Brueckner on Campbell & Campbell on that

Back in April we had a nice discussion here at the Garden of Joe Campbell's Analysis piece, "Free Will and the Necessity of the Past". Well, the discussion continues.

Anthony Brueckner has a reply to Campbell forthcoming in Analysis called "Retooling the Consequence Argument", which he has been kind enough to let us post here.

Also, Joe Campbell has drafted a reply to Brueckner's forthcoming reply that he has been kind enough to let us post here.

Enjoy!

Comments

[Joe Campbell]

Wow! Campbell's reply is really good!

(I'm just trying to get something started.)

[Mark Smeltzer]

Are you planning to publish your response? I'm curious about the endnote that says it was written for the GFP...

That aside, my impression was that you pretty much demolished his argument. So, there's not much to say until someone disagrees ;)

[Joe Campbell]

I sent the response into Analysis but that was just a few days ago. I'm not as confident about it as you are, so I'm interested to see what other folks think. I like Tony's paper a lot.

[Tony Brueckner]

I have some comments on the paper that Joe posted here.

At first, I couldn't quite see what was bothering Joe, but, with the help of Chris Buford, I think that I get it.

Mirroring van Inwagen's Third Argument, NTA attempts to establish a conclusion concerning a specific proposition concerning a specific action--viz. the proposition P=The judge does not raise his arm at t. NTA's conclusion is N*P,t, i.e., no one (including the judge) has any choice at t about P. NTA can be generalized, so to speak, along two dimensions. Just as was the case with VI's argument, we can formulate a version of NTA that applies to any human action you choose (just as the standard argument for skepticism about knowledge of whether I have hands has versions which apply to just about any proposition about the external world that I claim to know). So using various instantiations of the NTA-style argument, we can show that for any proposition to the effect that an agent S performs action A at t, S has no choice AT T about the proposition that S performs A at t (provided that there is a past relative to t, and of course assuming determinism as a background premise).

Joe, I think, raises the question: what about times earlier than t? Can we generalize NTA across times (in THAT second dimension of generalization)? For example, did the judge have a choice about P (which concerns time t) earlier than t? Joe seems to be saying that we cannot answer this question in the affirmative by appealing to the following principle: If S has no choice about P at t, then S has no choice about P at times earlier than t. Setting aside the assumption of determinism, Joe's Drunk Driver example appears to refute that principle: our DD has no choice at t about crashing his car at t, and yet it is plausible to suppose that S had a choice about crashing his car at t at an earlier time when he started pounding down shots of whiskey.

But NTA affords us the resources to make the generalization across times without any simple-minded appeal to the foregoing principle. Did the judge have a choice about P(=The judge does not raise his arm at t) at a time t* earlier than t? If there is a past relative to t*, we simply consider a time t**

Re Adam: I still think that it's OK to acquiesce to the result that NTA cannot show that, given determinism, Adam's first action is unfree. Why should determinism, viewed as a doctrine about how the PAST plus the laws necessitate the future, have any consequences regarding the freedom of an act performed at a time relative to which there IS NO PAST? (Also, as I said in a footnote in my paper, there is the possibility of using NTA with a backward-looking formulation of determinism.) Further, NTA does show that any subsequent action A of Adam's that is performed at a time t relative to which there is a past is such that Adam had no choice at t about performing A at t. Further, for any time t* earlier than t which is such that there is a past relative to t*, Adam had no choice at t* about performing A at the later time t. One weird feature of NTA is this: even though Adam had no choice at creation-time-plus-one-second about performing A at the later time t, NTA is unable to show that at creation time, Adam had no choice about performing A at t (because there is no past relative to creation time).

[Tony Brueckner]

Well, the penultimate paragraph in my previous post got lost in cyberspace. So I'll try to rewrite it:

But NTA affords us the resources to make the generalization across times without any simple-minded appeal to the foregoing principle. Did the judge have a choice about P(=The judge does not raise his arm at t) at a time t* earlier than t? If there is a past relative to t*, we simply consider a time t** earlier than t*. Given determinism, the conjunction of L and the proposition P** expressing the total state of the universe at t** necessitates P. So we just crank through another version of NTA, noting that N*P**,t*--no one has any choice at t* about P**, the total-state-of universe proposition regarding the earlier time t**. We then derive the conclusion that N*P,t*--no one, including the judge, has any choice at t* about the proposition that the judge does not raise his arm at the later time t.

[Joe Campbell]

Thanks for the comments, Tony!

One point I made in my original reply was that I'm suspicious about the kind of necessity involved in NTA. In short, I think that it is a kind of relative necessity and I question whether this kind of necessity is transferable.

In your post, you write: "So using various instantiations of the NTA-style argument, we can show that for any proposition to the effect that an agent S performs action A at t, S has no choice AT T about the proposition that S performs A at t ..." You distinguish this from the "foregoing principle" (FP): "If S has no choice about P at t, then S has no choice about P at times earlier than t."

Yet NTA is committed to these inferences: If S has no choice about P at t, then S has no choice about whether S at t (where S is a proposition describing the state of the world at t); if S has no choice about whether S at t and determinism is true, then S has no choice about whether P at t, for any proposition P.

But the relevant features of the relative past apply to all moments of time (other than the initial moment of time if there is one), so it follows that S has no choice about whether P at t', for any proposition P or any time t'. After all, this is the claim that the incompatibilist needs and it is supposed to follow given determinism and assumptions about the relative necessity of the past. Yet from this we can derive (FP) by conditional proof from the assumptions of NTA. (I leave the actual proof to the reader.)

Thus, I agree that "NTA affords us the resources to make the generalization across times without any simple-minded appeal to the foregoing principle." I only claim that given the assumption of determinism and the resources of NTA, (FP) easily follows. This is why Missed Flight is a problem for you.

I claimed that in order to prove that the free will thesis is false, you need to (1) establish that there is a lack of choice about some set of propositions and then (2) transfer that lack of choice onto all other propositions. In the case of the Third Argument, I showed that the case for (1) rested on a contingent assumption about the remote past, so incompatibilism doesn't follow.

You try to avoid this objection by dispensing with (1). I don't see how this gets off the ground. You can only transfer funds received. The best that this move gets is: our subsequent freedom is always contingent upon some previous freedom, given determinism. Yet why deny the initial freedom?

In closing, I cannot help but note that you are willing to concede that NTA does not prove incompatibilism. For over thirty years we were duped into thinking that some form of the Consequence Argument proved that incompatibilism is true yet now we are willing to admit we were duped yet somehow fail to appreciate the significance of the admission. Are we not philosophers? Doesn't the meaning of 'incompatibilism' matter? Doesn't it matter whether we have an argument for incompatibilism or for some related thesis?

If we were wrong for over thirty years about the soundness of the Third Argument, might it not be best to reflect on what additional assumptions are relevant and drop the endorsement of incompatibilism at least until the dust clears?

[Fritz]

Joe -- you write:

"If we were wrong for over thirty years about the soundness of the Third Argument, might it not be best to reflect on what additional assumptions are relevant and drop the endorsement of incompatibilism at least until the dust clears?"

What 30 year period do you have in mind as being the 30 year period during whick we thought the "Third Argument" (van Inwagen's modal version of the Consequence Argument right?) was sound?

Do you mean the 30 year period leading up to Widerker's Analysis piece on van Inwagen's Beta rule? or the 30 year period leading up to the Johnson/McKay Phil Topics paper showing, indirectly, that Beta is invalid? Or the 30 year period until Erik Carlson, among others, gave direct counterexamples to van Inwagen's Beta that built on the Johnson/McKay work? Or the 30 year period up to Lewis's "Are We Free to Break the Laws?", which didn't directly address the 3rd Argument but from which we can backwards engineer a quite serious challenge to that argumetn? Or the 30 year period leading to the modal fallacy charge against the 3rd Argument (contingent premise in conditional proof of strict conditional) from my Phil Perspectives paper? or...well, there are a lot of candidates here...though I won't try to exhaustively lsit the other natural candidates. -----

But do you mean the 30 year period before your recent Analysis paper? Surely you wouldn't suggest that until your recent piece we incompatibilists all thought van Inwagen's 3rd argument proved incompatibilism? That would require a "power over the past" (the power to erase the literature on the topic from the past 2+ decades!) that not even many compatibilists would buy into. :)

Some incompatibilists believe that a suitably modified "Consequence Argument" is a still a great argument for incompatibilism. Some incompatibilists disagree but endorse *other* arguments for incompatibilism (some direct, some indirect). Still others think that whether some version of the Consequence Argument is a strong argument depends on what it takes to be a "version" of the Consequence Argument, a matter that isn't so easily resolved.

But ok, what do you mean here? Perhaps I've missed something.

[Joe Campbell]

I'm interested in formal versions of the Consequence Argument (CA) of the sort noted by van Inwagen, which began in 1975. So the 30+ years refers to 1975-2007. It wasn't a typo. I meant
the CA, not the Third Argument which is just one formal version of the CA.

I don't think that any of the previous criticisms of formal versions of the CA were decisive. Van Inwagen responds adequately to the Johnson/McKay counterexample to Beta. Nor is Lewis's criticism decisive. (Well, personally I think it is but I recognize that it rests on several contentious, debatable claims.) Likewise with the other criticisms you note above.

I think that the argument in my Analysis paper -- which of course borrows from some of your own work -- is decisive in showing that formal versions of the CA cannot establish incompatibilism. I showed this to be the case wrt the Third Argument and the First Argument -- the Third Argument is not the only one that I mention in the paper. Granted you can still get something close: If determinism is true AND something else is true (say, there is a remote past), then no one has free will. But this thesis is NOT the thesis of incompatibilism. Can we at least admit that much?

My main point was that it took us 30+ years to note, decisively, that formal versions of the CA are incomplete, so we might want to be careful about which thesis or theses might be added to complete them. Some of these approaches are more favorable to incompatibilists but several of them are more favorable to compatibilists (I mentioned two of these in my original paper -- one by Eddy and one by Jason Turner).

[Tony Brueckner]

I have a couple of comments on Joe's most recent posting.

Re the principle Joe calls (FP): If S has no choice about P at t, then S has no choice about P at times earlier than t. FP considered entirely on its own seems implausible, given Joe's Drunk Driver case (DD has no choice about crashing as he crashes, and yet he seemed to have a choice about crashing earlier in the bar). But if you assume determinism, and you bring on stage the machinery of NTA, FP falls out as a consequence. This is because cranking through instances of NTA yields the following result:

(R) Given determinism, for any times t, t' relative to which there is a past: no one has a choice at t about what he does at t'.

The fact that FP falls out as a consequence of (R) does not seem to be a problem. It's just what an incompatibilist would expect.

As for being "duped" about whether a VI-style consequence argument can establish incompatibilism, I think that NTA comes pretty close (modulo, of course, objections to "No Power Over the Past", "No Power Over the Laws", and to Beta-like transfer principles). The only actions that are outside the scope of applications of NTA are those which are performed at a time relative to which there is NO past. As I said in the paper and in my earlier posting, this should come as no surprise, given that determinism is formulated as a thesis about how the PAST and the laws necessitate the future. Further, a backward-looking formulation of determinism can take care of such pesky apparently free actions as Adam's raising his arm at the first moment of time.

[Mark Smeltzer]

Tony,

Could you describe this "backward-looking" formulation of determinism? If it can wipe the floor with Joe's argument, more should probably be said than a fleeting remark...

[Tony Brueckner]

Mark: See footnote 2 of my posted paper for the reference to van Inwagen's formulation. The idea is that just as the past plus the laws necessitate a unique future, on forward looking determinism, the present (say) plus the laws necessitate a unique past.

[Mark Smeltzer]

Tony,

Are you aware of the many highly contentious mathematical assumptions required to make that assertion?

If so, which are you stipulating to be true in order to cover your bases? If not... well, that will speak for itself.

[Mark Smeltzer]

Tony,

I meant to add: even if the kind of "backward-looking" determinism you describe is a necessary consequence of determinism simpliciter (which is anything but proven), the argument would need to add the following premise to yield incompatibilism:

(R2) For any time t' and any proposition P, at t' no agents exist that have had a choice about P at any time prior to t'.

However, no one that believes in responsibility would accept that premise -- not even libertarians.

If R2 is rejected, R by itself poses no threat to compatibilism: R simply indicates that backwards causation is impossible. It says absolutely nothing about forward causation.

As stated here, R has another read where t' comes after t, but results in a bald denial of compatibilism, which compatibilists have no motivation to accept.

[Joe Campbell]

I don't want to deny the asymmetrical nature of determinism but it seems to have some bizarre consequences given the additional assumptions of NTA.

Suppose I'm sitting in an airport bar at time t waiting for an airplane. The plane leaves in one hour at time t* from a gate across from the bar. I order a beer at t. As it turns out, I do miss the plane. It seems to be the case that no one has a choice at t about whether I miss the plane at t*. Recall that for van Inwagen this means that there is nothing that anyone can do at t to render it false that I miss the plane at t*. Clearly, there is nothing I can do at t to, say, make it the case that I make the plane at t*. Again, given determinism and given that NTA is sound no one has a choice at t about whether any true proposition is true -- in particular I don't have a choice at t about whether I order a beer at t! (No I’m not making a mistake here; just read the comments above more closely.)

That is just crazy. I think it shows that there is something wrong with the assumptions behind NTA.

As I indicated in my reply to Tony, I think one thing that is wrong is the idea that the kind of relative necessity of the past (or the similar relative necessity of the future noted above) cannot be transferred onto the present. It makes sense to note that given that I didn't raise my hand yesterday there is nothing that I can do about it now. After all, the present time is remote relative to the past. This seems like a clear case of lost opportunity. But how can that lost opportunity be transferred to the present?

I asked this question in my reply to Tony but I don't think that he's answered it yet.

[Joe Campbell]

I should have said "one thing that is wrong is the idea that the kind of relative necessity of the past noted in NTA ... CAN be transferred onto the present."

[Tony Brueckner]

Re the recent posts by Mark and Joe:

I don't know whether a backward-looking version of determinism is true, and for that matter I don't know whether a forward- looking version is true either. The only reasons I brought it up in the footnote are these: VI suggested backward-looking determinism in the locus classicus of the Consequence Argument, and if that form of determinism is assumed to be true, then the action of Adam raising his arm at the first moment of time can be shown to be unfree by an NTA-style argument.

Joe's post ends with a question about transfer of powerlessness that I seem not to have answered. There is also the issue of RELATIVE necessity, which I have not directly addressed, in part because I didn't really get what Joe was worried about. So let me try to say something about these matters now. I didn't raise my arm yesterday, and as Joe says, there is nothing that I can do about it now. I have lost the opportunity to raise my arm yesterday. Joe then asks, How can that lost opportunity be transferred to the present? First of all, Joe is agreeing with the view that the past is necessary relative to now--no one now has any power over the past. Just as no one now has any power over what happened in the days of the dinosaurs, no one has any power over what happened yesterday. If that's what Joe means by "relative necessity", then he's talking about one of the basic notions in the original Consequence Argument and apparently not challenging the cogency of the notion. What of transfer? Well, again, we are talking about a basic notion of the Consequence Argument--Beta. In NTA, I have simply tweaked VI's Beta. My Beta* was just this: If no one has any choice about P at t, and P necessitates Q, then no one has any choice about Q at t. So if it is granted that no one now (me included) has any choice about the total state of the universe yesterday (the day that I lost the opportunity to raise my arm by not raising it) and about the laws, and if given determinism the past plus the laws necessitate that I am now typing, then no one (including me) has any choice now about my typing now. If Joe wishes to challenge Beta and Beta*, then he will be in good company. But that was not his objection to VI's 3rd Argument. The objection was rather that the argument assumes the existence of a remote past lacking humans. My NTA was designed to circumvent that objection. I would like to know exactly was is wrong with NTA and its generalization across times, which serves to generate (R).

[Joe Campbell]

Tony,

I don't think that relative necessity is "one of the basic notions in the original Consequence Argument." In the Third Argument P0 stands for a proposition expressing the state of the universe at some time t0, prior to the existence of any human beings. In the First Argument P0 stands for a proposition expressing the state of the universe prior to an individual J's birth. In the Third Argument, there is reason to believe that no one had a choice about whether P0 because there were no humans around to have such a choice. In the First Argument, J was not around to have a choice about whether P0. These are particular propositions that claim an absolute necessity that applies either to all persons at all times or a particular person at all times. A relative necessity that changes over time does not play a role in van Inwagen’s formal versions of the Consequence Argument (CA).

Here is a summary of my thoughts on the CA and your NTA. The former establishes an absolute necessity that is (for the sake of argument) transferable via Beta but the problem is that this absolute necessity cannot be established in all possible worlds, so it fails as an argument for incompatibilism. The latter establishes only a relative necessity, which is not transferable to all times. In the language of my response to your article, I’d say that the CA cannot get us the requisite grounding principle whereas NTA cannot get us the requisite transfer principle. Mind you, I’m still suspicious about Beta and, as you’ve admitted, even if NTA is sound, it doesn’t establish incompatibilism. Thus, versions of the criticism that I’ve leveled against the CA might apply to NTA and vice versa.

[Tony Brueckner]

Thanks, Joe. That clarifies things.

It seems that you are granting that no one has any choice now about past states of affairs--because they are past. And you have never raised any questions about our now having no power over the laws. So your objection to NTA seems to be that the transfer principle I use--Beta*--is invalid. It seems that you hold that we cannot, e.g., go from

N*P,now
N*(P-->Q),now

to

N*Q,now

How do the examples in your paper bear on the validity of Beta*? I don't quite see how Drunk Driver does. That's a case in which, setting aside determinism, it seems that I don't have a choice now about drunkenly crashing but I did have a choice earlier about drunkenly crashing.

Maybe you have something like this in mind. I deliberate about whether to go to the movies and decide at 9 AM to go later at noon. At noon I have no choice about having earlier decided to go to the movies, but, setting aside determinism, it seems that I do have a choice at noon about going to the movies. I can change my mind and override my earlier decision. So here our P is *I decide at 9 AM to go to the movies at noon*, our Q is *I go to the movies at noon*, and our t is noon. So we have N*P,noon and ~N*Q,noon. But we don't have N*(P-->Q),noon. So I guess I don't see how to use DD considerations to get a counterexample to Beta*.

I don't see why my current powerlessness about some past state of affairs S fails to transfer to, say, current powerlessness about some present state of a affairs S' when I am powerless about the connection between S and S'--I'm powerless about the fact that if S earlier obtained, then S' now obtains.

[Joe Campbell]

Dear Tony,

This is helpful. A few comments.

First, note that van Inwagen's 'N'-operator says 'no one has, or ever HAD, a choice about whether' whereas your 'N*'-operator says 'no one HAS a choice about whether.' I think that this is significant. Note that in my Drunk Driver example we tend to think that the agent is responsible because he HAD a choice even though he no longer HAS a choice at the time of the accident. What seems relevant about the past is not whether I have a choice now but whether I had a choice then.

Second, you are right to make the move you make. If I'm right, I should have a counterexample to Beta*. Well, I think that Missed Flight is a counterexample to Beta*. (It is a bit more complicated than this for I think that Missed Flight shows that something is wrong in NTA and now I'm admitting that Beta* seems to be what is wrong.)

I'm not so dull that I think that everyone will accept that Missed Flight is a counterexample to Beta*, so I have some explaining to do. Right now, I think either contextualism is true OR people are so much under the sway of incompatibilism that they fail to see clear counterexamples of Beta and Beta* when they are presented. Note that IF compatibilism is true, every free act is a counterexample to Beta. It isn't as if this issue is much different from the compatibilism/incompatibilism debate itself.

I acknowledge that neither option above is satisfactory but its been a bad day, so I'll have to leave it here for now.

Thanks, again, for your comments!