Suppose I act in a way that contributes to Bob's having a vicious character that deterministically produces wrong actions under a wide variety of circumstances. Am I responsible for these wrong actions of Bob's? Maybe and maybe not. I might have completely inculpably acted in such a way as produced this character in Bob. Perhaps my action caused Bob to undergo some minor temptation that he simply did not withstand. Perhaps I was brainwashed into brainwashing Bob. On the other hand, if with full responsibility I produced Bob's vicious character, and produced it so that it would deterministically produce such wrong actions, then I am fully responsible for those wrong actions of Bob's. The following seems right:
- x is responsible for the actions determined by y's character precisely to the extent that x is responsible for y's having a character that determines such actions.
Now, I will apply principle (1) in the special case where x=y. You might wonder if (1) is applicable in that case. But consider cases where I act so to induce a vicious character in some individual under some description and it turns out that that individual is me. For instance, I pay a brainwasher to kidnap a random person and brainwash that person into being a bank robber, and I turn out to be the random person. In those cases, (1) seems exactly right, too. It is plausible that (1) holds in general. (I think I am thinking about N-responsibility here.)
In the special case where x=y, we get the claim that one is responsible for the actions determined by one's character precisely to the extent that one is responsible for having such a character. And now add this:
- x is responsible for a state of affairs S only if S depends (causally or constitutively or by a chain of causal and/or constitutive relations) on or is identical with one or more of x's choices or actions that x is responsible for.
Now, in the case of someone all of whose actions are determined by her character, given (1) (in the special case x=y) and (2), if the individual is responsible for any action, we generate an infinite regress, as in Galen Strawson's argument against responsibility.
I think the compatibilist has to either deny (2), and insist that we can be responsible for a character that does not depend on any responsible choices or actions, or else has to distinguish in (1) between the case where x is non-identical with y and normal cases where x is identical with y (cases like the one where I hire the random brainwasher being abnormal).
As a libertarian, I also think the above arguments conflate derivative and non-derivative responsibility, but I do not think the compatibilist can really make much use of that distinction.