Accounts of free will purport to tell us what is required if we are to be free agents, individuals who, at least sometimes when we act, act freely. Libertarian accounts, of course, include a requirement of indeterminism of one sort or another somewhere in the processes leading to free actions. But while proponents of such views take determinism to preclude free will, indeterminism is widely held to be no more hospitable. An undetermined action, It is said would be random or arbitrary. It could not be rational or rationally explicable. The agent would lack control over her behavior. At best, indeterminism in the processes leading to our actions would be superfluous, adding nothing of value even if it did not detract from what we want.

This book examines libertarian accounts to see whether any can meet such challenges. It offers what may be called a conceptual assessment of libertarian views: an assessment of their adequacy apart from the questions whether there, in fact, exists, and whether we have evidence that there exists, the phenomenon that they purport to characterize. (Thus, the accounts considered here may be taken not to include any claim that we have free will. They may be understood as claims about what free will is or would be, if it does or did exist.) This type of assessment is one to which philosophers are particularly suited. It can be carried out by thinking and requires no specialized knowledge of natural science.

A conceptually adequate libertarian view must, first, provide a clear, intelligible account of the relevant phenomena. It cannot omit to say anything at all about something that is required for free will, and, of course, what it does say must be comprehensible. Second, an adequate account must require only things that are, in fact, possible, in a broadly logical sense. Third, it must present us with a characterization of something such that, if we have it, we have free will; the satisfaction of all the requirements of the account must suffice for the existence of free will. And finally, its requirement of indeterminism must not be superfluous; the required indeterminism must be of a sort and located such that, were it to exist, it would make a difference, a difference in whether free will does or could exist.

In recent years, it has come to be a matter of some dispute whether moral responsibility requires free will, where the latter is understood as requiring an ability to do otherwise.

I shall not take sides here in these disputes." I treat the thesis that responsibility is incompatible with the truth of determinism as a separate claim, and I call incompatibilism without this further claim "narrow." Narrow incompatibilism holds that free will, understood as indicated above, is incompatible with determinism, but it (at least) allows that responsibility and determinism may be compatible. I call the position that free will and determinism are not compatible but responsibility and determinism are compatible "merely narrow incompatibilism." A semicompatibilist may endorse merely narrow incompatibilism, but she need not, as she may remain uncommitted on the question whether determinism precludes the ability to do otherwise. I call the view that both free will and responsibility are incompatible with determinism "broad incompatibilism."

No noncausal libertarian account is adequate. Acting freely requires that one exercise active control over what one does, and (if free action is at all possible) it must be possible for a free action to be one that is performed for reasons. The exercise of active control, I argued, is a causal phenomenon, and noncausal accounts thus fail to provide an adequate account of it. They likewise fail to provide for acting for reasons.

If merely narrow incompatibilism is correct, then an unadorned, centered event-causal libertarian view provides a conceptually adequate account of free will. Such a view provides adequately for fully rational free action and for the rational explanation — simple, as well as contrastive — of free action. The indeterminism required by such a view does not diminish the active control that is exercised when one acts. Given incompatibilism of this variety, a libertarian account of this type secures both the openness of alternatives and the exercise of active control that are required for free will.

It is not necessary that an event-causal account require such additional features as self-subsuming decisions, a state of wanting more to act on certain reasons than one wants to act on any others, indeterminate, directed efforts of will, or parallel processing. Although some such features might figure in free actions, they are not crucial to the freedom of those actions. An unadorned centered event-causal account can be adequately defended against the charge of diminished control. It needs no further help, and gets none, from these further features.

If, on the other hand, broad incompatibilism is correct, then no event-causal account is adequate. An event-causal libertarian view secures ultimate control, which no compatibilist account provides. But the secured ultimacy is wholly negative: it is just (on a centered view) a matter of the absence of any determining cause of a directly free action. The active control that is exercised on such a view is just the same as that exercised on an event-causal compatibilist account. This sort of libertarian view fails to secure the agent's exercise of any further positive powers to causally influence which of the alternative courses of events that are open will become actual. For this reason, if moral responsibility is precluded by determinism, the freedom required for responsibility is not secured by any event-causal libertarian account.

This type of view nevertheless provides for some things of value that are secured by no view that is compatible with determinism. It provides for the varieties of difference-making and attributability that are precluded by determinism and that appear to be of some value, even if, all things considered, one might reasonably prefer not to have them in certain cases. Similarly, a centered event-causal account can secure the openness of alternatives that Is needed if a certain illusion is to be consistently avoided in deliberation. This, too, is something of value, as to be subject to illusion detracts from one's dignity.

Whatever freedom can be provided by an event-causal account can be secured for actions that are not decisions, as well as for decisions. Such a view can apply to both overt and mental actions. The freedom of a decision and of (later) doing what one has previously decided may come to the same thing. Libertarians who have restricted direct freedom to mental actions, or to certain types of mental actions, such as decisions, have been mistaken in doing so.

Given broad incompatibilism, if all that is required by an integrated agent-causal libertarian account is genuinely possible, then such a view is conceptually adequate. Unlike a traditional agent-causal view, which denies that the agent's having certain reasons causes her free action, an integrated account adequately provides for acting for reasons and for reason-explanation. Further, the requirement of agent causation is not otiose. It provides for the agent's exercising when she acts, in addition to the active control secured by by an event-causal view, a further power to causally influence which of the open alternatives will be made actual. In exercising this further power, the agent is literally an originator of her action, and neither the action nor her initiating the action is causally determined by events. No infinite regress of free actions is required, and an agent-causal view need not require that microlevel laws be superseded when one acts freely.

The sticking point for an agent-causal account is the notion of agent causation - of causation by a substance. Either of two causal realist views may be drawn on to clarify what is being claimed when it is said that a substance causes an event. It does not help in this regard to maintain that agent causation is exercised purposively and at will, though such claims may helpfully distinguish causation by an intelligent, rational substance from other substance causation. Finally, however, there are, on balance, reasons to think that substance causation is impossible. If that is so, then, given broad incompatibilism, it appears that free will is impossible.

It is an important question how we might live with the fact that we lack free will, if in fact we do lack it. Several recent works have addressed this question in an illuminating way. (I think particularly of Honderich 1988: part 3; Pereboom 2001; and Smilansky 2000.) I have made no direct contribution to this project here. However, if the argument of this book is correct, then the importance of this practical question is underscored.

Both types of account require, first, that determinism be false. But more than this, each requires that there be indeterminism of a certain sort (with some events uncaused, or nondeterministically caused) and that this indeterminism be located in specific places (in the occurrence of decisions or other actions, or at certain earlier stages in the deliberative process). What is our evidence with regard to the satisfaction of these requirements?

It is sometimes claimed that our experience when we make decisions and act constitutes evidence that there is indeterminism of the required sort in the required place. We can distinguish two parts of this claim: one, that in deciding and acting, things appear to us to be the way that one or another libertarian account says they are, and two, that this appearance is evidence that things are in fact that way. Some (for example, Mele 1995: 135-37) deny the first part. But even if this first part is correct, the second part seems dubious. If things are to be the way they are said to be by some libertarian account, then the laws of nature — laws of physics, chemistry, and biology — must be a certain way. And it is incredible that how things seem to us in making decisions and acting gives us insight into the laws of nature. Our evidence for the required indeterminism, then, will have to come from the study of nature, from natural science.

The scientific evidence for quantum mechanics is sometimes said to show that determinism is false. Quantum theory is well confirmed. However, there is nothing approaching a consensus on how to interpret it, on what it shows us with respect to how things are in the world. Moreover, there are deterministic as well as indeterministic interpretations of the theory, and in the view of many, the evidence we have does not decisively rule out the former. Perhaps the best that can be said for libertarianism here is that, given the demise of classical mechanics and electromagnetic theory, there is no good evidence that determinism is true. (For further discussion of these issues, see Loewer 1996 and the essays by Bishop and Hodgson in this volume, chs. 4 and 5, respectively.)

The evidence is even less decisive regarding the presence of the kind of indeterminism in exactly the places required by one or another of the libertarian accounts we have considered. Unless there is a complete independence of mental events from physical events, then even for free decisions there has to be indeterminism of a specific sort at specific junctures in certain brain processes. There are some interesting speculations in the works of some libertarians about how this might be so (see Kane 1996a: 128-30 and 137-42, and the sources cited there); but our current understanding of the brain gives us no solid evidence one way or the other on this question. At best, it seems we must remain, for the time-being, agnostic.

If our beliefs on the issue are to be (as I think they should be) guided by the evidence, then we will leave the question open whether we have what is characterized by any libertarian view. We will accept, then, that perhaps we never have more than one course of action open to us at any given time and we never exercise any greater degree of active control than what can be characterized by a compatibilist account. If this is in fact so, then we have never actually possessed something that many of us thought we had, and something that we may reasonably value. Though not the end of the world, neither is that a matter of indifference.