Wednesday, July 23, 2008

Kevin Timpe's critique of Frankfurt libertarianism

Timpe is a critic of use of Frankfurt counterexamples both by libertarians and by compatibilists. Of special interest is the fact that he points out, that in the Frankfurt cases there has to be a forking path. One path in which the controller does nothing (the actual case) and one in which the controller stops a process from going further. John Martin Fischer says that this is just a "flicker" of freedom and doesn't play a sufficient role in grounding moral responsibility. Why?


Mike Almeida said...

Of special interest is the fact that he points out, that in the Frankfurt cases there has to be a forking path

I don't see why. Have the person charged with causing you to change your mind (in the event that you decide to save the drowning swimmer) be a perfect predictor. An instant before the time you actually choose to save the person, he has predicted it, and sets off the device. There's not even a flicker of freedom in that case. You don't even get the chance to make the choice.

Victor Reppert said...

This example confuses me. You have a perfect predictor (which if you are a good Haskerite that takes out freedom right then and there), who realizes that you would do otherwise if he doesn't intervene, so he does intervene. Not free, not responsible, no problem.

I take it the interesting case is this there is an infallible predictor who would have set of the device if he had made an opposite prediction of your action.

But even here, one needs a little explanation as to why one should say "free and responsible" in such a case. Maybe we need a good middle knowledge guy. I'm not exactly a card-carrying open theist, but I must confess I don't have a good account reconciling freedom and foreknowledge.

If there's no forking path, what makes them free and responsible. Without the forking path, the example loses intuitive power.

Kevin Timpe said...


It sounds like you are describing the alternative sequence of a Frankfurt-case. If the predictor sets off the devise, then the person isn't morally responsible.

Consider then the actual sequence. If the predictor (Black) thinks that the agent (Jones) is going to do what Black wants him to, then Black does nothing. Think a bit more about Black here. There are 4 main options.

(1) Black is a perfect predictor, but only by chance. In this case, he could be wrong (even though he hasn't yet been, and might not be here). If he thinks that Jones will do what he wants him to and does nothing, then it is still open for Jones to do otherwise.

(2) Make Black an essentially infallible predictor via middle knowledge. Here, I think that there would (perhaps) be a case of no alternative possibilities. But, for other reasons, I think that Molinism is necessarily false, so this option doesn't do much work here (at least for me).

(3) Make Black an essentially infallible predictor via simple foreknowledge. In that case, it seems to me that there is still the ability to do otherwise for reasons I outline in my "Truthmaking and Divine Eternity." (The short answer is that I'm an Ockhamist, and think that in this case Jones would have counter-factual power over Black's belief/prediction).

(4) Black is an infallible predictor on the basis of something other than foreknowledge (either simple or based on middle knowledge). But then we enter the familiar waters of the debates about this mechansim--how is it that whatever the mechanism tracks is sufficient for the Jones' choice in a way that doesn't smuggle in that choice being causally determined?

Kevin Timpe said...


Here is a different way of making the claim that there must be alternative possibilities (of some sort or other). Think of the acutal and counterfactual sequences as states of affairs. In order for them to be different states of affairs, then there has to be some difference involved. Perhaps it's just that the intervener gets involved in the one but not the other; but ever here there are alternative possibilities. As McKenna and Widerker note, ‘strictly speaking, Frankfurt examples do not rule out all alternative possibilities since there do exist, built right into the examples, flickers of freedom.’

Of course, this doesn't mean that the remaining alternative possibilies do secure freedom/moral responsibility. That is why, I think, we're pushed toward the questions about robustness, such as Victor's original question.

(Sorry, Victor, that I haven't done anything to answer your question. I think that Fischer is wrong in terms of what is required for robustness, but I explain his thoughts on this matter in chapter 4 and 6 of my forthcoming book. I'd be happy to share the penultimate draft with you if you'd like.)

Mike Almeida said...

There are several worries here for me. (1) Victor, (and I won't spend time on this one) nothing Hasker has said on foreknowledge undermining freedom is remotely plausible to me. So I will set that aside.

(2) I didn't intend that the perfect predictor actually sets off the device. I just assumed we all know what happens in Franfurt cases and would gather that he never has to set off the device.

(3) Kevin, I don't see why anyone would have counterfactual power over the the predictor's belief in the Ockhamist case and not in the Molinist case. Molinists allow that, despite the Molinist knowledge, you could have done otherwise.

(4) Indeed, in lots of accounts of compatibilism, you have counterfactual control (though I really don't like this term) over the past. Lewis in 'Are We Free to Break Laws?' allows that we have such counterfactual control. Were S to raise his left arm, it would have been the case that the past was different. Though exactly how different, and how it was changed and where, is something that is determined by context. We have zero control over any of that.

Here's my worry. It cannot be counterfactual control that distinguishes these cases, since we plausibly have them in each of the cases above. So what makes the Molinist case any different from the simple forekowledge case wrt forking futures? Both allow abundant backtracking, so that can't be it. What distinguishes either from the compatibilist case? Compatibilists allow for abundant backtracking, too. Why aren't their futures forking?

Finally, you write,

Perhaps it's just that the intervener gets involved in the one but not the other; but ever here there are alternative possibilities

Yes, right, but there are alternative possibilities in the compatibilist case too and, I'm assuming, that doesn't get you freedom in your book.

Kevin Timpe said...


Regarding (2), sorry for the confusion. A failing of the principle of charity on my part.

Regarding (3), you're right. But I think that on the version of Ockhamism I favor, we have a stronger kind of counterfactaul power than in the Molinist case. Let me try to briefly explain, hopefully without taking us too far afield (if I haven't already done that). According to Molinism, I am not the ground or truthmaker for those true CCFs that are about me. I think that I am the ground(er) or truthmaker for those past beliefs that God has about what I will freely do (or the atemporal analogue if you prefer).

Regarding (4), what I want to say here is related to what I said immediately above--but it would take me more time (and thought) to spell out. Perhaps you can see roughly where I am going though.

I agree with you that on any of my (1)-(4) have alternative possibilities built right into them, which is why I think that there has to be a forking path in FSCs (by their very setup).

And regarding your last comment, yes. But they aren't the same kind of alternative possiblities as I think are at issue in the Frankfurt cases (alternative possibilities given the actual past and laws of nature). And I think that such alternatives are necessary (though insufficient) for freedom.

Mike Almeida said...

But they aren't the same kind of alternative possiblities as I think are at issue in the Frankfurt cases (alternative possibilities given the actual past and laws of nature).

Yes, good. It would move the discussion forward in really interesting ways if there were some principled way of distinguishing kinds of alternative possibilities--ways that matter to the freewill/responsibility question. Probably what counts as the relevant sort of alternative will depend on one's guiding intuitions in modal metaphysics, and that will also get us way off topic.