I'm not sure what you mean by (1), but in (2) you are still assuming that the probabilities will be fixed.  Many commentators have pointed out that one of the problems with Peter van Inwagen's original presentation of the rollback argument is that he assumes the probabilities will converge as the number of rollbacks approaches infinity.  The only reason to suppose this, however, is if you assume that the events in question are probabilistic, and this is the very thing which is at issue.  The response I give (and I believe Greg and miguel are saying this too) is that we should simply say that free will is sui generis, neither mechanically determined nor probabilistic.  If your metaphysics only has room for mechanically deterministic or probabilistic causation, then this is just evidence that you need to widen your categories.

Ouros wrote:

Then, two questions from what you've all said:

1) Can we really say that an agent is free when he do an action necessarily, even purposely, by an infinite good, say? ( I don't see in what other case an agent would "freely" AND necessarily do something.)
I would say that it's because X is an agent that in a case of an infinite good, he will necessarily choose the infinite good. But the fact that X is an agent isn't determinate by the agent himself, obviously. That would be meaningless.

2) Would you say that it's the motivation that fix the possible probabilities? If that's the case, then it seems that the agent isn't free, because they are external to him and fix what he could do.
If not, then how can we know what an agent would do, even without certainty? There would be nothing to observe, to measure.

 
1) If the action is entirely necessary, then I wouldn't call it free. That is to say, if "God created world X" in every possible world, then I wouldn't call it free, at least not in the libertarian sense. Perhaps a compatibilist could make sense of it, but I'm not a compatibilist.

2) You keep trying to frame free action into deterministic acts, even if under a specific probability. The motivation may be partly external to the agent, but that does not imply they force the agent to act; they don't even force the agent with a "0.85 probability" for instance. They don't force the agent at all, because finite goods are not sufficient to move a rational will that will always be simultaneously repelled by the finitude of the good in question. In general we can always say "to hell with chocolate" and refuse to eat it, no matter how desirable we might find it. But it requires a stronger willpower.

All the same, we can still use probabilities and predict people's actions in many different cases. But it's not in the same manner that we predict impersonal events; we might say there is an analogy involved in the use of probabilities. I know my friend as a rational person won't hit me; I may not be able to put an exact number on it, but I know it would be a very implausible scenario. And still this does not translate into either determinism, randomness or standard probabilistic events.

There's no logical incoherence involved. It's just different from impersonal causation; it's libertarian agent causation. We can't make sense of personal actions by modelling them in the same way we study rocks, elemental particles or chemical reactions.

Last edited by Miguel (4/24/2018 5:07 pm)