Suppose that you possess strong, perhaps decisive, evidence for a proposition that you’re considering. It is natural to think that that you thereby possess a reason to believe that proposition. What explains this? Why is there such a connection between evidence for (or against) propositions and reasons for believing them? What, in other words, are the grounds of epistemic normativity? One answer to this question is constitutivism about epistemic normativity, the view that epistemic reasons are grounded in the nature of belief. This view promises to provide a happy middle-ground between extreme versions of metanormative realism about epistemic normativity and extreme versions of anti-realism. In this paper, we reject this view, principally by focusing on the arguments that have been given in defence of it. The claim that belief is constitutively normative has been alleged to explain a number of distinctive psychological features of belief. We argue that arguments of this kind fail for a common reason: they either rely on an implausible claim about following prescriptions, or they don’t best explain the psychological feature of belief in question.