Andrew Bailey

Outstanding Undergraduate Paper:

Andrew Bailey (Biola University), "Some Unsound Arguments for Incompatibilism," with commentary by John Martin Fischer (University of California-Riverside).  Both the paper and the commentary can be found here.

Comments

[Malachai M. Nilsai]

I am a compatibilist, in the sense that I believe free will is the will as controlled by the conscious mind; and yet I recognize that both are the result of metaphysical and deterministic processes which create them. (My method of attacking incompatibilism would just be to accuse them of denying the obvious fact that agency arises out of deterministic conditions, and of conflating the phenomenon of free will with its underlying causes.) In any case, I am sympathetic to the goals of this paper.

Nevertheless, I'm not sure I understand the Fischer-Frankfurt argument presented in the first argument. Green is not responsible for the scientists performing some action, nor is he responsible for the scientists compelling the particular act of saving the child. Thus, Green is not responsible for the scientists' actions in saving the child. But none of this directly has anything to do with Green's moral responsibility to save the child; it just describes the fact that agency is a minimal requirement for him to be accountable.

We may only say that Green is possibly responsible when we suppose he has agency. By the nature of the example, this is robbed for him, and so too is he robbed of responsibility. In other words: let Green` be Green as an agent, while acting under his own volition, and let Green`` be him while acting under the influence of the scientists. Green` may possibly be held to account, and be morally responsible for his actions; Green`` may not, and is not. There is no question of whether or not he is "partly" responsible: Green` is, while Green`` isn't. They are different agents.

[Fritz]

The B* rule that you prove looks like a special case (isn't it?) of the Beta Box rule that you said is vulnerable to counterexample. That seems odd. The "N" in both is the same right? [Np = p and no one is even partly morally responsible for p]. In your discussion of N you give a particular reading of it, but that's open to the advocate of beta box too isn't it?

But perhaps I missed something reading too quickly...

[Andrew]

Fritz, you're right. As was pointed out to me at a recent conference, a proof (shorter, actually!) similar to the one I ran for B* can be used to derive Beta Box.

So given MR, Beta Box is a provably valid transfer principle. But if my strategy (employing Frankfurt-style cases to undermine MR) works, the validity of B* and Beta Box doesn't do anything for the incompatibilist, since the principles aren't really about moral (non) responsibility after all.

[Kevin Timpe]

Andrew,
In response to Fritz, you say that (1) "Beta Box is a provably valid transfer principle." However, on page 6 of the article, you write that (2) "Frankfurt-style Case 2 is a counterexample to Beta Box." Would you agree that (1) commits you to the denial of (2)?

Second, why is it the case that Beta Box isn't "really about moral (non) responsibility," particularly in light of Fischer's comment about your analysis?

[Andrew]

Both principles are provably valid given the analysis of moral (non) responsibility suggested by MR. And let me be clear: if the "derive a transfer principle from an analysis of moral responsibility" strategy is sound, then there can be no counter-examples to either principle. But I don't take that strategy to be sound, since the analysis relied upon (MR) doesn't capture the concept of moral responsibility; Frankfurt-style cases as applied to MR show this. Fischer's comments only nail the coffin further.

Defenders of Beta Box face a dilemma. If the 'moral responsibility' refered to in Beta Box is left unanalyzed, it's about moral responsibility--fair enough. But the principle's validity is subject to Frankfurt-style attack. If 'moral responsibility' is analyzed in a way at all similar to MR, then the principle is provably valid, but the 'moral responsibility' it governs the transfer of won't be what most people refer to when they talk about moral responsibility. So it's useless in arguments for incompatibilism with respect to everyday moral responsibility and determinism.

Since it looks like my response to Fischer's comments didn't get posted, I will paste them below:

John, thanks for your gracious comments. I agree with everything you say and have two replies.

First, I agree that MR is not a good account of moral (non)responsibility. But an account is not what is needed—only an analysis (a statement of the conditions for application of the concept). It’s needed, at least, to derive a transfer principle like B that has some connection to moral responsibility.

Second, I know of no way to repair MR to deal with the type of case you give (though perhaps a clever incompatibilist reader may be of help at Chisholming away your counterexample). And this says something about the strategy I discuss in section three of the paper. If the incompatibilist is to get a principle like B off the ground by the means I have suggested, a plausible route is to analyze non-responsibility in the terms of modal and counterfactual logic. But throwing around a few boxes ands diamonds (and predicates like O and P) doesn’t seem sufficient to the task. The logic of moral (non)responsibility is more complicated than that. Huemer makes a similar point in his 2000 Phil Review paper—showing conclusively, I think, that ‘free will’ just can’t be thought of in terms of access to possible worlds.

And even if this strategy (deriving a provably valid transfer principle from an analysis of (non)responsibility) could bypass the concerns you raise, Frankfurt-style counterexamples will still lurk in the shadows.

[Malachai M. Nilsai]

Andrew,

Again, the import of the Frankfurt counterexamples needs to be shown to have the power you think they do. I've given one explanation above, involving the bifurcation of agency, which seems to call into question just how effective the Frankfurt-Fischer examples are. Perhaps I'm mistaken, but that will require demonstration.

[Andrew]

Yentz,

I've reflected on your objection, and to the extent that I understand it, I may be sympathetic. In the typical Frankfurt-style case, the actual sequence involves genuine moral responsibility, while the alternate sequence (where the counterfactual intervener intervenes) does not--the hapless subject is surely not responsible for being *made* to do something. That's why I'm working on a Frankfurt-style case that involves genuine moral responsibility in *both* sequences. If this can be pulled off, I think I can undermine some of the central intuitions of those who object to the employment of Frankfurt-style cases in this way.

But that is a paper and a project for another day. =)

[Terrance Tomkow]

I am puzzled. If I am partially responsible for the truth of q (e.g. the child is saved) then surely I am at least partially responsible for the truth of any disjunct" ~p OR q" for any p (especially if p is true!). From which it follows that I am partially responsible for the truth of the *material* conditional "if p then q" (e.g. if the scientists are ready to intervene then child is saved). What am I missing?

[Andrew]

Terrance, if I'm reading you correctly, your line of response assumes what I call the True Conditional Non-Responsibility Principle (I hope the symbols come out right on the blog):

TCNP: (p & (p ⊃ q) & ~Nq) ⊃ ~N(p ⊃ q).

That is, given a true antecedent of a true (material) conditional, someone is at least partly morally responsible for the conditional’s being true if someone is at least partly morally responsible for the truth of the conditional’s consequent. Equivalently (given a true antecedent and true conditional), N(p ⊃ q)) ⊃ Nq.

But TCNP is invalid. My argument toward this end follows one independently offered by Stump and Fischer (2000). Recall the Green case, and let r be the conjunction of ‘the actual laws of nature obtain’ and ‘the child is saved.’ Given that q (‘the child is saved’) is a conjunct of r, there is an entailment relation between them:

s. NECC(r ⊃ q)

Now if anything is an uncontroversial rule in the logic of the N operator, it is this:

Α: NECC(p) ⊃ Np

By Rule A, we may infer from s that:

t. N(r ⊃ q)

But given TCNP, we must deny t, because Green is at least partly morally responsible for q. If we affirm Rule A (and it seems to me that the incompatibilist must), we must conclude that either s is false (as absurd a conclusion as any), or that TCNP is invalid. I choose the latter.

[Malachai M. Nilsai]

Andrew,

I suppose my intuition is to insist on what has been called the "control condition" in the paper by Julia Driver, which is just the idea that a person can't be accountable for what they have no control over. The implication for this case would be that Green's actual action is something he's responsible for, while the counterfactual would not be. In the former case, Green himself would be the agent we would rightly praise, while in the latter, we could not legitimately praise the agent Green, but rather we would have to praise the agent referred to as The Scientists. So I suppose it's very difficult for me to see how the agent Green can be responsible for the child in both cases. I would say that he is responsible in the real case, but absolved of responsibility in the counterfactual case. If we were to postulate otherwise, then it would seem to be a challenge to the control condition, which is not (I don't think) prudent.

After giving your paper another reading, I have a few other comments I'd like to offer; possible worries, not for you, but for van Inwagen.

Lurking behind the intuition that gives rise to (A) is the general idea that a necessary truth is not something that a person may be responsible for; but this is itself suspect. There are certain necessary truths which arise purely out of human stipulation; i.e., that all bachelors are unmarried men, because "bachelor" means "unmarried man". The former necessarily follows from the latter, but we have full control over whether or not we accept the latter or not.

But more specifically, from at least one compatibilist's perspective, (A) is wrong right from the getgo. For agency is understood by some compatibilists as an emergent phenomenon, something that arises from deterministic conditions. In which case, there would be some deterministic states of affairs (i.e., the agent's own actions) which the agent would at least partially be responsible for. Thus, (A) loses all its power, because a person is responsible for a state of affairs despite the fact that it arises by necessity. Thus, (3) cannot arise, and the rest of the argument fails.

[Terrance Tomkow]

The principle I was asserting that if if one is responsible for the truth of the consequent of a material conditional one is at least partially responsible for the truth of the material conditional. (I see no reason to restrict this to conditionals with true antecedents)

Your reply amounts to the argument that some material conditionals are necessarily true and we can't be even partially responsible for necessary truths.

But why on earth should I accept the principle that people can't be responsible for the truth of necessary truths? Consider the material conditional.

(C) If you respond to this comment then you will respond to this comment.

Now you can make this true by responding to this comment or by not responding to this comment. Either way, your behavior will have made this conditional true and I see nothing against saying that you are responsible for it's truth one way or another. Of course you *must* do one or the other and you are not responsible for the (modal) fact that you *must* do one or the other. But it does not in the least follow that you are not responsible for doing one or the other and it is your doing one (or the other) that will make C true.

[Kevin Timpe]

Terrance,

I see how, in the example posted at 3:12, I might be responsible for the truth of the anecedent, and I might be responsible for the truth of the consequent, but I don't see how I'm responsible for the truth of the conditional. After all, the following structurall parallel conditional is also true:

(C') If my twin sister responds to this comment then she will respond to this comment.

But I don't have a twin sister, so she can't be responsible for the truth of the conditional. How does Andrew's doing 'one or the other make' (C) true, when (C') is true despite my (non-existent) twin sister not doing 'one or the other'?

[Kevin Timpe]

Andrew,

Can you say a bit more about "a Frankfurt-style case that involves genuine moral responsibility in *both* sequences"? I realize that this is a 'project for another day', but I'd appreciate a few words on this if you have the chance.

[Terrance Tomkow]

In response to Timpe:

The principle I advanced was conditional. If Jones is responsible for p then he is partially responsible for p OR q. That is consistent with saying that Timpe's notional sister is not responsible for anything.

But Timpe's real problems are with Bailey, not me. Timpe says that even where Jones is responsible for p and responsible for q, Timpe doesn't see how that makes Jones responsible for "if p then q". But if *that* wouldn't do it what *would* make someone responsible for a material conditional and, if *nothing* would, then is it always true that N( if p then q) for all p and q? Bailey owes us an answer, I think.

[Andrew]

Terrance,

Not sure what to make of cases like (C); I'll have to sleep on it.

You suggest the rejection of Rule A. This is a strategy I'm open to, though I've never known an incompatibilist (or *anyone* in the literature, actually) to make this move. But doesn't it seem odd, saving Rule B at the expense of Rule A? Remember that without Rule A, van Inwagen's argument is invalid. And Rule A seems to encapsulate about as central an incompatibilist intuition as any-- if Rule A (and the intuition behind it) goes, it's hard to see how the incompatibilist project is ever going to get off the ground.

On responsibility for conditionals: I have no systematic account of what it is to be partly morally responsible for the truth of a material conditional. But I can say this much with confidence: the logic of moral (non)responsibility is a lot more complicated than is suggested by the standard incompatibilist toolbox of arguments and transfer principle. Additionally, it seems to me that the conditionals we generally ascribe moral responsibility for are *counterfactuals*, so looking to material conditionals to do much work for us in these debates is misguided.

Kevin,

I don't have much else to say at the moment on the Frankfurt-style cases that involve genuine moral responsibility in both sequences. But I'll let you know when I'm finished with a paper advancing this strategy. =)

[Terrance Tomkow]

On a different point, let me say something in defense of Bailey against Fisher's comments. I don't understand Fisher's objection but it is clear that it involves a non-standard reading of counterfactuals about chancy (or indeterministic) events

Fisher wants to say that where a indeterministic coin toss results in "Heads" it is nevertheless true that

(F) if S had done x the coin might have come up Tails.

for any "non-related" x (I take it he means "causally related").

Fisher says: " This is because it is simply true that the coin might have come up tails, and thus it is true that for any (unrelated) action S could perform, the counterfactual, "If S were to perform it, the coin might have come up tails" would be true.

But this is far from obvious. Suppose x is "Betting on tails". If F is true then it would follow.

(F') "If S had bet on tails, S might have won".

But this we would ordinarily take as false: If S had bet tails, S would have lost because, in fact, it came up heads.

Anyone interested in this issue might take a look at Jonathan Bennett's nice discussion in "A Philosophical Guide to Conditionals".