I. Introduction

IT REMAINS to consider the possibility of applying this general approach to Kant’s more specific specification of the “categories” of understanding. To be sure, the Deduction itself makes no effort in this regard, apart from its general argument that “forms of judgment” are required for the formation of concepts through the unification of manifolds in intuition. As to what the particular forms are, Kant simply appeals (B143) to a ready-made “table” alleged to display them (A70/B95). The categories are then supposed to be those very same forms. That is, they are supposed to be those same forms somehow functioning as concepts themselves, not merely as forms. Unfortunately, from our point of view, the table is a classification of the general forms involved in the uniting of already formed concepts into judgments. What we need to see is how the same functions might also, on a deeper level, be involved in the original formation of concepts from bodies of imaginative material. (This, it should be clear, would not be the same as showing that judgments involving the categories are logically entailed by, or presupposed by, judgments involving other concepts.)

II. From Matter to Form: General Structures

As we have seen, Kant is especially attentive to the role of a clearly non-logical kind of “necessity”—thus at least of a proto-causality—in the first-edition Deduction’s account of the concept of an “object.” Obviously, what I propose to take very seriously in the account is the fact that Kant presents it in terms of what is needed in order that a “synthesis of recognition in a concept” may bear on, but also go beyond, a synthesis of imaginative “reproduction”:

The word ‘concept’ might of itself suggest this remark. For this unitary consciousness is what combines the manifold, successively intuited, and thereby also reproduced, into one representation. . . . At this point we must make clear to ourselves what we mean by the expression ‘an object of representations’. . . . Now we find that our thought of the relation of all cognition to its object carries with it an element of necessity. . . . [T]he unity which the object makes necessary can be nothing else than the formal unity of consciousness in the synthesis of the manifold of representations. . . . But [a concept] can be a rule for intuitions only insofar as it represents in any given appearances the necessary reproduction of their manifold, and thereby the synthetic unity in our consciousness of them. The concept of body, in the perception of something outside us, necessitates the representation of extension, and therewith representations of impenetrability, shape, etc. (A 103-6; my emphasis)

It should be clear that Kant is not merely concerned, in this passage, with the logical relation of necessity whereby the proposition that something is, say, a body entails the proposition that it is shaped or impenetrable. If he were, then only the sheerest confusion could explain why he formulates the issue in terms of the kind of necessity that concepts or predicates entail with respect to “reproduction,” not merely with respect to other concepts or predicates. It should also be independently clear, from the context, that the relevant “unity of consciousness in the synthesis of the manifold of representations” is supposed to be some kind of unity with respect to immediately apprehensible appearances. The synthesis in question, Kant tells us, is “necessitated” by a concept in the very perception of an object as falling under it.

In this regard, the second-edition Deduction may seem less explicit than the first (B162-3). Kant considers there, as an instance of the “unity of the manifold,” to be effected by means of the category of causality, a relationship that he represents as obtaining between two states of freezing water. In this case, it is appearances already conceptualized as material objects (quanta of water), or as their states (whether frozen or liquid), that are said to be brought to an appropriate sort of unity of consciousness through categories. (The same goes for Kant’s discussion of “quantity” in the same passage.) But I have of course argued that the second edition, just as much as the first, requires an intellectual forming of materials on a level that is still “below” the level of conceptual representation. The passage does not speak to this point. At most, it can be intended to offer an example of a mode of conceptualizing appearances that is a necessary upshot of the more basic forming in question. The material for the latter cannot be concepts of objects and their states. It can only be a manifold of anticipations and retentions of possible appearances, in principle conceptualizable as objects and states. By the same token, any representation of “necessity” in regard to the (otherwise merely) anticipated and retained possible appearances cannot “yet” be the representation of causal relations as such.

In any case, in the first-edition Deduction, Kant proceeds to insist upon a “transcendental condition” for the relevant “necessity.” All necessity, he claims, is grounded in such a condition (A 106). The ground of this particular necessity is “transcendental apperception.” Now we have already seen that Kant equates transcendental apperception with some kind of consciousness of oneself as an enduring self, and the latter of course in some way involves “necessity” in its own right. It does so at least in the sense that consciousness of oneself as an enduring self is a necessary condition of the conceptualization of appearances. But now Kant seems to make an implausibly stronger claim. For he seems to claim that original self-consciousness involves, beyond its role as a necessary condition of something, a special consciousness of the necessity of something. Specifically, we now seem to be dealing, not simply with a (relatively) necessary consciousness of something identical, but with a consciousness of something that is itself necessarily identical: with something that “has necessarily to be represented as numerically identical.” This, in turn, is then held to require that we go beyond “empirical data” (A107). (In other passages as well, Kant seems to speak of the consciousness of one’s identity as the consciousness of something that is “necessary,” thus as an a priori mode of consciousness of some special matter of fact [e.g., A108, 109, 112, 113, 116].)

It has been argued, most ably by Guyer, that Kant indeed subscribes at this point to an implausibly strong view of self-identity.² But it is possible to offer a more plausible reading. (In making the point, I shall limit myself to the problem of anticipation in judgment. Later, I shall have more to say about the role of retention.) In conceptualizing an appearance as an object of a particular kind, I necessarily anticipate the possibility of additional appearances, relative to the satisfaction of certain conditions. This, I take it, is supposed to be evident directly from reflection on what concepts of objects are. But something else is now supposed to be evident from the nature of concepts. First of all, Kant now tells us that it is necessary, not simply that additional appearances be anticipated, conditional upon the satisfaction of particular conditions, but also that they be anticipated precisely as necessary, given satisfaction of those conditions. That is the first point. The second is one on which we have already touched in the preceding chapter.

In conceptualizing appearances, it is necessary, not simply that one anticipate additional appearances, conditional upon the satisfaction of certain conditions, but also that one (albeit pre-reflectively) anticipate those additional appearances precisely as ones that one might oneself eventually apprehend. (Indeed, we have seen that Kant seems to demand even more than this. He seems to demand, not simply the anticipation of appearances that one might oneself eventually apprehend, given the satisfaction of certain conditions, but of appearances whose very apprehension will contain a retention of having been anticipated in the first place. I shall have more to say about this point presently.) The conjunction of these points yields a harmless, but important, sense in which, in conceptualizing appearances, one necessarily represents, not simply one’s own identity through a course of possible perceptions, but one’s own “necessary identity” through such a course of perception. Obviously, this is not to say that one necessarily represents the necessity of one’s own existence.

Now Kant may also have more than this in mind, when he says that one’s identity is present to one “a priori.” But that is no cause for alarm either. If it means anything more than the truism that—like the representation of necessities in experience—we are dealing with something that can never be given “through empirical data” (since its representation involves necessity, and also because the transcendental self is not an intuitable object), then it may only be a way of indicating that one’s original self-consciousness is indeed anticipational in nature. Kant himself acknowledges a sense of the a priori that comes to no more than this (B2). He makes it clear that he will not speak in that sense in the Critique, when he is considering cognition as such. But he does not rule out the possibility of speaking, in other contexts, of a mere thought, or of a mere consciousness, or of a mere representation of some matter of fact “a priori,” when he merely means a thought, or a consciousness, or a representation whose ground is a mere (though appropriately rule-governed) anticipation. For obvious reasons, Kant generally does not use the term in this way. But at A108, in particular, there is surely no reason to suppose that what is in question is the possibility of knowing or cognizing one’s own continuing identity a priori, since Kant himself speaks merely of the possibility of thinking it a priori. I presume that Kant is simply intending to speak of a mode of thinking that is, in an important way, anticipational in nature. (Related uses of the term, in more explicit connection with causality, occur at A112 and B280.)

But why should Kant see a transcendental significance in the mere ability to think one’s own identity a priori, if all that he means is the mere ability to anticipate its continuation? To see the answer, what we need to remember is that Kant is concerned with a special way of anticipating one’s identity from the start, namely, with that mode of anticipation that is constituted in the very act of conceptualizing appearances. That indeed offers a puzzle demanding transcendental reflection: How can one manage to be engaged in the act of anticipating oneself, when one’s attention is directed toward objects (appearances) distinct from oneself? It is, I suggest, precisely this puzzle that Kant is attempting to address in this context:

For this unity of consciousness [i.e., transcendental unity of apperception] would be impossible if the mind in cognition of the manifold could not become conscious of the identity of function whereby it synthetically combines it in one cognition. The original and necessary consciousness of the identity of the self is thus at the same time [zugleich] a consciousness of an equally necessary unity of the synthesis of all appearances according to concepts, that is, according to rules, which not only make them necessarily reproducible but also in so doing determine an object for their intuition, that is, the concept of something wherein they are necessarily interconnected. For the mind could never think its identity in the manifoldness of its representations, and indeed think this identity a priori, if did not have before its eyes the identity of its act, whereby it subordinates all synthesis of apprehension (which is empirical) to a transcendental unity, thereby rendering possible their interconnection according to a priori rules. (A108; my emphases)

The trick is, of course, to see just how, in being conscious of nothing more than the manifold of representations, one may indeed at the same time be conscious of the unity of one s own “act”—that is, to see this without taking representations as states, in this context, potentially “ascribable” to oneself, as opposed to regarding them as the very appearances that are in principle conceptualizable as objects.

There is an additional difficulty to consider. It might seem that, on the proposed reading, Kant’s position is inconsistent. This is because, it might be alleged, the position requires the postulation of judgmental form on what ought to be a prejudgmental level. It is of course clear that judgmental form is needed in order to account for the difference between merely “sensibly” anticipating appearances and anticipating appearances as necessary (relative to certain conditions). The representation of anything as necessary, on Kant’s view, rests on the capacity for differentiating between the modal forms of merely “problematic” or “assertoric” judgment, on the one hand, and “apodeictic” judgment, on the other (A74-6/B99-101). Since these distinctions are irreducibly “judgmental,” they could not already have been available on the level of mere sensibility. But this, it appears, is no problem, since the representation of the necessity in question is introduced, in Kant’s reasoning, only at the point at which imaginative material is finally transformed by judgment. Consider, however, the originally postulated material. Presumably, even purely “imaginative” anticipation is not merely the anticipation of appearances as forthcoming. Rather, it is the anticipation of appearances as possibly forthcoming, relative to the satisfaction of conditions (in their turn represented as possible). This seems to imply that we need to introduce “forms of judgment” after all, including at least the forms of hypothetical and problematic judgment, on a level that is altogether below what Kant himself recognizes as truly judgmental. If we do not, then we will be unable to find that body of material needed for judgmental form to transform in the first place.

It is important to see that, in effect, this difficulty is one that we have already considered. We disposed of it when we decided to avoid verbal quibbles as to whether or not “animal” anticipation involves concepts or judgments. It is merely a verbal quibble whether animals have concepts and judge or whether only more elevated beings are capable of such functions. The substantive question concerns the relation between purely animal and other modes of “judgment.” We may concede that there is a sense in which animals engage in the representation of states of affairs as, say, “possible.” Whether or not Kant himself was clear on it, there is no need for him to have denied it. Nor does consistency require that he provide an account of what “judgment” of the sort amounts to: whether it could be treated in purely dispositional terms or perhaps functionally, for example. To be consistent, all that Kant needs to hold is that whatever such judgment amounts to, it can at most serve as material in distinctively human judgments. (If it is not plausible to hold even this much, then Kant would end up on one of the horns of the dilemma that I have been trying to avoid: (a) ultimately regarding both human and animal judgment in purely dispositional or functional terms; (b) implausibly isolating the two sorts of judgment from one another, within the mental life of those human subjects in which both obviously play an important part.)

This line of response may seem to entail that it is impossible to draw any real line between “lower” and “higher” capacities. For example, if we concede that animals are capable of representing states of affairs as possible, then why not have conceded, in the first place, that they are capable of representing states of affairs as necessary, given the satisfaction of conditions that are possible? Consistently with this, after all, we might continue to hold that, whatever such representation amounts to, it could at most relate to the relevantly higher level by serving as the mere material in judgments on that level. But then the problem would remain: How could we specify the difference between such “levels” in the first place? Certainly, we will have deprived ourselves of the option of saying that the difference rests, even in part, on a capacity for the representation of “necessity.”

So far as I can see, there are only two possible responses on the general account proposed. First, we might take the role of judgmental “form” to lie precisely in the fact that it does not merely serve to provide forms for “judgment,” as the latter is normally understood, but rather that it serves equally for the incorporation of (possibly) “judgmental” material in intuitions. Apart from the latter function, and regardless of the internal articulation of the material in question (whether, for example, it involves a “representation” of possibility, necessity, or what have you) we might then have some kind of “judgment,” but we would not have any kind of judgment that involves judgmental form. At the very least, this does not seem Kantian. It accords too great an importance to the form of intuition. In effect, it virtually equates the forms of judgment with the latter: what is needed for a full-blown (Kantian) judgment would simply be the incorporation of a certain sort of material by means of such form.

For a more promising approach, we need to take a closer look at the problem of self-consciousness. As it happens, this will eventually take us closer to seeing how the representation of objective nature, as subsumable under causal laws, is part of the constitution of empirical concepts. (We need to remember that the mere anticipation of appearances as necessary, relative to the satisfaction of certain conditions, does not suffice for the representation of causality involving objects or their states.) It is here that we finally need to return, as well, to my earlier suggestion concerning the role of “higher-order” modes of anticipation: anticipations not simply of appearances relative to conditions, nor even of appearances represented as necessary relative to conditions, but of appearances whose apprehension contains some kind of retention of their very anticipation in the first place.

In one respect, it should be clear that there is something lacking in any account of empirical concepts that takes the act of conceptualizing an appearance to consist in nothing more than a set of anticipations in regard to that appearance, or in regard to its apprehension—even when those anticipations are truly ingredient in the original apprehension in the first place. At the very least, what is lacking is the representation of the very same concept, as a potential ingredient in any number of additional apprehensions (or, alternatively, as “predicable” of any number of additional appearance), and of course in particular in those in fact anticipated.

This is not to deny that a particular set of anticipations, just like any concept, can be regarded as a repeatable “term,” multiply tokenable or instantiable in any number of intuitions: any number of intuitions might contain this or that particular set of anticipations. But it is still impossible to suppose that any such set can by itself constitute a truly conceptual mode of representation. (In what follows, I continue to abstract, for a while, from the role of “retention.” In addition, I also abstract from the normative dimension that Kant no doubt intended as part of his own notion of concepts as embodying “rules.”)³ If the intuition in question is to amount to genuine conceptualization, then it is plausible to suppose that at least the following is necessary: that I need to be doing more than merely anticipating whatever additional appearances are in question. Beyond this, I need to be anticipating those appearances precisely as appearances that are subsumable under a particular concept. It seems plausible to insist, in other words, that if I am not anticipating the additional appearances specifically as subsumable under whatever concept was supposed to have been applied in the first place, then my anticipation of those additional appearances will remain irrelevant to any would-be act of conceptualizing that appearance.

This may, of course, appear to imply that any account of conceptualization, in terms of the anticipation of appearances, is bound to be circular. I want to argue that it is not. As a first step, we need to see why the following suggestion is not a way out of the difficulty. It might seem that we can avoid circularity, simply by specifying that the anticipations peculiar to the conceptualization of appearances are to be regarded not merely as anticipatory of possible additional appearances, relative to certain conditions, but as anticipatory of additional appearances in whose very apprehension the given set of anticipations will likewise be an ingredient. The proposal simply appeals, as I have already suggested is necessary, to a second-order level of anticipational consciousness. At least one advantage of it will therefore be that it provides a criterion for distinguishing human from purely animal “judgment”: even if there is a sense in which the latter involves the anticipation of possibilities, necessities, conditions, and the like, it does not, we might propose, involve higher-order anticipation of the sort suggested.

Whatever weakness the proposal may prove to have, it is important to be clear about one part of it. It is important to appreciate the sense in which the suggestion does not, although it may seem to, presuppose a conceptually determinate mode of self-awareness, as a necessary condition for the formation of any concept. If it presupposed such a mode of self-awareness, then the account would, of course, be circular. But it may seem in fact to be circular in just this way. For it presupposes the ability to represent sets of “one’s own” anticipations, as a pre-condition of the formation of empirical concepts. However, we already have the materials to provide a response to this charge. The proposal, again, was this: that the anticipations peculiar to the conceptualization of appearances are simply anticipations of additional appearances, as appearances correlative with an apprehension that contains the very same anticipations as were originally in question. This was supposed to suffice, in a non-circular way, to meet the demand that conceptualization involves the anticipation, not simply of additional appearances, but of appearances that are in their own turn subsumable under the very same concept that subsumes the appearance originally in question. On the proposal, the role of the “same concept” is to be non-circularly filled by the set of original anticipations itself.

It should at least be possible to see, given our account of self-consciousness, that whatever “higher-order” awareness is required for this proposal cannot be presumed to rest on the possession of a self-concept. What we need to remember, for this point, is simply that an appearance that is apprehended through a set of anticipations is a different “object” for consciousness than an appearance that is apprehended through a different set of anticipations (or, if possible, through none at all). It is this fact, as I have suggested, that opens the door to recognition of the fact that the apprehension of appearances through sets of anticipations already is, by its very nature, a way of “apprehending” those anticipations themselves. But it is a way of apprehending them that does not presuppose any special kind of reflection on, not to mention a concept of, “oneself’ as a subject of apprehension. The apprehension in question is directly constituted in the original apprehension of appearances.

This provides an important part of what we need to develop the Kantian theory of concepts. But it is not enough. It is insufficient, because it is plainly wrong to suppose that the conceptualization of a given appearance requires the anticipation of additional appearances as apprehensible through the very same anticipations as were ingredient in the apprehension of the given one. For example, in conceptualizing an appearance as a rose bush, I may be anticipating the possibility of additional appearances, each of which is likewise conceptualizable as a rose bush, and each of which is anticipable as the upshot of the satisfaction of certain equally anticipable conditions (that is, each of which is anticipable as the upshot of certain anticipable ways of “stretching” the given appearance). But I surely do not need to anticipate the apprehension of those additional appearances as themselves eventually apprehensible through these very same anticipations. That would in fact be incompatible with any reasonable conception of a perceivable objective reality. At most, what would be reasonable is that the additional appearances be anticipated as apprehensible through anticipations appropriately related to the given anticipations.

What is needed, therefore, is the notion of anticipating the apprehension of additional appearances as containing a suitable transformation of the anticipations ingredient in the apprehension of a given appearance. I need to anticipate appearances as ones in whose eventual apprehension will be contained the anticipation of additional appearances that are at most appropriately related to the ones that were anticipated in the apprehension of the given appearance—and that are obtainable upon the satisfaction of conditions that are at most correspondingly related to those originally in question. Among other things, for example, the additional appearances need to be anticipated as apprehensible through anticipations relating to the possibility of getting back to (or to the possibility of having been arrived at from) the starting point of the original appearance, whereas the original appearance was at most apprehended through anticipations relating to the possibility of getting from there to those (and, of course, to the possibility of having been arrived at from still others).

In general, what counts as a “suitable” transformation of an original set of anticipations may simply be expressive of what one takes to be the conditions of empirical objectivity. It must therefore be embodied, in general, in one’s very concept of objective reality. But there is no circularity in conceding this: higher-order anticipations of just this sort are supposed to be constitutive of one’s concept of empirical objectivity in the first place.

Obviously, this requires a degree of sophistication beyond anything that we have so far considered. (Remember: What is in question is not merely the capacity for a systematic transformation of one’s own anticipations; it is the capacity for the anticipation of appearances as eventually apprehensible through such systematically transformed anticipations.) It is not implausible to suppose that this requires something beyond the capacity of most creatures. But while abilities of the sort require sophistication, they do not presuppose the capacity for any special kind of self-reflection on the part of anticipating subjects. Rather, the latter is supposed to be made possible through the former in the first place: on the present level of analysis, the anticipation of appropriately systematic transformations of “one’s own” anticipations, in the eventual apprehension of additional appearances, simply is a way of anticipating additional appearances. It is just a more sophisticated (though not merely more “complex”) way of anticipating appearances. Therefore, to whatever extent it may indeed be the case that distinctively human self-apprehension rests on the ability to conceptualize appearances—which is that part of Kant’s theory to which commentators’ attention is usually limited—this claim remains compatible with the suggestion that—in a sense that is usually overlooked—it is precisely through such conceptualization that distinctively human self-apprehension is originally constituted.

From here, it is a short distance to Kant’s notion of concepts as, or serving as, “rules” in relation to appearances (cf. A106). But the present account is different from what that notion may be supposed to entail. It may be supposed to involve the idea that concepts, as mental “predicates,” are quasi-linguistic items or terms, attachable to given perceptions or appearances (instantiable or “tokenable” by them), but only to be counted as representations to the extent that the subject is in command of the rules that “govern” those predicates. The metaphor suggests a notion that is very different from the one proposed. It suggests that the rules determinative of representational content function as part of a system of rules that govern would-be predicates in a purely external way, that is, by virtue of factors distinct from the way in which those predicates are actually constituted, as “terms,” in the first place. So far as I can see, this yields a purely mechanical account of understanding.

The proposal has the advantage of regarding the rules that “govern” concepts as internal to concepts themselves. But it does so without abandoning the idea that concepts are indeed a kind of “term” to be governed in the first place—which they need to be in order to be sufficiently like predicates. What we need to see is that the terms in question are sets of anticipations. Once we see this, then we are free to suppose that one part of such sets is simply a set of higher-order anticipations. The latter bear on transformation of those very sets, as they move from intuition to intuition. The “rules” that define the identity of concepts are just higher-order anticipations that bear on sets that include lower-order anticipations as well. The conceptualization of appearances is their apprehension through sets of the latter, appropriately embedded in the former. No more than this, then, needs to be involved in Kant’s notorious doctrine of the “imposition” of intellectual form on sensible material. What is noteworthy, as I see it, is simply that the forms in question are not merely externally related to the corresponding material. Their relation is not analogous, for example, to that between a “functional role” and a term that happens to “play” it.

These suggestions provide, finally, for a more substantial role for the phenomenon of “retention” as well. In doing so, they explain Kant’s suggestion, noted earlier, that experience requires, not simply the retention of earlier experiences, and not simply the retention of earlier experiences by experiences themselves, but the anticipation of experiences as themselves retaining the fact of their own anticipation. On one level, it might seem trivially obvious that something like this is required for experience. For if anticipations are going to be relevant to experience in the first place, then we need to be able to recall that anticipated appearances had in fact been anticipated. But while it may be essential that subjects retain their anticipations in this sense, it would not follow that such retentions are a part of the structure of conceptual acts. Nor would it follow that whatever anticipations are essential to conceptual acts are, at bottom, anticipations of acts that are in their own turn specifically retentional. (Here, I am ignoring the more primitive types of anticipation and retention that we considered in Chapter Three. That these more primitive types are also essential, not simply to experience in general, but to the structure of individual states or acts of consciousness, still falls short of entailing that they are specifically essential to the act of conceptualization. For the latter, we need to be concerned with the more “associational” type of anticipation and retention.)

III. Causality, Substance, and Community of Interaction

This take us at least significantly close to some of the full-blown categories that Kant claims to be counterparts of the purely “logical” forms of judgment. We may begin with the concept of causality. Kant’s notion of concepts of objects as “rules,” or as somehow serving as rules, is presumably supposed to be relevant to his view of the role of this concept in experience. In this respect, what we therefore need to recall is the way in which we have now expanded our account beyond an appeal to a notion of anticipational and retentional “rules” that bear merely on the anticipation and retention of appearances as possible upshots (or even as necessary upshots) of the possible advance of experience, given the satisfaction of certain conditions. Beyond this, we have in effect introduced the requirement that appearances, as potential subjects for conceptualization, must be regarded as subject to whole systems of rules, systematically correlated with all of the possible “points of view” from which any appearances are possibly obtainable in the first place.

It follows from this that every point of view from which appearances are anticipable as possible must be regarded as subsumable under an overarching system of rules. This is because the anticipation of any possible appearance, in connection with a given one, is necessarily correlative with the anticipation of the possibility of returning, via a rule-governed transformation of the originally anticipated path, from that appearance to the given one. This at least takes us well past the mere proto-causality with which we began. It is not implausible to suppose, it seems to me, that it takes us far enough past that notion to account for a concept of an objective “world of appearances,” i.e., for an objective nature in Kant’s sense:

The unity of apperception is thus the transcendental ground of the necessary conformity to law of all appearances in one experience. This same unity of apperception in respect to a manifold of representations (determining it out of a single one) acts as the rule, and the faculty of these rules is the understanding. All appearances, as possible experiences, thus lie a priori in the understanding, and receive from it their formal possibility, just as, insofar as they are mere intuitions, they lie in sensibility, and are, as regards their form, only possible through it. However exaggerated and absurd it may sound, to say that the understanding is itself the source of the laws of nature, and so of its formal unity, such an assertion is nonetheless correct, and is in keeping with the object [dem Gegenstande] to which it refers, namely, experience. (A 127; my emphases)

That the laws of appearances in nature must agree with the understanding and its a priori form, that is, with its faculty of combining the manifold in general, is no more surprising than that appearances themselves must agree with the form of sensible intuition a priori. For just as appearances do not exist in themselves but only relatively to the subject in which, so far as it has senses, they inhere, so the laws do not exist in the appearances but only relatively to this same being, so far as it has understanding. . . . As mere representations, they are subject to no law of connection save that which the connecting faculty prescribes. Now it is imagination that connects 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. All possible perception is thus dependent upon synthesis of apprehension, and this empirical synthesis in turn upon transcendental synthesis, and therefore upon the categories. (B164)

That something like the proposed reasoning is in fact Kant’s own is also suggested by the following comparison, offered by Kant himself, between (a) the ground of our knowledge that all matter has a degree of presence in space and (b) the ground of our knowledge regarding universal causality:

[a] This reveals the possibility of knowing a priori a law of alterations, in respect of their form. We are merely anticipating our own apprehension. . . . In the same manner, therefore, in which time contains the sensible a priori condition of the possibility of a continuous advance of the existing to what follows, [b] the understanding, by virtue of the unity of apperception, is the a priori condition of the possibility of a continuous determination of all positions for the appearances in this time, through the series of causes and effects. . . . (A210-11/B255-6; my emphases).

Of course, none of this is meant to amount to a proof of the Second Analogy— that all events are causally necessitated. At most it amounts to the conclusion that all objects are subject to causal laws. And all that it additionally attempts to do, with this conclusion, is to show how it can be grounded directly in reflection on the original constitution of concepts of objects out of what would otherwise remain purely imaginative anticipations and retentions. It seems obvious that the Second Analogy attempts to do more than this. That is not my concern here, although I have also argued elsewhere that, at least on a relatively sophisticated “phenomenalism” of the sort that I have ascribed to Kant, together with some additional (though false) assumptions that seem to be Kant’s, even that stronger inference can be regarded as valid.⁴

It may not be possible to come even this close in regard to Kant’s argument for a universal “community” of causal interaction in the Third Analogy, that is, with regard to the principle of reciprocal causality among all coexisting objects:

Things are coexistent when in empirical intuition the perceptions of them can follow upon one another reciprocally. . . . Thus I can direct my perception first to the moon and then to the earth, or conversely, first to the earth and then to the moon; and because the perceptions of these objects can follow each other reciprocally, I say that they are coexistent. . . . [But] the synthesis of imagination in apprehension would only reveal that the one perception is in the subject when the other is not there, and reciprocally, but not that the objects are coexistent, that is, that if the one exists the other exists at the same time, and that this is necessary, in order that the perceptions can follow one another reciprocally. Consequently, in the case of things which coexist externally to one another, a pure concept of the reciprocal sequence of their determinations is required, if we are able to say that the reciprocal sequence of the perceptions is grounded in the object, and so to represent the coexistence as objective. (B256-7; my emphases)

I shall presume that we may take the liberty of supposing that Kant is putting his own point somewhat poorly in this passage. This would seem to be necessary in order to find a line of thinking that is in fact plausibly ascribable to someone of Kant’s intelligence.

In the light of our results so far, it ought to be expected that we take very seriously Kant’s claims about the “reversibility” of perceptions. On some readings, that notion is given a surprisingly minimal role in Kant’s argument. In the argument as stated, Kant himself seems to suggest that judgments that conceptualize appearances as coexisting objects are just at bottom judgments about the possibility of certain in principle reversible perceptions, or at least—as I would prefer to put it—that they are judgments that contain, as their whole “material,” nothing more than anticipations (and retentions) of possibilities regarding reversible perceptions. On some other accounts, the role of the reversibility of perception becomes purely negative. For example, one might suppose Kant’s concern with that notion to be motivated simply by the need to show that the reversibility of perceptions is insufficient—as it obviously is—as a criterion for the determination of whether or not objects are in fact coexistent. Positively, one might argue, what is then required for the actual determination of such facts is not any sort of consideration that bears on “perceptions” at all, but rather an appeal to rules concerning “objects” directly.⁵ In any case, I want to attempt to give a more positive role to perceptual reversibility.

Let us begin with Kant’s reference to what mere “imagination” can give us in this passage. In particular, we need to take special care in regard to his claim that what imagination reveals (angeben) is, for example, merely that a certain perception is present, while some other is not. As we already now, Kant specifically connects the notion of imagination, in the Deduction, with the capacity for representing what is not present, not with the capacity for representing what is present while something else isn’t. So it would seem reasonable to suppose that, in the present passage, Kant is in fact intending to ascribe to the imagination, not simply the indication that some perception is present at a moment while some other is absent, but, more specifically, the representation of the very absence of the latter, while the former is present. In addition, we cannot help but presume that the relevant type of imaginative representation, in this case, is meant to be none other than some type of anticipation, namely, anticipation of the very possibility of the presence of what is in fact absent.

With these as our initial presumptions, it is then reasonable to conclude that Kant’s real point about what imagination can and cannot give us, in this passage, is not simply (and trivially) that a sequential awareness of perceptions is insufficient in order to account for the representation of objects as coexistent. The point must rather be, more informatively, that not even any number of additional (first-order) anticipations is sufficient to account for such representation. Someone, of course (for example, Hume), might well have supposed that it is sufficient. Kant wants to inform us that something more is needed.

Now obviously, one thing that is needed for the representation of objects as coexistent is the representation of a certain kind of “necessity.” That, of course, is needed for any representation of objects. Kant, I take it, wants to make it clear that something else is needed as well. This is evident from the phrase that I emphasized in the passage that I quoted. The additional factor has something to do with a representation of “reciprocity.” But what could be the subject of reciprocity in this case, if Kant is not to beg the question from the start, with respect to his establishment of the need for reciprocity among the states of objects? We can only suppose that Kant’s formulation is unclear. What can be in question, at this point, cannot be the reciprocity of objective states. It can only be the reciprocity of possible perceptions.

If the reciprocity in question concerns the possibility of perception, then it presumably amounts to more than the trivial two-fold fact (i) that at a certain moment a certain perception, say B, may be anticipated (and even anticipated as necessary, under certain conditions) while some other, say A, is actually present to perception and, reciprocally, (ii) that at another moment A may be anticipated while B is actually present. That would yield no more than the weakly “reciprocal” anticipation with which we began: imagination informing us that some perception is present while some other, which might have been present, is not. What Kant might additionally be supposing, I therefore suggest, involves a rather more complex type of anticipation. It involves the anticipation, not simply of “reciprocally” perceivable (first A then B, or first B then A) perceptions, but of reciprocally anticipable perceptions. That is, it involves a structure in which A, while perceived, anticipates B, not simply as necessary under certain conditions, but precisely as itself in turn anticipating A; and, reciprocally, it involves a structure in which B, while perceived, anticipates A, not simply as likewise necessary under certain conditions, but precisely as itself anticipating B.

In the terms developed earlier, what we have is the following situation. In conceptualizing some appearance A as a real object, we must, as we have already discovered in the Transcendental Deduction, be anticipating various possible modes of “stretching” A (qua appearance), so as eventually to arrive at upshots that are necessary, given the way that the stretching in question is anticipated as occurring. Now suppose that A (qua object) is conceptualized as coexisting with B. This implies that, among the manifold of anticipated stretchings of A, there will be at least one that is the “reciprocal” of one that would also have been anticipated in the act of conceptualizing B as coexisting with A. This is necessarily so because there will be at least one such stretching that culminates in an appearance conceptualizable as B itself. But, as we have also already seen in the Deduction, a special feature of the anticipations ingredient in conceptual acts is this: that they anticipate not merely the possible stretching of given appearances, so as to arrive at necessary upshots, but they anticipate the necessity of upshots that in their own turn “retain” the very possibility of reversing that anticipation, so as to arrive back at the original appearance as a necessary upshot as well.

It follows from this that, considering A and B together—as appearances conceptualizable as coexisting objects—the conceptualization of either of the two as coexisting with the other necessarily involves the anticipation of possible perceptions of the two that are connected by a single system of necessities. These necessities will, of course, in the first instance, bear on the “possibility of perception” itself. But for Kant, as I have interpreted him, judgments concerning the existence of objects and their states simply are judgments concerning (or, much better: simply are judgments that contain, as their “material,” nothing more than anticipations concerning) the possibility of perceptions. Therefore, judgments concerning the coexistence of objects entail that those objects are causally connected with one another.

Strictly, this reasoning cannot get us to the conclusion of the Third Analogy, as Kant appears to intend it. At most, it can get us to the conclusion that coexistent objects stand in a single causal system, that is, in a system to which a common set of causal laws applies. I do not see how to get from there to the stronger conclusion that coexistent objects, be it directly or indirectly, causally effect one another at every moment: “Each substance . . . must therefore contain in itself the causality of certain determinations in the other substance, and at the same time the effects of the causality of that other . . .” (A212/B259). On the other hand, it has been argued that Kant did not intend the stronger conclusion in the first place, or at least that he was merely careless in espousing it.⁶

It remains to attend briefly to the First Analogy, and to the category of substance or permanence as the “substratum” of all change. Here is a formulation of a part of the argument that is relevant to our purpose. It appears only in the “B”-edition:

[B] All appearances are in time; and in it alone, as substratum (as permanent form of inner intuition), can either coexistence or succession be represented. Thus the time in which all change of appearances has to be thought, remains and does not change. For it is that in which, and as determinations of which, succession or coexistence can alone be represented. Now time cannot be perceived on its own account [für sich]. Consequently there must be found in the objects of perception, that is, in appearances, the substratum which represents time in general; and all change or coexistence must, in being apprehended, be perceived in [an] this substratum, and through relation of appearances to it. (B224-5; my emphases)

The argument, added in the the second edition, seems to differ from a formulation that appears in both editions:

[AB] Our apprehension of the manifold of [the?] appearance is always successive, and is therefore always changing. Through it alone we can never determine whether this manifold [my emphasis], as object of experience, is coexistent or in sequence. For such determination we require an underlying ground which exists at all times, that is something abiding and permanent, of which all change and coexistence are only so many ways (modes of time) in which the permanent exists. And simultaneity and succession being the only relations in time, it follows that only in the permanent are relations of time possible. In other words, the permanent is the substratum of the empirical representation of time itself; in [an] it alone is any determination of time possible. . . . Now time cannot be perceived in itself [an sich selbst]; the permanent in the appearances is therefore the substratum of all determination of time. . . . (A182-3/B225-6)

Not surprisingly, there is a problem in all of this that is comparable to what we encountered regarding the “Refutation of Idealism.” The formulation in [AB] seems to rests on a consideration that is specific to the problem of knowing whether or not apprehended succession amounts to a case of successively apprehending some number of coexistent objects or states. (In the latter case, of course, we might also be dealing with the apprehension of a single object from different points of view, or with respect to different of its aspects.) What is “given” is, in an important sense, a succession of appearances. But the possibility always exists that the successive appearances (“this manifold”) really are a set of coexistent objects, or possibly even a single continuously enduring object. Passage [AB] then seems to focus on the problem of “determining”—in the sense of coming to know—whether or not this is the case. (We shall presently see a second respect in which the “determination” of time is necessarily involved in any “determination” of the objective status of appearances.) By contrast, [B] focuses on a problem that is supposedly involved in apprehending successions that are determinable in either way in the first place; this is clear from its initial premise, which simply rests on the fact that time is the form of “inner intuition.” Of course, if something is already necessary for an intuitional apprehension of succession in appearances, then it is also necessary for “determining” the objective status of what is thereby apprehended. But if the basic ground of argument is this, then it is perplexing why Kant originally formulates it at one remove from the point.⁷ I shall limit myself to arguing that the approach that I have adopted can at least explain two things. First, it can explain why, in its later and presumably clarified version, the argument gives priority to purely intuited succession. Second, it can explain why Kant’s argument appears to conflate two very different questions. The key is to be clearer than Kant himself was regarding the notion of a succession “of appearance(es).”

As we have already noted, Kant’s language is often suggestive of a misleading model, namely, one that involves the discrete apprehension of successive appearances. We may well be skeptical as to the sense in which, usefully to Kant’s own argument, the latter might be held to presuppose the representation of time as a “substratum.” Certainly, we may say that the discrete apprehension of successive appearances in some way presupposes the representation of a single time “in which” those successive representations are contained. Presumably, such a representation would then have to involve either (a) some kind of apprehension of representations as successive states of oneself (or as the intentional correlates thereof) or (b) some kind of apprehension of representations (or of their intentional correlates) as coexistent or successive objects (or as coexistent or successive states of some object or objects). In either case, we would need to presuppose a single time in which the items are apprehended as temporally related. But there is nothing in this that should incline us toward saying that time needs to be presupposed as the “substratum” of whatever coexistence or succession may be in question.⁸

If we like, we might of course always regard time as a substratum in which successive moments or stretches of time “inhere.” Such a substratum might then be regarded as in its own turn necessary as a background for the representation of coexistence or change among appearances. But this would still be different from regarding time as a substratum with respect to appearances themselves, apprehended as occupying, or as possibly occupying, such moments or stretches in the first place. If Kant cannot justify the stronger formulation, then his jump to a correlative “substance” as the material of appearances seems arbitrary.

It is at this point that we simply need to recall that the intuitive apprehension of stretches of time is the very material of any objectively determinable apprehension in the first place. This, of course, stems from my interpretation of anticipation and retention in terms of the imaginative grasp of appearances as stretching, or at least as possibly stretching, into the future or the past. I have argued that such a notion is needed in order to support Kant’s attempt to regard some kind of consciousness of a “synthesis of representations,” as the original and pre-judgmental mode of one’s own self-consciousness. That notion, I would now suggest, can also provide a single account both of Kant’s formulation of the argument in [B], as an argument concerning the role of time in “inner intuition,” and of his ease in sliding to the role of time, in [AB], as the substratum of appearances determined with respect to their status as full-blown objects.

In this way of looking at things, Kant would not merely be concerned, in [B], with our general ability to apprehend succession, in the sense of apprehending manifolds of distinct, successively apprehended representations that also happen to be “determinable” in principle (if they are not hallucinations) as either successive or coexistent objects or states of objects. Nor would he merely be concerned, in [AB], with the specific problem of such determination, over and above the mere fact of temporal apprehension. Rather, we would need to suppose that Kant is concerned with the specific kind of apprehension that is essentially involved in the apprehension of any objectively determinable object or state of affairs, that is, with the kind of apprehension that involves the anticipation of appearances themselves as stretching through time.

If we suppose that the manifold of representations that requires time as a substratum, not as a mere “background,” is the manifold qua imaginatively contained in the apprehension of appearances, insofar as the latter are candidates for objective determination, then it is no longer arbitrary to regard time as indeed that manifold’s substratum. For, at least in that specific case, stretches of time are the very object of such a mode of apprehension of the manifold in the first place. If we were only considering an isolated multiplicity of appearances, successively apprehended, then at most we would be entitled to say that the apprehension of those appearances is a way of apprehending (possible) regions of space, namely, a sensory way of apprehending them. The fact that the appearances in question were also apprehended successively would give no title to suppose, by analogy, that such apprehension is also a way of apprehending stretches of time. An awareness of time might be presupposed as a “background” for that apprehension. But there is no useful sense in which we would be entitled to insist that it is the very subject or substance of what is thereby apprehended.

It is important to see that this does not contradict Kant’s assumption that time cannot be perceived fur sich or an sich selbst. Kant is perfectly clear that time is an object of intuition. The only way it can be so, of course, is if stretches of time are an object of intuition. When Kant says that time cannot be “perceived by itself,” he cannot possibly mean to deny such points. Presumably what he means is that, apprehended only as objects of intuition, stretches of time are not yet apprehended as objective states of affairs. For that, they need to be conceptualized as such states of affairs. Now at this point Kant assumes that, whatever is conceptualized, within the realm of appearance, as an objective state of affairs must, at least in that capacity, be regarded as in principle sensibly perceivable in its own right, in such a way that the perception in question will have an eventual bearing on the “determination” (in the sense of “knowing-that”) of objectivity in the case. In this sense, intuitively apprehended time is not, as such, sensibly perceivable. Its apprehension is a necessary ingredient in the conceptualization of appearances as sensibly perceivable objects in space; indeed, the apprehension of manifolds of stretches of (possible) time is the only ingredient—in the sense of material—in the conceptualization of appearances as such objects. But that, we may suppose, is just Kant’s point in denying that time can be perceived in itself. While the apprehension of stretches of time is the only ingredient in the conceptualization of appearances as concrete objects, in the sense of providing the material for the latter, such apprehension could not possibly constitute the latter in its own right. For that, we need something more than the intuitive apprehension of stretches of time. But this is also compatible with recognizing that, in another sense, the conceptualization of appearances involves nothing more. In another sense, it does not involve anything more than a way of conceptualizing those very stretches of time.⁹

Thus supposing that Kant is focusing, not merely on the subjective apprehension of time, as in [B], nor merely on the problem of the objective determination (in the sense of “knowing-that”) of appearances, as in [AB], but precisely on the respect in which the latter is the former, objectively “determined,”¹⁰ we can understand why both of these considerations are present in a single argument, and also why Kant slides from one to the other. The slide is legitimate on one assumption: that the objective determination of appearances is the determination of time itself; it is not merely determination against time as a “background.” That it is not simply the latter follows from the proposed account of anticipation and retention as the matter of conceptualizing appearances. On that account, there is indeed an important sense in which time itself is always the “object” that one conceptualizes as matter in space. From this, it is at least not unreasonable to infer that something, if not time itself, must be the substratum for all appearances and their changes.

Of course, my interest in this study has in any case not been in arguments for specific categorial concepts, as Kant offers them in the Analytic of Principles. It is only in the more general claim that sensible material needs to be subjected to some kind of judgmental form, both as a necessary condition of the conceptualization of appearances as concrete objects and, precisely in that capacity, as constitutive of original self-consciousness. If, along the lines that I have sketched, a valid argument cannot be constructed for any of the Analogies of Experience, it is more important to assess the account I have offered as an attempt to shed some light, from a non-functionalist direction, on Kant’s general theory of understanding and consciousness in the Deduction.

IV. Appendix: Judgments about the Past

It will be useful to conclude with some speculation regarding a somewhat more specialized problem. It concerns the role of anticipation and retention in judgments about the past. It may seem plainly impossible to account for such judgments on the general theory of judgment that I have proposed. There may be a temptation to say that judgments about the present are constituted out of anticipations in regard to possible experience. But judgments about the past are difficult to construe in such terms. In addition, to the extent that “retention” is understood, as it often seems to be, as itself already a kind of judgment about the past, one would not normally suppose it to provide a workable supplement in this regard. However, neither anticipation nor retention is, in a relevant sense, intrinsically judgmental. It is therefore worth considering whether we might not indeed have the materials, along the lines proposed, for dealing with judgments about the past.

Since we are limiting our concern to the problem of predication in intuition, the problem would seem to concern the possibility of judging with regard to imagined stretches of time. In particular, it would seem to concern the possibility of judging, of some imagined stretch of time, (a) that it is past, (b) that it has certain contents, and possibly also (c) that it extends back a certain distance from the present. But apart from that special case, there may, of course, also appear to be a prior question concerning the possiblity of intuitionally imaginative judgments of any kind. I shall simply assume that, to deal with this, Kant will already need to have postulated a specially “modalized” form of anticipation and retention. Such modalization will be needed in order to account for the obtaining of a general and non-formal “sameness” of concepts, insofar as they are supposed to be predicable both of imagined and of perceived objects.

Somehow, merely imagining a rose bush is an activity that needs to mobilize just the “same” anticipations and retentions as would have been involved in actually seeing one. The only difference could be, as we might put it, that in the former case those anticipations and retentions are present in a purely “problematic” mode. That is to say, they are not simply “imaginative” by virtue of being contained as material in an instance of imagining. They are “imaginative” in the way that might be involved in an actual, though purely imaginative, mode of perception of some object as well. In the latter case, one need not, in an important sense, actually “have” (not even conditionally) a certain set of anticipations regarding the perceived object. All the same, one somehow manages to perceive the object through those very anticipations that one does not actually have. Once again, this simply shows something that Kant himself no doubt failed to appreciate, namely, the need for various “forms of judgment” on the level of purely imaginative material. As I have argued, concession of the point is perfectly compatible with recognizing the need for distinctively “higher” forms of judgment, namely, those involved in the incorporation of such material into full-blown concepts and judgments.

Now the first feature that we need to attend to, regarding judgments of imagined stretches of past time, is this: that they must involve the imagining of stretches of time that stretch up to (or back from) the present. It is important to see that this is not a matter of actually judging that the stretches in question are ones that include, or even simply reach up to, the present (as when I judge abstractly that the present is included in the twentieth century). If it were, then our attempt to appeal to such imagining, as part of an account of judgments of the past, would be circular. We must rather be dealing with a strict analogue of what we have been considering under the heading of “anticipation.” The latter involves the intuitional apprehension of a present appearance as (possibly) stretching into the future. We have already seen that this is not a matter of “judgment,” in any sense that threatens circularity. What we first need to acknowledge, therefore, is a corresponding mode of intuitional awareness in which the present is non-judgmentally apprehended as the continuation of something past. But we have already seen (Chapter Three, Section IV) that this general structure is any case necessary for any sort of experience whatsoever.

At least this much is already involved whenever one experiences—and, therefore, always—a state of consciousness as having had a past. As I took pains to note earlier, this can be regarded as a type of “retention,” but it is very different from the associational phenomenon that Kant usually connects with that title. What is crucial is to recognize that this type of retention is not at all a matter of judging that one’s current state of consciousness has a past. Nor is it a matter of apprehending some past state of affairs or object as somehow related to a second object or state of affairs that one identifies as the present moment. That would be no more plausible than regarding the anticipation of possible appearances, insofar as it functions as an ingredient in the intuitive apprehension of objects, as a matter of intuitively apprehending those appearances as objects as well. Nor need we, in either case, conclude that one is therefore committed to the recognition of some mysterious kind of immediate access to truths about the future or the past. To see this, we need simply remind ourself that the “objects” in question, at least as so far considered, are no more than the intentional correlates of consciousness: though they are intuitively apprehended, there is always the possibility that they did not, and that they never will, exist apart from the apprehension of them.

I take it to be independently evident that there is such a thing as non-judgmentally apprehending present appearances as termini of stretches of time. Again, we may simply regard the phenomenon as a counterpart of non-judgmentally apprehending those same appearances as in turn (possibly) stretching into the future. In the latter case, the total (intentional) object of consciousness will be the presently given appearances, presumably conceptualized in this way or that, and also experienced—though not necessarily conceptualized—as stretching in a manifold of ways into a manifold of possible futures. In the former case, the same appearances will be—however they are also conceptualized—non-conceptually experienced as stretching from a manifold of at least possible pasts. (In either case, their standard mode of “appearing,” in those earlier or later portions of the whole, might, of course, be very different from their mode of appearing in the present. This is compatible with the whole being regarded as a single, intuitively apprehended “object,” only one part of which is in fact presently appearing.)

Once we have recognized this, then we will have no reason to refuse recognition to a closely related structure of experience. In fact, the latter seems to be just as directly evident in experience as the former. What I have in mind is only the reverse of what we have already granted, namely, a case in which the focal object is no longer present appearances, hence a presently “appearing” object, apprehended as part of a whole that stretches beyond it into the past, but a completely past stretch, intuitively apprehended in imagination—though, again, not necessarily conceptualized—as part of a larger stretch, stretching beyond it up to the present. Its apprehension as part of a stretch of time that stretches beyond it up to the present is, of course, what guarantees that we are dealing with an intuition directed, not merely toward an imagined stretch of time, but toward a stretch of time imagined as actually past. So far, there is therefore no circularity in our account of judgments concerning stretches of past time: the “apprehension” in question, though pre-conceptual, is simply what constitutes one’s (imaginative) “judgment” that the imagined stretch is a stretch imagined in the past. (But there is, of course, no guarantee that what one imagines to be in that stretch of time was actually in it. What one imaginatively places, or even takes to have been, in the year 1900 may not have been in it; one may nevertheless be imagining it as present in that very year.)

What remains is to consider how such a stretch of imagined past time is specifically conceptualizable: How do we account for its conceptualization with respect to its contents (what one imagines, or even takes, to have existed, or even merely possibly to have existed, during the imagined stretch of time), duration (a past day? year? century?), or actual distance from the present (which day, year, or century?)?

Now consider, again, the case of merely imagining a rose bush. We have so far considered that sort of case in an oversimplified way. We may now consider various sorts of cases. For example: (1) (If it is possible) The non-temporal case: I imaginatively represent a region of space that I (purely “problematically”) conceptualize as containing a rose bush. (2) The temporal but non-historical case: I imaginatively represent a stretch of time (but without necessarily conceptualizing it as of any particular duration) that I conceptualize as containing a rose bush, but without experiencing it as actually stretching back from the present, that is, without apprehending it as having been a stretch of actually past time. (3) The minimally historical case, introduced in the preceding paragraphs: I imaginatively represent a stretch of time (but without necessarily conceptualizing it as of any particular duration) that I may conceptualize as containing a rose bush, and I experience that stretch as stretching back from the present (but without necessarily conceptualizing it as at any particular distance from the present).

What I have so far argued is that nothing more is needed, for (3), than (a) the purely general ability to conceptualize, at least “problematically,” in imaginative intuition, and (b) the ability to apprehend stretches of time as stretching back from the present. (Case (2), of course, only requires the first of these elements.) The second of these features is no more mysterious than the basic structure of anticipation and retention in the first place; the first seems to involve no more than the ability to incorporate anticipations and retentions, at least in a suitably “modalized” form, into purely imaginative intuitions. On the other hand, we still have not—not even with (3)—arrived at a judgment regarding the existence of an object in the past. We have at most made room for a case of merely imagining an object as existing in the past. It should also be clear that a case of (4), actually conceptualizing the stretch in question as of some particular duration, and conceptualizing its distance back from the present as of some particular length, would still not yield (5), a judgment regarding the existence of an object in the past. At most, it would yield a case of imagining an object as existing for some particular time in the past, at a particular distance from the present.

It may seem impossible to deal with this problem in the terms that are available to us. The only alternative would seem to be that our conceptualization of an object is no longer to be constituted through the incorporation of “problematically” modalized anticipations and retentions in imaginative intuition. For the latter would not appear to amount to the affirmation of any real existence in the past. Instead, it would appear that the judgment in question needs to be constituted through the incorporation, in imaginative intuition, of what we might call genuinely assertoric anticipations and retentions. This would seem to be demanded by the fact that the object is no longer merely imagined as occupying some imaginatively apprehended past, but is actually taken to have been in that past. But then this, though apparently required, seems impossible. If the anticipations and retentions in question are not merely “problematic” in mode, then we would seem to have a case of regarding an object as present, not merely as existing at some point in the past. Surely, one cannot apprehend an object “through” anticipations that one does not in fact have. But if one does in fact have them, and they are not problematically “modalized,” then one would seem to be actually anticipating in the very manner in which one normally anticipates in regard to present reality.

This objection may appear to be telling against an attempt to deal with judgments about the past in terms of anticipations and retentions as material in imaginative intuition. But I am not convinced that it is. The difficulty seems to be that, according to the proposed account, judgments about the past would turn out to be indistinguishable from judgments about objects in the present. But the proposal does not seem to me to entail this. At most it implies that judgments about the past might involve the very same anticipations as the ones that we would ordinarily have in regard to certain present objects. But in the sense that concerns us, the anticipations in question are not themselves judgments. They are at most the material for judgments.

It may also help to remember that we are dealing only with intuitional judgments. Thus we are not dealing, for example, with everything that might come under the heading of “having a belief,” or “thinking,” about the existence of an object in the past. The latter might be construed in any number of ways, perhaps of interest in empirical psychology, but of no interest here. However the notion is construed, it is no doubt implausible to suppose that, so long as one “has the belief’ or “thinks” that some object formerly existed, one must continue to have a particular set of (unmodalized) anticipations concerning possible perceptions of that object. But the present account carries no such implication. At most, it entails that, so long as one is actually judging (that is, actually predicating in intuition), to the effect that an object formerly existed, then one must be in a particular set of anticipational states bearing on possible perceptions of that object. In any case, it is crucial to remember that, even then, the states in question are not themselves judgments. They are not even states of “consciousness,” in the full-blown Kantian sense. At most, they serve as material through which certain types of states of consciousness are constituted. In the present case, they serve as material through which a past stretch of time is imaginatively apprehended.

Strictly speaking, that an object is apprehended “through” some set of anticipations regarding the possibility of perceptions no doubt implies that one is, for the moment, indeed “anticipating” the possibilty of those very perceptions, at least given the satisfaction of certain conditions. But we need not take this to blur the distinction between judging that an object formerly existed and that it actually exists now. The distinction will not be blurred, so long as we remember that one’s state of consciousness is never totally constituted by such anticipations in the first place. We may agree that, both in judging that an object formerly existed and in judging that the same object really exists now, one must be apprehending an intuitively (imaginatively or sensorily) apprehended object through the very same set of anticipations. But there may still be a crucial difference. In the one sort of case, one’s state of consciousness may be constituted through those anticipations serving as material in the conceptualization of given appearances; in the other, it may be constituted only through their serving as material in the conceptualization of a past (and merely imagined) stretch of time. Since the anticipations are not themselves ways in which objects are judgmentally present or past, there is no danger that the judgments that contain them must really be the same.

Now we may consider a second feature of judgments about objects in the past. This is that they are conceptualized as occupying times that are at a more or less definite distance in the past. What could this involve? It must in any case involve more than apprehending imagined time as merely stretching back from the present, as in (3). For the latter, we may need no more than a very primitive mode of imaginative “retention,” as an essential ingredient in imaginative intuition whose focus is a past stretch of time. For the former, we need a full-blown mode of conceptualization, namely, the conceptualization of an imagined past as at a certain distance from the present. Once again, our only recourse would appear to be to some special mode of “anticipation.” For reasons that we have already seen, this cannot, of course, be a merely “problematic” mode of anticipation. If it were, we would not be dealing with anything that amounts to genuine affirmation regarding existing in the past. What we seem therefore to require is yet another appeal to a special mode of anticipation. We might call it: anticipation-in-retention. But to account for it, we need to capitalize on a special feature of anticipation in general.

Consider the case of ordinary anticipation, as an ingredient in perception. One anticipates certain possible extensions of a given perception, as upshots of certain conditions. Obviously, this does not mean that one anticipates those upshots as immediately forthcoming. The antecedent conditions are presumably anticipated as requiring a certain time for their own satisfaction. Therefore, ordinary anticipations must already involve, on a pre-conceptual level, an anticipation of “having to wait” some more or less determinate length of time, in order for a given experience to stretch to the point of its anticipated terminus. Now we are attempting to regard intuitive judgments about an object in the past in terms of the ingredience—in an imaginative intuition whose object is a past stretch of time—of the same, or at least the same sorts of, anticipations as would ordinarily provide the material for conceptualization of objects in ordinary perception. To do this, we may now simply make use of the facts: first, that, as already conceded, even a minimally historical case such as (3) already contains the apprehension of a phenomenal present from which an imagined stretch can be reached by stretching “back” (or of a phenomenal present which can be reached from that stretch by stretching “forward”); second, that ordinary anticipations already involve the anticipation of having to “wait” for certain upshots. Putting the points together, we have the basic materials required for a case in which some past is imaginatively apprehended (as in (3)) as stretching back from a present that in turn needs to be awaited, in more or less determinate respects, from the point of view of that past. To recognize this as possible, we need simply permit a modification of the kind of anticipation that is already involved in one’s ordinary perceptions of objects. The modification must concern the fact that the merely imaginative apprehension of an object needs to apprehend it, not simply through anticipations of the same sorts of upshots as are ordinarily anticipable in perception, but also through the anticipation of upshots whose terminus would now be in the very present.

If sense can in fact be made of this modality of anticipation—where what is “anticipated” is anticipated as yet to come in the present—then it would follow that a “phenomenalistic” account of judgments about the past would not, as often supposed, need to rest on the conceivability of a subject’s traveling back into the past from the present, in order to obtain the relevant “possible (and, under given circumstances, necessary) perceptions.” It would only need to rest on the possibility of presently anticipating (in the apprehension of an imaginatively represented past) the possibility of arriving at the present from that past, in order to obtain certain possible perceptions. By the same token, the proposed account does not presume that past objects need to have left presently perceivable traces. At most it requires the “anticipation” of their having done so. (It should be clear, in other words, that the relevant anticipation of upshots to come in the present is not the same as the judgment that those upshots really are obtainable in the present. It is at most the “anticipation” of their obtainability, and even their necessary obtainability, given certain possible ways of stretching from past to present.)

What then remains to be explained is the representation of how long one would need to wait in order to arrive at the present, from an imaginatively apprehended past. But it seems to me that we now have at least the basic materials for dealing with this question as well. The materials are provided by two facts: first, that the present is already represented in any imaginative apprehension of the past (by virtue of the purely intuitive apprehension of the latter as stretched back from the former, or of the former as stretched forward from the latter); second, that anticipation already contains, in its basic structure, the anticipation of having to wait for anticipated upshots. The two together ought to allow for the construction of a mode of anticipation through which objects would be apprehensible, in imaginative intuition, not simply as in the past, but as at certain distances from the present. Once again, the relevant mode of anticipation will have to be what I have called “anticipation-in-retention.” The rest will simply need to hinge on precisely what is anticipated in this mode. Consider, for example, the imaginative apprehension of a stretch of past time through a body of material containing, among other things, the following anticipation: the anticipation of eventually being able to apprehend the present, if one were only to wait for the earth to revolve around the sun ten thousand times. Though it may at first appear to be circular to do so, we might simply propose that apprehending a stretch of past in this way—or at least in some such way—amounts precisely to apprehending it as what we would ordinarily call “ten thousand years ago.”

The reason the suggestion may seem to be circular is that we now appear to have introduced a particular set of concepts (sun, revolution, ten thousand, etc.), in order to specify what particular anticipations are in question. What we were trying to do, by contrast, was to explain how any conceptualization of the past is originally constituted in terms of certain sorts of anticipations in the first place. But the reason that this does not introduce circularity into the account is simply that we are not now dealing with the general problem of concept-formation. The account would be circular if the original material for the formation of concepts, as predicates in intuition, must always itself be something conceptually formed. Apparently contrary to Kant, I have granted that it must, in a sense, be “judgmentally” formed. But this is not the same as saying that it must already have been formed into concepts in the relevant sense. The latter involves a specific structure of judgmental forming, correlative with the constitution of original self-consciousness.

Nothing rules out the ingredience of concepts among the material for the formation of at least certain additional concepts in intuition. Thus, for example, the concept of “having to wait ten thousand years for the present” may simply not be an example of a concept that is formed out of anticipations of the sort that we have called purely “animal” in nature. Rather, it might be formed out of anticipations that are formed with the help of additional concepts that are only in their own turn formed out of such more primitive material. Many concepts are no doubt like this. (To be sure, in his discussion of analyticity, Kant himself gives the impression that concepts are, at least in many cases, directly formed out of other concepts; they are not merely formed out of material to which other concepts contribute [A6-7/B10-11]. But it should be evident that this formulation poses a problem of interpretation in its own right.) There will be no contradiction in admitting this, so long as the concepts that contribute to the formation of anticipations, out of which other concepts are then formed, are not—in the final “analysis”— themselves formed out of anticipations of that particular kind. In other words, there must eventually be a set of concepts that are directly formed out of anticipations (and retentions) to which concepts do not contribute. That this is the case does not seem implausible. (Again, this should not be confused with the implausible claim—apparently, and unfortunately, Kant’s own—that there must be concepts that are formed out of anticipations that are totally devoid of “judgmental form.”)

In these terms, it seems to me that it may be possible to account for the conceptualization of imagined times as stretches of real history, in terms of the general account of conceptualization as the incorporation of anticipations and retentions in intuition. To do it, we simply need to appreciate the following points: (1) The general possibility of imagining stretches of time in the first place is already conceded with the introduction of time as a “form of intuition.” (2) The possibility of imagining past times as stretching from the actual present is simply a “retentional” counterpart of the basic structure of anticipation that has already been conceded, namely, of its structure as involving the non-judgmental apprehension of appearances as (possibly) stretching into the future. (3) The basic structure of anticipation, already conceded, also includes the capacity for anticipating upshots for which it is necessary to wait some length of time. (4) There is nothing incoherent in the idea that the conceptualization of past times involves their apprehension through manifolds of anticipations, appropriately comparable to those through which present appearances are apprehensible. (This yields the basic structure of “anticipation-in-retention,” and it provides the basis for the representation of particular sorts of objects as occupying past times.) (5) In the light of (1)-(4), there is nothing incoherent in the idea of apprehending past times through the anticipations of upshots, where the latter would require waiting until the present for their obtainment. Finally, (6) there is nothing incoherent in the idea of apprehending past times through the anticipation of upshots, where the latter would require a conceptually determinate length of waiting. (This provides the basis for the representation of past times as at certain distances from the present. I assume that it also contains material sufficient to account for the representation of past times as of particular lengths.)