Kant's Intuitionism: A Commentary on the Transcendental Aesthetic
(Toronto: University of Toronto Press, 1995)
pp. xxiii, 465. $70.00
Kant's Intuitionism is the best book length study of Kant's views of space, time and sensible input in existence. Unfortunately, it is also one of the few such studies. However, it is an excellent piece of Kant scholarship by any standard. And it is highly original.
After usefully introducing intuitionism and the issues in the background of Kant's theory (nativism and empiricism, intuitionism and constructivism), Falkenstein divides his large book up into three parts. Part I is a long, thorough study of Kant's terminology. Part II takes up Kant's Expositions, one by one and in great detail. This Part is the heart of the book. Part III is called, and draws, Conclusions from the Above Concepts for Kant's big metaphysical and epistemological claims: that things as they are in themselves are neither spatial nor temporal; that we have no knowledge of things as they are in themselves (these two claims are in some tension, of course); that space and time are merely aspects of human cognition; that our spatiotemporal experience is nevertheless affected in some way by things as they are in themselves; and so on (the last two views are also in some tension).
Most studies of Kant on space and time focus on these big issues. By contrast, Falkenstein holds that,
There is ... a theory of space and time cognition to be found in the Critique of Pure Reason [that is] more fundamental than the theory of affection and stands independently of any problems there may be with it. [p. 142]
More fundamental than Kant's other big theses, too. Falkenstein presents Kant as developing an information-processing model of the mind, one in which the deliverances of the senses and the cognitive activities of the central systems work together to produce representations (pp. 138-42 give a nice summary). Falkenstein is not the first to argue that Kant's model of the mind still has things to teach us, nor to seek to separate that model from the big conclusions. Many people would like to have a Kantian model of cognition that is free of Kant's specific problems, Kitcher (1990) and Brook (1994) for example. What makes Falkenstein's book unique is that he thinks that he can isolate and extract something sensible and interesting from the most idiosyncratic part of Kant's model and the one part that has always seemed to be inextricably enmeshed in his overall system, namely, his account of cognition in time and space.
Falkenstein's interpretation is as original as his approach. On the standard interpretation, Kant starts from two theses:
1. Intuitions, the raw material of sensible experience and the content of the faculty that Kant called Sensibility, arrive in a totally unorganized state, a mere `booming, buzzing confusion' to use James' phrase. This view Falkenstein calls the `heap thesis' (p. 79).
2. Even though sensible experience is in the faculty of Sensibility, it is the Understanding that organizes it spatiotemporally, via acts of synthesis.
As Falkenstein sees it, both attributions to Kant are mistaken. Against (1), he argues that, though the organizing spatial and temporal structures are properties of the person, not the objects being intuited, a point on which Kant insisted, intuitions are nevertheless structured on arrival. Against (2), he holds that for Kant, organizing particular sensory inputs into a temporal and, many of them, a spatial array is not an act of synthesis. Rather, as Falkenstein puts it,
Sensory impression and the constitution of the subject [are] joint causes operating prior to any cognitive processing to deliver an integrated spatiotemporal sensory manifold as originally intuited representation. [p.93]
Originally intuited: our intuitions are arrayed in space and time as we receive them.
Not only is spatiotemporal structuring not an act of synthesis, Falkenstein tells us, synthesis requires that such structuring already be in place. Falkenstein usefully distinguishes two forms of synthesis, pattern recognition and assembly of sensible material into objects. To find a pattern in a manifold of items and a fortiori to assemble them into integrated, persisting objects, the items in the array must already be spatiotemporally located. In fact, according to Falkenstein, Kant went so far with the idea that intuitions are immediately given as spatiotemporally organized that he held this organizing to be "physiological," not "psychic" (by which Falkenstein means, not done by the central, concept-using systems) (p. 123).
Making this case against the standard interpretation is the burden of Parts I and II of the book. Falkenstein's account is extremely generative; it led me to rethink many long-held views and throws new light on topic after obscure topic:
the nature and extent of Kant's innatism;
the nature of Kant's faculty psychology and how it differs from earlier faculty psychologies;
the concept/intuition distinction;
the composition problem (why the infinite divisibility of space and time requires us to think of them as properties of the mind, not properties of things as they are in themselves);
what Kant could legitimately have meant by saying that things as they are in themselves are not spatial or temporal;
how the mind spatializes and temporalizes sensible input and why the heap thesis is suspect;
the localization problem (how represented particulars can be located at particular points in space);
how Kant can insist that space and time are not concepts and yet speak constantly of the concepts of space and time (Falkenstein handles this issue beautifully, pp. 63-71);
how we can make sense of Kant's apparent allowance that we do know one object as it is in itself, namely, the representing self;
and on and on and on.
All this would be more than enough for one book. However, Falkenstein does more. In the third part of the book, he goes after the big conclusions--the nonspatiality and nontemporality of things as they are in themselves, the unknowability of things as they are in themselves, the thesis that they affect us, and the thesis that space and time are properties of the mind, not the world. In every case, Falkenstein shows that there is something closer to good sense in them than most commentators have seen.
The book also has marked scholarly virtues. Unlike much work on Kant (including much of my own), Falkenstein places Kant in his time. He shows us to what Kant is reacting, what Kant could be expected to know, the status quo of scholarship on issues of interest to Kant, why Kant might not have felt a need to deal with a problem that seems pressing to us now, and so on and so forth. Equally, Falkenstein makes his exegetical principles clear, for example the principle that to understand Kant's German terms of art, we should look to see if they are natural translations of Latin philosophical terminology around in Kant's milieu. Even Falkenstein's translations often throw new light on familiar passages (Falkenstein did most of his own translating). And his extended treatment of difficult terms such as `intuition' (Anschauung), `sensation' (Empfindung) and `concept' (Begriff) is among the best in the literature.
One weakness of the book is that Falkenstein leaves us pretty much to ourselves when it comes to figuring out the exact relationship of Kant's model to contemporary work. It would have been good to have had at least a short discussion of what is still living and what is dead in Kant's work on perception and its sensible antecedents. Falkenstein's language is also a little strong sometimes. He accuses Kant of cowardice, indeed, of intellectual, psychological, and political cowardice (p. 19), for example, and calls an example of Walker's `inept' (p. 198). These are harsh judgments; indeed, Falkenstein himself partly backs off from the cowardice charge. Given the important work that Falkenstein does on terminology, it would have been good to have a glossary.
I am sympathetic to Falkenstein's general approach, less sympathetic to some of the details. Thus
I don't have broad, sweeping criticisms but I do have questions about some of the particulars. I
will begin with some questions about the relationship of Falkenstein's interpretation of Kant on
space and time to other aspects of Kant's system (Section I). Next I will comment on two
possible problems with Falkenstein's own account (Section II). Then I will explore two questions
about Falkenstein's treatment of the `big' issues in the third part of his book (Section III).
I. The Heap Thesis and Synthesis
Like most readers of Kant, I have always taken the heap thesis for granted in a casual sort of way--Kant, I thought, held that the raw material of sensible experience arrives without structure, we know not how or whence, and the mind then goes to work on it, imposing a temporal and in many cases a spatial matrix on it and articulating it categorially. The first activity one vaguely associated with the Syntheses of Apprehension, the second with the Synthesis of Recognition. Indeed, the full name of the former is Synthesis of Apprehension in Intuition, of the latter Synthesis of Recognition in a Concept.(1) So it is eye-opening to see this reading forcefully challenged.
In contrast to the heap thesis and the standard interpretation of Kant as holding that both spatiotemporal organizing and recognition in concepts are acts of synthesis, Falkenstein urges that for Kant information goes through two stages of processing on its way to becoming the content of sensible representations.
Stage 1: The mass of intuitions that impinge our sensible receptors are structured as a spatial and temporal array.
Stage 2: The Understanding then performs acts of synthesis in which patterns are recognized in the arrayed items and they are assembled into objects (pp. 98-100, 245).
To understand Falkenstein's alternative, it is vital to see that he means these two stages to be very different from one another. Whereas the second operates by way of deliberate acts of judgment, the first operates by way of innate filters or pathways in the receptors, filters or pathways that "determin[e] the manner in which sensations are received" (p. 11, italics in original). This difference generates the argument against the heap thesis:
The heap thesis [is] incorrect. For Kant, the manifold is ordered in the intuition, not just in `appearance' and the order of the manifold originates from the same source as its matter--the reception of the parts of the manifold. ... An ordered manifold of parts ... [is] the representation immediately given in sense intuition. [pp. 80-1]
(An appearance is a worked up representation that is the product of synthesis.) Falkenstein expresses the distinction between sensible structuring and acts of synthesis in a number of ways: innate structures in a receptor system vs. cognitive processing (p. 11); receiving (see `immediately given' in the quote) vs. processing (p. 60); the form in which intuitions are received vs. how we go on to process them (p. 67); the "immediate effect of impression of the senses [on] a subject with a certain sensory `constitution'" vs. spontaneous acts (p. 93); and so on. The distinction grounds an exciting, even dramatic alternative to the standard reading. How well does it stand up?
I think that Falkenstein's insistence that spatiotemporal structuring is something very different from recognition or assembly is correct and valuable. It does not follow, however, either that spatiotemporal organizing is not also an act of synthesis or that the heap thesis about the former is incorrect. I have questions about both.
1. Did Kant hold that spatiotemporal organization is not an act of synthesis? Recognition and assembly would seem to go with the Synthesis of Recognition but there is also the Synthesis of Apprehension, which Kant explicitly puts before the Synthesis of Recognition.
2. If not by an act of synthesis, how is the stuff of intuition organized spatiotemporally? And exactly when does it cease being a heap, upon arrival or when the receptors structure it? Only the former is a clear alternative to the heap thesis, but only the latter seems compatible with Kant's insistence that space and time are forms provided by us.
Synthesis of Apprehension Let's start with the Synthesis of Apprehension. Kant tells us that in this form of synthesis, "the mind distinguishes the time in the sequence of one impression upon another" (A 99).(2) Between this kind of synthesis and the further kinds that Kant calls Syntheses of Reproduction and Recognition, he seem to allow just the difference between spatiotemporal organizing and recognition/assembly that Falkenstein wants--but entirely inside the doctrine of synthesis. Falkenstein's suggestion is that the remark from the Synthesis of Apprehension passage just quoted is "not ... about intuited representations at all, but about representations of intuitions" (p. 76). Perhaps; but for whatever purpose, the mind is distinguishing the time in the sequence of impressions as presented, not in any subsequent representation of them. To use Falkenstein's distinctions, Kant is speaking here about distributing an array, not about tying items of an array together into a pattern or assembling them into a unified object. (Note the qualification: `the sequence of impressions as presented'. As I will argue in Section III, Kant is always talking about how things appear to us, or at least should always be. The actual order of impressions is something that we cannot know, even impressions as presented to us. All we can know, Kant thinks, is that this order is not temporal.)
One of Falkenstein's arguments for his distinction between synthesis and spatiotemporal organization is that there must already be an organized spatiotemporal array in intuition by the time synthesis begins. The job of synthesis is to find patterns in intuitions and assemble them into representations of single things. This would be impossible, urges Falkenstein, if particular intuitions were not already arrayed in space and time. Response: This may be true of Synthesis of Reproduction and/or Recognition, but there is another kind of synthesis, Synthesis of Apprehension, and it may be responsible for distributing items in the spatiotemporal array. If "raw sensation already exhibits spatiotemporal form before intellect or imagination make any contribution to it" (p. 83), it does not follow that it exhibits this form before Synthesis of Apprehension makes a contribution to it. Imagination enters Synthesis of Reproduction, intellect the Synthesis of Recognition, but the third element, intuition, is the job of Synthesis of Apprehension. Presumably that is why the full name for the latter, as we noted earlier, is Synthesis of Apprehension in Intuition. True, the Synthesis of Apprehension is about distribution in time alone, not also in space, but distribution is its job.
Strangely, Falkenstein has almost nothing to say about the Transcendental Deduction's threefold doctrine of synthesis. He discusses the four kinds of synthesis articulated in the Analytic of Principles (pp. 242-3) but neglects the three laid out in the Deduction, not even examining how the latter relate to the former. Yet the first three were hardly insignificant to Kant; to the contrary, when he rewrote the Deduction, he explicitly kept all of them.
As a result, Falkenstein may harbour an oversimple picture of what is going on in the different kinds of synthesis. The three kinds of synthesis, Apprehension, Reproduction, and Recognition, capture very different aspects of cognition. Only the third is about recognition or assembly or cognition of objects (Falkenstein speaks about the latter on pp. 165 and 243). The earlier two are about prerequisites of these activities--some would say all the prerequisites of them, including the temporal and any necessary spatial ones. If so, Falkenstein could be right to urge that the activity of arraying intuitions in space and time is very different from recognizing patterns or assembling objects, right moreover that spatiotemporal organizing is an immediate result of how our senses are constituted, and yet be wrong to deny that such arraying is an act of synthesis. The structure that our senses contributes may be precisely what Kant had in mind by the Synthesis of Apprehension. If so, spatiotemporal organizing is one form of synthesis for him.
Relationship to the Deduction of the Categories Falkenstein's analysis faces a still bigger problem in connection with synthesis. If spatiotemporal organizing is as independent of synthesizing activity as Falkenstein urges, how could the categories, unified consciousness, etc., possibly be necessary conditions of experience? Here is the problem. For Falkenstein, particular intuitions arrive already organized as a temporal and, many of them, a spatial array (p. 67). If so, why must we in addition organize them into objects, tie the objects together causally, and so on, if we are to experience them? Surely we could experience them just as they are when they arrive. Yet Kant insists that to experience objects, we must perform complex conceptual tasks on the material of these objects.
In particular, we must link the objects causally. The necessity of causal connection is no mere sideline in Kant. It is central to two of his deepest aims. It is the pivot on which his whole defense of the synthetic a priori status of physics turns. And his argument for the unity of consciousness hangs on it; we have this unity only if we are aware of an array of objects in "one single experience" (A 110), and to have this single experience, we must link the items in it causally. On Falkenstein's account, it is hard to see why we need to do things to have experience of at least simple objects such as spatial and/or temporal arrays.(3)
Falkenstein might try to answer this worry--i.e., that he may be making much of Kant's analysis otiose--in the following way.(4) There is every difference, he might urge, between items being temporally or spatially arrayed, on the one hand, and our seeing patterns in them or assembling them into something new, on the other. Array dots in any way you like, you won't see a triangle unless you also link some of them. Likewise but a fortiori for such complex relationships as causal ones. The trouble is, I can agree with him about this and still wonder why we couldn't be aware of an array as it arrives, without needing to find any patterns in it. If "a knowledge of temporal order ... serves as a basis for a knowledge of causal relations and not the other way around" (p 302), why is knowledge of causal relations a necessary condition of experience? At most it would seem to be a necessary condition of causal experience, and that is perfectly trivial.
Kant himself avoids this problem, at least in the eyes of some interpreters, by denying that cognition happens in separable stages. On this interpretation, Kant sees the whole of cognition from spatiotemporal distribution to recognition and assembly as a single interdependent process. If we turn from space and focus on time, we can see one reason for believing this. Perhaps a spatial array could just arrive already distributed but that is much harder to conceive of for time. We are only immediately (unmittelbar; unmediatedly) aware of one moment or at most a short span of time, though that moment or span continually changes (short span allows for the possibility of a specious present). If so, the only temporal location that an intuition could have as it arrives is `now'. Intuitions at other temporal locations would be either no longer existent or not yet in existence. Now, and this is the important point, to present either past or future intuitions to ourselves, we have to perform a cognitive act, either an act of memory (for past intuitions) or an act of imagination (for future ones).(5) That is to say, temporal organizing is not prior to and independent of other kinds of synthesis. Falkenstein's account seems to work better for space than for time.
Let me close these exegetic comments by saying that I do not consider that I have shown
Falkenstein to be wrong to deny that, for Kant, spatiotemporal organizing is an activity of the
Synthesis of Apprehension. I just don't think that he has shown that he is right. At least as good a
case can be made for the interpretation he is denying as for the one he maintains. Indeed, I am
not sure that there is enough in Kant's text for us to decide one way or the other.
II. Beyond Exegesis
Falkenstein's interpretation of Kant also throws up questions that go beyond exegesis. The first has to do with localization.
Localization If it is not an act of synthesis that arrays sensible intuitions in time and in space, then what does? Remember, for Kant space and time are features of us, not of the intuitions that our receptors receive; intuitions are not arrayed `in themselves'.
For Kant, localization is harder than it might look. Though the point can easily be understated, the pure forms of intuition are very, very different from particular intuitions, so different that it is not easy to see how the latter could come to be located in the former at all. It is not just that the forms exist prior to experience whereas particular intuitions are received by the senses. The forms allow us to discern necessity. As Kant puts it, "experience tells us ... what is, but not that it necessarily must be so." (A1). We need something a priori for that; it is the a prioricity of space and time that allows us to discern necessity by them (Brook, 1992).
Falkenstein tends to understate the distinctiveness of the forms of intuition because he thinks that Kant is concerned merely with hypothetical necessity (pp. 203, 268). But Kant did not think that the propositions of geometry and mechanics are merely hypothetically necessary; he thought that they are necessary and universal sans phrase (B14-18). Any cognitive system whose mere form is enough to establish necessary truths such as these is a very special cognitive system indeed! This distinctiveness raises a problem for Falkenstein's account.
If the forms of intuition are not just a priori but set the limits of the imaginable and the possible, then they are something very different from particular intuitions. The latter arrive a posteriori and never display any kind of necessity. This gap must be bridged by any account of how intuitional particulars get arrayed temporally and spatially. Falkenstein argues that for spatial localization of particular intuitions to be possible, intuitions must be ordered as they are `immediately given' to us. Otherwise, the intuitions would be coming from one source, the localization from another and Kant would have "no way to explain why we place matters in one location rather than another". (p. 85; see also p. 250) But Kant did hold that the source of space and time is different from the source of particular intuitions; the source of the former is a priori, of the latter a posteriori (as Falkenstein himself notes, p. 89). If so and if having distinct sources create a problem about localization, then Kant has a problem about localization. Saying that intuitions are organized "as immediately given" is no help at all.
The next point, and it is crucial, is that once space and time are separated from incoming intuitions in this way, neither perceptual pathways or filters nor (other) acts of synthesis can achieve distribution of intuitions in time and space by themselves. (As we saw, the action of perceptual pathways or filters may be a form of synthesis, but that does not matter for present purposes.) To distribute intuitions to distinct points in space and time, there has to be something in the incoming intuitions that our perceptual system can use to distinguish them (Falkenstein makes a related claim but seems not to have seen its significance [p. 85]). To insist--as Kant sometimes does--that intuitions have nothing spatiotemporal about them when they hit the senses comes close to being an a priori disproof of the possibility of spatial or temporal localization. I can see only one way out: we have to allow that there is something about intuitions as they arrive that allows their spatiotemporal organization. There has to be something in intuitions as they arrive that resists some forms of spatiotemporal organization and facilitates others.(6) Kant says or implies that it cannot be a spatiotemporal structure already in the intuitions. Space and time are forms provided by us. As Falkenstein argues, there is no nonspatial, nontemporal property from which we could infer spatial or temporal location, in the way that shades and saturations can be used to organize particulars in colour `space'. Fine. Then there must be something else in particular intuitions as they arrive that we can use to put them in different locations. Otherwise they could not be localized at all.(7)
The issue of localization has been around at least since the time of Kant and I don't pretend to have exhausted either the issue or Falkenstein's discussion of it. For example, I have said nothing about his interesting treatment of Steinbuch, a 19th century psychologist. I have said enough, however, to expose some potential problems for Falkenstein's account. If particular intuitions must arrive with something that allows us to localize them, then Falkenstein's argument against the heap thesis is both incorrect and incomplete and the argument against synthesis collapses. The argument against the heap thesis is incorrect because intuitions are not spatiotemporally structured as immediately given, not by senses constituted so to organize or by anything else. It is incomplete because, in addition to intuitions and structuring senses, we also require the idea of intuitions containing information that allows them to be localized. The argument against synthesis collapses because the processing necessary to exploit the information contained in arriving intuitions need have no special character. These processes could easily be simply some form of synthesis--even if the action of perceptual pathways and/or filters is not itself a form of synthesis.
Intuition/Understanding `Surely', it will be objected, `there must be more to Falkenstein's account than you are allowing. Synthesis and spatiotemporal organizing are activities of different faculties, namely, Understanding and Sensibility. What difference could be bigger?' I am not sure what this difference amounts to. Notwithstanding Falkenstein's many different ways of marking the distinction between spatiotemporal structuring as "the form in which the information is received" (p. 67) and acts of synthesis, the two are broadly similar in certain ways. Both involve order imposed by the mind and in neither is the imposition a conscious or even an intentional act. One kind of unconscious, undirected information processing is not all that easy to differentiate from another.
What about the distinction between synthesis as acts of the Understanding and spatiotemporal localization as a process in Sensibility? This scheme is only as clear as the distinction between Understanding and Sensibility that underlies it. What is this line such that the tradition locates spatiotemporal organizing as an act of synthesis on one side of it, Falkenstein as a filtering process on the other?
The fundamental idea is that the Understanding consists of activity, synthesizing activity, and intuition in Sensibility is passive receptivity. The first problem is that, following Kant faithfully, Falkenstein offers at least four different ways to characterize the intuitional and therefore the distinction:
immediacy (pp. 60, 67)
receptivity (p. 60)
sensibility (source in a sense modality such as vision or hearing or in something derived from such a modality such a dreaming or imagination) (pp. 29, 94)
physicality (p. 123).
Falkenstein also speaks of intuition as the original and primitive elements in cognition (p. 67).(8) Presumably one or more of these criteria anchors Falkenstein's claim that spatiotemporal organizing is in Sensibility and so is not an act of synthesis. Unfortunately, none of them does the job very well.
`Immediacy', as in `immediately given', is not Kant's most frequent way of marking the intuitional, though Falkenstein often uses it. The trouble is, the notion is far too broad. Most experience of whatever kind is immediate. `Immediate' means `unmediated' (the German term, as I have noted, is `unmittelbar'), i.e., not experienced via another representation. By this criterion, thoughts, judgments and beliefs can be just as immediate as any deliverance of sensible intuition. So being "an ordered manifold of parts ... immediately given in intuition" (p. 81) is quite compatible with a representation having undergone all sorts of cognitive processing. There is no possibility of distinguishing spatiotemporal organization of the manifold from synthesis here. (Flanagan makes a similar point about qualia: thoughts, etc., have qualia--there is something it is like to have them--just as much as visual experiences, dreams, etc. [1992, p. 67-8].)
Kant most frequently marks the understanding/intuition distinction by reference to receptivity (as in `receptivity of impressions, spontaneity of concepts' [A50=B74]). Here the problem is, what does he mean by the term? Does `receptivity' mean `affection by something other than myself'? This won't work for dreams, etc., though they are clearly intuitional for Kant. In any case, it puts space and time on the wrong side for purposes of Falkenstein's analysis; space and time are self for Kant, not other than self. Does `receptivity' mean: `influenced by something over which I have no control or awareness'? But this is true of most of cognitive life, not just space and time, including at least Synthesis of Apprehension and Synthesis of Reproduction. Does `receptivity' have to do with something being a building block of cognition prior any processing by the mind? Again, this puts space and time on the wrong side for Falkenstein's argument; they are aspects of the mind. Or does `receptivity' have to do with things to which I am passive consciously? If so, then again most of cognitive life is a matter of receptivity. In short, I can think of no sense of `receptivity' in which the spatiotemporal organization of the manifold of intuition would be something the mind receives, rather than something that results from activity of synthesis.
Let us take `sensible' and `physical' together; by `physical' Falkenstein really means `sensible'. Falkenstein contrasts the physical or physiological and the cognitive or psychic (p. 123).(9) (This distinction parallels a contrast he sometimes makes between sensory processes and the cognitive (p. 94), though in other places he views intuition as part of the cognitive.) What he has in mind in these distinctions is the general distinction between perceptual and what Fodor (1983) calls central systems. The sensible or physical systems are the processes and outputs that can be mapped fairly straightforwardly onto processes, etc., in perceptual systems (this seems to be why Falkenstein says that for Kant intuition is physiological). The central systems tie outputs of various sensible modalities together, describe them, etc. Falkenstein urges that Kant saw spatial and temporal arraying as happening in the sensible (we would now say perceptual) systems, not in the higher systems. In this he may well be right; indeed, this `physiological' reading of Kant on intuition is ingenious. I think that is does a better job of capturing what Kant really had in mind by the distinction between the faculty of Understanding and the faculty of Sensibility than any of the other suggestions. Now, how well does it support Falkenstein's distinction between spatiotemporal organization and activities of synthesis?
According to current vision research, distribution in a three dimensional spatial array happens in the perceptual systems (temporal distribution is more complicated) but so do a lot of things would have counted as synthesis for Kant on anybody's account: binding of detected quality features, for example, or computation of the 2½D object image from the two-dimensional array sent down from the retina (2½D is a term coined by David Marr (1982) for an image of an object that has dimensionality but not the full range of three dimensional properties of a finished object). In short, no clear functional difference between synthesizing activities and something else is correlated with the anatomical and physiological differences between perceptual and central systems.
Of course, we know a lot more about these things than Kant did. Even given the limitations in what he knew about perception, however, is it so clear that he did not think of the spatiotemporal structuring of intuitions as a form of synthesis?
Part of what supports Falkenstein's claim that spatiotemporal structuring is very different from any act of synthesis is his sense that, somehow, what is "immediately given in sense intuition" (p. 81) is already structured. If so, the structuring could not be an act of synthesis; synthesis is always of already-given intuitions. As we have seen, there is a big problem with this view: for Kant, particular intuitions are not spatiotemporally structured as immediately given; the arrival of particular intuitions and their being arrayed in time and space are two different events. Even if spatiotemporal organization is done by our sensory `constitution', it is still done to intuitions. Putting the point in physiological terms, though intuitions may be arrayed when they leave the receptor systems, they are not arrayed when they hit the receptor systems (though intuitions must contain information that allows them to be arrayed, if our earlier argument is right).(10)
All this seems fairly obvious, so why does Falkenstein say otherwise? One possibility is that he is confusing two different kinds of `immediately given' or `received' representations:
1. What is immediately given to receptors. This is the first element in cognitive processing and is what is immediately given without qualification. This material is not spatiotemporally arrayed as received.
2. What is immediately given to the central systems. This is what is immediately given to the discursive abilities. This material could be spatiotemporally arrayed as received by the system for which it is immediate.(11)
Falkenstein distinguishes what is done to items of type 1 from what is done to items of type 2, of course, but the idea that spatiotemporally organized states in Sensibility are immediately given is another matter. They may be immediately given to the central systems (2.) but they are not immediately given to receptors (1.). When Falkenstein says that spatiotemporally organized intuitions are "the representation immediately given in sense intuition", that they are the "immediate effect of impression of the senses [on] a subject with a certain sensory `constitution'", he can only be thinking of the second type of immediately given.
To summarize. 1. Intuitions may not be a heap by the time the senses get done with them but
they certainly have some heaplike properties upon reception. 2. When incoming intuitions are
arrayed, it is not easy to distinguish the process of arraying them from acts of synthesis, the
Synthesis of Apprehension in particular.
III. Kant's Big Theses
In Part III, the final Part of his book, Falkenstein explores the implications of his interpretation for some of Kant's big theses. He examines the theses that (1) things as they are in themselves are nonspatiotemporal, (2) we have no knowledge of them, (3) we are affected by them and (4) space and time are features of us, not features of things as they are in themselves. I agree with much of Falkenstein's analysis of these issues but have questions about particular aspects of his account of (1) and (2). Even about (1), the nonspatiotemporality thesis, and (2), the unknowability thesis, I agree with him that the hardest test is the mind itself: Kant seems to allow that we know a great deal about the mind as it is in itself and in particular that its denizens, representations, have temporal structure.
The Temporality of Representations as They are in Themselves Kant says that things as they are in themselves are nontemporal, the mind included. Here I want to ask two questions: (1) What does Falkenstein attribute to Kant on this issue?, and, (2) Can we make any sense of Kant's claim? The problem facing Kant is that representations, it would seem, do not just appear to us to be ordered temporally, they are ordered temporally. If so, one thing is temporal as it is in itself, namely, our representations and, by extension, our mind as the thing that has them.
Falkenstein first urges that for Kant even the mind only appears to have temporal structure. Even the mind as it is in itself might not be temporal (or spatial). That indeed seems to be right, what Kant believed. But now, what about representations? Even for representations, did Kant distinguish how they appear from how they are in themselves?
As Falkenstein recognizes, he did:
Thus, as Kant says, to another being, able to intuit the representing subject as it is in itself, those very things that for us occur successively over time ... would appear in some other way altogether, and neither the representing subject nor its representations would be thought, therefore, to be successive in time. [p. 351, my emphasis]
Representations as they actually are do not, or at the very least need not, have temporal order. Now comes the crucial question: What about representations as they occur in us?
Here there are two alternatives:
1. Whatever representations are like in themselves, by the time we are through with them they "occur successively in us".
2. Even as they occur in us, representations only appear to be successive. What they are actually like continues to be something that we cannot know.
The first position is the one that Falkenstein attributes to Kant. I think that the second is the one that Kant actually held.
Falkenstein has to attribute the first view to Kant. All along he has argued that intuitions are arrayed in space and time on receipt (e.g., p. 67). If so, they are arrayed. They do not merely appear to us to be arrayed.
Unfortunately, that is not what Kant says:
I can indeed say that my representations follow one another; but this is only to say that we are conscious of them as in a time-sequence. [A37=B54fn.]
This is the only such passage that I know of but it is extremely clear. Though Falkenstein does not quote this passage, he does imagine Kant laying out the view expressed in it, the view, that is, that even the temporal order that representations appear to have is not real, merely how they appear to us:
Representation A does not actually precede representation B in the subject's mind, ... it is merely represented as prior to B in time. The temporal precedence of A and B is itself merely the object of a representation, and not a manner in which representations actually occur. [p. 342]
His response is surprising: "this [view] is hopeless." (p. 342)(12) For Kant, he says, "our knowledge of temporal relations is ultimately grounded in the fact that certain representations actually occur ... before certain other representations." (p. 343, my emphasis) That is to say, for Kant, appearances not only represent but have temporal properties. And Falkenstein has to attribute this much temporal realism to Kant; he has held all along that for Kant intuitions actually are temporally (and, many of them, spatially) structured. Other things may merely be represented as being in time and/or space but representations are in time.
It is true that Kant talks like this at times but how could he possibly be entitled to do so? His considered view has to be that we no more know the order in which representations `actually occur' in us than we know how they are realized in the brain.
Let us turn now to question (2.): Can we make any sense of this idea of Kant's that the temporal order that representations appear to have might not represent how they actually are? Dennett (1991) has gone further in the direction of making this approach respectable than anyone else. Indeed, Dennett uses the footnote from the Aesthetic quoted above (A37=B54fn.) as the epigraph for a chapter. Start with a point made earlier. The only representations of which we are currently aware are the ones that have the temporal position `now' (and perhaps those in the specious present). All others, being either remembered or imagined, do not have an actual temporal position, thus could only be represented as having one. As Dennett and others have shown, the order in which we represent experiences as coming to us is often not the order in which they have actually come to us. As part of his realism about temporal ordering, Falkenstein urges that it is not possible "to induce experimental subjects to misperceive the order in which stimuli are actually impressed ..." (p. 171). In fact, it is easy. The psi phenomena is the best known example. Circles of two different colours are flashed on a screen one after the other at two different places. Even though the second circle and therefore the first stimulation by the second colour cannot take place until the second circle has been presented, to the subject the circle appears to move from the first place to the second and to change to the colour of the second circle halfway through. That is to say, the circle appears to change colour before the stimuli causing the appearance have actually arrived. Kant merely wants to say that, so far as anything we can know is concerned, what happens in cases like the psi phenomena could be true in all cases.
To summarize. The claim that the temporal order that representations appear to have may not be their actual order is Kant's view and it can be given sense. If so, Falkenstein may need to be modify his analysis of Kant on the spatiotemporal organization of intuition. Falkenstein talks about what does happen. Kant talks about how things end up appearing to us.(13) Unfortunately, to make this change, Falkenstein would have to change a lot of other things; his whole treatment of spatiotemporal structuring is realist. Moreover, constructing appearances is precisely the job that Kant grants to synthesis.
The Limits of Unknowability Kant claims that we do not know (or at any rate know that we know) anything about anything as it is in itself. Falkenstein notes that for Kant there would seem to be a further, particular problem about the knowability of intuitions. This further problem stems from what Falkenstein calls their blindness, as in "Concepts without intuitions are empty, intuitions without concepts are blind" (A51=B74). Falkenstein seems to read this and similar passages to be saying that, prior to conceptualization, we cannot know anything about intuitions, even as they appear (pp. 55-73). Now I am not sure that this is what Kant had in mind; he seems rather to be saying something to the effect that prior to conceptualization, intuitions do not give knowledge. But I will let that pass. I want to ask instead, does Kant's treatment of intuitions violate his general view that to get knowledge, we must apply concepts as well as have intuitions?
Kant's views on the unknowability issue are more complicated than might at first appear. Some kinds of knowledge of the mind as it is in itself are clearly available. When he sets out the necessary conditions of experience in the Transcendental Deduction, for example, he is setting out how the mind must be if it is to have experience, not merely how it must appear to itself. (Falkenstein agrees [p. 242].) The mind must actually use concepts, for example, not merely appear to itself to do so. Thus, Kant thinks that we can know at least some important things about the mind as it is in itself. (If he is right that nothing is temporal as it is in itself, then these actions of the mind would have to be nontemporal. One place where Kant actually makes some use of such an idea, to everyone's bewilderment, is in his Solution to the Third Antinomy [A538=B566--A558=B596].)
How can we reconcile Kant's view here about the mind with his fervent denial that we know anything about things as they are in general? My own way of reconciling the two goes as follows. First, Kant is denying that any unmediated intuition gives us knowledge of anything as it is in itself, including the mind. All knowledge of things as they are in themselves is inferential, for example by inferring the necessary conditions of experience. Second, even the knowledge that we can gain by inference is severely limited. Even inferentially we cannot learn anything about the mind's structure, not even something as basic as whether it is simple or complex, only how it must function (see the attack on the second paralogism [A353]).
That is not to say that we have no unmediated intuitive awareness of the mind. We just do not have any that gives us knowledge of it. Kant allows and can allow that there is at least one kind of unmediated awareness of things and that it does not yield knowledge (B158), the nonattributive awareness that we have of ourselves as ourselves, as the subject of our experiences. It is achieved via acts in which "we designate the subject ... only transcendentally, without noting in it any quality whatsoever--in fact, without knowing anything of it either by direct awareness or by reasoning" (A355). The important point about this kind of awareness of ourselves for present purposes is that it gives us no knowledge of ourselves.(14)
How then should we take Falkenstein's worry about Kant and the blindness of intuition? Kant could not allow that we have any immediate awareness of intuitions prior to conceptualization that yields knowledge; but then I am not sure that he ever does allow this. On the other hand, he can perfectly well allow that we can gain knowledge of unconceptualized intuitions by drawing inferences about what intuitions must be like if our experience is to be as it is--and this he does, repeatedly, using something like Humean distinctions of reason. But perhaps I am missing a problem that Falkenstein sees.
Concluding Remarks In these comments I have identified quite a number of places where either
I don't entirely follow or suspect that I don't agree with Falkenstein's reading of Kant on intuition.
In closing let me repeat what I said at the beginning: Kant's Intuitionism is a fine book, full of
detailed insights into a host of difficult issues in Kant's philosophy. It articulates an original (if not
entirely persuasive) reading of one of the most obscure texts in the whole of philosophy, the
Transcendental Aesthetic. If I seem to find many problems in Falkenstein's account, that is in no
small part because reading the book has forced me to get a lot clearer about a number of things
that I thought understood--clear enough to see exactly where I have questions about
Falkenstein's account. In addition, I have said little or nothing about the many excellent things in
the book with which I am in agreement.(15)
Brook, A. 1992. "Kant's A Priori Methods for Recognizing Necessary Truths", in: The A Priori Revisited, ed. Philip Hanson and Bruce Hunter, Canadian Journal of Philosophy Supplementary Volume 18, pp. 215-52.
Brook, A. 1994. Kant and the Mind. New York: Cambridge University Press
Brook, A. 1995. "Realism in the Refutation of Idealism", Proceedings of the 8th International Kant Congress Milwaukee: Marquette University Press, pp. 313-20.
Dennett, D. 1991. Consciousness Explained. New York: Little, Brown
Flanagan, O. 1992. Consciousness Reconsidered. Cambridge, MA: MIT Press
Fodor, J. 1983. The Modularity of Mind. Cambridge, MA: MIT Press
Kant, I. (1781/1787). Critique of Pure Reason (trans. N. Kemp Smith in 1927 as Immanuel Kant's Critique of Pure Reason. London: Macmillan Co. Ltd., 1963).
Kitcher, Patricia. 1990. Kant's Transcendental Psychology. New York: Oxford University Press
Marr, D. 1982. Vision: A computational investigation into the human representation and processing of visual information. New York: W. H. Freeman
1. 1. There is also the Synthesis of Reproduction in the Imagination. It places little role either in Falkenstein's account or in my comments--probably too little role.
2. 2. Kant, I. (1781/1787). Critique of Pure Reason (trans. N. Kemp Smith in 1927 as Immanuel Kant's Critique of Pure Reason. London: Macmillan Co. Ltd., 1963).
Falkenstein quotes this passage, too, and discusses the Synthesis of Apprehension briefly. However, he does not examine whether it might be a counter-example to his general thesis. His suggestion that it was deleted from the B-edition and therefore cannot be considered canonical is not altogether sound, incidentally. The passage was deleted but the notion of Synthesis of Apprehension most assuredly was not, and there is no reason to think that Kant changed his mind about its character (B-edition, §26). In fairness I must allow, however, that some of what Kant says about it there supports Falkenstein's general view.
3. 3. Interestingly, Falkenstein worries that Kant himself backs into this problem without noticing it (p. 68).
4. 4. Falkenstein considers the relationship of Kant's early view of space and time in the Inaugural Dissertation to the Transcendental Deduction but not his own view (pp. 68-9).
5. 5. As I argue elsewhere xx, this observation that we are aware of only intuitions of the present moment may also help to explain Kant's strange notion that "time ... cannot be a determination of outer appearance" (A33=B49; see A23=B37).
6. 6. The same argument holds, mutatis mutandis, for conceptual organization using the categories: there has to be something that resists some conceptualizations and facilitates others.
7. 7. Falkenstein's own thoughts on the matter may be taking a similar direction (`Response to Commentators', North American Kant Society/Canadian Philosophical Association Symposium on his book, Memorial University, June 3, 1997, and personal communications). As Falkenstein notes (p. 110), Kant says nothing directly to the point but Aquila (1989) is a recent commentator who urges that for Kant intuitions as they arrive already have something that allows their spatiotemporal placement (the issue has been around for a long time). Unfortunately, supporting texts are scarce. However, if intuitions prior to our contribution provided no information useful for sorting them spatiotemporally and conceptually, the result would be some form of solipsistic relativism, about the farthest thing from Kant's mind (though the risk of solipsism is real on some versions of the unknowability thesis and seems nearby at some points in Falkenstein's analysis [p. 316 for example]). In fact, contrary to Kant, this information may itself have to be temporal. As Falkenstein notes, Kant talks about becoming aware of intuitions only after they have been spatiotemporally organized, subject to synthesis, etc. (p. 141).
8. 8. In fact, Kant offers more than these four (or five) ways. He also maintains that intuitions are singular representations whereas products of the understanding can be common to a number of particulars. Falkenstein argues, convincingly to my mind, that this difference falls out of more basic differences between intuition and understanding and is not definitive in its own right. He also argues, again persuasively, that Kant equivocates over `intuition' here. In any case, singular representations could clearly be the result of a lot of cognitive processing and so are not at all the kind of thing that Falkenstein needs.
9. 9. Falkenstein's contrast between the physical and the cognitive is unfortunate; for any materialist, cognition will be just as physical as anything intuitional. This move does suggest that by `physical' here Falkenstein means `sensible', however. (He tries to avoid begging the question of whether the cognitive is physical [p. 119] but his terms belies his effort. The same is true of his attempt to treat the intuitional as physiological: all cognition is physiological for a materialist.)
10. 10. This point also raises a question about whether reception and distribution of particular intuitions are intuitional processes as Falkenstein defines the latter. "If you believe that a certain output is already contained in the input to a processor, ... then you are what I call `an intuitionist'." (p. 7) By this definition, Kant was not an intuitionist about sensible intuition. I think that this is merely a terminological problem and will say no more about it.
11. 11. Actually we can distinguish a third, too, though it is not relevant to Falkenstein's analysis:
3. What is immediately given to consciousness.
For Kant, these items would be the results of what is done to items of type 2. They are the immediately given of those things that are `something to me' (A111, A116, A120, B131-2).
12. 12. Falkenstein also takes me to task for advocating this reading (p. 378 n.3; I make the claim in 1994, p. 197 and elsewhere).
13. 13. Falkenstein's reading is clarified and perhaps modified in `Response to Commentators', North American Kant Society/Canadian Philosophical Association Symposium.
14. 14. Falkenstein may have the Refutation of Idealism wrong when he says that Kant maintains in it that we are aware even of ourselves only as we appear but my views on it are controversial (Brook, 1995).
15. 15. I would like to thank Lorne Falkenstein, Robert Stainton and two anonymous referees for this Journal for many helpful comments and suggestions. Lorne Falkenstein's contribution has been particularly important. Given how hard I push him on some issues, his willingness to push back again and again until I finally got clear about what I wanted to say was exceptionally valuable.