Hegeler Institute

KANT ON THE MATHEMATICAL METHOD Author(s): Jaakko Hintikka Reviewed work(s): Source: The Monist, Vol. 51, No. 3, Kant Today: Part I (JULY, 1967), pp. 352-375 Published by: Hegeler Institute Stable URL: http://www.jstor.org/stable/27902038 . Accessed: 31/08/2012 16:37
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

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

Hegeler Institute is collaborating with JSTOR to digitize, preserve and extend access to The Monist.

http://www.jstor.org

KANT ON THE MATHEMATICAL

METHOD

Reason

Method

Doctrine in the first chapter of the Transcendental of = B a In Kant this number 713 741) chapter proffers (A of further observations on the subject of the mathematical method. These remarks have not been examined very intensively by most students of Kant's writings. Usually they have been dealt with as

to Kant, "mathematical knowledge is the knowledge According gained by reason from the construction of concepts/' In this paper, I shall make a few suggestions as to how this characterization of themathematical method is to be understood. The characterization is given at the end of the Critique of Pure

a sort of on space and time, appendix to Kant's better known views Transcendental Aesthetic. In in the this paper, I also presented want to call attention to the fact that the relation of the two parts of the first Critique is to a considerable extent quite different from the usual conception of it. come back to Kant's characterization: To the first important term it contains is the word 'construction*. This term is explained by Kant by saying that to construct a concept is the same as to exhibit, a priori, an intuition which corresponds to the concept.2 to the transition is tantamount in other words, Construction, to an intuition which from a general concept represents the recourse to that this done is without concept, provided experience. It is not How is this term 'construction* to be understood? to meet it in a theory of mathematics, for it had in surprising Kant's time an established use in at least one part of mathematics,
lin tions con I shall use the standard of Pure Reason, referring to the Critique = edi second edition A = first edition (1787). All decent (1781), B of one or both of these editions. translations and give the pagination normally Macmillan

ventions

I shall in English, of the first Critique In rendering passages York: and New translation Norman Smith's (London Kemp

follow Co.,

1929).
2Loc. cit.

KANT

ON THE

MATHEMATICAL

METHOD

353

of such constructions which Kant gives in the Transcendental Aesthetic. What guarantee, if any, is there to make sure that the had seen constructions are always possible? Newton geometrical the only foundation of geometrical constructions in what he called But if this 'mechanical practice* (see the preface to Principia). is so, then the certainty of geometry is no greater than the certainty

viz. in geometry. It is therefore natural to assume that what Kant are the con in the passage just quoted primarily has in mind to say seem it may also structions of geometers. And plausible that the reference to intuition in the definition of construction is calculated to prepare the ground for the justification of the use

certain logical arguments which may be completely formalized in terms of modern only reason why Kant thought that logic. The is based on the use of constructions was that con mathematics structions were necessary in the elementary geometry of his day, in most cases almost directly from Euclid's Elementa. derived But this was only an accidental peculiarity of that system of ge set of axioms and ometry. It was due to the fact that Euclid's was incomplete. In order to prove all the theorems postulates to to prove, it was therefore not sufficient for Euclid he wanted a or to set out had a He out diagram carry logical argument. to our geometrical intuition figure so that he could tacitly appeal in this way could supply the missing assumptions which he which

of more or less crude 'mechanical practice*. It may seem natural to furnish a better to intuition is designed that Kant's appeal is no need to the geometrical constructions. There foundation to construct a figure on a piece of paper or on the blackboard, Kant may seem to be saying. All we have to do is to represent the means of imagination. This procedure would be required figure by outcome of the Transcendental the Aesthetic, if this justified by can be accepted. For what is allegedly shown there is that all the are due to the structure of our sensibility geometrical relations reason (our perceptual apparatus, if you prefer the term) ; for this in without can be any imagination they represented completely of sense-impressions. help This interpretation is the basis of a frequent criticism of Kant's It is said, or taken for granted, that con theory of mathematics. structions in the geometrical sense of the word can be dispensed with in mathematics. All we have to do there is to carry out

354

THE

MONIST

of geometry.3 This interpretation, and the criticism based on it, is not with out relevance as an objection to Kant's full-fledged theory of space, Aesthet time, and mathematics as it appears in the Transcendental ic. It seems to me, however, that it does less than justice to the way in which Kant actually arrived at this theory. It does not take a sufficient account of Kant's precritical views on mathe matics, and it even seems to fail to make sense of the arguments it by means of which Kant tried to prove his theory. Therefore does not give us a chance of expounding Kant's real argu fully ments for his views on space, time and mathematics, or of criticiz ing them fairly. It is not somuch false, however, as too narrow. We begin to become aware of the insufficiency of the above examine the notion of construction interpretation when we somewhat more closely. The definition of this term makes use of

had omitted. Kant's it is thus alleged, theory of mathematics, arose by taking as an essential feature of all mathematics some a was a which of defect in Euclid's par consequence only thing ticular axiomatization

the notion of intuition. We have to ask, therefore: What did Kant mean by the term 'intuition'? How did he define the term?What is the relation of his notion of intuition to what we are accustomed to associate with the term? The interpretation which I briefly sketched above assimilates Kant's notion of an a priori intuition to what we may call mental pictures. Intuition is something you can put before your mind's eye, something you can visualize, something you can represent

to your imagination. This Kant is not at all the basic meaning to give to the word, however. According himself wanted to his definition, presented in the first paragraph of his lectures on logic, from general concepts is an every particular idea as distinguished intuition. Everything, in other words, which in the human mind
3A tion paradigmatic to Mathematical statement of this view occurs in Bertrand Russell's Introduc

Allen and Unwin, (London: George 1919) , that the geometers of his day could not prove p. 145: "Kant, having but required an appeal to the figure, their theorems by unaided arguments, to which invented a theory of mathematical the inference reasoning according is never strictly logical, but always requires the support of what is called 'intui to say, there does not seem to be a scrap of evidence tion'." Needless for attrib Philosophy observed uting to Kant the 'observation* Russell mentions.

KANT represents nothing an

ON THE

MATHEMATICAL is an

METHOD

355

individual

'intuitive* about

intuitions

intuition. There so defined.

is, we may say, Intuitivity means

simply individuality.4 Of course, it remains true that later in his system Kant came to make intuitions intuitive again, viz. by arguing that all our human intuitions are bound up with our sensibility, i.e., with our faculty of sensible perception. But we have to keep in mind that this connection between intuitions and sensibility was never taken a as mere Kant by logical consequence of the definition of intuition. On the contrary, Kant insists all through the Critique of Pure Reason that it is not incomprehensible that other beings might have intuitions by means other than senses.5 The connection be tween sensibility and intuition was for Kant something to be not to be assumed.6 The proved, something proofs he gave for case the connection the of human beings) are pre assuming (in sented in the Transcendental Aesthetic. Therefore, we are entitled to assume the connection between sensibility and intuitions only in those parts of Kant's system which the Transcendental Aesthetic. are logically posterior to

of

My main suggestion towards an interpretation of Kant's theory the mathematical method, as presented at the end of the first Critique, is that this theory is not posterior but rather systematic Aesthetic. If so, it follows that, ally prior to the Transcendental within this theory, the term 'intuition* should be taken in the 'unintuitive* sense which Kant gave to it in his definition of the
10; Critique of Pure Reason are given by H. references reinen Vernunft (Stuttgart:

W.

4 See of 1770, section 2, e.g. Kant's Dissertation B 376-377; Prolegomena 8. Further 320 = zu Kants Kritik in his Commentar der Vaihinger A zum Spemann, 1881-1892), leichteren Gebrauch Vol. der

on 1798) Anschauung. 5 "We cannot assert of sensibility that it is the sole possible kind of intuition" B 43, A 34-35 = B 51, A 42 = B 59, A 51 = (A 254 = B 310). Cf. e.g. A 27 = "uns Menschen B 75 and the characteristic at B 33. phrase wenigstens" 6 The remarks of the Transcendental opening hard-and-fast connection between all intuitions out, however, Aesthetic and seem As

2, pp. 3, 24. Cf. also c.ch. E. Schmid, W rterbuch Kantischen Croker, Schriften (4th ed., Jena:

sensibility. to be taken as a statement of what Kant wants they have partly to prove. See H. J. Paton, Kant's Metaphysic of Experience (London: George Allen and Unwin, 1936), Vol. I, pp. 93-94.

to envisage a Paton points

356
notion. based

THE

MONIST

as In particular, Kant's of mathematics characterization on the use of constructions has to be taken to mean merely one is all the time introducing particular that, in mathematics, of general concepts and carrying out arguments in representatives terms of such particular representatives, arguments which cannot be carried out by the sole means of general concepts. For if Kant's is independent of his proofs for con methodology of mathematics necting intuitions and sensibility in the Aesthetic and even prior to it, then we have, within Kant's theory of the method of mathe* for assuming such a connec matics, no justification whatsoever tion, i.e., no justification for giving the notion of intuition any meaning other than the one given to itby Kant's own definitions. that the There are, in fact, very good reasons for concluding discussion of themathematical method in the Doctrine ofMethod is prior to, and presupposed by, Kant's typically critical discussion Aesthetic. One of them of space and time in the Transcendental in the work in which Kant should be enough: in the Prolegomena, wanted to make clear the structure of his argument, he explicitly at of mathematics appeals to his discussions of the methodology

the end of the Critique of Pure Reason in the beginning and dur to Aesthet which the Transcendental the argument ing corresponds the dependence of the latter on the former ic, thus making explicit. This happens both when Kant discusses the syntheticity of mathematics (Academy edition of Kant's works, Vol. 4, p. 272)

and when he discusses its intuitivity (ibid. p. 281; cf. p. 266). Another persuasive reason is that at critical junctures Kant in Aesthetic means by intuitions precisely what the Transcendental his own definitions tell us. For instance, he argues about space as follows: "Space is not a. .. general concept of relations of things in
general, but a pure intuition. For ... we can represent to ourselves

. . . in it, only one space. Space is essentially one; the manifold and therefore the general concept of spaces, depends solely on the it follows that an . . . intuition introduction of limitations. Hence underlies all concepts of space" (A 24-25 = B 39). Here intuitivity is inferred directly from individuality, and clearly means nothing more than the latter. But I am afraid that, however excellent reasons there may be for reversing the order of Kant's exposition and for putting the discussion of mathematics in the first Critique in the Methodenlehre

KANT

ON THE

MATHEMATICAL

METHOD

357

in concreto, i.e., in the form of individual instances, was the elaborate views on mathematics. starting-point of Kant's more or not my suggested reading of Kant's characterization Whether of mathematics is exhaustive or not, that is, whether or not intu ition there means something more than a particular idea, in any case this reading is the one which we have to start from in trying to understand Kant's views on mathematics. It is useful to observe at this point that the reading of Kant which I am suggesting is not entirely incompatible with the other, more traditional, interpretation. On one hand, a fully concrete mental picture represents a particular, and therefore an intuition cept in the sense of the wider definition. On the other hand, particular instances of general concepts are usually much easier to deal with than general concepts themselves; they are much more intuitive in the ordinary sense of the word than general concepts. The
7This writings has been out clearly and forcefully by E. W. Beth, to whose brought I do not I am greatly indebted, although fully share Beth's of Kant's evaluation der Urteile theories. See "Kants Einteilung voor Wijs und Nederlands Algemeen Tijdschrift synthetische," et la (Am

before the Transcendental Aesthetic, my readers are still likely to be incredulous. Could Kant really have meant nothing more than this by his characterization of the mathematical method? Could he have thought that it is an important peculiarity of the as distinguished method of mathematicians from the method of that the mathematicians make use of special cases philosophers of general concepts while philosophers do not? Isn't suggesting this to press Kant's definition of intuition too far? answer to this is, I think, that there was a time when The Kant did believe that one of the main peculiarities of the mathe matical method is to consider particular representatives of general 7 concepts. This view was presented in the precritical prize essay of the year 1764. Its interpretation is quite independent of the interpretation of Kant's critical writings. In particular, the formu lation of this precritical theory of Kant's does not turn on the notion of intuition at all. It follows, therefore, that the idea of themathematical method as being based on the use of general con

on Kant

philosophical in analytische begeerte logique sterdam:

en La 46 crise de la raison 253-264; Psychologie, (1953-54), oj Mathematics 1957) ; The Foundations (Paris: Gauthier-Villars, North-Holland 41-47. pp. Company, 1959), Publishing

358

THE

MONIST

In fact, if we have a closer look as presented at the end mathematics we shall see that many things Reason, in mind the notion of intuition as a distinction to general concepts. Usually,

as may two interpretations therefore don't disagree as widely two is the makes difference between first seem. What the really whether Kant sometimes had in mind, in addition to 'usual* in tuitions in the sense of mental pictures or images, some other that are actually used in mathematical individuals arguments. I This, think, is something we must make an allowance for. theory of of the Critique of Pure we become natural if keep particular idea in contra actual at Kant's

theory people read Kant's of the mathematical in the light of what he says in the method In other words, they read 'intuition* Aesthetic. Transcendental as if itmeant 'mental picture* or 'an image before our mind's eye* or something of that sort. But then it becomes very difficult to understand why Kant refers to algebra and to arithmetic as being based on the use of intuitions. The point of using algebraic symbols is certainly not to furnish ourselves with intuitions in the ordinary sense of the word, that is, its purpose is not to furnish ourselves with more vivid images or mental pictures. Scholars have tried to reconcile Kant's remarks on algebra and arithmetic with his critical doctrines as they are presented in the Transcen dental Aesthetic. The outcome of these attempts is aptly summed in a well-known essay on up, I think, by Professor G. D. Broad "Kant's where

and Philosophical of Mathematical Reasoning," Theory of he says that "Kant has provided no theory whatsoever we correct in This is if my reasoning."8 algebraic opinion quite read Kant's description of the mathematical method in the light Aesthetic. But then Broad's of what he says in the Transcendental view becomes, it seems to me, almost a reductio ad absurdum of Aesthetic is, in Kant's the assumption that the Transcendental at the end mind, logically prior to the discussion of mathematics of the first Critique. For on this assumption the statements Kant makes on arithmetic and algebra are not only deprived of their Aesthetic If the Transcendental truth but also of their meaning. were logically prior to Kant's methodology of mathematics, it would become entirely incomprehensible what on earth Kant could have
s Proceedings of the Aristotelian Society, 42 (1941-42), 1-24.

KANT

ON THE MATHEMATICAL

METHOD

359

meant by his remarks on arithmetic and algebra which so obviously are at variance with his professed theories. On the other hand, if we assume that by 'intuition* Kant any representative of an individual when he com only meant on and algebra, a number of things, although arithmetic mented

and division, constructions. For what hap addition, multiplication pens when we combine in algebra two letters, say a and b, with a functional sign, be this / or g or -f. or . or:, obtaining an expres b or a . b or a : b} These expres sion like / (a,b) or g (a,b) or a + sions, obviously, stand for individual numbers or, more generally, for individual magnitudes, usually for individuals different from those forwhich a and b stood for. What has happened, therefore, is that we have introduced a representative for a new individual. And such an introduction of representatives for new individuals, i.e. new intuitions, was just what according to Kant's definition happens when we construct something. The new individuals may be said to represent the concepts 'the sum of a and &*, 'the product of a and b', etc. therefore receive a natural meaning under my interpretation, quite apart from the question whether thismeaning is ultimately reconcilable with what Kant says in the Kant's remarks on algebra

not necessarily everything, becomes natural. If we can assume that the symbols we use in algebra stand for individual numbers, then it becomes is based on the trivially true to say that algebra use of intuitions, i.e., on the use of representatives of individuals as distinguished from general concepts. After all, the variables of elementary algebra range over numbers and don't take predicates of numbers as their substitution values as the variables of a form we can also understand what alized syllogistic may do. Then Kant had in mind when he called algebraic operations, such as

Transcendental Kant's

Aesthetic. We might of say that the purpose use of the term 'intuition* here is to say that algebra is sense: the only acceptable values of vari in Quine's nominalistic ables are individuals. Kant's present a somewhat more com I not shall deal with them fully here, although plicated problem. can be shown to square with the view I am they suggesting. is only one point that I want tomake here. remarks on arithmetic

There

360
and

THE

MONIST

12 by means of points, fingers, or some other suitable illus trations so that we can immediately perceive the desired equation. He goes as far as to say that equations like 7+ 5= 12 are immediate = B and indemonstrable not 164 easy to reconcile (A 204). This is a Kant with the fact that nevertheless described procedure which serves,whether we call it a proof or not, to establish the truth of the equation in question and that he said that his view ismore natural as applied to large numbers (B 16). I hope to be able to show meant later what Kant 12 by saying that equations like 7 -f- 5 = are 'immediate* and 'indemonstrable*. He did not mean that the an argument which we are equation can be established without a likely to call proof. 'Immediate* and 'indemonstrable* did not serve to distinguish from an articulated immediate perception argument, but to distinguish a certain subclass of particularly and ordi straightforward arguments from other kinds of proofs. The nary interpretation of Kant's theory therefore fails here too. Of the correct view I shall try to give a glimpse later. One good way of coming to understand Kant's theory of mathe matics is to ask: What were the paradigms on which this theory was modelled? The most obvious paradigm, and in fact a paradigm system of elementary ge recognized by Kant himself, was Euclid's ometry.9 In the beginning of this paper, we saw that a usual is based on a comparison criticism of Kant's theory of mathematics and Euclid's between Kant's It seems to me, how system. theory ever, that it is not enough to make a vague general comparison. It is much more useful to ask exactly what features of Euclid's Kant was interpretation of Kant's

In the case of the arithmetic of small numbers, such as 7, 5, 12, the ordinary reading of Kant's remarks is not without seems to be saying is that in order to Kant plausibility. What establish that 7 -f- 5 = 12 we have to visualize the numbers 7, 5,

presentation

thinking of in his theory. In view of the notion of intuition that I have suggested, Is there anything particular in Euclid's the question becomes: the that which idea is based mathematics encourages procedure on the use of instances of general concepts? particular
9 See Elementa, Books Edition of Kant's the Academy works, see Sir Thomas Heath's translation Elements (Cambridge: Vol. and the 307. Concerning The Thirteen commentary 2, p. University Press, 1926).

of Euclid*s

Cambridge

KANT

ON THE

MATHEMATICAL

METHOD

361

It is easy to see that there is. For what is the structure of a a proposition consists of five (or in Euclid? Usually, proposition sometimes six) parts.10 First, there is an enunciation of a general he 20 of the Elementa proposition. For instance, in proposition are manner two in sides taken together any says: "In any triangle greater than the remaining one." This part of the proposition was
Called

was called the setting-out or ecthesis proposition It is that in Latin perhaps no accident (cicOeais, expositio). in Kant used the German for setting-out (darstellen) equivalent term and that used the he his notion of construction, explaining con to that of mathematical exposition for a process analogous Euclidean
struction.

For instance, after having enunciated proposition 20, Euclid goes on to say: "For let ABC be a triangle. I say that in the triangle are greater than two sides taken together in any manner ABC the remaining one, namely, BA, AC greater than BC; AB, BC greater than CA; BC, CA greater than AB." This part of a

But Euclid never does anything on the basis of the enunciation alone. In every proposition, he first applies the content of the enunciation to a particular figure which he assumes to be drawn.

the irp Tacn*;.

example, the preparation reads as follows: "For let BA be drawn through the point D, let DA be made equal to CA, and let DC be joined." The construction was followed by the apodeixis or proof proper Tr Sei is). In the proof, no further constructions were carried out. ( a series of inferences were drawn con What happened was that

The setting-out or ecthesis is closely related to the following or third part of a Euclidean proposition. This part was called the or in stating machinery preparation (Karao-Kemj). It consisted that the figure constructed in the setting-out was to be completed lines, points, and circles. In our by drawing certain additional

(2) of the earlier propositions, and (3) of the properties of the figure which follow from the way in which itwas constructed. After having reached the desired conclusion about the par
io Heath, op. cit., Vol. 1, pp. 129-131.

cerning the figure introduced in the setting-out and completed in the 'preparation*. These inferences made use (1) of the axioms,

362

THE

MONIST

Kant explains his theory of the mathematical method. says that the superiority of the mathematical method over the one in geometry lies in the fact that the mathe philosophical can draw actual figures and carry out proofs in terms matician of such figures. For instance, if a philosopher (qua philosopher) tries to prove that the sum of the internal angles of every triangle is equal to two right angles, he is reduced, Kant says, to analysing to the concepts 'straight line*, 'angle*, and 'three*, and is unable in contrast, can draw a figure of get anywhere. A mathematician, a triangle, complete it by means of suitable additional construc tions (i.e., introduce suitable new lines, circles, etc. into the argu to be proved obvious. and thereby make the proposition ment) n = B A 716-717 (See 744-745.) to the setting-out or shows that, in addition This example Kant the also had in mind ecthesis of a Euclidean proposition, of which He part of the proposition which follows the ecthesis, viz. the prepa ration or 'machinery*. Setting-out and preparation were the two parts of a Euclidean proposition where constructions in the usual sense of the word were made; and we have seen that these two parts were also the ones in which constructions in Kant's abstract sense of the word were needed, i.e. where new individual points, lines, etc. were introduced. This, then, means that within geometry Kant's
the

returned to the general enunciation again, ticular figure, Euclid saying, e.g. "Therefore, in any triangle, etc." is compared When this structure of a Euclidean proposition account the agreement with Kant's of the mathematical method, is obvious. Kant's idea of geometry was, itmay be said, Euclidean in more than one sense of the word. When Kant says that it is the method of mathematicians always to consider general con the cepts in concreto, in a particular application, he has in mind setting-out or ecthesis of a Euclidean proposition where a general is 'exhibited* or 'set out* by means of a geometrical proposition is borne out by the examples by means particular figure. This

notion

of construction

coincides with

the ordinary usage

of

term

'construction*.

il We does matics, not

can lie

but

to Kant the peculiarity of mathematics according and postulates of the different branches of mathe of argumentation and demonstration. in the mathematical mode that in the axioms

see here

KANT This

ON THE MATHEMATICAL

METHOD

363

outcome of our comparison between Kant and Euclid supports what was said earlier. It shows that Kant's notion of a as a special case the usual geometrical construction accommodates notion of construction. Now the constructions of the geometrical kind need not take place in the human mind. More often than not, they are carried out on a piece of paper or on the blackboard. is common to all such constructions is that some new lines, What

points, or circles are introduced. If these geometrical entities are conceived of as individuals, they fit into Kant's general definition of an intuition. There is no need, therefore, to assume that the constructions of geometry mean for Kant something else than what we are prepared to call constructions. But this is not all we can get out of the comparison. If we have a somewhat closer look at the relation between Kant's theory of the mathematical method and Euclid's practice, the relation serves to suggest several insights into Kant's theory. Here I shall only mention a few of them.

is in geometry an ancient distinction between two (1) There kinds of methods. There is, on one hand, the method of assum a to desired result be achieved, for instance, of assuming that ing we have succeeded in making a desired construction, in the ordi nary sense of

'construction*. From these assumptions one then argues 'backwards', so to speak, to the conditions on which this construction is possible and to the ways in which it can be ef fected. This is called the analytic method. It was sometimes ascribed to Plato, but itwas not to be employed explicitly and systematically in a large scale until the analytic geometry of Descartes, the very name of which is derived from the 'analytic* method in question. The other method was the synthetic one. In applying it one tries to effect the desired result, for instance, to make a desired con

struction, by actually carrying out constructions. What distinguishes the two methods, therefore, is broadly speaking the fact that in the analytic method no constructions are made while the synthetic method is based on the use of actual constructions.12
12 We which are sometimes used have one to realize, however, in an at the could that analysis of the mere and difference in a of the directions in

is proceeding emphasized or not. One

expense thus

synthesis, respectively, was the questions whether constructions between a 'directional* and a

distinguish

364
Kant indicates geometry in particular of constructions. We

THE

MONIST

that what makes mathematics in general and use the is of intuitions, i.e., the use synthetic have seen that his notion of construction

coincides, in geometry, with the ordinary mathematical usage of term this is Kant's What that the 'construction*. means, then, and within is distinction between modelled, synthetic analytic mathematics at least, on a usage of mathematicians which was cur

by means of the introduction of new geometrical entities. This from the other usage which was based on the he distinguished of proceeding 'inversely* from a ground to a consequence. paradigm This difference is stated by Kant, if not in so many words, in a footnote to the first paragraph of his Dissertation of the year 1770.13 is another way in which an awareness of the respec (2) There and Descartes helps us to understand tive geometries of Euclid can make a particularly if we Kant. We interesting observation to with Descartes'. Euclid's Descartes, geometry According compare themain idea of the analytic geometry was a correlation or analogy between algebraic and geometrical operations. Just as all we need in arithmetic are the four or five basic operations of addition, sub
'constructional* paper, "Kant (or 'problematic') the Tradition and sense of analysis and Cf. my synthesis. in Deskription, Existenz of Analysis," und Pustet, (Munich: 1966).

methods of finding a desired proof or construction, or, in some cases, to separate two methods of exposition. What Kant needed was a distinction between two different means of carrying out a proof. For him, the paradigm of synthesis was precisely synthesis in the geometrical sense of the word, i.e. the completion of a figure

rent at his time. (Mathematicians to-day still speak of synthetic geometry, meaning geometry which turns on the use and study of geometrical constructions.) This suggestion is supported by Kant's own comments on the subject, which serve to narrow down his sense of synthetic so as to connect it explicitly with constructions sense. The distinction between analytic in an almost geometrical and synthetic in geometry was earlier often used to separate two

t, ed. P. Weingartner Analytizit 13 Cf. also, 5 (Academy Edition, Vol. 4, p. 276, footnote). We Prolegomena, can also say that Kant's remarks in effect serve to distinguish between the di sense of and rectional and the constructional (problematic) analysis synthesis, and to indicate that Kant opts for the latter. (See the article mentioned the preceding footnote and there.)

KANT

ON THE

MATHEMATICAL

METHOD

365

geometrical constructions. This gives, I think, the key to what Kant means by saying that simple arithmetical equations, such as 12 are 'immediate* and 'indemonstrable*. We see this ifwe 7 -j- 5 = 5= 12 is verified try to cast the argument by means of which 7 + into the form of a Euclidean proposition. Because of the analogy

and the extraction of roots, division, traction, multiplication, same we in need in geometry only a few basic the way exactly we are interested in here is What constructions, Descartes says.14 the analogy between algebraic and geometrical operations, in par ticular the fact that algebraic operations correspond to certain

and geometrical constructions, the between algebraic operations actual addition of 7 and 5 corresponds to the third stage, i.e., the or 'machinery*, of a Euclidean proposition. Kant's preparation to the numbers 7 and also show that, him, according explanations 5 must somehow be 'set out* or 'exhibited* before the actual opera to the 'ecthesis* of a Euclidean tion of addition, in analogy on "points or fingers" illus is what remarks his (This proposition. to But the proof proper, to the what, then, corresponds trate.) we to that have do in order to show that all apodeixis} Obviously, 5= 12 is to carry out the operation of addition; the proof 7+ to the mere observation proper is reduced to a mere minimum, that the result of the addition

of his to Schultz, dated November 25, 1788. The main difference is that, instead of using the terminology which pertains to the theorems of the Euclidean geometry, Kant in this letter makes use of the parallel terminology pertaining to geometrical problems. This is important, I think, over and above the interpretation of particular passages, for it shows how Kant intended the intu immediacy of arithme itivity of arithmetic to be understood. The 5 tical truths is not due to the fact that simple equations like 7 + = 12 are perceived to be true and not argued for, but to the fact
14 See La G om trie, the first few statements (pp. 297-298 of the first edition).

equals the desired result 12. In a one can say that no sense, therefore, perfectly good proof (prop no to that 7 -f- 5 = is , needed establish 12. This er) apodeixis in 'immediate* and the is 'indemonstrable* precise sense equation that it can be established by the mere preparation or 'machinery* of a Euclidean is all that Kant's statement amounts proof. This to. And the fact that this really was Kant's idea is shown by a letter

366

THE MONIST

that the only thing we have to do in order to establish such serves to explain is to carry out the computation. This equations account more Kant of said the his is why equations readily under stood in the connection of large numbers (B 16; cf.A 78 = B 104). (3) I suspect that a particularly perplexing passage in the first in the same Critique receives a natural explanation pretty much mean as on I statement arithmetic. the remarks the Kant makes way in B 14 to the effect that all the inferences (Schl sse) of the mathe are based on the principle of contradiction "which the maticians nature of all apodeictic certainty requires." This passage becomes very natural ifwe take Kant for his word and understand him as referring solely to the apodeictic or 'proof proper* part of a Euclid ean proposition. Taken is literally, the proof proper or apodeixis after all the only part of a Euclidean proposition where inferences are drawn. Taken in this way Kant's statement expresses precisely what he would be expected to hold on my interpretation, viz. that the distinction between on one hand apodeixis and on the other

hand ecthesis and the auxiliary construction separates the analytic and the synthetic parts of a mathematical argument. We have seen that in Kant's What have we accomplished so far? mathematical of the method, theory presented towards the end of the first Critique, one has to keep in mind the possibility that by intuitions Kant means particular representatives of general con cepts. We have seen that a number of things about Kant's theory of algebra, arithmetic, and point of view. But, itmay be are not sensible is ruled out argues there that all the use

from this geometry become natural said, the possibility of intuitions which in the Transcendental Aesthetic. Kant of intuitions in mathematics is based on the intuitions of space and time, and that these intuitions are due to the structure of our sensibility. There is, therefore, no room for intuitions that are not connected with left in mathematics

sensibility. I have no desire to deny that this is what Kant says. But I want to point out that the disagreement between the above interpreta of mathematics and his theory of tion of Kant's methodology Transcendental and the time in Aesthetic does not disprove space two parts of between The the my interpretation. discrepancy

Kant's system belies my reading of Kant only if the account of Aesthetic is correct. Kant mathematics given in the Transcendental

KANT claims

ON THE MATHEMATICAL

METHOD

367

can only be there that the use of intuitions in mathematics are due to our if we assume that all these intuitions understood now are intuitions, say the individual variables sensibility. If there or 'intuitions' of algebra, which have no relation to our sensi

stances was the starting-point of Kant's better-known theory that are due to our sensibility. all the intuitions we use in mathematics What is there in the notion of an intuition as an individual instance which made Kant think that this conclusion is inevitable? the r le of intuitions, in the sense of repre We have discussed sentatives for individuals, in algebra, in arithmetic, and in geom is the common feature of these uses which can only etry.What to Kant, by assuming that the mathe be explained, according is the common denominator matical intuitions are sensible? What

But then it remains to be explained how Kant came to enter tain the mistaken doctrine. I have implied that the notion of the as being based on the use of individual in mathematical method

bility, then the only possible conclusion is not that these alleged sense. The other intuitions are not intuitions at all in Kant's are Kant to that intuitions but that is say they genuine possibility just was wrong in saying that all the intuitions used inmathematics are sinnlich, i.e. due to our sensibility.

Aristotelian

of all themathematical 'constructions' we have discussed? me to seems that a natural generalization It is virtually (4) contained in the above analysis of Euclid's propositions. The most important part of a Euclidean proposition which is intuitive in sense is the setting-out, the ecthesis. Now Kant's this notion of ecthesis occurs not only in Greek geometry. It also occurs in the logic. Aristotle never explains explicitly what the pro cedure called ecthesis is, but we can perhaps say that it was a step in which Aristotle moved from considerations pertaining to a gen to considerations

pertaining to a particular repre sentative of this general term. For instance, in An. Pr. I, 2, 25al5 Aristotle seems to argue as follows: If no A is a B, then no B is an a particular b of this A. For if not, then some .B's are A's. Take eral term over kind. This and a B

particular b has both the property B and the property A shows, therefore, that it is impossible that none of the A's is as we assumed. This contradiction proves the conclusion. A seems to indicate that Aristotle (An. Pr. I, 41, 49b33ff.)

later passage

368
took the logical
metrical one.15

THE MONIST ecthesis to be essentially the same as the geo

I suggest that this notion of ecthesis offers a very good recon struction of Kant's notion of construction, i.e., of the notion of the exhibition of a general concept by means of particular representa tives. It agrees, as we see, very well with the way in which Kant defines the notion of construction. Its use in the Aristotelian logic may perhaps explain why Kant criticized (in the essay on the "false subtlety of the four syllogistic figures") certain parts of this logic. He went as far as to reject, in effect, all the syllogistic modes ex cept the first two modes of the first figure. The explanation may perhaps lie in the fact that the particular application of ecthesis I just outlined was calculated to prove one of the rules of conver sion which Aristotle needed in order to reduce all the syllogistic modes to the first two modes. Since the use of ecthesis was for Kant method of reasoning, he could not use typically mathematical it in logic in the way Aristotle did. For this reason, Kant could not to the two modes of Barbara reduce all the syllogistic modes and Celarent which he recognized as the basic ones, and was bound to reject all the others as being 'impure* and 'confusing*. The notion of ecthesis can be made precise in terms of modern in effect, identical with one of the most logic.16 It becomes,
15 Prior the notion of ecthesis Concerning A Revised and Posterior Analytics: in Aristotle, Text with see W. D. Ross, Aristotle's and Commen

Introduction

Clarendon Press, 1949) pp. 32-33, 412-414; Jan Lukasiewicz, tary (Oxford: The Formal Aristotle's of Modern from the Standpoint Logic (Oxford: Syllogistic The Clarendon Press, 1951), pp. 59-67; G nther Patzig, Die Aristotelische Syl & Ruprecht, 1959), pp. 166-178; B. Einarson, (G ttingen: Vandenhoeck logistik Terms American in Aristotle's "On Certain Mathematical Journal of Logic," Philosophy discussions, used in his may on here) 34-54, 151-172, esp. p. 161. As will be seen from these (1936), notion of the Aristotelian of ecthesis the precise (as interpretation an unambiguous to which solution is a controversial problem logic) I prefer I shall rely The be unavailable. (and which interpretation which 57

of modern instantiation" assimilates logical ecthesis to the "existential as fully here as it deserves. For for this interpretation argue logic. I cannot seems to me to be closely use of the term ecthesis in geometry, which Aristotle's An. Post. to the logical related I, 41 49b30-50a4; I, 10, ecthesis, cf. e.g. An.Pr. 76b39-77a2. 161 am here For footnote. presupposing further remarks Philosophical the on mentioned in the preceding interpretation cf. my paper, "Are Logical this interpretation, Beth's Review 74 (1965), 178-203, and E. W.

Truths

Analytic?,"

KANT

ON THE

MATHEMATICAL

METHOD

369

of his attempt to connect the mathematical method with sensibility. It was already suggested that the notion of construction may perhaps be identified with certain methods of proof in modern logic. If this is so, then Kant's problem of the justification of constructions in is not made obsolete by the formalization of geometry mathematics and other branches of mathematics. The distinction between intu itive and nonintuitive methods of argument then reappears in the formalization of mathematical reasoning as a distinction between

important rules of inference of quantification theory (existential instantiation). And, in terms of the notion of ecthesis so recon 5= 12 can be strued, we can see in what sense the equation 7 + said to be based on the use of particular representatives of general concepts, i.e., on the use of ecthesis. It would take us too far, how ever, to go into this question here.17 I shall conclude this paper by sketching very briefly and in terms how the reconstruction of Kant's notion of con un-Kantian sense struction in terms of ecthesis or in some similar way makes

two different means of logical proof. But does there remain any sense in which the use of such 'intuitive* methods is problematic? Would Kant have accepted such a reconstruction of the notion of intuition as a premise of his argument that all intuitions are due to our sensibility?

The answer to the questions is, I think, yes.We can see why it was natural for Kant to make the transition from the use of indi vidual instances of any kind to their connection with sensibility. I shall briefly outline two explanations. itmay be said, nothing was more natural for Kant Historically, than to connect individuals with the use of our senses. Aristotle already held that "it is sense-perception alone which is adequate for grasping the particulars" (An. Post. I, 18, 81b7). All knowledge, is obtained by means of particulars, must be therefore, which natural How of the this general Aristotelian perceptual. application
discussion and Formal demie of the relation Derivability," of ecthesis Entailment logic in "Semantic de Koninklijke Nederlandse Aka 13 (Amsterdam, N.R., Letterkunde, 18, no. van "Kant Vindicated," Pustet, and modern

van Afd. Wetenschappen, 1955), pp. 309-342. 17 Some remarks on these in my paper, points are contained Existenz und Analytizit in Deskription, t, ed. P. Weingartner

Mededelingen

1966).

(Munich:

370

THE MONIST

idea to the case of constructions in Kant's sense was, is perhaps shown by the fact that Alexander the Commentator already applied Aristotle's idea to the process of ecthesis. Alexander held that the term in introduced the ecthesis is given by perception, and singular that the proof by ecthesis therefore consists in a sort of perceptual evidence.18 And the general Aristotelian assumption about indi viduals and senses was echoed by Kant's German predecessors. Another, and perhaps a more important way of making Kant's

purely analytic. Now the distinction between these parts of a Eu clidean proposition corresponds, according to a widespread view of which Kant seems to have accepted, to a distinction between two kinds of principles of Euclid's system. The principles of construc tion are the so-called postulates, while the principles of proof prop er are called axioms It is significant that the (common notions). examples Kant gives of analytic principles used in geometry (B 17) obviously fall into the second category. (This shows, incidentally, that Kant's notion of construction in geometry was not, as sometimes has been suggested, something alien to the axiomatic treatment of geometry. The very examples Kant gives of geometrical construc tions are based either directly on Euclid's postulates, or else on explicit propositions Euclid has proved earlier; a fact of which Kant scarcely could have been unaware. In point of fact, themain
In Aristotelis I Priorum Librum of Aphrodisias, Analyticorum in Commentaria in Aristotelean ed. M. Wallies, Graeca, Vol. 2 104; Lukasiewicz, (a) 1883), p. 32, cf. pp. 32-33, 99-100, op. cit. pp. (Berlin and to justify the mathematical to explain ecthesis from an 60-67. An attempt ian doctrines. matical of Kant point of view also easily gives rise to striking anticipations to Theophrastus Thus we find, for instance, that according mathe us act of in "seem to have been, as it were, devised in the objects by figures . . ." and 18Alexander

ideas plausible may be derived from the division of the Euclidean into parts. We have seen that for him, the use of propositions constructions took place in the second and the third part of a Euclidean in the fourth part the argumenta proposition, while tion was purely nonconstructive or, which amounts to the same,

Commentarium,

Aristotelian

to have no nature in shapes and ratios, and 4al8 308-309 ff., pp. (Theophrastus, Metaphysica Plato's Philosophy Cf. also Anders Wedberg, of Mathematics (Stock Brandisii). & Wiksell, that Aristotle likewise holm: Almqvist 1955), p. 89, who emphasizes some of the most salient features of Kant's seems to anticipate theory of mathe vesting things with and of themselves matics.

KANT

ON THE MATHEMATICAL

METHOD

371

in Kant's construction needed favourite example, the theorem about the internal angles of a triangle, is based on the postulate of parallels which Kant himself had tried to prove.) Hence,

the distinction between intuitive and logical ways of was for Kant, within geometry at least, equivalent with reasoning the distinction between the use of postulates, i.e., principles of use and the of construction, axioms, i.e., principles of proof. What, the latter constitutes distinction? then, According to a wide-spread view which may be traced back to Aristotle and certainly back to the Greeks, postulates are assumptions of existence. Kant's problem of the justification of constructions, therefore, amounts to the prob lem of justifying the use of existential assumptions inmathematics. Stated in this form thewhole problem may seem spurious. There is certainly nothing that could prevent a mathematician from studying
assumptions.

axiom

systems which

incorporate

general

existential

pure mathematics (This appears particularly only. clearly from cf. Vaihinger's discussion of paragraphs 8-9 of the Prolegomena; these paragraphs.) We may ask: what happens when we apply to argument in the course of which a reality a particular mathematical i.e., a general existential assumption, has been used? postulate, In applying it, we have to introduce a representative for a new individual, as Kant puts it "without any object being present, either previously or now, to which it could refer." The introduction of the new representative for an individual is carried out a priori. The existence of the individual object in question, in other words, is not given by experience. Kant describes the situation by saying that the intuition or, in our terms, the representative for an indi vidual object precedes its object. The only thing tomake sure that there is any object at all corresponding to the representative is the general existential assumption. But it may seem as if there is no

sense if we are concerned with the problem only makes of mathematical reasoning to reality. But this certain applicability was is Kant concerned with in the Transcendental ly something in spite of the fact that he insists that he is speaking of Exposition, The

general justification for the application of existential assumptions at all unless we are in fact acquainted with the objects that are assumed to exist, which simply is not the case with applications of our a priori knowledge. It seems, as Kant puts it, impossible to

372

THE MONIST

intuit anything a priori. For in the absence of actual acquaintance there is in reality nothing to make sure that we can always find objects which the representatives we have introduced really stand for or that they have the properties we expect them to have.19 Kant's solution of this (real or apparent) problem consists in saying that there is one and only one case in which we can be sure that the individuals we have assumed to exist really do so and have the desired properties. This is the case in which we have ourselves created the objects in question or ourselves put the desired prop

erties and relations into them.20 And he seems to think that there is only one stage of our coming to be aware of objects in which this kind of 'putting properties into objects' can take place. Or, rather, there is only one stage in which we can 'put properties' into all (individual) objects. This stage is sensible perception. For sensible perception is the only way in which an individual object can 'make its way' into our consciousness. Outer sense is the only way in which we can become aware of external objects. For this reason, it is the only stage of our coming to know objects at which we can ourselves give spatial relations to all external objects. There in geometry must be due to fore, the spatial relations postulated the structure of our outer sense.
difficulty was emphasized by Kant's early critics. For instance, J. G. E. writes in his long paper, "Ueber die transscendentale Philo Aesthetik," 1 (1788), Kant's of an notion 117-149, as follows, apropos sophisches Magazin a priori ich (I) die Bemerkung intuition: kann nicht vorbeglassen, "Hierbey . . . nach Kants a dass eine Anschauung priori eigenen Erkl rungen nicht denk bar sey. Eine ist eine Vorstellung. Sollte sie a priori seyn, so Anschauung sste sie nicht von Objecte und eine werden, schlechtendings hergenommen ist doch nur m glich, sofern uns der Gegenstand gegeben wird, nur dadurch m glich, dass er das Gem ist wiederum th auf gewisse a unm glich, und kann afficiere. Eine Anschauung ist demnach priori 19This

Maas

Anschauung dieses aber Weise mithin

auch zum Grunde in Ansehung des Raumes nicht 134 liegen" (pp. does not realize, however, that the possibility of a successful use of 135). Maas a priori is precisely intuitions the problem Kant was trying to solve in the Transcendental Aesthetic,

20 in B xviii Kant as our new method of thought says that he is "adopting . . . the we can know a that of principle priori things only what we ourselves on the historical put into them." Cf. also B xiii-xiv. I have commented briefly of background and his Theory valtaa," this Kantian in "Kant's 'New Method assumption of Mathematics," 27 (1965), 37-47, and Ajatus, 84 (1964), 185-196.

of Thought' on in "Tieto

Valvoja,

KANT

ON THE

MATHEMATICAL

METHOD

373

I am putting forth this partial reconstruction only as a first to what Kant had in mind in the Transcendental approximation Exposition. This reconstruction is related fairly closely to Kant's 'transcendental argument' for his theory of space and time espe I have merely tried cially as it is presented in the Prolegomena. to fill in those steps which Kant does not himself emphasize in the relation of my partial re light of his general assumptions. The to Kant's other arguments for his views is more construction a longer discussion than I can undertake complicated, and requires
here.

to emphasize that I am not at all claiming that Kant's argument is correct. The main purpose which the reconstruction serves here is to suggest that Kant's problem of the possibility of and his attempted solution to the constructions in mathematics, problem, makes perfectly good sense even when by 'construction' one only means 'the introduction of a new individual representa a tive for general concept'. I want

structure of Kant's argument in the form presented here The is nevertheless worth a closer look. Its several stages may be rep resented in the light of what has been said somewhat as follows: reasoning is principally concerned with the (1) Mathematical existence of individuals. . to all (2) The results of mathematical reasoning are applicable a experience priori. In virtue of Kant's general 'Copernican' assumptions ("we can know a priori of things only what we ourselves put into them") (1) and (2) force us to conclude: (3) The is concerned existence of the individuals with which mathematical is due to the process by means of which

reasoning we come to know the existence of individuals Of

in general. course, what really matters is not the existence of the indi viduals as such (there are plenty of individuals existing in the but the existence of individuals having the world) appropriate relations to each other. Hence we may perhaps paraphrase (3) as follows:

(4) The mutual relations of the individuals with which mathe matical reasoning is concerned is due to the process by means of which we come to know the existence of individuals.

374

THE MONIST

It pletely implausible as applied to logic instead of mathematics. is in (5) that Kant really goes wrong. It is simply not true that we usually or always come to know the existence of individuals in the world by means of perception in the sense that perception is the whole of the process involved. It may even be asked whether any perception at all need be involved. When we come to establish the existence of a number of a certain kind, it ismistaken to assume that is always involved. perception (But is a number really an indi vidual? Maybe not; but certainly a number was an individual forKant when he called the symbols of algebra intuitions, i.e., rep resentatives of individuals. Kant's account of algebra stands or falls with the assumption that 'individuals' of the sort represented by the variables of algebra are also known by the sole means of per ception.) The concept of an inner sense to which Kant resorts here is one of theweakest points of his system.To think of all knowledge of individual objects as being due to perception is to succumb to a temptation which for Kant may have been very real but which it is important to get rid of. This is the temptation to think that the basic materials of human knowledge are given to us passive receivers who do not have to actively search for these materials. On this fallacious idea the human mind, often conceived of as a disembodied spirit inhabiting an alien machine, has to wait until the signals from the outside strike its receptors. (It is interesting in

as complemented doctrine of the mathematical method, by the results he thought he had achieved in the Transcendental Aesthetic. This line of thought (1) - (6) is not without interest and even without certain plausibility. Since we have seen that Kant's point can be translated so as to apply to modern logic, we are therefore led to ask what the corresponding argument will look like as applied to symbolic logic. Steps (1) - (2) and (4) do not seem tome com

These relations may be expected to be re systems of mutual flected by the structure ofmathematical reasoning. Now Kant has been seen to assume that (5) the process by means of which we come to know the exis tence of individuals in general is perception (sensation). From (4) and (5) it follows that (6) the structure of mathematical reasoning is due to the struc ture of our apparatus of perception. Now full and final (6) is in effect a basic feature of Kant's

KANT

ON THE

MATHEMATICAL

METHOD

375

this connection to observe the way in which Kant stressed the pas sive nature of perception, speaking, e.g., of how objects are given to us in perception.) The fact that the mind can indirectly spur a movement into is not thought to alter the situation themachine materially. Nor is the situation essentially changed by the fact that the human mind can in many ways actively according to Kant organize the raw-materials thus obtained, add to them and perhaps even modify them. I hope that I do not have to argue here that this picture is grundfalsch, thoroughly false. It is more interesting to ask for a better account. If perception is not the general concept which covers all that we want, what is? It seems to me that insofar as we can give a general name to all the processes by means of which we come to know the existence of individuals, they may rather be called processes of searching for and finding than acts of percep the accidental perception of tion, albeit we have to accommodate an object as well as the deliberate construction of an object as cases of 'searching' and 'finding* in this broad (broadest special possible) sense. Hence we have instead of (5) : (5) The process by means of which we come to know the exis tence of individuals is that of searching for them. Instead of (6) we thus have to conclude: a logical argument is due to the structure (6) The structure of of the processes of searching for and finding. My attempted partial reconstruction of themain point of Kant's as applied to modern symbolic logic philosophy of mathematics thus gives rise to an interesting suggestion instead of mathematics for our present-day philosophy of logic. The suggestion is to con as being essentially the logic sider the logic of quantification of the notions of searching for and finding (suitably generalized). It seems tome that this suggestion is likely to give rise to interesting and important considerations, if carried out systematically. JAAKKOHINTIKKA STANFORDUNIVERSITY AND THE UNIVERSITY OF HELSINKI