亨利希-康德先验演绎的证明结构

Henrich, Dieter, The Proof-Structure of Kant's Transcendental Deduction , Review of Metaphysics, 22:4 (1969:June) p.640 KANT'S TRANSCENDENTAL DEDUCTION 64 1 In the following, an attempt will be made to settle this conflict which has persisted more than 150 years.' We shall advocate the thesis that only the second edition develops a tenable argument and that the argument in this version corresponds more adequately with the specific structure of Kant's thought as a whole, than does that of the first edition. This position contradicts the most im- portant interpretations of Kant; moreover it proposes to re- evaluate the meaning of his work and to guide its reception in a direction other than that of speculative Idealism, Neo-Kantianism, o r Existential Philosophy. We will treat first another controversy which, compared with the debate over the value of the two editions, is only of minor importance, yet which is relevant here insofar as it ultimately leads back to this question and allows it to be answered: namely, the controversy concerning the structure of the proof in the second edition. In this edition the conclusion of the deduction seems to be drawn twice in two completely different passages. It is the task of a transcendental deduction to demonstrate that the categories of our understanding are qualified to provide knowledge of ap- pearances, as they are given to us in the unity of a synthesis of experience (B 123). The conclusion of section 20 reads: "Conse- quently, the manifold in a given intuition is necessarily subject to the categories" (B 143) . This conclusion does not seem to differ from the result of section 26, according to which "the cate- gories . . . are . . . valid a priori for all objects of experience" (B 161). Thus one is tempted to see two proofs of the same proposition in the text of the second edition. That leads, however, into direct conflict with Kant's unequivocal explication in section 21, which states that two arguments, rather than two proofs, are involved In this paper I shall discuss only the proof-structure of the Tran- scendental Deduction. An analysis of its premisses and the problems involv- ed in the application of its conclusion will be given in another paper. 642 DIETER HENRICH and that these together constitute the proof of the deduction. "Thus in the above proposition a beginning is made of a deduction of the pure concepts . . . . " "In what follows, 'something further' will be shown . . . . " "Only thus, by demonstration of the a priori validity of the categories in respect of all objects of our senses, will the purpose of the deduction be fully attained" (B 145). We can now formulate a criterion for a successful interpretation of the whole text of the deduction in this way: the interpretation must show that, contrary to the initial impression that the two conclu- sions merely define the same proposition, on the contrary, sec- tions 20 and 26 offer two arguments with significantly different results, and that these together yield a single proof of the transcen- dental deduction. We shall call this task the problem of the two-st eps-in-one-proof . In previous commentaries this problem has been either pro- nounced insoluble or else passed over in s i l ence .Vhe better commentaries claim that Kant's assurance that his deduction pre- sents two steps in one proof cannot be taken seriously, and that we are compelled to read the text as two distinct and complete proofs. Two proposals made on the basis of this double-proof theory merit our attention.' We shall call them the interpretation according to Adickes/Paton " and the interpretation according to Erdmann/ de Vleeschauwer " and shall examine them in that order. 1. In the preface to the first edition of the Critique, Kant himself distinguished an objective and a subjective side of the Cf., for instance, Norman Kemp Smith, A Commentary o n Kant's Critique of Pure Reason (London, 1918), p. 289; and A. C. Ewing, A Short Commentary on Kant's Critique of Pure Reason (Chicago, 1967), p. 120. In recent English publications on the Critique one does not find a discussion of this problem. Bennett and Strawson are writing from a point of view which does not focus on it. Wolff is interested almost exclusively in the first edition of the Deduction. " Erich Adickes, Kants Kritik der reinen Vernunft (Berlin, 1889), pp. 139-140; Herbert James Paton, Kant's Metaphysic of Experience (Lon- don, 1936), vol. I, p. 501. "enno Erdmann, Kants Kritizismus in der 1. und 2. Auflage der Kritik der reinen Vernunft (Leipzig, 1878); Herman de Vleeschauwer, "La deduction transcendentale dans I'ceuvre de Kant," in Werken uitgegeven door de Faculteit der Wijsbegeerte en Letteren (Ghent, 1937), vol. 3, pp. 24 et seq. KANT'S TRANSCENDENTAL DEDUCTION 643 deduction (A XVI). The objective side makes the validity of the categories intelligible, the subjective investigates their relation to the cognitive faculties in us which must be presupposed if these categories are to be used. According to Kant one can also dis- tinguish these two aspects as the demonstration that the cate- gories have validity, and the demonstration how they attain validity. Adickes and Paton propose that this distinction be employed in order to understand the division of the deduction into two arguments: section 20 completes the proof of objective valid- ity, section 26 demonstrates the subjective conditions of applica- tion. This proposal has the advantage of being able to invoke in its support certain fundamental Kantian statements about the deduc- tion-but there is no further evidence for it. For it is clear that the proposal cannot be applied to the structure of the second ver- sion of the deduction. In section 21 Kant clearly stated that the demonstration of the validity of the categories would be completed in section 26 (B 145). The title and conclusion of this section can be read in no other way. And the text itself contains no reflections about the interconnection of our cognitive faculties. The little word "how," which can indicate the distinction between a psycho1ogical and an epistemological investigation, a subjective and an objective deduction, only appears incidentally. In this context, however, we shall see that it must be understood quite differently, 2. The proposal of Erdmann and de Vleeschauwer likewise attempts to understand the second version of the transcendental deduction with the help of another observation of Kant's-this time of a distinction made in the first version of the deduction. In two corresponding trains of thought, Kant here elaborates the relation between the categories, whish can be developed from self-consciousness, and the given sensible representations. He distinguishes them as the demonstration "from above" and that "from below." In this way he implies a hierarchy of cognitive faculties, the highest of which is the understanding and the lowest sensibility-extremes between which the faculty of imagination establishes a relation of possible coordination, and between which the two proofs move in opposite directions. 644 DIETER HENRICH It seems quite natural to apply this distinction to the inter- pretation of the second edition. And indeed Erdmann and de Vleeschauwer propose that section 20 be understood as a deduction "from above," while section 26 is to be regarded as a deduction "from below. >, This proposal is in better agreement with the text of sec- tion 26, which has supplied the decisive arguments against the interpretation of Adickes and Paton. For Kant here actually proceeds from intuition, mentions the achievement of the faculty of imagination, and comes then to speak of the unity in the forms of intuition, which can be reached only through the categories and by virtue of the unification of the manifold in a consciousness (B 160). Nevertheless the two parts of the deduction remain unexplained by this proposal for the following reason: the struc- ture of the first argument in section 20 can in no way be con- ceived as a deduction "from abovev-and thus as a process which differs from the argument of section 26 insofar as its proof must be constructed in the opposite sequence. In section 20, just as in section 26, the manifold of a sensible intuition is mentioned first. Then it is shown that the manifold can assume the charac- ter of a unitary representation only if it is subject to the categories. Thus both arguments establish that a given intuition can become a unitary representation only \\hen the intellectual functions of the understanding are applied to it. Now as to whether or not 6 L this argument can properly be understood as a deduction from below": the forms of these proofs in no way make it possible to draw a meaningful distinction between the considerations of the two sections. Hence the failure of the only proposed interpretations-not only because they depart from Kant's assurance that there is one proof presented in two steps and attempt instead to find two dis- tinct proofs, but also and primarily because their arguments can offer no useful explanation of the distinction between the two proofs. We must search for another interpretation of the text. It should avoid both of these errors as far as possible and seek an understanding of the proof of the deduction that would require the two-steps-in-one-proof thesis. Moreover, it cannot derive sup- port, as do the proposals just discussed, from Kant's observations KANT'S TRANSCENDENTAL DEDUCTION 645 about the structure of the proof of the deduction, for they are valid only in the context of the first edition. Kant always allowed so many different trains of thought to influence him in formulating his central arguments that it is never possible to employ his com- ments unless he has explicitly related them to the passage of the text in question. But now, from the propositions of sections 20, 21, and 26, we can develop a proposal which will solve the problem of the two- steps-in-one-proof. Its plausibility stems from the fact that it makes intelligible many peculiarities of the text which must be neglected by all other proposals. Kant obviously attached importance to the fact that the result of the proof in section 20 contains a restriction: he established that intuitions are subject to the categories insofar as they, as intui- tions, already possess unity (B 143) . He indicates this restriction very clearly by writing the indefinite article in the expression "in an intuition" (in Einer Anschauu-ng) with the first letter capitaliz- ed. Norman Kemp Smith, the translator, has misunderstood this hint.Qe believes that Kant wanted to imply that some single intuition was subject to the categories. This interpretation, how- ever, leads to no meaningful emphasis in the course of the proof. Unlike English, in German the indefinite article (ein) and the word unity (Einheit) have the same root. This made it possible for Kant to express through the capital letter not the distinctness of any arbitrary intuition as opposed to others (singularity), but rather its inner unity. The result of the proof in section 20 is therefore valid only for those intuitions which already contain unity. That is: wherever there is unity, there is a relation which can be thought according to the categories. This statement, however, does not yet clarify for us the range within which unitary intuitions can be found. The restriction of the proof in section 20 is then discussed in that part of section 21 which makes reference to section 26. Here Cf. p. 160 of his translation, which shows also that, as a com- mentator, he could not find sense in Kant's text (cf. note 2 above). 646 DIETER HENRICH it is announced that the restriction just made in section 20 will be overcome in the paragraphs of section 26, i.e., the second part of the deduction will show that the categories are valid for all objects of our senses (B 161). And this is what actually takes place. The deduction is carried out with the help of the following reason- ing: wherever we find unity, this unity is itself made possible by the categories and determined in relation to them. In our repre- sentations of space and time, however, we have intuitions which contain unity and which at the same time include everything that can be present to our senses. For indeed the representations of space and time have their origin in the forms of our sensibility, outside of which no representations can be given to us. We can therefore be sure that every given manifold without exception is subject to the categories. At this point the aim of the proof of the deduction has been attained, insofar as the deduction seeks to demonstrate the ztn- restricted validity of the categories for everything which can be meaningfully related to experience. Perceptions, which arise erratically and which cannot be repeated according to determinate rules, would not make intelligible a coherent and systematic knowl- edge of experience. The only conceivable result' of a limited capacity for ordering our sense-data would be a diffuse and dis- continuous sequence of perceptions. It is certainly extraordinary to claim that our capacity for making conscious and thereby unifying our own sensuous repre- sentations should perhaps only be limited. However, its conceiv- ability is an immediate result of the fundamental argument of the whole Critique. It is implied that our consciousness has the peculiarity of being "empty." Everything of which we can become conscious must become accessible to us through media which do not immediately depend on this consciousness. Accord- ing to Kant, it is for this reason that consciousness must be under- stood as an activity, thus always a making-conscious whose neces- sary inner unity causes us to give it the name "I." But this activity always presupposes that something is present in the first place which is to be made conscious. Thus our consciousness can be found only together with a "passive," receptive faculty, which is distinct and in certain respects opposed to the spontaneity of con- sciousness; it can encounter intuitions only as given "before all KANT'S TRANSCENDENTAL DEDUCTION 647 consciousness. " Kant reformulates the task of the transcendental deduction with reference to this very distinction: it must demon- strate that categories are capable of taking up something given into the unity of consciousness. "Appearances might very well be so constituted that the understanding should not find them to be in accordance with the conditions of its unity" (B 123). If that is possible, then it can also be asked whether such a disproportion between consciousness and givenness can be excluded for all or only for part of the given appearances. The difference between these two possibilities also defines the difference between the result of the proof of the first and that of the second step of the deduction .' This question need not recur at every level in the analysis of the conditions of our knowledge. It could be that considerations are possible such as would establish rather quickly that the alter- natives with which the transcendental deduction has to deal are not three-termed but rather only two-termed: that therefore either no sensuous representations or else all sensuous representations are capable of being determined by the categories. Anyone familiar with Kant's work will suspect that Kant had good reason to assert this. But this amounts to saying that Kant also had an alternative way of constructing the proof of the transcendental deduction, other than the one which he actually used in the second edition. For in this construction he takes into account the possibility of a merely partial ability of the understanding to establish unity in the sensible representations. He excluded it only because we do in ' There was a thesis on the Transcendental Deduction by Friedrich Tenbruck (Marburg, 1944) never published, which came close to the con- clusion of this section. Pietro Chiodi, La Deduzione nell'opera di Kant (Torino, 1961), pp. 245 et seq. makes an attempt to bring the problem of the "how" (section 26) with that of the "that" (section 20) into a necessary connection so that together they build one chain of arguments. But one cannot distinguish the two sections on the basis of these two problems. And moreover, Chiodi's account of Kant's intention is highly formal and abstract and cannot be expressed in the language of the Transcendental Deduction itself. 648 DIETER HENRICH fact have unitary representations of space and time and therefore can also unify all representations of sense. Fortunately we can demonstrate that Kant himself was actually conscious of the fact that the transcendental deduction could also be constructed quite differently. His pupil Johann Sigismund Beck undertook in the year 1793 to publish a selection from Kant's writings.Qn the title page he was able to announce that this was being done on Kant's own advice. Kant was interested in making available a competent commentary which could also be used for lectures. But when Beck published the third part of his selections in the year 1796, he considered it necessary to undertake a funda- mental investigation in order to specify the standpoint from which Kant's Critique was actually to be evaluated. He had come to the opinion that the structure of the book promoted a false estimate of Kant's doctrine. Thus it would be necessary to begin with the productive activity of the understanding, in order to avoid the misunderstanding that Kant really wanted to speak of "given con- cepts" and of "objects which affect us." In Beck's opinion all this talk was only an accommodation to traditional doctrine and constituted preliminary concessions for the purpose of an introduc- tion into the system. With this interpretation, Beck approached, somewhat belatedly, Fichte's philosophical conviction. Naturally Kant could not bring himself to approve this. But since he was interested in Beck and in the effect of his writings, he was more willing to consider Beck's proposed alteration of the Critique than was his custom in comparable cases. In a letter to Beck's colleague Tieftrunk, he tried to show approximately what form the Critique might assume in an altered presentation.Yhus we see that Kant himself at one time proposed an alternative to the transcendental deduction of the second edition. It must begin with the doctrine of the categories as rules for the unity of a possible universal consciousness--corresponding to Erlaeuternder Auszug aus Kants krit ischen Schrif ten, vol. 1 (Riga, 1793); vol. 3 (Riga, 1796). Kants gesammelte Schrif ten, ed. Preussische Akademie der Wissen- schaften, Briefwechsel, vol. 3, letter to Tieftrunk 11th Dec. 1797 and the first sketch of this letter in vol. 4, pp. 468 et seq. Further evidence in vol. 5 of Kants handschriftlicher Nachlass in the same edition, reflections 6353 and 6358. KANT'S TRANSCENDENTAL DEDUCTION sections 16-18 of the second edition. Then it must demonstrate that intuitions a priori are presupposed in order that the categories can be applied at all to given sensuous intuitions. This becomes evident, when one considers that the categories can only be con- ceived as operators under which they can be applied. Without such a possibility of application an essential moment of their mean- ing is missing. The meaning of a priori concepts such as the categories can only be accessible a priori. But the only possibility of securing a meaning a priori for the categories is their application to a form of sensi