Constitutivism is a kind of metaethical theory according to which one can explain reasons or normativity in terms of what is constitutive of agency. Any constitutivist theory makes three basic claims: (1) that some feature is constitutive of agency, (2) that one can explain reasons or normativity in its terms, and (3) that doing so has plausible first-order normative implications. I consider the paradigmatic constitutivist theories of Christine Korsgaard and J. David Velleman and the more recent variant developed by Michael Smith, and I argue that each fails adequately to justify at least two of the three basic constitutivist claims. I then argue that a constitutivist strategy can nevertheless be adapted to explain the necessary connection between normative judgment and motivation. More specifically, I argue that practical deliberation has two constitutive features. First, it aims at proceeding in a rational way from premises to conclusions. Second, it has an internal connection with motivation: barring weakness of will, people are motivated to act in accordance with their deliberative conclusions. Because a person's normative beliefs guide the course of her deliberation, and her deliberation motivates her action, a person will be motivated to act in ways that correspond to her normative beliefs, which her sincere normative judgments express. This account provides a cognitivist explanation for a phenomenon often taken to be the most important evidence for non-cognitivism or expressivism.

On a standard philosophical conception, being autonomous is roughly equivalent to having some particular natural capacity. This paper provides argues that this conception is incorrect, or at least incomplete. The first chapter suggests that adopting an alternative conception of autonomy promises to resolve to several objections to the metaethical constitutivism, and so promises to provide highly desirable theory of moral reasons. The second chapter first motivates a broadly Kantian account of autonomous action, and then gives reasons to think that Kant's own development of this theory runs into damaging action-theoretic problems. The way to address these problems, I argue, is to modify Kant's account of autonomy in a way that leaves no room for the standard conception of autonomy to do any work.