For what it's worth, I think the upshot of my recent posts on regress arguments and free will is this: Whether compatibilism is true or not depends largely on whether
- It is possible that x is to some extent responsible for a deterministically caused state of affairs A even though x is in no way responsible for any of A's causes?
Are the answers to (1) and (2) (which will be given at the bottom) positive? I think not. Here is something that is more a statement of intuitions rather than an argument.
- If (1) is true, then it is possible for a person to be intentionally produced fully formed, and for its character, beliefs, desires and external circumstances to immediately causally determine a choice that the person is at least somewhat responsible for.
- But the consequent of (3) is false.
- Hence (1) is false.
Here is my intuition for thinking (3) to be true. If a choice is immediately causally determined by character, beliefs, desires and external circumstances that the agent (if one can even call her that) is in no way responsible for, then the only thing that matters for determining whether the agent is responsible for the choice is the extent, if any, to which the agent was responsible for this character, beliefs, desires and external circumstances. Nothing else about their causal history matters. The responsibility-relevant thing is the extent to which the agent is responsible for them.
My reason for thinking (4) to be true is even less developed and philosophical. It's one of those "arguments" by restatement of what is to be proved. It just seems right to say that if I choose the character, beliefs, desires and external circumstances so as to ensure that the person I produce in that state is thereby immediately caused to choose A, then that person is not free in that choice. (I should note, though, that at least one person who is very friendly to compatibilism—Mele—is committed to (4). So my intuition in (4) does not beg the question.)
Finally, since I think the real issue is about sources, not about determinism or even deterministic causation, I think the really interesting question is not so much whether (1) is true, but whether
- It is possible that x is to some extent responsible for a state of affairs A even though x is in no way responsible for any of A's causes?
I know, also, that Heath has a way of separating the responsibility question from (1) and (2) by vagueness about responsibility. I think that's mistaken, because more sophisticated versions of the 3-5 argument can be used to argue that responsibility does not increase in deterministic transactions. But that's rather more controversial. Anyway, I don't like non-epistemic vagueness when something important, like responsibility, is at stake.