1. I-Rule and Object-Rules

Rational empirical cognition (RE cognition) is, according to Patricia Kitcher, cognition of such a kind that the cognizer can give the reasons or grounds for it (Kitcher Reference Kitcher2011: 121). The I-rule, crucial to Kitcher's understanding of Kant, requires that all representations relevant to cognition belong with others to some continuing self (Kitcher Reference Kitcher2011: 123). The I-rule follows from the requirement that all representations that play a role in cognition must be potentially conscious, and that in turn is required because otherwise a cognizer could not know the grounds of his cognition and thus would not be a rational cognizer. In a complex argument at pp. 188–9 of Kant's Thinker, Patricia Kitcher argues that the I-rule functions in a unique way, different from all other rules:Footnote ¹

1. (1) The I-rule cannot be applied based on intuitions, because there is no intuition of the I.

2. (2) The I-rule can be applied only when some object-rule is applicable to the manifold of intuition, so that the understanding combines some representations and creates a relation of necessary connection among them.

3. (3) Categorial rules can only be applied via schematized rules.

4. (4) (3) does not apply to the I-rule. The I-rule is different insofar as it does not apply through a schema, but through the application of an object-rule.

5. (5) The I-rule is the highest and most general rule of cognition, but it is not an independent rule: it depends on the use of an object-rule for its application.Footnote ²

(1) and (3) are familiar Kantian claims and uncontroversial as such. (4) is on the one hand supported by Kant's remark that the I think belongs to the table of categories even though it does not have its own title there (CPR, A341/B399), on the other hand by the fact that there is also no schema for it. (5) spells out a consequence of (2). The crucial point thus seems to be (2). According to (2), applying the I-rule requires application of object-rules or, framed in terms of consciousness, self-consciousness requires some sort of consciousness of objects. Thus, according to (2), representations united in one self-consciousness also have to be unified by the categories. If this consequence could be established, the main aim of the Transcendental Deduction would be fulfilled – but it is in fact a highly problematic claim. Dieter Henrich was among the first to clearly state what is problematic here (Henrich Reference Henrich1988: 44–5): That all my conscious representations belong to one consciousness, and hence, that I must be able to accompany all of them with ‘I think’, is an analytic truth (except for representations that are impossible or ‘nothing for me’, CPR, B132). Now, this basic ‘togetherness’ of all representations that belong to one self-consciousness certainly is some form of unity or connection: all my representations have in common that they belong to the same I. However, the claim expressed in (2) is much stronger and more specific: It entails that the unity among my representations is based on certain principles, the object-rules or categories. Clearly, this is what Kant aims at, but the mere ‘togetherness’ of pure apperception does not directly yield this result. Further arguments would be required to establish these stronger forms of connection.

2. Does the Analysis of Cognition Imply the Existence of a Thinker?

As Patricia Kitcher points out, Kant accepts the widespread criticism that Descartes's cogito argument is intended as a syllogism and fails as such since ‘I exist’ is analytically contained in ‘I think’ and thus follows immediately (Kitcher Reference Kitcher2011: 194–6). However, according to Kant's own theory, the cogito does not seem to include any existence claims. He seeks to consider the cogito not as tied to any perception of an existent, but merely as a function that could be so applied (CPR, A347/B405). On the contrary, though, some important passages in the B Deduction and in the B Paralogisms suggest that the I-think does imply more than the continued existence of the actual cognizer, namely the possibility of a continued existence of cognizers per se. The first, analytic claim is that the actual thinker must exist just as long as her thoughts persist (i.e. perishes without the thoughts). The second, more ambitious one is that the thinker exists independently of her perceptions (and thus may exist longer than any of them, or without having any).

Usually, existence claims have to be synthetic and based on intuition for Kant. But here, Kant seems to suggest that, in the case of the I, existence can be established by mere thought, as Kitcher rightly observes. Kant argues that (1) in the original unity of apperception, I am not conscious of myself as I appear to myself, nor as I am in myself, but I am only conscious that I am (CPR, B157), and he goes on to argue (2) that the I-think expresses the act of determining my existence. Kitcher concludes: ‘Despite his critique of Rational Psychology, he thus came to appreciate that Descartes was nearly right about cogito ergo sum, that thought alone could establish existence’ (2011: 194). But what kind of existence could Kant have in mind here, and how would that sit with his transcendental idealism in general? Kitcher considers the possibility that Kant could here be claiming a strange ‘nonphenomenal type of existence’ for the apperceptive I, or he could be raising the possibility of some kind of ‘bare existence’ that is of no determinate kind. Neither option sits well with his overall theory.

Kitcher offers an attractive way to deal with this important puzzle that is based on a lengthy footnote at CPR, B422, in which Kant refers to the I-think as an ‘indeterminate empirical intuition’. Kant here obviously does not deny that cognition depends on intuitions, but he merely abstracts from any particular cognitive contents or intuition, though not from the fact that some kind of intuition must be given. Because of this abstraction, cognizers can imagine themselves as beings determined by thinking alone (or imagine themselves to exist only as thinkers, as in Descartes's second meditation). Hence, Kant need not postulate some odd sort of nonphenomenal existence.

Thus, according to Kitcher, that sort of existence is neither phenomenal nor noumenal nor determinate in any way, but just a consequence of the fact that self-conscious thought implies existence even when considered in abstraction from any particular form of intuition (but, of course, not in abstraction from all given perceptions). This is an elegant solution. However, I see one major exegetical difficulty: in the footnote at CPR, B422, Kant deals with the I-think as an empirical proposition, whereas in the context of the Deduction, he argues that my awareness of my existence is based on the original unity of apperception which is clearly a priori. Thus, it remains to be shown that Kant is really addressing the same issues in both contexts, despite the attractiveness of the proposed solution.

3. What Kind of Being is a Thinker?

Kitcher argues that the criticism of rational psychology Kant develops in the Paralogism chapter is a further elaboration of his positive theory presented in the Deduction. This is a controversial view that goes back to Wilfrid Sellars, among others (Sellars Reference Sellars1970/1); the opposite view is represented by Karl Ameriks who argues that Kant is indebted to the rationalist tradition more than his criticism suggests (Ameriks Reference Ameriks2000). However, I think this controversy is not only about how the Paralogisms and the Deduction are related but it is also a substantial philosophical one about what kind of being human subjects ultimately are for Kant. Whereas it is rather obvious that Kant rejects all attempts to hinge the subject on substances in the traditional sense (as, most prominently, in substance dualism, but also in materialism or Spinozistic monism), it is much less clear what his positive theory of the subject is. Broadly speaking, there are two possibilities:Footnote ³

1. (1) Minimal substantialism: Despite Kant's criticism of dualist metaphysics, Kant retains the idea that thought requires some substratum of an indeterminate kind. According to transcendental idealism, we do not know anything more about it, but we have to assume that it is there.Footnote ⁴

2. (2) Functionalism: The phenomenal self is a bundle or array of representations, based on the application of the categories (the categories understood as functions that bring about unity).Footnote ⁵ Since we have no cognitive access to the noumenal self, this option includes that there might be no substratum at all.

I am not sure where Kitcher stands here. Would she agree to one of these options or argue that there are more possibilities? A few passages suggest that she ultimately leans towards an epistemological reading in the sense that Kant replaces ontological claims about the self with merely epistemic ones. She writes, for instance: ‘The metaphysical proposition that any representation must belong with all others to a continuing self is replaced by the claim that in human cognition, empirical consciousness has a necessary reference to original apperception, that is to one (continuing) consciousness’ (Kitcher Reference Kitcher2011: 123). The problem I see with this reading is that it collapses, in my opinion, into one of the above options: If we assume that transcendental idealism does not allow for claims about the nature of the self, we necessarily have to allow for the entire range of metaphysical possibilities, including that there is no self distinct from the representations at all, i.e. that representations can exist by themselves, or that there is some kind of substratum. This is so because any epistemological reading of Kant does not solve the metaphysical questions involved but is merely silent about them. But that does not make these questions disappear.Footnote ⁶