What would count as evidence against free will?
I’ve said before that there can be no evidence for or against free will because it is a doctrine about the ontological status of things that don’t happen. A person with free will might very well do precisely the same things as a person without free will — the only difference being that the former could have done otherwise. But what “could have happened” is invisible to us; we can only observe what actually happens. Therefore, a person with free will and a person without free will are empirically indistinguishable. I’ve said before that this means all we can do is assume that we have free will (or not), and that it makes more practical sense to assume that we do. But what it would actually mean, if it were true, is that free will vs. determinism is a fake question, a distinction which makes no difference. If it is indeed true that the one belief is more practically useful than the other, then the two beliefs must be empirically distinguishable, at least in principle.
My argument that there can be no evidence for or against free will could also be used to argue that there can be no evidence for or against the proposition that nature is governed by laws. The sun rose in the east today, but could it have risen in the west instead? We believe that the sun and the earth move in accordance with fixed laws of gravity and that it is impossible for them to do otherwise — but isn’t it also possible that they are free, that they just happen to choose to behave in a uniform way but could just as easily choose otherwise? G. K. Chesterton somewhere discusses the possibility that every single day God freely chooses to tell the sun (or rather the earth) to “do it again.” Have we no empirical grounds for favoring Kepler/Newton/Einstein over Chesterton? After all, we only observe what happens, not what could have happened.
I’ve been conflating absolute proof with mere evidence. There can be no conclusive disproof of Chesterton’s “do it again” hypothesis, but we can and do have evidence against it. Suppose an astronomer predicts precisely where and when the sun will rise tomorrow. If the sun and earth are bound by the laws the astronomer thinks they are bound by, what is the probability of the prediction coming true? 1. If they could behave otherwise, what is the probability of the prediction coming true? Unknown, but necessarily less than 1. You can do the Bayesian math and see that, whatever prior probability we assign to the Chesterton hypothesis, it should be revised downward if the astronomer’s prediction comes true. Therefore, every successful prediction is evidence that things could not have happened otherwise. (How strong that evidence is cannot be calculated, though.)
When it comes to human behavior, things are less straightforward, since no one claims to be able to predict it. Determinists say it is predictable in theory but, due to the fantastic complexity of the human brain, not in practice. Indeterminists say it in not predictable even in principle. The two theories do not therefore make distinct predictions.
However, if someone were to make a detailed and accurate prediction of human behavior, comparable to the predictions of astronomers, that would be evidence for determinism and against free will. The more specific the prediction, the stronger the evidence (though, again, we cannot assess exactly how strong).
Prophecies like those featured in Greek tragedy would be relatively weak evidence against free will. One gets the impression that Oedipus and his parents were free to do many different things, but that some unseen power was seeing to it that, whatever they chose to do, the final result would be the same. (It could be compared to a chess master’s prediction that, whatever his novice opponent may choose to do, the master will still win in the end.) Much stronger evidence can be found in the Gospels, where Jesus says to Peter, “Before the cock crow twice, thou shalt deny me thrice” — and then, despite knowing the prophecy and being unwilling to do any such thing, Peter proceeds to fulfill it. If the story is true, it offers strong evidence against free will, since specific details of Peter’s behavior (which Peter, apparently erroneously, believed were under his control) were successfully predicted. No wonder the poor devil “went out and wept bitterly” at this revelation that he was, despite appearances, a robot!
What, then, would count as evidence for free will? Well, any failed prediction of human behavior. Granted, that seems like a very strange thing to say. If I am unsuccessful in predicting the behavior of a given system, that doesn’t mean the system isn’t governed by rules — it just means it isn’t governed by the particular rules I thought it was governed by. But, logically, that is evidence (very weak evidence) that it isn’t rule-governed at all — just as the fact that I wasn’t born on February 12 is evidence that I wasn’t born in February at all.
The problem is that none of this evidence is at all quantifiable, so it remains impossible to say whether, on balance, there is more evidence for free will or against it. In the end, then, there’s still nothing to do but to make an assumption one way or the other.