Kant's Transcendental Idealism

by Wilfrid Sellars
University of Pittsburgh

Published in Collections of Philosophy 6 (1976): 165-181.

[Note by editor: This surely must be one of the most difficult of Sellars' articles to find. I myself was unsuccessful in finding a copy until two years ago when both David Rosenthal and Willem deVries sent me copies of the typescript and published versions. There are minor differences between these. In this transcript, I have favored the typescript over the published version. Andrew Chrucky]

      1. When Kant mobilizes the position which he calls 'transcendental idealism' to resolve the antinomies, he describes as the doctrine that "everything intuited in space and time, and therefore all objects of any experience possible to us, are nothing but appearances, that is, mere representations which, in the manner in which they are represented, as extended beings or as series of alterations, have no independent existence outside our thoughts." He contrasts this thesis with that of the "realist, in the transcendental meaning of the term" who "treats these modifications of our sensibility as self-subsistant things, that is, treats mere representations as things in themselves" (A 490-1; B 518-9).

      2. Since Kant calls his idealism 'transcendental' in order to indicate that it enables him to account for the existence of synthetic a priori knowledge concerning objects in space and time, he has, strictly speaking, no use for the term 'transcendental realism,' since on his account of synthetic a priori knowledge we could have no such knowledge of spatio-temporal objects if they were things themselves. Nor did realists, as he sees it, claim that their realism accounted for the existence of such knowledge, although at least some of them would have taken realism to be compatible with it.

      3. The title of this essay is, of course, overly ambitious. In particular, it seems to promise as much discussion of 'transcendental' as of 'idealism.' My primary concern, however, is with the ontological aspects of Kant's idealism, and only incidentally (and by implication ) with epistemological issues concerning synthetic a priori knowledge. I note in passing, however, that since the ontological aspects of Kant's idealism concern in large part the ontology of mental states, and since, although epistemology is not psychology, it is mental states which are the proper subjects of epistemic appraisal, Kant's ontology of mental states is directly relevant to his epistemology and, consequently, to his transcendental philosophy.

      4. I shall be primarily concerned with the views Kant expresses in the first edition. I am, however, in whole-hearted agreement with his claim that there is a fundamental identity between the views expressed in the two editions. In the Preface to the second edition (B xxxvii ff.) he explains the changes he has made as a matter of removing "difficulties and obscurity" which "not, perhaps, without my fault, may have given rise to misunderstandings," and of omitting or abridging, to make room for new material, "certain passages which, though not indeed essential to the completeness of the whole may yet be missed by many readers as otherwise helpful." He suggests that the "loss . . . can be remedied by consulting the first edition," thus implying that the deleted material coheres with the new material. As regards ''the propositions themselves and their proofs," however, he claims that he has ''found nothing to alter," a statement which he shortly repeats with even greater force by characterizing the new edition as altering absolutely nothing in the fundamentals of the propositions put forward or even in their proofs," though he does grant that this [he hopes] more intelligent exposition . . . "here and there departs so far from the previous method of treatment that mere interpolations could not be made to suffice" (B xlii).

      5. It is surely implied by these claims that in Kant's eyes the refutation of idealism which was added in the second edition is not only compatible with the teachings of the first, but is implied by them. I think this is correct. The 'new' refutation simply applies the content of the Analytic of Principles, and, in particular, the Analogies to the topic of idealism, an application which is present in, though not highlighted by, the first edition refutation. And when Kant explains why he has relocated the refutation in the Postulates of Empirical Thought, his reason is a good one rather than, as Bennett characterizes it, a "silly" one.1 Anyone who studies the fourth Paralogism must feel the awkwardness involved not only in classifying the theses of Rational Psychology in terms of 'relation,' 'quality,' 'quantity' and 'modality,' but, in particular, the treatment of problematic idealism as the modal counterpart of the substantiality, simplicity and personal identity of the soul. The relation of our knowledge of material things to our perceptual experiences is far more at home in a section devoted to the thesis that "that which is bound up with the material conditions of experience, that is, with sensation, is actual.'' For this bound-up-ness, when one spells it out does have an inferential aspect and therefore does lead naturally to the Cartesian problem.

      6. I shall not, therefore, discuss the second edition refutation as such, since I hope to convince you that its familiar claims are indeed contained in the first edition, indeed in the first edition refutation itself. For in spite of its 'subjectivist' flavor, the latter contains a reference to and, indeed, a summary of the answer to the question "What is an object of representations?" raised first in the introductory passages of the first edition Transcendental Deduction, answered in highly schematic form, raised again is the Second Analogy, and this time given a fleshed-out answer which is repeated in the text of the second edition. And it is Kant's answer to this question which is central to the contrast he draws between his idealism and the idealisms he calls 'dogmatic' and 'problematic.'


      7. In the passage from the Dialectic with which I began, transcendental idealism is characterized directly as the view that "everything intuited in space and time, and therefore all objects of any experience possible to us, are nothing but appearances, that is mere representations which . . . have no independent existence outside our thoughts." It is characterized indirectly by its contrast with realism "in the transcendental sense," according to which these objects are "self-subsistant things," or "things in themselves." In a key footnote which occurs in the fourth Paralogism (A 375n) Kant writes

     We must give full credence to this paradoxical but correct proposition, that there is nothing in space save what is represented in it. For space is itself nothing but representation, and whatever is in it must therefore be contained in the representation. Nothing whatsoever is in space, save insofar as it is actually represented in it. It is a proposition which must indeed sound strange, that a thing can exist only in the representation of it, but in this case the objection falls, in as much as the things with which we are concerned are not things themselves, but appearances only, that is, representations.

Kant here does not use the contrast between existence 'inside' or 'outside' our 'thoughts,' but it is clear that he is treating the relevant sense of 'representation' as equivalent to 'thought,' for he has just written that while "something which may be (in the transcendental sense) outside us" -- i.e. which exists in itself -- "is the cause of our outer intuitions . . . this is not the object of which we are thinking in the representations of matter and of corporeal things; for these are merely appearances, that is mere kinds of representation, which can never be met with save in us . . ." (A 372). It is clear that to 'represent' is here a case of to 'think' and that Kant, who is elaborating the definition he has just given of 'transcendental idealism' as

the doctrine that appearances are to be regarded as being one and all representations only and not things in themselves (A 369)

is introducing in his comments on this definition the formula of the later definition in the Dialectic according to which the objects of intuition "have no independent existence outside our thoughts."

      8. Intuitions, in the relevant sense, are a species of 'thought' and when Kant says that appearances are ''mere kinds of representation", we certainly should not interpret this as meaning that appearances are mental acts of thinking. They are items which exist 'in' our thoughts, i.e. which are, in an appropriate sense, representeds rather than acts of representing. And when he adds that "appearances are not to be met with save in us" this must, of course, be construed in terms of his distinction between the 'empirical' and the 'transcendental' senses of 'in us' and 'outside us.'

      9. Now mental acts which are intuitings are not, of course judgings. But they are nevertheless thoughts. In the important classification of representations (vorstellungen) which Kant gives in an introductory passage of the Dialectic (A 319; B 376) Kant includes sensations as well as intuitions as representations. A sensation however is a "mere modification of the mind," whereas an intuition (though not a general concept) is an ''erkenntnis" ("Cognitio"). Its distinctive feature is that although, like a general concept, a cognitio, it is a singular one and does not refer to an object "mediately by means of a feature which several things may have in common" as do general concepts.

      10. Kant holds the interesting and important view which I have explored elsewhere,2 that an 'intuition of a manifold' as contrasted with a sheer 'manifold of intuition' is an 'erkenntnis' which presents as much as does an Aristotelian 'phantasm' (tode-ti) as a this-such, though it is not a judgment of the form '(this) is (such).'3 In a paragraph (A 79; B 105) of the Metaphysical Deduction of the Categories, which can almost be described as the Transcendental Deduction and the Schematism in embryo, Kant tells us, in effect, that intuitions of manifolds contain the very categories which can be found m the general concepts which we apply to these intuitions (and which we have, indeed, by "analytic thinking", derived from them) (A 78-9).

      The same function which gives unity to the various representations in a judgment also gives unity to the mere synthesis of various representations in an intuition; and this unity, in its most general expression, we entitle the pure concept of the understanding. The same understanding through the same operations by which in concepts, by means of analytical unity, it produced the logical form of a judgment, also introduces a transcendental content into its representations, by means of the synthetic unity of the manifold in intuition in general. On this account we are entitled to call these representations pure concepts of the understanding, and to regard them as applying a priori to objects -- a conclusion which general logic is not in a position to establish.

Thus singular judgments which express basic perceptions have the form

(This-such) is so and so

and the categories which are implicit in the general concept so and so can be true of the subject of the judgment, i.e. the object intuited, because the intuition of the this-such also contains the categories.

      11. It is essential to see that intuition is a species of thought, for any sense-datum like approach makes essential features of Kant's theory of knowledge unintelligible, e.g. the Schematism. Thus the categories apply to intuitions, because, although the content of sensations does not contain the categories, the content of intuitions (of manifolds ) does. This is the point of Kant's problem about homogeneity and of his solution.


      12. Let us take seriously then, the thesis that intuitions of manifolds are thoughts. And let us apply to them the ontological categories which the Cartesian tradition, rooted in scholastic tradition had applied to thoughts. An adequate discussion would call for a whole cluster of distinctions in which themes from Husserl, the early Brentano, Meinong, and the later Brentano would be inextricably involved. I shall use a bare minimum of distinctions and resolutely avoid probing into the deeper metaphysical issues involved.

      13. Descartes distinguishes between the act and the content aspects of thoughts. The content, of course, 'exists in' the act. And, of course, contents as contents exist only 'in' acts. On the other hand, there is a sense in which something which 'exists in' an act can also exist, to use Kant's phrase, 'outside' the act. In Descartes terminology, that which exists 'in' the act as its 'content' can have 'formal reality' in the world.4

      14. The concept of 'existence in thought' is, of course, a metaphor. So, of course, is the idea that a thought is true if its content 'corresponds' to something actual in the world. In the Cartesian tradition, this account of truth holds both of thoughts of individuals and thoughts of states of affairs. Thus a thought of an individual is true if there corresponds to it an actual individual, i.e. if the individual exists not only 'in the thought' as its content, but in the actual world. Similarly, the thought of a state of affairs is true if there 'corresponds' to it an actual state of affairs, i.e. if the state of affairs which exists in the thought as its content, also exists in the actual world.

      15. I have spoken of actuality where Descartes speaks of 'formal reality.' In Kant's terminology what Descartes means by 'formal reality,' and which Cartesians would equate with actuality, is 'existence in itself' or, to use the latinate term of the Prolegomena, existence 'per se.' Kant clearly accepts the Cartesian contrast between 'existence in thought' and 'existence per se,' -- so much so that he takes it for granted, as did most of his predecessors.5

      16. One final terminological point: The content of the thought of an individual was considered to be, in an important sense, an object of the thought. Indeed the thought would have in this sense an object even if nothing in the actual world corresponded to it. Later philosophers drew tidy distinctions between 'immanent objects' and 'transcendent objects,' and we are all familiar with the intuitive appeal of these metaphors. Philosophical terminology consists largely of metaphor added to metaphor in the hope that the mixture will crystalize out into clear and distinct categories.

      17. For Kant, then, an act of intuiting a manifold is a thinking of a this-such in space and/or time. The this-such is something that exists 'in' the act. The problem with which Kant is dealing can be characterized initially as that of whether individuals in space and/or time also have existence per se. Kant's answer, to anticipate, is that these intuited items exist only 'in' acts of intuition.6 That is, no items in space and/or time exists per se. He will nevertheless insist that some items which exist in acts of intuition are actual. This obviously requires a distinction between actuality and existence per se, which were conflated by his predecessors.


      18. It will be useful to connect Kant's concept of the 'intuition of a manifold' with that strand of contemporary perception theory which operates with fairly traditional concepts of intentionality. A familiar notion is that of a perceptual taking. Perceptual takings are, so to speak, thinkings which are evoked in our minds by our environment or, in limiting cases, by abnormal states of our nervous system. Perceptual takings are usually thought to have propositional form. One takes there to be a cat on a mat. I suggest that what is taken is best expressed by a referring expression, thus 'this cat on this mat.' We should think of perceptual takings as providing subjects for propositional thought, rather than already having full-fledged propositional form.

      19. Again, if we think of a taking as a special case of a believing, it is best to think of it as a 'believing in' rather than a 'believing that.' In a perceptual taking one believes in this cat on this mat, and may believe, for example, that this cat on this mat is a siamese. Thus construed, perceptual takings are in many respects the counterparts of Kant's 'intuitions of manifolds.' They represent this-suchs; and it is worth noting that although they are not explicitly propositional in form, they obviously contain propositional form in the sense in which 'that green table is broken' contains 'that table is green.'


      20. Little needs to be said at this stage of the argument about the 'problematic idealism' Kant attributes to Descartes. Problematic idealism regards the claim that material things and processes exist per se as a coherent one, but one which can be established only by an inference from our perceptual states; an inference from effect to cause. Mental states have a privileged position in that they not only can have existence per se. But at is known, indeed 'directly' known, that some mental state exist per se.

      21. Before looking at Kant's refutation of problematic idealism, it is worth pausing to ask: If this is 'problematic' idealism, what might 'dogmatic' idealism be? One would expect it to be the view that material objects in space and time could not have existence per se, i.e. that there was an absurdity or contradiction in the idea of the existence per se of material objects. Berkley certainly held this position, a fact which might strengthen the temptation to interpret dogmatic idealism along these lines.

      22. It is also worth noting that Berkeley did not deny that in a sense we (and a fortiori God) can conceive of material objects in space and time, where by the phrase 'material objects' I mean, so to speak, Lockean objects and not patterns of actual and counterfactual perceptual experiences.7 Berkeley can be construed as holding that our actual and counterfactual perceptual experiences are grounded in God's plan to cause us to have those experiences we would have if (a metaphysical counterfactual) there were material objects including human bodies with sense organs, and minds and bodies were able to interact. In less theological terms, this can be formulated as the view that what exists per se other than our minds is the causal ground of our actual and counterfactual perceptual experiences.

      23. To return to the main line of argument, it is only too clear that if we mean by 'dogmatic idealism' with respect to material objects the view that they cannot have existence per se, then Kant is a dogmatic idealist of the first water. Indeed as we have seen, Kant makes dogmatic idealism in this sense the very corner stone of his transcendental idealism.

      24. What then does Kant mean by dogmatic idealism? And is any such view to be found in Berkeley? It should be clear that the only answer to the first question which satisfies the requirements of the argument to date is that Kant means by dogmatic idealism the view that nothing spatial can be actual, where actual does not mean 'exists per se.' Indeed Kant's own idealism, while denying that material objects exist per se, nevertheless insists that some at least of the spatial objects which exist 'in our thoughts' and, in particular, in our acts and intuitions, or perceptual takings, are, in the critical sense, actual.

      25. But what of the second question? To what extent is Kant justified in attributing to Berkeley the view which he (Kant) would express by saying that no spatial items are actual?8 After all, if, as is often claimed, Kant in the first edition and on occasion in the second edition construes physical objects in terms of actual and counterfactual perceptual experiences, is he not in essential agreement with Berkeley?9


      26. At this point we must retrace our steps in order to advance. My discussion of Descartes' ontology of mental acts was not only schematic but radically incomplete. For Descartes had two paradigms of what it is to be a mental state. Let me begin with the one I have neglected. Sensation is a mental state, though one in which the body is intimately involved, and his paradigm of a sensation is a feeling of pain.10 To us, the obvious feature of a feeling of pain is that a feeling of pain simply is a pain. The existence of a feeling of pain is identical with the existence of a pain. Much more would need to be said to nail this point down,11 but after all the analytic work has been done, the fact remains that a pain is a kind of feeling. If we put this by saying that a pain is a 'content' or 'object' of feelings, this should be regarded simply as a (misleading) paraphrase of the above. The danger arises from the fact that this usage would tend to assimilate feelings of pain to Descartes' second paradigm, the clear distinct thought of an object. It is in connecton with the latter that he elaborates the distinctions which were inherited by Kant.

      27. It was obvious to Descartes that the mental state of thinking of a cube is not (at least in any ordinary sense of 'is') a cube. Thus, whereas a feeling of pain is a pain, a thinking of a cube is not a cube. Now in his systematic account of sensation, Descartes construes visual sensation on the model of pain. In terms of this paradigm, a sensation of red would be a case of red as a feeling of pain is a case of pain. His construal, however, as Gassendi saw,12 confronted Descartes with a problem. Surely the colors we experience have shape. But if the colors we experience have shape, then, if they are sensations construed on the model of pain, mental states can have shape. To get experienced color and shape together, Descartes, it seems, must either

  1. give experienced colors the same status as shapes (deny that either can be a modification of the mind);
  2. give the shapes of experienced colors the same status as colors (admit that modifications of the mind can have shape, as Gassendi thought we should).

      28. Descartes gives no clear account of the matter. His official account is that while the sensation of red as such is unextended, the shape we perceive, and which has existence 'in thought', is confusedly believed to be the shape of something which resembles the sensation of red. Such beliefs date from early childhood. They are confused, because nothing can be like a sensation of red without being a sensation of red, and hence a modification of the mind13. In effect, Descartes gives perceived red the status of believed in red, and by doing so gives it the existence 'in thought' which the shape also has. This move requires a distinction between the sensed red which we introspect, and which is not extended, and the perceived shaped red, which is a content rather than a modification of the mind.

      29. Now Berkeley, as is well known, also assimilated color to pain. What, then, of shaped colors? When one feels a pain, there is an actual case of pain. When one senses a color, there is an actual case of color. But when the mind senses a triangle of red, is there an actual case of shape? When directly confronted with this question,14 Berkeley's answer is no. The shape exists only 'by way of idea,' and it is clear that here the operative conception is a Cartesian cogitatio of which the shape is a content in a sense which does not entail that a mental state has a shape. But though Berkeley gives this answer, he is simply not clear about the status of shapes, and what stands out is his assimilation of the status of shape to that of color and, ultimately, to that of pain.15

      30. Now if Berkeley had consistently held that perceived shapes are not features of mental states, but have existence only 'in our thoughts' (in the Cartesian sense) he could nevertheless have argued ( perhaps on proto-Kantian grounds ) that no shapes have existence per se. Shapes would not exist 'outside the mind' in what Kant called the transcendental sense. They would exist only 'in' the mind, not as features of its states but as immanent objects. Berkeley, however, because of the slippery slope pain-color-shape, makes the quite different claim that shapes can not exist 'outside the mind' in the sense in which pains can not exist outside the mind. Even though he is not prepared to say in so many words that shapes are essentially features of mental states, he actually commits himself to this position.

      31. Thus it is not unfair on Kant's part to attribute to Berkeley the view that the concept of shaped items which are not mental states is an incoherent one, as incoherent as would be the concept of a pain which was not a mental state.

      32. Now it is as evident to Kant as it was to Descartes that neither space itself nor any spatial object can be a modification of the mind or of a mental state. Thus, while Kant denies that either space or any spatial object has existence per se, and argues that the idea that they do is an incoherent one, he also argues that the idea of shapes which are not features of mental states is itself a coherent one. Thus, if shapes which are not features of mental states do not exist per se it is not because shapes are essentially features of mental states, but because shapes belong in space, and neither space nor anything in space can exist in itself.

      33. We can now begin to appreciate why in Kant's sense of the term Berkeley was a 'dogmatic idealist' and to interpret the following passage:

      Berkeley . . . maintains that space, with all the things of which it is the inseparable condition, is in itself impossible and he therefore regards the things in space as merely imaginary entities (B 274).

Given that Kant is clearly aware that on his own view neither space nor objects in space can exist in themselves, the gravamen of the charge against Berkeley must be that the latter's reason for holding this is such as to require, as his (Kant's) reasons do not, that ''things in space are merely imaginary entities." For, once again, Kant dearly has up his sleeve the view that material things, though they necessarily lack existence per se, which was the traditional concept of actuality, can in the critical sense be actual. On my interpretation Kant would be recognizbng that Berkeley's reason for rejecting the existence per se of space and spatial objects would be that shape and extension are essentially features of mental states. The concept of spatial items which are not mental states would be a figment of the philosophical imagination. They would be radically imaginary, not just imaginary in the empirical sense.

      34. Kant obscures the justice of his characterization of Berkeley's position in the remark which follows, in which he claims that

     Dogmatic idealism is unavoidable if space be interpreted as a property that belongs to things in themselves. For in that case, space and everything to which it serves as a condition, is a non-entity (B 274).

Here the direct connection with Berkeley is lost. But that is as it should be, for Kant is merely pointing out that one can arrive at the conclusion that the concept of actual spatial items other than mental states is incoherent by a route other than Berkeley's. For if space is a property which must belong to things in themselves, then, Kant has argued, there can be no such thing. Hence, given the classical interpretation of actuality as existence per se, it would follow that the concept of an actual spatial object is incoherent. And if the only alternative to hoping that space is a property of things in themselves were, as Berkeley thought it was, the view that spatial items are features of mental states, then the concept of actual spatial items which are not features of mental states would also be incoherent. We would be faced by a dilemma.

      Either space is by nature a thing in itself or a property of things in themselves or the spatial items which underlie our concepts of space and spatial items are (features of) mental states (sensations). On either alternative the concept of actual spatial items which are not mental states is incoherent.

Kant has prepared the way for an escape through the horns of this dilemma. He points out that the argument of the Aesthetic enables him to avoid the view that space is either a thing in itself or a property of things in themselves without agreeing with Berkeley that the concept of spatial items which are not mental states is incoherent.

      35. Yet even after he has established to his own satisfaction that the concept of an actual spatial item which is not a mental state is coherent, he still has to make the essential point that the concept of actuality does not coincide, as it traditionally did, with that of existence per se.

      36. I take it to be clear, then, that Kant holds that no spatial item (notice the cautious use of the term 'item' as contrasted with the richer term 'object') is a mental state. It can, however, be argued that in the first edition and on occasion in the second edition Kant held that material objects are, to use an anachronistic turn of phrase, logical constructions out of mental states which, though not spatial, are representations of spatial items -- i.e. of which spatial items are the content or immanent objects. After all, he does write that

     In our system, on the other hand, these external things, namely matter, are in all their configurations and alterations nothing but mere appearances, that is, representations in us, of the reality of which we are immediately conscious (A 371-2).

I think that this interpretation is a mistake, though it must be confessed that because of his failure to make it clear when he is using 'representation' (and, in particular, 'intuition') in the sense of act of representing, and when he is using these terms in the sense of content represented, it has some initial plausibility.

      37. The crux of the matter as Kant clearly saw is his account of what it is to be an object of acts of representing. He formulates this account schematically in a passage (A 104) which prepares the way for the first edition Transcendental Deduction, and develops it in a more full-bodied way in a key passage in the Second Analogy which occurs in both editions.16

      38. Our primary concern is with perceptual acts or takings. But in the first passage referred to above, Kant makes his key point in a way which abstracts, as the passage in the Second Analogy does not, from essential aspects of perceptual takings. Nevertheless, the concepts for which Kant is preparing the way is that of rules for generating perceptual takings.17

      39. The term 'rule' is a dangerous one, for it suggests deliberate activity or, at least, activity which would be deliberate if it weren't so hasty and, in the ordinary sense, thoughtless. Actually the most useful concept is that of a sequence of acts of representing which can reflectively be classified as conforming to a rule which is (at least in principle) graspable by thought. The rules in question must, according to Kant, be available, if one is to recognize that one's acts of representing belong together as an intelligible sequence.18

      40. Now it might be thought that by introducing the concept of a rule-governed sequence of perceptual representings, i.e. acts of perceptual taking, I am giving hostages to the view that material objects consist of rule-conforming sequences of perceptual takings. That I am not is implied by the fact (which I hope to make clear) that even the most tough-minded transcendental realist grants that veridical perceptual takings have the coherence which Kant is attempting to clarify by the concept of rule-conforming sequences. Roughly, Kant's transcendental realist thinks of the perceiver as deriving these rules by induction from experience, whereas Kant thinks that induction itself presupposes an antecedent grasp of these rules.


      41. We can now turn our attention to Kant's initial explanation of what it is to be an 'object of representations.' He asks us to consider the intuitive representation of a triangle. Here the rich implications of the concept of a perceptual taking are laid aside, for the moment, and we are given an explanation which could concern a construction in pure geometry. For the essential point he wants to make is that while the object of the intuitive representing is indeed a triangle, the triangle is not an existent per se, and that although the content triangle specifies sequences of representing which count as coming to represent a triangle, the object of the representing of a triangle is not the sequence of representings which culminate in the representing of the triangle. A triangle is neither a mental act of representing a triangle, nor is it a sequence of mental acts each of which represents a part of a triangle.

      42. Before turning to the passage in the Second Analogy which is essentially a development of the passage on which I have been commenting but one which does take into account the specific character of perceptual takings, let me elaborate briefly on the triangle example in a way which will make for a smooth and easy transition. After all, Kant's account of 'drawing figures in thought' i.e. in pure intuition, is an idealised version, ascribed to the mind, of drawing figures on paper. Now if we take seriously the three-dimensionality of space, it strikes us that to represent a triangle in space is always to represent it from a point of view. Thus, what we represent is

this equilateral triangle facing me straight on


this equilateral triangle at such and such an angle to my (metaphorical) line of sight

      43. Now it is by no means an original idea on my part that intuitive representings of figures in three-dimensional space are essentially point-of-viewish. But its importance has been underestimated. For it means that we must distinguish between the figures -- which are not point-of-viewish -- and the total content of the representing of the figure, which total content is point-of-viewish, thus

equilateral triangle facing me straight on

Let me repeat. Equilateral triangles are not point of viewish, but they are, so to speak, intuited in perspective. The representing has a content which specifies a point of view.

      44. Thus, the object of a representing of an equilateral triangle from a certain point of view is simply the equilateral triangle. But, according to Kant's position, according to which we 'construct' or 'draw' figures in space, the concept of an equilateral triangle must specify not only a sequence of representings in which we represent one line, then, continuing to represent that line represent another line at a sixty degree angle, then, continuing to represent these we represent the third side. It must also specify in an intelligible way what it means for two representings to be representings of an equilateral triangle from different points of view, i.e. representings which have the contents

equilateral triangle face on

equilateral triangle at such and such an angle to my 'line of sight'

      45. The notion of representings which have contents of this form obviously builds on pervasive features of perceptual takings. I do not simply perceptually accept a house; the content of my perceptual acceptance is something like

this house over there facing me left-edge-of-front-wise.

      46. The point I am makung is simple, but it is so essentially a part of a larger story that I shall have to disguise the torn edges to put it across. When, in the Second Analogy, Kant says (A 191; B 236) that "the object is that in the appearance which contains the condition of this necessary rule of apprehension," he is commenting on the example of a house which he has just introduced:

. . . immediately I unfold the transcendental meaning of my concepts of an object, I realize that the house is not a thing in itself but only an appearance that is a representation . . . (A 190-l; B 236).

Here it is dear that he means by 'representation' something represented i.e. a content or immanent object of an act of representing.

      47. Nevertheless, although he calls the house an appearance, where this clearly does not mean that the house is an act of representing, he does say that

The appearance . . . is nothing but the sum of these representations (of apprehension)

and it might be thought that Kant is characterizing the house as a sum of acts of representing. It is therefore important to note that he characterized the representations in question as ''that which lies in the successive apprehension." (A 191; B 236, emphasis mine) This must surely be his way of warning us that 'representation' here means represented rather than representing i.e. Kant is relating the content house to the contents of successive acts of apprehension.

      48. Kant is answering the question 'in what sense is the house the object of successive acts of apprehension.' His negative answer is that it is not qua house in itself. The actuality of the house qua object is not its existence per se. What is his positive answer? As I see it, he is telling us that the house qua object is that aspect of the content of the perceptual takings which explains (together with certain other factors) the belonging together as state of the perceiver of certain perceptual takings (apprehendings). But that aspect of the content of these perceptual takings is simply the content house which they share, thus

house over there left-front-edgewise to me
house over there facing me
left side of house over there facing me
etc., etc.

      49. As in the geometrical example, all the representings are representings of a non-point-of-viewish object -- the house -- from a point of view, i.e. representings of house-from-a-certain-point-of-view. It is, in this sense, that the house is the sum of the point of viewish appearances. But if the total content of a perceptual act is point-of-viewish, it is because it is the content of a perceptual act. Thus, while the content house is not a point-of-viewish content, it explains (together with certain other factors) why such and such perceptual representings with contents which can be subsumed under the rubric


take place. Thus, the concept of a house as a perceptible object essentially involves a reference to perceptual acts, i.e. to the perceptual takings of a perceiver.

      50. Nevertheless the concept of a house as perceptible object is not the concept of the sequences of perceptual takings (actual and counterfactual) which (together with other factors) it explains. To pull my points together in one sentence,

      The object of a perceptual representing of a house is the non-perspectival content house; yet as the sort of item that can be the object of a perceptual representing, it must provide rules for explaining (together with other factors) why such and such sequences of perceptual takings with perspectival contents were necessary.

      51. It will have been noticed that I have written in several passages above that the content house is the source of rules which explain together with other factors why such and such sequences of perceptual takings were necessary. What do I mean by 'other factors'? The answer should be obvious, and peeks out from almost every page of the Analogies. House by itself can generate no explanation of the occurrence of a sequence of perceptual takings. It is only house in such and such relations to a perceiver which can do this. And this, obviously, means to a perceiver's body, and, of course to sense organs. The essential structure of the content of perceptual takings is not just

house from a certain geometrical point of view

but, to make a complicated point in a simple way

house in front of my sightful eyes
ship in water moving to the left of my sightful eyes.

In my argument I have thinned out this mutual involvement of object, circumstances and embodied perceiver into a ghostly 'object from a point of view.' But Kant took seriously the fact that perceivers are embedded in a spatio-temporal system of interacting substances. In other words the doctrine of the double affection of the self, far from being a problematic feature of Kant's critical idealism is an essential feature of it, and is present in and, indeed, an essential feature of the argument of the Second Analogy.18

      52. Kant denies that material things and processes exist per se, but he holds that in the critical sense they can be actual as contents which make an essential contribution to the explanation of the patterns in which perceptual experiences occur. But the deeper thrust of Kant's transcendental idealism is the thesis that the core of the knowable self is the self as perceiver of material things and events. And if it is relatively easy to see that the distinction between actual and non-actual material things and events is tied to the concept of an actual sequence of perceptual takings, it has (until recently) proved less easy to see that the distinction between actual and non-actual sequences of perceptual takings, i.e. between perceptual takings which are correctly and those which are incorrectly taken to have occurred in one's mental history, is tied to the concept of actual material things and events.

      53. Kant saw that the concept of an object of perception contains a reference to the perceptual takings which are the criteria for its actuality. He also saw that the concept of a perceptual taking, as the taking of an object, contains a reference to material things and events which, if actual, would imply its own actuality. The actuality of perceptual takings and the actuality of material things and processes are not logically independent. And since, for Kant, the concept of matter-of-factual truth concerns the agreement of what we represent with what is, in the critical sense, actual, rather than, as traditionally with what exists per se, he can pay his respects to what he calls "the nominal definition of truth" while giving it a radically new interpretation.


      1 Kant's Analytic, p. 166.

      2 Science and Metaphysics, London and New York, 1968, Chapter 1.

      3 One should also bear in mind Ockham's concept of a perceptual intuition, and his claim that God could bring it about that the object intuited not exist.

      4 I have avoided Descartes use of the phrase 'objective reality' in connection with contents, since it has, of course, a quite different meaning in Kant and in contemporary philosophy.

      5 Including the British Empiricists who, in the process of muddling through to important philosophical insights, muddied many classical distinctions.

      6 A more complete discussion of this point would have to comment on those passages in which Kant allows that things in themselves are objects of perceptual intuitions. After all, if the latter contain the pure categories, and if the latter constitute 'the pure concept of an 'object' ["The pure concept of this transcendental object which . . . (concept) in all our knowledge is always one and the same" (A 109)], then things in themselves would 'correspond' (in the Cartesian sense) to intuitions, though not qua intuitions of spatio-temporal items. see A 490-l; B 518-9 quoted in the first paragraph; see also Science and Metaphysics pp. 42 ff.

      7 I deliberately use this vaguer expression 'perceptual experiences' instead of 'perceptual takings,' since one of the essential differences between Kant and Berkeley concens the analysis of what it is to be a perceptual experience.

      8 Notice that to pinpoint the issue I have retreated from the phrase 'material object' to 'spatial item.'

      9 Notice once again that I deliberately used the vaguer phrase 'perceptual experience,' for in the last analysis everything will hinge on how this phase is interpreted.

      10 Another paradigm is to be found in his use of 'sensation' in connection with seeming to see in the second meditation. But while there is clearly a close connection between sensing and seeming to see an object, they are not identical, nor does Descartes equate them m his developed philosophy of mind.

      11 I have discussed this topic in "Metaphysics and the Concept of a Person," in The Logical Way of Doing Things, edited by Karel Lambert, New Haven, 1969 (reprinted as chapter XI in Essays in Philosophy and its History published by Reidel, Dordrecht, Holland, 1974).

      12 Philosophical Works of Descartes, eds. Haldane and Ross, Dover, 1934. Vol. II, pp. 196-7.

      13 Principles of Philosophy, LXVII-LXX.

      14 Principles of Human Knowledge, section 49 (Fifth Objection).

      15 A more penetrating account would demonstrate that Berkeley initially conceved of all 'perceptions' as in the 'understanding' a move which, by taking thoughts as the paradigm of mental states, constituted exactly that blurring of the distinction between sensibility and understanding so obviously present in Spinoza, Leibnitz and the Wolffians -- which Kant was to regard as a key philosophical error.

      16 I have in mind the passage (A 190; B 235) introduced as follows: ". . . it is a question for deeper inquiry that the word 'object' ought to signify in respect of appearances when these are viewed not in so far as they are (as representations) objects, but only in so far as they stand for an object."

      17 It might be helpful, here, to think of rules for generating sequences of acts of imagining which would be the counterparts of perceptual takings, if they had their source in outer sense.

      18 An elaboration of this theme would require an exploration in detail of Kanl's conception of the transcendental unity of apperception as a necessary correlate of the intuition of objects.

      19 As noted above (paragraph 22) even Berkeley came to see that the intelligibility of the patterns in which perceptual experiences occur involves the concept of embodied perceivers in a materia world. However, since he was convunced that material objects could not exist per se, he gave them existence in God's Understanding and Providence as essential features of his plan for causing us to have the experiences we do.