Campbell on the Consequence Argument

I just got the most recent issue of Analysis in the mail today, and very much enjoyed reading an article by our very own Joe Campbell, "Free Will and the Necessity of the Past".  A very nice piece of work, Joe.  For those interested, I'll summarize and discuss the piece a bit below the fold.  (But you should really go read it, too!)

Joe argues that the "necessity of the past" premise that features prominently in van Inwagen's formal presentation of the Consequence Argument is either unsupported or, if it is supported, its support renders the conclusion of the argument weaker than is usually supposed.  That premise is as follows (where the claim 'Np' is read as 'p and no one has, or ever had, any choice about whether p' and 'p0' is some true proposition about the remote past):

Necessity of the past premise: Np0

Joe says that van Inwagen gives us two reasons to think that Np0 is true.  First, we might think it's true simply because "no one can change the past".  Second, we might think it is true because it is a statement about the remote past (where 'remote'  here designates some time before there were humans).  Whereas the first reason does not support the premise, the second reason renders the conclusion of the argument weaker than is usually supposed.

First, Joe points out that from the claim that no one can change the past, we're never going to be able to infer that no one has, or ever had, and choice about some true proposition about the past.  For consider some proposition about the not-so-remote past that I arguably did have a choice about: whether to drink that one night that I got drunk and ran over someone on the way home.  Sure, it's true now that I can't change the fact that I drank that night.  But it doesn't follow from the fact that I can't change it that I never could change it.  Intuitively, I could have done something about it that night -- I could have chosen not to drink.

Of course, van Inwagen is talking about a proposition about the remote past, before any humans were born.  So perhaps it is the "remoteness", as Joe calls it, rather than the "pastness" that gives support for Np0.  But if that's right, then Joe points out that the conclusion of the Consequence Argument will no longer be full-blooded incompatibilism.  Full-blooded incompatibilism is that there is no deterministic world with free agents, but the most the "remoteness" defense can get us is that there is no deterministic world that also has a remote past with free agents.  (Of course, since our world does have a remote past, the argument still gets us the conclusion that if determinism is true of our world, our world doesn't contain free agents, but as Joe points out, this is less than full-blooded incompatibilism.)

I think Joe's argument is fascinating and, it seems to me, right on the money.  Consider a world according to which every agent exists at every instant of time.  At this world, there will be no remote past, and no remote future, either for human beings in general or even for any individual human being.  (Where a 'remote' past or future for an individual human being is just some time at which that individual doesn't exist.)  And further suppose that this world is deterministic -- that the state of the world at any one time together with the laws entails the state of the world at any other time.  Despite the fact that determinism is true at this world, we can't run the Consequence Argument.

This is not to say, of course, that the people in this world are free agents.  But we'll need to resort to some other argument in order to show that they aren't, if they aren't.  I'm actually curious as to what our resident incompatibilists think of this world.  Need an incompatibilist be a full-blooded incompatibilist?  What's lost if the incompatibilist admits that there are deterministic worlds with free agents, so long as these worlds are only the ones where each agent exists for every instant of time?

Comments

[Cihan Baran]

Hey Neal,

Interesting post. It reminds me of the question I asked Kip few days ago: If there was agent-causation that was deterministic, would you think there'd be moral responsibility? (He called the very idea "non-sensical monstrosity"(haha) but I am hoping the question is not such a dumb one.)

The worry, a bit like Kant's antinomies, I have is that if agents sprung to existence out of nothing, then there was some causation that caused them to spring into existence and this serves as the distant past. If there was no such cause, it seems as if the agents in this universe caused themselves to exist - and hence we seem to have agent-causation.

So I am not very comfortable with the idea of "a world according to which every agent exists at every instant of time". It's not so much a "remote past" that I worry about but a "remote cause" that's effective prior to my existence.

[Eddy Nahmias]

Ned Markosian has an interesting published essay (and an interesting unpublished essay) arguing that agent causation and determinism are entirely compatible. I think he has certainly succeeded in showing that the idea is *not* a "non-sensical monstrosity." Indeed, I believe he has offered a reason to think that it's not determinism that is threatening to free will but something else (perhaps a reductionistic picture of some sort) that suggests we agents (and our reasons) are not properly causal... Oh, there I go again!

Joe's very interesting paper also suggests that it cannot be determinism *alone* that is incompatible with free will. Combine Joe's argument with Fritz Warfield's claim that the proper incompatibilist position is not the contingent claim, “If determinism is true then there is no freedom,” but the stronger claim, “Necessarily, if determinism is true then there is no freedom” (2000: 169). If Joe is right, then the latter claim is false. The incompatibilist would have to argue, instead, for the conclusion: "Necessarily, if determinism is true in a universe with a remote past, then there is no freedom."

Perhaps this addition makes no difference, but now suppose Markosian is right. Then the proper incompatibilist conclusion would be: "Necessarily, if determinism is true AND there is a remote past AND there is no agent causation, THEN there is no freedom." Perhaps this addition makes no interesting difference either, but it sure makes the incompatibilist position take longer to to say or write ...

[Joe]

Thanks a lot for the post, Neal! And thanks to Eddy for noting Markosian's work on compatibilist agent causation, which I've always like and which I think is very much underappreciated by contemporary philosophers. Markosian's original paper is called "A Compatibilist Version of the Theory of Agent Causation," Pacific Philosophical Quarterly 80 (1999), pp. 257-77. He has a sequel, which he presented at the Bled Conference this past summer. Also, Michael Nelson offers a similar theory in "Breakin' the Law," which he presented at INPC 2006. They both borrow liberally from Randy Clarke's work.

The response by Cihan is interesting but in the end it won't work. As I indicate in a footnote in the paper, following a suggestion by Jason Turner, it is easier if one imagines a world in which time is circular. Suppose that in that world there is a man, oscillating Adam, who is eternal. Adam has always existed and always will exist. He spends his time growing ‘old’ and then getting ‘younger.’ Adam always ‘begins’ with powers and capacities comparable to a 25-year-old man and then develops powers and capacities comparable to a 50-year-old man at such time he regresses back to the state at which he ‘began.’ Adam is in the grips of an oscillating eternal recurrence: for Adam, there was no explicit beginning and there will be no end. Hard to see what reason we have for thinking that there is a remote cause in the works here. Nor can the infinite regress worries that Galen Strawson notes in the basic argument crop up.

The problem, as I now see it, is that arguments for incompatibilism require two things: a grounding principle, which establishes a lack of freedom or control, and a transfer principle, which transfers that lack of freedom or control to all of our other actions, given the thesis of determinism. It is easy to come up with a grounding principle given contingent features of the actual world, and p0 is one example. Yet p0, even when joined with a proposition expressing the conjunction of the laws of nature, cannot establish full-blooded incompatibilism.

Which brings us back to Neal's questions: Need an incompatibilist be a full-blooded incompatibilist? What's lost if the incompatibilist admits that there are deterministic worlds with free agents, so long as these worlds are only the ones where each agent exists for every instant of time?

My own answers are 'Yes' and 'A lot,' respectively, for full-blooded incompatibilism just is incompatibilism.

[Justin Capes]

Neal,

I think Fritz is right about what the *real* incompatibilist conclusion is. However, as an incompatibilist myself, I'm inclined to think that less-than-full-blooded incompatibilism is a very significant position. I'm perfectly happy with incompatibilism-or-something-close-enough. After all, isn't what we care about most OUR world, or at least ones relevantly similar to it? Is there reason to care about worlds that are very unlike the actual world? I'm not as sure about this; but if determinism does in fact rule out freedom in this world, this seems to me an interesting and substantive conclusion, even if it is not equivalent to the strict incompatibility claim.

[Derk Pereboom]

Joe – do you think that this manipulation argument might help the cause of full-blooded incompatibilism?

Imagine a divine manipulator, outside of time, who creates the circular-time world. She also creates Adam so that he exists at every time in that world, and so that his every state causally inevitable given the past and the deterministic laws of that world. In so doing, the divine manipulator, by virtue of these creative acts, renders fixed each of Adam’s states. Suppose Adam eats the forbidden fruit, and all the prominent compatibilist conditions are in place. It would seem that the divine manipulator would be wrong to claim Adam is blameworthy for eating the forbidden fruit.

Now imagine everything about this scenario is the same, except there is no divine manipulator, and the deterministic circular time world is not the result of any cause. This world, with Adam in it, just exists and was never created. (Still, it’s forbidden to eat that fruit.) Is there a difference between this scenario and the one with the divine manipulator that would justify the claim that Adam is now responsible in this one? It seems that there isn’t.

[Joe]

Derk,

Manipulation arguments are tricky. I'm not sure what to say about them.

Initially, I agree that the two cases are similar in all relevant respects. In other words, Adam is blameworthy for eating the forbidden fruit in the first case iff he is blameworthy for eating the fruit in the second case.

Why think that Adam is not blameworthy in the first case? We are told that "his every state causally inevitable given the past and the deterministic laws." I am a compatibilist, so this is irrelevant to me. We are also told "the divine manipulator, by virtue of these creative acts, renders fixed each of Adam’s states." I'm not sure how this adds anything significant to what we already know. In virtue of being a compatibilist, I think that Adam’s freedom is not settled by whether or not his action is "causally inevitable." Knowing the causal inevitabilities certainly puts the divine manipulator in a good position to be able to "render fixed each of Adam's states." Thus, if determinism is true, then manipulators might have an advantage and I shouldn’t think that manipulation alone is a relevant factor. The devil is in the details.

I’d like to change the example to one that is more familiar: Even if Bill's affair with Monica were "causally inevitable" given the "manipulation" of some agency like the Republican Party it would not, by virtue of these facts alone, follow that Bill was not blameworthy. We would need to know more about the situation: Is there a significant range of situations in which Bill might not have performed the act? Is the act something that Bill clearly did, as opposed to something that was done to him? Did he have the requisite abilities and capacities while performing it? If the answers are 'yes,' 'yes,' and 'yes,' then I might consider Bill to be blameworthy for his action.

Just noting that the Republican's set him up and that the action was manipulated and inevitable is not enough to get Bill off the hook. What matters are the explanatory details. The same holds for Adam. I need to hear more about his situation before I judge him.

Justin,

Let’s distinguish between the Traditional Problem -- whether or not we have free will -- and the Compatibility Problem -- whether or not free will is compatible with determinism. By no means am I suggesting that the Analysis paper makes significant headway on the Traditional Problem. (Though I have another paper that tries to close the gap.) The conclusion of the Analysis paper is not 'we have free will' or 'it is more likely than not that we have free will.'

In terms of the Compatibility Problem, however, my findings are significant. I wouldn’t say that they are more significant than Warfield's, for in many ways my aim was to shed light on that paper. But Warfield’s concession is quite a concession, much more of a concession than you seem to admit.

Pereboom’s strategy hinges on the comparative similarity between manipulation cases and cases of causal determinism. But my results show that we can just as well go in the other direction. According to the ‘real incompatibilist,’ Adam is not blameworthy for his actions yet a very similar guy who performed the same actions but who happened not to be eternal would (apparently) be blameworthy for at least some of his actions. What is the relevant difference? This is the question that haunts the ‘real incompatibilist.’

The 'real incompatibilist' alters the definition of ‘incompatibilism’ in a significant way. She is admitting that it isn't determinism alone but determinism given certain contingent features that is incompatible with determinism. As Eddy suggests, we are a long way from identifying these additional contingencies and saying that they are more in keeping with the incompatibilist spirit than the compatibilist spirit is presumptuous.

[Justin Capes]

Joe,

What do you think about this? Let us suppose, for the sake of argument, that you are right, that the consequence argument does not prove

INC: Nec, if determinism is true, no one has free will

but only

WEAK: If determinism is true, no one has free will.

I agree that this would be an interesting conclusion with respect to the compatibility question. However, the reason the compatibility question is important (as I understand the dialectical situation) is that a compatibilist answer is supposed to help us with the Traditional Problem. But even if incompatibilists have failed to establish INC (with the consequence argument), as you claim, they still seem to have established WEAK, and the later seems just as troublesome when it comes to the Traditional Problem as the former, at least for denizens of worlds like ours. Compatibilism, in this case, doesn't help. If WEAK is true, then no one acts freely (in the actual world) if determinism is true, and this is so whether or not it is determinism per se that precludes free will. This sort of compatibilism seems to me to somewhat trivialize the compatibility question.

Moreover, suppose the advocate of the "incompatibilism-or-something-close-enough" view agrees that in worlds very unlike our own (e.g. worlds with no remote past)free will is compatible with determinism. That person might well say something like this: "great, compatibilism is true, but determinism still must be false if I (or my counterparts in relevantly similar worlds) am to be free (and morally responsible). So, Consequence style arguments still seem to raise the same sorts of difficulties for compatibilist believers in free will (and moral responsibility), even if they don't establish INC.

[Joe]

Justin,

You raise some good points.

For now, let me just ask: What is wrong with my version of the Pereboom argument? What is the relevant difference between actual world Adam and oscillating Adam? Why is it that you can prove one to be unfree yet not the other?

[Cihan Baran]

Joe,

I am still not convinced, so let me tell you my worries. See when you have funny things happening like circular time, it's best not to use words like "prior" or "before", since it is ambiguous what these words mean if they mean anything at all.

So perhaps, here is a better distinction. If the agent acts the way she does due to "extrinsic" causes - causes that don't originate in her -, she is not morally responsible.

Now back to your temporally circular world. Who/what caused this world? If it wasn't the agent, then the agent can not be morally responsible.

If it was the agent, then the agent has magical powers of self-creation. And if you have such magical powers, yes, you are indeed morally responsible.

I am convinced by Frankfurt-style cases that it is not merely causal determinism or pre-determination that rules out moral responsibility. However, I just haven't been able to find such a possible world. (This was my motivation in asking Kip that question.)

So suppose that there is a god and then this god endows us with Cartesian-souls that have magical powers of self-causation. Yet, due to god's foreknowledge, everything we do is determined. I have the intuition, from Frankfurt-style cases and what not, that we would be responsible in such a world.

Note that my favorite physicist Schrödinger once said "At the price of mystery, you can purchase anything". Though, Russell added "it won't be by honest toil but by theft." (This is quoted in Kane's introductory Free Will book.)

[Joe]

Cihan,

I'm not sure what the mystery is that you are talking about. I'm not suggesting that the key to free will is circular time. The circular time model is offered to illustrate the incompleteness of the Consequence Argument as an argument for incompatibilism. Here we have determinism but no clear way to show that Adam lacks free will.

You ask: "Who/what caused this world? If it wasn't the agent, then the agent can not be morally responsible." But why think this? What general principle are you resting this claim on? A theory of free will which requires that one be the cause of the world in order to be morally responsible for his actions is far more mysterious than anything that I can come up with!

[Cihan Baran]

Joe,

I didn't mean to call your views "mysterious". (Although if one has to postulate "circular time" to get moral responsibility, that seems a little metaphysically suspect.) I just don't think that mere causal determinism rules out free will but when I try to come up with a possible world that's determined and has moral responsibility, I come up with mysterious worlds.

[Joe]

Cihan,

I hope you don't think that I was offended in any way! It is just that most compatibilists believe that they offer a less mysterious account of free will than incompatibilists. No doubt we might be wrong about that!

You're thinking of my paper in the wrong way, though. You act as if incompatibilism is the default position and folks like myself must explain how on earth compatibilism is possible. As an explanation of the possibility of compatibilist free will, the introduction of circular time would be a bit odd. But I'm not explaining compatibilism. Were I to give a positive account of what free will is I would not, at any point, make reference to worlds with circular time.

I'm not explaining compatibilism. I'm merely responding to an argument for incompatibilism. Suppose I ask you, Why on earth do you think that free will is incompatible with determinism? No doubt you would offer some reason. For many, those reasons are similar to van Inwagen's Consequence Argument. I've shown that whatever that argument does, it doesn't support the thesis of incompatibilism, strickly speaking.

You have two choices: revise your view to something like WEAK or base your view on some other argument, like Pereboom's Four-Case Argument (which is a kind of manipulation argument). (You could also find some flaw in my argument or say that belief in free will is primative and not held on the basis of argument.)

If you go the WEAK route, you are faced with a new challenge, as I noted above. You must explain what is significant about having a remote past? How is it that I can easily be shown to be unfree, if determinism is true, yet Adam cannot be shown to be unfree? There doesn't appear to be any relevant difference between the cases, as you note. Then why is it that your reasons for accepting incompatibilism do not have the result that Adam is unfree?

As I also suggested above the compatibilist now has an anti-manipulation argument at his disposal: There is no reason to think that Adam lacks free will but there is no relevant difference between Adam and someone like Adam who has a remote past, so there is no reason to think that a person in a determined world with a remote past is unfree -- contrary to what the Consequence Argument suggests.

You would respond by noting that it goes both ways: You, in fact, are trying to point out to me that you're not convinced of my argument BECAUSE of this same similarity. At this stage, I'm willing to accept that we're in a dialectical stalemate. But I don't see a clear advantage for the incompatibilist.

[Justin Capes]

Joe,

You say, " if you go the WEAK route, you are faced with a new challenge, as I noted above. You must explain what is significant about having a remote past?" But why think that? Suppose I accept WEAK on the basis of the consequence argument. Why can't I just say, "I don't know what's significant about having a remote past, but I know that if a world has one and determinism is true, then no one does anything freely." I can *know* that WEAK is true on the basis of the argument (if it is indeed sound) without knowing why having a remote past is important to that outcome. I certainly have a question to answer that the compatibilist doesn't, but that wouldn't undermine my confidence in WEAK--afterall, the consequence argument is a very powerful argument, even as an argument for WEAK.

Your response to the manipulation case strikes me as leading to a dialectical stalemate at best, though my own intuitions go with Pereboom's in that if there is indeed no relevant differences between the two cases, then Adam can't be free (or morally responsible). It is simply too (way too) counterintuitive to accept a view of free will with the consequence that I could be manipulated the way Adam is in Derk's example and still be free. But, of course, this won't move a compatibilist, who apparently does not share my intuitive convictions on this matter.

One last thing. In the world without a manipulator, there is nothing to prevent Adam from having buck-stopping agent-causal powers is there? If not, then perhaps that's the difference.

[Joe]

Justin,

There are some good points here, again. I don’t want to dismiss them but I disagree. Sorry to all for the long post!

You ask, in response to my challenge to the proponent of WEAK, “Why can’t I just say, ‘I don’t know what’s significant about having a remote past, but I know that if a world has one and determinism is true, then no one does anything freely’.”

Two things. First, my paper is one of a handful of responses to arguments for incompatibilism. Here are some others:

G.E. Moore, Ethics, Ch. 6 (1912)
Keith Lehrer, “Preferences, Conditionals and Freedom.” In van Inwagen (ed.), Time and Cause (D. Reidel, 1980)
David Lewis, “Are We Free to Break the Laws?” Theoria 47: 113-21 (1981)
Jonathan Westphal, “A New Way With the Consequence Argument, and the Fixity of Laws.” Analysis 63: 208ff. (2003)
John Perry, “Compatibilist Options.” In Campbell, O’Rourke, and Shier (eds.) Freedom and Determinism (MIT, 2004)

I understand that you think that the argument in support of WEAK is a good one but I think that it is a bad one. I admit its role in the history of philosophy and glorify it for that reason alone. I am not joking here! In this respect, it is a wonderful, important, ingenuous argument. I’ve even confessed that I’m obsessed with it. I just think it is unsound. Yet responding to the argument is no easy task.

My guess is that most proponents of the Consequence Argument have read only the Lewis paper; it would be rare to find someone who has read at least three of these articles and nearly impossible to find one who has read them all.

In short, I do not think that the Consequence Argument is as good as you think that it is. As I’ve said before, I have never thought that the argument was good. The argument, in my mind, does little more than pump a few widely held intuitions about free will, intuitions that if left unchecked will eventually lead one to free will skepticism. That alone provides a pretty good reductio to the argument, in my mind.

Second, I don’t see how you can say, on the one hand, “I can *know* that WEAK is true on the basis of the argument (if it is indeed sound) without knowing why having a remote past is important to that outcome” yet say, on the other hand, “if there is indeed no relevant differences between the two cases, then Adam can't be free (or morally responsible).” Why can’t I know that there is no reason to deny that Adam is free in the one case and then infer that there is no reason to deny that persons in determined worlds with remote pasts are free? Why can’t I say: “It is way too counterintuitive to accept an argument for incompatibilism that can’t even prove that incompatibilism is true”?

Again, I don’t see a relevant difference here.

[Cihan Baran]

Joe,

Here is the relevant difference.

I am not so sure of this but agents in the circular-time world seem to be the sources of their actions. (I am not sure because I don't know if the agent in your paper self-caused himself/herself to existence. I don't know if I can make sense of the notion of circular-time but just for the sake of argument...)

Agents in other deterministic worlds are not the sources of their actions.