecans to (leu-' only that we know these things as t( hey are in themselves. BuLt wha-tt hie is tirvilng to sax' seems clear enotugh, and might, haxe been caught by Ia p)hrase like "intel- lectualized inttuition." Kant's point is that the mere Occurrence of intulitions with either temlpo-al l)hperties or even the pli- iomenollogical forml of' spatiality is not itself' enough to plroi(le

empirical knowledge ot' the temporal relations ot' these rel)resenta- tions; for this puLI-pose the presuppositiomi that at least some o'f these represent endulrilng objects, including ones o wn bo(ly, is all essential element ot the interpretation that mLuSt be pll(e(l onI the sensory content of the representations. Ani interpretation mu1LISt be placed on intutition before suLch intuitions can be Lse(l for tdhe em11- pir-ical determination ot' one's OWIn experience. thtis before qtues- tiois ot' immflediate or- inferential statuLIs Canl exCe11 pio)pely arise.

langttage of intentionality not available to Kant might have helped him to avoid imlisleading impressions here, butt even without

it he has pointed to something important. In fact, it Was coi-rect fOr Kant to contrast himself to the "skeptical idealist' by (Ienvying that otir knowledge of' indleplendent reality in general is inferential. For on the Humean model of' causal infeerence which Kant 1himi-self

382x

KANi' INTENTIONS IN THE REFUTATION OF IDEALISM

accepted, the premise f'or any causal inf'erence must be logically independent of' the conclusion to be inferred, but the potential premise of such an inf'erence is otir knowledge of' our own tem- porally discrete representations, andl on Kant's final account knowledge that external o1)jects exist is already implicated in this premise. At, the same timee, it was also correct f()r Kant at least to suggest that knowledge that external objects exist might not usefully l)e described as immediate. For such a description could be taken to mean that such knowledge is simply given in intuition, when in fact intuition by itself' yields no knoNledge of' anything. Rather, the knowledge that external objects exist is a presupposi- tion which must be brrought to bear- on intuition. (Confusing as it may be, the language of "construction" or "intellectual intuition" is at least an improvement on the misleading (lichotomy between immediate and inferential knowledge.

This is not the place to assess the ultimate f'tb-ce of Kant's mature refutation of' idealism, which would require an overall assessment, of' Kant's transcendental deduction of categorical princilples from conditions of' self'-consciousness, or indeed of' transcendental argu- ments in general (if' in fact these are distinct). But I hope to have shown that there is better evidence f'tr a realistic interpretation of' this refutation than is generally known, and that Kant did at least attemlpt to develop tactics f'()r this argument which are both more detailed and also more intimately connected with the most success- f'ul aspect of the Transcendental Deduction than has usually been realized.5) )

University of Pennsylvania

-Vi)ork nI this paper wias begun du-ring my tellUre of a National Endow- ment for the Humanities Fellowship f'ori Independent Study and Research inl 1978-79, and I would like to thank the Endcoment f()r- its (patient) Sl1PI)()Iot. Some (f that perio(l was spent Uinder the auspices ()f University College, Oxf)rdl, and I would also like to thank the University (College Seimii C(mmiion Room fI()Io their hospitality. I hlad the privilege ()f (liscuLIss- ing Somie ()f the ideas f()r- this paper with the late (Thirethi Evans of' thatl College, and I Wotild here like to commemorate that wNarm CInCd brilliant main WhoII iS m11uCh missed, even by those who had barely come to know himl.

Earlier versions of this paper were Iread at Northwestern University, the

University ()f (Califon() 1ica at San Diego, and the University (of icagh , CIII( p-ti()iS ()f its petIlltinmate f(rm at the Ulniversity of' Oklahoma and the University (f Arkansas. I had patrticulaIrly useful conversations of this last (raift with oh(n Bir(), (CIis Swoyer, and Tony BrI-ueckner.

3 8 3