Abstract

After laying out five criteria for a successful libertarian account of free will in chapter 1, in chapter 2 I examine the four best recent alternatives to traditional agent causation theories, by Carl Ginet, Stewart Goetz, Randolph Clarke, and Robert Kane. I find that each of these proposals clearly fails on its own terms. In addition, I highlight other problems they face. First, all are theories of free action done for a reason, and not free action simpliciter. It seems best to account for the latter first. Second, the proposals by Clarke and Kane depend crucially on the notion that events of an agent's having of reasons "probabilistically cause" her free actions. I develop an argument that there is the notion of probabilistic causation is at least as problematic as agent causation, so long as we are conceiving of causation in a realistic, non-reductive way. Third, the proposals of Clarke and Kane inherit problems plaguing all causal theories of action. The situation warrants a careful reconsideration of the prospects of agent causation theories. In chapters 3 and 4 I give a critical account of the theories of Richard Taylor and Roderick Chisholm, who re-popularized agent causation theories in the late fifties and sixties. Ironically, both later abandoned agent causation. I lay out the development of and motivations for their theories of agent causation, as well as their stated reasons for abandoning them. In chapter 5, I lay out my theory of agent causation in detail, which owes more to Thomas Reid and William Rowe than to Taylor or Chisholm. I argue that this account succeeds where the recent theories of Timothy O'Connor and William Rowe do not. In the sixth chapter I spell out the different ways in which an event can be controlled by a person, address problems of action individuation, and show why the account meets the five desiderata of chapter 1