Many authors argue that facts about how we form beliefs reveal constraints on what we ought to believe. Most famously, Nishi Shah and David Velleman (2005) argue that one cannot deliberate about whether to believe that p without deliberating about whether it is true that p, since the deliberative question “Should I believe that p?” is transparent to the factual question “Is it true that p?”. The best explanation for this, they claim, is that it is part of the concept of belief that one ought only to believe that p if it is true that p. I shall present reasons to reject both claims. First, deliberative questions are not always transparent to factual questions, since ignorance and uncertainty frequently prevent us from answering the former by way of the latter. Second, classical dispositionalist theories of belief provide better resources for explaining why only certain ways of deliberating result in the formation of beliefs. Such accounts have no implications about what one ought to believe, however, so we cannot appeal to facts about how we form beliefs to draw normative inferences about what we ought to believe.