1 Keith Allen Korcz, ‘The Causal-Doxastic Theory of the Basing Relation,’ Canadian Journal of Philosophy 30 (2000) 525-50. All further references to Korcz will be to this article.

2 Though Korcz does not say this, one may claim that we hold people at fault for failing to believe p when they believe r and should have seen that r entails p and that this gives us reason to think that the meta-belief is sufficient to establish a basing relation between r and p. Here, however, the problems with drawing conclusions based on deontological considerations remain. It is highly unlikely that beliefs about basing explain such intuitions. This is especially true since we are not even considering one's belief that p; rather, we are interested (for, e.g., moral, prudential, or epistemic reasons) in one's coming to believe that p. Reasoning is sufficient to do this.

3 Keith Lehrer, ‘How Reasons Give Us Knowledge, or The Case of the Gypsy Lawyer,’ Journal of Philosophy 68 (1971) 311-13

4 In ‘How Reasons Give Us Knowledge, or The Case of the Gypsy Lawyer,’ Lehrer argues that the lawyer had to have a correct answer to the question ‘How do you know that p?’ in order to know that p. It is important to see that such a theory of the basing relation is not interpersonal. In particular, what one would say under certain circumstances is irrelevant to whether one's belief is doxastically justified. In a later presentation of the case, however, he claims that the lawyer's justification for believing that p is based on the good reasoning. Perhaps this is what has encouraged the common view that his theory of basing is interpersonal. See Keith Lehrer, Knowledge (New York: Oxford University Press 1974), 122-6.

5 It is not until after the presentation and evaluation of the case that Korcz stipulates that no such causal relations obtain. Since he is appealing to our intuitions about the case as it is presented, one's initial intuition may be explained by one's implicitly supposing that the lawyer's belief does come to be causally sustained by his good reasoning, as typically happens in these kinds of cases.

6 Put differently, even after the addition of this evidence I may be believing p because of and only because of the reasons I had all along. The belief has not now come to be based on more or different reasons.

7 Korcz makes no claims about the justificatory status of the doubtful lawyer's belief. Nevertheless, one might reason as follows. The persistent lawyer's belief is better justified than the doubtful lawyer's belief. If this is the case, then it must be based on something better — i.e., the meta-belief. Therefore, it must be based on the meta-belief. Such reasoning fails, however, as it is subject to the considerations developed above regarding the case of the persistent lawyer. Further, it is also unlikely that basing is needed, or even appropriate, to explain such a justificatory difference. It seems, rather, that a theory of propositional justification can do all of this work, even if we suppose that such a difference exists.

8 Here and throughout this paper a reason is understood exactly as Korcz has understood it — that is, as a c ausally efficacious state of a person. Such states include beliefs as well as perceptual states (526). According to Korcz's notation, therefore, ‘r’ is shorthand for one of two things: either ‘belief that q, where q is evidence for p,’ or ‘perceptual state s, where s is evidence for p.’

9 The following argument is a reformulation of a displayed argument that Korcz presents. I have formulated the premises more carefully than Korcz formulated them in his paper in order to highlight the argument's structure and to make more explicit the mistaken reasoning. They do, however, still capture the meaning of each of the premises.

10 I think this is the best reading of (P1), but an alternate reading is as follows. (P1*): If S shows that S's belief is justified by r, then it is possible that S's believing that p is justified on the basis of r. To see that (P1*) is not relevant to our present concerns, consider the following. Two conditions are sufficient to make (P1*) true, and neither of them have to do with intuitive considerations regarding what it is to show that one's belief is justified. In the actual world either S shows that S's belief is justified by r or S does not show that S's belief is justified by r. If the latter, then the antecedent is false, and the conditional is trivially true. If the former, then it is plausible to think that the conditional is true, for it is plausible that some possible world exists in which S's believing p is justified on the basis of r. This, however, is not to tell us anything else about which world this is or about what else is true in that world. In particular, it does not tell us whether in that world S's belief that p is causally related to r or, more importantly, whether in that world S even shows that her belief is justified by r. These are the considerations which are relevant to showing that a non-causal condition needs to be included in an analysis of the basing relation, and they are left unaddressed by (P1*). What we want is for a possible world to exist in which both the antecedent and the consequent obtain, and this is what (P1) captures.

11 In the following I have omitted for ease of exposition any reference to the standing condition that S believes that p. Of course, doing so does not affect the crucial point here.

12 If we substitute (P1*) for (P1) in the argument from showing the argument still fails. To see this, substitute for (P1’) (P1*’): (A • B), and consider the following model. In the actual world A, ∼B, and C are all true, and in the only other possible world (a world accessible from the actual world) ∼A, B, and ∼C are all true. On such a model both of the premises are true while the conclusion is false, and therefore the inference from (P1*’) and (P2’) to (C1’) is invalid. Note also that formulating (P1) as a conjunction will not help either. Nor will formulating it as a material conditional with reference to an appropriately specified set of conditions in which one shows one's belief to be justified (thereby avoiding the modal language). In short, on all plausible interpretations of (P1), the inference is invalid.

13 Here, again, note that ‘p’ represents a proposition while ‘r’ represents a (causally efficacious) state of a person that is evidence for p. Since space does not allow me to consider (RCT) here, in what remains I will only focus on (NCC). As a result, I will not explicate (RCT) in this paper. For its explicit formulation, see Korcz (547).

14 That is, everything I believe is made justified by r.

15 Contrast the following that I think the non-causal theorist should be committed to. If the meta-belief had contained p (i.e., Ezekiel saw an armadillo) as a constituent, then, all else being equal, p would have been based on r. This, of course, is importantly different from Ezekiel's meta-belief in the example under consideration.

16 I would like to thank Richard Feldman for his very helpful comments on earlier versions of this paper. Thanks also go to John G. Bennett for bringing to my attention some parts of this paper that needed to be clarified.