Mele on Pereboom

In the latest issue of Analysis, Al Mele gives  us a critique of Pereboom’s ‘four-case argument’ for incompatibilism. Pereboom’s strategy is to present four cases, beginning with an entirely manipulated agent and ending with a moderately reasons-responsive agent, but which form a relatively smooth continuum. He argues that the first agent isn’t morally responsible for their actions, because they are causally determined to act as they do, and that the changes introduced into each subsequent scenario do not alter that fact. If agent 1 fails to be morally responsible for their actions, because they are causally determined to act as they do, then agent 4 – a moderately reasons-responsive agent – cannot be responsible either if their actions are causally determined.

Mele claims that the argument fails, because in none of the scenarios presented by Pereboom is it true that causal determinism explains why the agent fails to be morally responsible. Consider case 1, in which (in Pereboom’s version) the agent is manipulated so that his every momentary state is produced by outside deterministic mechanisms operated by neuroscientists. Suppose, Mele asks, that the neuroscientists exercised indeterministic control in this scenario, so that there is a small chance that instead of producing the desires and beliefs that will motivate the action the neuroscientists want, instead the agent will be incapacitated. Surely the agent is no more morally responsible for their action in this scenario than in the first? So causal determinism cannot be doing the work here.

This seems right. But this does not show, as Mele claims, that the four case argument for indeterminism fails. There is a way to revive the strategy, in the face of Mele’s objection. Rather than insist that it is causal determinism that explains the lack of moral responsibility in cases 1-3 (and therefore 4), an incompatibilist should insist that it is causation that is doing the explanatory work here. Probalisitic (or other indeterminist) causation is still causation, and moral responsibility (the incompatibilist should insist) is incompatible with moral responsibility.

After all, why would anyone think that determinism is incompatible with moral responsibility? It’s not (or shouldn’t be) because we think that what happens is settled long before it happens, but because the agent doesn’t control what happens. If I am pushed by a non-deterministic force, which might not have pushed me at all, I am pushed as surely as if I were pushed by a deterministic force. In other words, event-causal libertarian adds nothing to moral responsibility that isn’t compatible with determinism.

I’m not claiming that the four-case argument for incompatibilism will work, if we tweak it as I suggest. But it seems to me that understanding it in this way will refocus the debate where it should be: on the question of the kind of control agents can exert over their deliberations and intention-formation, in both deterministic and indeterministic worlds.

Comments

[Robert Allen]

Neil,

"Probalisitic (sic) (or other indeterminist) causation is still causation, and moral responsibility (the incompatibilist should insist) is incompatible with moral responsibility."

I think you mean to say that Probabilistic (or other indeterminist) causation is still causation, and CAUSATION (the incompatibilist should insist) is incompatible with moral responsibility. But why should anyone think that causation is not compatible with "control"? Haven't Fischer and Ravizza put this notion to rest by drawing a distinction between regulative and guidance control?

[Kip Werking]

Pereboom’s “four-case argument” (which he refers to elsewhere as a “generalization strategy” and credits R. Jay Wallace with that name) is, of course, not just an argument for incompatibilism, but more specifically what he calls “hard incompatibilism.” The label “hard” distinguishes his view from hard determinism, which asserts the truth of determinism. Hard incompatibilism claims that both determinism and indeterminism are incompatible with free will. Or rather “indeterministic causal histories” are incompatible with free will—Pereboom argues that agent-causation is coherent but just not the case. On this last point, I disagree with him and defend Galen Strawson’s a priori case against free will (see our discussion: http://gfp.typepad.com/ilibertarian_accounts_of_/2004/09/chapter_9.html#c2945842).

Considering the above, it is surprising that Mele would criticize Pereboom’s generalization strategy for relying upon determinism to do its work. It does no such thing. Indeed, the argument is easily modified to accommodate Mele’s concern about indeterminism and, in that case, works just as beautifully. Agents who are manipulated with 99% likelihood are just as innocent as agents who are manipulated with 100% likelihood—and their innocence, not determinism, is all that Pereboom argues for.

Mele’s argument illustrates a tremendous stumbling block towards the adoption of skeptical or pessimistic accounts of free will: the belief that determinism, rather than chance, is especially threatening to our freedom. Neither the major proponent of the a priori argument against free will and moral responsibility (the Basic Argument), Galen Strawson, nor the major proponent of the a posteriori argument (the generalization strategy), Derk Pereboom, believes this to be true. Rather, both of their views accommodate indeterminism. Perhaps Ted Honderich, with his enthusiasm for determinism, has aggravated this problem. Nevertheless, so many compatibilists seem impressed by the discovery that indeterminism threatens our freedom too, and therefore conclude that the skeptical position, which has traditionally been characterized as hard determinist, must be mistaken. Dennett, for example, often says that indeterminism can provide no more free will than determinism does. Perhaps the question turns upon how much freedom the philosopher assumes must exist before philosophizing. This attitude does skeptics a disservice.

More Pereboom:
Honderich’s site hosts Meaning In Life Without Free Will
http://www.ucl.ac.uk/~uctytho/dfwVariousPereboom.htm
Pereboom’s website hosts his recent (Forthcoming, Midwest Studies 29, 2005) reply to critics:
http://www.uvm.edu/~phildept/pereboom/Replytocritics.pdf

[Kip Werking]

Correction: The generalization strategy is just an argument against the sufficiency of compatibilist conditions for free will and moral responsibility, not an argument for hard compatibilism itself.

[Neil]

Robert,

I think it is question-begging to take Fischer & Ravizza as having established that guidance-control is all that is necessary for free will. That is precisely what is at issue in the debate, with folk like Randolph Clarke (at least in some of his moods!) claiming that it isn't. In any case, I think I can appeal to authority here: Fischer & Ravizza don't think that guidance control is sufficient for free will, just for moral responsibility.

[Neil]

I should acknowledge that in the four-case argument, the point is to establish the incompatibility of causal determinism and moral responsibility. So Pereboom isn't agreeing with F&R here. But even if we limit the point of the claim in this way, it is still question-begging to claim that F&R have shown that determinism is compatible with control. They have certainly shown that it is compatible with a notion of control, but plenty of folk think it's not the relevant one. F&R may be right, but (as they acknowledge) they don't have any knockdown arguments.

Kip,

For what it's worth I agree with you (and disagree with Pereboom) about agent causation. I suspect that it's impossible - indeed, incoherent - but it wouldn't help if it were. I take it that Mele was deliberately restricting his focus to the argument Pereboom presented, and not to the overall strategy in which it took its place. As I recall, Pereboom's staregy is 'divide-and-conquer', so it makes sense to take a piecemeal approach to criticising his work.

[Derk Pereboom]

Here’s some of what I say in “Defending Hard Incompabilism,” forthcoming in Midwest Studies 29 (2005), about Mele’s argument:

“I agree that the deterministic aspect of the cases is not essential to Plum's lacking moral responsibility in them. In fact, in Living Without Free Will I present an indeterministic manipulation case intended to show that certain kinds of undetermined agents will not be morally responsible. In my view, in either sort of case, deterministic or indeterministic, the agent lacks the sort of control required for moral responsibility. These two sorts of manipulation arguments show that one can lack this sort of control by virtue of being causally determined, and one can lack it by virtue of being a certain sort of indeterministic agent. I also agree that there are many more ways in which one can lack the sort of control at issue. The point of the four-case argument is that determination by factors beyond one's control is sufficient for non-responsibility, for the reason that it precludes the kind of control required for moral responsibility. This point is consistent with the claim that there are other conditions, potentially the theme of other manipulation cases, that are also sufficient for non-responsibility. So determinism’s not being essential to Plum's lacking moral responsibility does not undermine the argument.”

So I agree, for the most part, with what Neil and Kip said about this. One concern with a remark of Neil’s: I do think that control is what’s at issue, but not that causation is what undermines control. On the one hand, one way an agent might lack the sort of control required for responsibility is if her action is completely uncaused. Then she would not cause it, and consequently she would lack the relevant sort of control. On the other hand, in my view it has not been ruled out that if an action were agent-caused, the agent would have the control required for moral responsibility. On agent causation, Neil says “I agree with you [Kip] (and disagree with Pereboom) about agent causation. I suspect that it's impossible - indeed, incoherent - but it wouldn't help if it were.” What I think is that we can’t now discern, in our conception of agent causation, an internal inconsistency, nor an inconsistency with anything else of which we can be reasonably confident – an external inconsistency. (At least I haven't seen such inconsistencies demonstrated anywhere in the literature.) Thus, for all we know, agent causation is coherent. But this is compatible with agent causation really being incoherent, for the reason that our conception of agent causation may not be an ideal – we may not have a maximally clear or complete conception of agent causation (e.g., maybe it’s a conceptual truth that all causation is event causation, despite the fact that we can’t now see that it is). A fortiori, agent causation may, for all we know, be metaphysically impossible. Perhaps this is not so different from what Neil thinks, except for the “but it wouldn't help if it were” part.

[Neil]

Thanks for the feedback, Derk. Perhaps "incoherent" isn't the right word with regard to agent causation. I don't think it is incosistent (which is, I suppose, a primary connotation of "incoherent"). Rather, I have no idea how it's supposed to work. The whole idea is rather mysterious. Two other points: I think the incompatibilist should say that causation *by forces outside the agent* (alone) is incompatible with whatever the incompatibilist is an incompatibilist about, for the reasons Clarke gives in his book. Such causation, probabilistic or deterministic, is incompatibile with control as the incompatibilist should understand it. So that puts the burden of a libertarian defence on agent causation. The reason I think that agent causation wouldn't help is that we get a regress of controlling. Whatever is agent caused, the agent does not exercise control over her causing. My deliberation, or intention formation, is not something I can control in anything except a compatibilist sense, because any control conditions would spark a regress, So I don't have alternative possibilities, in an incompatibilist sense, with regard to my agent causing my decisions or my deliberations. So agent causation adds nothing to compatibilist freedom. I know this is rather obscure at the moment. I'll defend it properly when I have the time to sort it out to my own satisfaction.

[Kip Werking]

“I take it that Mele was deliberately restricting his focus to the argument Pereboom presented, and not to the overall strategy in which it took its place. As I recall, Pereboom's staregy is 'divide-and-conquer', so it makes sense to take a piecemeal approach to criticising his work.”

Neil, you’re absolutely right. I’m not sure why I originally wrote that the generalization strategy argued for hard incompatibilism. Perhaps I remembered this incorrectly because, as Mele charges and Pereboom agrees, determinism itself is not “essential to Plum's lacking moral responsibility”. As I stated above, that one could easily modify the argument so that the agent was shown to be innocent even in an indeterminate universe seems intuitive enough. But as you remind me, Pereboom adopts a “divide-and-conquer” approach: the Luck Objection against event-causal libertarianism, the generalization strategy against compatibilism, and our best scientific theories against agent-causal libertarianism.

This brings me to a related criticism of the generalization strategy: does anyone agree with Fischer’s claim that Plum is morally responsible but is not blameworthy? This distinction seems to be too weak to do the work Fischer demands of it. Furthermore, I would agree with Pereboom that the criticism seems to rely upon a counter-intuitive sense of moral responsibility.

This last criticism raises the issue of semantics. I would agree with McKenna that the compatibilist’s best strategy is to generalize backwards. McKenna’s strategy is similar to Fischer’s, but his argument seems more explicit and persuasive. Pereboom’s reply is less convincing here—because the term “moral responsibility” is ambiguous enough to suggest that Plum is guilty as well as innocent.

“The compatibilist needs to make clear that once the manipulation is so qualified that all an agent’s current time-slice compatibilist-friendly structures are properly installed through a process of manipulation, then the role of the manipulator begins to shrink into the background; we are simply left with a normal person who happened to be brought into existence in a very peculiar manner.”

Ultimately, we are left with Watson’s dilemma: even the most rational and reason-responsive of agents might yet be the puppets of a controller. In that case, do we hold the agent morally responsible? Fischer and McKenna appeal to the rationality and reason-responsiveness of the agents. But the intuition that they are not morally responsible for the actions persists. Perhaps, as Shaun Nichols and Joshua Knobe’s fascinating “Responsibility and Determinism: Empirical Investigations of Folk Intuitions” (http://gfp.typepad.com/online_papers/2004/12/shaun_nichols_j.html) argues, one would have the intuition that these manipulated agents are innocent until one focuses upon their immoral actions, which arouse the reactive attitudes.

[Robert Allen]

Neil,

That's what you were talking about: moral responsibility, not free will. The claim in question is that if an action is caused, then it's agent is not morally responsible. I said in response that if the causing entails guidance control, then a case can be made that she is morally responsible. Now other conditions may have to be added here. (I have added one myself: that the agent be able subvert the will of anyone who has instilled a value in her.) But at least it has been shown that whether or not an agent is morally responsible depends upon HOW her action has been caused. Causation per se does not rule out moral responsibility.

[Neil]

Robert,

I agree with much of this. Yes, whether an agent is MR for an action depends upon how it is caused. But I disagree that anyone has "shown" (a success word; here understood to mean "with such rational force that dissent is not reasonable") that causation by forces outside the agent is compatible with MR. Causation by the agent's reasons (or something like that) might be a necessary condition for MR, but the question whether causation *of* these reasons by forces beyond the agent is compatible with it is, if not quite an open question, at least not an entirely closed one (yes, yes, I know that "open" and "closed" exhaust the logical space. It just seems to me that to call something an open question is to imply that no one has any right to hold a view one way or another, whereas to say it is not entirely closed is to imply that rational dissent from views is in order).

[Al Mele]

I’m glad people are finding my *Analysis* paper interesting. The structure of my argument is very simple. Here is Derk’s punch line about his 4-case argument in his book: “The best explanation for the intuition that Plum is not morally responsible in the first three cases is that his action results from a deterministic causal process that traces back to factors beyond his control. Because Plum is also causally determined in this way in Case 4, we should conclude that here too Plum is not morally responsible for the same reason” (p. 116). Now, intuitions are caused. So it’s natural to apply Mill’s methods to the first sentence in the quoted claim. Should we be confident that the deterministic causation in the first three stories is causing the (alleged) intuition? A natural test is to eliminate the deterministic causation from the stories and hold another feature of the stories constant – the manipulation. When we do that (in the way I do it), it certainly looks like people will have the same intuition about the modified stories that they have about the original stories. So we should *not* be confident that the deterministic causation in the first three stories is causing the (alleged) intuition. Now look at the second sentence in the passage I quoted. There you have it, in a nutshell.

I’m certainly not claiming that there are no better arguments for incompatibilism. In fact, I float an argument for incompatibilism that I claim is much better in my new book ms, *Free Will and Luck.* It first appeared in this blog in June under the title “the zygote argument.”

Al Mele

[Derk Pereboom]

Suppose a dam at one end of a reservoir would burst if the reservoir were filled with more than one billion gallons of water, because the dam could not withstand the pressure that volume of water would exert. Suppose the reservoir is in fact filled with more than one billion gallons of water, and the dam bursts. It is natural to say here: “what explains the dam’s having burst is the water pressure.” Suppose someone says at this point: “if the reservoir were filled with more than one billion gallons of oil, it would have burst. So by the method of differences, it isn’t the water pressure that explains the dam’s bursting.” To this we’d want to respond: some true causal explanations lay out actual sufficient conditions for an event’s occurring, and so the explanation in terms of water pressure is true. But in response to the point about the method of differences, we can also point out that there is an explanation common to both the water pressure and the oil pressure cases that is in a sense deeper than the water pressure explanation. It’s that liquid pressure higher than a certain level would cause the dam to burst. (There are even deeper explanations.)

By analogy, in case 2 it is correct to say that Plum’s decision’s being deterministically caused by factors beyond his control explains why he is not morally responsible. This explanation sets out the sufficient conditions, in that case, for Plum’s not being responsible. At this point Mele points out that changing determinism a kind of indeterminism would not alter the verdict that Plum is not morally responsible. But, by analogy with the example of the dam’s bursting, this is compatible with the truth of the deterministic explanation for Plum’s lack of responsibility in case 2, and with a deeper explanation for why he is not morally responsible that is common to both cases, viz., that due to the causal circumstances of his decision, he lacks the sort of control required for moral responsibility.

[Neil]

Okay, but all (relevant) determinism is causal determinism (whereas not all pressure is water pressure). It's the causation that's doing the work in "Plum", not the causal determinism, which can only affect MR by being causation.

[Derk Pereboom]

Neil -- But I don't think it's been ruled out that we can be morally responsible for agent-caused action. Suppose it turns out that we indeed can be. Then (as I see things) while causal determination of a decision will be one kind of causal circumstance sufficient to undermine moral responsibility, and the kind of indeterminism that Al adduces will be another, it won't be by virtue of its being a kind of causation that causal determination is sufficient to undermine moral responsibility. Then it won't be causation that's doing the work. However, if moral responsibility for agent-caused decision can be ruled out, then it might be that things are as you say. For me, and for many others, some sort of causation by us is intuitively a key requirement for moral responsibility, since, intuitively, only by some sort of causation by us could the right sort of control be secured. It would be paradoxical if meeting this requirement all by itself was also sufficient to undermine moral responsibility. (But that's not to say that this isn't also how things are.)

[Al Mele]

Two things. (1) I don’t find Derk’s analogy apt. Suppose that there are two kinds of thing messing around behind a dam – ghosts and goblins (just as, in Derk’s cases, we have two kinds of thing – deterministic causation and various kinds of manipulation). We wonder: Did the ghosts break the dam? Did the goblins break the dam? Did they break it together? So we haul the goblins to an intact dam a lot like the one that was broken and we leave the ghosts behind. The dam breaks. Now we try the same thing with the ghosts. The dam doesn’t break. . . . Make your own causal judgment. In my paper, I say how I think people will respond to certain versions of Derk’s cases in which the deterministic causation is subtracted. Suppose we subtract the manipulation from Derk’s cases and keep the deterministic causation. Incompatibilists will make incompatibilist judgments about the cases and compatibilists will make compatibilist judgments.

(2) Neil, for grounds for thinking that uncaused action is conceptually impossible, see chapter 2 of my *Motivation and Agency* (OUP, 2003).

Al Mele

[Derk Pereboom]

The analogy is meant to show that an explanation that sets out the sufficient causal conditions for X can be true even if turns out that altering some features of those sufficient causal conditions, and not others, would also result in X; and also that in such cases there can be a deeper explanation, common to both situations, for X. For this purpose I think the analogy works. In accord with this analogy, it is compatible with altering the causal determinism in case 2 to Al’s indeterminism, and getting Plum’s non-responsibility in both cases, that causal determinism does explain his non-responsibility in case 2, while a deeper fact, such as the presence of causal circumstances that precludes responsibility-relevant control, explains his non-responsibility in both cases. In my broader story, causal determinism precludes moral responsibility because it is a type of causal circumstance that precludes responsibility-relevant control. So my view won’t be undermined by pointing out that there are cases of non-responsibility in which causal determinism is absent, manipulation is present, while the manipulation generates another type of causal circumstance that precludes responsibility-relevant control.

On Al's last point in (1), if you can get people generally to have the intuition that a non-manipulated but causally determined agent is morally responsible -- under conditions of reasonable reflection, where they are sufficiently cognizant of the agent’s causal determination –- then my view faces a very steep challenge. One thing my four-case argument attempts to do is to set up a situation in which an agent is causally determined but not manipulated (case 4), in which those engaged in the thought-experiment are sufficiently cognizant of the fact that he is causally determined (via the presentation of cases 1-3). My argument would fail if after being presented with cases 1-3, reasonable and reflective people would generally have the intuition that Plum is morally responsible in case 4.

[Robert Allen]

"In my broader story, causal determinism precludes moral responsibility because it is a type of causal circumstance that precludes responsibility-relevant control."

Derk,

Why should a Compatibilist accept this either? She could just demonstrate that it's not CD that's "preclud(ing) responsibility-relevant control," in any case you might offer that elicits the intuition that the agent involved is not in control of himself, but manipulation, employing Al's strategy as before. If a neuroscientist were manipulating my brain (as in Plum 1), that alone would keep me from being in control of myself.

[Derk Pereboom]

Robert -- How about this: suppose that Ann has agent-causal free will, and has a strong desire to help the stranded motorist and a weaker desire to drive off. The neuroscientist then manipulates her brain so that her desire to drive off is equal in strength, or even stronger than, her desire to help. But, exercising her agent causal free will, Ann decides to help. My intuition is that she is morally responsible for helping, despite her being manipulated. My diagnosis is that the manipulation in this case does not generate a type of causal circumstance that precludes responsibility-relevant control, whereas the manipulation in case 2, and in Al's indeterministic parallel case, do generate such a causal circumstance.

[Neil]

Derk,

I'm not sure what role the manipulation in your case (in response to Robert) is actually playing. A neuroscientist manipulates Ann's desire to drive off, but she helps instead. So the manipulated desire plays no role in the actual sequence. In your original cases, though, the manipulation does play a role in the actual sequence. Either you think that agent-causation is a sufficient condition of MR, all by itself and no how the agent is manipulated, or you think that manipulation on the actual sequence precludes MR. So I'm not sure that your example works as a response to Robert.

BTW, you and Al both misunderstood my point. I too think that action must be caused. I had earlier qualified the causal claim with the addition "causation *by factors external to the agent*" I inadvertently dropped the qualification.

[Derk Pereboom]

The neuroscientists’ procedure in my example does have an effect on the actual sequence, since it generates an actual-sequence desire that would not have existed absent this procedure. So what would it be that rules out this procedure as a kind of manipulation? If the alternative proposal is that, by definition, a process counts as manipulation only if it generates causal circumstances that preclude responsibility-relevant control, then I’m happy to say that, in case 2 and in Al’s parallel indeterministic case, manipulation explains Plum’s non-responsibility. But manipulation would then explain Plum’s non-responsibility only by virtue of the fact that, by definition, a process counts as manipulation only if it generates causal circumstances that precludes responsibility-relevant control. Does anyone have a characterization of manipulation that allows it to play precisely the role in explaining non-responsibility that Al and Robert have in mind?

[Robert Allen]

Derk,

That's the great thing about agent causal FW, the sort an immaterial substance would exercise: as MC Hammer would put it, "you can't touch this." On the other hand, if I am my brain or some part of it, then, in Paul Churchland's phrase, someone else could get their "hands on the levers of control."

I have characterized a manipulated agent as someone who is incapable of subverting the wills of those who have instilled values in him, that is, to use Al's apt term, someone who cannot "shed" the values of his influences and replace them with others that they do not intend for him to have. (The replacement part is my emendation of Al's view in response to an objection of Tomas Kapitan. Cf. his "Autonomy and Manipulated Freedom," in Phil. Persp. 14.) If you instill a value in me, to get me to do something, and I act from it instead of excising it from my mind-set, as I am capable of doing, then you are not controlling but influencing me, ala advertisers. Only if I were incapable of excising it could you count on me doing your bidding: then you or someone/something else would have destroyed my will. (I develop this idea in "Castle's Choice: Autonomy and Transcendence," available @ http://home.twmi.rr.com/robertallen/.)

Despite what you said @ Wheaton about the belief in souls being "weird," (or maybe because of it), I would like to develop a dualistic theory of agent causation. But until such time as I do, my belief in my ability to subvert the will of anyone who has influenced me will be what makes me think that my own will is free.

[Derk Pereboom]

So maybe it turns out that on that sort of definition of manipulation, an agent’s being manipulated is sufficient for non-responsibility. Manipulation would then ensure the presence of causal circumstances that preclude responsibility-relevant control. Then I would agree that manipulation explains Plum’s non-responsibility in case 1 and in Al’s parallel indeterministic case. But this does not all by itself help show that causal determinism is not also sufficient for and explains non-responsibility. For none of this undermines the claim that manipulation would be one way to ensure causal circumstances that preclude responsibility-relevant control, while causal determination is another. On the other hand, if after being presented with cases 1-3, reasonable and reflective people generally would have the intuition that Plum is morally responsible in case 4 (in which the agent is causally determined but not manipulated), that would make for a strong argument against causal determinism being sufficient for non-responsibility. But part of the challenge here is to point out some responsibility-relevant difference between, say, case 2 and case 4, that would explain why Plum is not responsible in case 2 but responsible in case 4; or else, a satisfying explanation of why Plum is responsible in case 2 despite strong intuitions to the contrary, and this I haven’t yet seen.

[Robert Allen]

Derk,

"For none of this undermines the claim that manipulation would be one way to ensure causal circumstances that preclude responsibility-relevant control, while causal determination is another."

Again, why should anyone think that causal determination all by itself precludes R-RC (rather than enhances it, considering the Mind objection to Prof. Kane's Libertarianism)? The R-RC difference between Plum 2 and 4 is that the former but not the latter lacks the ability I defined in my last post. If the manipulators have programmed Plum 2 so that any critical thinking in which he might engage is guaranteed, as you specify, ("(I)f it threatens to result in a character that runs afoul of their plans ....") to result in an endorsement of the values they have instilled, then he simply cannot overcome their influence; he must remain under their control. Compare that to the tenuous 'hold' a typical parent has on his/her children.

[Seth Shabo]

This is my first visit to the Garden. Nonetheless, I'll hazard an interpretation of Pereboom's argument that differs slightly from the ones suggested here. I'll then address Mele's critique in light of that interpretation.

As I see it, Pereboom's attack on compatibilism has three prongs. 1. His first case directly impugns the sufficiency of the leading compatibilist conditions. 2. Because these conditions are satisfied even in this first case, compatibilists will find it difficult to draw a principled line anywhere among the four cases. 3. Pereboom suggests a plausible explanation for why Plum isn't responsible in case 1, viz. Plum's not the ultimate source of his conduct (the manipulators being at least more nearly ultimate). If the underlying source incompatibilist principle--that moral responsibility requires ultimate origination--is valid, this will reinforce the conclusion that Plum isn't responsible in case 4 either. (At this stage, it's a further question whether Pereboom's version of this principle, which holds that ultimate origination is incompatible with both causal determination _and_ indeterministic event-causation, is the right version...)

I submit that on this interpretation, Pereboom can say, pace Mele, that causal determination needn't play a direct role in explaining our intuition that Plum isn't responsible in case 1 (where, among other things, Plum's reasoning process is manipulated). Rather, the main explanation will be that Plum is clearly not the ultimate originator of his conduct in that case. If this explanation is right, it generalizes to other deterministic cases--and perhaps to some indeterministic ones (maybe even to all event-causal ones...). Finally, the explanation does seem plausible. That is, it does seem plausible to attribute our intuition that Plum isn't responsible in case 1 to the scenario's making it vividly apparent that he's not the ultimate source of his conduct. (Granted, the explanation is somewhat philosophically sophisticated, but that needn't count against its being a good clarification of our gut reaction...right?)

[Robert Allen]

"If the underlying source incompatibilist principle--that moral responsibility requires ultimate origination--is valid, this will reinforce the conclusion that Plum isn't responsible in case 4 either."

But it is not valid. Pereboom, Strawson, and Honderich have set the bar way too high, requiring self-creation, an impossibility, for FW. Am I supposed to believe that I lack a free will because I did not do what no one COULD do: develop a character ex nihilo?

On my view, there is one uncreated being who is ultimately responsible for the existence of the rest of us. But since neither that being nor any other being is in control of me- I can subvert anyone's will- my will is free.