Abstract

A familiar form of scepticism supposes that knowledge requires infallibility. Although that requirement plays no role in our ordinary epistemic practices, Barry Stroud has argued that this is not a good reason for rejecting a sceptical argument: our ordinary practices do not correctly reflect the requirements for knowledge because the appropriateness-conditions for knowledge attribution are pragmatic. Recent fashion in contextualist semantics for 'knowledge' agrees with this view of our practice, but incorrectly. Ordinary epistemic evaluations are guided by our conception of a person's standing with regard to the reasons that there are for and against the truth of a belief. Thus the objection from our ordinary practices is sound: fallibility is not an epistemological shortcoming, and a convincing sceptical argument must use only requirements which figure in ordinary epistemic practice