A third set of issues concerns the ways in which a better understanding of love and caring might contribute to our understanding of free or autonomous action. Even when an action is motivated by the agent’s own desires, it may represent a failure of autonomy—as when a drug addict is motivated to act, against her better judgment, by addiction-based desires (Frankfurt 1971). Some have argued that we can understand autonomous actions as those which are motivated, not by just any desire, but by a desire whose motivational efficacy is endorsed by a further, “second-order desire” (Frankfurt 1971, 1999). Others have argued that we can understand autonomous actions as those that are motivated by certain beliefs about what is good or valuable (Watson 1975). Both of these possibilities, however, are subject to powerful objections (Bratman 2007). But if we accept an account of love and caring that steps outside the confines of belief-desire psychology, another possibility emerges: might we understand an autonomous action as one that is motivated by what an agent most deeply cares about (Taylor 1992, Helm 2001, Seidman 2009, Jaworska 2009)? Given that action motivated by addiction is commonly understood as a paradigm of non-autonomous action, and given that addicts systematically ignore what they care about in their addiction-driven decisions, close attention to the extensive empirical literature mapping out the neural mechanisms of addiction promises a fertile ground for testing this possibility; it could help us understand precisely how caring, and therefore autonomy, is undermined in such cases—and, by contrast, how caring functions in healthy cases of autonomous choice.