Abstract

False-premise-inference (FPI) cases—where a subject appears to acquire inferential knowledge by inferring from a false premise—challenge the standard view that inferential knowledge must be grounded in true premises. Existing responses, knowledge from falsehood and knowledge despite falsehood, prove unsatisfactory. We defend a third alternative: knowledge from partial truth. In FPI cases, the subject’s inferential knowledge is epistemized not by the false premise as a whole but by a true proper part of its content. Drawing on a fine-grained theory of propositional content, we show how to isolate this knowledge-generating component. Our approach resolves the puzzle by preserving the natural reading of the FPI cases while upholding the principle that knowledge-generating inferences are truth-conducive.