I. Introduction

IN THE preceding chapter, I distinguished between the imaginative synthesis involved in the “anticipation and retention” of immediately upcoming and receding appearances and the kind of anticipation and retention involved in what we may more naturally call the “association” of appearances. The latter, under the title Reproduktion in Kant, may appear more readily to lend itself to Kant’s own interest in the possibility of a representation of an order of natural objects. This is because “association” is in effect what provides the material for the formation of concepts of objects in the first place. In the context of the Deduction, the importance of the other, more apparently immediate, anticipation and retention is best seen as an effort toward a more adequate notion of the relevant form of (embodiment of) anticipation and retention in concepts (or at least in “conceptions”) of objects. In any event, it is in his treatment of what we more ordinarily regard as imaginative “association” that Kant attempts to establish a direct connection with our capacity for representing an anticipable order of nature. In this chapter, I try to clarify some of the issues that are involved in this attempt.

It is clear that a relevantly Kantian connection between our capacity for association and for the representation of an anticipable order of nature needs to have some bearing on the problem of conceptual representation in particular. In this regard, we can at the very least say, for example, that a being incapable of even the more immediate anticipation and retention will be unable to apply, or even to develop, any empirical concepts. In itself, the point is fairly trivial. But what is remarkable is that, in certain passages, Kant appears to consider it a significant start toward some very strong conclusions. He does this at A100 as well as in a (supposedly) more systematic presentation of the same material at A119-23. These are the passages on which mainly I want to concentrate. In at least the first of them, Kant is often thought to be confused. I want to argue that in both cases he is attempting to present an essential but a generally neglected aspect of his theory.

It is in the second of the passages in question that Kant also distinguishes between productive and reproductive imagination and ascribes an a priori and a transcendental status to the former. Earlier, he had ascribed an a priori status to reproduction (A102); now he refuses it on the ground that this “rests on empirical conditions” (A118; cf. A121). In addition, Kant expands upon the notion of an objective order within the world of appearances, supposedly already presupposed by our merely empirical ability to associate. He refers to this order as the “affinity” of appearances (A122-3; cf. A113), and he attributes it to the transcendental capacity of imagination itself. (He also distinguishes, at A114, between a transcendental and an empirical affinity.) But later he also seems to attribute the affinity of appearances to a distinctively intellectual capacity. There are at least two ways in which he might be supposed to waver on this. He might be shifting, on the one hand, between an attribution of affinity to the faculty of imagination (whether productive or also reproductive) and an attribution of affinity to a more intellectual faculty altogether. On the other hand, Kant might be consistent in regarding affinity as the product of imagination and simply be wavering with respect to the question whether imagination, or at least a genuinely productive version of that faculty, is not itself already something intellectual. We may also be charitable if we speak of mere “wavering.” It is more common to accuse Kant of blatant self-contradiction.¹

To get clear regarding affinity, and regarding the issue of productive and reproductive imagination, it is crucial to distinguish between the question of an anticipable order in the world of appearances, as some sort of “ground” of the associability of the latter, and the question of an essential structure internal to ordinary empirical apprehension. In the light of this distinction, Kant’s ascription of a transcendental status to productive imagination may seem to foreshadow a disturbingly mysterious doctrine. Even the ascription of such a status to merely reproductive imagination might seem to be less problematic. Kant’s point at A100-102 might then simply be that it is a necessary condition of the apprehension of appearances, or at least of any conceptualized apprehension of appearances, that the apprehending subject anticipate and retain instances of other possible apprehensions as well. In this sense, even merely reproductive imagination (or at least an imagination that is not mysteriously productive) may easily count as a “transcendental” faculty. The concession would rest on a perfectly harmless notion of the transcendental. But unless Kant is guilty of self-contradiction, his subsequent claim that only productive imagination is genuinely transcendental would seem to involve a much less modest notion. Here, Kant appears to invite our tolerance, not simply of ordinary anticipation and retention, internal and perhaps essential to any experience of anticipable order, but rather of something responsible for generating such an order to begin with. The “transcendental” may now seem to be much more than a necessary condition of experience. In itself, the latter might be boringly ordinary, something of which we detect countless examples in the course of everyday experience: ordinary anticipations and retentions, for example. But how could we detect the very activities supposed to have been responsible for generating the very order of phenomenal nature itself?²

We have seen that reproductive imagination might be in one sense transcendental, or at least possessed of a transcendental dimension, while in another sense not. This is a way of accounting for Kant’s apparent contradiction. But the price to pay would be high. It requires the supposition that, in contrast to the faculty of merely reproductive imagination, Kant presumed the existence of a mysterious faculty of productive imagination, responsible for generating the manifold of appearances in the first place. Only once generated, it seems, could the reproductive faculty undertake its course of anticipating and retaining.

A second approach is possible. According to this approach, there need be no question as to whether or not, and in what sense, reproductive or productive imagination is transcendental. Rather we simply ask: In what sense is reproductive imagination itself something productive (and not simply reproductive but “transcendental’)? By now it should be clear what I mean by the suggestion. We have already seen that, fairly trivially, anticipation and retention are necessary conditions of experience, and therefore transcendental in that sense. What we need to see, very untrivially, is that they are also able fully to serve as conditions of experience only to the extent that they are incorporable as material within particular experiences, that is, within empirical intuitions. Only thereby can they be material for the acts that conceptualize appearances.

To the extent that imaginative material is ingredient in intuitions, the world of appearances is apprehended through that material. That is, it is apprehended through that material in a sense analogous to that in which appearances are apprehended through “sensations.” This is a special sense. It involves the notion of an aspect of appearances that necessarily corresponds to, and that is an intentional correlate of, the specific character of the material in question. The latter is, of course, just our ordinary anticipations and retentions. Their correlate could then be nothing other than a corresponding order apprehensible in the world of appearances. In this way, reproductive imagination is indeed productive as well. And it is productive in a sense that is much more substantial than what we have so far granted in conceding it to be “transcendental.” (At the same time, to whatever extent the suggestion may favor some brand of “phenomenalism,” it will in any case favor no supposition of the manufacture of objects out of subjective material.) The main lines of my interpretation will be laid out in the next section. In the final section, I apply them in order to clarify distinctions among several concepts of affinity, as well as some related concepts in the Critique of Judgment.

II. From Association to Affinity

We are ready to take a more detailed look at Kant’s discussion of association at A100-102. The main line of argument also reappears at A120-3, in explicit connection with the notion of productive imagination. The argument is puzzling. In it, Kant tries to show that one’s mere possession of a capacity for associative reproduction presupposes that appearances themselves (die Erscheinungen selbst) are subject to rules that correspond to one’s modes of association. It will be helpful to divide the passage into two parts:

[A] It is a merely empirical law, that representations which have often followed or accompanied one another finally become associated, and so are set in a relation whereby, even in the absence of the object, one of these representations can, in accordance with a fixed rule, bring about a transition of the mind to the other. [B] But this law of reproduction presupposes that appearances are themselves actually subject to such a rule, and that in the manifold of their representations a coexistence or sequence takes place in conformity with certain rules. (A 100)

It is not immediately clear what [B] is meant to assert. That “appearances are themselves” subject to rules might mean that phenomenal realities (real objects in space and time) are subject to rules. Or it might only mean, more weakly, that all experiences of phenomenal realities are subject to rules. In that case, perhaps the claim might simply be that reproduction “presupposes” an order among appearances themselves, in the sense that the former could only arise in consequence of regularities among experiences in the first place. This may seem confirmed by the sentence that immediately follows: “Otherwise our empirical imagination would never find opportunity for exercise appropriate to its powers, and so would remain concealed within the mind as a dead and to us unknown faculty.”

The most obvious problem with this reading concerns the strength of the alleged presupposition. If Kant is aiming at the conclusion that appearances necessarily are rule-bound, then the argument seems irrelevant. Now in [A], Kant in fact speaks of a “merely empirical” law, for which in [B] he purports to provide a ground in appearances. In addition, the second paragraph of the section begins with what appears its intended conclusion: “There must therefore be something which, as the a priori ground of a necessary synthetic unity of appearances, makes this very reproduction [selbst diese Reproduktion] possible” (A 101). So one might well conclude that Kant is seeking a ground, internal to appearances themselves, that is causally necessary for the original formation of reproductive associations. But surely, there is nothing incoherent in the idea that we develop patterns of association that are not in any way causally derived from corresponding perceptual regularities.

Furthermore, a closer look at the argument seems to make it clear that Kant in any case assumes from the start, at least for the sake of argument, that associations indeed always derive from a corresponding order in appearances. Premise [A] seems to incorporate that assumption. [B] then attempts to add, as a stronger conclusion, not (or at least not simply) that associations necessarily derive from regularities in appearances, but that appearances are themselves necessarily subject to the regularities already presumed in [A] to be in question. In other words, [A] seems already to suppose that there needs to be some regularity or other among appearances, in order that associations be possible in the first place. Much more strongly, [B] then appears to conclude that whatever regularities are in question must really be more than “regularities.” For they must be such that, given the particular facts of association, appearances must be subject to them (and not merely to some regularity or other).³

Similar difficulties arise in regard to the supposedly (cf. A98, 115) more systematic exposition at A120-3. Again, after covering the ground corresponding to a mere synthesis of “apprehension,” Kant introduces the notion of empirical reproduction. He makes it clear, once more, that the latter is already involved in the former. It is involved to the extent that the retention of passing experiences is a necessary condition even for the experience of single images (A121). The discussion of associative reproduction then proceeds to draw a distinction that was not so explicit earlier. It distinguishes two senses in which one might speak of the ground (Grund) of associative reproduction. Kant calls one of these grounds the “empirical ground” of reproduction. Somewhat confusingly, he also refers to that ground as the “association of representations” itself:

If, however, representations reproduced one another in any order, just as they happened to come together, this would not lead to any determinate connection of them, but only to accidental collocations; and so would not give rise to any cognition. Their reproduction must, therefore, conform to a rule, in accordance with which a representation connects in the imagination with some one representation in preference to another. This subjective and empirical ground of reproduction according to rules is what is called the association of representations. (A121)

Kant does seem to argue, not merely to assert, in this passage that associative reproduction needs some kind of “grounding” in an order of appearances. But at the very least, that argument seems premature. It merely presupposes what Kant is aware needs yet to be argued, namely, that objective “cognition” is necessary in the first place. Apart from this, no transcendental consideration is evident in the argument. Kant will of course argue, eventually, that it is a necessary condition of experience that we retain and associate appearances. He will also argue that it is necessary that objects exhibit an order such that our perceptions of them also, in general, exhibit a corresponding order. All of these propositions involve transcendental considerations, and from them it of course follows that it is necessary for experience that appearances contain an order that is at least sufficient for the derivation of corresponding anticipations and retentions. In any case, whether prematurely or not, no such considerations justify the assertion that, by their very nature, imaginative associations must in fact be “grounded” in that order. That appears to involve, once again, an assumption on Kant’s part regarding mere causal origins.

If there is an interesting argument concerning the notion of “grounding,” at this point, it must rather concern the notion of an “objective”—but also non-causal—ground of imaginative association. This is, of course, just Kant’s move, again, from whatever order is represented in our imaginative associations—however the latter happen to be generated or derived (but assuming that they are in fact “empirically” grounded)—to a corresponding order in the realm of objects. Here is the passage:

Now if this unity of association had not also an objective ground which makes it impossible that appearances should be apprehended by the imagination otherwise than under the condition of a possible synthetic unity of this apprehension, it would be entirely accidental that appearances should fit into a connected whole of human knowledge. For even though we should have the power of associating perceptions, it would remain entirely undetermined and accidental whether they would themselves be associable; and should they not be associable, there might exist a multitude of perceptions, and indeed an entire sensibility, in which much empirical consciousness would arise in my mind, but in a state of separation, and without belonging to a consciousness of myself. (A121-2)

This passage introduces a factor that Kant had held in abeyance in his earlier, less “systematic” discussion. There he had at most implied that some kind of associability of perceptions is a necessary condition of the employment of empirical concepts, and perhaps even of their development in the first place. After that, in his account of “recognition in a concept,” he connected the relevant notion of concept with that of a necessary unity of self-consciousness. In the later passage, he streamlines the treatment by indicating from the start that imaginative association is in some way a necessary condition for a unity of self-consciousness. But what interests us now is not this aspect of the argument. Our concern is not yet with the thesis that imaginative association is a necessary condition of self-consciousness, nor even with the thesis that the latter itself presupposes some kind of order among appearances. Our concern is rather with the claim that some such order is quite directly a necessary condition of imaginative association as such. What is the point of insisting, not simply that an orderly world is non-causally necessary for a “unity of consciousness,” but that it is—equally non-causally—necessary for the very occurrence of association as such?

It seems unlikely that Kant would consider it important to establish that imaginative associations need to “correspond” to a worldy order, merely for the sake of having an account of where those associations come from in the first place. Nor should he, at least at the present point, be concerned to establish the fact, even in order to provide a ground for the eventual “origin” of objective concepts or cognitions. (Apart from unclarity as to how either concern involves transcendental issues, even to assume the relevance of the second would be question-begging.) What I suggest is that Kant had a different notion in mind, with regard to the way in which associations need to “correspond” to a worldly order. This notion, as argued, needs to bear on the very nature of the relevant sort of association, not merely on the question either of its causal antecedents or of its logical presupposition by understanding. What reflection on that nature reveals is an important sense in which, even if the relevant order is in fact presupposed by understanding, it must be regarded as more than a correlate of the activity of understanding.⁴

If Kant did not intend something like this, then it is difficult to see why he should be so centrally concerned with the problem of merely imaginative association. If he were concerned only to establish that a necessary condition of the relevant sort of “unity of consciousness” is that appearances be subject to conceptualization, and that this in turn implies that they exhibit a regular order, there would be no reason for him to consider the nature of imaginative association as such. Naturally, if he could establish the truth of these claims, then he would be able to establish the transcendental necessity that appearances exhibit an order that “corresponds” to the order represented by imaginative association. But that would be anti-climactic with respect to the main point. And it would not, of course, explain why the argument begins precisely with a consideration of imaginative association.

On the other hand, we need to concede that Kant is in fact not merely concerned with imaginative association simpliciter. If he were, then it would indeed be difficult to establish a connection with the corresponding transcendental reflections. What we must assume is that Kant means to deal from the start with that particular form of association that is—or will at least eventually be—required as a foundation for understanding, hence for any relevant sort of unity of consciousness. In effect, in other words, we may read Kant’s arguments precisely as attempts to distinguish the two “types” of imaginative association. (That would explain why the argument begins with mere association, and also why it appears to confuse the question of association with questions that properly arise only at a later stage of reflection.)

We ought not to forget that association simpliciter is a rather primitive business. Non-human animals, for example, ought to be as capable of it as they are of most kinds of sensation. Kant, as we have already seen, concedes that they are. But just on that basis, Kant could not hope to establish any interesting conclusion concerning the world “as it appears” to such creatures. One might, of course, argue that such creatures at least live in a world that is able to provide a sufficient (causal) basis for their imaginative associations, hence in a world that itself exhibits a generally corresponding order. But that sort of argument would be possible only after the demand for universal causality had already been established. Even then, the conclusion could be no more than a plausible hypothesis. More important, unless it could be established that causal relations only obtain among appearances, not among things considered “in themselves,” there would not even be anything in such an argument to bear precisely on what concerns Kant in the first place, namely, on the question of an order in the world as it appears. For all that would have been done, we might have merely succeeded in supporting the hypothesis of an order internal to things in themselves.

Consider again Kant’s conclusion regarding the “objective ground” of association. Here is how he describes that ground: It is a “ground which makes it impossible that appearances should be apprehended by the imagination otherwise than under the condition of a possible synthetic unity of this apprehension” (emphasis added). Now we know why Kant introduces the unity of apprehension at this point. This is because he holds, and he will independently argue, that it is necessary for unity of consciousness that our conceptualizations represent a corresponding unity as obtaining among the objects (appearances) of which one is conscious. But what is striking, from the perspective of this reflection, is that Kant does not then simply conclude that, corresponding to one’s capacity for imaginative association, there is an objective “ground which makes it impossible that appearances should be apprehended by the understanding otherwise than under the condition of a possible synthetic unity of this apprehension.” As it is, Kant does not merely say that the order demanded by the relevant unity of consciousness, and to which imaginative associations at least generally correspond (and indeed, on the basis in question, why require more than a merely general correspondence?), must be an order that is intellectually apprehensible. Instead, he says that the order needs to be apprehended by the imagination.

We need, then, to distinguish two levels of reflection in Kant’s arguments. These might be seen as resting, respectively, on a more and a less concrete notion of the kind of “correspondence” that is supposed to be in question, between an order of things as represented in one’s imaginative associations and an order internal to the world of appearances itself. Both of them bear on the conclusion that some worldy order must correspond to the order that is represented in imaginative association. But only one of them rests on considerations that bear on the particular nature of the latter. The other rests on considerations more specifically concerning self-consciousness and understanding.

It is from the standpoint of the second of these levels that Kant can maintain the following:

This objective ground of all association of appearances I entitle their affinity. It is nowhere to be found save in the principle of the unity of apperception, in respect (in Ansehung) of all cognition which is to belong to me. According to this principle all appearances, without exception, must so enter the mind or be apprehended, that they conform to the unity of apperception. Without synthetic unity in their connection, this would be impossible; and such synthetic unity is itself, therefore, objectively necessary. (A 122)

But it is from the standpoint of the first level that Kant is also required to say that pure apperception is something that needs to be “added [emphasis added] to pure imagination, in order to render its function intellectual” in the first place (A124).

It is obvious, both in the first- and in the second-edition Deduction, that Kant sees the faculty of understanding as capable of functioning only through the faculty of imagination. Notoriously, he is not successful in making it clear what the connection is supposed to be. This may be what accounts for the fact that the discussions of imagination that we have so far been considering seem to conflate two entirely distinct lines of consideration. These two lines, and their uneasy mingling, are also evident in the argument at A100-102. There Kant proceeds as well to connect the question of imaginative association with that of empirical concepts. But he does not do it as explicitly as later, nor in explicit connection with the problem of unitary consciousness:

If cinnabar were sometimes red, sometimes black, sometimes light, sometimes heavy, if a man changed sometimes into this and sometimes into that animal form, if the country on the longest day were sometimes covered with fruit, sometimes with ice and snow, my empirical imagination would never find opportunity when representing red colour to bring to mind heavy cinnabar. Nor could there be an empirical synthesis of reproduction, if a certain name were sometimes given to this, sometimes to that object, or were one and the same thing named sometimes in one way, sometimes in another, independently of any rule to which appearances are themselves subject. (A100-101)

From these reflections, Kant thinks he has shown that imaginative association needs “grounding” in a “necessary synthetic unity of appearances.” At this particular point, again, it remains unclear what the necessity is supposed to involve. At most it would seem to be relative. The unity in question would be necessary for the possibility of association itself, which in turn is necessary for the employment of empirical concepts. In any case, Kant proceeds to suppose that this unity must— in its own turn—have an a priori ground. This, he finally claims, can only lie in the phenomenal nature of appearances:

What that something is we soon discover, when we reflect that appearances are not things in themselves, but are the mere play of our representations, and in the end reduce to determinations of inner sense. (A101)

There are a number of puzzles here, besides the general difficulty inherent in any attempt to move directly from imaginative association to an actual order among appearances. For one thing, Kant might appear to beg important questions in the move.⁵ For example, he might appear to beg the question whether appearances need to be sufficiently regular to allow for the development of material-object concepts in the first place. Apart from this, the point about “names” appears trivial, irrelevant, or confused. And the point about “determinations of inner sense” is obscure. I shall return to some of these points later on. In any case, what may seem a merely question-begging appeal to empirical concepts must be, once again, just a harmless matter of anticipating the further course of Kant’s own argument. What is important is not to suppose that this exhausts Kant’s point.

No doubt, in the passage in question, Kant is anticipating the transcendental necessity that we develop empirical concepts that are reasonably well connected with the regular course of experience. But this accounts for only one part of Kant’s reasoning in the passage in question. In addition, Kant also argues, on the basis of nothing more than our mere ability to associate, that the world must exhibit an order that is correlative to the associations in question. If Kant’s point is only that the world must exhibit an order correlative to our ability to conceptualize it, then he is surely guilty of severe overstatement. Imaginative associations are in some way necessary for the formation of empirical concepts. It would seem a dull joke to argue, just on that basis, that the world must therefore exhibit an order that is correlative to imaginative associations. So the question remains what the point really is, given that Kant has not yet argued for the necessity of objective concepts in the first place. Whatever it is, it has got to reflect more than so obvious a connection between the employment of empirical concepts and one’s capacity for imaginative association.⁶

I have proposed that Kant’s point is to argue that whatever sort of worldly order one may establish as transcendentally necessary—that is, establish as necessary for the very possibility of a unitary consciousness—it must not be an order that is merely a correlate of the conceptualization of experience, hence only indirectly a correlate of imaginative association. Now in a way, this may seem obvious. Obviously, whatever causal order is reflected in the conceptualization of experience cannot reflect concepts alone. For it must surely have emerged out of some sort of encounter between our general capacity for conceptualization and the particular course of experience. The latter concerns the specific detail of the sensations or intuitions available. It is presumably this that in turn is reflected in the associations that we eventually manage to develop. But that is precisely the difficulty. The difficulty is to make sense of the notion of such “encounters.” We may be tempted to adopt too simplistic an approach to the notion. In particular, we may be tempted to consider it a matter of merely external connections between sensations or intuitions as “stimuli,” on the one hand, and whatever conceptual “responses” these are able to elicit, on the other.

The farthest we may be inclined to go, in overcoming the notion of a merely external connection between intuitions and sensations, on the one hand, and our conceptual responses to them, is to rely on a vague picture of the latter as somehow “imposing” an order on the former. But what could this mean? At least part of it is, no doubt, that one’s conceptual responses manage to generate the conception of a generally orderly world, and a commitment to its reality on the whole. That world is then, in its own turn, represented—and perhaps necessarily so—as containing at least enough order to have accounted for the corresponding order of experience in the first place. But the question would still remain: To what extent would such a “representation” ever amount to any real awareness of the world as containing an order? Everything we have so far been told is compatible with holding that one is “aware” of that order in a purely conceptual sense: the only qualification being that the conception in question has been “stimulated” in a certain way.

Now I have already argued, in the first chapter, that immediate (intuitional) awareness cannot be grounded in purely external connections between sensory and conceptual elements. It cannot merely involve the capacity of one’s sensory intuitions to “stimulate” appropriate judgments or conceptions regarding reality (or would-be reality). It must rather involve the embodiment of judgments and conceptions in one’s very intuitions. In the earlier discussion, we were of course only considering the general problem of one’s intuitive awareness of “objects.” In effect, what we are now discovering is that Kant’s solution to that problem, in terms of the role of imaginative associations in intuition, at the same time provides the solution to an additional problem, namely, to the problem of one’s awareness of a whole world of objects. Kant himself does not put the issue in quite these terms. In any event, the central issue is now precisely that of accounting for the possibility of awareness, not simply of objects, but of a natural order of objects.

One may see more plausibility in the view that one’s “awareness” of an order of nature, as a set of regularities to which objects are subject, is a purely conceptual matter than in the corresponding claim regarding awareness of those objects themselves. Naturally, there is no denying that, in either case, the relevant conceptions and commitments arise in “response” to sensory input. But one may be more inclined, in one of the cases, to concede that, however they arise, what they are is only conceptions and commitments.

Few people are prepared to deny that we really and directly “see” concrete objects. Indeed, even those who hold that what we really see is something altogether different—for example, sense data, or even aspects of our own sensory states—will still be happy to concede that we see material objects. They will simply inform us that “seeing” such objects is nothing more than judging or believing or conceiving in certain ways, in response to certain sensations. But while few may be comfortable denying that we see material objects, and holding that our apprehension of them is a purely conceptual matter, it may be tempting to suppose it a purely conceptual accomplishment that we “see” an order of nature. We might simply suppose that the latter involves the fact that regularities among the objects that we see has somehow stimulated development of an appropriate “conceptual scheme.” What we are discovering, now, is how such a position is excluded by Kant’s account of our apprehension of objects in the first place.

To summarize, then, with regard to the arguments that Kant himself gives us: We need to distinguish two levels of reasoning. One of these—perhaps the most prominent—directly concerns conditions for the employment of empirical concepts. Thus it ultimately concerns conditions for an appropriate “unity of consciousness.” But it rests on no particular conception of the nature of imagination, and it rests on only the most abstract view of the latter’s relation to the faculty of understanding. On this level, the most that we could say is that the ability to form imaginative associations is a necessary condition of the ability to form concepts. Likewise, it is a condition of the latter that objects themselves exhibit an order that “corresponds,” as something objective, to mere associations. These are, of course, significant conclusions, although, at this stage in the argument, the second of them emerges prematurely. In particular, the “corresponding” order in question in the second conclusion is just the order of nature itself—regarded as a correlate of the conceptualization of nature (which Kant has, of course, not yet argued to be necessary).

The second level presupposes a more specific notion of the role of imaginative association. More precisely, it rests on a notion of the special sort of association that is (or will be eventually) required to play a role in the conceptualization of experience.⁷ What the second level of reasoning adds is this: that any imaginative associations that are to be relevant to the formation of concepts must satisfy an additional condition, beyond their merely (causally) grounded connection with intuitions and conceptions, on the one hand, and with a real order of nature, on the other, and beyond the mere (logical) fact that associations are, in general, necessary for conception. What is needed, of coure, is that specific associations actually become ingredient in the intuitions to which they are “connected.” So long as they are not, then they are not in a position to serve as material for the formation of any concepts.

The crux of Kant’s argument may then be seen as lying in the conclusion that he draws from these reflections. What it amounts to, as I have already suggested, is this: that whatever sort of world may be correlative with the understanding’s grasp of things, hence with whatever “unity of consciousness” will eventually be in question, must be a world that is perceivable in the first place precisely through a manifold of imaginative associations. To put the point in other words, the very “order of nature” must be a perceptible aspect of the world of appearances. This must be so, according to the proposed reasoning, simply because appearances themselves—at least qua candidates for conceptualization—need to be apprehended through associations in a way that is analogous to their apprehension through mere “sensations” in the first place. This of course goes well beyond the conclusion that associations are necessary for the formation of concepts, and that they must in turn be causally connected with the order of nature represented by means of those concepts.

Kant’s distinction between the objective and the empirical “grounds” of association, as well as his notion of “affinity,” might now be interpreted in accordance with this reading. The empirical ground of association is simply whatever causal relations in fact happens to connect one’s associations with a corresponding actual order of nature. As we have seen, Kant either takes it for granted, at this stage of the argument, that there must be such a grounding, or else he takes this to follow from an argument (concerning necessary conditions of conceptualization) that he could only be anticipating at this point. As for the “objective” ground of association: it is simply a non-causal—but equally non-logical—relation of intentional correlation. In particular, it is the relation of intentional correlation that “connects” the order of associations, to the extent that they are actually ingredient in perceptions themselves, with the order of nature apprehended through those perceptions. The “affinity” of appearances, finally, may be regarded as the corresponding intentional correlate. That is, it may be regarded as what corresponds, in appearances, to the associations through which those appearances are apprehended. (In the next section, we shall need to distinguish several notions of affinity.)

Apart from whatever more general help this analysis may offer, in an analysis of the arguments in question, it may also help with an odd turn of phrase at A100 (although it is one altered by Kemp Smith in translation). There Kant claims, as we saw, that the objective basis for association must lie among appearances themselves, in such a way that “in the manifold of their representations a coexistence or sequence takes place in conformity” with rules. Now were Kant merely arguing that the imaginative representation of regularities among appearances must be in some way based upon regularities detectable in appearances themselves, then why would he immediately blur the very distinction in question (namely, between the realms of imaginative representation, as such, and of appearances thereby represented) by substituting for a reference to the order inherent in the latter, a reference to the order inherent in our representations of the latter? What Kant actually says would seem more easily to fit our own reflections. After all, what we in the first place want to distinguish from the level of (mere) imaginative association is itself just a special way of associating, hence indeed a kind of “representation”: the kind that is internal to sensory intuitions. Only from here, as its intentional correlate, do we finally arrive, in the argument in question, at the actually perceptible order of appearances themselves.

This reading requires that we consider Kant’s notion of the “manifold” of appearances, and of intuitions, along the lines that I suggested in the preceding chapter. In the present context, for example, it may seem that the primary distinction is that between associations as such and regularities detectable among a manifold of appearances and intuitions. But if by empirical imagination Kant means the mere ability to “bring to mind” a manifold of items associated with a given appearance, then there is another distinction to consider. For we still need to distinguish that general ability from the more specific capacity for apprehending appearances that, in an important sense, “contain” a manifold of additional appearances from the start, that is, as a correlate of the anticipations and retentions in their apprehension. In this case, we would not be dealing with a manifold of distinct appearances or intuitions, but rather with a manifold in any particular one.

Kant himself emphasizes, in both the earlier and the later presentation, that all appearances do contain a manifold. But he may be thought to be dealing only with what I have called the more “immediate” sort of imaginative association, that is, with anticipation and retention of what is immediately upcoming or receding. It may be easier to suppose that this is so in the earlier passage. There Kant speaks of the manifold within any single intuition in connection with the “synthesis of apprehension.” He does not explicitly attribute it to imagination, but to intuition itself, and he reserves the treatment of “reproduction” for a separately numbered section. Though he then reintroduces the more immediate kind of anticipation and retention, and concedes that it, too, involves reproduction, he does not appear to suggest that, like the more immediate kind, what we more usually regard as associative reproduction involves a manifold in any given intuition, not merely relations among distinct ones.

In this respect, the later treatment seems different. Kant begins, at A120, by observing that “every [single] appearance contains a manifold,” and by attributing to the imagination that special sort of apprehension that is involved in its incorporation into a single image (Bild).⁸ He also (A121) indicates that this requires the reproduction of bygone perceptions (just as it presumably requires the anticipation of others). His language may suggest, at this point, concern with the case of merely immediate anticipation and retention. One might then expect him to do two things: to introduce the other sort of anticipation and retention as well, and to indicate that, corresponding to the latter, we represent not only “images,” but the very world of appearances as itself objectively ordered. What may at first seem amazing is that Kant instead claims that whatever it is he has already introduced already supports that notion. This can only indicate that, at least with respect to the prospect of conceptualizing appearances, even apparently external associations must be as internal to the appearances in question, or at least to their apprehension, as the parts of an image to the image itself, or the parts of an ongoing sequence to that sequence.

The point might also be seen in a later passage, on which I commented earlier. Kemp Smith translates it this way:

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 unity) acts as the rule, and the faculty of these rules is the understanding. (A127; emphasis added)

We are, of course, not concerned here with the notion of a rule as such. But consider the emphasized passage. As noted earlier, what the original says is that the unity of apperception, in respect to a manifold of representations (in Ansehung eines Mannigfaltigen von Vorstellungen), involves, or even in some sense is, the capacity es nämlich aus einer einzigen zu bestimmen: “out of’ a single representation.

The proposed reading may also make sense of a turn of phrase in the argument at A100-101. There Kant says, again, that if cinnabar did not behave in a sufficiently regular way, then the imagination would lack the “opportunity when representing red colour to bring to mind heavy cinnabar” (emphases added). One might have supposed his point to be more adequately served by observing that imagination would lack the opportunity, when perceiving a certain color—that is, a particular “appearance”—to bring to mind other appearances. Indeed, in what way does mere imagination, at least of the relatively primitive sort that must concern us at this point, even manage to bring objects to mind at all, as opposed to a manifold of possible appearances of objects, that is, a manifold that only the understanding may ultimately succeed in taking as objects (or at least as the appearances of them)? But suppose that, in Kant’s view, conceptualizing appearances as objects (for example, conceptualizing intuitively apprehended colors as cinnabar) essentially involves the elevation, to properly conceptual level, of a manifold of imaginative associations that are themselves internal to the very apprehension of those appearances. In that case, Kant could insist, the regular behavior of cinnabar must indeed have been discoverable in the color of cinnabar itself, that is, in any intuited patch of color that is conceptualizable as cinnabar. This would no longer be the obvious, and in any case merely empirical, claim that neither imaginative associations nor empirical concepts could develop in the first place, apart from regularities among appearances.

Finally, this reading allows us to account for Kant’s claim that his argument rests on conceding that appearances “are not things in themselves, but are the mere play of our representations, and in the end reduce to determinations of inner sense” (A101). According to the account that I have proposed, the affinity of appearances is, at least as we are so far concerned with it, the intentional correlate of our apprehension of appearances, insofar as the latter contains a manifold of imaginative anticipations and retentions. But this notion was itself derived from an analysis of the apprehension of appearances themselves. Thus with respect at least to this level of reflection, appearances must be regarded precisely as intentional correlates of the apprehension of them. On another level, obviously, appearances will need to be regarded as more than this. The latter occurs when we begin to consider that some appearances, although apprehended, might not be anything real. Of course, Kant thinks he can argue, although has not yet done so, that it is necessary for experience that, at least in general, one take the things that appear in experience to be real, and to have real features that occasionally conflict with what they appear to have. The distinction in question will in any case concern us in the next section.

III. Kinds of Affinity

At A113-4, Kant distinguishes between merely empirical and transcendental affinity. In the same passage, at A113, he also refers to affinity in general—thus apparently including the empirical—as “the ground of the possibility of the association of the manifold, so far as it lies in the object” (emphasis added). From this it of course follows that even merely empirical affinity is something “objective,” or at least something that “lies in the object.” It is therefore not, for example, to be equated with association itself. Neither, presumably, is it to be equated with the merely empirical ground of association. The latter seems to involve a purely causal notion, and Kant does not call it “affinity” at all. At A121, as we have seen, he explicitly distinguishes it from affinity. In any case, we now appear to face a distinction between two sorts of “affinitative” grounds of association, both of them “in the object”:

The ground of the possibility of the association of the manifold, so far as it lies in the object, is named the affinity of the manifold. I therefore ask, how are we to make comprehensible to ourselves the thoroughgoing affinity of appearances (through which they stand, and must stand, under unchanging laws)?

On my principles it is easily explicable. All possible appearances, as representations, belong to the totality of a possible self-consciousness. . . . The representation of a universal condition according to which a certain manifold can be posited in uniform fashion is called a rule, and, when it must be so posited, a law. Thus all appearances stand in thoroughgoing connection according to necessary laws, and therefore in a transcendental affinity, of which the empirical is a mere consequence. (A 113-4)

Kant is, of course, ultimately concerned with the possibility of representing a real order of nature. As should be equally clear, any such representation needs to be grounded in something more than affinity as I have so far characterized that notion. At the same time, we might appreciate why Kant puts the distinction precisely in terms of distinction between two kinds of “affinity.” According to the account proposed, a representation of affinity in objects, in the full-blown sense that must ultimately be in question, requires nothing more than a kind of enrichment of the affinity in appearances so far considered, that is, of the affinity that is a mere intentional correlate of imaginative association.

Representation of an order of nature obviously requires more than the ingredience of imaginative associations in intuition. It requires an intellectual capacity that imagination cannot provide, namely, a capacity for incorporating associations into distinct empirical concepts. Only once this has been accomplished is the correlate of imaginative apprehension, with its order of what might so far be considered mere “regularities,” elevated to the status of that lawful order that is the correlate of a truly intellectual mode of representation. We have so far abstracted from this level of representation. As we have seen, Kant himself did not always keep the levels distinct in his own thinking. But the distinction may help in the interpretation of some additional passages.

It may, for example, be precisely a transition between these two sorts of “order”—from a realm of imaginatively apprehended regularities to one of intellectually represented law—that Kant has in mind in the following formulation:

That the affinity of appearances, and with it [mit ihr] association, and through this in turn [durch diese endlich] reproduction according to laws [nach Gesetzen], and so [folglich] experience itself, should only be possible by means of this transcendental function of imagination, is indeed strange, but is none the less an obvious consequence of the preceding argument. For without this transcendental function no concepts of objects would come together into an experience. (A123)

Kant proceeds to note that something still needs to be “added to pure imagination,” namely, pure apperception, “in order to render its function intellectual” (A 124). Apart from the latter, apprehension of the affinity in question—though it involves the apprehension of something internal to appearances themselves (which are thereby represented as an sich assoziabel), and therefore involves the apprehension of something that is in its own way “objective” (A 122)—still falls short of the apprehension of a truly law-governed realm, and therefore of a truly “objective” realm.

We may also note that Kant seems to regard affinity in this passage as somehow more intimately one with association itself (mit ihr) than with the corresponding order of nature (durch diese endlich). In this case, the affinity in question must be the imaginative affinity that I discussed in the preceding section. I have argued that there is an important sense in which it constitutes an “order of nature,” even though there is another sense in which it does not. How, in any case, does it relate both to “transcendental affinity” and to “empirical affinity”? Transcendental affinity, I suggest, is just the order of nature considered precisely as an order of nature, hence, in Kant’s view, considered as a correlate of the understanding’s conceptualization of appearances. Empirical affinity is that same order. But it is that order considered as the correlate of mere “apprehension,” that is, considered as a correlate of intuition, insofar as the latter is not purely conceptual, but has ingredient in itself a set of imaginative anticipations and retentions suitable for the understanding’s conceptualization of appearances in the first place.

I have already tried to make it clear how both of these sorts, or aspects, of affinity—both that of the fully objective order of nature and that of “nature” as a correlate of imaginative apprehension—do indeed involve an order that is represented “in objects.” The only remaining puzzle is to make sense of the claim that the latter is a “consequence” of the former (A114). But that, I think, can now be simply put. The point is that the only reason for expecting imaginative associations ever in fact to become ingredient in intuitions themselves, not to remain merely mechanically or externally “associated” with them—and thus the only reason for ever considering a merely empirical affinity in the first place—is precisely in order that intuitions might be subject to conceptualization, and so precisely in order that a genuinely transcendental affinity might ultimately be in question. As Kant himself puts it, though productive imagination needs to have something “added” to it, in order to render its function intellectual, it nevertheless—or rather just for that reason—appears to aim at the very possibility of having that something added (zu ihrer Absicht hat [A123-4]; imagination considered insofar as it bloss auf die ursprünglich synthetische Einheit . . . geht [B151]).

Paton adopts a very different approach to the claim that the empirical is a “consequence” of transcendental affinity. Transcendental affinity, in his view, is the general lawfulness of nature, insofar as it is demanded by the categories. By contrast with transcendental affinity, empirical affinity is the particular instance of this lawfulness, as it is in fact to be found in nature. The former is a consequence of the latter, simply in that it is a particular instance of the latter.⁹ However, this suggestion seems odd. It seems odd in view of Kant’s insistence that, while empirical laws are “special determinations” of pure laws of the understanding, they are precisely not “derivable from them” (A127; cf B165). If they are not “derivable” from them, it seems odd to say they are “consequences” of them.

In my view, transcendental affinity is, in a sense, empirical affinity as well. It is simply the latter considered as elevated, and necessarily so, to a truly intellectual level. This proposal at least has the advantage of explaining Kant’s tendency to fail to distinguish the two. Put differently, it helps to explain his tendency to conflate the two levels of reflection that I distinguished in my earlier analysis of his arguments. Once again: Compatible with this kind of “identification” of the two affinities in question, empirical affinity may said to be a “consequence” of transcendental affinity, simply in that the only possible ground for associations ever finding their way into intuitions in the first place, as opposed to remaining merely externally attached to (or “associated” with) them, is precisely in order to render transcendental affinity possible, that is, to render possible a conceptually accessible order of nature.

With regard to the question of “grounding,” in any case, and in view of the difference between the present approach and the one that Paton proposes, it is not surprising that Paton is led, but that we need not be, to see the only possible reason or ground that Kant himself might be likely to offer, as a reason or ground for empirical affinity, as a reason or ground that arises from an independent order of things in themselves.¹⁰ If something like this is indeed the ultimate “ground” of empirical affinity, then why didn’t Kant say that empirical affinity is a consequence of it, instead of saying that it is a consequence of transcendental affinity? In fact, as Paton himself understands the notion, empirical affinity ought not to be any more a consequence of transcendental affinity than it is of an independent order of things in themselves.

Apart from the discussion of affinity in the first-edition Deduction, Kant also uses the term in different ways in other contexts. Though there is presumably a close connection among these notions, it is important to be clear how they differ. This is especially so in view of their inevitable relevance to an assessment of the first Critique in relation to the overall development of Kant’s thought. Apart from our explicit concern with the notion of affinity, it will therefore be useful to comment briefly on some of the more general issues raised by the distinction between “reflective judgment,” as it is treated in the third Critique, and the general problem of “understanding,” as Kant had already analyzed it.

Two of the additional passages that speak of “affinity” occur in connection with Kant’s treatment of the Ideas of Reason in the Transcendental Dialectic (A572/B600, A657/B685ff). Another of them occurs in the “First Introduction” to the Critique of Judgment. (By contrast with all three of these passages, the A766/B794 discussion of Hume’s confusion between affinity and association enters directly into the ambit of the argument in the Deduction as we have already examined it.)¹¹ In the “First Introduction,” the connection with the question of “purposiveness” in nature is prominent. The latter is apparently something that, for better or worse, was neglected in the analysis of the concept of nature in the first Critique. In the passages from the Dialectic, a similar point is at issue. The passages suggest a question concerning the extent to which Kant’s analysis of the faculty of understanding in the Deduction, and of its allegedly “constitutive” demands upon any experience of nature, might stand in need of correction or augmentation in terms of an autonomously “regulative” faculty of reason.

It is often claimed that, in moving from the level of mere understanding to that of transcendental reason, and then on to the principle of autonomous “judgment” in the third Critique, Kant achieves successively more adequate points of view. Gradually, it may seem, he overcomes a narrow favoritism, on behalf of the mere understanding, that typified the first Critique. Whatever one’s attitude toward this claim, and apart from the precise use of the terms in various contexts,¹² it does seem clear that there is some important connection between the third Critique’s treatment of the “purposiveness” of nature, with respect to our efforts to comprehend it, and the first Critique s treatment of the objective “affinity” of appearances. But it is important to be clear what the connection really is. It does not, so far as I can see, indicate any need on Kant’s part to withdraw from the position elaborated in the Deduction.

While it seems to me a bit overstated to claim that the concept of transcendental purposiveness is already present under the title of “transcendental affinity” in the Deduction,¹³ in a certain sense we can concede that it is so. But in an important respect, the ground of comparison has nothing to do with reflective judgment at all. It concerns an issue that arises on a level below that of judgment proper. With regard to the connection between “purposiveness” and affinity, as we have been considering the latter, the most important comparison seems to me in fact to lie only in a very specific aspect of the third Critique. What we need to attend to is Kant’s treatment of aesthetic judgment. To see the point of comparison, however, we also need to appreciate the sense in which the latter is not a kind of “judgment” at all for Kant—at least in the usual sense of that term. It is rather a unique kind of intuition. Finally, and just for that reason, we need to appreciate the extent to which the originally suggested allusion to Kant’s comments on “affinity” in the Dialectic are misleading.

There is no doubt a connection between the regulative demands of reason, as they are treated in the Dialectic, and the transcendental principle of (reflective) judgment in the third Critique. But the point is that our awareness of the former is, as such, of a purely rational kind. It is the awareness of a purely rational demand. It is not a matter of perception or intuition. Indeed, even our awareness of the purposiveness of nature, as it is explained in the third Critique, is only in one special case a matter for actual perception. It is so only in the special case of a truly aesthetic grasp of the purposiveness of nature.¹⁴ On the other hand, I do not, of course, claim that the apprehension of affinity, as it functions in the Deduction, is itself a form of aesthetic apprehension. The latter is essentially mediated by a feeling of pleasure or displeasure. No such mediation is in question in regard to the apprehension of affinity in the Transcendental Deduction. Despite this difference, we can appreciate the sense in which the issues are connected by reflecting more deeply on the nature of aesthetic apprehension.

It is tempting to regard aesthetic apprehension, as Kant conceives it, as a complex involving three factors. First of all, Kant speaks of a special relationship that obtains, at least on certain occasions, between the faculties of imagination and understanding. I shall take the liberty of putting the point in terms of the notions that I have so far developed in this chapter. In these terms, the relationship concerns the suitability of the former as provider of “material” for eventual formation into concepts by the latter. This is not a relationship that obtains in virtue of the latter’s actual formation of a concept. An aesthetic judgment, Kant says, is nicht auf Begriffe gegründet.¹⁵ The relationship in question is one that obtains simply in virtue of the fact that the corresponding material is suitable for the formation of a concept. Kant refers to this suitability as an “agreement” (Angemessenheit, Einstimmung, Proportion, Übereinstimmung, Zusammenstimmung) or “harmony” between the faculties of imagination and understanding.¹⁶ Significantly, he also refers to that suitability in terms of a corresponding harmony between the harmony of the faculties themselves and apprehended objects.¹⁷

The harmony or suitability in question is only the first of the three elements involved in any instance of aesthetic apprehension. The second is a feeling of pleasure of some kind. (I ignore the case of disharmony, hence of a corresponding displeasure.) More specifically, the pleasure is of a kind that Kant seems to describe as having been produced (bewirkt, gewirkt, zur Folge haben) precisely by the relevant harmony of the cognitive faculties.¹⁸ Finally, there is, it seems, the element of “judgment” proper. On one reading of Kant, it is through the specific office of this third element that the others in an aesthetic judgment become its elements in the first place: the aesthetic judgment is a judgment about a certain feeling of pleasure, namely, a judgment to the effect that a certain feeling of pleasure has been produced by nothing other than the harmonious working of the appropriate cognitive faculties.

On this reading, we cannot appreciate the connection between aesthetic apprehension and the apprehension of affinity. The latter, I have argued, is the apprehension of an order of things through the ingredience, in intuitions themselves, of an imaginative material that is suitable for the “application” of empirical concepts to the objects of intuitions. That same order, we have seen, can also be represented conceptually. But the point of the doctrine of affinity, as Kant treats it in the Transcendental Deduction, is to make clear that a prior (though not necessarily temporally prior) condition for the conceptual representation of any order in nature is a special kind of preconceptual representation of that same order. Clearly, this does not mean that the apprehension of affinity is simply to be equated with that of aesthetic apprehension. What is missing, at the very least, is the demand that, in aesthetic judgment, the order in question be apprehended with the right sort of pleasure. Despite that difference, we should be able to notice an important similarity.

Like properly aesthetic apprehension, the apprehension of the affinity of appearances can be regarded in two ways. It can be regarded as an apprehension of two different kinds of harmony or fittingness. First of all, as the apprehension of an order in nature, the apprehension in question must involve the awareness of some kind of “affinity,” or of some kind of fittingness, that the very elements of nature have for each other. But it might also be said to involve the awareness of a very different kind of affinity. In apprehending any such affinity among objects of perception, one may also be said to apprehend an affinity that their internally perceptible order has for us. In a sense, after all, what one is apprehending is simply an intentional correlate of imagination. That is to say, what one is apprehending is a correlate of the actual ingredience in one’s own intuition of one’s own imagination, namely, its ingredience in that special mode that is demanded, but not sufficient, for the “application” of concepts to intuitions in the first place. Thus there is a sense in which the apprehension of the order of nature in question (or at least the apprehension of what is in principle conceptualizable as an order of nature) is also the apprehension of a harmony or fittingness, not only of the things that are encompassed within nature’s own bounds, but of the order of nature itself, with respect to the working of one’s own cognitive faculties. In that sense, the apprehension of the affinity of appearances is as much an apprehension of a purposiveness in nature as truly aesthetic apprehension. The only difference would be that the purposiveness in question is not apprehended with that special pleasure that is characteristic of aesthetic apprehension.

This parallel would fail in the view of aesthetic judgment that I have so far sketched. In that view, aesthetic “apprehension” might amount to two different things, but in neither case would it be comparable to the apprehension of affinity as I have presented it. First of all, aesthetic apprehension might simply be the experience of a pleasurable sensation that is in fact caused by the harmonious interplay of one’s cognitive faculties in the apprehension of appearances. In that case, we would have two elements in relation: an apprehension of appearances and a feeling of pleasure, caused in a certain way. But there would be no sense in which one has apprehended, precisely through that pleasure, a harmonious or purposive aspect of appearances themselves. In a sense, perhaps, one may be said to have apprehended, “through” that pleasure, a certain kind of purposiveness. At least, one will have gotten a feeling that is caused by the purposive interplay of one’s own faculties. But the fact would remain that the object of that feeling—if we may speak of an “object” in a case that is constituted by purely causal relations—is not an aspect of appearances. It is just something internal to oneself. This seems to run counter to Kant’s own view of aesthetic apprehension.¹⁹

The second alternative fares no better. According to the approach in question, the second thing that we might mean by aesthetic apprehension is just aesthetic “judgment” proper. But aesthetic judgment would then be importantly distinct from the element of apprehension “with pleasure.” The latter is just a feeling of pleasure caused in a certain way. The former is a judgment to the effect that the feeling is in fact so caused. No such judgment (being purely intellectual)—any more than the feeling itself (regarded as mere effect)—is comparable to a genuine apprehension of some order in the world of appearances. That is why, according to the approach in question, aesthetic apprehension would not be comparable to the apprehension of affinity as I have construed it. So far as I understand it, it is also why that approach is inadequate to Kant’s own view of aesthetic apprehension.

I have argued in detail elsewhere against this approach to Kant’s account of aesthetic judgment.²⁰ I will indicate the main considerations here, and then draw the connection with our problem concerning affinity. The first consideration is that Kant aims at a number of identifications of elements that the account in question is constrained to regard as distinct. First of all, Kant persistently tends to speak as if an aesthetic judgment is not at all distinct from the feeling of pleasure that, according to the account in question, that judgment is supposedly about. More specifically, though it may seem oddly, Kant appears to regard feelings of aesthetic pleasure as if they themselves served as the very predicates of aesthetic judgments.²¹ That is the first point. The second point is that Kant seems to speak, not simply of an aesthetic pleasure as if it were itself the predicate through which an object of aesthetic appreciation is apprehended in the first place, but also as if that pleasure were not distinct from the very harmony of the faculties that is supposed, according to the account I am criticizing, to be its cause.²²

It is not absurd to suppose that cognitive activity and the establishment of appropriate relations among cognitive factors are not merely capable of generating pleasurable sensations, but are themselves pleasurable “sensations,” at least on certain occasions. Kant himself describes pleasure in general (or at least any pleasure that one takes in some representation) in a way that seems to regard it as indistinguishable from the representation in which that pleasure is taken:

Pleasure is a state of the mind in which a representation is in harmony with itself as ground [mit sich selbst zusammenstimmt als Grund], either in order merely to maintain that ground [diesen bloss selbst zu erhalten] (because the state of mutually supporting mental powers in a representation maintains itself), or in order to produce its object. In the first case, the judgment regarding the given representation is an aesthetic judgment of reflection.²³

The approach that I am proposing is suited to recognize this point as well as the others. This is because it is based on an analogy between the role that sensations play in intuitions and the role of imagination, at least insofar as the latter is relevant to (hence in “harmony” with) the faculty of understanding, hence relevant to the possibility of intuitions being subject to conceptualization in the first place. This is not to say that, according to my approach, the ingredience of imaginative material in an intuition, in a way that is favorable to the eventual conceptualization of the latter, is ipso facto a pleasure of any sort. It is at most to say that the ingredience of imaginative material in an intuition is the sort of thing that could be a pleasure on some occasion. In general, and quite apart from the case of specifically aesthetic pleasure, the sensations ingredient in an intuition might or might not be pleasurable. But when they are, there is no reason to suppose that their pleasurable feeling is, strictly speaking, a distinct state from those very sensations themselves. It simply is those sensations, now become pleasures. The same, we must now conclude, goes for the purely imaginative material in an intuition.

I have already argued for important analogies between the imaginative material in an intuition and what one is more usually inclined to regard as the “sensations” in them. I now suggest that we continue to take the analogy seriously. I suggest that we suppose that, at least on certain occasions, the imaginative material in an intuition is so well suited to the application of a concept to that intuition—quite apart from the question as to what concepts, if any, are in fact applied to it—that the material in question becomes an actual pleasure in the intuition. When it does, then it comes to serve as a kind of “predicate” in intuition. In the preceding section I argued, in effect, that the ingredience of mere imaginative material in an intuition already serves as a kind of (pre-conceptual) “predicate” in its own right. It does so simply in the sense that the appearances that are apprehended through that material are necessarily apprehended as characterized in a certain way; they are necessarily apprehended as part of an order of nature (or of what is in principle conceptualizable as nature). That aesthetic pleasures are able to function as predicates in intuition may thus now be seen as no more than an extension of Kant’s account of affinity in the Deduction.

Now let us return to the more general problem of the relationship between the first and the third Critique. This concerns us, beyond the specific problem of aesthetic judgment, with the general problem of the relation between “reflective” judgment, in the third Critique, and the kind of purely “determining” judgment in regard to which the analysis of judgment was pursued in the Deduction of the first. As I have already indicated, it should not be surprising to find a connection, quite apart from the problem of aesthetic judgment, between Kant’s concern with affinity in the first and with the problem of reflective judgment in the third Critique. For the latter is simply the problem of “getting” from intuitions, and whatever they contain, to the actual formation of empirical concepts applicable to those intuitions. That, it seems, was nothing other than the problem of the Transcendental Deduction as well. But it is important to appreciate the extent to which, though related, the problems confronted in the two works are also distinct.

The transcendental principle of reflective judgment concerns the guidelines that need to be followed in the formation of concepts that are able to bear on the actual course of experience as we find it. In this regard, Kant argues that we need to postulate a systematic whole, within which the various laws of nature will form, among themselves, an intelligible system.²⁴ It appears to be precisely the intelligible character of this holistic system that Kant then refers to, in the “First Introduction” to the Critique of Judgment, as the “affinity of particular laws under more general ones.”²⁵ Now in the first Critique, Kant had of course argued that there need to be laws of nature, and that we must regard every single event as occurring in accordance with those laws, even if we have no idea as to what they actually are. In addition to this, the idea of systematic interrelation also arose in the first Critique. But it seems primarily to have been the idea of a systematic interrelation of the various parts of nature, as it is treated in the Transcendental Dialectic of that work. What is new, or at least more prominent, in the third Critique is the idea of a systematic interrelation of the various laws of nature themselves.²⁶ In particular, Kant now seems to be maintaining that a necessary condition for any experience of nature at all—that is, a condition that is every bit as necessary as the causal principle of the first Critique—is the condition that nature be represented, not simply as conforming to laws, but also as conforming to whatever more systematic demands we need to make in endeavoring to discover what those laws really are.

This new demand may seem not merely to enlarge the problematic of the first Critique. It may seem to threaten one of its central claims, namely, that the faculty of understanding is alone the source of “constitutive” demands regarding the experience of nature. But to resolve the difficulty, we simply need to distinguish two “constitutive” problems. The problem of the Transcendental Deduction, I have argued, is that of constituting concepts, just as such, out of the sort of material that is independently available, apart from that actual constitution, among the internal ingredients in intuitions themselves. Now we have considered only one side of this problem. We have not been concerned with the particular factors that need to be involved precisely in elevating that sort of material to the status of genuinely conceptual representation. We shall see, in Chapter Six, that Kant regards this question as one that concerns the structure of a particular mode of “consciousness.” In virtue of the consciousness in question, one’s imaginative anticipations and retentions will be in an appropriate way transformed, or at least appropriately “structured” in consciousness. At least one part of this structuring will entail that they are now to be regarded as anticipations and retentions of certain possible courses of experience as necessary courses of experience (relative to sets of antecedent conditions, themselves represented as at least possible). This much, I take it, is already meant to be established in the first Critique, on the basis of an analysis of the faculty of understanding as such.

Now this is obviously very closely related to what Kant regards as a necessary function of “reflective judgment” in the third Critique. The connection, as I see it, is this. It is that the basic functions of understanding that Kant regards as essential to the constitution of conceptual consciousness as such suffice only to accomplish one part of what is necessary for genuine knowledge regarding the objects thereby conceptualized.²⁷ This is the accomplishment that is reflected in the earlier work’s argument that the conceptualization of particular intuitions necessarily involves the representation of certain courses of experience as necessary ones (relative to sets of conditions that are anticipated and retained as themselves at least possible). Obviously, whatever “representation” of necessities are sufficient for the conceptualization of an intuition, as such, will not suffice for a genuine assessment of the cognitive value of that act of conceptualization. It could not possibly do so, to the extent that it abstracts from any question as to what the particular laws might be, with respect to which represented necessities obtain. Rightly or wrongly (I do not propose to discuss it), it is specifically the latter problem that is held, in the third Critique, to require a transcendental principle that had not previously been discovered, and to require holistic considerations so far ignored.

We may concede that what remains, as a problem for “reflective judgment,” involves a principle that is as much “constitutively” necessary as that of understanding as such, as considered in the Deduction. But it does not follow that Kant had been mistaken in supposing that the powers of understanding, as considered in the Deduction, were sufficient for the problem there at hand. Whether we say that its powers are “constitutive” or not, it should be clear that the principle of reflective judgment aims at the “constitution” of something in addition to what had been of primary concern earlier. As I have suggested, and despite the impression that Kant himself may give in some formulations, we can only presume that the later principle does not aim merely at constituting concepts as such, but rather at constituting instances of concrete empirical knowledge.²⁸ For the latter, as we have noted, we need to go beyond the mere representation of certain anticipable and retainable courses of experience as necessary. We need, in addition, to arrive at a conception of the particular laws by virtue of which those courses of experience may be necessary. This in turn requires more than the imaginative representation, internally to the conceptualization of an appearance in intuition, of further appearances whose intuitions are in their own turn obtainable, and necessarily so, from the given one (or—emphasizing “retention”—from which the given one was itself obtainable). In addition, it is arguable, it requires a more or less specific conceptualization of those further appearances as objects in relation to the given one. If it did not involve the conceptualization of additional objects, as objects in relation to the given one, and not merely the imaginative representation of further “appearances,” then the original act of conceptualization could not have contained any real understanding of whatever sort of “necessity” might in fact have been represented in the first place. This in turn, it is arguable, involves holistic considerations that were ignored in the first Critique.

Whether or not Kant was justified in demanding a transcendental supplement to the work of understanding to account for them, these considerations may at least help us get clear concerning the place of the systematic and holistic elements introduced both in the Transcendental Dialectic and in the Critique of Judgment. They should make it clear why we should not have expected to find that supplement earlier, and why its absence indicates no defect in Kant’s procedure. Certainly, we can see why Kant should feel that the problems involved in any attempt at achieving a conceptual representation of the whole of nature, at least in terms of the representation of a system of natural laws, are problems that arise only in regard to the problem of knowledge as such. The first Critique was not unconcerned with that problem. But it is clear that it did not attempt to uncover more than certain of the necessary conditions of knowledge. In particular, it was primarily concerned to uncover, in the Transcendental Deduction, only the conditions sufficient for the conceptualization of intuitions. To put the point another way, the Deduction was concerned only with the nature of a concept, and with the general concept of a concept’s “application” to intuition. The problem of reflective judgment arises in the context of a different problem.

To sum up: We have, in effect, discovered that the notion of “reflective judgment” is ambiguous. In one sense, it is already a notion that was in question in the first Critique, but not under that title. (It was in question precisely in the problem of “getting” from the mere material for the formation of concepts to their actual formation.) In another sense, it raises a problem neglected in the first Critique, but it offers a solution that is in no way incompatible with the theory presented earlier. In any case, looking at the matter in the terms that I have been proposing in this chapter, we may conclude that the transcendental principle of reflective judgment in no way indicates a diminishment of the uniquely “constitutive” powers of understanding. We should also be clear that what Kant specifically calls the lawful “affinity” of nature, in the “First Introduction” to the third Critique, is not quite the same thing as the kind of “affinity” already considered in the Transcendental Deduction. The latter involves an “order of nature” whose representability rests on the faculty of imagination, and it is at most a necessary condition of the conceptualization of objects. The former, though resting on the latter, involves a representation of a purely conceptual sort. For this reason, finally, while Kant’s additional comments on affinity in the Dialectic of Pure Reason, and even his treatment of the Ideas of Reason generally, may well be inseparable from his later development of the problem of reflective judgment, they too are oriented toward a problem that is legitimately distinguishable from the problem that concerns us.