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 first 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 first instalment, Sacha Golob critically discusses James Conant’s recent article on the Transcendental Deduction, soon to be followed by an essay from Dennis Schulting and James Conant’s reply to both


By Sacha Golob

James Conant’s recent article, ‘Why Kant Is Not a Kantian’, offers a sophisticated and provocative account of the relationship between sensibility and understanding. It is also an account that I think is mistaken. One consequence is that Conant is unable to do justice to both the differences and the deep continuities that exist between us and non-rational animals. Kant’s own views in this regard, I argue, were both more flexible and more attractive.

Before proceeding, a disclaimer. Given the systematic nature of Kant’s thought, such arguments raise many related questions, textual and conceptual. For example, I am sceptical as to how much can be achieved without first establishing whether we should read Kantian perception in relationalist or representationalist terms, and that in turn begs the question of how we demarcate those approaches and how we view hybrid accounts such as Schellenberg’s (2015). Much will also hang on the details of how one parses notorious texts such as A99 or B160–1n, or how exactly one sees the proof structure of the Deduction. I cannot do justice to these issues here; if the reader is interested, I treat the topics in a recent and forthcoming series of papers.[1]

1. An Overview of Conant’s Position

Conant’s basic aim is to argue for a deep unity between understanding and sensibility. More specifically, his approach

requires seeing how each of these faculties depends on its relation to the other to be the sort of faculty that it is in a finite rational being. (p. 117)

Insofar as sensibility can operate independently of understanding, for example in non-rational animals (henceforth simply ‘animals’), it would thus have a radically different nature: the introduction of rationality, to borrow a helpful phrase from Boyle, has a “transformative” effect on intuition.[2] To put it another way, Conant opposes what he calls a “layer cake” model where sensibility remains unchanged irrespective of its relation to conceptual processes. He writes:

Just as in a layer cake with a lower layer of chocolate and an upper layer of vanilla[,] the fact that there is a layer of vanilla sitting on top of the chocolate does not affect the internal character of what it is to be chocolate. So, too, according, to the deep-seated assumption: just because, in the human case, there happens to be a layer of cognitive functioning, which involves “additional” capacities (say, the capacities to employ concepts and make judgments) sitting on top of our merely animal nature, does not alter or otherwise affect the internal character of the capacities which make up the lower level—the human animal’s capacities to be sensibly affected by and desire objects in the world. (p. 77)

As Conant notes, Boyle (2012) has launched a similar assault on such ‘tack on’ theories of rationality; readers familiar with the phenomenological literature will also recall Brandom’s (1999) rejection of ‘layer cakers’ who treat propositional intentionality as a merely ‘optional superstructure’.

The question of the links between understanding and sensibility has, of course, been central to the reception of the Critique of Pure Reason since its publication, and many of the canonical readers of that text, most obviously Hegel and Heidegger, have argued for a stronger form of inter-faculty unity than the standard ‘two stems’ story. As Conant acknowledges, his position might be sloganised, roughly but not falsely, as holding that ‘There is only one unity!’ (p. 112). I think Conant does a masterful job of refuting some crude objections to his view—surely Kant distinguishes between the unity characteristic of intuition and that characteristic of concepts?—and I shall not rehearse such worries here. Instead, I will press a quite different line of attack. To do that we need to sharpen up the claims under consideration.

Conant offers several formulations of his core thesis. The first I will call Alteration. Conant describes the ‘primary purpose’ of his paper as denying that

our faculty of sensibility could have the very same character that it presently does independently of its entering into cooperation with our faculty of understanding. (p. 76)

But we need to get clear on what the “very same character” amounts to: after all, even the most enthusiastic layer caker will accept that token intuitions in the layer cake have various modal properties, such as being potentially conceptualisable, which those in an untiered, purely sensible, structure lack. The problem is exacerbated by the fact that Conant at times makes what is prima facie a much stronger claim: that the form of sensibility when coupled with understanding is “utterly distinct” from that of sensibility taken alone (p. 79). Clearly two things might lack the “very same character” without being “utterly distinct”—they might have some similarities and some differences. Conant also glosses his view in terms of the ‘internal character of the manifold’. Here his opponent is defined by the following commitment:

What is crucial to the assumption is the following idea: that the internal character of the manifold constituting the bottom layer remains unaffected by the introduction of the upper layer. (p. 77)

But problems remain: we cannot simply say that the opponent believes, and Conant denies, that the ‘internal character’ of the manifold is unaltered when placed in relation to understanding. This is because if Conant is right, it becomes hard to retain a grip on the relevant sense of ‘internal’: the upshot of his position would be that what was thought to be internal is in fact relational, a function of a relation to another faculty. Given these difficulties in formulating Conant’s position, let us start simply from the following rough formulation and develop it as the argument progresses:

(Alteration) The intuitive character of the manifold experienced by subject s is necessarily altered if s possesses conceptual capacities.

Conant endorses Alteration, whilst the classic ‘layer caker’ would deny it.

The second main formulation offered by Conant is in terms of the relationship between us and animals. He initially makes the point in methodological form. His opponent believes and Conant denies that

we can start with a conception of that creature’s capacities and arrive at a fully adequate conception of “the human capacity to acquire knowledge” (what Kant calls our Erkenntnisvermögen) by simply supplementing its repertoire of capacities with those that figure in ours but not in its. (p. 78)

This methodology is closely tied to the original ‘layer cake picture’: the opponent assumes that he or she can simply plaster the human top on to the animal base, a practice which in Conant’s eyes will yield only an incoherent chimera. Now, at this point, a natural question arises—surely Conant acknowledges some overlap between us and animals? After all, Kant does:

[A]nimals also act in accordance with representations [Vorstellungen] (and are not, as Descartes would have it, machines), and in spite of their specific difference, they are still of the same genus as human beings (as living beings). (KU, AA 5:464)

Conant’s response is that talk of overlap is not entirely prohibited. Rather, his claim is that any appeal to overlap must be purely ‘generic’:

At the level of a formal characterization of what it is to have a faculty of receptivity, qua cognitive capacity, there is nothing which may figure as a highest common factor across the capacities of two creatures only one of which is a rational creature. On the conception which I will attribute to Kant, if the two creatures here under consideration may be said to have a capacity in common, it is only because of their “having something in common” at a very generic level of description—a level of description of “the” capacity in question which completely abstracts from the manner in which it specifically figures in the exercise of their respective forms of cognition. (p. 79)

Conant frames the issue here in terms of specific vs abstract levels of description, but that doesn’t seem to get his position quite right. Suppose, for example, I thought that the abstract specification of the shared capacities did all the explanatory work, and that the specific ways in which it was realised marked fairly trivial differences: after all, there are countless specific differences between the way in which the understanding figures in my cognition and in yours (you have concepts I lack, you will be synthesising different manifolds and so on) and yet the whole Kantian project is founded on the idea that some abstract characterisation, emphasising the structural similarities, should be privileged. So, a position on which humans and animals had something in common only at a very generic level of description, but where this generic level also did all the explanatory work would surely run counter to everything Conant wants. This is one reason, I suggest, why Boyle is right to frame the issue in explanatory terms, rather than in terms of ‘specific vs abstract’:

The claim that rational animals have a distinctive kind of animal mind thus implies that rational capacities for perception and desire cannot be explained as: the kinds of capacities for perception and desire to be found in nonrational animals, supplemented with a further, independent power to regulate these capacities in the light of reflective reasoning. (2012:424)

Given this, the following seems a fair summary of Conant’s view:

(Generic) If humans and animals may be said to have a common sensibility, it is only because this “having something in common” holds at a sufficiently generic level of description so as to play only a limited explanatory role.

We now have a sharper formulation of Conant’s position: he endorses both Alteration and Generic.

From these claims, as Conant eloquently shows, multiple consequences follow. To take a textual example, Alteration implies that the Aesthetic is at best a preliminary analysis of intuition: clearly, it cannot “furnish us with the full story about the nature of our faculty for sensory apprehension” if that the intuitive character of the manifold depends on capacities not even mentioned there (p. 117). Conant argues that standard readings of Kant are overwhelmingly ‘layer cake’ approaches. He gives the example of Allison, but he could, as he notes, have used many others: Conant’s charge is that Allison recognises a basic level of sensory receptivity, prior to the involvement of the understanding, and so sharable between us and animals. Allison thus

insist[s] that one must distinguish two sorts of intuition—those that figure in the first layer of the layer-cake picture of sensory cognition and those which figure in the second layer, once the former have come to be reshaped in the light of their interaction with our higher cognitive capacities. The first sorts of “intuition”, according to Allison, are nonconceptual modes of apprehending an object; these require no involvement of the understanding. The second sorts of “intuition” are such that they can come into view for us as episodes of consciousness with objective purport only once they have been informed by the categories. (p. 94)

Whilst there are passages which seem to allow such a position—particularly the infamous passage at A90–1/B122–3—Conant reads Kant as holding that it is ultimately unintelligible, a pedagogically useful but incoherent “philosophical fiction” (pp. 100–2).

2. “Alteration”–The Compromise Conant Overlooks 

I want now to sketch some possible lines of response to Conant. I begin by showing how one can endorse Alteration, just like Conant, whilst nevertheless retaining significant parts of the layer cake story. I claim that retaining these parts is important both for Kant’s larger argument structure and for his ability to make sense of animals.

Before proceeding, a terminological clarification is needed. The Kant literature has recently seen extensive debates over a view called ‘nonconceptualism’; indeed, there are few topics which have attracted more attention in the last decade.[3] Such nonconceptualism is a much stronger view than the one Conant associates with the term; stronger in the sense that it entails far more. Conant uses ‘nonconceptual’ for any putative intuitions defined independently of the understanding (p. 94, 109).[4] But Kantian nonconceptualism is standardly understood as the view not just that there could be such intuitions, but that these can represent spatio-temporal particulars. Allais’s classic paper gives a helpful introduction to the view:

Consider a creature whose actions indicate that it sees a located, spatially unified thing that it can discriminate from other things, and which it can track […]. The creature might be able to perceptually discriminate and act on a spatially unified thing without understanding the thing’s causal unity—perhaps it does not represent the plant growing on the rock as a separate thing from the rock, since they appear spatially continuous, and perhaps it could not recognise or track the thing if it were to be changed by collisions with other things or natural processes of growth and transformation. On this account, the non-human animal (assuming it lacks concepts but has some way of representing space) that perceives its environment represents the world in the sense that it has relational mental states that present it with parts of the world—it does not have an inner display of nonintentional, raw sensations. (2009:406)

In line with my earlier disclaimer, my aim here is not to provide a full assessment of such approaches: that would require an analysis of the relevant notions of intentionality. But even this brief sketch can serve as useful foil and so I shall follow the literature in reserving ‘nonconceptualism’ for the view that it is possible to be intentionally related to spatiotemporal particulars in the complete absence of conceptual capacities.

Why does this matter? After all, nonconceptualism implies a stronger version of the kind of separability between understanding and intuition than that which Conant takes himself to have disproved. The answer is that it helps us see that the situation is more complex than Conant allows. Consider the following nonconceptualist position: the categories are not necessary for the representation of spatiotemporal particulars, but solely for the representation of a very specific subset of spatiotemporal relations. A familiar example to start: given the necessarily successive nature of apprehension, it is only in virtue of categorial synthesis that subjects can represent the distinction between a succession of perceptions and a perception of succession, between the house and ship cases (A99, A189–90/B234–5). The relation of objective succession is thus one member of the relevant subset: whilst a dog may see both the house and the ship, it is unable to represent the temporal relation that distinguishes them.[5] Next, a less familiar example: it is only in virtue of categorial synthesis, in this case involving the categories of quantity, that subjects can represent various mereological relations—for example, the fact that the parts of something together constitute its whole (A142–3/B182). A dog, by contrast, may associate the parts of something, in the sense that the image of one calls to mind the other or that the image of one part calls to mind the image of the larger object. But animals are unable to represent those parts as parts, for example unable to see them as constituting a whole. This is because, as Kant puts it in the Prize Essay:

[T]he representation of a composite, as such, is not a mere intuition, but requires the concept of a composition [Zusammensetzung] […] that is not abstracted from intuitions […] but is a basic concept, and a priori at that. (FM, AA 20:271; trans. modified)

The upshot is that animals will be unable to form basic mereological beliefs: ‘if one part is removed from a whole, then the size of the whole will be reduced by the size of that part’ and so on.

Two points can now be made. First, on this story, the categories make possible what are in a significant sense intuitive relations. In every case the challenge is created by the fact that apprehensive synthesis “is always successive” (A189/B234; see also A99 and B160). The result is that some mechanism is needed to represent properties which that successive synthesis obscures. Some of these relations are straightforwardly temporal: do the roof and door exist simultaneously? Others such as the Axioms, are more complex: when the parts of something are seen one after another, how can one represent them as constituting a single, simultaneous, contiguous whole? Second, this fact is compatible with nonconceptualism. Animals can, for example, identify and track a specific particular, such as a prey item; what they lack is just the ability to represent the relevant subset of spatiotemporal properties relations enabled by the categories.

My aim is not to say enough to convince you of this story here. Rather, what I want is to highlight its dialectical impact: returning to Conant, we can see why things are more complex than they might first appear. Recall his endorsement of:

(Alteration) The intuitive character of the manifold experienced by subject s is necessarily altered if s possesses conceptual capacities.

The model just sketched can also endorse this given only the premise that ‘objective succession’, a occurs after b, is an intuitive property and thus part of the manifold’s ‘intuitive character’. Is this premise true? Well, it is certainly a temporal property, and such a basic one that ‘no one’ will fail to recognise it (A190/B23). One might respond that it is a property, as opposed to say a singular term, and as such cannot be intuitive. But this is a misunderstanding of Kant’s distinction: such properties are effectively what Kant calls ‘marks’ and these are both “in the thing [Ding]” and a “partial representation […] considered as the ground of cognition” insofar as they are recognised by a rational agent (Log, AA 9:58). What is distinctive about the categories is that they correspond to a small subset of intuitive marks which are generated by, rather than simply recognised with, the understanding.[6] The result is that it is possible to endorse Alteration whilst allowing a much stronger form of nonconceptualism than Conant considers. By extension, it is possible to endorse Alteration whilst retaining a layer cake picture. There is indeed a layer of sensibility that is unaltered by the addition of the understanding, and so shared with animals: the perception of particulars in the sense sketched above. There is also another layer of sensibility that is radically altered by rationality and this is what the categories delineate: this is necessary for object awareness in a more full-blooded sense.

A huge amount of ink has been spilled on Kant’s remark on animals and I do not want to replay those issues here. But we can note that my approach is in line with the most important texts: the ox can see both the door of its stall and then the back of it just as I can, although it lacks my ability to recognise the relevant marks and the attendant inferential relations (DfS, AA 2:59–60). Indeed, it is only because both it and I can see ‘door, stall, door’ that the problem of representing objective succession even arises, the problem which the Second Analogy is designed to fix. In short, positions such as the one sketched retain some layer cake elements because they treat understanding as transforming some but not all spatiotemporal content. Doing so has attractive consequences: it allows an account of both animal capacities which does not reduce them to a ‘blooming, buzzing confusion’, and of the precise point at which our perception differs from theirs.

Conant misses this move because he thinks of the layer cake story exclusively as treating the categories as conditions for objects to be thinkable (p. 107). But perhaps all this really shows is that Alteration was too weak all along? Perhaps what Conant really meant was this:

(Universal Alteration) All aspects of the intuitive character of the manifold experienced by subject s are necessarily altered if s possesses conceptual capacities.

This fits with his remarks about sensibility coupled with understanding being “utterly distinct” from sensibility taken alone (p. 79). As noted above, this is very different from saying the two instances of sensibility lack the “very same character”: the compromise I just proposed effectively exploits the fact that layer cakers can agree that the two would lack the “very same character” whilst still insisting that some elements remain unaltered by the chocolate sauce.

Universal Alteration is a coherent position, but it comes at an exorbitant cost. When I walk my dog, my experienced manifold includes various spatial properties: for example, that the ball I just threw is up ahead. The manifold which my dog experiences surely also contains the same spatial relation: that is precisely why he is able to retrieve the item. Of course, my manifold is also saturated with all types of other associative and inferential material, and I have a much more sophisticated conception of us and of our location. But to deny an underlying, shared spatial awareness would require the postulation of some radically alternative mechanism to allow poor old Rover to get his toy. Given the well attested biological similarities between my eyes and his, and the fact that my compromise approach can still embrace Alteration, there is no good reason to do so.[7]

3. “Generic” and the Question of Explanatory Continuity

I want now to expand the discussion by turning to the broader question of animal-human continuity. We are used to locating Kant between and thus beyond rationalism and empiricism. But there is another methodological divide in the modern period which is just as important. On the one side, you have authors such as Hume who regards our continuity with animals as a “touchstone” by which one “may try every system” (1978: On the other side are thinkers who reject an appeal to the same explanatory apparatus even when animal behaviour closely mimics its human counterpart. Heidegger, at least in some of his texts, is a good example of this: whatever apparatus is needed to explain how an animal might open a door, there is an “abyss” (Abgrund) between that case and my own.[8] One can immediately see that this debate is not simply another version of the well-worn empiricist vs rationalist controversies: Heidegger is scarcely a rationalist. Conant, in endorsing Generic, effectively takes a hardline position similar to Heidegger’s. But Kant’s own position is much more nuanced: just as with rationalism and empiricism, he offers a delicately balanced compromise. Consequently, I shall suggest, we should reject Generic as a reading of Kant.

As noted, Kant is explicit that both animals and humans possess representations. Indeed, he is clear that both possess conscious and unconscious representations:

The field of sensuous intuitions and sensations of which we are not conscious, even though we can undoubtedly conclude that we have them; that is, obscure representations in the human being (and thus also in animals), is immense. (Anth, AA 7:135)

It follows there must be some line in both us and animals which marks the boundary between conscious and unconscious. Of course, the nature of this line might be radically different in the two cases: this is what Generic implies and it is why, for example, Freud was so hedged on whether animals had an unconscious. But if we look at the example of unconscious states which Kant himself immediately gives, it is very likely that he understood it as operating in the same way in both cases. If I perceive a man, Kant says, I must also represent those parts that constitute him “since the representation of the whole […] is composed of these partial ideas” (Anth, AA 7:135). However, I am not typically conscious of those parts: I may see a man’s face “even though I am not conscious of seeing his eyes, nose, mouth etc.” (Anth, 7:135). Construed in this way, there is no reason why the line between the human and the animal unconscious should differ at all: the same reasoning would equally apply to my seeing the man’s face and my dog doing so. So, we have a prima facie continuity, and one with some explanatory significance: it is part of why the Kantian unconscious is not the Freudian one.

Now take another example, that of inner sense. As McLear (2011) has documented, Kant’s stance on inner sense and animals is complicated by a pre-Critical tendency to conflate it with apperception. But it seems plausible that the considered Critical position would be to allow animals inner sense if only because of the links Kant postulates between it and time: otherwise, Kant would have to attribute representations to them, whilst denying that their experience was sequential, a position McLear rightly describes as “scarcely credible” (2011:11). So animals must be allotted both time and inner sense. But what exactly is the latter? Well, Kant is clear: it is the form in which we intuit our own act of “attention” (B156 and B156n.). As with the unconscious, there is no reason to think this aspect of the story shouldn’t also carry over to animals: after all, if animal time is not also the form of auto-affection, it risks reclaiming some kind of realist status. By extension, it is a good exegetical hypothesis to assume that animals likewise experience time through acts of attention. And you can now see how this all ties together. Both me and my dog see the man’s face without the parts being conscious; depending on the situation, our attention may shift with the result we are now suddenly conscious of, say, his teeth.[9] These shifts in attention, shifts by which the manifold of the face is first perceived as a manifold, thus mark the boundary between conscious and unconscious states. Furthermore, it is precisely these attentional shifts which also generate the experience of succession. Kant’s name for this combined process is the ‘synthesis of apprehension’:

Every intuition contains a manifold in itself, which however would not be represented as such if the mind did not distinguish the time in the succession of impressions on one another; for as contained in one moment no representation can ever be anything other than absolute unity. (A99)

We can now add a final, vital move: in humans, such shifts in attention are susceptible to conceptual control and analysis. When I acquire a concept, I am thus literally conscious of new aspects of the scene before me: hence Kant identifies a schema as a “rule for apprehension” (Refl. 2880, AA 16:557).[10] By contrast, in animals the shifts of attention are dictated by instinct and association (Br, AA 11:52).

If we now return to Generic, we can see that such considerations would imply that it is false:

(Generic) If humans and animals may be said to have a common sensibility, it is only because this “having something in common” holds at a sufficiently generic level of description so as to play only a limited explanatory role.

On the picture I offered, the overlapping material plays a very significant role indeed. Now, of course, you might reject the story sketched; it is very much only a sketch. But what I want to stress is that there is no need to do so simply because you want to maintain a ‘transformative’ role for rationality. Rationality transforms the system outlined in that it takes it over for its own ends: the way in which we auto-affect, i.e. the way in which our attention shifts over the manifold, i.e. the way in which we become conscious, is embedded and so transformed within a rationally explicable and criticisable framework, that of concepts functioning as “rules for apprehension”. So as in Section 2, there is an attractive compromise to be had without going as far as Conant does.

4. Two Possible Objections—Ambiguity and Imposition

I want to close with two objections which Conant raises to theories besides his own. The first is that such theories must read terms such as ‘intuition’, ‘form’ and ‘synthesis’ as ambiguous (p. 95). I confess I am happy to be guilty as charged here. Surely one point we can all agree on is that the Critique is an almost uniquely complex text in which Kant is engaged in multiple arguments with multiple opponents using multiple premises at any one time. It seems utterly unsurprising that some of his terms, terms for which he rarely supplies explicit definitions and which are often linked to arguments whose incredible power he is visibly struggling to control, should display a consequent ambiguity. That is not to say that their meaning simply varies arbitrarily; rather, as Longuenesse (1998:223–4) stressed with regard to ‘form of intuition’, it is typically relational and is a complex product of the specific argument in play and certain base functional properties. To take a famous example, the one that so preoccupied Hegel and Heidegger, try to define ‘imagination’ stably across the A and B editions. Second, Conant suggests that positions which retain layer cake elements, as mine avowedly does, reduce the categories “(as Hume had claimed) to nothing more than mere subjective imposition” (p. 110). Conant makes this claim because he assumes that layer cake models must regard the categories as necessary for thought alone (p. 107). As he concludes:

If that is the case, then all the categories would represent are mere conditions on the thinkability for us of that which is given to us in such a self-standing form of consciousness. (p. 110)

I agree this is a threat, and I have argued elsewhere that it is indeed a problem for forms of nonconceptualism such as Allais’s.[11] But it does not apply to my own position on which the categories are necessary conditions on the representation of certain spatiotemporal relations, rather than conditions on concept formation. Equally, my solution makes clear why Kant is able to answer Hume: the relevant spatiotemporal relations, such as objective succession and the simple forms of mereological composition noted, are precisely those assumed by Hume in his own account; to take the obvious example, his account of causation assumes objective succession and asks what is added to it.[12]

5. Conclusion

My aim has been to sketch an alternative to Conant’s approach; I have also tried to do so without resting my case on notoriously slippery texts such as A90/B122–3. The proposed alternative recognises the enormous changes which understanding makes to sensibility: thus it can accept Alteration. At the same time, it is much more flexible on our relationship to animals; hence its retention of some ‘layer cake’ elements and its rejection of Generic. Kant is often criticised for smuggling theological assumptions back into the Critical system; claims like Generic give such charges plausibility. The proposed compromise, in contrast, allows Kant to take seriously the countless explanatorily significant continuities between us and animals identified by the modern sciences.

Invited: 7 June 2017; received: 25 August 2017.


[1] Golob (2016a), (2016b) and forthcoming.

[2] Boyle cited by Conant, p. 80.

[3] The seminal pieces are Allais (2009) and Hanna (2005); Schulting (2016) brings together recent articles by many of the key figures on both sides of the argument.

[4] This is partly because Conant introduces the issues via Allison who never gets to grips with either the nonconceptualism literature or with Kant’s position on animals (the bibliography for Allison 2015, for example, contains neither the Allais nor the Hanna nor any of the key conceptualist responses such as those by Ginsborg).

[5] Of course, the dog does not possess the concepts <house> or <ship>; in Kantian terms, it intuits the particulars but fails to recognise their marks (DfS, 2:59).

[6] One might dig one’s heels in and insist that, insofar as they are so generated, they are not really intuitive but rather conceptual. But this is surely not a move Conant can make since, given the interdependence he sees between intuition and understanding, it would imply that on his account there were no genuinely intuitive relations (this is effectively the crude reading of his ‘one unity!’ slogan which he rightly fends off at p. 112). Another option would be to argue that property is not intuitive because it is not ‘immediately perceivable’. But that way lies a very tangled thicket given the basic nature of the property as stressed in the Second Analogy, and the fact that some highly intellectual properties would then become intuitive in virtue of cognitive penetration.

[7] A full treatment of the issue would require extensive discussion of B160–1n to show how the relationship between empirical intuitions and the pure forms plays out in animals. This is, unfortunately, beyond the present paper.

[8] Heidegger (2001:192) and (1976:326). I say “some of his texts” to avoid the debate surrounding notions such as weltarm.

[9] Thus, as Kant stresses having introduced the link between inner sense, time and auto-affection, “every act of attention can give us an example of this” (B156n; original emphasis).

[10] For very helpful discussion of this phrase see Longuenesse (1998:116–7).

[11] See Golob (2016b:51–3).

[12] One reason why Conant misses this, I suspect, is because of his overall understanding of the Deduction. He describes it like this:

What is at issue in “the Deduction” that concerns us in this paper are not claims about what is the case, but rather considerations regarding what is necessarily involved in the apprehension of anything’s ever possibly being the case. (p. 105)

I think this is too broad: the transcendental conditions on representing ‘something’s being the case’ are given by Kant’s treatment of marks and judgement, and there would be no need for much of the Deduction if that were all that was in play (thus e.g. Frege can offer a perfectly coherent account of how we can represent “anything’s ever possibly being the case” without any discussion of e.g. causality or synthesis).


Allais, L. (2009), ‘Kant, Non-Conceptual Content and the Representation of Space’, Journal of the History of Philosophy 47(3): 383–413.

Allison, H. (2015), Kant’s Transcendental Deduction. An Analytical-Historical Commentary (Oxford: Oxford University Press).

Boyle, M. (2012), ‘Essentially Rational Animals’, in G. Abel & J. Conant, Rethinking Epistemology, vol. II (Berlin: De Gruyter).

Brandom, R. (2002 ), ‘Dasein, the Being that Thematizes’, in H. Dreyfus & M. Wrathall (eds), Heidegger Re-Examined (New York: Routledge).

Golob, S. (2016a), ‘Kant as Both Conceptualist and Nonconceptualist’, Kantian Review 21(3): 367–91.

——— (2016b), ‘Why the Transcendental Deduction Is Compatible with Nonconceptualism’, in D. Schulting (ed.), Kantian Nonconceptualism (London: Palgrave Macmillan), pp. 27–52.

——— (forthcoming), ‘What Do Animals See?’, in L. Allais J. Callanan (eds), Kant and Animals (Oxford: Oxford University Press).

Hanna, R. (2005), ‘Kant and Nonconceptual Content’, European Journal of Philosophy 13(2): 247–90.

Heidegger, M. (1976), Wegmarken (Frankfurt a.M.: Klostermann).

——— (2001), Einleitung in die Philosophie (Frankfurt a.M.: Klostermann).

Hume, D. (1978), A Treatise of Human Nature, ed. L. Selby-Bigge P. Nidditch (Oxford: Clarendon Press).

Longuenesse, B. (1998), Kant and the Capacity to Judge (Princeton: Princeton University Press).

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

Schellenberg, S. (2015), ‘The Relational and Representational Character of Perceptual Experience’, in B. Brogaard (ed.), Does Perception Have Content? (Oxford: Oxford University Press), pp. 199–218.

Schulting, D. (ed.) (2016), Kantian Nonconceptualism (London: Palgrave Macmillan).

© Sacha Golob, 2017.

Sacha Golob is a Senior Lecturer in Philosophy at King’s College, London. His research focuses on the intersection between Kantian and post-Kantian philosophy and contemporary work on the philosophy of mind, aesthetics and philosophical methodology. Among publications in British Journal of the History of PhilosophyEuropean Journal of PhilosophyKantian ReviewProceedings of the Aristotelian Society and in various edited volumes, he is the author of Heidegger on Concepts, Freedom and Normativity (Cambridge UP, 2014) and co-editor of the Cambridge History of Moral Philosophy (Cambridge UP, forthcoming).