There is an epistemologically significant foundationalist structure in our empirical thinking, and it divides the propositions we accept into two classes, basic and derivative. An account of this structure needs to answer two questions: which propositions are basic, and whence do basic propositions acquire their warrant? A natural and ancient answer to these questions is that basic propositions are observational and that these propositions gain their warrant from perceptions. I will critically examine this view and also a variant view offered by Wilfrid Sellars. I will put forward an alternative conception, dialectical foundationalism, that sees the distinction between the basic and the derivative as issuing from a distinction between two kinds of reasoning, derivational and dialectical. This type of foundationalism provides a more accurate account of our empirical thinking, I will argue, than its traditional and Sellarsian counterparts.