GFP Reading Group: Haji's "The Manipulation Argument"

Two weeks have come and gone since the posting of Haji's article, "The Manipulation Argument".   It's a wonderful piece and I greatly appreciate the opportunity to present some of my own reactions to it. 

In broad strokes, Haji tackles three of the most central issues currently surrounding such arguments:

(A)      The proper interpretation of the Manipulation Argument (i.e. the formal structure shared by all manipulation arguments), and the relation between it and particular manipulation arguments.

(B)       Michael McKenna’s recent distinction between ‘hard-line’ and ‘soft-line’ response strategies, and the soundness of his argument that every ‘credible’ manipulation argument (including Pereboom’s Four-Case Argument) requires a hard-line reply. 

                                 

(C)       The dialectical role of “ultimate origination” principles in the success of manipulation arguments, especially that of Pereboom’s Principle O in the Four-Case Argument.

My comments are a bit long (sorry), but I hope that they will help kick off a great reading group discussion:

Download GFP_HAJI_demetriou.pdf

Comments

[Joe Campbell]

I haven't had a chance to read either paper carefully, but here are a few questions.

(A) What is the structure of the manipulation argument? My guess is close to McKenna's, offered by Kristin:

1. In situations of type A (manipulation cases), no one is morally responsible for his actions.
2. Situations of type A are relevantly indistinguishable from situations of type B (determinism cases).
3. Therefore, no one is morally responsible for his actions is situations of type B (e.g., moral responsibility is incompatible with determinism).

I wonder what others think.

(B) What role does Principle O play in the manipulation argument?

Principle O: If an agent is morally responsible for her deciding to perform an action, then the production of this decision must be something over which the agent has control, and an agent is not morally responsible for the decision if it is produced by a source over which she has no control. (Pereboom, 2001: 4, 43)

I once thought that Principle O was a background assumption, similar to what Ish seems to think. Kristin seems to think that, given the structure of manipulation arguments noted above, Principle O is a conclusion of sorts. This seems to be what both Pereboom and McKenna think, too.

But if the manipulation argument has the structure noted in (A), how can Principle O be the conclusion?

(C) The ways in which manipulation cases are relevantly indistinguishable from determinism cases cannot be because such cases are deterministic. This would be question begging (cf. some of Ish's comments).

It must be something like cases in which the production of the action is (ultimately?) something over which no agent has control. But if compatibilist agent causation accounts (Markosian, Nelson) are possible, how can this conclusion be reached in a non-question begging way? And if compatibilist agent causation accounts are impossible? That's another argument, isn't it?

[Randy Clarke]

Joe, your suggestion about the structure of manipulation arguments seems right to me. And I think Kristin is right about how the argument is supposed to support Principle O.

I don't see that charges of question-begging get us anywhere here. If I reject a premise of the argument, I can say which one I reject, and I can try to say why I think it's wrong.

[Joe Campbell]

Randy,

I wasn't trying to say that the Manipulation Argument is bad because it begs the question. I'm just trying to figure out the structure of the argument. I don't see how it could be an argument for both Principle O and Incompatibilism. Of course, one could get an argument for Incompatibilism given that one has Principle O but that argument would look an awful lot like the Consequence Argument. So part of my wonderment concerns the extent to which the Manipulation Argument offers us more than the Consequence Argument.

One way to think of it all is the Manipulation Argument is an argument for a Beta like principle (e.g., Principle O) and that you need some other principle or set of principles -- about the necessity of the past, etc. -- in order to get to Incompatibilism.

[Kristin Demetriou (Mickelson)]

Hi Joe!

I think what's leading to the confusion is that the 4-CA can be seen in two different ways.

On the first, it’s a simple manipulation argument and follows the type of structure that you've noted above. On this view, as expected, the 4-CA ends with the conclusion that the tested CAS fails because it does not provide a principled difference between the victim of menacing manipulation and the normal person in a deterministic world (so both seem to lack responsibility).

However, in CH. 4 of *Living Without Free Will*, Pereboom goes beyond the simple manipulation argument and offers a diagnosis of its success as well. That is, he considers possible ways that a compatibilist might respond to the argument and then, finding none, offers a diagnosis to explain why no response is available. This is when he introduces Principle O, arguing that his 4-CA illuminates that the poor compatibilists have been committed to an incompatibilist principle of origination all along. It is on the basis of his *diagnosis* of the success of the 4-CA (rather than the 4-CA on its own), that Pereboom offers the strong anti-compatibilist conclusion he is famous for, i.e., that compatibilism is in principle untenable.

So, technically, the 4-CA itself does not end with Principle O as a conclusion, but Pereboom’s overall discussion of the 4-CA ultimately gets him there. It's very natural to use the name '4-CA' to refer to Pereboom's overall argument, but your comments suggest that we should be taking more care to distinguish the basic 4-CA from the larger lesson Pereboom draws from it. (So, for instance, we might want to distinguish between the ‘4-CA’ and the ‘4-CA+Diagnosis’ in our comments).

[Joe Campbell]

Kristin,

Thanks! This is very helpful! I'm not sure that it makes my worries go away but I don't want my obsession with the consequence argument dominate this discussion!

Let me switch to another topic, which I don't think that you have covered. (I'm teaching summer school now, having just finished the spring semester, and I'm still trying to play softball and finish papers, etc. so I have yet to do more than scan the paper. Sorry if I missed something! I'm just trying to get a discussion going.)

What, exactly, is the relevance of determinism to manipulation arguments? I have two reasons for thinking that determinism is a non-issue. First, imagine an imperfect manipulator. He implants a manipulation device but it is imperfect. It works only 90% or perhaps only 20% of the time. Either way, on this one occasion it works. Is the absence of determinism relevant to the judgment that the manipulated person lacks control? I don't see why.

Admittedly I got this thought from reading Mele's Analysis criticism of the 4-case argument, but I don't see why it doesn't apply to his zygote argument, as well. The suggestion, though, is that what is relevant is the causes of one's actions, not determinism or indeterminism.

Second, libertarianism is not without its own troubles. Luck problems abound for libertarian accounts. So what, exactly, is so precious about libertarianism that makes it immune to the manipulation argument? The fact is, no libertarian account has successfully illustrated the key to free human action.

The problem is with free will -- compatibilism is bad because, given determinism, our actions seem to trace back to causes beyond our control; libertarianism is no better since anything short of a sufficient cause leaves open the possibility of chance or luck playing a role, and a similar lack of control over our actions.

I know that Derk and others think that the libertarian agency theory might be of some help but exactly how is that so? Well, one could say "If the action traces back to the agent, then manipulation is impossible." But the real question is, Why is this a better response than the one that a compatibilist might give -- especially given the problem of luck? If the action is the product of a reason-responsive mechanism (with the appropriate causal history), then manipulation is impossible. Similarly, as Markosian and Nelson suggest, why wouldn't a compatibilist agency theory work? Is the key that the action traces back to the agent? If so, then Markosian and Nelson should work. Or is it important that the action not trace back to factors beyond the agent's control? (I know I've gotten sucked back into my original point but I still wonder if the manipulation argument is doing double duty.)

Why does the libertarian agency theorist response sound better, as a response to our worries about manipulation, than any of the compatibilist responses? Given the points above, I don't see it.

Why isn't the manipulation argument an argument for free will skepticism -- given that it is worth considering, for you just need to add the claim that indeterminism won't help -- as opposed to an argument for incompatibilism?

[Note: Perhaps in my last post I should have noted the Direct Argument and van Inwagen's principle (B), instead of the Consequence Argument and his principle (Beta). If we're talking about free will, then the latter; if we're talking about moral responsibility, then the former.]

[Kristin Demetriou (Mickelson)]

Hey Joe,

Sorry my response was slow in coming…maybe now that grading, graduation, and general end-of-semester celebrations are over, more people will have time to jump in.

So, you’ve asked: What is the relevance of determinism to manipulation arguments? Why is libertarianism immune to manipulation arguments? Why isn’t the MA an argument for skepticism?

I think you’re definitely right that manipulation arguments can be mounted against (at least some) libertarian accounts, so I agree that determinism is not an essential aspect of such an argument and that not all libertarian accounts have immunity. So, I think that your version of MA is a plausible account of the most general structure of manipulation arguments.

Where we part ways, I think, is that I believe that it is important to distinguish determinism-based manipulation arguments from indeterminism-based manipulation arguments. It’s been a while since I looked at the response to the 4-CA from Mele that you mentioned, but I remember thinking that little is gained by discovering that (for many people) the same intuitions follow from both deterministic & indeterministic versions of the manipulation. The reason is that it does not follow from the fact that people have the same moral intuitions to both versions that the causal features of the stories don’t matter.

On the contrary, I think that determinism and indeterminism threaten freedom and moral responsibility in different ways—your comments about “the problem is with free will” suggest that you’ve noticed this—and so the challenge set forth by each type of MA will be different. Notably, this means that it might turn out that compatibilists can defuse the threat of manipulation arguments while libertarians cannot, or vice versa. Since, like you, I’ve already given up on the possibility of a viable libertarian account of freedom for other reasons, I guess I decided my time is best spent trying to understand the anti-compatibilist challenge and doing my best to defend compatibilism in light of it. However, your comments are making me realize that I've overlooked some potentially useful strategies because of the way I was approaching this debate.

So, why am I unwilling to see the MA an argument for skepticism? First, I think that it would pretty tough to run a manipulation argument against agent-causation libertarian accounts, so I think the MA has a limited scope in that way. Second, for those of us who feel that we are making forward progress against manipulation arguments, such a concession to MA proponents seems premature. But, as someone who has a dog in this fight, my response probably isn’t surprising…am I being overly optimistic in your view?

[Joe Campbell]

Kristin,

Thanks! I mean to reply to the other discussion, too, but I've been busy. Two quick points here.

First, as I see it, the problem with libertarianism is just that it doesn't help. Indeterminism cannot increase the level of freedom or moral responsibility in a given world. I'm sorry but I still can't find a better way to make this point. But if I'm right, then any reason for denying compatibilism is a reason for denying the free will thesis.

Second, I don't see how agent causation helps. Further, I don't see why agent causation is essentially indeterministic. People keep saying this yet there are compatibilist agent causation accounts in print: Markosian, Nelson, Kant (I would say).

According to the agent causation account, agent causation is a primitive, unanalyzable concept. How a thesis based on an unanalyzable concept -- the thesis of agent causation -- can be deemed incompatible with determinism -- or any other consistent thesis -- is beyond me. Can someone explain this to me? I don't see why I can't take my favorite compatibilist theory -- Kadri Vihvelin's -- and simply add agent causation to it. It is primitive for crying out loud. What is to keep me from merely stipulating it?

And even if one could explain the incompatibilism of the agent causation view and determinism, we seem to have moved on to something else -- maybe the ultimacy argument. I don't see how this incompatibilist thesis is supported merely by considerations voiced in the manipulation argument.

[Al Mele]

Hi Joe and Kristin,

Joe wrote: “What, exactly, is the relevance of determinism to manipulation arguments? I have two reasons for thinking that determinism is a non-issue. First, imagine an imperfect manipulator. He implants a manipulation device but it is imperfect. It works only 90% or perhaps only 20% of the time. Either way, on this one occasion it works. Is the absence of determinism relevant to the judgment that the manipulated person lacks control? I don't see why.
Admittedly I got this thought from reading Mele's Analysis criticism of the 4-case argument, but I don't see why it doesn't apply to his zygote argument, as well.”

Here’s what I say about this in a paper entitled “Manipulation, Compatibilism, and Moral Responsibility” (forthcoming in a special issue of the *Journal of Ethics*):

“Some students have asked me why a version of my test of Pereboom’s best-explanation premise does not undermine the second premise of the zygote argument. Some speculated that people who have the intuition that Ernie, in my deterministic story, is not morally responsible for A-ing and does not freely A would also have that intuition about him in an indeterministic version of the story modeled on my indeterministic versions of Pereboom’s stories. They suggested that this would be a problem for premise 2 of the zygote argument. Would it? Well, it would threaten the claim that the best explanation of the intuition about Ernie in the original story zeroes in on determinism. But the zygote argument has no best-explanation premise. Instead, it claims that, concerning free action and moral responsibility of the beings into whom the zygotes develop, there is no significant difference between the way Ernie’s zygote comes to exist and the way any normal human zygote comes to exist in a deterministic universe. The truth of that claim is in no way threatened by the imagined fact about intuitive responses to the imagined indeterministic story.”

Kristin wrote: “It’s been a while since I looked at the response to the 4-CA from Mele that you mentioned, but I remember thinking that little is gained by discovering that (for many people) the same intuitions follow from both deterministic & indeterministic versions of the manipulation.”

If you keep in mind that the four-case argument is presented as having the form of an inference to the best explanation and then apply orthodox tests for evaluating inferences to the best explanation, you should see what is gained. (If you reformulate the argument as something other than an inference to the best explanation, then, of course, any objections raised to it should be appropriate to its revised form.)

Al

[Kristin Demetriou (Mickelson)]

Hi Al,

Thanks for the reply. I understand where you’re coming from with regards to the general guidelines for assessing the quality of an argument to the best explanation. However, you seem to be assuming that the 4-CA should be judged by an agnostic about the true theory of free will; I am not one, nor do I see any good reason to defer to the agnostic when it comes to evaluating the success of the 4-CA.

When I approach the 4-CA, I approach it with a great deal of baggage, though I do not consider my baggage a burden: I am a proud compatibilist, a soft determinist, and anti-libertarian. I believe that I have developed these beliefs in a philosophically respectable way.

So, when *I’m* confronted with the indeterminism-based 4-CA and determinism-based 4-CA, I am not searching for a best explanation from a neutral standpoint. I judge the quality of a purported best explanation in light of the justified beliefs about freedom and responsibility that I currently have. Given my commitments, the 4-CA sets up a dialectic in which the very general guidelines you mentioned are not in play. Why? Consider:

As an anti-libertarian, I think that indeterminism disrupts the agency required for freedom. Since I have good *independent* reasons to think that indeterminism mucks things up, I can provide justification for judging any manipulation which introduces indeterminism into the causal story leading up to a choice or action as freedom-undermining. Unsurprisingly, then, I immediately conclude that all of the Plums in the indeterminism-based 4-CA are lacking freedom and moral responsibility *because of the indeterminism*.

As a compatibilist, I think that determinism is compatible with freedom. As a result, I look at the determinism-based 4-CA and (initially) find myself stumped: I thought that determinism was compatible with freedom, but I now find myself unable to explain what could account for my intuition better than the diagnosis Pereboom offers. I mean, indeterminism isn’t present, so it can’t be that (though that would have worked). My preferred CAS is (purportedly) satisfied, so it can’t be a failing there. But, gosh, what else can I point to that would explain my intuition that Plum1 & Plum2 are not morally responsible without undermining my compatibilist commitment to Plum4 being a free & responsible agent? Pereboom seems to have revealed to me that determinism, too, undermines freedom—though clearly it must do so in a different way than indeterminism.

In short, then, the reason I said “not much is gained” is because I was thinking about the unique challenge the 4-CA presents for me and about saving my own skin--and I am unable to see how your critique could help me & like-minded theorists defeat the 4-CA.