In his “Farewell to the Direct Argument” (Journal of Philosophy 2002), David Widerker argues that ‘direct arguments’ for incompatibilism all implicitly depend on the assumption that determinism rules out the ability to do otherwise, and therefore collapse back into traditional ‘indirect arguments’ for incompatibilism. According to Widerker, “the hopes of proponents of the direct argument to be able to prove that determinism excludes moral responsibility without the notion of avoidability was unrealistic in the first place. To see this, it is enough to consider one central assumption of the direct argument, to wit, that no one is morally responsible for the laws of nature. Why should one accept it if not for the fact that no one could have prevented them from obtaining?” (324).
I’m curios about Widerker’s argument for a number of reasons, not least of which being what it would mean for the debate if it were true. For one, it would seem to make non-PAP libertarianism (such as that endorsed by Stump and Zagzebski, among others) unstable. But I’m not aware of any reactions to this argument in the literature. Are they there? If not, this would seem to be a crucial point that needs to be addressed, right?
I don't have the Widerker paper in front of my and haven't read it for a year or more but here's what I think about your post given my memory of the Widerker paper. My memory may be way off on this.
Here are two issues worth addressing that, so far as I know, have not been addressed in the professional literature. [But perhaps I'm unaware of something someone has forthcoming?]
1. Does Widerker "get direct arguments right" before he criticizes them? That is, does he present them correctly and in their best light? I remember thinking that the way he spelled out discussion from van Inwagen and me on direct arguments didn't seem quite right to me. Peter and I have both talked about transfer principles concerning "partial moral responsibility" (and lack of it) and I think I recall Widerker treating us as if we were talking about transfer principles for moral responsibility and lack of responsibility). This difference makes a big difference in the details of the argument (in the logic of the transfer principles).
2. Is the "non-PAP" libertarianism of Stump, Zagzebski and others (a) the same view? and (b) coherent?
This one is of interest independent of the relevance of Widerker's discussion to the issue but perhaps that connection is a nice entry to the issue. Terminologically this view has always annoyed me because I've always thought PAP is a moral responsibility principle and not a metaphysics of freedom principle but the PAP of the non-PAP libertarians isn't that familiar PAP - it's something "inspired" by the old PAP and more directly relevant to the metaphysics of freedom.
***
I think both of these issues, and related issues, are well worth exploring.
Posted by: Fritz Warfield | February 03, 2005 at 04:52 AM
A few comments on Widerker on the direct argument.
1) Kevin, I’m not sure why you say that Widerker believes that direct arguments “implicitly depend on the assumption that determinism rules out the ability to do otherwise.” As I see it, Widerker argues that all versions of the direct argument incorporate an invalid inference rule. He does so by providing counterexamples to a variety of non-responsibility transfer principles. Widerker’s central claim is that there are “asymmetries between the notion of avoidability and that of moral responsibility. For example, although no agent’s performance of some action in the past is now unavoidable, he may nevertheless now be morally responsible for it. And although an agent cannot be now responsible for an act he has not performed yet, that act might now be avoidable.” (320, fn. 9) For this reason, one cannot directly argue that moral responsibility is incompatible with determinism. Rather one must argue for this conclusion in the traditional, indirect way by assuming a principle like PAP.
2) Fritz is correct that Widerker’s own non-responsibility transfer principles are different than those used by both Fritz and van Inwagen. Widerker defines the non-responsibility operator as such:
NR(p) =df. “p and no one is (now), or ever has been, morally responsible for the fact that p.” (317)
As Fritz notes, he and van Inwagen talk about “partial moral responsibility.” But some of the Widerker’s counterexamples were originally put forth by Fischer and Ravizza (1998), and in these examples the relevant operator is specified in terms of “partial moral responsibility.” See for instance Fischer’s “The Transfer of Nonresponsibility” in my co-edited volume Freedom and Determinism (MIT Press, 2004). There are, of course, reformulations of non-responsibility transfer principles and additional counterexamples, so this issue needs to be looked at more closely.
3) Non-PAP libertarianism is a kind of source incompatibilism. But source incompatibilism is a much broader view that includes varieties of free will skepticism as well as libertarianism. Pereboom, for instance, is a source incompatibilist.
Source incompatibilists (a) accept the source theory of free will on the basis of Frankfurt-type examples, (b) believe that determinism is incompatible with moral responsibility, and (c) require some version of the direct argument to prove incompatibilism.
I’ve been working on a general argument, inspired by Widerker’s work, which attempts to show that source incompatibilism is incoherent. Here is a rough idea of my argument.
In Frankfurt-type examples an agent S is claimed to be morally responsible for the occurrence of some event e even though he could not have done otherwise. In such examples, the relevant event e must be the inevitable result of one of two possible causal chains: c1 (the actual, indeterministic chain that has its source in S) or c2 (the counterfactual chain). Let C1 be the proposition that chain c1 is causally efficacious and results in e, let C2 be the proposition that chain c2 is causally efficacious and results in e, and let E be the proposition that agent S brings about e.
In any adequate Frankfurt-type example, that (C1 or C2) entails that E. Proponents of Frankfurt-type examples must claim that S was partly morally responsible for the fact that E. But such proponents must also claim that S was not even partly morally responsible for the fact that (C1 or C2), for otherwise the fact that E was not unavoidable for S. Thus, it seems that Frankfurt-type examples also provide the basis for counterexamples to agent-relative non-responsibility transfer principles.
Obviously there is a lot of work to be done but hopefully this is enough to encourage you, Kevin, to pursue your investigation further!
Posted by: Joe | February 08, 2005 at 06:38 AM
"In any adequate Frankfurt-type example, that (C1 or C2) entails that E. Proponents of Frankfurt-type examples must claim that S was partly morally responsible for the fact that E. But such proponents must also claim that S was not even partly morally responsible for the fact that (C1 or C2), for otherwise the fact that E was not unavoidable for S. Thus, it seems that Frankfurt-type examples also provide the basis for counterexamples to agent-relative non-responsibility transfer principles."
Joe,
Talk about hoisting someone on their own petard! You have shown that source incompatibilists cannot accept the moral of FCs and then turn around and use "agent-relative non-responsibility transfer principles" to "prove incompatibilism." That is, having eschwed avoidability as a necessary condition for MR, they must also reject the idea that one cannot be responsible for something unless one is also responsible for ALL of its antecedents. The source incompatibilist must either give up the claim that S is responsible for E even though it is unavoidable or concede that the fact that S is not responsible for C1 or C2 does not mitigate his responsiblity for E. If she does make the latter concession, then the causal connection between, say, the Big Bang and my writing this post cannot be cited as a reason for denying that I am responsible for it.
Posted by: Robert Allen | February 11, 2005 at 08:34 AM
Thanks, Robert!
I just finished rereading Widerker’s “Farewell to the Direct Arugment” and I have two minor corrections to make to what I said above.
First, Widerker responds to Fritz’s point about the change in Widerker's formulation of the non-responsibility operator in fn. 6 of “Farewell.” I overlooked this in my original comments.
Second, it seems that Kevin is closer to the truth than I thought that he was in characterizing the conclusion of "Farewell." Here are Widerker’s own words: “the direct argument ... depends for its plausibility on the assumption (IVD) of the incompatibility of determinism with the freedom to avoid acting as one did.” (323) This needs a bit of explaining.
There are two main parts to “Farewell.” In part I, Widerker does what I said that he does: he shows that various versions of the direct argument incorporate an invalid inference rule by providing counterexamples to a variety of non-responsibility transfer principles.
In part II, Widerker anticipates a reply that the proponent of the direct argument might make, e.g., to restrict reformulations of these transfer principles and circumvent the counterexamples (see fn. 14, for instance). But at this point it is open to the compatibilist to provide a meta-counterexample to any such reformulation. Borrowing from a previous example, “Fate,” the compatibilist might claim that even though (the relevant agent) Jones’s action is the result of deterministic processes he is nonetheless morally responsible for his action since (a) Jones knew he was acting wrongly and (b) Jones could have avoided doing what he did. (As Widerker points out (fn. 15) the claim here isn’t that (b) is a necessary condition for moral responsibility but that (a) and (b) together provide a sufficient condition for moral responsibility.) Widerker writes: “the dialectical situation has changed. For now it is clear that the validity of the improved version of transfer NR ... requires that determinism rules out avoidability.” (323)
Thus, I still think that it is incorrect to say--as Kevin wrote--that, according to Widerker, “‘direct arguments’ for incompatibilism all implicitly depend on the assumption that determinism rules out the ability to do otherwise.” There are many versions of the direct argument that do not depend on this assumption but such versions incorporate invalid transfer principles. I think that it is closer to the truth to say that, according to Widerker, any sound version of the direct argument is going to implicitly depend on the assumption that determinism rules out the ability to do otherwise.
I should also note that the final argument in my first set of comments is inspired by “Farewell” as well as Widerker’s libertarian response to the Frankfurt examples (sometimes called the Kane/Widerker dilemma). See the article by Widerker in Tomberlin (ed.) Philosophical Perspectives, 14, Action and Freedom (Blackwell 2000). In the same volume Pereboom has a nice piece where he surveys the relevant literature up to 2000 and responds to the Kane/Widerker dilemma. See also Fischer’s “Recent Work on Moral Responsibility,” Ethics 110 (October, 1999). Condensed versions of both the Pereboom and Fischer articles can be found in Kane’s Free Will (Blackwell, 2002). There are also some nice articles by and exchanges between Widerker, Fischer, Stump, and others in Faith and Philosophy (1995-6).
I think that if you combine Widerker’s criticisms of the direct argument and the Frankfurt examples you get a very interesting and compelling criticism of source incompatibilism.
Posted by: Joe Campbell | February 12, 2005 at 11:23 AM
Here’s how I see it: In standard Frankfurt-style arguments, the conditions of the case are sufficient for the action at issue. Given the source incompatibilist’s endorsement of some such Frankfurt-style argument, she can’t also claim that if in some situation conditions hold that are sufficient for an action, that’s sufficient for the agent’s non-responsibility. The source incompatibilist would therefore avoid any direct argument that specifies such a condition for non-responsibility. At this point she contends that it’s the nature of the actual sequence that settles whether the agent is morally responsible. In my own view, to be morally responsible, the agent would have to be an agent-cause, and (in standard cases) she would have to exercise agent-causal free will in the actual sequence. Now a number of people, myself included (also Hunt, Mele/Robb, McKenna), have argued that it’s possible to construct a successful Frankfurt-style case in which there is no causal determinism in the actual sequence. For example, a successful Frankfurt-style case can have this feature when the trigger for intervention is a necessary condition for any robust alternative possibility. My Tax Evasion case works this way. This feature allows the agent to exercise even agent-causal libertarian free will in the actual sequence. If so, then at no point in this sequence will the agent be causally determined to make the decision. And the decision will then not be an inevitable result of any causal chain in the actual sequence. To complete the case for source incompatibilism, I develop the argument from manipulation cases designed to show that deterministic (and also event-causal indeterministic) conditions suffice for non-responsibility. (I don’t press a direct argument at this point.) Endorsing this manipulation-case argument is compatible with endorsing the Tax Evasion argument, and so I don’t see that there’s any problem for source incompatibilism here.
Posted by: Derk Pereboom | February 12, 2005 at 06:26 PM
Thank you for your comments, Derk! From your contribution we can see how unique your own position is and, thus, how diverse the group of source incompatibilists is. Nonetheless I think that my general point applies to your position, as well. This is not the place to provide exhausting details, so let me just try to show that your own arguments in support of incompatibilism (between determinism and moral responsibility) commit you to the view that the agent in Tax Evasion is not morally responsible for his actions.
As you note above, you do not use a version of the direct argument to argue for incompatibilism. Rather, you use the argument from manipulation cases. This argument is too subtle and complex for me to discuss in detail here. So let me just quote your endorsement of a principle you refer to as “the fundamental incompatibilist claim”:
“… if one’s action results from a deterministic causal process that traces back to factors beyond one’s control, to factors that one could not have produced, altered or prevented, then one is not free in the sense required for moral responsibility.” (249, “Determinism al Dente,” Free Will, edited by Pereboom (Hackett 1997))
This passage comes directly after you give the argument from manipulation cases. There is a similar quote directly before you give the argument (246) and another similar quote later in the essay (252).
My best attempt to characterize the fundamental incompatibilist claim is as follows:
If NC[S]p & p entails q, then NR[S]q,
where ‘NC[S]p’ means p is a fact over which S has no control, and ‘NR[S]p’ means S is not morally responsible for the fact that p.
When I say that S has no control over the fact that p, I mean that p is a fact that is not produced by S, cannot be altered by S, and cannot be prevented by S. More specifically, let’s say that a fact p is not produced by S iff p is a true proposition and S does nothing to make it the case that p, and a fact p cannot be altered or prevented by S iff p is a true proposition and there is nothing that S can to make it the case that not-p.
Tax Evasion is an ingenious Frankfurt-type example from your (Derk’s) “Alternative Possibilities and Causal Histories,” reprinted in part as “The Explanatory Irrelevance of Alternative Possibilities” in Kane’s Free Will (page references are to this latter volume; both sources are referenced in more detail in my previous blog contribution). The case involves a man named ‘Joe’ (not me, of course!) who is about to commit tax evasion. Here are the concluding two sentences of Tax Evasion.
“But to ensure that [Joe] choose to evade taxes, a neuroscientist now implants a device which, were it to sense a moral reason occurring with the specified force, would electronically stimulate his brain so that he would choose to evade taxes. In actual fact, no moral reason occurs to him with such force, and he chooses to evade taxes while the device remains idle.” (119)
(There are some important details of the case that I’ve left out and I invite the unfamiliar reader to review Derk’s article in more detail. It really is an ingenious example!)
Let a be the action that Joe performs, that is, Joe’s act of tax evasion, which I call ‘a.’ Let c1 be the actual, indeterministic causal chain that has its source in Joe. Note that, since c1 is actual and indeterministic, I am not making any assumptions about the truth of determinism in this argument. More specifically, Joe’s action a is not the result of a deterministic causal process and we may even assume that a is a result of Joe’s “agent-causal libertarian free will.” Let c2 be the counterfactual causal chain, that is, the causal chain that would have resulted in a were the neuroscientist’s implanted device to sense a moral reason occurring in Joe with the specified force.
Let C1 be the proposition that chain c1 is causally efficacious and results in a, let C2 be the proposition that chain c2 is causally efficacious and results in a, and let A be the proposition that agent S brings about a.
Consider now the disjunction (C1 or C2). First, (C1 or C2) is true. Second, Joe did not make it the case that (C1 or C2); the neuroscientist made it the case by implanting the device ensuring that either c1 or c2 will be causally efficacious and result in a. I am not denying that Joe made it the case that C1 but by the time that C1 was made the case, (C1 or C2) has already been rendered inevitable. This very fact is built into the example. Third, S cannot make it the case that not-(C1 or C2) since he cannot make it the case that not-C2. Thus, (C1 or C2) is a fact over which S has no control. More formally:
NC[Joe](C1 or C2).
But (C1 or C2) entails A.
Hence, NR[Joe]A.
That is, given the two premises and the fundamental incompatibilist claim it follows that Joe is not morally responsible for his act of tax evasion. QED
Posted by: Joe | February 13, 2005 at 08:21 AM
Sorry! Instead of "let A be the proposition that agent S brings about a" I meant to write "let A be the proposition that Joe brings about a." And instead of "Thus, (C1 or C2) is a fact over which S has no control" I meant to write "Thus, (C1 or C2) is a fact over which Joe has no control." I was playing Yu-Gi-Oh with my son, Lake, while finishing this post!
Posted by: Joe | February 13, 2005 at 08:36 AM
Joe: Thanks very much for the interesting challenge.
You say:
“My best attempt to characterize the fundamental incompatibilist claim is as follows:
If NC[S]p & p entails q, then NR[S]q,
where ‘NC[S]p’ means p is a fact over which S has no control, and ‘NR[S]p’ means S is not morally responsible for the fact that p.”
I think that my version of the fundamental incompatibilist claim can’t be stated this way. As I see it, by virtue of accepting standard Frankfurt-style arguments, the source incompatibilist agrees that it’s possible that conditions hold that are sufficient for or entail that an action is performed, or render it inevitable, while the agent can still be responsible for this action. But crucially, there’s a distinction between conditions holding that are sufficient for or entail that an action is performed, or render it inevitable, and those conditions causally determining the action. This is illustrated by these recent sorts of Frankfurt-style cases, such as Tax Evasion. So in Tax Evasion (as embellished in my last post), Joe’s action does not result from a deterministic process that traces back to factors beyond his control, since his decision results from an exercise of his libertarian agent-causal power; but still, as you point out, there is a condition relevantly beyond Joe’s control that is sufficient for and entails that the action is performed, and renders it inevitable. If this distinction can’t bear the weight the source incompatibilist wants it to, then the view is jeopardy. But I think that it can bear this weight. As the manipulation argument indicates, it is intuitive that for an agent to be morally responsible for an action, it cannot be produced by a deterministic process in the actual sequence that traces back to factors beyond this agent’s control; and, at the same time, as Tax Evasion shows, it is intuitive that an agent can be responsible for an action despite the fact that conditions beyond his control hold that are sufficient for or entail that the action is performed, or render it inevitable; and nothing about Tax Evasion undermines the moral of the manipulation argument as I just stated it. (It's highly significant to the source incompatibilist's argument that in Frankfurt-style cases the key feature of the condition that renders the action inevitable has no causal role in the actual sequence.)
Posted by: Derk Pereboom | February 13, 2005 at 05:26 PM
Thanks for the reply, Derk! No doubt we’ve reached an impasse in the debate but let me just make a few quick points and then I’ll drop the issue.
As I understand your last post, a better way to characterized the fundamental incompatibilist claim (as opposed to the way that I characterize it above) is as follows:
If NC[S]e & e is a set of conditions that causally determines a, then NR[S]a.
Call this way of characterizing the fundamental incompatibilist claim ‘DV’ (for Derk’s version) and my original way of characterizing the claim ‘JV.’ I see several problems with preferring DV over JV.
First, DV is clearly question begging. If one has DV, then one has a pretty quick argument for incompatibilism (between determinism and moral responsibility).
1. There exist a set of conditions e which occurred at sometime in the remote past, at a time before any human beings came into existence.
2. If determinism is true, then e causally determines every subsequent event and thus every subsequent action.
3. DV
4. Therefore, incompatibilism is true.
Clearly the compatibilist will reject DV. Moreover, given DV the manipulation argument is unnecessary. The real question is whether the manipulation argument is intended to provide justification for the acceptance of DV or whether DV is presupposed in the manipulation argument. If the latter, then the manipulation argument is also question begging.
Second, there is no substantive reason for accepting DV and rejecting JV. Let E be the proposition that a certain set of conditions e occurs and let A be the proposition that a certain action a occurs. Then if e is s set of conditions that causally determines a, it follows that E entails A. Given a standard definition of ‘determinism’ (van Inwagen 1983, 65), wherever you have causal determination you also have entailment.
Third, in as much as incompatibilism is the intuitive view, as you suggest, our intuitions hardly stop there. It is likely that people find fatalism to be as problematic as determinism, where ‘fatalism’ is the view that certain events are inevitable regardless of their prior causes. This suggests--as the traditional incompatibilist has maintained all along--that people find determinism problematic because determined events are inevitable, and that people are motivated to accept DV on the basis of JV.
Finally, here is an example that drives home all of the points made above. It is a slight variation on Tax Evasion that I call ‘Fatalistic Tax Evasion.’ The set-up is exactly the same as Tax Evasion but for one difference. Recall that in Tax Evasion the story goes like this: “to ensure that [Joe] chooses to evade taxes, a neuroscientist now implants a device which, were it to sense a moral reason occurring with the specified force, would electronically stimulate his brain so that he would choose to evade taxes.”
In Fatalistic Tax Evasion, the device also senses if and when Joe has a moral reason with a specified force. But instead of stimulating the brain so that Joe chooses to evade taxes, the device merely wipes out Joe’s recent memories and effectively puts him back into a state where Joe begins his decision process over again. In other words, the device never forces Joe to choose to evade taxes it merely never allows him to choose not to evade taxes. The only situation allowed by the device is one in which Joe (eventually) chooses to evade taxes. Suppose that Joe does so choose. My guess is that folks with incompatibilist intuitions will also have the intuition that Joe is not morally responsible for his action, so it cannot be DV that is driving those intuitions.
Posted by: Joe | February 15, 2005 at 07:55 AM
Joe – for the record, here’s my view about what your say: The compatibilist will indeed reject DV, since it’s a fundamental incompatibilist claim. And because it’s a fundamental incompatibilist claim, it needs to be supported by argument against the compatibilist. In this way, a defender of DV would avoid begging the question. I intend the manipulation argument to support DV – and in my account the manipulation argument does not presuppose DV. Even if DV entails JV, JV doesn’t entail DV, as is illustrated by Tax Evasion. The substantive reason for rejecting JV is provided by Frankfurt-style cases, e.g., Tax Evasion. People might find JV initially plausible, but Frankfurt-style cases provide what might sometimes be a surprising reason to reject it. My intuition in Fatalistic Tax Evasion is that Joe is morally responsible (supposing that Joe chooses by exercising his agent-causal free will) – I’d be interested to see how other incompatibilists would respond!
Posted by: Derk Pereboom | February 15, 2005 at 09:15 AM
Thanks, Derk! I was pretty sure you'd say this!
Posted by: Joe | February 15, 2005 at 10:16 AM
Derk,
Though my personality is starting to disintegrate, the incompatibilist in me thinks that in Fatalistic Tax evasion, Joe is free (if he's the agent-cause) but not morally responsible for his evasion, because he couldn't have avoided it.
Posted by: Scott Ragland | March 03, 2005 at 01:40 PM
In Fatalistic Tax Evasion, the device also senses if and when Joe has a moral reason with a specified force. But instead of stimulating the brain so that Joe chooses to evade taxes, the device merely wipes out Joe’s recent memories and effectively puts him back into a state where Joe begins his decision process over again. In other words, the device never forces Joe to choose to evade taxes it merely never allows him to choose not to evade taxes. The only situation allowed by the device is one in which Joe (eventually) chooses to evade taxes. Suppose that Joe does so choose. My guess is that folks with incompatibilist intuitions will also have the intuition that Joe is not morally responsible for his action, so it cannot be DV that is driving those intuitions.
Posted by: genuwine | April 08, 2006 at 12:24 PM