Knowledge and belief fully display the pragmatic features that make of them different concepts only in third-person epistemic attributions. This is the main thesis of this paper, which has three sections. In section 1 I argue, following a pragmatic reading of Gettier, that agents on their own lights cannot tell the difference between what they know and what they believe that they know. The reason lies on the pragmatic peculiarities of normative notions, which according to Brandom’s normative expressivism amount to saying that first-person epistemic claims lack the required complexity to ground a complete contrasting analysis of the concepts of knowledge and belief. Section 2 deals with the norms of assertion and elaborates in more classical terms something that follows from Brandom’s treatment of assertions, i.e. that assertions are expressions of belief that must be taken as knowledge claims. Finally, in section 3, I propose to explain the link between third person ascriptions and first person avowals by borrowing one of Ramsey’s hints on truth ascriptions to derive the role of the latter from that of the former. First-person epistemic claims, I suggest, are essentially the result of reactive actions, being their role dependent upon the functioning of third-person attributions