ANIL GOMES & ANDREW STEPHENSON (eds) | Kant and the Philosophy of Mind. Perception, Reason, and the Self | Oxford University Press, 2017


By Yoon Choi 

Kant and the Philosophy of Mind, edited by Anil Gomes and Andrew Stephenson, is a welcome collection of previously unpublished work on Kant, ranging over a selection of topics central to both Kant’s philosophy and to current debates in philosophy of mind. All contributions are primarily interpretative in aim, and most are deeply rooted in Kant’s texts and the secondary scholarship. Several also make connections with current philosophical and psychological work, sometimes to shed light on Kant’s views (e.g. Lucy Allais and Katherine Dunlop) and sometimes to bring Kant’s views to bear on ongoing debates (e.g. Patricia Kitcher and Ralph Walker). The resulting volume thus presents Kant “as engaged in the philosophy of mind”, as Gomes puts it (p. 6), and advances our understanding of Kant’s account of intuition, his theory of judgement, and his views on the self, self-awareness, and self-knowledge. Some may say the volume focuses on a handful of topics at the expense of representing the full range of work on Kant’s theory of mind. That is not wrong, but the editors are inclusive in other ways, and their priorities result in a volume that is exceptional in one way: it captures several substantive debates, in which contributors engage intensively with each other rather than presenting a series of different views on a topic. Indeed, every essay takes up or is taken up to some degree by another, and even when this takes the form of a passing footnote, it generates continuity and unity to the volume as a whole and conveys a sense of common purpose running through the disagreement. This must be the result of careful editorial design and encouragement; and it is, in my view, a real achievement.

In this review, I discuss all contributions, though some only briefly. I proceed in order with just one exception, grouping essays thematically in a way that is compatible with but not identical to the divisions and pairings the editors indicate in their Introduction.

The editors’ brief ‘Introduction’ provides compact and clear previews of arguments to come. The other tasks normally fulfilled by an Introduction are taken up by Gomes in ‘Kant, the Philosophy of Mind, and Twentieth-Century Analytic Philosophy’ (Chapter 1). Thus Gomes begins with a clear overview of Kant’s work, introducing the topics discussed in the rest of the book, situating them within Kant’s theory as a whole, and providing extensive references to the secondary literature (§1.2). Gomes then goes on to trace the way Kant’s theory enters the Anglo-American philosophical tradition, identifying P.F. Strawson and Wilfrid Sellars as the main formative influences in the reception of Kant’s views of the mind (§1.3). (This invites the observation that the Sellars school is not represented in the volume, but perhaps none of the contributors identify themselves as Strawsonians, either.) Gomes concludes by characterising the present-day debates that have emerged from this tradition. He thus illuminatingly reveals how much Kant’s views have been transformed in being taken up but also how much they have thus shaped the philosophical problem-space we currently inhabit.

The Nature of Intuition

The first and largest cluster of essays explores the nature of intuition, addressing two questions: By what process are intuitions generated (Chapters 2 and 3)? And do they depend on the presence and existence of the objects they represent (Chapters 4–6)?

Producing Intuitions

Allais (Chapter 2) and Dunlop (Chapter 3) agree that some kind of processing is involved in the formation of intuitions, and that it is not to be identified with the activity Kant calls ‘synthesis’, but they disagree about how best to characterise it.

Allais begins ‘Synthesis and Binding’ by arguing for a non-conceptual and relational view of intuitions (§2.2). She emphasises that Kant assigns a specific cognitive role to intuition: “Intuitions are singular and immediate representations that give us objects” (p. 31). She then argues that only if intuitions are non-conceptual can they give us singular objects, and only if they “put us directly in touch with objects”, as the relational view of intuition holds, can they give us objects immediately (pp. 31–2). How does this relate to the question of how intuitions are produced? As Allais notes, there are some passages in the Critique of Pure Reason—most notably, the discussion of the threefold synthesis in the A-Deduction (A98–110)—that are frequently taken to say straightforwardly that intuitions are produced by synthesis. But synthesis is the kind of cognitive processing that brings concepts to bear on representations. Thus to defend her claim that intuitions are non-conceptual, Allais takes a close look at the putatively problematic passages and argues that synthesis never actually features in Kant’s thought as “something that produces intuition, but rather, [as] something that is done to intuitions” (p. 38, emphasis added)—something that takes place at a higher level of cognitive processing than intuition formation does. Allais’s main aim in this essay is to defend the conclusion just reached (p. 44). But she goes further and asks how intuitions are formed, if not by synthesis. Here she turns to the idea of ‘binding’, a process posited by empirical psychologists whereby sensory input becomes representations of perceptual particulars, which is a prima facie attractive model for the transcendental process whereby sensation is taken up in intuition. Since binding is furthermore a non-conceptual process that can be understood in a way that does not render the relation between the representation generated and the object represented indirect (see Allais 2015:110), this compelling account of intuition-formation turns out to fit particularly well with Allais’s non-conceptual, relational interpretation of Kant’s theory of intuition, and adds to the already-substantial case for it (see Allais 2015).

One question that arises for me concerns the role of consciousness and the relation between intuition and consciousness. According to Allais, an object of intuition is “something that is before the mind, something that features in consciousness” (p. 30). But—and this is a complexity Allais herself alludes to (pp. 26, 44)—Kant often writes as if intuition alone does not make us conscious of an object, but must be apprehended for consciousness to arise. Thus although Allais calls objects of intuition “perceptual particulars” (e.g. p. 27), Kant distinguishes intuition and perception and takes the latter to be the conscious representation of what is given by the former. For example, Kant says that

the synthesis of apprehension [is] the composition of the manifold in an empirical intuition, through which perception, i.e., empirical consciousness of it (as appearance), becomes possible. (B160)[1]

How much would this affect Allais’s account? I’m not sure. On the one hand, it does not seem that an object must be phenomenally present to me for it to count as ‘given’ to me in a cognitively meaningful sense. We could think of an unapprehended intuition as access-conscious but not phenomenally conscious, for instance (Block 1995). On the other hand, Kant says objects must be phenomenally present if we are going to think about them, and it seems that intuitions should ‘give us’ what we need to start thinking (e.g. Log, AA 9:33). I return to this point again later.[2]

Dunlop raises a different concern in ‘Understanding Non-Conceptual Representation of Objects’ (Chapter 3), considering whether Allais’s theory of intuition formation provides a strong challenge to both empiricists and conceptualists (§3.3). An account of intuition provides a good challenge to empiricists, Dunlop says, if the processing it postulates is “far enough removed from […] operations countenanced by empirical philosophers” (p. 55). Since binding is a process that empiricists can comfortably endorse, this, according to Dunlop, reveals a vulnerability in Allais’s account. Dunlop further suggests, if I have understood her correctly, that an account of intuition succeeds as a challenge to conceptualists if it shows that we can represent, without concepts, objects of a robustly objective kind, objects that are objective enough for the conceptualist to think they could only be represented with concepts (pp. 53–4). Dunlop takes this to reveal another weakness in Allais’s account, for Allais’ objects of intuition “fall short of objecthood”: perceptual particulars include such things as shadows and ripples of water, and as Dunlop observes, “it is not clear that the conceptualist must require the use of concepts” in the representation of such things (pp. 53–4). Dunlop then argues that Allais’s account could be shored up on both counts by replacing binding with what Dunlop calls the capacity for “primitive representation of objects”: a surprisingly sophisticated ability to individuate, track, and seemingly identify and re-identify objects that is seen in infants between 2 1/2 and 4 months old (§3.4).

It seems to me that Dunlop hits upon an elegant way to assess and even compare empirically-informed interpretations of Kant’s transcendental theory. A question such as “Does this account of intuition preserve anti-empiricist force?” can rule out proposals; a question such as “Which account of intuition has maximal anti-conceptualist force?” can rank competing proposals. Though I think this is an ingenious and perhaps widely applicable approach, I shall challenge some details of the way Dunlop implements it in response to Allais. First, is Dunlop right to look for Kant’s challenge to empiricists at the level of intuition? Isn’t it at the level of cognition, not intuition, that Kant wants to pick his battle with the empiricist? Second, I’m puzzled by what Dunlop says about conceptualism. I take it that Kantian conceptualists are not conceptualists because conceptualism is an independently attractive account of intuition, but because conceptualism is taken to be necessary to account for the possibility of experience. If this is right, I do not see how any account of intuition proposed by the nonconceptualist could by itself challenge the conceptualist. And given the reason the conceptualist would reject nonconceptualist accounts of intuition, the details of the account seem irrelevant. Thus I find myself a bit dialectically lost, and am not convinced that Allais’s binding-based account has the specific weaknesses Dunlop points to. Nevertheless, Dunlop’s proposal might still be an improvement, and she provides a promising model for figuring out whether that is true.

Intuition and Object-Dependence

The next three essays debate whether intuitions depend on the existence of their objects. According to a strong object-dependence theory, one can only have an intuition of an object that is present to you. (This is my own simplified formulation; I think the right reading of ‘present’ makes it compatible with the variations given by Stefanie Grüne, p. 67, Colin McLear, p. 89 and Stephenson, p. 108). We have already seen this view championed by Allais (Chapter 2); here it is defended and extended by McLear (Chapter 5) and challenged by Grüne (Chapter 4) and Stephenson (Chapter 6). I use Grüne’s term “object-dependence” throughout, replacing, as necessary, McLear’s “presence-dependence” and Stephenson’s “particular-dependence”.

In ‘Are Kantian Intuitions Strongly Object-Dependent?’, Grüne challenges several arguments put forward in support of the strong object-dependence view (p. 68). She offers competing readings of texts that strong object-dependence theories point to (§§4.2–4.4), and she considers what kind of object-dependence is necessary if intuition is to fulfil its cognitive function (§4.5). Though Grüne’s attention thus remains focused on rebutting the strong object-dependence view, some positive arguments emerge, too, and give us some indication of her own view (p. 68n.5). She argues, for instance, that a priori intuitions are not object-dependent at all (pp. 70–5); she also offers an ingenious but complicated argument that entails, among other things, that the cognitive role of intuition is not that of giving us only present, existing objects, as the strong object-dependence view claims. I take a close look at this argument and its surprising implications next.

I begin by reconstructing the argument. I simplify Grüne’s complex analysis substantially; I hope this does not distort Grüne’s view. The argument seems to begin with a claim I shall call the ‘objective-validity requirement’. I formulate it using Grüne’s own words as long as possible, for reasons that will become clear.

1. “For cognizing an object that falls under a concept F it is necessary that one can prove that F has objective reality” (p. 78).

Grüne then analyses what it means for a concept to have objective reality. She argues that a concept has objective reality just in case it refers to an object that has real possibility. But an object has real possibility just in case it has formal possibility, and it has formal possibility just in case it “agrees with formal conditions of experience”, viz. our forms of intuition and thought. This yields:

2. A concept F has objective reality just in case it refers to an object that agrees with formal conditions of experience.

Grüne defends this argument at length (pp. 79–80, 81–2). Since the issue is far too complicated for me to address here, I accept Grüne’s analysis. The objective-validity requirement can then be re-formulated as follows:

3. Therefore, “for cognizing an object that falls under a concept F it is necessary that one can prove that F” refers to an object that agrees with formal conditions of experience.

Now the question becomes, what is involved in supplying the requisite proof? Grüne argues that it is far too stringent to require intuition of an existing object. It is enough to require an “intuitive representation”, where this includes things like hallucinations and imaginings as well as intuitions of existing objects. For according to Grüne, all intuitive representations are “compatible” with the forms of intuition and the forms of thought; intuitive representations thus give us only objects that agree with the formal conditions of experience (p. 80). So:

4. To prove that a concept F refers to an object that agrees with formal conditions of experience, it is necessary and sufficient to have an intuitive representation of the object.

This is probably the most controversial part of Grüne’s argument. To defend the claim that intuitive representations are compatible with the forms of thought, she needs the premise that categorially-governed synthesis is involved in the production of intuitive representations—a premise that won’t be granted by nonconceptualists (p. 80). But granting this too, we can reformulate the objective-validity requirement again:

5. Therefore, “for cognizing an object that falls under a concept F it is necessary that one can” have an intuitive representation of the object.

The problem is that Grüne seems to draw a significantly stronger conclusion, for she says:

If showing that a concept refers to really possible objects is sufficient for proving the objective reality of the concept and thus is sufficient for cognizing these objects, then we are able to cognize objects of dreams and hallucinations. (p. 81, emphasis added)

Though her formulation is conditional, I think it is clear from the context that she is drawing out the implications of what she takes herself already to have shown. And the trouble, obviously, is that the objective-validity requirement gives us only a necessary condition for cognition, not a sufficient condition (p. 78). Of course, there might be something I missed in Grüne’s argument. In particular, before first stating the objective-validity requirement, Grüne quotes Kant’s familiar claim that cognition requires intuition and a concept (A92/B125, quoted on p. 78). I’m not sure how she sees the relation between this claim and the objective-validity requirement, but perhaps it allows her to strengthen the objective-validity requirement as necessary. But if her thought is that an intuitive representation of an object can do double duty—that it proves the objective reality of the concept of the object and it is the intuition required for cognition—this would beg the question against the strong object-dependence view.

However this argument is to be clarified, the account of cognition that Grüne takes it to generate is certainly “very different”, as she mildly puts it, from current conceptions of cognition (p. 81). On this view, we cognise existing objects through intuition of existing objects, but we also cognise non-existing objects through intuitive representations of non-existing objects: “We are able to cognize objects of dreams and hallucinations” (p. 81). Pace the strong object-dependence theorist, the function of intuition is thus not best understood as that of giving us existing objects, since that leaves out tons of things we can cognise. The implication, though Grüne does not outright say this, is clearly that ‘intuition’ is better understood in the broader sense of ‘intuitive representation’—and thus as not object-dependent. But to raise just one question about this view, how do we understand the singularity of an intuition of a non-existing object? Does it still give us particulars? Can there be particular non-existing objects? The only way I can think of to preserve a sense for singularity is by individuating non-existing objects by their being objects of specific intuitive representings. To use Grüne’s example, when I hallucinate a purple-spotted cow, we can distinguish the non-existing cow I hallucinate from the non-existing cow you hallucinate by the fact that one is my hallucination and one is yours (p. 81). But this would limit intuitions of non-existing objects to objects of inner states. And it brings Grüne’s view close to Colin McLear’s—a surprising conclusion, given the fundamental differences in their accounts. There may be a sense of singularity that avoids this conclusion, if the conclusion needs avoiding. But for now I turn to McLear.

McLear takes intuitions to be strongly object-dependent, and in ‘Intuition and Presence’ (Chapter 5), he defends this view by first undermining an intuitively attractive competitor (§5.3) and then answering one of the strongest objections to it: viz. that it cannot give a good account of hallucination and imagination (§5.4). I focus on the second half of his argument. Why is the strong object-dependence view taken to be unable to provide a satisfactory account of hallucinations and imagination? Because when I hallucinate a flower, there is just no flower for me to be given. What can the strong object-dependence theorist say about this case? What McLear proposes is the following: when I see a flower, I have an outer intuition of a flower that exists; and when I hallucinate a phenomenally indistinguishable flower, I have an inner intuition of a mental state that exists, perhaps a mental image of a flower (pp. 97–8). There is certainly something elegant about this proposal. There is nothing ad hoc about extending a claim about outer intuition to inner intuition, and with that simple manœuvre, McLear does not just neutralise the trouble cases, but turns them into central elements of his view. Since the theory is nevertheless subject to vigorous challenge by Andrew Stephenson, I bring him into the discussion before returning to McLear.

I discuss two arguments Stephenson sets out in ‘Imagination and Inner Intuition’ (Chapter 6). The first concerns all strong object-dependence views. Stephenson argues that strong object-dependence views are committed to the following claim, which he calls “particularity”:

The object of intuition [is] the very same object that exists and is present to the subject of intuition. (p. 111)

But in the B-Deduction, as Stephenson notes, Kant says that “imagination is the faculty for representing an object in intuition even without its presence” (B151, as translated by Stephenson, p. 113). And Stephenson argues that the original unambiguously specifies that it is the object of imagination itself that is not present to the subject of imagination (pp. 112–14). The particularity claim is thus straightforwardly incompatible with B151. Though I find Stephenson’s ultra-close reading of the text impressive, it seems to me that the defender of the strong object-dependence theory is not entirely out of options. She could argue that B151 concerns imagination specifically in its transcendental function—after all, the preceding sentence is explicitly about the transcendental synthesis of imagination—and that the claim of strong object-dependence is limited to empirical intuition (e.g. McLear, pp. 90–1). Admittedly, this leaves similar passages that do not permit a similar defence (e.g. Anth., AA 7:153; see Stephenson pp. 114–15), but perhaps it provides some room for manœuvre.

The second argument I look at is specific to views that, like McLear’s, combine a relational view of intuition with strong object-dependence and extend the account to cases of inner intuition (Stephenson pp. 120–3). I won’t present Stephenson’s argument in full; the first half generates the dispute that concerns me. Stephenson argues that those who take intuitions to be relational and strongly object-dependent—where this includes Allais as well as McLear—must take intuitions to make their objects phenomenally present. This means there must be a phenomenal character to having an intuition, a phenomenal character part-constituted by the properties of the object of intuition (Stephenson pp. 120–1). I’m not sure that such a view is really forced on Allais and McLear, and McLear denies that it is (McLear pp. 101–2), but one virtue of such a view is that it makes clear how an intuition ‘gives us’ an object for cognition: the phenomenal character of the intuition carries information about the properties of the object. What does McLear propose instead? He accepts that an intuition must make an object at least “sensibly present” (this, I think, is supposed to be quite a bit weaker than phenomenal presence), but he denies that we have to explain how an intuition gives us an object by appealing to its ‘sensory character’ at all. But what is the alternative? McLear suggests we could think of an object as given by the fact that it “reveal[s] the metaphysical […] compossibility of whatever properties may be presented” (McLear p. 102). I’m not sure this does not end with an appeal to sensory character, but McLear offers it just as a passing suggestion, so I close by pressing a different point. I noted, in my discussion of Allais, that there is some textual evidence suggesting that Kant does not take intuition to involve consciousness. An account of intuition that does not account for its cognitive role by appealing to its phenomenal character would help with this textual problem. However, the more we move away from the idea that an intuition makes its object immediately and phenomenally present to me, the more difficult it is for me to see the advantages of a strong object-dependence view.

Self-Awareness and Self-Knowledge

The discussion of intuition has already drawn us onto particularly tricky terrain: Kant’s doctrine of inner sense. This is taken up directly by Ralf Bader (Chapter 7), whose account of inner sense explains both how outer objects become temporalised and how temporal self-awareness emerges. Andrew Chignell (Chapter 8) picks up the story from there. Bringing the faculty of understanding to bear on inner sense, he asks whether empirical self-cognition is possible. His account of inner sense-based empirical self-knowledge is usefully contrasted with the apperception-driven account that follows (Chapter 9).

In ‘Inner Sense and Time’, Bader addresses a notorious puzzle concerning inner sense: time is the form only of inner sense, but the spatial objects of outer sense must have temporal properties, too! How do they “get in time”? Bader answers this question by providing a detailed model of inner sense. Two functions are central to the story: the temporalisation of appearances by what Bader calls “reflexive awareness” (§7.2) and the ordering of our representations by the “reappropriation of mental states” (§7.3). I summarise Bader’s proposal before raising a few questions.

On Bader’s view, representations are essentially content-carrying, intentional states, but they are not conscious states. I am not aware of an object just in virtue of representing it. How, then, do I become aware of an object that I’m representing? By attending to the representation (pp. 126–7). When I attend to a representation, Bader says, I become aware of what the representation is representing; Bader calls this “an act of reflexive awareness” (p. 127). The term can be misleading, for it can be read as characterising a kind of awareness we have of something by reflecting on that very thing. In the act Bader has in mind, what I reflect on (my representation) is not what I’m thereby made aware of (the object of my representation); what I reflect on is just, as Bader often puts it, a “vehicle” or “medium” that carries representational content (p. 127). Reflexive awareness is key to Bader’s account, for Bader argues that “time [is] the form of awareness” (p. 126). This means that when I’m aware of an object, I represent it as temporal; more specifically, I represent it as existing now (p. 128). It is clear how Bader plans to account for the temporalisation spatial objects, but I spell out the story.

In cases of outer intuition, I am (noumenally) affected by objects and thus given a manifold of intuition. A kind of pre-cognitive processing (not unlike Allais’s binding) takes up this manifold and produces representations; since the form of outer intuition is space, these are representations representing a manifold of objects ordered in space. Recall that representations are not themselves conscious states. However, through reflexive awareness of these representations, I become aware of spatial objects. And since time is the form of awareness, I become aware of these spatial objects as temporal, as existing now (pp. 127–8). Furthermore, I can be aware of many things at once; and when I am, I am aware of all of them as existing now, as simultaneous (pp. 130–1). What about the case of inner intuition? This involves the second function of inner sense, which has not yet been mentioned: reappropriation. Bader argues that reappropriation is an act of inner intuition whereby some of my acts of representing are “reappropriated” and made into objects of representations themselves. It involves (noumenal) self-affection and the processing of a manifold of the intuition I am given into representations. But since the form of inner intuition is temporal, I end up with representations that represent a temporal manifold of my representings. As always, representations are not themselves conscious states. But if I become reflexively aware of my inner representation, I become aware of its object: a temporal series of my representings (pp. 132–3).

I’m not entirely sure I have gotten the details of “reappropriation”, but I start with a question, which leads me to suggest a minor modification. Bader says that reflexive awareness of an object is awareness of “temporalized representational contents”: I am aware of the object as existing now (p. 127) and “the object itself […] ends up being in time” (p. 128). But this means, first of all, that reflexive awareness brings in time without involving inner sense or inner intuition, although time is supposed to be the form of inner sense. Moreover, in what sense can such awareness be characterised as awareness of “temporalized representational contents”? When I’m aware of an object as existing now, how is the temporality represented? The case of simultaneity provides an instructive comparison case. When I’m aware of several objects at once, I’m aware of them as simultaneous. And simultaneity is obviously a temporal notion. But how is the temporality represented? Insofar as it is, it is represented by the spatial co-presence or relatedness of the objects I’m simultaneously aware of. Similarly, insofar as the nowness of an object is represented in an act of awareness, it seems to be represented by something spatial: basically, the object’s presence to me. Thus the sense in which temporality can be said to be represented by an object of awareness does not seem to leave me with the “object itself […] in time” (p. 128).

One modification to Bader’s model eliminates this issue, and has, it seems to me, a few other advantages. If Bader allows for an act of reflective awareness to not only make us aware of an object, but also to affect us, thereby generating a representation of inner intuition, viz. a representation of <the act of representing the object as existing at the time of representation>, i.e. if Bader combines reflexive awareness and reappropriation, then awareness of an outer object always involves inner sense and intuition, and every object of outer awareness also gets represented in time. This modification would also preempt the question “What is the explanation of why and when some representings get ‘reappropriated’?” And it helps make sense of several further things Kant says. For instance, Kant says that every act of attention involves inner sense (B156–7n.). Attention seems to be an example of reflexive awareness. If Bader takes acts of reflexive awareness to be self-affecting and thus to involve inner sense, he could easily accommodate these examples. Bader could also account for the many passages in which Kant connects the synthesis of apprehension with the generation of time (e.g. “I generate time itself in the the apprehension of the intuition” [A143/B182; also A145/B184]). Apprehension appears to be a kind of reflexive awareness; it “generates time” because it is self-affecting and thus involves the representation of a temporal manifold.

I can think of very few serious and sustained attempts at understanding inner sense; Bader’s interpretation of Kant’s theory of inner sense sheds some light on one of the murkier corners of the Critical theory of experience. The analysis he provides seems to me to be consistent with a wide range of often puzzling texts, and to have a lot of explanatory power. Bader, for instance, combines it with an account of the difference between forms of thought and forms of intuition to provide an elegant explanation of why objects that are never given to us but which may be thought about, like God, do not end up in time (pp. 128–9). Bader thus leaves us with a powerful account of how functions of sensibility generate self-awareness structured by subjective temporal relations of simultaneity and succession (p. 135).

The understanding is then brought into play by Chignell. In ‘Can’t Kant Cognize Himself?’ (Chapter 8), he challenges a prevalent (though not consensus) view according to which Kant’s theory of mind does not leave room for the possibility of substantive empirical self-knowledge. According to Chignell, inner sense makes possible cognition not only of our mental states, but also (and thereby) of “something else”: the empirical “self or […] mind” (p. 138). How does Chignell defend this claim? An opening section turns to the textual evidence and lays out a textual basis for the view (§8.2). Though Chignell’s interpretations are persuasive, it seems to me unlikely that this debate could be settled on narrowly textual grounds. Too many passages cannot be interpreted without making the very assumptions that are up for debate. For instance, it is hard to have an argument about whether a reference to “self” concerns the empirical self when one party to the debate does not think Kant grants the existence of empirical selves (see e.g. A107, quoted on pp. 140–1). Chignell’s discussion of the texts thus provides him with textual cover for his view, but it is his non-textual arguments, especially his “parity argument”, that must do most of the convincing (pp. 142, 145–6).

According to the parity argument, outer sense allows us to cognise not only states of outer objects but also thereby the outer objects themselves, so why should the same not be said in the case of inner sense? This argument essentially claims explanatory higher ground. It is on those who would deny this common-sensical position, Chignell argues, to show what is wrong with it (e.g. p. 145). But I’m not sure Chignell’s opponents should grant him this advantage. There are some obvious differences between outer sense and inner sense: they have different forms and different objects. And if we cannot get further than these obvious differences, that is because of the plain fact that one is much more mysterious than the other. It remains very unclear how inner sense gets populated, what the denizens of inner sense are, whether inner sense has a proprietary manifold, etc. Assuming that there is good reason for such widespread bafflement, it seems one could just turn the tables and deny that it is reasonable to assume parity between two things when one of them is so much more mysterious than the other.

It might seem that Chignell still retains a field advantage, for is it not another plain fact that we each have a lot of knowledge about ourselves? Empirical knowledge that must have involved inner sense? I think the strongest challengers to Chignell’s view agree with this. What they deny is that the way we acquire this knowledge is akin to the way we cognise outer objects. Chignell offers a good example for illustrating this. He notes that there are cases in which we can order our mental states without referencing anything external. For instance, “When I close my eyes and count off numbers in my head […] I am aware that I must have said ‘10’ after I said ‘9’ because I know something about the order of the natural numbers” (p. 152). This seems right, but first of all, it appears that my ability to order my mental states is parasitic on my knowledge that I’m counting. And how do I know that? (Some, such as Kitcher, would argue that it is through apperception; cf. Chignell p. 139 and p. 139n.3.) More importantly for present purposes, it does not seem that the way I use the knowledge that I’m counting to order my mental states involves anything like synthesising my inner states in accordance with the categories to bring my inner intuitions to concepts. The resulting knowledge certainly depends on inner sense, and is knowledge of my inner states, but is it inner cognition? A similar point could be made for temporal determinations of my perceptual states. Whatever more is needed to move from the perceptually-based judgement “The ship is sailing downstream” to the judgement “I am perceiving the ship sailing downstream”, it is not like the process involved in getting to the judgement “The ship is sailing downstream” in the first place. Though there is much more to discuss in Chignell’s essay, including an exceptionally clear and helpful discussion of the Refutation of Idealism (§§8.4–8.5), I turn now to introduce Kitcher’s radically different view of self-knowledge.

In ‘A Kantian Critique of Transparency’ (Chapter 9), Kitcher presents an account of Kant’s theory of self-knowledge that is grounded on apperception rather than on inner sense (§9.2). She argues that on Kant’s view, even rudimentary instances of concept application involve the self-conscious and rule-governed production, reproduction, ordering and combination of mental representations. Counting, for instance, requires consciously producing and combining units in conformity with the counting rule. Because of the self-conscious nature of this process, when a cognizer applies, say, the concept nine by counting up nine units, her judgement is always accompanied by awareness of the grounds of her judgement: in this case, the partial representations combined (the nine units) and the rule followed (pp. 163–4; Kitcher is elaborating on A103–4). Moreover, because a cognizer produces and combines her representations in accordance with a rational rule, this generates rational and thus necessary connections among all her representations: a unity of self-consciousness (pp. 164–5). This, Kitcher says, is essentially the argument given in the A-Deduction; it moves from an “analysis of conceptual cognition to the theory of apperception” (p. 165). And the account that emerges ties together cognition and self-knowledge such that it is not possible to have one without the other: I cannot make a judgement without knowing my grounds; I cannot draw an inference without knowing my premises; and I cannot engage in any of these conceptual activities without being conscious of myself as doing so.

Kitcher goes on to defend and develop this view. But her central aim is to use it to challenge what she calls the “transparency thesis” (§§9.1 and 9.3), so I focus on her objection. A preliminary difficulty to flag is that there are many different transparency theories. The variations among them are substantive, and some internal debates turn on issues closely related to Kitcher’s concern here (see e.g. Byrne 2011 and Boyle 2011). Since Kitcher talks mostly about Gareth Evans, I take her challenge to be directed specifically at his account (p. 171). As Evans famously puts it,

in making a self-ascription of belief, one’s eyes are, so to speak, or occasionally literally, directed outward—upon the world. If someone asks me ‘Do you think there is going to be a third world war?’, I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question ‘Will there be a third world war?’ (Evans 1982:225–6, quoted on p. 158)

What about this claim makes Kitcher say that Kant is the “natural enemy” of transparency views (p. 172)? As best I can tell, Kitcher takes Evans to be enjoining us not only to direct our gaze at the world, but also to avert our gaze from ourselves—to screen out everything internal (pp. 167-68). As she puts it, he requires us to be “concerned only with external and not with mental factors” (p. 172). If the transparency view says what Kitcher says it does, then it does seem both problematic and deeply at odds with her Kantian account; but I think another, more attractive reading is available. It is true that Evans says such things as that there is “no need for the inward glance” (1982:226, quoted at p. 159). But he also says, in the passage Kitcher quotes, that we are only “occasionally literally” directed to look outwards. And in rejecting the “inward glance”, Evans is only rejecting introspection—making myself the object of attention. We are told to look outward not in order to avoid all “mental factors”, but so that we put them in play and catch them or their effects in action. If I quit searching through my memory to see whether I like a certain fruit and just take a bite instead, this is in order to see how I like itto catch my response.

It seems to me that Evans, far from being at odds with Kitcher, is a useful ally. What I’m not sure about is whether Kitcher and Chignell could be allies. To what extent are inner sense-based accounts of self-knowledge and apperception-based accounts of self-knowledge compatible? Could they be complementary, perhaps even mutually dependent? Or are there fundamentally different accounts of apperception lurking beneath and leaving them at odds?

The Nature of Judgement

In ‘Judging for Reasons’ (Chapter 10), Jessica Leech takes up Kitcher’s analysis of cognition in order to shed light on Kant’s account of the modality of judgements. According to Leech, the modality of a judgement identifies the role the judgement plays in reasoning (p. 174). To use her own example, in the syllogism

If Socrates is wise, then Socrates has a virtue

Socrates is wise

Therefore, Socrates has a virtue

“Socrates is wise”, as tokened in the major premise, is a problematic judgement, but as supplying the minor premise, it is an assertoric judgement. “Socrates has a virtue”, by contrast, is an apodictic judgement. In this essay, Leech’s aim is to address a problem this view runs into. For it entails that if one assumes, as is plausible enough, that every judgement must have some quality, quantity, relation, and modality, then every judgement must be a part of a course of reasoning. And this seems wrong; it rules out what seems to be an undeniable fact, namely, that we find “judgments sometimes just popping up in isolation”, as Leech puts it (p. 174). It seems to me that there is another puzzle, too, for according to Leech’s analysis, problematic judgements do not occur independently but only as embedded in complex judgements. In any stretch of reasoning, then, we will find acts of assertoric judgement and acts of apodictic judgement—but nothing to describe as an act of problematic judgement. This seems odd, and perhaps not entirely independent of Leech’s original puzzle.

Leech finds a resource to help solve this problem in Kitcher’s interpretation of cognition (§10.4). Since, according to Kitcher, all concept use involves reasoning, it follows that all judgement is part of a course of reasoning, and it is only fitting that every judgement has a modality. Though Kitcher’s account thus fits nicely with Leech’s and provides independent motivation for the claim that every judgement has a modality, it is not clear to me how this addresses Leech’s opening puzzle. What do we end up saying about those random passing judgements that pop into my head? But Leech does not stop with Kitcher’s account. Noting that Kitcher’s account of cognition is quite demanding, she sets out an alternative (§10.5). Her account begins with the claim that “judging requires combination”, and combination, in turn, requires unity of consciousness and awareness of this unity. But both unity of consciousness and awareness thereof, Leech argues, are provided by “the activity of judging for reasons”. Thus “judging has to be judging for reasons” and is “part of a course of reasoning” (p. 184). This argument is utterly lucid, but it completely eludes my grasp. To begin with, I’m not sure whether Leech intends for this to be taken as an interpretation of the opening moves of the B-Deduction or whether it is offered as more of a loose re-construction. She gives us a well-marked textual trail (pp. 184–5), but that seems inconclusive. Second of all, Leech’s proposal seems to hold out no hope of resolving her opening puzzle. If judging has to be judging for reasons, how could we accommodate random judgements? Finally, it is not clear what makes Leech’s account different from Kitcher’s: it seems to be the same account, and thus bound to be exactly as demanding. Leech’s discussion provides some further clues: she says that whereas Kitcher’s account appeals to “the nature of concepts”, she starts from “considerations of unity, i.e. the transcendental unity of apperception” (p. 183); she also characterises this unity as requiring “sufficient representations which have been combined through application of the logical functions of judgment” (p. 186, my emphasis). I’m not quite sure how to think about this characterisation of transcendental unity, but this just returns me to my first difficulty: is this Kant’s conception of transcendental unity, or a Kantian conception of transcendental unity?

But there is another line of thought that Leech clears the way to, but does not take. The unity of apperception is, on Kitcher’s account, a unity forged by double-strength, conscious and rational relations obtaining among our representations: my mental states are connected by the logical relations between their contents and by my awareness of making the transition. On some accounts, just one of these might be sufficient—at least sometimes, under certain conditions. So perhaps we can make simple inferences without being conscious of the mental transition, and the unity among my judgements will be preserved by the logical relation between their contents. Perhaps I can also make some mental transitions that are not logical but that remain unified by the fact that I am conscious of making the transition—for instance, when I set myself to brainstorm a topic. Though I’m not sure that this proposal helps with random judgements, it is weaker than Kitcher’s view and perhaps still defensible as Kant’s view.

Jill Buroker turns to another aspect of judgement in ‘Kant on Judging and the Will’ (Chapter 11), questioning whether Kant takes the will to play a legitimate role, directly or indirectly, in theoretical judgement. Buroker concurs with Andrew Chignell that the will is directly involved only in coming to have the kind of attitude Kant calls Glaube, in which one takes a proposition to be true on the basis of non-epistemic, subjectively sufficient grounds (§11.3, citing Chignell 2007a, b). To this Buroker adds a further, indirect but crucial, role for the will. Drawing on work by Alix Cohen, Buroker argues that the will must drive and direct the examination and improvement of our epistemic maxims (pp. 201–2, citing Cohen 2013).

Buroker’s analysis depends on understanding judgement as something that “involves thinking a proposition under some mode of assent” (p. 192). To judge is to assent. Buroker simply states this claim here, but it does not seem obvious to me. Is it not possible to make a judgement that does not count as knowing or having an opinion or being persuaded? For instance, say I’m attracted to a metaphysical claim. Kant says one cannot have an opinion about metaphysical claims. I certainly do not know anything yet; nor have I persuaded myself, because I do not take myself to have subjectively sufficient grounds. But I’m not just entertaining a thought, either. Where does this kind of judgement fall? Everyday empirical judgements are puzzling, too. I think that Buroker takes subjective grounds of assent to be the grounds I would state if asked to produce my grounds (p. 193). But it seems that many of my everyday empirical judgements are such that, if I were asked to produce my grounds, I would instead retract the judgement, as the prompting gets me to realise that my grounds of judgement were inadequate. But if this is the correct description of what is going on—if I’m retracting a judgement because I realise my grounds of judgement were inadequate—this suggests that I really did make a judgement, and that the judgement had grounds that I could later recall, but that these grounds cannot be understood as subjective grounds, since they are not the grounds I state when asked to produce my grounds. This suggests to me that there is not just room but need for the idea of judgement that does not yet involve assent. However these questions are resolved, I think Buroker’s conclusion remains untouched: reflection on the relation between judgement and will should, she argues, lead to a deeper understanding of the relation between theoretical and practical reason and a reaffirmation of the primacy of the practical (§11.5).

The Nature of the Self

The last section of the book focuses on the nature of the self. Tobias Rosefeldt (Chapter 13) offers a detailed analysis of the First Paralogism. Both Ralph Walker (Chapter 12) and Paul Snowdon (Chapter 14) discuss the Paralogisms within the larger context of transcendental idealism—a view that Walker finds “much that is right” with (p. 206) and Snowdon finds wholly implausible (p. 260). This is the only place I disagree with an editorial decision: it makes more sense to me to begin with Rosefeldt’s close analysis of the thinking subject and then to pan out to consider Walker (Chapter 12) and Snowdon (Chapter 14) together, so I proceed accordingly.

In ‘Subjects of Kant’s Paralogism’, Rosefeldt challenges a widely accepted interpretation of the First Paralogism (see p. 223n.3). Here is the argument, as presented in the B-edition and translated by Rosefeldt:

What cannot be thought otherwise than as subject does not exist otherwise than as subject, and is therefore substance.

A thinking being, considered merely as such, cannot be thought otherwise than as subject.

Thus a thinking being, considered merely as such, does not exist otherwise than as subject, and is therefore substance (B410–11, quoted at pp. 222–3).

According to the popular reading, what vitiates the argument is an ambiguity in the term ‘subject’: the major premise understands it in contrast to ‘predicate’ and thus states a familiar definition of substance as absolute subject; the minor premise takes ‘subject’ in the sense that contrasts with ‘object’ and thus expresses, as Rosefeldt puts it, “a fact of which we become aware in self-consciousness” (p. 223). Rosefeldt rejects this reading. There is clear textual evidence, he says, that ‘subject’ must be taken in the bearer-of-predicates sense throughout (A348, p. 223). Where, then, is the fallacy? On Rosefeldt’s proposal, the premises of the Paralogism, disambiguated, say:

What “cannot be thought otherwise than as subject if […] given in intuition does not exist otherwise than as subject, and is therefore substance.

A thinking being, considered merely as such, “cannot be thought otherwise than as subject […] in mere thinking and independently from how [it] is given in intuition” (p. 229, boldface added).

(It is important to remember to read the minor premise as a claim about how a thinking being represents herself in thoughts about herself [pp. 223, 233].) Rosefeldt makes a compelling case that his interpretation is more faithful to both the letter of the text (pp. 223–4) and its systematic spirit (pp. 224–5, 233–6) than the standard reading. He also situates his analysis within an account of the distinction between the logical and the real that is clearly of wide-ranging significance (§§13.3, 13.5, 13.6).

But there is one way that the ‘I’ appears in the Paralogisms that I’m not sure Rosefeldt has accommodated. Kant says that the Paralogisms arise when the ‘I think’ is taken problematically (B406), i.e. “only in its mere possibility, in order to see which properties might flow from [it]” (A347). What the rational psychologist is really doing is mistaking a mere “logical exposition of thinking” for a “metaphysical determination of the object” (B409). I take it that the former is Rosefeldt’s logical ‘I’; the latter is certainly the rational psychologist’s ‘I’. But Kant goes on to say that there is another way to say ‘I think’: assertorically, in an empirical proposition that asserts my existence in my consciousness of my thinking, for “apperception is” Kant says “something real” (B419, also B418, B422n.). This ‘I’ of apperception, as I call it, seems to be something distinct from the logical ‘I’. I think the same distinction is found in the B-Deduction. When Kant says that “the ‘I think’ must be able to accompany all my representations”, he adds that this representation of myself—‘I think’—is an act of spontaneity produced by apperception (B132). The ‘I’ of apperception is again distinct from the logical ‘I’. But in what way “distinct”? Rosefeldt says that “when we are conscious of ourselves as thinking beings, we represent ourselves by the representation ‘I’” (p. 226). His view seems to be that apperception is represented by the logical ‘I,’ and there is thus no need to introduce an ‘I’ of apperception. But the passages I quoted seem to me to suggest a different relation: apperception produces the logical ‘I’. So when I represent myself using the logical ‘I’, I take this self-representation to be produced by something that exists and is the spontaneous ground of this self-representation. It seems to me that there are many passages in which Kant talks about the ‘I’ in ways that are hard to accommodate with a purely logical conception of the ‘I’. But it may also be true that it is just as hard to give a coherent account of anything thicker.

In ‘Self and Selves’ (Chapter 12), Walker fills out the picture of the Kantian self, setting the thinking ‘I’ of the Paralogisms alongside other conceptions of the self. He begins by addressing two problems that he takes to threaten the metaphysical cogency of Kant’s view: the problem of noumenal agency (§12.2) and the limits transcendental idealism sets on knowledge claims (§12.3). He then turns to characterise the Kantian self. He proceeds by comparing and contrasting Kant’s view with Locke’s: Kant’s noumenal self with Locke’s real essence, Kant’s human being with Locke’s man, and Kant’s unity of apperception with Locke’s continuity of consciousness (§12.4). This turns out to be a highly illuminating comparison, though I would contest some of Walker’s claims. For example, according to Walker, Locke associates personhood and personal identity with continuity of consciousness, whereas Kant associates them with the human being (p. 214). But it is significant that Kant’s human being is not just a body or organism, as Locke’s man is. Kant’s human being is a body plus a soul, a unity of apperception. It is thus not just an object of outer sense: both body and soul must persist for the human being to persist (B415), and when Kant attributes personhood to the human being, it is not without also connecting it with a unity of apperception. In his final section, Walker explores some of the implications of Kant’s view, considering, for instance, the problem of other minds, usefully framed within a Kantian moral framework (§12.5).

Thus on Walker’s evaluation, there are insights still to be gleaned from Kant’s theories (p. 204), and though transcendental idealism is a coherent, even attractive metaphysics (p. 206), Kant’s views can be usefully considered without committing to the full apparatus of transcendental idealism (p. 206). A diametrically opposed view is presented by Snowdon, who frankly declares in ‘The Lessons of Kant’s Paralogisms’ (Chapter 14) that Kant’s ideas “do not really help us make progress in understanding the nature of the self” (p. 246). According to Snowdon, Kant’s project is thoroughly infected by his irremediable metaphysical commitments. Nothing is salvageable, not even Kant’s normally celebrated challenges to rational psychology (p. 253).

Though Snowdon and Walker clearly see Kant in thoroughly different ways, there is one thing they have in common. Snowdon asks:

If Kant’s conception of the categories is, roughly, that they are the fundamental concepts that humans must employ in their thinking about the world, why was the concept of the first person […] not on that list? (p. 247)

And more generally:

Why […] cannot there be […] synthetic a priori knowledge about ourselves? (p. 253)

These are objections that Walker echoes. According to Walker, Kant “obfuscates things by his reluctance to say […] we can know certain truths about the ‘I’ as subject” (p. 213). For it is a condition of experience that there is an ‘I think’ (p. 212); that is a bona fide piece of synthetic a priori knowledge, as is the fact that this self “is unitary, and endures through time; that it is active” (p. 207). According to Snowdon, Kant should assimilate self-knowledge to empirical knowledge (p. 261). According to Walker, Kant should assimilate self-knowledge to the rest of our metaphysical knowledge (pp. 212–13).

But it seems to me that there might be good reason why Kant does not do either of these things. Self-consciousness, after all, is unlike any other relation the self stands in. This does not mean Snowdon and Walker are wrong, but only that a move that assimilates the self to other metaphysical categories requires special defence: it is not fully justified by appeal to similarities between the self and other things; it must also preserve the distinctiveness of the self. This volume sheds light on some general features of the mind: the nature of intuition, the nature of judgement. It also draws attention to some features of the mind’s special relation to itself: the nature of inner sense, self-awareness, and self-knowledge. It thus provides the tools to consider whether, in what way and to what degree the self is distinctive from the rest of nature—an issue that certainly continues to occupy us today.


[1] See also i.a. A115, A116, A119–20, A320/B376, B164–5; FM, AA 20:276; Anth, AA 7:128.

[2] Cf. McLear pp. 101–3, Stephenson p. 106 and pp. 120–3, and Bader, Chapter 7.



Allais, L. (2015), Manifest Reality: Kant’s Idealism and his Realism (Oxford: Oxford University Press).

Block, N. (1995), ‘On a Confusion About a Function of Consciousness’, Behavioral and Brain Sciences 18: 227–87.

Boyle, M. (2011), ‘Transparent Self-Knowledge’, Proceedings of the Aristotelian Society Supplementary Volume 85: 223–41.

Byrne, A. (2011), ‘Transparency, Belief, Intention’, Proceedings of the Aristotelian Society Supplementary Volume 85: 201–21.

Chignell, A. (2007a), ‘Belief in Kant’, the Philosophical Review 116: 323–60.

——— (2007b), ‘Kant’s Concepts of Justification’, Noûs 41: 33–63.

Cohen, A. (2013), ‘Kant on Doxastic Voluntarism and its Implications for Epistemic Responsibility’, Kant Yearbook 5: 33–50.

Evans, G. (1982), The Varieties of Reference (Oxford: Clarendon Press).

© Yoon Choi, 2018.

Yoon Choi is Assistant Professor of Philosophy at Marquette University, USA. She received her Ph.D. from the University of Cambridge.