Abstract
I argue that DeRose's attributor contextualism cannot straightforwardly preserve the widespread view that, when a subject believes q solely on the basis of competent deduction from p, knowledge of q requires knowledge of p. I present a novel challenge to the compatibility of this widespread view with DeRose's contextualism, then argue that the tension can be resolved in only one of two ways: if DeRose rejects the widespread view or if DeRose accepts the existence of a range of contextualism-specific Gettier-style cases