Knowledge and Skepticism

3. Truth Tracking and the Problem of Reflective Knowledge

3 Truth Tracking and the Problem of Reflective Knowledge Joe Salerno In ‘‘Reliabilism Leveled,’’ Jonathan Vogel (2000) provides a strong case against epistemic theories that stress the importance of tracking/sensitivity conditions. A tracking/sensitivity condition is to be understood as some version of the following counterfactual: (T) ~p o! ~Bp (T) says that s would not believe p, if p were false. Among other things, tracking is supposed to express the external relation that explains why some justified true beliefs are not knowledge. Champions of the condition include Robert Nozick (1981) and, more recently, Keith DeRose (1995). To my knowledge, the earliest formulation of the counterfactual condition is found in Fred Dretske’s conclusive reasons condition (1971), which says, s would not have had the reason that she does for believing p, if p were false. Vogel contends that any such counterfactual condition on knowledge will render the theory of knowledge too strong. He believes that there is at least some possible reflective knowledge that cannot satisfy the counterfactual— namely, the possible knowledge that one does not believe falsely that p. The alleged impossibility of such reflective knowledge is taken by Vogel to be a decisive objection to the tracking theories advocated by Dretske, Nozick, DeRose,1 and others. The criticism finds its roots in Vogel’s earlier work (1987), and recurs in papers by Ernest Sosa (2002, 1996). Sosa suggests that the externalist idea behind tracking is on target, but that Nozick’s counterfactual is a misbegotten regimentation of the idea. In its place Sosa offers his own counterfactual ‘‘safety’’ condition, which he feels properly captures the externalist idea. Sosa’s counterfactual is not the topic of this paper. I mention it only to point out that the criticism that constitutes the subject of my investigation is meant to do a lot of work. In Sosa’s case the criticism is meant to motivate his own counterfactual analysis, and in Vogel’s case the criticism promises to be a silver bullet against a theory that has recently found renewed life in the work of Keith DeRose. It will be argued here that the criticism is misguided. My belief that I do not believe falsely that p can track the truth. The Counterexample to Truth Tracking Consider the proposition that I do not believe falsely that p. I should be able to know a proposition like that. Sometimes I know p and double-check my sources, thereby coming to know, additionally, that I do not believe falsely that p. Vogel’s concern is that this kind of epistemic confidence cannot track the truth: If my belief that I am not mistaken were false, I would still believe that I’m not mistaken. Vogel begins by translating the claim of epistemic confidence, ‘‘I do not believe falsely that p’’ as shown. Call it (EC): (EC) ~(Bp & ~p) Formally, (EC) reads, ‘‘it is not the case that both I believe p and p is false.’’ The tracking condition, again, says, if p were false it would not be believed that p: (T) ~p o! ~Bp Knowing (EC) then requires satisfying the following instance of (T): (T*) (Bp & ~p) o! ~B~(Bp & ~p) It says this: If I were to believe p falsely, then I would not believe that I do not believe p falsely. Evaluating this counterfactual is tricky business. Nevertheless, Vogel argues that clearly it is not satisfiable, because he thinks, ‘‘If you believe p, you believe that you do not falsely believe p’’ (2000, 611). Formally, (*) Bp ) B~(Bp & ~p)2 Of course, if (*) is valid then (T*) is not satisfiable. In the relevantly close worlds where the antecedent of (T*) is true, you believe you are not mistaken . That is because, by (*), in every world where the antecedent of (*) is true, you believe you are not mistaken. So if (*) is valid and (EC)—that is, ‘~(Bp & ~p)’—properly regiments the claim ‘‘I do not believe falsely that p,’’ then it is not possible to track the truth of this claim. Therefore, if Vogel’s logical resources are in order and tracking is a necessary condition on knowledge, then, absurdly, one cannot know that one does not believe falsely that p. 74 J. Salerno I Believe Higher-Order Consequences of What I Believe This is a good place to point out that both the assumption that (*) is valid and the assumption that (EC) best captures the expression of epistemic confidence are questionable. I address...