Compatibilism and Modality

In a recent article (“Free Will and the Problem of Evil,” Religious Studies 40 (2004): 437-456), James Cain makes what seems to me to be a rather bold claim regarding free will and the truth of causal determinism.  Cain first distinguishes between three kinds of compatibility: metaphysical, conceptual and epistemic.  He caches these out in terms of statements.  Statements S1 and S2 are metaphysically compatible if and only if there is a possible world where both S1 and S2 are true.  Statements S1 and S2 are conceptually compatible if and only if it is conceptually possible that both S1 and S2 are true; that is, the negation of the conjunction of S1 and S2 is not an analytic truth.  Finally, statements S1 and S2 are epistemically compatible for a person p at time t only if nothing in p’s epistemic body of evidence at t rules out its being the case that both S1 and S2 are true (for more on these, see pages 438f).

Now, for the claim.  “The standard arguments for compatibilism, at most, support a conceptual or epistemic compatibilism” (443) and that the “typical arguments for compatibilism generally do not provide support for metaphysical compatibilism” (444).  By ‘typical arguments’, Cain has in mind three general strategies for arguing for compatibilism:  paradigm case arguments, conceptual analysis arguments, and arguments employing Frankfurt cases (he specifically mentions Fischer and Ravizza’s arguments in this regard).  It is this third sort of argument for compatibilism that I’m most concerned with.

Now, admittedly semi-compatibilism, versus traditional full-blown compatibilism, makes things a little dicey here (In footnote 24, Cain writes that “I do not claim that Fischer and Ravizza argue for metaphysical compatibilism”).  So let’s just stick with compatibilism. 

Cain thinks that FSCs give “little reason” (448) that there is a possible world where an action is both determined and free.  To make things a little easier, he assumes that there is some correlation between brain states and volitions/choices, as Stump and others do (though he doesn’t think this assumption is crucial for his more general point).  According to a typical understanding of an FSC, the agent in question actually makes some choice even though she couldn’t have done otherwise; that is, had the agent not made the choice ‘on her own’ as it were, the implanted device (or whatever) would have brought about that very same choice via directly stimulating the brain.  This is where Cain takes issue with FSCs.  He writes:  “The answer to the question of whether a Frankfurt device would cause there to be a choice if it caused the atoms in the brain to move just as they might move if a choice was made is as follows: it depends on what sort of causal powers (if any) are essentially active in choosing, and on whether the device brings it about that those powers are properly exercised” (449).  He considers three possibilities:  (1)  choosing does not essentially involve the exercise of any particular causal powers, (2) choosing essentially involves the exercise of certain causal powers, and those causal powers can be exercised by the manipulation device, and (3) the causal powers involved in choosing may be such that they “issued in any one of two or more mutually exclusive alternatives, and which of those alternatives comes about is not fully determined by factors outside the exercise of that power” (449).  Now, if either (1) or (2) were true, then FSCs will have shown free will and causal determinism to be metaphysically compatible.  However, “under the assumptions of case 3, the story told by the Frankfurt example may not be metaphysically possible” (450).  And until it can be shown that FSCs aren't examples of (3), "there is no reason to accept Frankfurt examples as providing a way to show that one's making a given choice is metaphysically compatible with its being impossible for one not to make that choice" (450).

In general, I'm not usually one to support compatibilism.  But I'm struck both by Cain's argument against metaphysical compatibilism, as well as the dialectic at work throughout the article.  Has anyone else read it?  Thoughts?

Comments

[Terrance Tomkow]

Could I here raise a prior question about Frankfurt cases?

It seems to me that Kadri Vihvelin (USC) demonstrated some time ago (CJP 2000) that Frankfurt's arguments turn on a modal fallacy. No matter how you tell these stories, Frankfurt Style Cases simply fail to describe agents who "cannot do otherwise".

If that's so then FSC cases have nothing to tell us about free will and the whole of the debate about FSC has been a snare and a delusion.

To my knowledge no one has answered Vihvelin's arguments. So my question is: does anyone have an answer to Vihvelin or does the Frankfurt literature just keep stumbling forward out of inertia?

You can find Vihvelin's paper at her web site: http://www.uscphilosophy.org/faculty/facultydetail.cfm?Faculty_ID=12

[John Fischer]

Ok, Terry, I really should address that paper. I read it with great interest, and, as with everything Kadri writes, I found it both insightful and challenging. I don't recall whether I had a specific way of responding, or a general sense of where to respond; I'll try to take up this challenge soon.

Meanwhile, allow me to express the regret that my spouse is not as loyal and supportive...

[jon kvanvig]

John, your sad emotional state makes me want to help...

So, here's what I think is the right response to the Vihvelin paper. Using her coin example and the reductio on the claim that the coin will come up tails, she's right that you can't get that this is impossible by adding her premise (2). I'm not happy with her explanation, so I'll give my own. The reason is that if the second premise is contingent, then all that follows from deducing a contradiction between the assumption and premise (2) is that the coin won't come up heads, not that it can't.

That gives us a recipe for fixing the reductio. Here's one way. Just make the Frankfurt-intervener's existence, intentions, and powers more modally extensive than that of the coin. I.e., specify that the counterfactual intervener exists in at least all the worlds that the coin does, and has the same intentions and ability to carry them out in at least all the worlds in which the coin exists. The objective probability of the coin coming up heads, given only intrinsic features and laws, will still be .5, but the second premise in the reductio will now be a necessary truth if true at all (since it is a modal claim, to the effect that in any world in which the coin exists, the intervener exists and carries out the plan in question). Then the reductio shows, not just that the coin will come up heads, but that it has to.

I put this point into the coin-tossing example, but it's pretty easy to see why it needs to go this way in cases of action.

But I still think that the key question here is whether the beginning of a trying is itself an action...

[Terrance Tomkow]

As I read him, Jon Kvanvig is agreeing that Frankfurt interveners do not deprive agents of the ability to do otherwise. He proposes a different style of story in which "Kvanig Interveners" somehow make it metaphysically necessary that agents do what they do. That might be an interesting discussion, but it has not much to do with Frankfurt. Is everyone else prepared to so readily give up on Frankfurt style cases?

[jon kvanvig]

I'm not sure why you say that my view "has not much to do with Frankfurt." It does. Here's how. Suppose we determine what kind of modal strength one wishes for the claim that one can't do otherwise (personal impossibility, counterfactual impossibility, nomological impossibility, logical impossibility, etc.). Frankfurt's examples did not specify any particular kind of modal strength involved in the claim that one can't do otherwise. That allows a Vihvelin-style response that the argument fails: just read the necessity required in the conclusion of the reductio to exceed that of the Frankfurt-claim about the example (her premise (2)). One can block that response by constructing a case where the second premise has the strongest kind of modal strength, since then it will be impossible to require the conclusion to have a more demanding kind of modal strength than the second premise.

[Terrance Tomkow]

Frankfurt told a story in which someone just like us, cannot do other than he does. But, Frankfurt argued, this person is still responsible for what he does (because he's just like us).   Vihvelin shows that you can only conclude that the agent in a Frankfurt story can't do otherwise if you commit the fatalist fallacy.  You say. "No big deal. Let's just tell a story in which it is metaphysically impossible for someone to do otherwise."   That would certainly get you where the fatalist fallacy is supposed to take you but:
a) while it is true that Frankfurt is cagey about the relevant sense of "could have done otherwise" no one could possibly have taken is story as describing an individual for whom doing otherwise was metaphysically impossible.  His was a contingent intervener.
b)  You haven't told the story yet. It seems to me  the only way such a story can be told is by describing an agent who is very different from us. He will have to have a different metaphysical essence from ordinary folk. Would we want to say that someone for whom doing otherwise is metaphysically impossible was responsible nevertheless?  I dunno. Tell your story.
c)  Frankfurt stories are typically taken as showing something about the compatibility of responsibility and nomological determinism. Will a Kvanvig story convince us of the compatibility of freewill and logical/metaphysical determinism?  If so it will be a very interesting story. It will certainly make look foolish all those people who have spent careers worrying about the compatibility of responsibility with mere nomological determinism.  I look forward to hearing it.