One of the many interesting aspects of Kant's Thinker is that Patricia Kitcher does not only want to give a reconstruction of the Kantian account of the relation between self-consciousness and our cognition of the world but also argues that this account is philosophically convincing and has the potential to solve certain problems for which contemporary discussions have no satisfying solution. The argument for this claim is found mainly in chapter 15. Some parts of this chapter are inspired by a discussion Kitcher and I had some years ago and deal with an objection I raised against her interpretation then. In this commentary, I will try to show that the thrust of my earlier objection is still untouched by Kitcher's response to it.

Before I do so, I want to recapitulate two claims Kitcher makes in the central chapter of her book, i.e. in chapter 9. The first claim is that Kant does not only claim that we are conscious of ourselves as the identical subjects of the various kinds of representations that make up our cognition of the world but that he also gives a subtle argument to the effect that we could not have cognition of the world without such self-consciousness. He ‘argues for apperception’ – to quote the title of chapter 9. The second claim is one about what the argument is. In Kitcher's view this argument can be summarized as having two main premises. The first premise claims that we would not be rational cognizers if we were not able to have beliefs for certain reasons, i.e. on the ground of other beliefs. The second premise claims that we could not believe something on the ground of other beliefs if we were not conscious of these other beliefs as our grounding beliefs and hence conscious of these beliefs as belonging to one and the same subject as the grounded belief. Although I have qualms about whether these two claims – the claim that Kant argues for apperception and the claim about how he argues for apperception – correctly represent the argumentational structure of the transcendental deduction, I will confine myself here to the question of whether the argument Kitcher ascribes to Kant – no matter whether he held it or not – would be convincing.

My main objection in our earlier discussion was that Kitcher overintellectualizes the requirements for rational cognition. It could be rephrased in the following way. I agree with the first premise of the argument for apperception that in order to be rational cognizers we have to be able to hold certain beliefs on the ground of other beliefs. But I am sceptical about the claim made in the second premise that a person can only believe something on the ground of another belief if she is able to recognize the grounding belief as one of her own and as grounding the first belief. Imagine a 3-year-old girl who hears the sound of her parent's car in front of the house and correctly concludes ‘Mama is coming!’ This girl is rational insofar as she forms a belief on the ground of another belief. But it might well be that she is not yet able to recognize this grounding belief as such. I think we can distinguish two ways in which she might fall short of the requirements Kitcher's theory imposes on her.

1. (i) She might believe something for a certain reason without recognizing that this reason is a reason, for example, if she has not yet acquired the conception of some things being reasons for others. Let us call this the possibility of unreflected reasons.

2. (ii) She might believe that something is a reason for something else without being able to self-ascribe either her original belief or the belief that grounds it. For example, she might believe that the sound of the car speaks in favour of Mama's arrival but lack the ability to self-ascribe her own beliefs because she has not yet acquired a conception of herself as a believer. Let us call this the possibility of reflected but not self-conscious reasons.

If the first possibility were actual the contents of that person's mental life would display a certain rational structure without the person's recognition that they have this structure. If the second possibility were actual, the person would recognize a certain rational structure among the contents of her mental life without recognizing these contents as contents of her own mental life.

In section 3 of chapter 15, Kitcher presents a line of thought that she takes to be a successful reply to the objection just mentioned (cf. pp. 253ff.) Since the argument is rather subtle, complex and ‘a little unusual’ (to use her own words; cf. p. 254), and since it combines a direct answer to the objection with a discussion and critique of various contemporary accounts of self-consciousness, I will first try to make explicit its fundamental structure. In a second step I will try to explain why I am still not convinced by it.

The core idea of Kitcher's new argument can be spelt out by means of the following claim:

Given (C), it is easy to argue for the conclusion that believing something on a ground and hence rational cognition always implies self-consciousness, because it is undeniable that some believers are in fact conscious of the fact that they believe something on the ground of other beliefs they have. However, the ‘if not always, then never’-claim (C) is, of course, far from being self-evident.

As far as I can see, Kitcher's strategy to defend (C) is to show that some prominent attempts to explain how we can become conscious of our own grounding beliefs as such without assuming that we are already conscious of them fail and that they have to fail for systematic reasons. The attempt that features most prominently in Kitcher's discussion is that of Kieran Setiya (Reference Setiya2011). Setiya suggests that we can use a strategy that other authors have used in order explain our ability to become conscious of our own beliefs in an analogous way in order to explain how we can become conscious of the reasons for our beliefs as such. The original strategy is to explain the ability to self-ascribe beliefs through an acquisition of an epistemic rule such as the following:

Analogously, Setiya suggests the following epistemic rule features in our ability to recognize the beliefs that are our grounds for holding certain other beliefs:

Kitcher argues that appealing to (RULE 2) will not offer a convincing explanation for our ability to know that we believe something on the ground of another belief. Her objection is that the showing-relation alluded to in (RULE 2) can only be understood as a relation between beliefs, namely as a relation that holds between two beliefs just in case the one makes the other more credible to the believing subject herself. But then believing that a certain fact q ‘shows’ that p would already involve thinking about q and p as something that the subject herself believes and hence (RULE 2) would only work in a case where the subject already is conscious of the grounds for her beliefs as grounding beliefs that belong to herself. Since any explanation analogous to that of Setiya would face a similar problem, we should simply accept that we can only become conscious of the fact that we believe something on the ground of another belief, if using this other belief as a reason already is a mode of being conscious of the original belief as one's own. Or so Kitcher argues.

Let us now see whether Kitcher's argument is able to rule out the two possibilities I have mentioned before as threads to her theory, the possibility of unreflected reasons and the possibility of reflected but not self-conscious reasons. I want to raise two objections.

First, it seems clear that Kitcher's discussion of Setiya's (RULE 2) is an argument against the second possibility – the possibility of reflected but not self-conscious reasons. As I have explained above, this possibility would be actualized if someone believed that something is a reason for something else without being able to self-ascribe either her original belief or the belief that grounds it. Kitcher's argument against this possibility is based on the claim that the grounding relation expressed by the predicate ‘x is a reason for y’ (in Setiya's terminology, the showing-relation) can only be understood as a relation between beliefs, which implies a cognitive attitude of a thinker towards x and y. However, I think that this is far from obvious and in fact false. There are respectable alternative grounding-relations that do not hold between beliefs and do not yet involve cognitive attitudes of a thinker. One such relation is that which holds between a certain fact and a certain state of affairs just in case the first makes the obtaining of the second sufficiently likely. The fact that the car has arrived in front of the house makes it sufficiently likely that Mama is coming, for example. And the fact that a dog is in a room makes it sufficiently likely that an animal is in the room. In both cases we are dealing with conditional objective probability, in the second case even with one that is grounded in metaphysical implication, and it seems far from clear that we have to refer to beliefs or other mental states in order to understand what this kind of probability is.

Hence, we can use this relation in order to reformulate (RULE 2) in such a way that an explanation of self-consciousness by means of it would escape Kitcher's criticism:

By means of (RULE 2*) we could explain how a subject that has reflected reasons – e.g. believes that the fact that the car has arrived makes it sufficiently likely that Mama has arrived – learns to become aware of her belief that the car has arrived as a reason for her belief that Mama has arrived.Footnote ³

My second objection consists in the observation that Kitcher's argument seems entirely designed to argue against the possibility of reflected but not self-conscious reasons. However, this possibility was only one part of the original objection. The other part alluded to the possibility of unreflected reasons, i.e. the possibility that someone is a rational cognizer in that she believes something for a certain reason but does not recognize that this reason is a reason, for example because she has not yet acquired the conception of some things being reasons for others. Even if Kitcher's argument against the possibility of reflected but not self-conscious reasons were sound and we could not recognize that something speaks in favour of something else without being aware that it is a reason for us, this would have no bearing whatsoever on the question of whether someone could believe things for reasons without recognizing these reasons as such. And hence the original objection that there could be agents such as our 3-year-old girl who believe things for certain reasons without being aware of these reasons as such and as theirs stands unrefuted.

One might wonder whether Kitcher could argue in an analogous way that it would again be impossible to explain how such a person could ever become conscious of her own reasons as reasons, and hence impossible to arrive at a level of reflection on which (RULE 2*) could then help her to come to self-consciousness. Again I think that one could find an epistemic rule that makes the development from unreflected to reflected reasons understandable, namely the following:

One might object that a rule like this is problematic because it could require of someone to believe that the fact that q makes it sufficiently likely that p even if, in fact, this is not the case. However, although this could certainly happen, it does not speak against (RULE 3) – any more than the fact that a rule like ‘If you believe that p, and believe that if p, then q, you are entitled to believe that q’ could license someone to believe q although it is not the case that q. What both cases show is that the mentioned rules have to do with the internal rationality of believers and can be overruled by other rules that have to do with the fitting of one's beliefs with the world.