Suppose that in present circumstances C, which includes both external and internal (e.g., character, mental state, etc.) circumstances I decide to take out the trash. Compatibilists (and maybe libertarians as well) have to say that the causal history of how I came to be in C is relevant to determining whether I was responsible for deciding to take out the trash. On a fairly standard view, I am responsible for the decision only if I came to C via a process that I "owned" in some important sense—a process that involved the normal self-directed formation of character.
But why is the causal history of C at all relevant to grounding my responsibility, or lack thereof, for the decision? I want to suggest a very simple answer: The causal history of C partially grounds facts about what aspect of C I am responsible for, and facts about what aspect of C I am responsible for very naturally at least partially ground facts about how far I am responsible for the decision I make in C. The causal history of C is relevant precisely because it helps determine how far I am responsible for relevant features of C, such as my character, mental state, etc.
If this is right, then the following supervenience principle is very plausible:
- If x and y make the same decision in the same way in the same circumstances, and are responsible for these circumstances in exactly the same way (degree, etc.), then x and y are responsible for their decision in exactly the same way (degree, etc.).
But while (1) is very plausible, it has a very controversial consequence:
- If x and y make the same decision in the same way in the same circumstances, and neither is in any way responsible for the circumstances, and y is in no way responsible for the decision, then x is in no way responsible for the decision.
- I can't be responsible for anything at all without being or having been responsible for some decision.
Suppose my first responsible decision is made in circumstances at t0. Imagine a swamp being that comes into existence, out of the random confluence of swamp matter, at t0 in exactly the same internal and external state as I had in t0 (so an environment might need to come into existence from the swamp). If determinism is true, that swamp being is not responsible for the decision he makes. This is generally acknowledged by compatibilists, since compatibilists insist that an appropriate history is needed for freedom given determinism (without determinism, it is plausible that no history is needed; if x comes into existence at t0, with a sufficient understanding of the moral features of options A and B, and chooses between A and B in the way libertarians say that choices are made, then x may well be free). But then by (2) neither was I responsible for the decision in the same circumstances, since the swamp being and I are alike in not being responsible for any aspect of the circumstances at t0. And so I wasn't responsible for my first responsible choice, which is a self-contradiction. So determinism needs to be rejected.
I think (3) is a weak point in the argument. But the reliance on (3) can be eliminated when we strengthen (1) to apply to events other than decisions.