Truth-tracking and the value of knowledge

Sensitivity-based approaches to knowledge have not fared well in recent discussion, due primarily to perceived failure on the issue of preserving epistemic closure and to competition from the closely related notion of safety that both purportedly preserves closure and is easy to confuse with sensitivity. Recent work shows that the closure problem may not be decisive against a properly articulated truth-tracking account (Roush 2005), and thus there is some motivation for taking sensitivity-based theories more seriously, especially once we notice that safety-based theories do not preserve closure. Even so, there remains the issue of the value dimension of knowledge. As I have argued, even a theory of knowledge that is counterexample-free should be viewed with suspicion if it cannot accommodate or explain the value of knowledge (Kvanvig 2003). Modal epistemologies, which include sensitivity-based ones, are especially suited to the task of explaining the value of knowledge. The question is whether such approaches can sustain the promise which they present. As I will argue here, the answer is ultimately negative, but negative in a way that highlights the central importance of modal dimensions of the nature of knowledge. In section i, I lay out a variety of value problems, describing the logical landscape of value issues and identifying a special value problem on which I will focus here. That special value problem concerns the relationship between knowledge and its parts, and I use this problem to motivate taking seriously probabilistic accounts of truth-tracking and sensitivity over standard counterfactual approaches.