The ‘frame problem’ according to Jerry Fodor concerns the question of how a cognitive system like ours can frugally and reliably determine what is relevant for its processing tasks. A number of authors have pointed to heuristics as a solution to this question. A heuristic solution, however, leaves unanswered important epistemological problems concerning the rationality and truthfulness of human cognition. In this dissertation I will try to address these ‘epistemological relevance problems’ on the basis of a heuristic account of reasoning. Thereby, I will argue that we need to look beyond internal computational processes to our technological and social wherewithal to understand the characteristic success humans have enjoyed in their epistemic endeavours. More concretely, it will be my claim that collaborative reasoning allows us to enter a cumulative selection process in the direction of optimal solutions.