1 Philosophical Studies 2006; 131: 729–42CrossRefGoogle Scholar; all quotations in the text are from this article.

2 BonJour, Laurence and Sosa, Ernest, 2003. Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues. Oxford: Blackwell PublishersGoogle Scholar.

3 Sosa also discusses a view he calls “Cartesian internalism.” We will discuss the prospects for this version of internalism later in the paper.

4 We will not discuss the parts of Vogel's article that concern the distinction between subjective and objective justification. We think that the main issues between Vogel and Sosa do not depend upon that distinction.

5 We follow Vogel's numbering.

6 The inference here is not from absence of truth to absence of justification. Instead, what grounds Sosa's Intuition is the view that your counterpart's belief is not the result of a truth-conducively reliable process. At your counterpart's beginning, he has no working memorial system at all. In a normal human, it is possible for a working memorial system to deliver mistaken outputs that have the feel of genuine memories. In such case, Sosa can allow that these outputs give rise to justified and yet mistaken beliefs. Thanks here for prodding from an anonymous referee.

7 We will consider the question of how the “puzzle” unfolds when the primary focus is the propositional justification, if any, that B has for your counterpart.

8 An anonymous referee pointed out that in some cases of doxastic justification – those not involving inference – diachronic considerations will be irrelevant. But in BC, as set up by Sosa, it seems that such considerations do come into play, as higher-order beliefs about reliability come into play after the initial formation of your counterpart's belief of B.

9 Note that (8’) is false as it stands. Suppose that your counterpart believes R(B) at t' and that this belief is doxastically justified. It does not follow that he also believes B at t' (though it so happens that in BC he does believe B at t'). The suitably modified general principle would be something like: if you're doxastically justified in believing R(X) at some time, and you infer to a belief of X at some later time, then the latter belief is also doxastically justified. In the text, we will ignore this complication.

10 We should stress that it is not a disaster if Sosa accepts (10’’), the claim that your counterpart has propositional justification for B at t. What Sosa needs to do is to avoid commitment to a contradiction concerning either doxastic justification or propositional justification, and we think that we have shown, contra Vogel, how to do this. An editor of this journal suggested that Sosa should not countenance (10’’), since its defense in the text, in effect, had Sosa allowing a justified status for a reliability belief that could figure in an inference involving CT (one that in fact does not occur in the version of the case considered in the text in the discussion of propositional justification). After all, said the editor, your counterpart has no memories to justify a conclusion of the reliability of B's source. If Sosa were to follow this path, there would still be no obvious way of sticking him with a contradiction regarding propositional justification.