This article considers the different approaches to cognition in conversation analysis and discursive psychology. Its points are illustrated through a critical but appreciative consideration of an article by Drew in which he uses conversation analysis to identify ‘cognitive moments’ in interaction. Problems are identified with Drew’s analysis and the conclusions he draws. In particular, he a) presupposes a dualistic division between depth and surface; b) makes circular inferences from conventional conversational patterns to underlying cognitive entities; c) presupposes that the underlying (...) cognitive entities influence conduct. It is argued that none of these things is required by conversation analysis; rather Drew is imposing cognitivist assumptions on conversational materials. Discursive psychology’s approach focuses on cognitive issues in terms of how they are constructed and oriented to in interaction; its virtues are pressed. (shrink)