I. Introduction

SO FAR our attention has been mainly directed toward the first-edition Deduction. In Chapter Three, I observed that the later version, perhaps even more than the first, seems clear in regarding the primary notion of conceptual “synthesis” as involving a “manifold of representations” that are in some way ingredient in intuitions themselves. It seems clear, in other words, that we are not there concerned with a synthesis of representations that are merely externally connectible, by means of concepts, to additional representations. (As we shall see more clearly later, we are not, for example, merely concerned with representations that are connectible in virtue of being “ascribable” to a common subject.) Appropriately construed, the point can be used to support the reading of Kant that I have proposed. But I have yet to defend it in terms specific to the second-edition Deduction.

Whether or not, as often suggested, the second edition departs from the first, it seems plainly to attribute transcendental synthesis to the faculty of intellect alone. Thus it appears to degrade reproduction, and imaginative association, to contingent psychological factors. When it is regarded transcendentally, or “productively,” imaginative syntheses seem to be nothing other than instances of the understanding itself at work upon a manifold of intuition. This provides an obvious challenge for our reading of the Deduction.

The present chapter will not exhaust our consideration of the issues raised in the second-edition Deduction. In the next, I focus on the problem of synthesis in both versions of the Deduction, with specific attention to the notions of consciousness, unity of consciousness, and self-consciousness. For the present, we may speak more generally of synthesis and of unity of synthesis without considering these other notions in detail. This may seem unwise, since these notions are obviously crucial to the Deduction. In particular, one is often inclined to formulate the central problem of the Deduction in terms of a problem concerning the “ascribability” of manifolds of states of consciousness to subjects. To a certain extent, I do not disagree with this. I agree that Kant’s central argument for the necessity of applying categories to objects is that applying them is necessary for self-consciousness, in particular for consciousness of oneself as an individual to whom representations are “ascribable.” I shall argue later, however, that most readings go wrong in their neglect of synthesis as an irreducible form of one’s consciousness of appearances themselves—that is, of objects.¹

In the second section of this chapter, I try to show that the issue is complicated by some ambiguities that must have been difficult for Kant himself to resolve. One of them bears on a shift made possible by a notion that I have taken to be central to Kant’s entire enterprise, namely, by the notion that objects of possible cognition, qua objects of possible cognition, may always be considered as intentional correlates of the apprehension of them. Another ambiguity may account more specifically for the impression that, at least in the second edition, conceptual “synthesis” involves purely intellectual functions. The ambiguity that accounts for the impression will be the same ambiguity that, as we saw in Chapter Three, accounts for Kant’s hesitation as to whether concepts are “rules” or merely embody or contain rules.

I postpone, until the fourth section, a consideration of the overall structure of Kant’s argument in the Deduction, in particular its division into two stages. I then suggest that the relationship between stages be read as involving a transition from synthesis regarded “noetically”—that is, as an operation applied to a manifold of anticipations and retentions, ingredient in intuitions—to synthesis regarded “noematically”—as the intentional correlate of such operations. (As we shall see, there is a second, derivative logical inference that might also be intended: from the correlate of the mere apprehension of appearances—taken without regard to any questions of reality—to the correlate of that same apprehension, taken as the real world of nature.) Our discussion will also benefit from a reconsideration of Kant’s distinction between judgments of “perception” and of “experience” in the Prolegomena. Far from rejecting the distinction in the Deduction, it is precisely what is needed, in that terminology or another, for a grasp of the latter’s structure.

II. The Ambiguity of Synthesis

We have already seen the possibility of two points of view concerning concepts and the syntheses they involve. On one of them, we are concerned with concepts as a kind of “term,” already at our disposal, and available for “application” to intuitions. In that case, we can only ask what sort of “synthesis” is involved in that application, insofar as it concerns the relating of a given intuition to others (or, alternatively, insofar as it concerns the relating of one correlative appearance to others). Construed in this way, we are likely to regard synthesis purely intellectually. For we are likely to regard it purely in terms of actually (however explicitly) conceived connections among (possible) intuitions or appearances—or at least in terms of connections that one is logically committed to representing conceptually—in any instance of concept-application. In that case, the “transcendental” structures of synthesis might simply be regarded as the most general and necessary ways in which one is able to think about things (insofar as one is required to do so in terms of relations among appearances).

The second point of view is very different. It considers how concepts are “possible” in the first place, in a sense that does not merely bear on their logical implications, nor on the logical implications of their application to objects. Instead, it considers what concepts, hence conceptual acts, are as such. From this point of view, we may of course continue to focus on what Kant himself regards as purely “intellectual.” The distinctively conceptual dimension in any conceptualized intuition is indeed constituted by the “application” to it of a purely intellectual function. The question is how to regard that function. Obviously, we cannot regard it in terms of the application of already available concepts to objects, that is, on the model of predication. I shall argue that we must rather be concerned with the “application” of a special function of consciousness that, while not in itself conceptual (at least in any ordinary sense), suffices for the original constitution of a concept in intuition. (Specifically, it must suffice for the constituting of a concept out of a manifold of material to be found within the intuition in question.)

It may be misleading to speak of the “application” of anything at all in this case. In any case, however such functions may otherwise be described, they must function by constituting some kind of unity with respect to a manifold of material ingredient in intuitions themselves:

Understanding is, to use general terms, the faculty of cognitions. This consists in the determinate relation of given representations to an object; and an object is that in the concept of which the manifold of a given intuition [das Mannigfaltige einer gegebenen Anschauung] is united. (B137)

To be sure, shortly later Kant also says that the relevant problem of unity is one that concerns the “ascribability” of a manifold of distinct representations to a subject:

This proposition . . . says no more than that all my representations in any given intuition [in ingendeiner gegebenen Anschauung] must be subject to that condition under which alone I can ascribe them to the identical self as my representations, and so can comprehend them as synthetically combined in one apperception through the general expression, ‘I think.’ (B138)

But whatever we make of the connection between unifying a manifold in a single intuition, and ascribing distinct intuitions to subjects, the fact remains that the former is depicted, at least in these passages, as an operation with respect to manifolds within single intuitions.

However we provide the details, a number of other passages make the same general point:

The transcendental unity of apperception is that unity through which all the manifold given in an intuition [alles in einer Anschauung gegebene Mannigfaltige] is united into [n.b.] a concept of the object. (B139)

The manifold given in a sensible intuition [Das mannigfaltige in einer sinnlichen Anschauung Gegebene] is necessarily subject to the original synthetic unity of apperception. . . . All the manifold, therefore, so far as it is given in a single empirical intuition [sofern es in Einer empirischen Anschauung gegeben ist], is determined in respect of one of the logical functions of judgment, through which it is brought to one consciousness. (B143)

A manifold, contained in an intuition which I call mine [das in einer Anschauung, die ich die meinige nenne, enthalten ist], is represented through the synthesis of understanding as belonging to the necessary unity of self-consciousness; and this is effected by means of the category. (B144)

Of course, all these passages are neutral as to what the basic material is that needs to be combined within single intuitions. In this regard, Kant may seem to hold a different view from the one that I have proposed. For example, he illustrates the claim “that all the manifold of intuition should be subject to conditions of the original synthetic unity of apperception” by reference to some kind of manifold contained in space and time themselves:

Space and time, and all their parts, are intuitions, and are, therefore, with the manifold which they contain, singular representations (vide the Transcendental Aesthetic). Consequently they are not mere concepts through which one and the same consciousness is found to be contained in a number of representations. On the contrary, through them many representations are found to be contained in one representation, and in the consciousness of that representation; and they are thus composite. The unity of that consciousness is therefore synthetic and yet is also original. The singularity of such intuitions is important in the application (vide §25). (B136n)

There is a similar passage in a footnote to §26 as well. (It is generally assumed that Kant meant to refer ahead to this passage, not to §25.) The later section is the one that is in fact supposed to provide the second of the two “stages” of the Deduction as a whole. This has led to the supposition that the transition between the stages is simply a transition from the need for a purely intellectual synthesis in any instance of cognition as such to the need for the application of such synthesis to perception in particular, in order to generate spatiotemporal unities as objects of the latter.² How an intellectual synthesis might in fact account for the unity of intuitively apprehended space and time, as opposed to the unity of space and time in mere thoughts and judgments³—and what sort of “manifold” that synthesis could be applied to, in order to yield such results⁴—remain unanswered questions. In both passages, in any event, Kant seems to focus on synthesis in regard to the manifold of sub-regions in an intuitively apprehensible region of space or of time. Other accounts might postulate yet other “material” as well.⁵

One might suppose that Empfindung is meant to satisfy this need. It is at least something sub-intuitional. But it cannot satisfy the need now in question. Its role, in Kant, is to account for the fact that objects are presented in intuition in irreducibly sensory manners. Other things being equal, it accounts for the fact, for example, that a region of space is seen as filled with a certain color, and is not (whether correctly or not) merely imagined or thought to be so. But what we are now considering is functions required for the conceptualization of objects. This cannot be sensation, because we are capable of conceptualizing objects in pure imagination. It cannot be sensation for another reason as well. Sensation cannot account for the temporal dimension in sensory apprehension. To be sure, sensations belong to our “inner” condition; to that extent, they are “temporal.” But the ingredience of sensation in an intuition is not what confers a temporal dimension on an object of intuition. Apart from whatever temporal concepts might be in question for the purpose, it is the temporal form of intuition that does that. (Of course, it is essential that the relevant concepts specifically bear on that form. But this is accommodated through the role Kant assigns to anticipation and retention in the structure of conception itself.)

Kant indeed suggests that ordinary objects, or at least the “appearances” of them, are somehow formed through the synthesis of an unusual kind of material, unavailable to ordinary awareness. But this may be due to no more than a tendency to shift between the standpoint of cognitive states and the intentional correlates of them. In any event, the proposed account requires nothing more strange than anticipations and retentions as the basic material for synthesis. Of course, these are at most material on the “noetic” side of consciousness. Something needs to “correspond” to them on the noematic side as well. But there is no reason to suppose that, whatever this is, ordinary appearances are made out of it, in anything like the sense in which cognitive states might be made out anticipations and retentions. To the extent that we are dealing with mere intentional correlates, the supposition would be nonsense.

Now we may look at the second of the passages to which I earlier referred. It is the footnote to a passage almost identical to the note to B136. Beyond raising the issues on which I have been commenting so far, it might be considered a difficulty for my original reading of the Aesthetic, thus for extending that reading into the Deduction:

Space, represented as object (as we are required to do in geometry), contains more than mere form of intuition; it also contains combination [Zusammenfassung] of the manifold, given according to the form of sensibility, into an intuitive representation, so that the form of intuition gives only a manifold, the formal intuition gives unity of representation. In the Aesthetic I have treated this unity as belonging merely to sensibility, simply in order to emphasize that it precedes any concept, although, as a matter of fact, it presupposes a synthesis which does not belong to the senses but through which all concepts of space and time first become possible. For since by its means (in that the understanding determines the sensibility) space and time are first given as intuitions, the unity of this a priori intuition belongs to space and time, and not to the concept of the understanding. (B160n)

In the Aesthetic, Kant argued that our ability to apprehend regions of space and time—whether through sensory apprehension of identifiable regions, apprehended as filled with matter, or merely through the imagining of would-be regions—requires a pre-intellectual intuitional “form” in the cognitive states involved. Presumably, the apprehension of such regions amounts to the apprehension of particulars or individuals of a sort. In one sense, it would therefore involve a distinctive “unity” that concepts cannot provide. At the very least, it would do so because it involves the apprehension of unities, namely, the apprehension of the particulars in question. But Kant now says that the form of intuition “gives only a manifold.” What does this mean? It may seem to imply that the original demand for “intuitional” form could not of itself have been a demand for a special function of “unity.” At most, it will have amounted to the demand for an extraconceptual sensible “context” into which concepts and judgments might first introduce any real unity in the first place.

As before, the question would remain as to the nature of the “material” contained in such contexts. Apart from that, it is important to be clear what the passage in fact admits and excludes. In particular, even if Kant is granting that an unorganized manifold is all that intuition has of itself to offer,⁶ it is clear that it is neither empirical concepts nor any others to which he attributes the eventual responsibility for unity. The responsibility is unambiguously attributed to the faculty of understanding, insofar as the latter is able to “determine” sensibility. But despite this, Kant says, the unity in question is one that “precedes any concept.” In addition, Kant says, it “belongs to space and time.”

It is not clear what these claims can mean, if the unity in question does not stem from intuition itself. Allison offers a suggestion:

[T]his cryptic remark can be taken to express the view that the intuited unity of space and time is distinct from the conceptual unity that is imposed upon representations in a judgment. . . . In the case of a judgment the unity belongs to the pure concept of the understanding, whereas in the case of the intuition it belongs to the intuited content.⁷

Presumably, the “intuited content” is an intuitively apprehended space or time. It is obvious that its unity is not like the unity in a judgment or a concept. The point should have been too obvious to have needed stating in the first place. To be sure, Kant had devoted a separately numbered argument, in the Aesthetic, to distinguishing the way in which a concept contains representations “under” itself from the way in which an intuition contains representations (B40). But in both cases, I suggest, what Kant is trying to tell us must rather concern the structure, not of “intuitions” in the sense of objects of intuition, but of intuitional states as such.

It is obvious that the unity of a spatiotemporal region is different from the unity “in” a concept or a judgment. Even Leibniz—whom one might naturally take as Kant’s foil in this case— could grant the point. At most, Leibniz might claim, the unity of a spatiotemporal region is the intentional correlate of the unity embodied in a concept or a judgment.⁸ What is not obvious is not this, but rather the distinction between structure internal to intuited objects and structure internal to intuitional states themselves (insofar as the latter are in fact conceded to have an autonomously functioning structure). Kant plainly states, after all, that the unity in question “belongs to space and time” precisely in order to illustrate his claim that the unity is involved in an act by which the understanding “determines the sensibility.”

Now, Kant often speaks of the understanding, or of concepts, as “determining” objects or appearances. In that case, he presumably means that those objects or appearances are (at least “problematically”) subsumed, in intuition, under particular predicates. But this could not be the sense in which understanding determines sensibility. The latter concerns an “effect” of some sort on a subject’s internal condition, not merely the intentional correlate of such effects with respect to objects. In any event, however the relevant unification is supposed to be effected, the suggestion that it forms intuited objects (or even mere “appearances” of objects), out of manifolds of some material, could not help to make any sense of Kant’s theory. It leaves us doubly mystified: as to what the unification in question involves and as to what the manifolds themselves are composed of.

In addition, Kant says that the act of unification is one “through which all concepts of space and time first become possible.” So he specifically sees that act, not as accounting (or at least not only, or not directly, as accounting) for the unity of intuited objects or “contents,” but rather for the constitution of one’s concepts of objects. In whatever manner we take “intuited contents” to be generable out of manifolds, it is impossible to suppose anything like that to be relevant to the generation of concepts themselves.

There is another difficulty in the passage. Again, as to the claim that “the form of intuition gives only a manifold”: we would need to suppose that, as so far explained, Kant must be putting his point in an extremely odd manner. Why should he say that the form of intuition gives only a manifold? Why not simply say that intuition itself does so? If his point really is that intuition provides unorganized material for understanding to unify, then why should Kant say that the form of intuition is what provides that material?

One might suppose that Kant only means that intuition, by its very nature, is what provides the unorganized manifold. But it is surely clear that, throughout the Critique, the notion of intuitional “form” plays a weightier role than this. Indeed, in the passage that we are examining, Kant explicitly contrasts the mere form of intuition with full-blown “formal” intuition. It would be an uncharacteristic pun, if by the first he just meant intuition itself. In addition, it is not the case that, in Kant’s view, intuition by its very nature is what “gives” the appropriate material. If the material is “sensations,” the point should be obvious from the start. For there is a purely imaginative kind of intuition, as well as what involves sensations. But even if the relevant material is presumed to be autonomously imaginative, the point would be the same. It is not intuition as such that requires the ingredience of such material: the requirement stems from the demand that intuitions be conceptualized. If it is not intuition as such that requires the material to be given, then it is difficult to see why Kant should claim that the very “form” of intuition is what “gives” it.

Perhaps another suggestion is this: We might take Kant to be saying, not that intuition as such “gives” only a manifold, but that space and time do so. The understanding is then needed in order to synthesize this manifold. But that is not possible either. Of course, we might always take it as no more than the claim that full-blown cognition presupposes some kind of “synthesis” of spaces and times, and of appearances in space and time. But the question remains why Kant should say that space and time themselves are what “give” the items thereby to be synthesized. If anything “gives” the manifold in question, it could hardly be space and time. That would require taking space and time as things existing in themselves. At most, it seems, we should say that they contain whatever manifold is in question.

Certainly, this was a point on which Kant had insisted in the Aesthetic: that spaces and times necessarily contain smaller spaces and times. But Kant could not, of course, be saying that the mere containment of spaces and times within larger spaces and times is what provides the understanding with its original material for synthesis. A central doctrine of the Aesthetic is precisely that our original access to the parts of space and time is by means of the introduction of “limitations” into the latter (A25/B39). So while space and time necessarily contain a manifold, and while it is a manifold that the understanding will eventually need to synthesize, it makes no sense to suppose that the manifold is one that is originally given, as material for such synthesis, simply in virtue of being contained in that very space and time. Any relevant synthesis of the parts of space and time, however necessary it may be, must be the upshot of a more fundamental kind of synthesis.

So we need a very different explanation of Kant’s claim that the manifold of intuition is given by the “form” of intuition. The explanation, I suggest, is that Kant wants to emphasize— obviously—the eventual role of the former in acts of conceptualization as such, and that—less obviously—any manifold of material that actually plays such a role in the first place must be supposed to enter into intuitions, in a sense that can be formulated only in terms of the notions of matter and form so far presupposed. Kant’s suggestion that the “form” of intuition is what “gives” the material in question may simply be a way of saying that, in operating on intuitions, so as actually to conceptualize the latter, understanding cannot operate merely in regard to a manifold of distinct intuitions. Nor can it operate merely in regard to a manifold of some other, perhaps sensory or even autonomously imaginative, material. Rather, it must operate on a manifold that is appropriately contained within intuitions. That the latter condition is satisfied might well be put by saying that we need to be dealing, not simply with something that is “given in” intuition, but with something that is given through the very form of intuition. (It should be equally clear, of course, how the proposal applies to a second claim that we have encountered in this passage, namely, that the material in question is not only given through the “form” of intuition, but is a material “through which concepts of space and time first become possible.”)

There is an obvious shift of focus between the Aesthetic and the Analytic. The latter is concerned with the possibility of conceptualization as such. This suffices to account for at least a good part of the apparent “departure” from the Aesthetic. Another factor may also contribute. For the purposes of the Aesthetic, there was no harm in considering the pure “form of intuition” as such as correlative (on the noetic side) with space and time as “objects” in their own right. As we saw in Chapter One, there is a point to such formulations, even if we concede, as Kant does in the present passage, that intellectual functions are also needed in order for space and time, and for regions and appearances in them, actually to be given as “objects.” Apart from what may be verbal quibbles concerning the latter term, the substantive point is the same: whether or not, apart from concepts, intuitions give “objects,” conceptual acts are required to operate, not merely on this or that material, nor even on manifolds of specifically imaginative material, in order to generate awareness of objects. They need rather to operate on manifolds of material that are ingredient in the appropriate way in intuitional states.

Whether a state is in fact an intuitional state is determined neither by the nature of the material ingredient in it nor by the operations upon it. If the latter are indeed required for a true apprehension of “objects,” it remains the case that those operations could never of themselves determine that their target material is at the same time material through which an object is to be apprehended in intuition. It is not implausible to suppose that it is the difficulty of these notions, and of the choice of terminology for conveying them—together with the general shift of focus between the Aesthetic and the Analytic—that accounts for Kant’s apparent uncertainty as to whether transcendental synthesis is merely a direct application of understanding to intuitions or whether it does not also essentially involve an independently imaginative component.⁹

On this account, there is a sense in which neither “conceptual” syntheses nor “unities” of synthesis are purely intellectual. That is, there is a sense other than the one that is already obvious: that conceptual syntheses or unities of synthesis are always “responses” to, or based on responses to, concrete sensory or imaginative conditions. Depending on what one takes to be the internal structure of “response,” concession of the latter may or may not be sufficient. One advantage of the proposed view is that it in any case uncovers an ambiguity in the concession. Accordingly, it can help us to see the ambiguity in the question whether “conceptual” synthesis indeed involves an “imaginative” dimension.¹⁰ Kant, for example, says the following:

This synthesis of the manifold of sensible intuition, which is possible and necessary a priori, may be entitled figurative [figürlich] (synthesis speciosa). . . . But the figurative synthesis, if it be directed merely to the original synthetic unity of apperception, that is, to the transcendental unity which is thought in the categories, must, in order to be distinguished from the merely intellectual combination [Verbindung], be called the transcendental synthesis of imagination. . . . But inasmuch as its synthesis is an expression of spontaneity, which is determinative and not, like sense, determinable merely, and which is therefore able to determine sense a priori in respect of its form in accordance with the unity of apperception, imagination is to that extent a faculty which determines sensibility a priori; and its synthesis of intuitions, conforming as it does to the categories, must be the transcendental synthesis of imagination. This synthesis is an action [Wirkung] of the understanding on the sensibility. (B151-2)

Spontaneity is, of course, officially attributed to the understanding (cf. A51/B75). As a mode of intuition (B151), imagination is apparently not in the sense in question “spontaneous.” But in this passage, Kant ascribes imaginative synthesis to a faculty of spontaneity. A page later, he also claims that what “determines” sensibility—as he had already maintained imaginative synthesis does—is always the understanding:

Thus the understanding, under the title of a transcendental synthesis of imagination, performs this act upon the passive subject. . . (B153). . . . which is possible only through the consciousness of the determination of the manifold by the transcendental action [Handlung] of imagination (synthetic influence of the understanding upon inner sense), which I have entitled figurative synthesis (B154).

Unsurprisingly, Kant draws a conclusion regarding the very capacity for a “synthesis of apprehension.” In the first edition, this level of synthesis, or this aspect of synthesis, had been explicitly distinguished both from that of reproductive imagination and from the synthesis of recognition in a concept. Kant now ascribes it to the exercise of an intellectual faculty:

In this manner it is proved that the synthesis of apprehension, which is empirical, must necessarily be in conformity with the synthesis of apperception, which is intellectual and is contained in the category completely a priori. It is one and the same spontaneity, which in the one case, under the title of imagination, and in the other case, under the title of understanding, brings combination [Verbindung] into the manifold of intuition. (B162n)

This is certainly compatible with preserving some autonomy for imagination, and not having to reduce it to the work of understanding. For example, to claim that transcendental imagination is an action of understanding is compatible with supposing that empirically “reproductive” imagination also provides an essential element in conceptualization. As we have already seen, however, a concession of the sort is ambiguous. It might, on the one hand, mean something that nobody would think to deny, namely, that acts of empirical conceptualization essentially “refer” to empirically reproducible possibilities, or in some way involve conceptions or judgments regarding them. On the other hand, it might mean something more interesting. It is in any case compatible with the passages in question to claim that even “reproductive” imagination is much more literally ingredient in any act of conception as such. The passages are compatible with maintaining that, to the extent that concepts are themselves an element in intuitive apprehension (and do not merely embellish the latter through attachment of quasi-linguistic “terms” of some kind), acts of conception as such—and so, in a sense, the corresponding “concepts”—must actually be made out of imaginative material.

It is possible to hold that, in a certain sense, conceptual syntheses are, as such, purely “intellectual,” and also to hold that empirical concepts are nothing but imaginative material, intellectually transformed or elevated. (To say the latter, of course, is just to say that, in another sense, conceptual syntheses are not as such purely intellectual.) Obviously, there must be something that is purely intellectual in Kant’s view. For there must at least be the functions needed for the embodiment of imaginative material, not merely in intuitions, but in concrete concepts in action. This is something purely formal and purely intellectual. But it is only a matter of terminology whether we limit our notion of “transcendental synthesis,” or of transcendental Verbindung, to this formal element as such or instead consider it specifically in its role in regard to imaginative material. Certainly, to the extent that empirical concepts embody anticipations and retentions, it is at least permissible to say that they embody imaginative “syntheses”—beyond whatever intellectual functions of Verbindung are in them. What the latter can at most add is a special kind of unity of the former:

But the concept of combination includes, besides the concept of the manifold and of its synthesis, also [noch] the concept of the unity of the manifold. Combination is representation of the synthetic unity of the manifold.¹¹

An analogy may help. We might say that the difference depends on whether we are oriented toward the animate “body” of a concept as such or rather toward its animating “soul.” What is important to see, in these terms, is just that Kant’s understanding of the connection between the relevant souls and bodies, or between the form and the matter of concepts, is more like Aristotle’s and less like Plato’s notion of such relationships. The Platonic model may seem to be suggested by Kemp Smith’s translation of the passage quoted from B152. Kemp Smith interprets Kant as saying there that conceptual synthesis is an “action” of the understanding on sensibility. But perhaps he meant only to suggest that it is the effect of some action.¹² Kant’s choice of the noun (Wirkung) allows for either. In any event, Kant seems to prefer a different term altogether when he specifically wants to express the idea of an action in such cases. Thus at A57/B81, he refers to the pure concepts of understanding as Handlungen of pure thought.¹³

If the distinction between matter and form in conception were a distinction between two distinct items or sorts of items—only externally related or at most related in that one of them “refers” to the other—then, despite whatever imaginative “dimension” might be conceded to conception, any relevant synthesis literally internal to conceptual acts, as such, might indeed be unambiguously ascribed to a faculty of pure understanding. And sometimes Kant appears to adopt this approach. But other times he attributes synthesis to an imaginative capacity that is distinct from understanding; the latter is merely responsible for the unity of the former’s synthesis. Where the relationship is indeed one of matter to form, it is not surprising that the upshot is thus ambiguously describable. For in that case, it would be as proper to say that conceptual syntheses are anticipations and retentions as it would be to say that sensory intuitions (at least in abstraction from their conceptual components) are manifolds of sensations. In a sense, a sensory intuition is, as such, only sensations. It is sensations appropriately “animated.” Analogously, the conceptual syntheses introducible into sensory intuitions are just a certain portion of the latter’s ingredient anticipations and retentions. They are the latter—appropriately “animated.” (A non-sensory intuition can of course have no other ingredient, in the relevant sense.)

It may seem to be only in the first edition, or perhaps more prominently there, that Kant credits imagination with the primary role of “synthesis” in conception, and understanding only with the “unity” of that synthesis.¹⁴ But if not in the same terms, the second edition makes the same point. Kant says, for example, that he would call all intellectual combination (Verbindung) a form of “synthesis”— except for the fact that “the concept of combination includes, besides the concept of the manifold and of its synthesis, also the concept of the unity of the manifold.” In particular, “combination is the representation of the synthetic unity of the manifold” (B130; Kant’s emphasis). Thus the thoroughgoing identity of apperception, far from being identical with a synthesis of representations, “contains” (enthält) such a synthesis and is only possible through the consciousness of that synthesis (B133), or at least through the consciousness of its possibility (B134); “it is imagination that connects (verknupft) the manifold of sensible intuition; and imagination is dependent for the unity of its intellectual synthesis upon the understanding, and for the manifoldness of its apprehension upon sensibility” (B164).

The specific issues raised by appeal to a “consciousness” of synthesis need to be postponed to the next chapter. Until then, we must also postpone a full understanding of the “unity” Kant has in mind. But two comments may still be made about these passages. First, we may note the ambiguity in Kant’s reference to “the unity of its [that is, of imagination’s] intellectual synthesis.” We might, of course, take the point to be that, qua transcendentally imaginative, imaginative synthesis is a function of the intellect. But the genitive might also be read as the objective genitive: imagination depends on the intellect for the unity of whatever intellectual synthesis is directed toward itself. This in turn is compatible with two positions. It is compatible with holding that imagination is responsible for no synthesis at all. But it is also compatible with holding that it is responsible for some type of synthesis, but just not for a distinctively intellectual type of synthesis. (The latter reading is viable even if the genitive is possessive.) In any event, it would be a quibble to claim that, apart from the intellect, imagination is responsible for no synthesis at all, in view of Kant’s concession that it is at least responsible for some kind of Verknüpfung, and of his hesitation in identifying intellectual Verbindung with synthesis precisely because every instance of the former must always include the latter.

The second point is this. Kant says that the “manifoldness” of imaginative apprehension stems from sensibility. This may seem to imply that it does not stem from imagination itself, and thus it may seem to contradict the suggestion that conceptual syntheses unify imaginative anticipations and retentions. It might even seem to imply that conceptual syntheses unify mere sensations. But we need to remember that, to the extent that it is an ingredient in intuitions, imaginative material is indeed analogous to sensations. We need only remember, in addition, that to the extent that such material is also embodied in concepts, “applied” to intuitions, it is responsible for connecting, thus responsible for “synthesizing,” a given intuition with a manifold of other possible (past and future) ones. No purely intellectual function does that on its own. In any particular context, Kant might be focusing on one or the other of these aspects of the transcendental function of imagination. Each is essential to the latter’s contribution to conception.

However we finally elaborate the notion, the unity of consciousness required with respect to the “manifold contained in an intuition which I call mine” always includes in itself (in sich schliesst) a synthesis of the manifold given in that intuition (B144 and note). In turn, the “whole power” of understanding, in any such unity of consciousness, consists “in the act whereby it brings the synthesis of a manifold . . . to the unity of apperception” (B145). This act, Kant tells us, is effected by means of the categories, or by means of the “pure forms” of understanding. Thus, in the first instance (though not necessarily first in temporal order), the understanding does not unify a manifold in virtue of applying empirical concepts to it. Whatever original unity it introduces is something much purer than this. In that this unity is partly constitutive of concepts in the first place, applying concepts to intuitions could at most be the upshot of (though not necessarily temporally subsequent to) the presence of such a unity in intuition. Any determinate conceptual “content” is only the upshot of the “imposition” of a correspondingly indeterminate form. We shall have to do more to clarify what such “imposition” involves. But we need at least to be clear that, in any case, what the unity in question is supposed to unify is not in the first instance manifolds of distinct intuitions. It is rather manifolds of a certain material in intuitions. It is only through unification of the latter that the understanding is able to unify the former, in any sense that could be said to involve the “application” or “predication” of concepts.

That any truly conceptual synthesis presupposes a pre-conceptually synthetic consciousness of some kind, seems also to be what Kant is arguing in the following passage:

The analytical unity of consciousness belongs to all general concepts, as such. If, for instance, I think red in general, I thereby represent to myself a property which (as a characteristic [als Merkmal]) can be found in something, or can be combined with other representations; that is, only by means of a presupposed [vorausgedachten] possible synthetic unity can I represent to myself the analytic unity. . . . Consequently [a representation which is to be thought as common to different representations] must previously be thought [vorher gedacht] in synthetic unity with other (though, it may be, only possible) representations, before I can think in it the analytic unity of consciousness, which makes it a conceptus communis. (B133n)

In other words, the employment of concepts as Merkmale (or as representative of Merkmale) cannot by itself constitute, because it is in its own turn constituted by, the consciousness of connections among possible intuitions. Presumably, it is constituted by a consciousness of connections between a given candidate for predication in the first place, and a manifold of other possible candidates. This in turn requires that one be “given,” internally to intuitions themselves, a manifold of anticipations or retentions of other possible ones.¹⁵ (The specifically anticipative dimension of that “constitutive” consciousness might provide a harmless explanation of Kant’s claim, at B134, that the synthetic unity of the manifold is given a priori [a priori gegeben: contrast Kemp Smith’s “generated a priori”]. The point need simply be that it is anticipatively “given.”)

That the unity originally constituted by intellectual functions is not in the first instance a unity of Merkmale is suggested by Kant’s choice of comparison in the case. He refers to his discussion of “qualitative” unity in §12. In every instance of cognition, Kant had said in that section, “there is unity of concept, which may be entitled qualitative unity, so far as we think by it only the unity in the combination of the manifold of our cognitions; as, for example, the unity of the theme in a play, a speech, or a story” (B114). By contrast to the mere unification of Merkmale in a concept, in a subject of predication, or even in a merely represented subject of predication,¹⁶ Kant’s suggestion appears to involve the unity that a set of items enjoys (the components of a play, a speech, or a story) by virtue of their all representing—indeed, their all being made to represent—a common subject matter. That is very different from their all representing (not to mention their being, or even their possibly being) the predicates of a common subject. As we shall see more clearly later, it is also more like the sort of unification that needs to be “imposed” on anticipations and retentions, in order that they be truly embodied in acts of conception. Those anticipations and retentions need to be “unified” in two respects. They of course need to be unified by virtue of being one and all ingredient in a particular intuition. But they also need to be unified by virtue of being (taken to be) anticipations and retentions regarding a common subject, represented through an intuition.

The other examples of synthesis Kant offers make it difficult to construe that notion in terms of the unification of Merkmale. For example, he offers the case in which, in thinking of some line, one “draws it in thought” (B154; cf. B162 [“I draw as it were the outline of the house”], A102, A162-3/B203). The parts of a line that one “synthesizes,” when drawing it in thought, are surely very different from the Merkmale of a line. Of course, it may be difficult to see why Kant should say that drawing a line in thought is a necessary condition for thinking it in the first place. One might suggest that, in saying it, Kant must indeed have something in mind like the unification of Merkmale, and simply puts it badly. But I have, in effect, already suggested an alternative, in commenting on Kant’s more general claim concerning the need for synthesis in the apprehension of space as an “object” (B160n). It is simply that the capacity for conceptualizing figures as lines (which we may take to be, in Kant’s view, at least a necessary condition for “thinking of’ lines in the first place), presupposes the capacity for giving special expression¹⁷ to a set of anticipations regarding the possibility of drawing lines. Specifically, it presupposes the capacity for giving “expression” to certain anticipations, by means of the latter’s very ingredience in an act of (at least) imaginative intuition. This notion of line-drawing anticipation, not simply as expressible in conceptual acts, but as embodied in intuition, was no doubt a difficult notion for Kant to convey. It is not implausible to suppose that he sometimes overstated it. The point is not that we need to draw a line, not even in thought, in order to have it in mind. But there does need to be much more “in mind”—having to do with the act of drawing—than the mere ability to “think” in some way or other.

III. Judgments of Perception and Experience

We have yet to consider the overall structure of the second-edition Deduction. In the next section, I argue that it rests on the distinction, drawn in §§18—20 of the Prolegomena, between judgments of perception and of experience. In §19 of the Deduction, Kant appears to imply that he had been mistaken in drawing that distinction, at least in terms of a distinction between two kinds of “judgments.” My own view is that, while in strict Kantian terms, they are not judgments at all, Kant had an important point in treating judgments of perception as judgments. The relevant points in the Prolegomena account are in any case four.

First, a judgment of perception involves nothing more than some kind of “connection” of representations in a perceiver’s perceptual state:

Empirical judgments, so far as they have objective validity, are judgments of experience, but those which are only subjectively valid I name mere judgments of perception. The latter require no pure concept of the understanding, but only the logical connection of perception in a thinking subject. (P. 298) [A judgment of this sort] only expresses a relation of two sensations to the same subject, that is, myself, and that only in my present state of perception [nur in meinem diesmaligen Zustande der Wahrnehmung]. . . . [I] do nothing but refer two of my sensations to each other. (P. 299) The foundation is [Zum Grunde liegt] the intuition of which I become conscious, that is, perception (perceptio), which pertains merely to the senses. But in the next place, there is judging (which belongs only to the understanding). But this judging may be twofold: first, I may merely compare perceptions and connect them in a consciousness of my state [in einem Bewusstsein meines Zustandes]. . . . The first judgment is merely a connection of perceptions in my mental state [bloss Verknüpfung der Wahrnehmungen in meinem Gemütszustande], without reference to the object. (P. 300)¹⁸

Second, a judgment of experience can only be generated from a corresponding judgment of perception. The former is what the latter becomes, when it is appropriately “determined” by the logical forms of judgment:

All our judgments are at first [zuerst] mere judgments of perception; they hold good only for us (that is, for our subject), and we do not till afterward [nur hintennach] give them a new reference (to an object) and desire that they shall always hold good for us and in the same way for everybody else; for when a judgment agrees with an object, all judgments concerning the same object must likewise agree among themselves, and thus the objective validity of the judgment of experience signifies nothing else than its necessary universal validity. (P. 298)

According to Kant, however, there are also two kinds of judgments of perception. One of them cannot—not even through adding pure concepts (nicht solche Warhnehmungsurteile. . .die jemals Erfahrungsurteile werden könnten, wenn man auch einen Verstandesbegriff hinzu täte)—ever become a judgment of experience. The other, of course, can: Wahrnehmungsurteile, die durch hinzugesetzten Verstandesbegriff Erfahrungsurteile werden (p. 299n). The distinction relates to Kant’s distinction between primary and secondary qualities and need not detain us.¹⁹ The point remains that a judgment of experience always involves some operation in regard to a judgment of perception:

For instance, when I say the air is elastic, this judgment is as yet [zunächst] a judgment of perception only; I do nothing but refer two of my sensations to each other. But if I would have it called a judgment of experience, I require this connection to stand under a condition which makes it universally valid. (P. 299)

Before, therefore, a judgment of experience can come from a judgment of perception [ehe aus einem Wahrnehmungsurteil ein Urteil der Erfahrung werden kann], it is requisite that the perception should be subsumed under some such concept of the understanding. . . . The judgment that air is elastic becomes universally valid and thereby finally [dadurch allererst] a judgment of experience. . . . (Pp. 300-1)

Third, Kant describes the requisite “operation” as an operation on perceptions or intuitions themselves. Judgments of perception are themselves perceptions:²⁰

[A]ll empirical judgments are not judgments of experience; but, besides the empirical, and in general besides what is given to sensuous intuition, special concepts must be superadded—concepts which have their origin wholly a priori in the pure understanding, and under which every perception must be first of all subsumed and then by their means changed [verwandelt] into experience. (P. 298)

Quite another judgment therefore is required before perception can become experience [ehe aus Wahrnehmung Erfahrung werden kann]. The given intuition must be subsumed under a concept which determines the form of judging in general relatively to the intuition . . . , which does nothing but determine for an intuition the general way in which it [emphasis added] can serve for judgments [zu Urteilen dienen kann] . . .. [I]t is requisite that the perception be subsumed under some such concept of the understanding. (P. 300)

The fourth point bears on the particular way in which the forms of judgment are supposed to “determine” perceptions, and thus convert them into judgments of experience. They do this, Kant says, by determining and converting, not simply perceptions themselves, but the very connections among perceptions that constitute judgments of perception:

[W]hen through the concept of the understanding the connection of the representations, which it gives to our sensibility, is determined as universally valid, the object is determined through this relation, and the judgment is objective. . . . But if I would have it called a judgment of experience, I require this connection to stand under a condition which makes it universally valid. I desire therefore that I and everybody else should always connect necessarily the same perceptions under the same circumstances. (P. 299)

But this judging may be twofold: first, I may merely compare perceptions and connect them in a consciousness of my state; or, secondly, I may connect them in a consciousness in general [in einem Bewusstsein überhaupt] . . . [which] connects empirical consciousness of intuition in consciousness in general [in einem Bewusstsein überhaupt]. (P. 300).

Putting the points together, Kant appears to describe the relationship between judgments of perception and experience in terms of two different ways in which perceptions might be “connected,” in particular connected within a given perception.

There is a way, very different from my own, in which one might attempt to put these points together. One might suggest that the difference between judgments of perception and experience is that the former involve judgments about the subject itself, and how things merely seem or appear, while the latter involve judgments as to how things really are.²¹ According to this approach, we do not take Kant at his word, in his claim that judgments of experience need to be formed out of judgments of perception (not to mention out of perceptions). But perhaps we might do better with his claim that judgments of experience objectively “determine,” not simply perceptions, but subjective connections among perceptions, and indeed that they determine the very connections involved in judgments of perception. We might simply take the claim to be that a judgment of perception is a judgment about certain connections among perceptions—namely, a judgment as to how they actually connect in one’s experience—while a judgment of experience is a judgment about the way in which they necessarily connect in anyone’s experience.

Apart from slighting Kant’s own suggestion that the subjective connections thereby “determined” are those that actually constitute judgments of perception—as opposed to being connections that judgments of perception merely refer to—we would have to pay a price for this reading. For it would now follow that judgments of experience are not about objects. Instead, they would be about perceptions. Kant says a number of things that seem to imply just this. These are, of course, the things that suggest a phenomenalist reading of Kant. In Chapter Two, I tried to show how the kind of phenomenalism entailed by such pronouncements is in fact compatible with denying that judgments of experience are about perceptions. Putting that issue aside, there is an additional difficulty, in the proposed reading, concerning judgments of perception. The reading is incompatible with Kant’s insistence that judgments of perception do not involve “forms of judgment.” As I shall propose, it is possible to see why Kant should regard judgments of perception as “judgments,” even if they do not involve the forms of judgment. Obviously, they could not be judgments in the official Kantian sense. But it is hardly plausible to suppose that Kant forgot this. On the reading so far suggested, it is in any case impossible to see why Kant should insist that judgments of perception do not involve categories.

A second proposal avoids this final problem. According to it, judgments of perception are merely associations of perceptions—as opposed to judgments about associations of perceptions.²² There is ample reason to suppose that the “connections” involved in judgments of perception are indeed associations:

But if I would have it called a judgment of experience, I require this connection [emphasis added] to stand under a condition which makes it universally valid. I will [will] therefore that I and everybody else should always connect necessarily the same perceptions under the same circumstances. (P. 299)

“This connection” can only be a certain pattern of associations. Were it not, then it would be difficult to see how one could ever demand, in the conversion to judgments of experience, that this very same connection be universalized. What would there be, in that case, that might happen to obtain peculiarly in me, or in anyone else in particular, but that also might be subjected to the demand that it obtain universally? In addition, having already excluded the possibility that the relevant sort of “connection” of perceptions is only a judgment about a connection of perceptions, how could the reference to “circumstances” be relevant? We could only be talking, it seems, about patterns of association, insofar as they are, as they always are, relative to circumstances.

Consider also the following:

If all our synthetical judgments are analyzed so far as they are objectively valid, it will be found that they never consist of mere intuitions connected only (as is commonly believed) by comparison into a judgment; but that they would be impossible were not a pure concept of the understanding superadded to the concepts abstracted from intuition, under which concept these latter are subsumed. (P. 301)

Admittedly, one may wonder how, prior to subsumption under pure concepts, any sorts of concepts are “abstractable” in the first place. But then, that would be no more problematic than Kant’s suggestion that judgments themselves are possible in such a case. Contrary to Kant’s official terminology, we might simply need to distinguish (as he does) between different notions of a “concept’’ and a “judgment.” Terminology aside, there is no reason for that to pose a problem for Kant. What we should note is rather the following.

Kant says that it is commonly believed that judgments of experience “consist of mere intuitions connected only. . .by comparison into a judgment.” That is, it is commonly believed that objective judgments are reducible to judgments of perception. But who could Kant really be thinking might believe such a thing, not to mention embrace it as an item of common opinion? To whom could it possibly seem evident that judgments about objects are no more than subjective “connections” among perceptions (as opposed to being judgments about such connections—which in any case surely cannot be supposed a matter of common opinion)? Kant’s terminology suggests Hume: that the “connection” in question involves “comparison” is mentioned three times in the course of §20. It is in fact not unreasonable to take Hume to hold, not that judgments about objects are really judgments about perceptions, or even about associations of perceptions, but rather that such judgments are themselves associations. Habits of association might naturally be presumed to arise from a kind of “comparison” (of cases repeated in experience). In addition, Kant consistently describes Hume as having substituted mere “habit” for judgment; he does not simply describe him as erring with regard to the objects of judgment (p. 258).²³ So it does seem reasonable to conclude that judgments of perception are mere associations of perceptions.

But some problems still remain. First, the suggestion that judgments of perception are mere associations of perceptions does not explain why Kant should call judgments of perception a kind of “judgment.” Second, it does not explain how judgments of perception could ever become judgments of experience, nor why all judgments of experience must “first” be judgments of perception. Third, it does not explain how the logical functions that convert the one to the other can be functions exercised with respect to given perceptions: associations of perceptions seem to be one thing, perceptions themselves another.

I think that we can avoid these problems. The proposal is simply this: that judgments of perception are associations of perceptions—but only so far as such associations are ingredient in given perceptions, as a kind of material for the apprehension of appearances. Such ingredience does not really amount to “judgment,” nor to the employment of empirical concepts, nor to the apprehension of “objects,” in the strict usages that Kant tends to favor. But it is understandable why Kant nevertheless suggests that such ingredience of associations amounts to a kind of judgment. In a sense, it does indeed amount to a kind of “predication” with respect to appearances. For insofar as appearances are apprehended through associations, and the latter are not merely externally attached to (or “associated” with) the intuitions through which appearances are apprehended, the correlative appearances must be apprehended as correlatively characterized. This is what I argued in the last chapter.

The point holds, we should note, even if we refuse to admit that, apart from conceptualization, appearances are ever “apprehended” to begin with. Even with that restriction, it remains the case that it is a necessary condition for any such apprehension that the apprehension be formed from material ingredient in intuitions. Once again, the way the correlative object is “characterized,” through such an act, is not solely a function of something intellectual. It is equally a function of the purely associative material on which the former operates.

Apart from its bearing on the Deduction, one upshot of this approach is that it allows us to confirm at least part of a suggestion made by Beck in regard to the distinction in question. It concerns the relation between the distinction and Kant’s theory of “judgments of taste.” In this respect, Beck suggests, “the Critique of Judgment seems to have grown out of the doctrines of the Prolegomena which were rejected in the second edition of the Critique of Pure Reason”²⁴ Apart from the question of the Prolegomena’s relationship to the Critique, my account suggests a rather different ground of comparison. According to Beck’s proposal, judgments of taste may be regarded as standing to judgments of “agreeableness” in a relation analogous to that in which judgments of experience stand to those of perception. In both cases, the second term of the relation is something purely subjective. Because of that, something needs to be “superadded” to it, in order to yield an instance of objective judgment. In the case of judgments of perception, what needs to be added is subsumption under the categories. What gets thereby subsumed, Beck suggests, is a merely “intuitive image” (and so an item that was really not any kind of “judgment” to begin with). Analogously, what is added to a judgment of agreeableness, to make it a judgment of taste, is subsumption “under the understanding as the faculty of concepts in general.” (What gets thereby subsumed, apparently, is also a “faculty,” namely, that of imagination.)²⁵

Beck does not make it clear how what Kant describes as the “harmony” of imagination and understanding, essential to a judgment of taste, can be comparable to a case of subsumption. Certainly, it is difficult to see how “subsumption” under a faculty can be significantly comparable to the subsumption of something under concepts or predicates. In addition, it is not, of course, the case that a judgment of agreeableness could ever be transformed into a judgment of taste for Kant. (But strictly speaking, Beck presumably takes this lack of analogy to apply as well to the Prolegomena distinction.) In any event, Kant himself says that in a judgment of taste it is a non-conceptualized pleasure in some relation between the faculties that is supposed to function in a way analogous to the functioning of a predicate in judgment, but without the need for actual conceptualization. This provided the basis, in Chapter Four, for a comparison between judgments of taste and a perception of “affinity” in nature. As we by now have seen, it also provides a basis for comparison between judgments of taste and judgments of “perception.” (Obviously, the analogy is compatible with an additional disanalogy, to which Beck also calls our attention: judgments of perception are not “objective,” while judgments of taste are. The difference simply reflects the fact that it is purely contingent factors that determine what human beings can associate with what. In Kant’s view, at least, it is not purely contingent factors that determine that human beings take pleasure in the harmonious workings of their faculties.)

IV. The Structure of the Deduction

I have dwelt on the Prolegomena distinction for the following two reasons. First, I disagree with those who hold that §19— culmination of the Deduction’s first stage—rejects that distinction. Second, independent evidence, from throughout the Deduction, supports the view that its structure turns on a shift in perspective (from the noetic to the noematic), the point of which could only be seen if judgments of experience are nothing but judgments of perception, as I have interpreted them, transformed by means of the forms of judgment into judgments of experience. As I shall argue, the Deduction turns from a conclusion regarding unification within intuitions, required for their transformation into judgments of experience, to a conclusion, in its second stage, regarding a correlative structure among the objects of intuitions.

The problem of the Deduction’s stages was made the focus of particular recent attention by Henrich. Henrich maintains that the key lies in the fact that §20 terminates the Deduction’s first stage with a restricted conclusion: “All the manifold, therefore, so far as it is given in a single [sofern es in Einer] empirical intuition, is determined in respect of one of the logical functions of judgment, and is thereby brought to a consciousness in general [zu einem Bewusstsein überhaupt]” (B143). Thus the first stage infers the need for categories as a condition for a sub-class of logically possible representations, namely, of appropriately “unified’’ ones. Culminating in §26, the second then extends the conclusion to sensory intuition generally, and to its objects. It does this on the basis of the fact that sensory intuition is representation in space and time, and, as the Aesthetic has already argued, space and time necessarily involve a “unity” of representation. According to Henrich, in other words, the ultimate aim of the Deduction is to exclude “the possibility of a merely partial ability of the understanding to establish unity in the sensible representations.” It does so, he suggests, on the ground that “we do in fact have unitary representations of space and time and therefore can also unify all representations of sense.”²⁶

The main objections to Henrich appear to be three. I am in agreement at least with the first; the others conceal an ambiguity to which both Henrich and opponents may be victim. The first objection is that the text fails to support the claim that, by the end of §20, Kant had not yet concluded that the categories apply to sensible intuition generally.²⁷

We need to be careful in formulating the second objection. What is sound in it is this: that the mere concession of whatever unity was available from the Aesthetic does not in fact suffice for the purposes of the Deduction.²⁸ The objection may give the impression that, according to Henrich, it is sensibility that provides the unity to which the Deduction then appeals. But to the contrary, his own claim is that the unity of space and time, laid down as necessary in the Aesthetic, is a product of understanding.²⁹

If the unity in question is a product of understanding, then the question of course remains as to how the Aesthetic, confined to sensibility, could have been entitled to introduce it. But this is a question that Henrich himself regards as legitimate. The apparent difficulty is merely indicative of the “synthetic” method employed by Kant in the Critique.³⁰ Alternatively, I would propose, we might simply have distinguished between two kinds of “unity” from the start. In any case, this point also relates to the third of the objections against Henrich. The objection is that it is plainly unintelligible, for Kant, to suppose that intuitions might lack a unity of their manifold. But if it is unintelligible, then so is Henrich’s suggestion concerning the distinction of stages.³¹

On the account that I have proposed, the Aesthetic argues for a pre-intellectual “unity” in intuition. It does this by arguing that a preconceptual “form of intuition”—internal to (would-be) cognitive states themselves—is necessary for the employment of concepts in the apprehension of any possible instances of concepts. (As I have argued, it is, strictly speaking, compatible with this to deny that the form in question is sufficient for such apprehension, at least as a mode of fullblown “consciousness.”) This is what we examined in Chapter One. Now it might appear that Kant rejects this position in the Deduction. For he there claims that the very “unity of space and time” is a product of understanding. But I have argued that this claim, as intended in the Deduction, does not in fact contradict my reading of the Aesthetic.³² What concerns us now must in any case be a notion of “unity” that had not been in question in the Aesthetic at all.

In a way, my suggestion concerning the Deduction’s stages will be similar to some others. Certainly, a number of readings take it that the Deduction’s first stage regards the categories as necessary for a certain sort of unity, while its second concludes that the categories are necessary for a “correlative” unity involving sensible objects. So far as I can tell, however, the view that Kant’s inference is from one kind of unity to a correlative kind of unity generally involves supposing that the first requires a purely logical or judgmental kind of unity. Validly or not, the second stage then demands application of this kind of unity to the manifold of intuition.³³ In a way, I do not disagree. But if the unity in question in the first stage is supposed to be purely “logical,” then it cannot be so in any ordinary sense, certainly not in the sense that is generally assumed. In particular, it must be a unity formed out of a manifold of sensible material. Insofar as the material, though “sensible,” is distinct from mere sensation, we might perhaps still say that the corresponding unity is purely “logical,” but only in a misleading sense.

In the following passage Kant states what has been done, in the Deduction’s first stage, and what remains to be done in order to complete the Deduction:

Thus in the above proposition a beginning is made of a deduction of the pure concepts of understanding, in which, since the categories have their source in the understanding alone, independently of sensibility, I must still abstract from the mode in which the manifold for an empirical intuition is given, and must direct attention solely to the unity which, in terms of the category, by means of the understanding, enters into the intuition. In what follows (cf. §26) it will be shown, from the mode in which the empirical intuition is given in sensibility, that its unity is no other than that which the category (according to §20) prescribes to the manifold of a given intuition in general. Only thus, by demonstration of the a priori validity of the categories in respect of all objects of our senses, will the purpose of the deduction be fully attained. (B144-5)

The contrast between what the categories prescribe “to the manifold of a given intuition in general” and what they prescribe with respect to “the mode in which the empirical intuition is given in sensibility,” may suggest that the operative distinction is between a general case and its specific application: for example, from sensible intuition in general to distinctively human (spatiotemporal) inituition.³⁴ (Kant himself points out, in the sentence that follows this passage, that the first stage had at least been concerned with sensible intuition specifically.) But there is ample evidence that the first stage was, if not already explicitly directed toward spatiotemporal intuition, at least regarded by Kant as merely trivially applicable to the latter.

We have already examined a passage, from the Deduction’s first stage, that specifically applies Kant’s reasoning to the case of the manifold of spatiotemporal intuition. The passage was from a footnote to §17. It ended with an anticipatory reference to the second stage as well (actually to §25, but presumably intending §26). Kemp Smith translates the reference in the following way: “The singularity of such [spatiotemporal] intuitions is found to have important consequences (vide§25)” (B136n). I deliberately translated that sentence in a much more awkward way. Kant himself says that the singularity in question is wichtig in der Anwendung (“important in [the] application”). I shall later suggest how the phrase may be read in support of my approach. The question remains as to exactly what can be merely anticipatory about the reference to spatiotemporal manifolds, at this point, and about the question of “application.”

In at least one respect, I agree that the point of the second stage can be put, as Kant puts it, in terms of a problem concerning the “manifold given in spatiotemporal intuition.” But this is only if we mean by the latter: the manifold of portions of space and time themselves, together with the manifold of possible spatiotemporal “appearances,” insofar as these may be regarded as genuine “objects” of awareness. It is with one or with both of these that the footnote in question is concerned. There, for example, Kant specifically discusses space and time as “objects.” In addition, in the section to which the note is attached (§17), he is concerned with a distinction between space as an “object” and space—or something that he calls space—taken in some other way altogether. Whatever exactly he means by the latter, he puts it by referring to space as a “form.” I have already suggested what this distinction must involve. It is a distinction between space itself, and the spatial structure of objects appearing in it, and the intuitional form internal to representations of space. What Kant calls “space,” taken as a form, is not space at all in the present context. It is simply the form of spatial intuition as such. A closer look confirms this:

Thus the mere form of outer sensible intuition, space₁, is not yet a cognition [noch gar keine Erkenntnis]; it supplies only the manifold of intuition a priori for a possible cognition. To know anything in space₂ (for instance, a line), I must draw it, and thus synthetically bring into being a determinate combination of the given manifold, so that the unity of this act is at the same time the unity of consciousness (in the concept of a line); and it is through this unity of consciousness that an object (a determinate space₂) is first cognized. (B137-8; subscripts added)

Kant makes a point of saying that “space,” as a form, is not yet a cognition. This alone makes it clear that he is using the term space (indicated by means of the first subscript) in a technical way. In any ordinary sense, it would be absurd to suppose that space ever is or could become a cognition. As I did in an earlier discussion of a related passage, I therefore conclude that Kant means by “the mere form of outer sensible intuition, space,” not space at all, or even spatial form as such, but the irreducibly intuitional form of our representations, or would-be representations, of space (or of spatial form as such).

It might seem more natural to conclude that Kant simply means that the mere form of intuition is not in itself a proper object of cognition. But that would be mistaken. In Kant’s own view, space and spatial form are as proper an object of cognition as anything else. To be sure, insofar as space and spatial form have not yet been conceptualized, they are not full-blown objects of cognition. But that is not a point that concerns the “form” of intuition specifically. It applies as well to “appearances,” and to anything that we might regard as mere “matter” in intuition. In any case, the question would still remain: what could be the point of saying that, while space is not yet an object of cognition, it provides a manifold that needs to be synthesized, in order for it actually to become an object of cognition? In what sense could space, insofar as it is not an object of cognition, provide or contain any sort of manifold at all? The suggestion threatens to return us to some of the more objectionable readings already rejected.

A more plausible reading will simply assume that Kant is intending to distinguish, in §17, between two very different sorts of manifold and two different sorts of unifying structure. He is distinguishing, on the one hand, between a unifying structure that needs to be “applied” to a manifold internal to spatial representations and a structure that needs to be regarded as (thereby) applying to manifolds in space as an object. The latter must depend on the former. A necessary condition for apprehending space, or objects in space— as objects—must be the establishment of some kind of unity within cognitive states themselves. In the section in question, Kant emphasizes the latter dimension several times. First, in the passage that I have quoted, he directly refers to the unity of the “act” of cognition (which he also calls unity of “consciousness”). Second, he equates the unity of the act of cognition with a unity within one’s very concepts of space or of objects—not with a unity internal to space or to objects themselves. In both cases, this sort of unity is a necessary condition for a truly cognitive apprehension of space as an object. A similar distinction, and the same conclusion, is drawn just before the passage that we have been examining:

Understanding is, to use general terms, the faculty of cognitions. These consist in the determinate relation of given representations to an object. An object, however, is that in the concept of which the manifold of a given intuition is united. . . . Consequently it is the unity of consciousness that alone constitutes [ausmacht] the relation of representations to an object. . . . (B137)

This reading would make it clear what is and what is not merely anticipatory about the footnote to §17. What is anticipatory is its reference to a structure internal to space itself (and time) as object of cognition. What is not anticipatory is its reference to spatiotemporal “intuition” as such. (To be sure, in most sections of the Deduction, Kant speaks of a manifold of sensible intuition in general, and does not consider spatiotemporal intuition at all. But this, I take it, only indicates that he regards the more specific application as following merely trivially from the general proposition. A triviality of this kind could not be what defines the distinction of stages.) In addition, this reading would make more literal sense of the conclusion to the footnote in question. Kant’s point would then be, not simply that the “singularity” of space and time, as objects of cognition, will have an important “consequence” when we eventually reach §26. The point would be, more specifically, that the consequence itself turns on the question of the application (Anwendung) of cognitive structures to objects.³⁵

As noted, something like this has been suggested, in at least a general way, by a number of commentators. But there is a difference. Typically, when the distinction between the Deduction’s stages is taken to rest on a distinction between structures internal to cognition and structures involving objects, the first stage is taken to be merely concerned with structures internal to thoughts about objects, or to mere concepts or judgments. In a way, this restriction does of course characterize the Deduction’s first stage. But what is crucial is to recognize the fact while also acknowledging another. What we need to see is that the first stage, with its demand for “unity of consciousness” in thoughts and concepts, already presupposes a doctrine regarding the unification of sensible manifolds. In particular, it presupposes a doctrine regarding the unification of manifolds within spatiotemporal intuitions themselves.

One way to combine the points might be thought to lie in a suggestion rejected earlier.³⁶ We might suppose that the requisite unity, internal to any concept that is truly applicable to objects, is a unity established by means of the conversion of intuitively apprehended features or “impressions” into Merkmale, and thereby into concepts. In my view, the requisite unity is of course established by means of the unification of a very different sort of material, namely, imaginative associations. Apart from the intrinsic intelligibility of it, at least one advantage of that alternative should be obvious. It will allow us to see the point of Kant’s own treatment of judgments of perception and experience. That distinction, as we shall see, is the key to Kant’s argument in §§18-19 of the Deduction. An examination of these sections reveals that the Deduction’s first stage rests precisely on the demand for conversion of the former sort of “judgment” into the latter. But in that case, only one task could possibly remain for the Deduction’s second stage. The second stage must proceed to conclude that, whatever structures are essentially involved in that conversion must in turn be mirrored in the objects of experience themselves.³⁷

Now let us return to the beginning of the Deduction. What we need to note, first of all, is that the section (§15) begins squarely from the noetic perspective. That is, it begins from the perspective of the structure of cognitive states or acts, as opposed to that of objects or appearances (or even of potential objects or appearances). The noetic perspective is evident from Kant’s characterization of the “manifold” that stands in original need of unification through understanding.

He describes it, namely, as a manifold that concerns nothing more “than the mode in which the subject is affected”:

The manifold of representations can be given in an intuition [in einer Anschauung] which is purely sensible, that is, nothing but receptivity; and the form of this intuition can lie a priori in our faculty of representation, without being anything more than the mode in which the subject is affected. But the combination [Verbindung] (conjunctio) of any manifold [eines Mannigfaltigen überhaupt] can never come to us through the senses, and cannot, therefore, be already contained in the pure form of sensible intuition. . . . For it is an act of spontaneity . . . understanding. (B129)

The noetic perspective might also be evident from the fact that the manifold in question is said to be contained within a single intuition (in einer Anschauung). Although the latter phrase is ambiguous, we have at least seen that the Deduction as a whole is in fact dominated by the notion of the unification of manifolds within single intuitions. And we have seen that the notion is specifically operative in §17.

I have of course already argued that we can make sense of the demand for unification within single “intuitions,” when the latter are regarded as instances of cognitive or potentially cognitive states or acts. It is much more difficult to make sense of the demand, when it is construed as bearing on a manifold supposed to be ingredient in appearances themselves, or in intuitable regions of space. That is why, in the latter case, one is more apt to emphasize—despite Kant’s suggestions to the contrary—the need for some kind of unification among distinct intuitions. In any case, Kant himself does not explicitly characterize the items comprising the manifold. Eventually, through an examination of §§18-19, I shall provide additional reason for holding that it can consist only of imaginative anticipations and retentions. For now, we must be content with the fact that Kant simply begins, in §15, with the claim that the manifold in intuition needs to be unified intellectually.

Why does the manifold in intuition need to be unified intellectually? The following would seem to be Kant’s only reason for saying so:

To this act the general title ‘synthesis’ may be assigned, in order to indicate at the same time [um dadurch zugleich] that we cannot represent to ourselves anything as combined in the object which we have not ourselves previously combined, and that of all representations combination is the only one which cannot be given through objects. (B130)

Kant may appear merely to be saying that a necessary condition for the awareness (representation) of relations among objects is that the understanding somehow combines those objects, or at least their appearances, qua objects of representation. But depending on what we had already packed into the notion of “awareness,” that might be trivially true. As it stands, it seems in fact merely to amount to the claim that any truly conceptual awareness of relations among objects is dependent upon conceptualizations of those objects. A much more significant claim would be the one that I have proposed: that a necessary condition for the awareness of relations among objects needs to lie in the effecting of combination within the very cognitive acts or states through which they are to become objects of representation in the first place. It is precisely this, I suggest, that is indicated by Kant’s use of the phrase um dadurch zugleich in the passage. (Kemp Smith simply has Kant say that we indicate by the title of synthesis that we cannot represent anything as combined in the object “which we have not ourselves previously combined.” What Kant himself says, again, is that the title in question is meant at the same time to indicate this truth.) The formulation seems to imply a comparsion. If so, it is presumably a comparison between synthesis, as a combination of something “in the object” itself, and some other, correlative kind of synthesis. On my own reading of the Deduction’s stages, it would then be the reference to the former that is merely anticipatory in the context of §15.

Now if we do not in fact take the “synthesis” in question to involve, in the first instance, some kind of combining internal to objects, or appearances, themselves, then it may appear that the real beginning of the Deduction is postponed to §16. And this, of course, is where a problem concerning “ascription” first arises. It arises in the claim that “it must be possible for the ‘I think’ to accompany all my representations” (B131). Here, at least, our focus seems more univocally to be on cognitive states or acts in their own right, or at least on would-be cognitive states or acts, as opposed to their objects or would-be objects. But what exactly is the problem? At the very least, it may seem clear that it is not what I have suggested it is. For it appears to be a problem that concerns some kind of a synthesis of a manifold of distinct intuitions, not a synthesis involving manifolds ingredient in single intuitions. In particular, it appears to be a problem concerning the conditions under which a manifold of distinct intuitions are (representable as) one and all “ascribable” to a single subject, namely, oneself. I do not, of course, propose to deny the centrality of this problem for the Deduction. But—though it may seem implausible to say it—we need to continue to postpone (until the next chapter) our detailed examination of that problem.

What is crucial to see, long before we turn to the problem of “ascription” for its own sake, is that a problem concerning the ascription of representations, in regard to a potential manifold of distinct representations, arises directly out of the demand for a certain kind of unification within single intuitions. Suppose that the manifold for the latter is in fact a manifold of anticipations and retentions. Even a non-human animal, we have granted, is able to “anticipate” the approach of possible appearances, and to “retain” other ones. But Kant’s point, in the present context, might simply be taken to be this: that while it is necessary for the formation of a truly cognitive act or state that appearances be somehow anticipated and retained, it is not sufficient that they be anticipated and retained on a purely animal level. Instead, they need to be appropriately “combined” or connected in the very (and what would otherwise remain the mere) process of anticipating and retaining them. In particular, the appearances in question need to be combined or connected in at least the following way: they need to be anticipated and retained precisely as appearances of which one may oneself eventually come to obtain intuitions (or—to emphasize “retention”³⁸—as appearances of which one might already have obtained intuitions). It is this notion that we need to examine in detail in the next chapter.

Whatever else in cognition may require “(self-)ascription,” with regard to a manifold of distinct intuitions ascribable to single subjects, the conceptualization of intuitions calls for a structure of self-ascription within any single intuition. For, as we have already seen at length, it is only anticipations and retentions within a given intuition that could be relevant to that intuition’s status as subject to conceptualization in the first place. Thus the present account can indeed explain the centrality of the problem of self-ascription, without abandoning the suggestion— in evidence throughout the Deduction—that the central problem concerns the Verbindung of manifolds internal to any given intuition. In other words, it is precisely the suggestion most prominent in the first-edition Deduction—that the manifold in question is a manifold of anticipations and retentions—that in fact permits the connection between these two undeniably central notions in the Deduction. But then, why was Kant so vague? If the relevant manifold, within single intuitions, is indeed a manifold as down-to-earth as that of ordinary “anticipations and retentions,” why not just have said so?

The reasons for Kant’s vagueness may be three. The first is something that I already indicated in Chapter Three. We may suppose that Kant was himself unclear regarding the very status of anticipations and retentions as “representations.” If so, then it would have been natural for him to have tried to avoid the issue altogether, and to have focused instead, as we have seen he does, on questions of “synthesis” regarding objects or appearances directly. Given the correlation between noetic and noematic perspectives, there is indeed, in a way, nothing wrong with doing just this. The second reason for Kant’s vagueness may be that he feared that the Deduction would otherwise appear objectionably psychologistic. For example, Kant may have found it difficult to explain exactly how cognitions could be “formed” out of something as subjective as anticipations and retentions— without reducing them to a variety of mental “particular,” namely, to particular bundles of associations in individual subjects. But it should be clear to us that the danger is avoidable. Finally, the third reason for Kant’s vagueness may be that he wanted to state his argument in a way that was applicable to sensible intuition generally (B145), and from there let it (trivially) apply to spatiotemporal intuition in particular. (This, of course, is not to offer that trivial application as the key to the two stages of the Deduction.) In fact, Kant does seems to say, although some consider it carelessness, that the combination of a sensible manifold, whether pure or merely empirical, is indeed an instance of a more general kind of Verbindung, applicable to manifolds that might not even be sensible at all (B130)³⁹ In that case, Kant could not appeal to “anticipation and retention,” in anything like the sense in which we are familiar with these things. In any case, our own examination of §§18-19 will eventually provide independent support for the claim that Kant regards (non-categorial) concepts as one and all, in a being whose intuition is at least temporal in form, “formed” out of manifolds of that very being’s anticipations and retentions.

The question of Kant’s actual argument remains. Where does Kant argue—or at least what is the argument—that conceptualization of an intuition requires the combination of a manifold in that intuition? If §16 does not contain the argument—but rather, as I have suggested, merely an entailment of its conclusion—then §15 ought to have contained the argument directly. As we shall see more clearly in our examination of §19, the answer is simply that, only by means of actually forming any concept, can the understanding be said to “employ” any concept in the first place. But §15, I suggest, had in its own way already made the point. It had made it in its famous observation that “where the understanding has not previously combined, it cannot dissolve, since only as having been combined by the understanding can anything that allows of analysis be given to the faculty of representation” (B130).

What is crucial is that Kant here assumes that understanding can do nothing other than provide some kind of intellectual “form” for given material. The point is also repeated in the next section: Understanding “is nothing but [emphasis added] the faculty of combining a priori, and of bringing the manifold of given representations under the unity of apperception” (B135). The point, I think, is generally not taken sufficiently seriously. Taken seriously enough, the needed conclusion follows directly from it. For if the understanding can only provide form for given material, then the relevant material needs to be available, and actually ingredient, in every intuition to which any concept is actually “applied.” Naturally, there may appear to be an alternative to the conclusion. The apparent alternative would be to suppose that the understanding functions by directly “combining” a manifold of distinct intuitions: a given one, subject to whatever conceptualization is in question, “together” with a manifold of others to which the given one is thereby regarded as appropriately related. But the argument, against that alternative, would presumably be this: that all that we could possibly mean, by the latter “combination,” is just the fact that a manifold of intuitions (or, correlatively, of intuitable appearances) have been subsumed under a particular concept or concepts. That would presume, impossibly, the legitimacy of appealing to the notion of “subsuming” under concepts, as an explanans, in the first place—when it is that very notion that the Deduction proposes to explicate in terms of the notion of “combination.”

Thus Kant’s central assumption, in the Deduction’s first stage— enunciated, at least implicitly, in §15—is that the “application” of concepts to objects (or appearances) can in its own turn be nothing other than an “application” of the very faculty of understanding itself For nothing else is given to the understanding to “apply” to objects—not even concepts! But this, taken strictly, entails that nothing is given for the understanding to “apply,” beyond its own intrinsic ability to combine some body of material into concepts in the first place. Obviously, this requires a notion of combination that cannot be borrowed from an already operative notion of applying, or subsuming under, concepts. In turn, this requires a notion of the combination of a manifold that is internal to intuitions themselves. For the only sense in which the understanding could “combine” distinct intuitions is precisely by conceptualizing given intuitions as combined with distinct ones, that is, by “applying” already constituted concepts to them.

As I argued in Chapter Three, this position was already implicit in Kant’s treatment of the nature of understanding as a “faculty,” in the sections prior to the Transcendental Deduction itself. Quite apart from what we have already seen in the first-edition Deduction, it is therefore not arbitrary to read the position “into” §15. The relative obscurity of the latter may be supposed simply to stem from Kant’s refusal to characterize the nature of the manifold that is supposed to be available to understanding for combination into concepts. But once again, what can in any case be seen in §15—and throughout the Deduction—is that the need for some such combination lies precisely in the fact that the latter is supposed, in the first instance, to be constitutive of any cognitive state or act as such, out of the constituents of such would-be acts or states, and not to be constitutive of objects or appearances out of constituents of the latter. Surely—in itself as well as in the light of the first edition—no other material is a plausible candidate for this sort of constitution, besides imaginative anticipations and retentions, so far as these are actually ingredient in intuition.

Unfortunately (it may seem), Kant is emphatic that mere “associations” have nothing to do with his project in the Deduction:

Insofar as imagination is spontaneity, I sometimes also entitle it the productive imagination, to distinguish it from the reproductive imagination, whose synthesis is entirely subject to empirical laws, namely, of association, and which therefore contributes nothing to the explanation of the possibility of cognition a priori. The reproductive synthesis falls within the domain, not of transcendental philosophy, but of psychology. (B152)

I have sometimes spoken of anticipations and retentions as “associations.” This passage may therefore appear incompatible with the proposed interpretation.

This is the section (§24), partly examined earlier, in which Kant draws a distinction between synthesis intellectualis and synthesis speciosa, or “figurative” synthesis. The former is purely intellectual, the latter involves a Wirkung of understanding on sensibility: “As figurative, it is distinguished from the intellectual synthesis, which is carried out by the understanding alone, without the aid of the imagination.” As I have read Kant’s distinction, the purely “intellectual” synthesis consists in whatever operations are required in order to form a conception out of a manifold of material potentially ingredient in a given intuition, and thus required in order to conceptualize any intuition at all. This is provided by the Kantian “forms of judgment,” supposedly already enumerated in the Metaphysical Deduction (§9), and considered in abstraction from their role as actually formative of conceptions. By contrast, I take it that synthesis speciosa is just those same forms—but now regarded as actually “applied” to a manifold of sensibility, that is, as actually formative of a conception out of a manifold of material in intuition. Kant also calls the “figurative” synthesis the “transcendental synthesis of imagination” (B151). It is imaginative, I propose, precisely because its material is essentially imaginative.

In this same passage, as we have now seen, Kant in turn equates “productive” imagination with the “transcendental synthesis” in question. From this it would follow that what I have called the mere material in transcendental synthesis, hence in “figurative” synthesis, could not be provided by the productive imagination in its own right. It would likewise seem to follow that, if there is indeed a distinction between imaginative material and intellectual form, internal to productive imagination itself, then the relevant material could be provided only by reproductive imagination. But then Kant plainly says, in the passage just quoted, that the synthesis of merely reproductive imagination “contributes nothing to the explanation of the possibility of cognition a priori.” This may seem to contradict my proposal. But it does not. The claim that productive or figurative synthesis does, but merely reproductive synthesis does not, contribute to cognition a priori is perfectly compatible with the view that reproductive synthesis (that is, anticipations and retentions) contributes to cognition precisely by virtue of being converted, by means of synthesis intellectualis, into instances of “figurative” synthesis.

We are finally ready to turn, then, to §§18-19, that is, to the sections that correspond to Kant’s discussion of the conversion of judgments of perception into judgments of experience in the Prolegomena. The first of these sections distinguishes between an “objective” and a merely “empirical,” or a merely subjectively valid, “unity of consciousness.” It begins with the claim that objective unity is that by which “all the manifold given in an intuition [in einer Anschauung] is united into a concept [in einen Begriff]” (B139).

Again, the notion of some kind of manifold, capable of ingredience in particular intuitions, but also susceptible of formation into concepts predicable of intuitions, is central to my own reading. But it is often underestimated in a reading of this passage.⁴⁰ Kemp Smith, for example, obscures the suggestion by translating “into a concept” as “in a concept.” Ignoring this distinction, as well as the reference to the containment of manifolds in single intuitions, one might suppose Kant’s point simply to be that any concept necessarily represents a manifold of possible intuitions in “connection” with a given one. That would, of course, formulate a Kantian claim. But Kant’s own formulation appears to suggest something deeper.

Kant does not in fact describe the empirical “unity of consciousness,” in §18, in exactly the same terms that he had used to describe judgments of perception in the Prolegomena. But this is only because he is not entirely consistent in §18. What I have argued, regarding the Prolegomena, is that “judgments of perception” are intuitions in which anticipations and retentions are ingredient in a special way. They are ingredient, that is, in such a way as to provide potential material for the “judgments of experience” into which they may eventually be converted. Now often, Kant refers to anticipations and retentions, in connection with given intuitions, as mere “associations” in regard to those intuitions. Conformably with this usage, Kant thus also says, in the present passage, that the empirical “unity of consciousness” is constituted “through association of representations” (B140). This, at least, conforms with the suggestion that the distinction Kant is attempting to draw, between empirical and objective consciousness, is indeed just the familiar distinction between judgments of perception and the objective judgments containing them as mere material. On the other hand, in the same passage, Kant also speaks of the empirical unity of consciousness in rather different terms. For he sometimes seems to regard it, not as the kind of “unity” that is involved in the mere anticipation and retention of possible intuitions, in connection with a given one, but as nothing distinct from the actual course of intuition itself: “Whether I can become empirically conscious of the manifold as simultaneous or successive depends on circumstances or empirical conditions” (B139).

Kant may simply be wavering between two different contrasts: between the kind of unity represented by means of objective judgments and a mere “unity” of representations in a single course of experience, on the one hand, and between the kind of unity represented by means of objective judgments and a mere unity of association, on the other. Or, tying the two distinctions a bit more closely together, we might simply suppose that Kant is contrasting the kind of unity represented by means of objective judgments with two very different kinds of “association.” In the one case, he might be contrasting it with the phenomenon of anticipation and retention as such; in the other case, he might be contrasting it with the phenomenon of actually “having in mind” what is anticipated and retained. This might simply reflect Kant’s own uncertainty as to the nature of the former in the first place. But whatever Kant’s uncertainty, he surely did not suppose that anyone ran the risk of confusing objective judgments themselves with the streaming of actual consciousness, or with instances of the mere togetherness of items in a stream of consciousness. Even Hume, after all, tried to make it clear that the crucial factor is not the fact of mere togetherness in a stream of consciousness, but that the items in question are together because “associated.” In any event, the phenomenon of anticipation and retention is clearly central to Kant’s concern in §18. This is evident, for example, from his appeal to the case of verbal association: “To one man, for instance, a certain word suggests one thing, to another some other thing” (B140).

A puzzle remains in the section. Kant says that the empirical unity of consciousness is “merely derived from the [objective unity] under given conditions in concreto” (B140). This is the claim that he illustrates by means of the example of verbal association. The claim may seem to pose a problem for any of the plausible alternatives: whether empirical unity is regarded as the actual course of consciousness, as the course of consciousness as it is merely anticipated and retained, or even, finally, as anticipations and retentions themselves. But we might explain Kant’s point in the following way. In themselves, anticipations and retentions might be merely externally (perhaps causally) connected with certain of a subject’s intuitional states (analogously to the connectibility of particular associations with particular words). On the other hand, those same anticipations and retentions might be actually connected in those states so as to constitute at least a “judgment of perception.” Now, in the latter case, there could be only one reason why anticipations and retentions should be so connected in intuition. The reason could only be that they might then provide material for eventual judgments of experience. The latter, we may presume, are what directly constitute the “objective unity of consciousness.” So it would indeed follow that the merely empirical unity of consciousness is “derived” from the objective unity “under given conditions in concreto. “The explanation simply parallels the one that I gave earlier of a corresponding claim concerning the “derivation” of empirical from transcendental “affinity.”

Now let us turn to §19. Here it is clearly “association,” again, that is supposed to provide the appropriate contrast with judgments of experience:

I find that a judgment is nothing but the manner in which given cognitions are brought to the objective unity of apperception. This is what is intended by the copula ‘is’. It is employed to distinguish the objective unity of given representations from the subjective. . . . Only in this way does there arise out of this relation [emphasis added] a judgment, that is, a relation which is objectively valid, and so can be adequately distinguished from a relation of the same representations [emphasis added] that would have only subjective validity—as when they are connected according to laws of association. In the latter case, all that I could say would be, ‘If I support a body, I feel an impression of weight’; I could not say, It, the body, is heavy’. Thus to say ‘The body is heavy’ is not merely to state that the two representations have always been conjoined in my perception, however often that perception be repeated; what we are asserting is they are combined in the object, no matter what the state of the subject may be. (B141-2)

It may be objected that Kant is confused in this passage, inasmuch as he attempts to illustrate pre-objective associations by reference to full-blown judgments, that is, by reference to judgments that clearly refer to objects (for example, to bodies), and even to oneself.⁴¹ But then no one could illustrate associations of the relevant sort without doing just that. What we should presume to be important, in Kant’s attempts, is just that what they are supposed to illustrate is neither a type of objective judgment (in the relevant sense) nor the mere phenomenon of the actual togetherness of representations in a stream of consciousness. What is in question is at most supposed to be the anticipation of representations. Naturally, such anticipations could not possibly be described apart from the employment of concepts that are, in the relevant sense, “objective.” But that does not imply that they are in themselves dependent upon concepts (in the relevant sense). Indeed, it is perfectly compatible with supposing that they are the mere material for the forming of any concepts in the first place.

Only the view that associations, of the intended minimal sort, are what provide the original material for the formation of concepts can make sense of Kant’s claims in this passage. For only that view can make sense of the claim that an objective judgment first arises “out of” the very relation that is constitutive of an instance of association. And only that view can make sense of the claim that an objective judgment involves an objective relationship among the very same representations that a purely associative “judgment” connects in a merely subjective manner:

Dadurch allein wird aus diesem Verhältnisse [emphasis added] ein Urteil, d.i. ein Verhältnis, das objektiv gültig ist, und sich von dem Verhältnisse eben derselben Vorstellungen [emphasis added], worin bloss subjektive Gültigkeit wäre.

One normally, of course, thinks of judgments as originally formed out of concepts, somehow placed in relation to one another. That in fact might be said to be Kant’s own “official” view. But the fact is that he begins the section precisely with the question as to what such a relation is supposed to “consist” of in the first place: worin dieses Verhältnis bestehe. Kant then proceeds to equate this question with a question as to the real force of the copula. This may suggest that he is merely inquiring as to what any judgment asserts, in virtue of the connection between the concepts that it contains. But the subsequent discussion is notably silent on this particular point. Kant does not in fact proceed to tell us what judgments have, in general, to say about objects, in particular about the objects that they are supposed to be about. Rather, he refers only to the function of judgments as somehow unificatory with respect to one’s representations of objects. At one point he also refers, not to mere representations, but to “cognitions”: “a judgment is nothing but the manner in which given cognitions are brought to the objective unity of apperception” (B141).

This last phrase may suggest that Kant is indeed intending to talk about the bringing of various concepts to a “unity of apperception.” But the subsequent discussion makes this reading impossible. For it becomes clear that Kant is at most talking about a bringing to the objective unity of consciousness of the very same representations that, in any particular instance, might well instead have constituted a purely subjective unity of consciousness. It could not be (objective) concepts that Kant is talking about here. Indeed, in the very heading to the section, Kant seems almost plainly to tell us that concepts need to be originally formed precisely by means of whatever structures are about to be introduced to the reader:

The logical form of all judgments consists in the objective unity of apperception of the concepts [emphasis added] which they contain.

It should be clear that the point does not require denying that the “forms of judgment” are constitutive of relations among concepts. To deny the latter would be to reject the Metaphysical Deduction. What I am arguing is simply that the Transcendental Deduction has a more fundamental aim, namely, to argue that, in every instance, the intellectual operations that are constitutive of relations among concepts must at the same time be formative of the very concepts thereby related. While the “forms of judgment” essentially constitute unities among concepts, this can only be a (necessary) upshot of their constituting concepts themselves, as actively functioning elements in cognition. (Attention to the notion of cognitive “activity” is crucial in this reading. There are obviously many ways the term concept might be used. When it is used, for example, to stand for something that one is merely supposed to be able to “possess,” and then on occasion to “use,” then concepts might turn out to be any number of things: perhaps even mere dispositions of certain sorts.) As Kant puts it elsewhere, “a judgment of experience is that (perception [emphasis added]) out of which a concept [emphasis added] of the object arises [entspringt].”⁴²

Although the following passage, from the Metaphysical First Principles of Natural Science, does not speak specifically of “concepts,” but rather of “cognitions,” it seems to confirm this reading:

[T]he categories, which are thought, are nothing but mere forms of judgments insofar as these forms are applied to intuitions (which with us are always sensible only), and that by such application our intuitions first of all obtain objects and become cognitions [emphasis added]. . . . [A]s to how experience is possible by means of these categories, and only by means of them. . . can be solved almost by a single conclusion from the precisely determined definition of a judgment in general (an act by which given representations first become cognitions [emphasis added] of an object.⁴³

The point, it should be clear, is not simply to apply cognitions or judgments to intuitions. In that case, the former might most naturally be regarded as originally formed through the combination of something called “concepts.” In turn, the judgments thereby formed would presumably at most refer to (or be in some way “about”) the intuitions in question. But Kant’s point is, of course, that the latter need to be turned into cognitions or judgments in their own right. This, I have argued, could only be done by converting some material within intuitions into the very concepts “available” for judgment about them in the first place. In any case, as Kant informs us in the continuation of the passage just quoted: he will “take the earliest opportunity” to make all of this clear to the reader. Presumably, he meant to do so in the second-edition Deduction. The improvement, he tells us, is only to concern “the manner of the presentation and not the ground of explanation.”

We might briefly notice the same point by returning to §17, in connection with a passage on which I have already commented:

To know anything in space (for instance, a line), I must draw it. . . so that the unity of this act is at the same time the unity of consciousness (as in the concept of a line); and it is through this unity of consciousness that an object (a determinate space) is first cognized. (B137-8)

I have already emphasized the fact that the unity of consciousness of which Kant is speaking is a unity supposed to be internal to the act of cognition itself. What we need now only notice, in addition, is that the unity in question is not described as a unity among concepts. It is rather said to be a unity within any concept as such, “as in the concept of a line.” The same, we saw, might be supposed to be Kant’s point in the notoriously difficult footnote to B160. An intellectual synthesis is there said to be required for the mere apprehension of space and time, or of spaces and times, as objects. But it is also a synthesis, Kant says, that “belongs to space and time” themselves, not to concepts. Quite to the contrary of the latter, it is a synthesis through which “concepts of space and time first become possible.” Such a synthesis can “belong to space and time,” I have argued, only to the extent that it is an operation required for the very forming, out of a manifold ingredient in spatiotemporal intuitions, of whatever concepts might be regarded as “applicable” to intuitions.

We are finally ready to return to the question of the Deduction’s two stages. So far we have seen some reason for saying that, apart from occasional glimpses ahead, the first stage adopts a purely noetic approach to cognition. Its emphasis, that is, is on the need for combinations of manifolds within (would-be) cognitions themselves. Accordingly, any question that concerns the relevant “application” of categorial structures must so far be limited to the question of their “application” to intuitions as opposed to intuitable objects. But the term intuition is of course ambiguous. Most notoriously, it is ambiguous as between representations and objects represented. For this reason, the heading of §20, and its text, may simply leave us unclear whether the manifold that is so far said to be “subject” to the categories is supposed to be merely a manifold within would-be cognitions, or rather (or perhaps in addition) within appearances as (would-be) objects:

All sensible intuitions are subject to the categories, as conditions under which alone their manifold can come together into one consciousness. . . . All the manifold, therefore, so far as it is given in a single [Einer] empirical intuition, is determined in respect of one of the logical functions of judgment, and is thereby brought to a consciousness in general [zu einem Bewusstsein uberhaupt]. Now the categories are just these functions of judgment, insofar as they are employed in determination of the manifold of a given intuition. (B143)

If the conclusion is in fact meant to be drawn with respect to objects, then it is at least surprising that Kant should speak so vaguely of “intuition.” Certainly, in any case, the emphasis on combination in (or even “into”) a subject’s consciousness suggests the noetic aspect. The same goes for Kant’s reference, in the section that follows, to the “manifold contained in an intuition which I call mine” (B144). And it of course goes for §16 as well, the section that introduces the problem of “the original synthetic unity of apperception” in the first place. What is particularly mine, or in principle “ascribable” to me, could hardly be intuitions qua objects. It could only be intuitions qua representations of objects.

The specific concern with the applicability of categories to objects enters with §22: “The category has no other use for the cognition of things than its application to objects of experience” (B146). But the point, in that context, is so far merely limitative, since it has not yet been explicitly stated that the categories necessarily apply to objects at all. The question of limitation is also the subject of §23. It concerns the restriction of categories to sensible intuition, and the purely negative character of any conception (at least by us) of a non-sensible faculty of intuition.

The first of these two sections had simply begun with the distinction between thinking and cognizing. For the latter, Kant says, we require intuition. It might seem that this is what provides the key to the Deduction’s stages. For it might be taken to suggest that the first stage concerns a purely “logical” notion of an object (mere “thinking”), while the second concerns, for the first time, the notion of a genuine object of cognition. But we have already seen that the first stage is in fact already concerned with the unification of manifolds within sensible intuitions, and even specifically spatiotemporal ones. So if the Deduction’s structure is indeed to turn on a distinction between what is merely thought and what is genuinely cognized, then the former must at least involve “objects” that are in the first place “thinkable” only through the very act of unifying manifolds in intuition.

It is perhaps possible, at least on one level, to read the Deduction as turning, in its second stage, from a concern with one kind of “object” to another. And if we do, then of course we need to suppose that the first stage had only dealt with objects in a certain kind of abstraction. But apart from the fact that this kind of distinction would be less fundamental than another (namely, than the distinction between the stages as concerned with noetic and with noematic “unification,” respectively), it would be misleading to put the point by saying that the first stage had dealt with objects only qua objects of “thought.” If we say it, we cannot mean to deny that it had dealt with objects precisely as apprehensible through the unification of manifolds within intuitional states. What we must mean is simply that the first stage had only been dealing with objects regarded in abstraction from their reality. If any “objects” at all could be said to be in question in the first stage, then they would not yet constitute a real world of nature. At most, they would merely be the intentional correlates of the very act of apprehending them. In these terms, the second stage might then be regarded as arguing that the categories are indeed applicable to the real world of nature. But in a very important sense, this would hardly mean that the first had been concerned with objects of a purely “logical” sort.

If we did distinguish the stages in this way, then it might seem that we would be committed to supposing that the first could only have legitimated the application of the mathematical categories to objects. For Kant tells us that only the dynamical categories concern actual existence (A160/B199; cf. A178/B220). But the conclusion does not follow. The first stage may not yet have concerned the applicability of categories to nature. But it may nonetheless have argued that any truly cognitive apprehension of a would-be nature presupposes the application, to the (would-be) cognitions in question, of all of the forms of judgment. Even the apprehension of mere illusions as illusions, for example, might arguably require the representation of at least a manifold of possible situations in which the apprehended “objects” might not have been illusory at all. (We need to remember that, by “cognition,” we are concerned with what may also be an instance, of miscognition.)

In any case, I do not propose to adopt this approach to the Deduction’s stages. It could perhaps be done. But this way of drawing the distinction, in terms of a distinction between kinds of objects, would at best be parasitical upon another. So far as it relies on the notion of noetic/noematic “correlation,” it would simply fail to emphasize a more fundamental inference that Kant first needs to draw. It is the latter to which I now turn.

After the discussion of synthesis intellectualis and synthesis speciosa, in §24, and of self-consciousness in the appendix to that section and in the section that follows, §26 finally informs us:

We have now to explain the possibility of knowing a priori, by means of categories, whatever objects may present themselves to our senses [Kant’s emphasis]. . . in respect of the laws of their combination, and so, as it were, of prescribing laws to nature, and even of making nature possible. (B159)

We have seen that it is tempting to locate Kant’s emphasis, in this passage, in a question regarding the objects that present themselves to our senses in particular. But by now we should see that a natural reading is the alternative that I have proposed. We have so far been concerned, Kant tells us, with the necessary structures in one’s apprehension of objects, insofar as apprehensions are cognitive (or at least would-be cognitive) states; we have now to be concerned with objects as such, that is, with the things that “present themselves” to apprehension. This distinction seems to be confirmed by what Kant had said in the sentence preceding the one just quoted:

In the metaphysical deduction the a priori origin of the categories has been proved through their complete agreement with the general logical functions of thought; in the transcendental deduction we have shown their possibility as a priori cognitions of objects of an intuition in general (cf. §§20, 21). We have now to explain. . . .

In other words, the first stage had demonstrated the necessity of application of the categories to cognitions of objects; the second must demonstrate—in a different but correlative sense of the term—the necessity of their “application” to objects.

It is the final paragraphs of §26 that then specifically formulate Kant’s points in terms, not simply of the application of categories to “intuitions,” or to cognitions or would-be cognitions of objects, but rather to appearances themselves, and “to nature, the sum of all appearances” (B163). The general question that then arises—as to how it is that objects should necessarily conform to anyone’s concepts at all—then provides the topic of §27. The concluding “brief outline,” finally, summarizes the whole:

It is the presentation of the pure concepts of the understanding, and therewith of all theoretical cognition a priori, as principles of the possibility of experience; of these, however, as the determination of appearances [emphasis added] in space and time in general [Stage Two],—and finally, this as following [emphasis added] from the principle of the original synthetic unity of apperception, as the form of the understanding [Stage One; emphasis added] in relation to space and time as original forms of sensibility. (B168-9)

In §21, Kant also describes the relationship between the stages. In the preceding section, he says:

. . . beginning is made of a deduction of the pure concepts of understanding; and in this deduction. . . I must abstract from the mode in which the manifold for an empirical intuition is given [emphasis added], and must direct attention solely to the unity which, in terms of the category, and by means of the understanding, enters into the intuition. In what follows (cf. §26) it will be shown, from [Stage One; emphasis added] the mode in which the empirical intuition is given in sensibility, that its unity is no other than that which the category (according to §20) prescribes to the manifold of a given intuition in general. Only thus, by demonstration of the a priori validity of the categories in respect of all objects [Stage Two; emphasis added] of our senses, will the purpose of the deduction be fully attained. (B144-5)

Once again, Kant’s reference to the “mode in which the manifold for an empirical intuition is given” may suggest that he is now to be concerned with the special features of human sensibility. And B160 (with its footnote), as we know, in fact specifically considers the problem of space and time as objects (although this was also anticipated in §17 and its footnote). But an alternative reading is clearly available, with respect to Kant’s reference to the “mode in which” objects are “given.” On that reading, Kant is for the first time to be concerned with the mode in which any kind of manifold is to be given as a real object to which forms of understanding are applicable. Thus the emphasis is on a “given” manifold, and the consequent need for its unity, that is no longer a “given” manifold in the sense that was first in question. Rather, any relevant unification with respect to what is “given” to our senses is now to be regarded as a unification with respect to objects themselves, not merely with respect to a body of material (potentially) available for formation into cognitions. (It is independently evident that Kant was comfortable speaking of the “given” in both of these ways. Sensations, for example, are “given” for Kant, and so are the “appearances” apprehended through sensations.) This, it should be clear, is perfectly compatible with seeing the logic of the transition as lying in the fact that the relevant “objects”—even when those objects are taken to be altogether (empirically) real—are merely the intentional “correlates” of (possible) cognition.

We might also note that §17, which we have already seen to anticipate §26’s attention to the distinction between space and time as “forms” (aspects of cognition) and as “objects,” had itself been content merely to draw the relevant distinction (between cognition and object). That is, it did not itself explicitly conclude that the forms of understanding, just by virtue of their “applicability” to the manifold in cognitive states, must also be “applicable” to possible objects of such states. Rather, at least in the passage already examined, any relevant unification is still regarded as unification with respect to a manifold internal to the act of cognition itself. Here is Kant’s own more general formulation in the same section:

Understanding is, to use general terms, the faculty of cognitions. These consist in the determinate relation of given representation to an object; and an object is that in the concept of which the manifold of a given intuition is united. Now all unification of representations demands unity of consciousness in the synthesis of them. Consequently it is the unity of consciousness that alone constitutes the relation of representations to an object, and therefore their objective validity and the fact that they are cognitions; and upon it therefore rests the very possibility of the understanding. (B137)

Obviously, Kant does not suppose that, on the basis of this, any considerable effort will in fact be required for finally completing the Deduction, that is, for moving from the structure of cognition to the structure of correlative “reality.” As it happens, the completion requires only a page plus a footnote in §26.

In §26, the following is the first step in Kant’s explanation of why the categories necessarily apply to objects:

[1] First of all, I may draw attention to the fact that by synthesis of apprehension I understand that combination of the manifold in [emphasis added] an empirical intuition, whereby perception, that is, empirical consciousness of [emphasis added] the intuition (as appearance), is possible. (B160)

In this passage, Kant calls attention to, or at least plays upon, the ambiguity of intuition. On the one hand, we have been concerned with “intuition” as a kind of cognitive state (or would-be cognitive state). But on the other hand, we have thereby also been concerned—although it has so far not been emphasized—with “intuitions” in the sense of the “appearances” that are the (would-be) objects of the former. Presupposing this distinction, the passage then summarizes the Deduction’s first stage: the apprehension of appearances, to the extent that it is cognitive, requires combination of a manifold in that apprehension. (Kant does not bother to repeat that the forms needed for this, at least in their “application” to the manifold, are nothing other than the categories of understanding; §20 had made that clear.)

The second step is this:

[2] In the representations of space and time we have forms of outer and inner sensible intuition a priori; and to these the synthesis of apprehension of the manifold of appearance [emphasis added] must always conform, because in no other way can the synthesis take place at all. (B160)

Following this observation, Kant proceeds, as we have seen, to elaborate on the distinction between the mere “form” of spatiotemporal intuition and space and time as objects of intuition.

I have, of course, already argued that, here and in the corresponding portion of §17, Kant’s references to the “forms” of intuition are not intended as references to space and time themselves, nor even to the spatiotemporal forms of appearances. Rather, they are intended as references to the intrinsic intuitional forms in one’s apprehension of (possible) spaces, times, and appearances. Read in this way, [2] then simply serves the dual function of reminding us (a) that “apprehension,” except when it is “pure,” is always apprehension through a body of appropriate material and (b) that objects as “appearances” are the necessary correlates of such apprehension. Though we have not so far been explicitly concerned with the notion of correlation, we are now to be concerned precisely with these correlates, and with the spatiotemporal “forms” that are apprehensible in them.

This noted, the final inference is indeed an effortless one:

Thus the unity of the synthesis of the manifold, without or within us, and consequently also a combination to which everything that is to be represented as determined in space or in time must conform, is given a priori as the condition of the synthesis of all apprehension—not indeed in, but with these intuitions. . . .[1] the categories are conditions of the possibility of experience, and are therefore [3] also [auch] valid a priori for all objects of experience. (B161)

In his translation of this passage, Kemp Smith eliminates Kant’s use of auch in the last sentence. This, of course, obscures the crucial inference from the applicability of categories as conditions of all possible experience (Stage One: premise [1] above) to their “applicability,” in what must be a merely correlative sense, to all of possible objects of experience (Stage Two: conclusion [3] above). The inference is mediated by a single consideration, namely, the one presented in premise [2].

The Kantian “forms of judgment,” available to us from the Metaphysical Deduction, are the forms intended to serve the relevant function of unification in sensibility. As I have argued at length, but primarily on the basis of an examination of §§18-19, and of Kant’s distinction between judgments of perception and experience, their function is therefore to unify bodies of “associations” internal to intuitional states. Regarded precisely as so functioning, the same forms are of course what Kant calls the “categories.” In one sense, therefore, the Deduction’s first stage had already demonstrated the necessary “applicability” of categories to sensibility. But at that stage, the manifold in question was simply within (would-be) cognitive states themselves. Completion of the Deduction required an inference to their correlative application to objects. The inference, we have seen is almost effortlessly accomplished. But that is what we ought to have expected, if the objects in question are indeed to be regarded as mere “correlates” of their very apprehension in the first place. And that, as Kant informs us, in his general “Observation” on the first stage (§21), is precisely how objects are to be regarded: “Now things in space and time are given only insofar as they are perceptions (that is, representations accompanied by sensation)” (B147).