From Scott A. Davison's "Craig on the Grounding Objection to Middle Knowledge" in the recent issue of Faith and Philosophy:
"In fact, we are probably never in a position to say with much confidence that any particular action is free (in the Libertarian sense employed by Craig and the Molinists). This is because for all we know, actions that seem to be free might very well have hidden causes."
Davison attributes this point, which he calls an old one, to Dennett among others. It simply will not do, however. I recognized the pattern of the argument at once; one comes across it in external world skepticism, and many other skeptical positions:
We cannot be confident that things that seem X are X because, for all we know, there very well might be something hidden that makes things seem X even though they are really not X.
You can plug anything into X; nothing in the form of this argument makes that impossible. Which means, if we accept this argument in one case, we have to accept it in other cases where we could say "for all we know" there might be these 'hidden causes' that are severing appearance and reality. And they are legion: other minds, external world, any scientific conclusion in any science (this is why Newton famously insisted that we should not make hypotheses - by 'hypothesis' he meant precisely this sort of 'for all we know' postulation of possible hidden causes in order to pre-empt our conclusions from the evidence), the appearance of evidentness in mathematics (because 'for all we know' there might be some hidden cause making it only seem to be the case that 2+2=4), you name it.
One finds this sort of argument, which is a lousy one, used by compatibilists quite a bit. In putting forward these arguments, they are being selectively obscurantist in ways they would be horrified to find in other places. For instance, C. A. Campbell in his Gifford Lectures On Selfhood and Godhood once claimed that we experience the fact that some of our actions are uncaused. Now, I think that this is a claim that is either false or misleading; but the actual argument I have seen compatibilists use against it is precisely a version of the above argument. This, as I said, simply will not do; it is an absurd argument (and ultimately self-defeating, because this argument form can be used against itself). The proper response to it should always be: "Prove to me that these 'hidden causes' actually exist in the relevant cases! Otherwise, you've got no more of an argument here than someone who believes evolution is false because 'for all we know' God just made things seem like evolution is true; or someone who believes that 'I have two hands' is false because 'for all we know' we might live in the Matrix."