JAMES CONANT | ‘Why Kant is not a Kantian’ — Philosophical Topics 44(1) (2016): 75–125 / “Die Einheit des Erkenntnisvermögens bei Kant” —In A. Kern & C. Kietzmann, Selbstbewusstes Leben. Texte zu einer transformativen Theorie der menschlichen Subjektivität, pp. 229–69 | Berlin: Suhrkamp, 2017 / ‘Kant’s Critique of the Layer-Cake Conception of Human Mindedness in the B Deduction’—In J. O’Shea (ed.), Kant’s Critique of Pure Reason: A Critical Guide (Cambridge UP, 2017), pp. 120–39


This is the second essay in a series of critical engagements with important recent articles or papers by prominent Kant or Hegel scholars, including two or more critics and a reply by the author. In this second instalment, Dennis Schulting critically discusses James Conant’s recent article on the Transcendental Deduction; Conant’s reply to both this and Sacha Golob’s essay, published earlier, will be published in due course




By Dennis Schulting

In an illuminating new essay ‘Die Einheit des Erkenntnisvermögens bei Kant’ (Conant 2017a), James Conant critically addresses what he argues is a widespread assumption in modern philosophy, namely, the assumption that our rational capacity to know is a capacity that is somehow “added” or tacked on to the capacity that we humans share with other animals, that is, our receptive capacity for sensations, our sensibility.[1] This is the so-called “additive” theory of cognition, more specifically of the relation between sensibility and the understanding.[2] He addresses this assumption by looking at the main argument of Kant’s Transcendental Deduction. Let me say upfront that I think Conant’s paper is one of the very few long-form pieces on the central thrust of the Deduction that I have read from the last twenty years or so, if not longer, that are as rhetorically strong as they are, on the whole, both interpretatively and philosophically appealing. I believe it is one of those papers that will, or at any rate should, be seen as a standard reference in the same way that Dieter Henrich’s influential article on the ‘two-step’ procedure of the B-Deduction has been—Conant indeed also refers to Henrich’s now famous ‘two-step’ proposal, but thinks that his own construal avoids what, in Conant’s view, can be seen as the delusive nature of Henrich’s overall framework, which suggests that there are indeed two independent, separably intelligible “steps in a proof” (Conant 2016:111).

I quite agree with the general tenor of Conant’s paper, namely that the Deduction should not be read as if the two stems of knowledge, sensibility and understanding, were connected in the way suggested by what he aptly calls the “layer-cake conception of human mindedness” (2016:78/2017a:232), whereby concepts are somehow “added” or tacked on to pre-given manifolds of representations to constitute acts of cognition. This is not to say though that I agree with all of the arguments he presents in support of this critical view, or even with the main argument he mounts in support of undermining the layer-cake conception of human mindedness. I think Conant oversells his rightful critique of the layer-cake conception by underestimating the modal nature of Kant’s reasoning in the Deduction and the compatibility between it and a minimally or relatively nonconceptual interpretation. Conant thinks that a rejection of the layer-cake conception of human mindedness entails an unqualified rejection of nonconceptualism of any sort. I think there is a third possible route, which likewise rejects the layer-cake conception of human mindedness, but is still compatible with a kind of minimal nonconceptualism about the relation between sensibility and the understanding, which also avoids the problems of Conant’s own positive reading (which I come to in the course of this discussion). To put it succinctly, I think Conant’s reading of the Deduction is too strongly conceptualist, and unnecessarily so. In this notice, I shall address some of the main points on which I diverge from Conant’s reading.

1. The “Additive” / “Disjunctivist” Theory of Cognition and Kant’s Deduction

In the essay, Conant debunks what he calls the “additive Auffassung des Verhältnisses zwischen Sinnlichkeit und Vernunft qua kognitiver Vermögen” (2017a:231),[3] in other words, the view that our rational and sensible capacities are wholly separable capacities when it comes to the formation or possibility of knowledge, and the view that, if we abstract from the unique and characteristic rational capacity that humans have, humans and other animals share the same sensible capacity to perceive their empirical environments and that when we make a claim to knowledge, our capacity to know “adds” something to our sensibility. The general, philosophical thrust of Conant’s story behind this rejection does not strike me as particularly controversial, or novel, for that matter (Conant himself points to McDowell’s influence), and it is also intuitively plausible. So I am not going to probe this here. Rather, I want to look at Conant’s interpretation of Kant’s Transcendental Deduction as a way of shoring up his philosophical critique of the “layer-cake conception of human mindedness”, that is, seeing Kant’s Deduction as an argument that debunks the “additive” / “disjunctivist” conception of the relation between our cognitive capacities.

Though I think Kant’s main aim in the Deduction is indeed to argue for the intimacy between the cognitive capacities that together enable knowledge, I do not think Conant’s attempt in enlisting Kant in an effort to refute the “additive” theory of cognition is entirely successful; for in trying to present Kant as a critic of the “additive” theory of cognition, he ignores an important modal element in Kant’s argument that has to do with what elsewhere (Schulting 2017a) I have called Kant’s “radically subjectivist” outlook, which involves central aspects of his transcendental idealism, which Conant does not discuss (but it is not necessary for my account here to enter that debate). The argument that Kant’s argument in the Deduction should not be seen as an “additive” / “disjunctivist” theory of cognition may be seen as a major obstacle to regarding Kant as a nonconceptualist. But I shall argue below that while Kant is indeed not a so-called “additive” theorist of cognition, this does not, or at least not on its own, undermine the view that Kant is a nonconceptualist insofar as the mere intuition of objects or our mere sensible receptivity independently of the understanding is concerned, a view the real possibility of which Conant explicitly dismisses. Instead, Conant proposes the “transformative” view of cognition, whereby our sensibility is wholly transformed by and in virtue of the capacities of our understanding, in other words, by our use of concepts for any claim about empirical objects. This uniquely and completely differentiates us, as rational animals with sentient capacities, from non-rational sentient beings. Conant writes the following:

What happens when we move through the Stufenleiter of forms of cognition, from considering the form of cognitive capacity involved in nonrational animal cognition to a form that essentially involves the use of concepts and the formation of judgments, is that the sort of animal under consideration is one whose nature is transformed through and through. A corollary of this Kantian thesis […] is the following: The possibility of something’s being given to the sensory consciousness of a rational animal, if that animal’s awareness thereof is to be conceived as an integral moment in the exercise of its overall capacity for rational cognition, requires that that capacity for sensory affection radically differ in its internal character from that of any nonrational animal. It requires that we come to see how the capacity for sensory affection in the rational animal exhibits the marks of the form of its capacity for cognition and thus how the episodes of such sensory consciousness are themselves shaped by the manner in which they are, ab initio, such as to be apt to bear on rational reflection on how things are. (2016:78–9/2017a:232; my underlining)

Das Schichtkuchenmodell des menschlichen Geistes zurückzuweisen heißt zu denken, dass die Natur des Vermögens zu sinnlichem Bewusstsein im menschlichen Tier kraft der Relation, in der es zur Vernunft stehen muss, verändert wird. Sein innerer Charakter qua Vermögen der Sinnlichkeit ist formal von demjenigen verschieden, der sich in jedem beliebigen Tier mit einem kognitiven Vermögen ohne vernünftige Form findet. (2017a:232–3)

These remarks suggest, first, that there is something intrinsic about our human sensibility that makes it apt for or amenable to rational reflection, as if our sensations were teleologically intended to be the content of rationally formed perceptions and ultimately judgements about how objects affecting our minds are, that is, that sensations have an inbuilt capacity to be intentional rational mental content (if so, this would eo ipso invalidate any kind of essentialist nonconceptualism). Secondly, notwithstanding Conant’s legitimate worry that what he calls the “layer-cake conception”—whereby the sensible and rational parts of our cognitive capacities are absolutely separated and the rational part is as it were imposed on (“added” to) a pre-given sensible content—is very problematic, the view that, as Conant suggests, humans and other animals have radically different kinds of sensibility (and that this is indeed Kant’s view) does not strike me as immediately persuasive. In what sense are human beings qua sentient beings then still like other animals? Conant remarks that

our sentient cognitive faculty, as we encounter it in act […] represents a faculty whose form is utterly distinct in character from any whose exercise might manifest itself in the sensory life of a nonrational animal—even if, when investigated from a merely physiological point of view, that animal’s sensory equipment might reveal itself to be in countless respects physiologically indistinguishable from our own. (2016:79/2017a:233; my underlining)

This does not help much, at least not stated as such, in clarifying the precise difference between human and non-human animal sensibility and why they are different, unless the ‘formal’/‘physiological’ distinction[4] is supposed to play an operative role here.

I think the problem here lies in the fact that, at least prima facie, Conant looks in the wrong direction for the solution to the familiar Kantian problem of how to see our sensibility and our rational capacity, the understanding, as linked up. The link between sensibility and the understanding has got nothing to do with the question whether, from a factual perspective, human beings and animals share or don’t share a common sensibility, nor with analysing the question of how, in a kind of bottom-up process, our sensibility is supposedly geared toward entering into our rational reflections or judgements about objects. Such an analysis would require probing into the putative natural disposition of our sensibility toward rationality—and this seems to go wholly against Kant’s sui generis transcendental or Copernican approach to the analysis of knowledge. Neither is it of course true, as Conant is right to point out, that these our judgements are somehow imposed upon (or “added” to) our sensibility after the fact along the “layer-cake” model, as if our sensibility were something wholly alien to our rationally formed judgements (whatever one understands by ‘imposition’, and we shall have cause to return to this infelicitous terminology below). These questions are informed by a much too naturalist, and hence, I believe, an in essence rather unkantian understanding of the possibility of knowledge and what constitutes it and any accounts thereof.

What I believe is instead fundamentally at issue is that, for Kant, the link between sensibility and the understanding is one whereby from the subjective perspective of the understanding (and thus judgement [A69/B94]) whatever sensibility delivers to us is taken, by us as cognisers or judgers, as necessarily contributing to our knowledge of objects and as intimately conjoined with our capacity for the use of concepts in virtue of that capacity, under the assumption that (1) there are such objects, and (2) sensibility provides us immediate access to them, and (3) we have knowledge of them—such ‘taking as’ is to be seen in terms of Kant’s theory of transcendental apperception, and, as said, I have elsewhere labelled this reading of Kant’s theory of knowledge “radically subjectivist” (see Schulting 2017a), whereby the epithet ‘radical’ should be seen as differentiating the subjectivism at issue from psychological construals of subjectivism. There is an implicitly modal element involved here, which suggests that sensibility and the understanding are necessarily and intimately linked only if from this connection objective cognition or knowledge should be seen to arise. Sensibility and the understanding are not connected or conjoined in a modally absolute sense, i.e. simpliciter.

This modal constraint on our cognition is often wholly ignored—this, I am inclined to believe, is because of an almost universal naturalist bias in readings of Kant’s Deduction argument, or of Kant’s philosophy in general. Sensibility does not lead to judgements about objects as a matter of course as if it were its natural disposition to do so, and nor is it the case that sensibility always leads to judgements about objects. It is not as if the choice were between arguing that sensibility and understanding either are or are not, in one way or another, always related to each other. Rather, they are related only under a certain condition, which means they are not related at all if that condition is not met. My last point should not be taken as simply saying that, necessarily, for every tokening of sensibility there is a necessary condition under which it obtains and that without it sensibility is impossible. The conditionality at issue is not an absolutely necessary condition on sensibility, as space and time are. Rather, it concerns the subjective condition of thought that takes the sensible tokening as necessarily conjoined with itself (i.e. with thought itself). This condition, which points to the subjectivism I suggested earlier as an alternative reading of the relation between sentience and sapience, is a conditionally necessary condition on sensibility, namely insofar as sensibility provides the necessary connection of our conceptuality to the world.

But I’m getting ahead of myself. Let’s first turn to Conant’s own take on things and focus on his ‘formal’/‘physiological’ distinction with respect to human sensibility and how the formal side of sensibility is intimately linked to the understanding. The “change” or “transformation” that sensibility is said to undergo in human cognition, and about which Conant speaks in the above-quoted passage, leads him—under the influence of work by Matthew Boyle (which I must admit I’m not familiar with[5])—to label his own reading of the Kantian cognitive model the “transformative conception of human mindedness” (2016:80/2017a:234), which is not coincidentally also reflected by the subtitle of the German volume in which Conant’s essay appears (i.e. “Texte zu einer transformativen Theorie der menschlichen Subjektivität”).

2. The Three “Puzzles” in Relation to the Transcendental Deduction

Conant argues that there are three main puzzles in relation to the topic at hand, and that these all hang together. Rather than being concerned in general with refuting sceptical doubt, Conant thinks—and I believe quite rightly—that Kant’s Deduction has got to do with solving fundamental puzzles that concern the very possibility of knowledge, and is not about “whether we can have this or that bit of knowledge” (2016:83/2017a:237). I thus agree with Conant that the Deduction is not about whether we have knowledge under various (non-transcendental) conditions or indeed whether we are justified in believing that p about objects or about the conditions or warrants for a justified true belief p, or, more radically, whether there are even objects out there, but rather about how it is possible in the first place that we have knowledge in general and can make particular knowledge claims, judgements as Kant calls them. The three puzzles at issue in the Deduction, according to Conant, are (I paraphrase):

1. What is the relation of the formal conditions of sensibility, as expounded in the Transcendental Aesthetic, to the formal conditions of the understanding, as expounded in the Transcendental Analytic?

2. What is the relation between the A- and B-versions of the Transcendental Deduction?

3. What is the relation between the first and second halves, the so-called “first” and “second step” (Henrich 1969), of the B-Deduction?

I think this breakdown is right as far as the first and third puzzles go, but I would argue that the question of the second puzzle is much less important for an explanation of Kant’s main argument in the Deduction, it being more of a philological than an interpretative issue; notwithstanding the apparent structural differences between the A- and B-Deductions, I think that the thrust of the central argument of the A-Deduction can be read along the same lines as the main argument in the B-Deduction (I have done so myself in Schulting 2017a, ch. 6), so that a comparative examination of the two versions won’t reveal any substantial additional problems with respect to resolving the other two puzzles—it appears, further on in his text, that Conant in fact thinks the same. At any rate, I do not think that not being able to immediately resolve the second puzzle will have an impact on resolving the first and third puzzles. The first and third questions can thus be answered without answering the second. But indeed the first and third question hang together: having an answer to the first puzzle enables one to answer the third. Conant however polemically suggests that “most accepted solutions to the first puzzle render the second and third puzzles insoluble” (2016:81/2017a:235). This makes one certainly curious as to what his own proposed solution to the puzzles amounts to.

3. The Relation Between Sensibility and the Understanding

Nonconceptualists might be taken to argue that we first sensibly apprehend an object in order to only then make judgements about it. Assuming that this is their view, this doesn’t work, for among many other reasons it invites an explanatory regress as to how sensibility and the understanding are a priori connected such that there is a case of knowledge properly so called or a claim to knowledge properly so called, as would appear to be a requirement on Kant’s account of possible knowledge or cognition.[6] A post hoc imposition of a judgemental structure upon a pre-given sensible manifold is a posteriori, not a priori (even if of course the manifold is given independently). Also, as Conant crucially notes, if our first cognitive encounter with the world is purely sensible, and does not reflect in any way the structure of our understanding of the world, which only in the second instance “works that raw sensory matter up into something fit to be a candidate for entering into a relation of objective purport between how we take the world to be and how it is”, then we are left with a “picture in which our forms of understanding always operate at an unbridgeable remove from” reality (2016:84/2017a:239). Conant thinks this results in the Humean conclusion that “[t]he forms of our understanding—categories such as substance and causality—now appear, at best, to involve mere subjective projections onto something already given, something to which the unity of thought is external” (2016:84/2017a:239).

But—and this is the operative question—does refuting this externalist picture of the relation of the understanding to sensibility, and hence to the world, necessarily mean that sensibility is always already conceptually informed, as Conant suggests? Conant says that the empiricist thesis that “[w]hat is given to the senses does not as such exhibit the form of thought” (2016:84/2017a:239) is contradicted by Kant. But this, I believe, is not the case (cf. B129–30, 134). Nonconceptualists are not wrong about the fact that mere intuition is not conceptually laden; they are wrong in claiming that sensibility is “self-standingly intelligible” (2016:84/2017a:240) and is a sui generis mode of objective cognition properly so called and they are wrong about how intuitions and the understanding are related in cases of genuine cognition, or at least they are negligent in explaining that relation.

My own ‘radically subjectivist’ (henceforth ‘subjectivist’ for short) reading of the relation between sensibility and the understanding has it that sensibility and the understanding are intricately related at the fundamental level insofar and only insofar as the understanding takes them so. This means that it is certainly really possible for a mere sensible manifold of representations not to be conceptually informed, namely for all occurrent cases where the understanding does not actually take up the sensible manifold as referring to a certain object, in an actual judgement about that object. Sensibility does after all not contain the element of necessary combination that is needed for conceptual recognition and thus for empirical knowledge (see the B-Deduction, §15). As said, there is a modal element involved in our knowledge, that is to say, sensibility is conceptually informed by the understanding at the fundamental level if and only if I, as judger, take what is receptively given as contributing to the content of my judging, that is, judge the content that is given in sensibility to be the intentional content that says something about the world. Importantly, this does not at all amount to a relativist position about knowledge, nor does it leave a gap between my judging and the world, for my judging in any occurrent case of judging involves staking a claim about the very existence and the objectivity of the thing about which I judge that so and so. There couldn’t be an issue of a remaining gap in principle between my judgement that so and so and things being so and so (I say “in principle”, for my judgement p could of course still be empirically false; this concerns the difference between a judgement’s necessary objective validity, its incontrovertible transcendental truth, and its having a truth value).[7]

The understanding informs sensibility not as a matter of course, as if Kant wanted to say to Hume[8] that “what is the case in cognition is precisely the opposite of what you claim”. Kant does not say that necessary connection can after all be found in sensibility. This would be to ignore Kant’s clear statements to the contrary (B130, 134) and to misunderstand his relation to Hume. However, the subjectivist reading does also not mean that Kant is just a Humean, who believes that knowledge is merely what we project on to the world, but has got nothing to do with the world itself and that we always “operate at an unbridgeable remove from” the world. Kant is not in any way acceding to the empiricist, who, as Conant puts it, “insists that our cognitive access to the world must, in the first instance, be purely sensory” and that there is “a self-standing sensory way of knowing what is—one which can operate independently of the exercise of a capacity for thought” (2016:83/2017a:238).

Conant says that the task of the Transcendental Analytic is to show that the form of really existing things is no other than the form of consciousness of the thinking, judging and experiencing subject. That is to say,

the form of sensory consciousness as it figures in sensory apprehension of an object cannot as such bear no internal relation to the form of the capacity which we exercise in engaging in acts of thought and judgment about that same object” (2016:85/2017a:241; boldface added)

In the German version Conant renders the underlined phrase more precisely as “[d]ie Form einer Episode des sinnlichen Bewusstseins […] qua sinnliches Auffassens eines Gegenstandes”. But the claim Conant makes here is ambiguous between saying that

A: the form of an episode of consciousness qua sensible apprehension of an object is not unrelated to the form of the capacity to think or judge about that same object


A*: the form of an episode of consciousness qua sensible apprehension of an object as an object is not unrelated to the form of the capacity to think or judge about that same object

In A*, the form of an episode of consciousness is indeed not possible without the form of thinking, since the categories first enable an objective apprehension of the thing, namely to apprehend a thing as an object that I perceive as existing—the ‘taking as’ is a function of the capacity to think or judge. But on reading A, the form of an episode of consciousness is not so dependent, for there is no specification on this reading of a further constraint that goes beyond the fact that an episode of consciousness is apprehended qua sensibly apprehending a merely indeterminate object of perception. So technically, reading A is false, for a mere sensible apprehension of an object is not identical to, or dependent on, a judging about an object. Much of course depends here on how we are to gloss the term ‘object’.

The distinction between a mere sensible apprehension of an (indeterminate) object and a sensible apprehension of an object as an object amounts to the distinction between concept-free “simple seeing” and conceptually loaded perception respectively, whereby one understands perception in the strict sense of objectively valid, veridical perception. “Simply seeing” x is like stepping on x, it is not the same as actually perceiving x (Cassam 2007). Perceiving x would involve concepts (including categories) and thus the capacity for the use of them, simply seeing x does not. As Quassim Cassam (2007) has argued persuasively (following Fred Dretske), not knowing what a cup is prevents one from perceiving a cup but not from seeing a cup (that is, seeing an undetermined object that is a cup). That, however, Conant wants to read the apprehension of an object in the broader sense of A (and not in terms of A*) is made clear by his conclusion to Section III: “What is given through the senses, simply in virtue of being intuited, already exhibits a form which is not simply other than that which the categories prescribe” (2016:85/2017a:241; emphasis added). In other words, it appears that for Conant even simple seeing requires a form prescribed by the categories.

4. Ways of Reading the Goal of the B-Deduction

One of the problems with getting to grips with Kant’s Transcendental Deduction is determining, first, what its goal is. This involves establishing the precise relation between the Deduction—in particular if we look at the so-called ‘second step’ of the B-Deduction—to what has already been shown in the Transcendental Aesthetic (Puzzle 1). And if we look at the B-Deduction only, it involves determining the much-discussed relation between the first and second steps (Puzzle 3). Conant talks about four “choice points” (Entscheidungspunkte), that is, central points in the argumentation, where one must choose to either make a right or a left turn, which decides on one’s further interpretative path through the thicket of arguments.

4.1 A Restrictive or Non-Restrictive View of Subjectivity

First, with respect to the role of subjectivity, according to Conant one adopts either a restrictive or non-restrictive view of subjectivity (this is the first “choice-point”). That is, a restrictive view of subjectivity sees the conditions of cognition or knowledge as limiting conditions or constraints on the possibility of cognition or knowing. But the terminology of ‘restrictive’ or ‘limiting’ is misleading. What is at issue in Kant’s account are rather constitutive conditions of cognition. Conant quotes Pippin in this regard (2017a:243–4),[9] but he seems to read Pippin’s comments about subjectivity in a psychological sense. Conant says:

If one begins by understanding Kant’s conception of finitude to be a restrictive one [as Conant believes Pippin just accepts], then it is almost impossible to avoid eventually sliding into […] an impositionist reading of the First Critique—a reading according to which the categories of the understanding are taken to impose certain forms of unity on an exogenous matter. (2016:88 /2017a:244–5)

It appears that Conant here reads Pippin as if the latter believed that the conditions of cognition are limiting in a merely subjective, psychological sense: they constrain our knowledge merely to what we are able to know, and do not concern the way that objects are constituted. But, on Kant’s account, the necessary, subjective conditions of finite experience are of course not psychological conditions. If these conditions are not only subjective but also objective conditions as Kant says (A158/B197) (and Conant himself quotes McDowell on this), then there cannot be a question of an “impositionist” reading, because the conditions of thought first constitute objectivity, as Conant himself reasons,[10] so there is nothing that is at first only objectively given on which the forms of thought are subsequently imposed, exogenously. Conant is of course right to say that a certain use of unfortunate terminology (“imposition”, or indeed Kant’s own “Anwendung” or the language of “addition” in §15 in the B-Deduction) does suggest such an “impositionist” reading (2016:88/2017a:245).[11]

4.2 Two-Stage or Anti-Two-Stage

The second “choice-point” is probably a decisive one when it comes to refuting the “layer cake model”. Conant says that most Anglophone readings espouse the two-stage reading. He differentiates between 3 versions of the two-stage reading:

(i) According to the standard reading, there are two temporally separated episodes in apperceptive consciousness, that is,

a first apperceptive stage in which a manifold of bare sensory consciousness is constituted, followed by a second stage in which it is then synthesized and brought into accord with the unity prescribed by the categories of the understanding. (2016:90/2017a:246)

Presumably, the first “apperceptive stage” is the analytic unity of consciousness or apperception, while the “second stage” is the synthetic unity of consciousness or apperception. But this is, in my view, based on a mistaken, that is, metaphysically intemperate and logically confused reading of the principle of transcendental apperception, for there just is no such two-stage apperception, and if some commentators hold such a view,[12] they are simply wrong. Moreover, if it were Kant’s view, there would be a problem about regress: how do we get from the first to the second stage? Given that we are already talking about transcendental apperception as an original, a priori form of consciousness more original than which there is none, there could not be a third kind of a priori principle of apperception which enables the transition from the first to second stage.

I cannot discuss apperception here in any detail (see Schulting 2008, 2012a, 2012b, 2015a, 2016, 2017a, 2017c, and forthcoming). What I can say here is that the analytic and synthetic unities of apperception that Kant formally distinguishes are simultaneous, coextensive in all actual cases of cognition of which apperception forms part and parcel, and reciprocally dependent. The synthetic unity of consciousness is the logical ground of analytic unity of consciousness and any synthetic unity of consciousness or apperception necessarily entails an analytic unity of consciousness or apperception. You can’t have an analytic unity of apperception without synthetic unity of apperception. And if you don’t have an analytic unity of apperception, you don’t have any transcendental apperception at all—you might have empirical apperception, but I don’t think that is what Conant has in mind, since (1) empirical apperception is purely contingent, associative, and subcognitive, and there is no teleological power of any sort in empirical apperception which makes it amenable to being conditioned, governed or subject to transcendental apperception (cf. B133, A107) and (2) he appears to refer to empirical apperception in his second variant reading, which we consider next. So this variant (variant 1) of the two-stage reading can simply be dismissed, despite the fact that, at least according to Conant (2016:91/2017a:247), this appears to be the most widespread version of the two-stage reading.

(ii) The second variant is equally mistaken. This variant says that the first stage is a subconscious or sub-apperceptive stage and the second stage is the apperceptive stage, i.e. it concerns the not less popular view that the principle of transcendental apperception and hence the categories first bring the manifold to consciousness. This suggests that everything that is not apperceived in the transcendental way is unconscious. By implication, any tokening of empirical apperception or consciousness is subject to transcendental apperception as a general principle of consciousness. This would entail that any representer who is either in principle (animals, infants) or occurrently not in possession of (her) apperceptive capacities is eo ipso unconscious (i.e. the intensity of consciousness equals a magnitude of 0). This is often thought, but it is an implausible view. And Kant does not hold it.[13] The principle of apperception is not a principle of consciousness; not even Leibniz held the view, though he is often portrayed as if he did, and neither is transcendental apperception the necessary principle of empirical apperception simpliciter (see Schulting 2015a).

(iii) Fortunately, Conant concentrates on the third variant of the two-stage reading. Conant thinks that sensibility and thought are two logically but not temporally separable components of cognitions. This is the most promising reading, and if modified, the right one. Conant thinks that reading the Aesthetic and the Deduction as separably enabling two forms of cognition, namely sensibility or the intuiting of an object, on the one hand, and thinking an object, on the other, requires the first variant, which says that there are two ‘apperceptive stages’. I do not think it does. For in order to be sensibly aware of an indeterminate object in space and time, I do not need transcendental apperception at all, and such an episode is nonetheless a form of cognition, or at least a form of sensory awareness (see the Stufenleiter distinctions). The indeterminacy implies that the ‘I’ as identical subject of perceiving is not involved in the perception. There is just a perception (in the sense of ‘simple seeing’) of some object going on in someone’s mind, for which there is no occurrent apperceptive ‘I’ strictly speaking who is aware of the perception as being her perception, and hence of a determinate object (cf. Conant 2016:107). The esse and percipi of the perception collapse into one. Conant writes:

Such a form of sheer sensory awareness is taken, on the standard variant of a two-stage reading, to be what is given to us first in the process of cognition—where “first” means that this stage of cognitive processing is taken to be prior in a temporal as well as a logical sense. A proponent of the standard variant will generally go on to read the Transcendental Analytic as introducing what is taken to be a further requirement on genuinely objectively valid representations of objects—one that comes into play when these elements of an episode of sensory consciousness are “brought under” concepts— where this business of “bringing things under concepts” is construed as both a temporally and logically posterior stage in the cognitive process to mere sensory awareness of the object. (2016:91/2017a:247)

Conant is right that some nonconceptualists on Kant can be read in this way, and that they are mistaken to read Kant in this way. But I believe Conant is wrong about why they are mistaken. That the stages of cognition (intuition of an object and thought of object) are indeed not temporally subsequent and thus separate, does not imply that sensible episodes of consciousness (sensible apprehensions) are always already combined with a cognitive thought or judgement, i.e. that any and all sensible apprehension is always already conceptual, or, is a content that is necessarily thought by a judging subject (cf. 2016:85/2017:241), or that a sensible manifold is always already unified in the strong sense that Kant means. Conant cites C.I. Lewis and says that “Lewis insists” that “[e]xperience always comes to us as unity” (2016:91/2017a:248), and that this unity is the unity of the understanding. But this just conflicts with §15, in the B-Deduction, where it explicitly says that (necessary) unity must be added to the manifold, for it is not contained in it as a matter of course. So the manifold of representations in an intuition as such does not come to us in a unity. We have to add the unity to it (see Schulting 2012b:141ff.). We have to be careful though to read this ‘adding’ not in an impositionist way, the way that Conant rightly rejects.

4.3 The Sense(s) of ‘Anschauung’

For the third “choice-point”, Conant concentrates on the term <Anschauung>, which he says can be, and often is, read in two senses, but should in fact not be interpreted thus. He reflects on one of the most prominent readings of the Deduction by Henry Allison. Allison cites British philosopher and Kant scholar W.H. Walsh, who argues that sensible intuition is “only ‘proleptically’ the awareness of a particular” (Allison 1983:67). Allison says—and quite rightly, I think—that we must distinguish between indeterminate or nonconceptual and determinate or conceptual intuitions (1983:67). Conant calls this distinction a distinction between two kinds of intuition (2016:95/2017a:250), but that is not entirely correct, for intuitions can become determined, as Allison indeed says, but there are no ontologically distinct kinds of intuition, which have nothing in common. A modal aspect is involved here which is ignored by Conant, who focuses on an explanation by way of classifying putatively different types/kinds of intuition, whereby the indeterminate intuition is supposedly the first stage of experience (mere sensory awareness or sensation) and determinate intuition represents the second stage of fully-fledged experience (Erfahrung in Kant’s sense). Conant speaks of the “reshap[ing]” (2016:94/2017a:250) of intuitions, from the indeterminate kind to the determinate kind of intuition. To be sure, Conant criticises this view. But I believe there is more similarity between a view such as Allison’s that says that intuition can become determinate and Conant’s own “transformative” view of intuition or sensibility than the latter would have it.

To be clear, Conant rejects what one could call the ‘commentarian’ approach, as if one had to look at the text with a pen in hand and “begin to introduce a little subscript for each occurrence of the term ‘intuition’, clearly indicating which sort of intuition is Kant’s topic where” (2016:95/2017a:250–1). This would indeed be a silly practice, but I believe no one really believes that we must interpret the term Anschauung differently at each point where it appears in Kant’s text, and accordingly make definitional distinctions, which are then rigorously abided by throughout the text of the Critique—or as Conant puts it: “Es sieht so aus, als ob die Aufgabe des verantwortungsvollen Kommentators von ihm verlangt, überall Unterscheidungen einzuführen, wo Kant dies versäumt hat” (2017a:251). However, I do believe we have to be careful not to think that instead of doing so we can blithely disregard the variant contexts in which the term Anschauung appears, and that Anschauung always and unambiguously means ‘intuition’ without noticing if this refers to an appearing object (as opposed to a thing in itself) of which I have such and such knowledge or just means an aggregate of representations, any arbitrary intuition of something undetermined (cf. B34/A20). Philosophical language is not an exact science (though some would like it to be), so there is undoubtedly a need for hermeneutics, whether one likes it or not, also, and I’d say especially, in the case of Kant’s notoriously elusive text called the Transcendental Deduction.

I should like to note though that I share Conant’s general sentiment here because all too often, and increasingly, Kant commentators show a penchant for filigree classification and definition, which runs the risk of slicing up and compartmentalising Kant’s reasoning rather than explaining it. With such a strategy, which models itself after a formalist conception of philosophy, one often fails to see, or plainly misjudges, the bigger picture. However, a sentiment does not an argument make. Moreover, contrary to what Conant suggests, Kant himself often differentiates between indeterminate and determinate intuitions or manifolds of representations, most explicitly at a crucial juncture in the B-Deduction at B160–1, including its notorious footnote.[14] And Kant also distinguishes between intuition and the unity thereof (see also e.g. the capitalisation of “Einer” at B143 and B144–5; this is discussed further below).

Nevertheless, to take the way the Deduction and the Critique in general is written at face value and not to try to rewrite it by trying to foist clear definitions on it, whenever it seems that Kant is contradicting himself, is a point well-taken, particularly also because Kant’s transcendental philosophy does not work according to the procedure of first providing clear definitions and basing its subsequent arguments on them—a fundamental point that so-called ‘analytically oriented’ readings of Kant often seem to forget. (I often have the impression that Kant scholars approach Kant’s transcendental philosophy as if it were a scholastic enterprise such as Wolff’s, with its neat and expansive classificatory schemas and analytical definitions, which only need to be mapped rigorously, ‘analytically’, in order to understand his thought.)

Conant is entirely right to warn against reading the Deduction (or the Critique in general) as if it concerned a mathematical treatise, starting with definitions, on which what follows necessarily depends. Rather, as he says, the concepts employed and delineated by Kant find their complete expression only at the end of a dialectical process of philosophical reflection, having completed the task of the Critique (2016:96–7/2017a:253). One should of course be careful not to go to the other extreme of analytical filigree, namely, what one could call interpretative globalism, in virtue of which the text is read in conformity with a particular broadly defined view one has in mind to which the text must be adapted, rather than letting it speak for itself, warts and all (i.e. with its apparent contradictions).

But having said that, an Anti-Two-Stage reading, the kind that Conant propagates, is not forced on us (nor of course is disagreeing with Conant on this point necessarily proof of a Two-Stage reading, the one that Conant rejects). This becomes clear if we focus on his analysis of the notion of Anschauung. Conant writes the following: 

[The notion] “intuition” […] is initially glossed by Kant as “immediate singular representation” […]. It is then shown over a great many pages that the initial gloss is not to be understood as the characterization of a self-standingly intelligible form of representation—that its very possibility requires the involvement of a capacity whose exercise cannot be restricted to the production of such representations. (2016:97/2017a:253–4)

Conant subsequently makes it clear that the capacity at issue here is the intellectual capacity, which is also involved in the production of intuitions. But I don’t think this can be right: intuition does for its production or possibility not depend on intellectual conditions or the participation of the intellectual capacity (cf. B122–3). I do not think that Kant argues that the involvement of the understanding is a necessary condition on the very possibility of intuitions as such.

Conant reasons that either intuition is blind, and so not, in contrast to nonconceptualist interpretations of Kant, already referring to an object, or an intuition concerns a representation with objective purport that can only be specified with the help of the understanding (2017a:254). But this omits a third possibility, which the nonconceptualist position alludes to, namely to specify what it means to have a blind intuition. A blind intuition is still immediately related to an object but blindly. A blind intuition is not an oxymoron. The Kantian nonconceptualist, at least in one sense of nonconceptualism, tries to specify the counterfactual situation in which the conditions under which the relation is objectively valid fail to apply, but there still is an intuition of an object in some other sense.[15]

4.4 One or Two Kinds of Unity?

For the fourth “choice-point” Conant points to Kant’s distinction between the subjective and objective unities of consciousness, only the latter of which has objective validity, according to Kant (B139–40). Conant argues that one should not read Kant as if it were possible to pass from a merely (Humean) subjective unity of consciousness, which is only subjectively valid, to an objective unity of consciousness by adding something to the connection of representations that provides them an element of objectivity (Conant 2016:98/2017a:255).

Conant refers to Lewis White Beck’s distinction between L and K-experience (Beck 2002), reflecting the apparent ambiguity in Kant’s notion of experience (2016:98–9/2017a:255), where L-experience stands for ‘Lockean experience’ and K-experience for ‘Kantian experience’. Lockean experience is equivalent to a merely subjectively valid unity of consciousness, whereas K-experience is equivalent to Kant’s objective unity of consciousness. According to Beck the goal of Kant is to change L-experience into K-experience. I do not think this is what Kant intends or does; while I agree there is a difference between both, the goal is not to change L- into K-experience. L-experience reflects the de facto givenness of our mental states and how representations are prompted in the mind in accordance with the necessary form of inner sense, time; apart from the explanation given in the Aesthetic, this givenness of states and representations needs as such no further transcendental explanation—L-experience just is the aggregate of sensations one has in any arbitrary manifold of representations in intuition. The real explanandum in Kant’s account in the Deduction is K-experience, only its conditions are examined, not the conditions of L-experience, nor the conditions of the latter being changed into K-experience. L-experience and K-experience are not to be seen as two different kinds of experience; rather, L-experience is the sensory state or states which we would be in, were the conditions necessary for K-experience not to be satisfied. (And this is not just a hypothetical possibility, it is a real metaphysical possibility. See further below.)

This ties in with Conant’s observation, with reference to A111–12, that L-experience is not a distinct form of experience, which could exist without the categories. To an extent, one could have sensations, sensibly apprehend indeterminate objects without the categories (the possibility of which Conant however rejects) but this would indeed not be a temporally separable kind of experience in the strict sense; it is just that having sensations is something which we must assume for a fact even to be able to talk about K-experience, while it is not something that necessarily entails K-experience (or the transcendental conditions on K-experience, for that matter). There is no sense in which a temporal transitional process from L-experience to K-experience is at issue, and to this extent I concur with Conant (for a detailed account, see Schulting 2017a, ch. 7).

Conant also mentions Beck’s reference to the often discussed passage at A90–1/B122–3, where Kant considers the apparently hypothetical case that our appearances do not conform to the functions of thought or the categories. It is a pity that Conant does not quote the recent literature on the interpretation of this passage (see Ginsborg 2008, Allais 2016, Allison 2012, Gomes 2014, Golob 2016, Schulting 2015b, 2017a). Conant says that there are two logically possible options, but only one is really possible. The possibility of the conditions of sensibility not corresponding to the conditions of thinking (the categories) is not after all a real possibility, Conant says (cf. Gomes 2014); and he calls this, somewhat cheekily, the “Phew! reading” (2016:101/2017a:258) of §13 of the Deduction:

It is to read Kant as seeking to show that the possibility that is entertained here in §13 [i.e. the passage at A90–1/B122–3] is to be unmasked as a merely apparent possibility. (2016:101 /2017a:259)[16]

Conant points out that the B-Deduction was written precisely to make it as clear as possible that “the Transcendental Aesthetic does not present us with a separate and independent condition for objects to be given to our senses” (2016:110/2017a:263). This seems indeed Kant’s goal if we look at the passage in §21 of the B-Deduction, where he says that he is going to examine how things are given in sensibility in the so-called ‘second step’. But note that the discussion in the ‘second step’ consistently concerns objects of perception; so the emphasis should lie on objects, not as mere objects of sensible apprehension but as determinate objects of which we have experience, about which we make judgements and express beliefs (for textual evidence and how to gloss it, see Schulting 2017a, ch. 7). Significantly, Conant points to B143, the transitional section between the first and second steps of the B-Deduction:

Also ist alles Mannigfaltige, so fern es in Einer empirischen Anschauung gegeben ist, in Ansehung einer der logischen Funktionen zu urteilen bestimmt, durch die es nämlich zu einem Bewußtsein überhaupt gebracht wird.

Conant comments:

This looks to say that in order to meet the conditions on something’s so much as being an intuition it must be subject to the categories. So it now looks as if the following idea is to be rejected as well: the idea of something’s being subject to the conditions on intuitional unity while not being subject to the conditions on categorial unity. (2016:107/2017a:259)

As Conant notes (but only in the English version of his text; 2016:107), unusually, Kant here capitalises the inflected indefinite article Einer. Why? For it concerns unitary intuition, not just the constitution of any arbitrary intuition or its existence condition, and this is at variance with Conant’s reading, who seems to think that intuition is always already a unitary intuition—in the above quotation “conditions on something’s so much as being an intuition” is elided into “something’s being subject to the conditions on intuitional unity” (my underlining). But Kant’s employment of the limiting conjunctive sofern indicates this is not the case: it is only to the extent in which the manifold in an empirical intuition is united, that the manifold of that intuition stands under the functions of thought and is determined in accordance with them, that is, subject to categorial unity. John McDowell is right here: “Intuitions just happen, outside the control of their subjects. But their unity is intelligible only in the context of apperceptive spontaneity” (2009:72; emphasis added); though this might just be taken to mean that the intuition is still, qua unified intuition, subject to the categories.[17] And indeed McDowell’s reasoning is in fact similar to Conant’s. McDowell writes the following:

The B Deduction is framed to avoid a certain objection. Kant wants to establish that experience has its objective purport by virtue of being informed by the pure concepts of the understanding. The objection is that that ensures only thinkability. But a condition for objects to be thinkable is not thereby a condition for them to be capable of being given to our senses […] Kant organizes the B Deduction so as to forestall this objection. The essential move is to deny that the Transcendental Aesthetic offers an independent condition for objects to be given to our senses. […] The unity constituted by conformity to the requirements of our sensibility, which is the unity of the pure formal intuitions of space and time, is not a separate unity, independent of the unity that consists in being informed by the categories. (2009:73–4; boldface added)

In a very recent essay, McDowell argues:

In the second part of the Deduction […] Kant needs to exclude this apparent possibility of presence to my senses that, because it does not conform to the requirement of synthetic unity, does not count as presence of objects to me. He needs to show that synthetic unity is already a condition for presence to my senses, not a merely additional condition for what (illicitly on this view) I count as presence of objects to me. (2017:317; boldface added)

McDowell reasons that “cases of sensory consciousness […] are describable as presence of objects to subjects” (2017:317), which as such entails the involvement of categorial unity; “presence to our senses without categorial unity” would not be “presence to us”, or “presence of objects in a demanding sense” (2017:317). Like Conant, McDowell argues that the ‘second step’ of the Deduction excludes the possibility of mere “presence to our senses without categorial unity”, presumably because he thinks that “presence to our senses” eo ipso entails categorial unity in virtue of its describability alone. But too much weight is put on the sense of the possessive pronoun “our” if it is read in the strict sense, as McDowell seems to do, as already entailing the subjective agent’s identity that is first introduced by the categories, which is question-begging: the sensing subject is not ipso facto a thinking subject.

Analogously, Conant writes that “[t]he second half of the B Deduction aims to show that the formedness of our sensibility, treated in the Aesthetic, cannot be in view fully independently of the form of apperceptive spontaneity, treated in the Analytic” (2016:112/2017a:264; emphasis added). The emphasised phrase adds some ambiguity to Conant’s statement. Does he mean that the form of sensibility cannot be examined separably by the apperceiving agent of cognition—or, by the philosopher who carries out or reflects on the analysis of the Deduction—or does he mean that the form of sensibility (i.e. its unity) is no other than the form of apperceptive spontaneity? Only the second option refutes the two-stage-model. But the fact that the philosopher who analyses the possibility of “presence to the senses” himself must be able to describe this in virtue of his (the philosopher’s) capacity to apply categories does of course not imply that the sensible subject to whose senses something is presented—the object of the philosophical description—must be able to describe this presence to himself (the sensible subject) in virtue of the capacity of a thinking subject. Conant says further:

So the crucial step involves showing that the unity constituted by conformity to the requirements of our form of sensibility, which is the unity of the pure formal intuitions of space and time, is not an utterly separate form of unity of manifold—one that could be in place altogether independently of the sort of unity of manifold that consists in our perceptual experience’s being informed by the categories. (2016:123–4n.54/2017:264–5n.26)

McDowell says likewise:

Synthetic unity is presupposed already in the requirement of spatial and temporal order that first came into view, in the Aesthetic, as required by our sensibility. There is only one unity, the synthetic unity that is intelligible only in terms of the unifying power of the spontaneous intellect. And the apparent possibility of presence to the senses that is not presence to the understanding is unmasked as the mere appearance of a possibility. (2017:318)

According to Conant and McDowell, the unity of sensibility, the unity of the pure formal intuitions of space and time, is thus no other than the unity that is prescribed by the categories. To be sure, Conant rejects a reductive reading, i.e. the reading that says that intuitional unity is just a conceptual unity, but other than rejecting a reductive reading he does not clarify wherein the correspondence between intuitional and conceptual unity does consist.

Of course, it is not at issue in the Deduction whether the unity of intuition is reducible to a conceptual unity or isolatable or a completely self-standing unity let alone “completely identical” to a conceptual unity; no serious conceptualist reading of the Deduction would claim that.[18] But the suggestion that the correspondence is not a complete identity is non-explanatory. One must be able to explain wherein this correspondence consists and this implies an understanding of the extent and modality of the involvement of apperception in the constitution of this correspondence, which at the same time explains the dependence, and the extent of this dependence, of the unity of intuition (space and time) on the unity of the understanding. It is not very helpful saying that the same unity of the understanding is the unity of sensibility without the latter being reducible to the former (which is trivially true), without thereby indicating how the shared original synthetic unity comes about. This unity is not simply given in sensibility, and so not to be identified with the given unity of space (and time), but due to an act of the understanding, which in some sense needs to be added (cf. §15) to the manifold, since it is not contained in the manifold as such.[19] This unity or combination comes about through an act of the self (apperception) who takes the particular manifold before her as unified, namely as her combined manifold. This is the subjectivism that I would argue is necessarily involved in establishing the unity of sensibility in the sense of a necessary synthetic unity in virtue of an original act of synthesis by the understanding, which produces spatiotemporally bounded objects.

This is not a case of literally—and this is what Conant rightly dismisses as a possible reading of Kant—a posteriori imposing (in time) a unity on a pre-given (in time) matter or manifold. This cannot be the case, since the self-activity of the understanding unifying the manifold is an a priori act. The unifying act is a priori because (1) this is required for any manifold to be a determinate, objectively valid representation, and (2) if it were not a priori, but an a posteriori ‘imposition’, it would be difficult to fathom how the sensible manifold and the act of synthesis by the understanding were in fact necessarily connected (this hangs, of course, on Kant’s identification of ‘necessary’ and ‘a priori’ [B4]). A necessary connection cannot be established or determined by just successively apprehending manifolds of representations or looking at how such manifolds are more or less randomly aggregated in the mind.

Reason (2) points to regress problems, which I have frequently indicated in my own papers dealing with nonconceptualism (Schulting 2010; 2015b; 2017a, ch. 5). The regress issue is the following: given that a priori synthesis is an original combinatory activity that connects two different kinds of cognition, sensibility and understanding, at the fundamental level at least in one sense, namely, with respect to their shared unitary character, if a priori synthesis, as a unifying act of the understanding, as Kant claims it is (§15), were an ‘additional’ imposition on a ‘pre-given’ manifold, then the question arises as to how the supposedly imposed synthetic unity itself is a priori synthesised with the pre-given manifold. This would putatively be an a priori synthesis*, namely the kind of a priori synthesis that connects a priori synthesis with a sensible manifold. But then the question would arise as to how a priori synthesis and a priori synthesis* are a priori synthesised. Would it be an a priori synthesis** that a priori synthesises both a priori synthesis and a priori synthesis*?

It is easy to see that this is going nowhere and makes a mockery of the notion of ‘a priori’: a priori synthesis was precisely designed by Kant to avoid these kinds of regress, i.e. it should serve as a regress blocker. A priori synthesis is a synthesis, carried out by the thinking self (the understanding), more original or more fundamental than which there is none (on Kant’s account, at least). There is just one kind of original or a priori synthesis, and it is the one that binds sensibility and understanding at the fundamental level, insofar as their shared necessary unitary character is concerned, namely the unity that is necessary for possible knowledge. The ‘insofar as’ clause indicates that sensibility and understanding are irreducible: there is still a sense in which sensibility is independent of the understanding, namely insofar as it is a real possibility that there is no occurrent subject of understanding, a judging subject, that actively takes a sensible manifold as together constituting a unified intuition that is a condition on possible knowledge.

This does not mean that on my reading to deny McDowell’s rejection of the real possibility of “rogue appearances” (McDowell 2017:318–19) means to claim that “there may be more to a reality that is empirical, in that it impinges on our senses, than what we can judge” (2017:319). What we can judge is all that there possibly is about empirical reality, which sensibly affects us; there is no in principle gap between what we can judge and what there is in empirical terms. To deny rejecting the real possibility of what McDowell terms “rogue appearances” just means to deny that any presence (of appearances) to the senses must be a presence to the understanding (and judging) subject because otherwise, as McDowell says, “we would not even be entitled to the idea of presence to the senses” (2017:319), for the ground that McDowell gives here confuses the empirical level with the transcendental level of explanation: there is no requirement for presence to the senses to be a real possibility that some thinking agent entertains the idea of presence to the senses and apply it in her own case.

5. Two Kinds of Synthesis? On How to Read the ‘Leitfaden’ Passage (A79)

Conant’s own portrayal of a priori synthesis is somewhat less felicitous, as it seems to ignore the active aspect of the original synthetic unity and thus to underplay the identity of the a priori synthesis as being a subjective “action” or “function” in virtue of transcendental apperception (A108) that binds the unity of intuition and the unity of the understanding. He also seems to be talking about two distinct kinds of such synthesis, which to me seems at least misleading, and potentially problematic for the same reason that I gave above with respect to regress issues. Conant writes with reference to the Leitfaden at A79/B104–5[20]:

This form of unity—categorial unity— characterizes both the manner in which objects are given to us in intuition and the manner in which concepts are combined in judgments. To say it can be in act in these two different ways is not to claim that the two sorts of synthesis in no way differ from each other [in the German version, the last clause reads: “…die beiden hier thematisierte Formen der Synthesis bloß zu identifizieren”, D.S.]. (2016:114/2017a:266)

But this begs the question as to the extent in which both forms of synthesis, which unite intuitions and concepts respectively, are non-identical and whether this view does not conflict with Kant’s claim that it is the same function (of the understanding, i.e. the function of synthesis) that operates on both levels, namely on the level of intuition and concepts. It seems to me that Conant underplays the identity of the act of synthesis that first constitutes the “categorial unity” on both levels. He writes further:

Hier liegt eine Form der Einheit vor, die auf zwei unterschiedliche Weisen auftauchen kann.[21] I take this to be the point of the following famous passage: “The same function which gives unity to the various representations in a judgment also gives unity to the mere synthesis of various representations in an intuition” (A79/B104–5). This “function” grasped in its most abstract form is the original synthetic unity of apperception. This is Kant’s most abstract characterization of the unity of the understanding. […] A synthesis of concepts in a judgment is one way of making this highest form of unity more determinate in cognition; a synthesis of a manifold into an intuition is another way of making this highest form of unity more determinate in cognition—both presuppose the involvement of the understanding. (2016:114/2017a:266; boldface added)

The last sentence is important, but Conant does not explain how it is possible that one function (the understanding) enables two different ways in which unity is brought about in the concepts and in the manifold of intuition of a particular judgement o is F, say. It looks like he has merely moved the dualism between sensibility and the understanding to a different level. That is to say, now it apparently concerns a dualism between one kind of synthetic unity and another kind of synthetic unity, supposedly between a synthesis of concepts and a synthesis of sensible manifolds in intuitions. But there are no two kinds of synthesis, rather there is just one synthesis in virtue of one original act of a priori synthesis in a judgement, and so one kind of function of synthetic a priori unity which as such unites both concepts, by means of analytic unity (concepts are only analytically united among each other), and empirical intuitions, by means of synthetic unity (cf. the second sentence in the Leitfaden passage at A79).

Importantly, the original synthesis does not create two synthetic unities on two separate or separable levels. It creates one synthetic a priori unity, which holds two different kinds of representations in a judgement together: namely a unity between concepts (F and G, say), on the one hand, and the manifold of representations in an empirical intuition, on the other. There is no synthetic unity on the conceptual level as such, in abstraction from the connection to the manifold in the intuition. The analytical unity among concepts, which, in its most basic form, is always a relation of subordination or a relation of genus and species, is rather grounded on a synthetic unity in the content of representations, whether they be intuitions or further concepts—this synthetic unity enables the combination of two different concepts F and G.

This is so even if the combination merely amounts to an identity relation of sorts between F and G, whereby reference to the content of an underlying intuition is otiose, such as in analytic judgements, for example the relation between <bachelor> and <unmarried> in the judgement “a bachelor is an unmarried man”: these are not the same concept, clearly, but their relation in the analytic judgement “a bachelor is an unmarried man” is based on the knowledge that there is an equivalence between the use of the two concepts <bachelor> and <unmarried>. Their similar use under similar conditions makes that a priori synthesis or an original synthetic unity is necessarily involved in being able to make the true judgement that “a bachelor is an unmarried man”, but no reference to the intuition of an actual bachelor is required, and nor is the identity relation between the concepts as such strictly speaking a synthetic unity (in the sense that Kant means); rather, that relation is purely analytic.

Of course, that does not mean that all conceptual relations are analytic judgements; it just means that all conceptual relations, no matter what, exhibit merely as such an analytic unity, which is made possible, implicitly, by a synthetic or combinatory act of understanding that establishes that they belong to each other in some specific sense (whether they form synthetic or analytic judgements). Syntactically, a judgement might be said to show a kind of synthesis of concepts, namely in the way that a concept F is combined with a concept G (or how two or more judgements are combined to form a disjunctive judgement, say), but this is of course not the a priori synthesis Kant talks about. There is nothing a priori synthetic about how a judgement looks on the surface. (Hence, Kant says at B140–1 that purely looking at the conceptual relation in a judgement does not explain what a judgement is.)[22]

The point is not to deny two distinct/differentiable kinds of unity but to explain how and, crucially, to what extent the two kinds of unity share a common characteristic, namely the original synthetic unity of apperception that is at issue in the Deduction, specifically in the Leitfaden passage at A79. Nor can there be a case—and I suspect Conant believes this, given his emphasis on the differentiation of two kinds of synthesis on two levels—of the distinction between figurative synthesis and intellectual synthesis that Kant makes at B151 as supposedly mapping one-to-one on the distinction between the intuitional and conceptual levels respectively. Often the intellectual synthesis is identified as having to do with how mere concepts are bound together. But this is misleading. Intellectual synthesis is the synthesis of the content of any manifold of representations in abstraction from the spatiotemporal nature of that manifold. So intellectual synthesis, as much as figurative synthesis, figures below the line in a synthetic objectively valid judgement S is P, whereby the line is the imaginary division between the conceptual level (“S” and “P”) and the level of intuition(s) of an object underlying the subject term of a judgement S is P. Conant seems to espouse a different view:

[S]ince the categories inform the exercise of our sensory as well as our judgmental capacities, this requires that we view both a synthesis of a manifold into intuition and a synthesis of concepts into a judgment as not only involving at some level of abstraction the same form, but as involving at any level of specification a fully and determinately identical form—so that the form of perceptual experience must be in all respects identical to the form of judgment. […] [A] move that appears to license the conclusion that the form of our sensory consciousness as such for Kant is always already judgmental in character. (2016:115/2017a:267; boldface added)

The way Conant puts this is at least misleading: Kant does not say that there is a synthesis of concepts into a judgement. It is rather the one act of synthesis (the function of the understanding) which unites concepts, “by means of analytic unity”, and also unites the bloßen Synthesis in intuition (B105),[23] “by means of synthetic unity”. There is no synthetic unity of concepts “above the line” (the terminology is from McDowell 2009:5). Concepts are as such strictly speaking merely analytically united, as I argued above and just as Kant says in the Leitfaden that the understanding “in Begriffen, vermittelst der analytischen Einheit, die logische Form des Urteils zu Stande brachte” (A79/B105; emphasis added). The phrase “vermittelst der analytischen Einheit” is significant. The logical form of a judgement is just expressive of the surface grammar of how concepts are related as species and genus; of course, they are syntactically bound together, but this is not the a priori synthetic unity that Kant means as being the transcendental ground of a judgement as a whole, i.e. both its logical form and its intuitional or, more precisely, transcendental (=pure intuitional) content (cf. beginning of §13).

Contrary to what Conant seems to suggest, it is not as if the a priori synthesis, which is what is meant by the “same function” of the understanding, created a synthetic unity above as well as below the line, where above the line is the conceptual space or form, which is shown by the syntactical structure of a judgement S is P, and below the line is the intuition of some x, of which is predicated both S and (through SP. Rather, the a priori synthesis that is the set of the logical functions of the understanding creates the form of a categorical judgement (as the basic form of any of the 11 other generic types of judgement), which happens by means of analytically subordinating concepts (S under P), whilst thereby synthetically (“by means of synthetic unity of the manifold in intuition”) combining the content of such a judgement, i.e. representing whatever is intuitionally represented under S as also represented under P.

So, the relation purely between the concepts <cup> and <chipped> in the judgement “This cup is chipped”, say, is analytic—as any mere relation of concepts is—namely purely qua logical form in which concepts are syntactically bound in the categorical judgement, as subject and predicate. The concept <cup> is placed within the extension of the concept <chipped>; that is, the thing of which is predicated <cup> belongs to the domain of all possible objects of which <chipped> is predicable, whereby, if we abstract from the connection with the thing judged about, the relation purely between the concepts <cup> and <chipped> is and remains analytic.

Again, this does not mean that the judgement is analytic, as “This cup is chipped” is obviously not an analytic judgement. (Klaus Reich long ago pointed out that it is a mistake to think that the distinction between analytic unity and synthetic unity, which Kant makes in the Leitfaden, maps onto the distinction between analytic and synthetic judgements.) An analytic judgement can be understood purely by looking at the formal relation between the concepts, i.e. their analytic unity; one can ignore the transcendental ground of this relation, for the reference to an underlying object (x) is otiose for the insight in this formal relation (cf. B133–4 note). However, for any non-analytic, that is, synthetic, judgement, one needs to look beyond the formal, analytical, relation between concepts. Here is where the a priori synthesis comes in most crucially,[24] which makes it possible to understand how concepts have some objectively valid reference to an intuited object of experience outside the conceptual context; if concepts fail to have relation to an object of experience, and cannot be explained via conceptual analysis otherwise, they lack objective validity, and though internally consistent are thus speculative at best.

The moral of this story is that the purely conceptual relation between concepts considered as such is not a synthetic unity, though it is grounded on an a priori synthesis. This thus goes against the view, espoused by Conant and others, that the synthesis of the understanding creates two kinds of synthesis on both the conceptual and intuitional levels. In my view, as I hope I have shown, it doesn’t. There is just one kind of a priori synthesis that unites representations on both the conceptual and intuitional levels uniquely by means of analytic unity and uniquely by means of synthetic unity respectively.

6. Concluding Remarks

Conant’s reading of the B-Deduction argument assumes a modally absolute construal of the relation between sensibility and understanding, whereby no sensible episode is not informed by the understanding (or at least by one kind of synthesis, as Conant suggests). But this reading is not forced upon us, and as I have tried to explain above, I believe it is the wrong reading. Rather, on the subjectivist reading sensibility and understanding are a priori linked in virtue of an original synthetic of apperception if and only if the understanding takes them to be so linked, and this happens in judgements only. This thus does not imply that sensibility itself is propositional or conceptually laden or informed, or that the understanding is involved in the production of intuitions per se; sensible content or an intuition is conceptually laden, so to speak, only to the extent that it forms the content of an empirical judgement about a spatiotemporal object that I perceive as bounded in space etc. The argument in the Deduction concerns an analysis of the possibility of knowledge in terms of possible judgements, given the fact of such knowledge. This reading commits one to a moderately conceptualist take on the relation between sensibility and the understanding, insofar as only the two together, in a judgement, constitute cognition or knowledge, and it suggests a residual nonconceptualism about intuitions, for intuitions are never as such, independently of how the understanding takes its relation to intuitions, conceptually informed.

On this reading, however, the understanding does still determine sensibility inwardly, to use a Pippinian phrase, so my reading is a ‘conceptualist’ reading, namely insofar as the categories as pure concepts are the necessary ways in which such determination takes place. This way, one avoids the conundrum of Conant’s transformative reading, namely how to combine a rejection of empiricist dualism whilst not succumbing to a rationalist conceptual reductionism. For sensibility is conceptually or categorially determined or informed only insofar as the epistemic agent, the subject of understanding, takes it to be so determined. The added benefit of this reading is that the passages in Kant which, to nonconceptualists, seem to point to Kant’s nonconceptualism can indeed be read as supporting some minimal form of nonconceptualism but at the same time as wholly compatible with a mainstream conceptualist reading of the dialectic of Kant’s chief argument in the Deduction.

Acknowledgements: Thanks are due, as usual, to Christian Onof for his helpful comments on an earlier draft of this essay, and to Jacco Verburgt for proofreading the penultimate version.

Received: 1 September 2017; proofread: 24 September 2017.


[1] A longer English version of Conant’s paper was published in the journal Philosophical Topics (it says that it is published in 2016, but to my knowledge the article came out only this year), under the somewhat generic title ‘Why Kant is not a Kantian’, and a shorter version appears in the volume Kant’s Critique of Pure Reason: A Critical Guide, edited by James O’Shea (Cambridge UP, 2017), under the more apt title ‘Kant’s Critique of the Layer-Cake Conception of Human Mindedness in the B Deduction’ (Conant 2017b). When I wrote this notice, I relied on the German published version, which came out last January in a Suhrkamp volume (see Conant 2017a). Insofar as the German version and the longer English version (2016) overlap, in most cases I’ll provide the English quotations from Conant (2016) only and the German references after the slash.

[2] Conant calls it thus in the German version of the article, on which I relied. In the English version he refers, somewhat confusingly, to both the conjunctivist and disjunctivist conception of the relation between sentience and sapience. It is however the latter which is the “additive” view that he rejects.

[3] In the longer English version of the article, Conant talks about “a disjunctivist conception of the relation between sentience and sapience qua cognitive capacities” (2016:79), which he contrasts with his own “conjunctivist” conception of that relation.

[4] This distinction comes out better in the German version.

[5] Boyle also has an essay in the volume in which Conant’s piece appears. See Boyle (2017).

[6] I do not distinguish between the terms ‘knowledge’ and ‘cognition’, as is sometimes standard, for as said above we are dealing here with Kant’s transcendental account of knowledge itself, not merely with how we, as epistemic agents, have a capacity for knowledge let alone with actual instances of (empirical) knowledge and all kinds of non-transcendental conditions for achieving it, but also with how objects themselves are first capable of being known. Kant’s account concerns possible knowledge. The suggestion is often that Kant’s account does not or cannot deal with the objects of our cognition, but this is just mistaken (see Schulting 2017a, chs. 2, 3, 4).

[7] See Schulting (2017a), chs 3 and 4. Cf Conant (2016:83): “The Kantian problematic is concerned, in the first instance, not with the distinction between truth and falsity, but with what it is to stick your neck out in thinking, which Kant calls the objective validity of judgment, with what I will sometimes call the objective purport of judgement.”

[8] Cf. Conant (2016:83–4/2017a:238ff.).

[9] In the English version of the article he quotes Guyer (Conant 2016:87).

[10] See the two paragraphs ending section V in the English version of his article (2016:89), which are not in the German version.

[11] But notice that the German original at B131, hinzukommen, does not have the active connotation of ‘adding’, as if some thing or agent has to literally tack something on from outside; hinzukommen just points to a transcendental-logically required extra element.

[12] And indeed quite a few commentators hold such a view or a view like it. See e.g. most recently Dyck (2017).

[13] So have I argued in Schulting (2012a) and (2012b). Cf. McLear (2011).

[14] See the account by Onof and Schulting (2015).

[15] I have argued elsewhere (Schulting 2015b; 2017a, ch. 5) that most Kantian nonconceptualists however believe that the reference to an object in an intuition, independently of the categories, is already in some way objectively valid, and I argued that this reading is mistaken—for Kant, objective validity is a function solely of the categories and judgement.

[16] Cf. my discussion in Schulting (2017a), ch. 5.

[17] Cf. Land (2014) and my discussion of Land in Schulting (2017a), ch. 7.

[18] See Onof and Schulting (2015) for more detail.

[19] Onof and Schulting (2015) distinguish between the unity that the understanding prescribes to the manifold in a spatial intuition for the latter to be a determinate intuition of a spatial object and the unicity of space itself, which is not a function of the intellectual unity that the understanding prescribes. The unity of determinate spatial wholes and the sui generis unicity of space as such (and the same holds mutatis mutandis for time) are irreducible and should not be conflated.

[20] I wrote earlier on the guiding thread passage at A79/B104–5 in Schulting (2017a:108–9, 203ff., et passim), specifically in the context of my account of McDowell’s reading of it (2017:203ff.), and more in detail in Schulting (2012b), ch. 5. My account below should be seen as an elaboration on that earlier account.

[21] The parallel passage in the English version is at p. 114 (“It is merely to claim that there is a level of description of form at which they have something generically in common”), but I find the German rendering clearer, since it includes a reference to unity, which is crucial in the discussion at hand.

[22] “I have never been able to satisfy myself with the explanation that the logicians give of a judgement in general: it is, they say, the representation of a relation between two concepts. […] it is not here determined wherein this relation consists” (B140–1).

[23] Kant means by this the empirical synthesis of representations in a sensible (spatiotemporal) manifold.

[24] This is not to say that a priori synthesis has no role in the grounding of analytic judgement (see again earlier above) or the principles of sheer logic. Even the possibility of general logic is based on a priori synthesis (see Schulting 2017b).


Allais, L. (2016), ‘Conceptualism and Nonconceptualism in Kant: A Survey of the Debate’, in D. Schulting (ed.), Kantian Nonconceptualism (London/New York: Palgrave Macmillan), pp. 1–25.

Allison, H. (1983), Kant’s Transcendental Idealism. An Interpretation and Defense (New Haven: Yale University Press).

——— (2012), Essays on Kant (Oxford: Oxford University Press).

Beck, L.W. (2002), ‘Did the Sage of Königsberg Have No Dreams?’, in L. W. Beck, Selected Essays on Kant, ed. H. Robinson (Rochester, NY: University of Rochester Press), pp. 85–101.

Boyle, M. (2017), ‘Wesentlich vernünftige Tiere’, in A. Kern & C. Kietzmann (eds), Selbstbewusstes Leben. Texte zu einer transformativen Theorie der menschlichen Subjektivität (Berlin: Suhrkamp), pp. 78–119.

Cassam, Q. (2007), The Possibility of Knowledge (Oxford: Clarendon Press).

Conant, J. (2016), ‘Why Kant is not a Kantian’, Philosophical Topics 44(1): 75–125.

——— (2017a), ‘Die Einheit des Erkenntnisvermögens bei Kant’, in A. Kern & C. Kietzmann (eds), Selbstbewusstes Leben. Texte zu einer transformativen Theorie der menschlichen Subjektivität (Berlin: Suhrkamp), pp. 229–69.

——— (2017b), ‘Kant’s Critique of the Layer-Cake Conception of Human Mindedness in the B Deduction’, in J. O’Shea (ed.), Kant’s Critique of Pure Reason: A Critical Guide (Cambridge UP, 2017), pp. 120–39.

Dyck, C. (2017), ‘The Principles of Apperception’, in G. Motta & U. Thiel (eds), Immanuel Kant: Die Einheit des Bewusstseins (Berlin and Boston: de Gruyter), pp. 32–46.

Ginsborg, H. (2008), ‘Was Kant a Nonconceptualist?’, Philosophical Studies 137(1): 65–77.

Golob, S. (2016), ‘Why the Transcendental Deduction is Compatible with Nonconceptualism’, in D. Schulting (ed.), Kantian Nonconceptualism (London/New York: Palgrave Macmillan), pp. 27–52.

Gomes, A. (2014), ‘Kant on Perception: Naive Realism, Nonconceptualism, and the B-Deduction’, Philosophical Quarterly 64(2014): 1–19.

Hanna, R. (2011), ‘Kant’s Nonconceptualism, Rogue Objects, and the Gap in the B Deduction’, International Journal of Philosophical Studies 19(3): 399–415.

Henrich, D. (1969), ‘The Proof-Structure of Kant’s Transcendental Deduction’, Review of Metaphysics 22(4): 640–59.

Land, T. (2014), ‘Spatial Representation, Magnitude and the Two Stems of Cognition’, Canadian Journal of Philosophy 44 (5–6): 524–50.

McDowell, J. (2009), Having the World in View: Essays on Kant, Hegel, and Sellars (Cambridge, MA: Harvard University Press).

——— (2017), ‘Rationalism Without Dogmatism’, in Z. Adams & J. Browning (eds), Giving a Damn. Essays in Dialogue with John Haugeland  (Cambridge, MA: MIT Press), pp. 311–28.

McLear, C. (2011), ‘Kant on Animal Consciousness’, Philosophers’ Imprint 11(15): 1–16.

Onof, C. & D. Schulting (2015), ‘Space as Form of Intuition and as Formal Intuition: On the Note to B160 in Kant’s Critique of Pure Reason’, the Philosophical Review 124(1): 1–58.

Schulting, D. (2008), ‘On Strawson on Kantian Apperception’, South African Journal of Philosophy 27(3): 257–71.

——— (2010), Kant, non-conceptuele inhoud en synthese, Tijdschrift voor Filosofie 72(4): 679–715.

——— (2012a), ‘Non-Apperceptive Consciousness’, P. Giordanetti et al. (eds), Kant’s Philosophy of the Unconscious (Berlin and New York: de Gruyter, 2012), pp. 271–302.

——— (2012b), Kant’s Deduction and Apperception (Basingstoke and New York: Palgrave Macmillan).

——— (2015a), ‘Transcendental Apperception and Consciousness in Kant’s Lectures on Metaphysics’, in R. Clewis (ed.), Reading Kant’s Lectures (Berlin and New York: de Gruyter, 2015), pp. 89–113.

——— (2015b), ‘Probleme des „kantianischen“ Nonkonzeptualismus im Hinblick auf die B-Deduktion’, Kant-Studien 106(4): 561–80.

——— (2016), ‘On an Older Dispute: Hegel, Pippin and the Separability of Concept and Intuition in Kant’, in D. Schulting (ed.) Kantian Nonconceptualism (London and New York: Palgrave Macmillan), 227–55.

——— (2017a), Kant’s Radical Subjectivism. Perspectives on the Transcendental Deduction (London and New York: Palgrave Macmillan).

——— (2017b), ‘Analyticity, Analytic Philosophy and Kant’s Synthetic A Priori: Comments on Robert Hanna’s Cognition, Content, and the A Priori’, Critique (May issue)

——— (2017c), ‘Apperception, Self-Consciousness and Self-Knowledge in Kant’, in M. Altman (ed.) The Palgrave Kant Handbook (London and New York: Palgrave Macmillan), pp. 121–43.

——— (2017d), ‘Gap? What Gap?—On the Unity of Apperception and the Necessary Application of the Categories’, in G. Motta & U. Thiel (eds), Immanuel Kant: die Einheit des Bewusstseins (Berlin/New York: de Gruyter), pp. 89–113.

——— (forthcoming), ‘Apperception and Object. Comments on Mario Caimi’s Reading of the Deduction’, book symposium on Mario Caimi, Revista de Estudios Kantianos 4(1) (May 2019).

© Dennis Schulting, 2017.

Dennis Schulting is a former Assistant Professor of Metaphysics and its History at the University of Amsterdam. His work concentrates on Kant and German idealism. His most recent journal publications were published in KANT YEARBOOKKANT-STUDIEN, the PHILOSOPHICAL REVIEW and KANTIAN REVIEW. He is the editor of Kantian Nonconceptualism (Palgrave 2016). His second monograph, entitled Kant’s Radical Subjectivism: Perspectives on the Transcendental Deduction, was published last July. Book symposia on the latter are forthcoming in Kantian Review (March 2018) and Tijdschrift voor Filosofie. Schulting is currently finishing his third monograph, entitled Reflexivity and Representation. Essays on Kant and German Idealism.