This thesis argues against the Humean theory of practical reasons, criticising its foundations in philosophical and moral psychology. It develops a realist account of value-based reasons, underpinned by a distinctive cognitivist moral psychology, and a non-causalist account of the rational explanation of action. Contemporary Humeans reject Hume's own theory of thought, but this leaves the Humean theory of practical reasons without justification for a conception of desire as non-cognitive and not open to fundamental rational evaluation. Two possible strategies for filling this justificatory gap are (i) an appeal to grammatical considerations about the attribution of desires and their content, or (ii) an appeal to distinctions in respect of direction of fit. I argue that neither is successful. Kant's moral psychology provides the key to an alternative account, but is unsatisfactory due to its acceptance of a theory of thought which is relevantly similar to Hume's, and of non-compulsory rationalist presuppositions. Separated from these aspects, Kant's insights open a path to developing a conception of desire as essentially rationally evaluable. I argue that, in addition to such a conception, we should accept an account of rational attitudes as constitutively normative. On the basis of these two views, I argue that desire is a kind of evaluative belief. An independently plausible account of reasons takes them to be evaluative facts, and this neatly connects to the normative philosophical psychology. I consider the implications of such a view for the rational explanation of action, arguing that while causal theories of action and action-explanation are unacceptable, the normative philosophical psychology allows the development of non-causal alternatives to them. The non-causal account of action and action-explanation leaves space for an explanatory role for reasons themselves, beyond that provided by merely psychological explanation, as well as an explanatory role for an agent's character and emotions.