This post is part of my series on free will.
One of the great things about Ekstrom's Free Will is the way she anticipates objections. There were many times as I was reading the book that I started to think, "But what about...?" and almost immediately found her posing the same question and answering it. At the end of Chapter 4, she gives a whole list of potential objections to her approach. She doesn't pitch herself softballs, either - she does a great job of putting the opposing case as strongly as possible. For the most part, I think she succeeds in answering these objections.
The most difficult issue she faces is the problem of control. The indeterminism in a libertarian account has to occur somewhere - but where should it go? If there is indeterminism in the process of deliberation, so that it is a matter of chance whether some consideration occurs to the deliberator, then the outcome of the deliberation seems to be a matter of pure luck, rather than something that is under the deliberator's control (p. 121). Likewise, if there is randomness after the deliberation is complete, then we have the same problem. So the only logical place for indeterminacy to occur is during the deliberation process itself. This all seems correct to me.
But then, given the same mental state before the deliberation begins and the same external inputs, the deliberator could come to a different conclusion. This, of course, is what the libertarian wants: the possibility of different outcomes. But if the outcome is not determined by the state of the deliberator or the external inputs, then it seems to be still out of the deliberator's control. Ekstrom sees this, and effectively punts: you have to have indeterminism somewhere, she says, or revert back to determinism - which she has already rejected.
Ekstrom makes no attempt to explain why there should be indeterminism at this point, apart from a brief comment that it might arise via amplification of quantum randomness (p. 124, citing Robert Kane). She just insists that we must have indeterminism to have free will, and it must occur in this way if it is to make any sense.
Kane's approach has the opposite problem. He locates the indeterminism in quantum events occuring in the brain, so we know why (in the reductionist sense) the indeterminism is there. But this puts it too early in the causal chain - random quantum events in the brain are outside the control of the agent, and so their outcomes are purely a matter of luck. (Dennett argues this point against Kane in Freedom Evolves.)
So it seems that indeterministic accounts of free will have a serious problem: We can get indeterminism from quantum physics, but we can't get it where we need it for free will.