2.1 The Power of Judgement in the Critique of Pure Reason

According to Kant's presentation in the Analytic of Principles of the Critique of Pure Reason, the cognitive role of the power of judgement is to apply the abstract rules supplied by our faculty of concepts (the understanding) ‘in concreto’ (Kant Reference Kant1998: 269; A134/B173). First, where the abstract rules of the understanding are a priori concepts (as in the case of the categories or of mathematical concepts), a set of necessary and sufficient conditions for the application of those concepts is also given a priori (see A67/B92). This set of conditions, accordingly, constitutes a set of rules for the application of rules. In the case of transcendental logic, these are the ‘sensible conditions under which pure concepts of the understanding can alone be used’ (Kant Reference Kant1998: 270; A136/B175). Kant presents these rules as well as the judgements that ‘derive a priori under these conditions’ (ibid.) in the first Critique's chapters on schematism and on the axioms of the understanding, respectively. When the power of judgement thus stands ‘under universal transcendental laws, given by the understanding … the law is sketched out for it a priori and it is therefore unnecessary for it to think of a law for itself in order to be able to subordinate the particular in nature to the universal’ (Kant Reference Kant2000: 67; KU 5: 179; my emphasis). In its a priori use, then, the power of judgement has no principle of its own.

Second, where the abstract concepts of the understanding are empirical concepts, no principles governing their use can, moreover, be given at all. This is so, because there can be no a priori rules by which to judge whether an object falls under a given empirical concept in concreto – and because a demand for empirical rules by which to judge whether an object falls under a given empirical concept would lead to an evident regress of rules (A133/172B; KU 5: 169). Accordingly, Kant declares the capacity for applying empirical concepts to objects ‘a special talent, which cannot be learned, but only practiced’ (Kant Reference Kant1998: 268; A133/B172). In its empirical use, then, the power of judgement has no principle of its own, either.

The power of judgement thus operates either with a borrowed principle, or with no principle at all – but never with a transcendental principle of its own. Accordingly, the power of judgement plays a peculiar role in Kant's system of transcendental logic in the first Critique. Lacking the requisite principled grounding, it cannot be considered a proper transcendental-logical analogue to our empirical-psychological capacity to judge. Consequently, transcendental logic does not run entirely parallel to general logic (A131/B170), whose division into an analytic of ‘concepts, judgments, and inferences’ (Kant Reference Kant1998: 267; A130/B169) neatly coincides with the division of our cognitive psychology into ‘understanding, power of judgment, and reason’ (ibid.). By contrast, Kant has little choice but to assign the purported ‘transcendental power of judgment’ (Kant Reference Kant1998: 268; A132/B171, caption) to the transcendental use of the understanding: ‘We can, however, trace all actions of the understanding back to judgments, so that the understanding in general can be represented as a faculty for judging’ (Kant Reference Kant1998: 205; A69/B94; my emphases).

2.2 The Determining Power of Judgement in the Critique of the Power of Judgement

As its title indicates, by the time Kant writes the Critique of the Power of Judgement, his assessment of the prospects for a Critical justification of the power of judgement has undergone a significant transformation. Kant now characterizes ‘the power of judgment as an a priori legislative faculty’ (Kant Reference Kant2000: 66; KU 5: 179, caption) and proposes a genuinely ‘transcendental principle’ (Kant Reference Kant2000: 68; KU 5: 181, caption) for it.Footnote 2 Two factors help explain how this change in judgement's transcendental fortunes comes about.

First, Kant now discerns a new cognitive capacity within the power of judgement at large. He begins by explaining that the power of judgement ‘overall’ (überhaupt; Kant Reference Kant2000: 66; KU 5: 179) is the capacity to establish and endorse subsumption-relations between universals and particulars. It is the ‘capacity of thinking the particular as contained under the universal’ (ibid.). He then distinguishes two ways in which this can be accomplished: either by descending from the universal to the particular, or by ascending from the particular to the universal. Kant explains that judgement consists in the former, analytical (KU 5: 407) descent to particularity, ‘[i]f the universal (the rule, the principle, the law) is given’ (ibid.). And he notes that, where the universal is thus given, ‘the power of judgment, which subsumes the particular under it … is determining’ (ibid.). Kant here has in mind predicative judgement or the application of extant (empirical or pure) concepts to sensible representations of objects. In short, determining judgement in the third Critique is what Kant had called ‘judgement’ simpliciter in the first Critique. By contrast, Kant explains that judgement consists in the latter, synthetic ascent to universality, ‘[i]f, however, only the particular is given, for which the universal is to be found’ (ibid.). And he notes that, where the universal is thus yet to be found or formed, ‘the power of judgment is merely reflecting’ (ibid.). This second way of ‘thinking the particular as contained under the universal’ is more perplexing than the first, principally because, as presented, the notion of an ‘ascent to universality’ is ambiguous between an empirical-psychological reading and a transcendental-logical one (see 3.2, below). As a preliminary guide to Kant's distinction, however, we may simply say that, while determining judgement is associated with the application of concepts, reflecting judgement appears to be associated with the formation of concepts. It is the power of judgement in the latter, reflecting employment, for which Kant proposes a transcendental principle in the third Critique (KU 5: 184, 186).

Second, while this turn to reflection does not, of itself, stop the threat of a regress (prima facie, rules for the formation of rules threaten no less of a regress than rules for the application of rules), the proposed principle for the reflecting power of judgement is, moreover, of a unique sort. It is not an objective rule for judging whether a given concept is the appropriate concept to be formed under given circumstances (KU 5: 169), but a subjective rule that guides the reflecting power of judgement in the formation of concepts, no matter the circumstances. The principle's distinctive characteristic is its procedural nature as a judgement-determining principle. As a second-order methodological principle with no ontological import of its own, it threatens no regress of first-order rules.

Given that at least one dimension of the cognitive role of the power of judgement (namely, reflecting judgement) is thus supposedly backed by transcendental principle, and given that the other dimension (namely, determining judgement) was previously transcendental-logically adrift, it is perhaps natural to regard the power of judgement as now happily Critically grounded in toto.

This, moreover, seems consistent with Kant's newfound confidence in the overall shape of his transcendental-logical project. No longer does he think that transcendental logic is lacking in its fidelity to an empirical psychological inventory of our cognitive capacities. First, Kant now claims that the title of the Critique of Pure Reason was essentially a misnomer: ‘it was strictly speaking the understanding … which was to be established in secure and unique possession [of its a priori concepts] against all other competitors in the critique of pure reason, generally so called’ (Kant Reference Kant2000: 56; KU 5: 168; Kant's emphasis). The ‘so called’ Critique of Pure Reason, accordingly, is really a Critique of the understanding and its principles (i.e. the categories), only (for criticism, see Brandt Reference Brandt1989: 183). Second, following his work on practical reason in the Groundwork for the Metaphysics of Morals and the Critique of Practical Reason, Kant is no longer bound (as he was in the first Critique) by the circumstance that a genuinely transcendental (i.e. constitutive) use of theoretical reason must be dialectical (and that only theoretical reason's apodictic or hypothetical use can be legitimate, A131/B170). Kant can now consider reason a ‘faculty of cognition’ (Kant Reference Kant2000: 83; KU 5: 198) in its own right, with a transcendental principle and a ‘domain’ (freedom or ‘the practical’) within which it is duly legislative (KU 5: 168). Add to this the proposed Critical grounding of the power of judgement, and Kant can claim to have arrived – in perfect architectonic simplicity – at a Critique of (the faculty of) concepts, a Critique of (the faculty of) judgement, and a Critique of (the faculty of) reason (KU 5: 198).

Yet, even in this somewhat airbrushed picture, the determining power of judgement cuts an awkward figure. For Kant is perfectly explicit that its status remains unchanged in the wake of the introduction and Critical grounding of a reflecting power of judgement. First, in its empirical use, the determining power of judgement necessarily remains as principle-less as it ever was (KU 5: 169). Second, in its transcendental use, the determining power of judgement continues to be heteronomous (ibid.). The determining power of judgement therefore cannot be subject to transcendental critique.

That this is indeed Kant's position is evident, moreover, from his continued association of determining judgement with the use of the understanding. In his extended discussion of the nature of our understanding in § § 76–7, Kant repeatedly explains that the determining use of the power of judgement – namely, the analytic descent from the universal to the particular – is, in fact, an operation of the understanding. Kant's discussion, in these sections, stands in the service of exposing a certain shortfall of the understanding. But along with Kant's diagnosis that the understanding, unaided by a transcendentally grounded reflecting power of judgement, cannot do its job of determining the particular (section 4 below), it becomes fully evident just what the indigenous job of the understanding is supposed to be – namely, determining the particular. Determining judgement remains firmly identified with an exercise of the faculty of concepts: ‘Our understanding, namely, has the property that in its cognition, e.g. of the cause of a product, it must go from the analytical universal (from concepts) to the particular (of the given empirical intuition)’ (Kant Reference Kant2000: 276; KU 5: 407; my emphasis; see 4.2, below).

Incidentally, Kant's identification of determination as the necessary task of the understanding – and not of the transcendental power of judgement – cannot be explained away by suggesting that Kant here takes the understanding ‘in a wider sense’ (Kant Reference Kant2000: 25; EE 20: 222), as referring to the ‘upper cognitive faculty in general’ (Kant Reference Kant2000: 26; EE 20: 223). To be sure, taken in this wider sense, the notion of the understanding would include the transcendental power of judgement. Kant could then say that determining judgement is both an exercise of the understanding and of the power of judgement, without openly contradicting his division of the upper cognitive faculties into understanding, power of judgement, and reason (KU 5: 198). Yet, Kant's discussion of the ‘property’ (Kant Reference Kant2000: 276; KU 5: 407) of our understanding in § § 76–7 – or of its ‘peculiarity’ (Kant Reference Kant2000: 276; KU 5: 406), or of the ‘sort’ (Kant Reference Kant2000: 274; KU 5: 404) of understanding it is – pointedly contrasts the understanding's analytic descent to particularity with the principle of the reflecting power of judgement which governs our mind's synthetic ascent to universality (KU 5: 407–8). Accordingly, if Kant, in § § 76–7, took the notion of the understanding in a wider sense, this would mean either that the power of judgement does not belong among our ‘upper cognitive faculties’ after all or that, so far as it does belong, its judgement must be an analytic descent from the universal to the particular. Either of these consequences would be detrimental to Kant's project in the third Critique.

Since determining judgement is, thus, an exercise of the understanding (narrowly construed as the faculty of concepts) which is not amenable to Critical treatment, it follows that there can be no such thing as a Critique of the (Reflecting and Determining) Power of Judgement or a Critique of the Power of Judgement (Overall). While it makes sense to speak of a ‘power of judgment overall’ (Kant Reference Kant2000: 66; KU 5: 179) from the perspective of an empirical-psychological taxonomy of cognitive capacities, the idea makes no sense from a transcendental-logical perspective – since the determining power of judgement cannot have genuine transcendental-logical status. This, to be sure, does not settle whether the reflecting power of judgement, for its part, has genuine transcendental-logical status – but it does mean that, if there is to be a Critique of the Power of Judgement at all, then it can, at most, be a Critique of the (Reflecting) Power of Judgement.

3.1 A ‘Moment’ of Reflection

The main proponents of the view that the reflecting power of judgement is closely tied to the determining power of judgement are Paul Guyer (Reference Guyer2000: p. xlvii; Reference Guyer2005: 11–13), Béatrice Longuenesse (Reference Longuenesse1998: 163–6, 195–7), and Henry Allison (Reference Allison2001: chs 1–2). The most straightforward version of the idea is Guyer's. On his tolerant account, not all judgements are both determining and reflecting. Judgements in which ‘only two terms are involved’ (e.g. demonstrative judgements in which a universal, such as ‘— is white’, is directly related to an empirical intuition) are ‘either determinant or reflective but not both’ (Guyer Reference Guyer2005: 12). Which one it is depends on whether the universal in question is antecedently given or has yet to be found. In more complex cases, however, where an antecedently given universal (e.g. ‘causation’) can only be applied to a sensible particular through ‘intermediate concepts’ (e.g. specific causal laws) that ‘have to be discovered’ (ibid.), Guyer believes that ‘reflective judgment may be needed to find those concepts and thus complete the task assigned to determinant judgment’ (ibid.). Determining judgement, accordingly, is both determining and reflecting, whenever ‘intermediate’ concepts, required for a given determination, have yet to be formed.

Longuenesse's version of the idea is more sophisticated and has broader application. According to her account, reflection – in the form of imaginative syntheses governed by Kant's amphibolous ‘concepts of reflection’ – plays an indispensable role in the generation of the logical form (as well as judgeable content) of empirical judgements. On this view – principally based on Longuenesse's reading of the Critique of Pure Reason, and developed, in the context of the third Critique, by Henry Allison (see below, 3.2) – all empirical judgement is both determining and reflecting (Longuenesse Reference Longuenesse1998: ch. 6; Allison Reference Allison2001: ch. 1).

Fortunately, it is not necessary, for present purposes, to analyse either of these accounts, in great detail. That there is a close relation between reflection and determination in judgement may, instead, simply be taken for granted. Of interest, rather, is whether the type of reflection that enters into this relation is indeed the type of reflection operative, also, in Kant's third Critique. The question is important, since the unscrutinized assumption that these forms of reflection are of the same type entails the dominant view that reflection, in the third Critique, cannot be sui generis.

Little mystery attaches to the nature of the ‘moment’ of reflection supposedly involved in all (or some) empirical judgements. Longuenesse sets the tone by identifying that ‘moment’ with the mental processes involved in the formation of empirical concepts Kant describes in §6 of the Jäsche Logik. The formation of an empirical concept, according to Kant's explanation there, is a complex empirical-psychological process, comprising three distinct mental acts: (i) an act of ‘comparison’ (namely, surveying a range of presently – or previously – given empirical objects); (ii) an act of ‘reflection’ (namely, noting similarities among those objects); (iii) an act of ‘abstraction’ (namely, disregarding dissimilarities among those objects). At the heart of Longuenesse's account of the intimate relation between reflection and determination lies the view that the ‘reflective aspect’ present in all empirical judgement is this threefold ‘progress from sensible representations to discursive thought: the formation of concepts through comparison/reflection/abstraction’ (Longuenesse Reference Longuenesse1998: 164).

The guiding question of the present section can now be formulated, with greater precision, as follows. Granted that all empirical judgement does involve an operation of empirical-psychological, concept-forming reflection: is this notion of a ‘progress from sensible representations to discursive thought’ (based on the Jäsche Logik) relevantly similar to Kant's notion of an ‘ascent from the particular to the universal’ in his account of reflection in the third Critique?

To ask the question appears to be to answer it. Any suggestion that Longuenesse's ‘progress’ may not in fact be Kant's ‘ascent’ seems ripe for rhetorical dismissal, or worse. Longuenesse asks ‘How could one fail to recognize, in this presentation of the activity of judgment, the features of reflective judgment later described by Kant in the first introduction to the Critique of Judgment’ (Longuenesse Reference Longuenesse1998: 163). Allison, coming up hard against the rhetorical paucity of Kant's presentation of reflecting judgement in the introduction to the third Critique, strengthens the proposed link to Kant's lectures on logic by declaring flatly that ‘[i]n order to understand the mechanics of this type of reflection it is necessary to turn from the third Critique to the Jäsche Logik’ (Allison Reference Allison2001: 21). Guyer, although he does not mention the Jäsche Logik, apparently has Kant's account of empirical-psychological, concept-forming reflection in mind as well, when he insists that, according to Kant's introduction to the third Critique, the job of reflecting judgement is ‘to find [intermediate] concepts’ (Guyer Reference Guyer2005: 12). After all, concepts do not grow on trees and ‘finding’ them, for Kant, does involve comparison, reflection and abstraction.

Yet Kant's account of reflection in the third Critique certainly does not entail that the designated job of reflection is to ‘find’ concepts. This view is based on the unwarranted assumption that by ‘the particular’, from which the reflecting power of judgement is tasked to ascend to the universal, Kant means spatio-temporally and categorially synthesized intuitions. This, however, is not what Kant means by ‘the particular’ in the third Critique – certainly not in the context of his discussion of the transcendental-logical significance of reflecting judgement's ascent from the particular to the universal (see section 4, below).

But even setting aside these considerations, which will soon take centre stage, following Longuenesse's lead gives reason for pause. For, both in the main text of §6 of the Jäsche Logik and in its associated ‘Remark’, we are reminded that the ‘Logical Acts of Comparison, Reflection and Abstraction’ – which are supposed to help us understand the cognitive operations of the power of judgement in the third Critique – are in fact ‘three logical operations of the understanding’ (Kant Reference Kant1992: 592; JL 9: 94; my emphasis). Specifically, they are logical operations of the understanding, narrowly construed as the faculty of concepts. After all, the Jäsche Logik is Kant's treatment of general logic, and the first section of its Doctrine of Elements, the analytic of concepts (in which Kant discusses the logical nature and empirical-psychological formation of universals), is fully distinct from the subsequent analytic of judgement and the analytic of inferences.

This suggests two things. First, if reflecting judgement in the third Critique is interpreted as associated with an ‘operation of the understanding’, then, given that determining judgement in the third Critique, moreover, is an operation of the understanding (see 2.2, above), it is hardly surprising that commentators take reflecting and determining judgement in the third Critique to be mere aspects of the same underlying capacity to judge (Guyer Reference Guyer2000: p. xlvii). Yet, the interpretative automatism of explaining Kant's notion of reflection in the third Critique in terms of Kant's notion of reflection in the Jäsche Logik, evidently begs the question against Kant's ‘intent’ in the third Critique to provide a theory of reflection that is as fundamentally distinct from determination – namely, separated by transcendental principle – as things get in Critical philosophy. If reflection in the third Critique were indeed identical to reflection in the Jäsche Logik, and if reflection in the Jäsche Logik, qua ‘operation of the understanding’, did indeed neither admit of nor require principled governance (let alone transcendentally grounded principled governance), then Kant's intent to present a Critique of the (Reflecting) Power of Judgement would not merely be ‘somewhat misleading’ (Allison Reference Allison2001: 18) but fully disingenuous.

Second, on the view that reflection in the third Critique is closely related to reflection in the Jäsche Logik (hence, closely related to the ‘moment’ of reflection detected in all empirical judgement), the alternative must then be that Kant realized that the concept-forming ‘operation of the understanding’ discussed in the Jäsche Logik requires Critical grounding after all. Allison himself takes this route. This, to be sure, turns the explanatory dependence Allison proposed between Jäsche Logik and third Critique on its head. More importantly, when followed through to its natural conclusion (see section 4, below), this reversal of explanatory strategy issues in a decidedly negative answer to the question pursued here: the type of concept-forming reflection at issue in the Jäsche Logik does not stand in need of Critical grounding in the third Critique (or elsewhere), and the type of reflection that does stand in need of Critical grounding in the third Critique is not the type of reflection at issue in the Jäsche Logik.

3.2 The Particular as Such

Allison gives two explanations why Kant should have come to believe that empirical-psychological acts of concept formation, like the ones described in the Jäsche Logik, require transcendental grounding after all. His first explanation is that Kant sought to ward off a threat posed by the contingency of the empirical order determined in our concept-forming acts. The threat is that, even with the transcendental machinery of the Critique of Pure Reason firmly in place, empirical regularities among spatio-temporally and categorially determined objects could in theory (‘as far as one can judge a priori’; Kant Reference Kant2000: 70; KU 5: 183) be so highly localized or fleeting that no unified theoretical cognition (hence, depending on how pervasive the disunity, no experience) of nature would be possible. That is to say, since we cannot deduce empirical from transcendental order, it follows that the circumstance that transcendentally justified, top–down categorial judgements structure all of experience offers no guarantee that our contingent, bottom–up inductive cognitions of nature will in fact – even within those spatio-temporal and categorial constraints – issue in a coherent picture of reality. More to the point, the fact that our inductive procedures seemingly do issue in such a picture can, even for the author of the Critique of Pure Reason, be nothing but a lucky coincidence. Right beneath the calm surface of our deceptive experience yawns the abyss of a reality ‘in which something like Hume's “uniformity principle” does not hold’. Allison calls this, evocatively, the ‘specter … of “empirical chaos”’ (Allison Reference Allison2001: 38). And he contends that Kant, in the third Critique, realized that exorcizing this spectre requires that our inductive procedures, too, be transcendentally justified.

The problem, of course, is that Hume did just fine with an inductive justification of the inductive principle and it is unclear why anything more should – or how more could – be required. The belief that nature supports inductive inferences may have considerable heuristic value (the practices of modern science embody it), but it is hardly a condition of the possibility of experience. This has prompted Guyer to conclude that, without importing substantive premises that are decidedly not part of Kant's argument, Kant's attempted justification of our inductive capacities must be considered a failure (Guyer Reference Guyer2005: 68; Reference Guyer1997: 43).

Sensing that Kant's answer to Hume has to consist in more than wanton exorcisms and transcendental foot-stamping, Allison explains that the threat that Kant's transcendental grounding of our concept-forming procedures in the third Critique is supposed to address is not simply the contingency of empirical generalizations with respect to transcendental laws. Rather, Allison notes that in his recapitulation of the transcendental deduction of the principle of purposiveness in § § 76–7 Kant addresses the ‘even more fundamental [and] endemic … contingency that the “particular, as such” (als ein solches) has with respect to the universal supplied by the understanding’ (Allison Reference Allison2001: 38). We may, in short, observe a deepening, over the course of the third Critique, in Kant's conception of the threat that needs transcendental address. It evolves from a problem about the contingency of a given particular with respect to a given universal – to a problem about the contingency of the particular as such with respect to conceptual universality as such.

In detecting a threat at this more fundamental level, Allison suggests that the real problem Kant seeks to solve in the third Critique is not the contingency of empirical conceptualizations vis-à-vis transcendental conceptualizations. Instead, it is the contingency of sensible particularity untouched by our conceptual capacities vis-à-vis the type of conceptual universality those capacities can alone supply. The problem with Allison's first formulation of a ‘specter of chaos’ is, then, that it remains a spectre of empirical chaos. Such a spectre presupposes a ground-level of empirical order, namely, of sensibly given, spatio-temporally and categorially determined objects on which our concept-forming psychologies come to be exercised. Allison now suggests (correctly, I think) that a rigorous transcendental accounting for the problem of sensible particularity cannot simply take a notion of the particular ‘qua spatiotemporal entity or event’ (Allison Reference Allison2001: 39) for granted.

But if such a ground-level of spatio-temporal and categorial synthesization is not presupposed, can the type of reflection that is required in order to first attain that level still be the type of concept-forming reflection discussed in the Jäsche Logik? In one obvious sense it cannot. Empirical-psychological concept formation does presuppose an outer reality of spatio-temporally and categorially constituted objects. By contrast, a more fundamental reflective ascent whose job is to first attain that ground-level of synthesization is a transcendental task empirical consciousness, happily, never confronts. Accordingly, that more fundamental reflective ascent cannot be governed by ‘empirical laws … of association’ (Kant Reference Kant1998: 257; B152), or the laws of (what Kant calls) the reproductive imagination. The reflective ascent in question must, instead, be one accomplished by (what Kant calls) the productive imagination and, hence, belong to the ‘transcendental synthesis of the imagination’ (Kant Reference Kant1998: 256; B151).

While this turns the discussion of reflection's synthetic ascent to universality from a focus on empirical-psychological acts to a focus on transcendental-logical acts, this is a shift which Allison and Longuenesse both can and do welcome. Longuenesse, in particular, argues that the mental acts of ‘comparison, reflection, and abstraction’ and Kant's A-deduction characterization of the transcendental synthesis of the imagination in terms of a ‘run[ning] through and then hold[ing] together’ (Kant Reference Kant1998: 229; A99) of sensible manifolds, are, respectively, psychological and transcendental manifestations of the same fundamental synthetic capacity (Longuenesse Reference Longuenesse1998: 206). Transcendental syntheses, on this view, are presupposed in all efforts at concept formation (Longuenesse Reference Longuenesse1998: 196). Whether the task is to go ‘from the particular to the universal’ or ‘from the particular as such to the universal as such’, the same synthetic capacities are arguably in play.

But if the synthetic ascent for which Kant, in the third Critique, seeks transcendental grounding really is, as Allison proposes, the ascent to universality ‘from the particular as such’, then, given that this is an ascent accomplished by the ‘synthesis of the imagination’, we must ask what the newfound need in the third Critique for grounding that synthetic activity transcendentally can possibly consist in. Remember that in the first Critique the imagination was a ‘blind but indispensable function of the soul’ (Kant Reference Kant1998: 211; A78/B103) whose operations could not be further grounded without launching a regress not unlike the one discussed above (see 2.1).

In response, Allison assures us (in a passage invoked earlier) that there is indeed room for further transcendental argument, because ‘the transcendental deduction [in the first Critique] was not concerned with the “particular as such” but merely with it qua spatiotemporal entity or event’ (Allison Reference Allison2001: 39). The issue, in other words, was not already settled in the first Critique. Moreover, Allison assures us that no regress looms, since the transcendental principle now considered is one by which judgement only legislates to itself (Allison Reference Allison2001: 41).

And yet, despite these reassurances, Allison's account remains inconclusive. This is because the exact nature of the problem that Kant's new transcendental argument is supposed to address remains unclear. First, Allison clouds matters when he continues to identify the problem posed by the ‘particular as such’ (i.e. by the particular not considered ‘qua spatiotemporal entity or event’) with the earlier ‘specter of empirical chaos’ (which, since it concerns disorder at the empirical level, very much presupposes the particular ‘qua spatiotemporal entity or event’). Second, beyond appeal to that earlier spectre – which, as we saw, leaves the Humean justly unperturbed – Allison simply does not say what the alleged deeper ‘unavoidable contingency in the fit’ (Allison Reference Allison2001: 39) between the universal and the ‘particular as such’, that our spontaneous synthetic abilities must negotiate in their apprehension of sensible manifolds, could possibly consist in. But without a credible transcendental-philosophical problem that reflection – limited to a mere ‘moment’ of transcendental synthesis – must address, the routine and facile dismissal of Kant's systematic intent in the third Critique to provide Critical grounding for a reflecting power of judgement not so limited (but, instead, sui generis) remains without justification.

4.1 The Problem of the Synthesizability of Unsynthesized Intuitions

There is indeed a spectre lurking in the shadows of the relation between our synthetic capacities and the ‘particular as such’. It is a problem that, albeit present in the Critique of Pure Reason, Kant did not already address there and that calls for just the sort of transcendental solution Kant presents in the third Critique.

The locus classicus for Kant's conception of the intersection of our synthetic capacities with sensible particularity is his account of the ‘synthesis of apprehension’ (Kant Reference Kant1998: 261; B160) in the transcendental deduction of the categories of the first Critique. Of specific relevance here is Kant's formulation of this synthesis in the A-deduction, where it is presented as the first of three interconnected syntheses of the productive imagination, namely, as the ‘synthesis of apprehension in the intuition’ (Kant Reference Kant1998: 228; A98, caption). Perhaps Kant's best-known claim about that synthesis (alluded to earlier) concerns its cognitive role of turning sensible manifolds into unified intuitions by ‘run[ning] through and then hold[ing] together’ (ibid.) their elements. But it is Kant's claim about a precondition of these synthetic acts of ‘running through’ and ‘holding together’ at the start of his discussion that is my main interest here. Kant notes, as ‘a general remark, on which one must ground everything that follows’ (ibid.), that for the synthesis of apprehension to be possible, further syntheses (specifically temporal syntheses) must already be presupposed.

Kant seeks to justify this idea with an intriguing line of thought. He explains that, metaphysically speaking, ‘[e]very intuition contains a manifold in itself’ (ibid.). But he insists that this does not mean that, epistemically speaking, intuitions are eo ipso represented as containing manifolds. That they are so represented is, instead, a consequence of the circumstance that intuition, qua modification of the mind, is subject to ‘the formal condition of inner sense, namely time’ (ibid.). To prove the point, Kant observes, counterfactually, that ‘if the mind did not distinguish the time in the succession of impressions on one another’ (ibid.), then the manifold of an intuition ‘would not be represented as such’ (ibid.). A manifold of an intuition does not an intuition of a manifold make! Pursuing the counterfactual, Kant explains that in the absence of succession, the manifold of a given empirical intuition would be ‘contained in one instant’ (ibid.). And he concludes that, absent succession, ‘no representation can ever be anything other than absolute unity’ (ibid.; my emphasis).

‘Distinguishing the time in the succession of impressions’ is, however, itself a cognitive capacity and thus entails yet another act of synthesis. Specifically it presupposes an act of pure spatial intuition which Kant famously illustrates with the image of drawing a line in thought. Thus, ‘the synthesis of the manifold in space, if we … attend solely to the action in accordance with which we determine the form of inner sense [illustrated by drawing lines in thought] first produces the concept of succession at all’ (Kant Reference Kant1998: 258; B155). As a consequence of the mind's spontaneous, interdependent and apparently equiprimordial spatial and temporal syntheses, a scenario in which the mind is in touch with an unsynthesized ‘given’ with respect to which it does not already ‘distinguish the time in the succession of impressions’ (Kant Reference Kant1998: 228; A99) does not arise. As Graham Bird puts the point, Kant does not ‘endorse a … bottom–up epistemological foundation for experience in his descriptive metaphysics. Kant's account is holistic, not atomist, and certainly not empiricist’ (Bird Reference Bird2006: 264).

Yet, as Kant's contemplation of the metaphysical reality of intuitions (in contrast to how they are represented by us) makes clear, it does remain an important transcendental-logical presupposition of Kant's holism that sensible particularity – precisely because, for us, it can never be not synthesized – actually be synthesizable. In the first Critique, Kant does not appear too concerned about this presupposition. The spontaneity of our transcendental imagination, in conjunction with the thesis that intuition – metaphysically speaking – ‘contains a manifold in itself’ (i.e. Kant's principle of the ‘affinity of the manifold’ (Kant Reference Kant1998: 235; CPR A113), which has been aptly characterized as a transcendentally necessary, formal, yet material condition of the possibility of cognition (Westphal Reference Westphal2004: 87), seems warrant enough for the assumption that sensible particularity is indeed synthesizable.

In his discussion of the ‘particular as such’ in the third Critique – i.e. of empirical intuition not considered ‘qua spatiotemporal entity or event’ (hence, not considered subject to spatial, temporal, categorial, let alone empirical determinations) – Kant revisits the issue. For good reason. When sensible particularity is thus considered untouched by our synthesizing capacities, a problem readily presents itself. After all, Kant, as noted, insists that if the mind did not ‘distinguish the time in the succession of impressions on one another’ (Kant Reference Kant1998: 228; A99) then the manifold of an intuition ‘could never be anything other than absolute unity’ (ibid.). It follows that the ‘particular as such’ – qua unsynthesized manifold of an intuition – must be considered an absolute unity. The problem is that something that must be considered an absolute unity cannot then be considered capable of being ‘run through’ and ‘held together’ – at least not without further argument. The circumstance that (metaphysically) this absolute unity ‘contains a manifold in itself’ helps little, so long as (epistemically) that manifold is beyond our reach.

One will object that the transcendental imagination needs neither antecedent prompt nor roadmap and will, simply by dint of its spontaneous epistemic activity, come across the elements that, ex hypothesi, are metaphysically there to be run through and held together. Yet – and this, I propose, is Kant's foundational insight in the third Critique – while this is an accurate description of the transcendental-logical process of synthesizing intuitive manifolds, it is not a full transcendental accounting of its epistemic cost.

The spontaneous activity of ‘running through and holding together’ has a relational structure (it is a ‘running from — through — to — ' and a ‘holding — together with — ’). Even granting that the relata entering into the relation can in fact be identified only in and through the synthetic activity itself, the possibility of this synthetic activity requires more than blind spontaneity and the metaphysical assumption that a manifold really is contained in intuition. The latter, after all, is an assumption on the part of the transcendental philosopher thinking about imaginative synthesis, not a presupposition on which the transcendental imagination itself may rely in its spontaneous synthesizing. As far as the spontaneous synthetic activity of the transcendental imagination is concerned, rather, the epistemology of as yet utterly unsynthesized intuitions (namely, absolute unity) is their metaphysics. This immediately makes clear that the relational structure of the synthetic activity itself masks an unwarranted epistemic assumption. The relational structure of transcendental synthetic activity requires that discrete, reidentifiable (even if as yet unidentified) relata be available to enter into it. This presupposition is coeval with the synthetic activity, not a product of it. But the presupposition of the availability of discrete, reidentifiable relata in intuition – thus inscribed into the logical structure of acts of the transcendental imagination – is the very assumption that, ex hypothesi, is denied us when confronting what (for the transcendental imagination) must be absolute unity and, therefore, devoid of identifiable elements. It follows that the spontaneous synthetic activity of the transcendental imagination is itself predicated on a hidden – and thus far unsupported – assumption of the synthesizability of sensible particularity.

The philosophical problem posed by ‘the particular as such’ is, accordingly, the problem of the precise nature (and transcendental justification) of this assumption of the synthesizability of intuitive manifolds. This, I now seek to show, is the problem Kant's transcendental principle of the new reflecting power of judgement is supposed to address in the Critique of the Power of Judgement.

4.2 The Limitation of the Understanding

The main evidence in favour of the interpretation that the fundamental epistemic problem Kant addresses in the Critique of the Power of Judgement is the problem of the synthesizability – qua absolute unity – of unsynthesized empirical intuitions is Kant's discussion of the shortcomings of the understanding in § 77. The main task of § 77 is to tie Kant's extended meditation on the nature of conceptual possibility and sensible actuality in § 76 – a lengthy aside, entitled, simply ‘Remark’, in which Kant presents the transcendental deduction of the principle of the reflecting power of judgement in a sensationist key – to the main argument of the Dialectic of the Critique of the Teleological Power of Judgement. The idea is to show that the very limitations of the understanding, which call for the transcendental deduction of a general epistemic principle of reflection, have additional resonance in teleological contexts, thus justifying special appeal to that principle in those contexts. This combination of general epistemic with specific teleological concerns is the central reason why Kant's characterization of ‘the particular’ in § § 76–7 oscillates between, on the one hand, focus on ‘the particular as such’ – i.e. qua sensible manifoldness ‘which merely gives us something, without thereby allowing us to cognize it as an object’ (Kant Reference Kant2000: 272; KU 5: 402) – and, on the other hand, focus on particular objects ‘e.g., the cause of a product’ (Kant Reference Kant2000: 276; KU 5: 407).

That Kant, in § 77, is principally concerned with particular objects and events (specifically, organisms and organic processes), accordingly, does not entail that it is no longer possible – or indeed necessary – to keep the general epistemic considerations on which Kant's argument is based distinct from the specific teleological use to which these considerations are now being put.

Kant's recapitulation of the general epistemic problem in § 77 is remarkable for its unprecedentedly strong (as well as thrice-repeated) characterization of the understanding's cognitive limitation in its relation to ‘the particular’. Kant's characterization of this limitation makes his discussion in § 77 both the clearest and severest statement in the third Critique of the philosophical problem that the reflecting power of judgement is called upon to address. The most striking of Kant's three formulations of the limitation of our understanding is the following:

On a cursory reading, the passage may appear to be little more than a third-Critique version of Kant's first-Critique principle that all analysis presupposes synthesis (B130). As such, it would seem to confirm Longuenesse's notion of the intimate relation between concept-forming syntheses and conceptual determination. On closer inspection, however, determining judgement's dependency on the concept-forming syntheses of comparison, reflection and abstraction is clearly not the limitation of the understanding Kant has in mind here. After all, Kant's claim is not that the understanding fails to determine the manifoldness of empirical intuition when the universal has yet to be formed. Instead, Kant claims that the (unaided) understanding ‘determines nothing’ in the particular precisely in its indigenous operation of bringing extant analytic universals to bear on the particular in fully formed empirical judgements (‘in its cognition, e.g., of the cause of a product’). As Kant puts it in the first of the three passages that make the point: ‘through the universal of our (human) understanding the particular is not determined’ (Kant Reference Kant2000: 275; KU 5: 406; my emphasis).

This claim would be baffling indeed, if by ‘the particular’ Kant meant spatio-temporally and categorially determined entities or events. On the available interpretations of what it means to ‘determine’ something in the third Critique, the claim would then either simply be false or, else, state no problem at all. For the understanding certainly does subsume sensible particulars under its empirical concepts (even if such subsumption were to yield only episodic object-consciousness along the lines of Allison's ‘empirical chaos’). The understanding is, moreover, constitutive of the objects of cognition (even if those constitutions amounted only to fleeting groupings of causally interacting substances in space and time). And while the understanding certainly fails to supply the matter of its intuition and, hence, to make its objects actual, the inability of our discursive intellect to accomplish this last type of determination (characteristic of certain forms of intuitive intellection) hardly amounts to an epistemic problem within Critical philosophy.

A first step towards unravelling the mystery of our understanding's alleged utter inability to determine the manifoldness of the given empirical intuition is to understand that manifoldness in the metaphysical sense Kant contemplates in the A-deduction. That this is the correct approach is evident, given Kant's appeal to the cognition of a contrasting, intuitive intellect, immediately following his account of our understanding's limitation. The intuitive intellect Kant envisions is one for which both the manifold elements of an intuition as well as the totality of their interrelations are immediately given without need for synthesis.Footnote 3 Kant contends that the cognition of such an intellect would not be cognition of discrete things, hence, that ‘if our understanding were intuitive, it would have no objects except for the Real’ (das Wirkliche; Kant Reference Kant2000: 272; KU 5: 402). Kant explains that in an intuitive intellect's ‘intuition of a whole as such’, hence, in its representation of ‘the Real’, ‘there is no contingency in the combination of the parts, in order to make possible a determinate form of the whole, which is needed by our understanding’ (Kant Reference Kant2000: 276; KU 5: 407; my emphasis). This is to say that the difference between our own discursive intellect and an intuitive intellect lies in the respective way each wrests the representation of a whole from the manifoldness of intuition (one through partially contingent determinations; the other through an unmediated ‘seeing’ of totality). It follows that, in his comparison of the two types of intellection, Kant must use the notion of the ‘manifoldness of an intuition’ in the same sense – or else there would be no comparison.

But then the manifoldness to which the intuitive intellect has unmediated, non-synthetic access cannot be spatio-temporally synthesized manifoldness (i.e. it cannot be our own transcendental imagination's ‘manifoldness represented as manifoldness’). Instead, this manifoldness to which our own intellect has access only through representing it as manifoldness (i.e. synthesizing it) must be the manifoldness that, metaphysically speaking, an intuition ‘contains … in itself’ (Kant Reference Kant1998: 228; A99). Thus, it is the metaphysical reality of ‘the manifoldness [of the given empirical intuition]’, apart from our synthetic representation of this manifoldness as manifoldness, with regard to which, Kant now observes, our (unaided) understanding determines nothing.

Initially, however, identifying the manifoldness with regard to which our understanding determines nothing along the lines that Kant's comparison between intuitive and discursive intellect dictates only seems to deepen the exegetical puzzle presented by Kant's account of the limitation of our understanding in § 77. To be sure, apart from being represented as manifoldness, an intuitive manifoldness must, for us, be absolute unity – but this hardly stops our spontaneous synthetic capacities from synthesizing. Moreover, on the transcendental philosopher's metaphysical assumption (i.e. that what, for us, must be absolute unity nevertheless does contain at least some discrete, reidentifiable elements in itself), there is then no a priori reason why that spontaneity should not be cognitively rewarded (if, perhaps, only in the long run – and however fleeting the resulting cognitions may turn out to be). Accordingly, it seems that even if we do interpret Kant's ‘particular as such’ as ‘manifoldness not represented as manifoldness’, it is still false to say (as Kant does so emphatically) that our understanding ‘determines nothing’ in it.

Kant's complaint in §§76–7 about the ‘entire heterogeneity’ (Kant Reference Kant2000: 272; KU 5: 401) – as well as Allison's complaint about an unavoidable ‘contingency of fit’ (Allison Reference Allison2001: 38) – between conceptual universality and sensible particularity thus seems exaggerated. First, if we take this complaint to express the worry that an additional transcendental principle is needed in order to help our conceptual capacities gain at least minimal traction on an otherwise forever slippery sensible particularity, then the complaint is misplaced. The transcendental assumptions of the spontaneity of the understanding and of the affinity of the manifold (suitably integrated with the rest of Kant's transcendental aesthetic and logic) name all indispensable requirements for progress on the transcendental-logical road to cognition. Second, if we take the complaint to express the maximalist worry (which so exercised the idealists) that we have no independent criterion by which to judge that the understanding's determinations fix any ultimate truths (i.e. the worry that the understanding does not ever truly determine objects), then the response must surely be that aspirations to absolute knowledge have no place in a duly Critical epistemology. In the end, then, there seems to be no sense in which Kant's contention in § 77 – that the understanding ‘determines nothing’ in sensation – can possibly come out true in Critical philosophy. Despite Kant's increasingly dramatic language, no credible cognitive problem in need of Critical solution has thus far emerged.

At this juncture, the distinction between, on the one hand, an identification of the distinct steps of the transcendental-logical process by which an as yet unsynthesized intuitive manifoldness first comes to be synthesized and, on the other hand, a thoroughgoing accounting for the epistemic cost of the steps thus identified becomes important. The latter accounting requires (whereas the former identification does not require) an answer, specifically, to the question of the cost of assuming spontaneous transcendental syntheses. It requires, first, an answer to the question what kind of cognition the assumption of the synthesizability of sensible particularity (necessarily expressed in the transcendental imagination's spontaneous syntheses) is. It requires, second, that this cognition be Critically grounded – lest there remain an unjustified and quite possibly unjustifiable assumption at the very heart of Kant's Critical epistemology.

For the purposes of the present paper – i.e. in order to show that Kant, in the third Critique, seeks Critical grounding for a power of reflection sui generis – we must focus on Kant's response to the first of these requirements. It is a matter of subsequent inquiry whether the cognition thus identified as standing in need of Critical grounding can in fact be so grounded.

The pivotal but enigmatic passage from § 77 can now be explained – along the lines of section 4.1, above – as Kant's response to this first requirement. Kant's point is that the determination of sensible manifoldness as synthesizable (i.e. as comprised of discrete, reidentifiable relata available to be run through and held together) cannot be a determination on the part of the understanding. The impossibility attested here lies in the fact that the syntheses of the understanding are themselves relational acts of ‘running through’ and ‘holding together’ manifoldness, which inescapably presuppose – and therefore cannot, on pain of regress, first propose – the synthesizability (or run-through-ability) of intuitively given material. When Kant says that the understanding ‘determines nothing with regard to the manifoldness [of the given empirical intuition]’ (Kant Reference Kant2000: 276; KU 5: 407; my emphasis), he means just this. The one ‘determination’ of sensible manifoldness that the determining power of judgement in principle cannot make is the one which does not have the manifold elements of the given empirical intuition for its objects but the manifoldness of that empirical intuition itself: that it be manifoldness (hence, determinable and synthesizable). This foundational ‘determination’ is forever a mere presupposition manifest in the relational structure of the understanding's spontaneous syntheses – never a product of those syntheses.

This, to be sure, means that the presupposition in question is not a ‘determination’ in the strict sense of a determining judgement, at all. For this reason, Kant says that the understanding must ‘expect’ (ibid.) it from elsewhere – namely, from a kind of judgement ex hypothesi distinct from the understanding's own (determining) form of judging. That Kant concludes the passage from § 77 by saying that the understanding must expect this determination ‘from the subsumption of the empirical intuition … under the concept’ (ibid.) does not run counter to the present reading. Kant here evidently has in mind the kind of ‘subsumption’ characteristic of exercises of the reflecting power of judgement. After all, reflecting judgement is a form of judgement – a synthetic, combinatorial act – which sets particular and universal into a logical relation to one another (see below). At any rate, if we did not bill this ‘subsumption’ to the reflecting power of judgement, then Kant's point about the limitation of the understanding in § 77 would (in the absence of alternative forms of judgement) be plainly self-contradictory (Kant would say that the understanding cannot determine the particular, yet, must expect that determination from its subsumption of the particular under the universal – hence, from determining the particular).

A model for the kind of non-determining subsumption that is thus called for by Kant's diagnosis of the understanding's necessary inability to represent sensible particularity as synthesizable is – entirely predictably – supplied by Kant's explanation of the cognitive operation of reflecting. In his crucial definition of reflecting in the First Introduction to the Critique of the Power of Judgement, Kant notes that reflecting need not be identified, merely, with a comparison of ‘given representations with others … in regard of a concept thereby made possible’ (Kant Reference Kant2000: 15; EE 20: 211). That would be the familiar notion of reflecting, relevant in considering the understanding's concept-forming syntheses of ‘running through’ and ‘holding together’ sensibly given material with the aim of ‘recognizing’ that material in a concept. But it is clearly not the only notion Kant considers (Mertens Reference Mertens1975: 95). Kant now insists that the operation of reflecting can also consist in the comparison of given representations ‘with one's cognitive faculty in regard of a concept thereby made possible’ (ibid.). This evokes Kant's conception of ‘transcendental reflection’ in the Critique of Pure Reason – roughly, the philosophical desideratum of assigning different types of representation to their appropriate cognitive sources (Kant Reference Kant1998: 366–7; A260–1/B316–17). But in the context of the third Critique this form of reflection assumes a different – and altogether heightened – significance. It no longer merely covers the desirable (but contingent) epistemic posture Kant believes should be adopted by the transcendental philosopher (Westphal Reference Westphal2004: 16–17).

The principal instance, in the third Critique, of the idea of comparing given representations with one's cognitive faculty in the service of conceptual cognition is the foundational assumption – necessarily inscribed into the structure of the transcendental imagination's spontaneous syntheses themselves – of the synthesizability of sensible particularity. For this assumption just is the comparison or ‘holding together’ of sensible particularity (ex hypothesi considered as yet unsynthesized or ‘as such’) with our faculty of concepts (specifically, with its structural demand for discrete, reidentifiable relata) in regard of a concept thereby made possible, i.e. with the aim of enabling conceptual cognition. Unlike determining synthesis, which can only address the elements of a given sensible manifoldness (comparing them with each other in concreto), reflecting synthesis addresses the given sensible manifoldness as a whole (considering it according to the type of representation it is and comparing that with our faculty of concepts). Accordingly, it can arrive at the requisite assumption of the availability of reidentifiable elements without launching a regress.

Seen through the lens of this interpretation of § 77, reflecting judgement's task of ‘ascending from the particular to the universal’ is then not the task of ascending from an unsynthesized sensible manifoldness considered synthesizable (the particular) to that same manifoldness, now synthesized (e.g. recognized in a concept; the universal). This latter operation is accomplished by the transcendental imagination's spontaneous syntheses without transcendental-logical difficulty and is an ascent for which transcendental grounding in the form of a Critique of the Power of Judgement is neither necessary nor possible. Reflecting judgement's task of ‘ascending from the particular to the universal’ is, instead, the task of ascending from an unsynthesized sensible manifoldness considered unsynthesizable – qua absolute unity that must, ex hypothesi, be devoid of discrete, reidentifiable elements (the particular ‘as such’) – to that same manifoldness, now considered synthesizable, hence, a candidate for conceptual cognition (the universal). Importantly, this ascent is an act of judgement (a ‘subsumption’ of sensible manifoldness under the conditions for object-cognition as such) that does not represent unsynthesized sensible manifoldness objectively (i.e. as synthesizable), but subjectively (i.e. as if it were synthesizable). An exercise of ‘transcendental reflection’ (in the original first-Critique sense) readily shows that Critical philosophy must account for the cognitive source and transcendental-logical warrant of this most fundamental of cognitive ascents. In other words, Kant's discussion in § 77 pointedly calls for what Kant had claimed to be presenting all along: a Critique of the (Reflecting) Power of Judgement.