Is Incompatibilism Intuitive?

During the past year and a half, I have been working on a research project with Eddy Nahmias, Steve Morris, and Jason Turner. The project--which was an attempt to probe folk intuitions concerning free will and moral responsibility in an empirically informed manner--has produced three papers. The third paper--entitled "Is Incompatibilism Intuitive?"--has just been accepted for publication by Philosophy and Phenomenological Research. Since we talk a lot about the relevance of folk intuitions to the free will debate--I thought some of you may be interested in checking it out. Here's the abstract:

Incompatibilists believe free will is impossible if determinism is true, and they often claim that this view is supported by ordinary intuitions. We challenge the claim that incompatibilism is intuitive to most laypersons and discuss the significance of this challenge to the free will debate. After explaining why incompatibilists should want their view to accord with pretheoretical intuitions, we suggest that determining whether incompatibilism is in fact intuitive calls for empirical testing. We then present the results of our studies, which put significant pressure on the claim that incompatibilism is intuitive. Finally, we consider and respond to several potential objections to our approach.

Comments

[Neil]

Just a pointer to the debate on this paper happening over at the Garden of Forking Paths:

http://gfp.typepad.com/the_garden_of_forking_pat/2005/03/is_incompatibil.html

[Terrance Tomkow]

"Incompatibilists believe free will is impossible if determinism is true..." No they don't. They believe that free will is incompatible with determinism which is something quite different. If you exercise this sort of modal sloppiness in your questioning of "lay people" I suspect your results will be meaningless.

[Brendan]

Terrance,

If determinism is true, then we cannot have free will. That is, it cannot be the case that we both have free will, and our actions be determined. That is exactly what incompatibilists are committed to. They think determinism and free will are two "incompatible" theses: both cannot be true. Thus, free will is impossible if determinism is true.
If you think otherwise, you owe some sort of explanation of what the term 'incompatible' refers to.

[Terrance Tomkow]

"If determinism is true, then we cannot have free will. That is, it cannot be the case that we both have free will, and our actions be determined. That is exactly what incompatibilists are committed to. "

There are two claims here. a) "if determinism is true, then we cannot have free will." (b) "it cannot be the case that we both have free will and our actions be determined". (b) is indeed what incompatiblists are committed to. (a) is equivalent to the claim I was criticizing "Incompatibilists believe free will is impossible if determinism is true" and this is *not* what incompatibilsts are committed to. If you think these two claims are equivalent then you are conflating "P then N~Q" with "N~(P&Q)". These are not equivalent. Indeed this confusion is of a piece with the fatalist fallacy.

I return to my point. Before one condescends to ask "plain folks" what they think about philosophical issues (and the paper in question reeks of such condescension) one should oneself have a proper grasp of those issues.

[Neil]

Genuine question, Terrance. I see that "P then N~Q" and "N~(P&Q)" are not equivalent (in part because the second licenses the inference "Q then ~P" and the first doesn't). But I don't see why incompatibilism can't be described as the thesis that "Determinism therefore necessarily not freedom", as well as "necessarily, not determinism and freedom". Doesn't the second entail the first?

[tnadelhoffer]

Sorry I have been slow to join the frey--I have been out of town presenting a paper at the SSPP in Raleigh-Durham. In any event, I find Terrance's tone to be counter-productive as well as unnecessary. For instance, to say things like: "Before one condescends to ask "plain folks" what they think about philosophical issues (and the paper in question reeks of such condescension) one should oneself have a proper grasp of those issues." is both inappropriate--given our tone in the paper--and utterly unhelpful. As P.M.S Hacker once correctly pointed out, "burning straw men often creates more heat than light." In any event, I second Neil's request for some clarification from Terrance. Even if I concede that there is obviously a difference between "P then N~Q" and "N~(P&Q)"--I nevertheless fail to see why the former is not an adequate characterization of incompatibilism. I suppose I would appreciate it if Terrance would specify how he would test the claim that people are naturally incompatibilists. More specifically, I would like for Terrance to specify precisely what is wrong with our argument in the paper that the prediction P is a proper characterization of incompatibilism. Ultimately, I think that it would be more helpful for Terrance to criticize the actual studies we ran rather than picking one sentence out of two lengthy papers to suggest that we are modally sloppy. Better yet, if Terrance is convinced that our studies produce meaningless results, perhaps he could suggest what studies we should have run--since he obviously feels he has better insight into how this sort of thing should be carried out (preferably without condescending).

[Brendan]

I would think that it would be more accurate to represent what I said as (a)(P then ~Q) and (b)~(P & Q) without modal operators. Then as (a*) N(P then ~Q) and (b*) N~(P & Q). Why think that when I add N to (a) I get (P then N~Q) and not (a*)?
(a) and (b) are logically equivalent. If you use DeMorgan's rule on (b) we get:

(c) (~P or ~Q)

Then by material implication we get

(a) P then ~Q

I am not the best with modal logic(I skipped that unit in logic class), but I imagine the modal versions (a*) and (b*) would be logically equivalent as well. Also, I think it is entirely vague given my earlier post where I was placing the modal operator. I should have been more clear. But given my above discussion, I would not translate it as it has been by Terrance.

[Terrance Tomkow]

"I don't see why incompatibilism can't be described as the thesis that "Determinism therefore necessarily not freedom", as well as "necessarily, not determinism and freedom". Doesn't the second entail the first?"

No! N~(P &Q) does NOT entail (P then N~Q). If it did then the following inference would be valid:
N~(P&~P) therefore (P then N~P)
If this were valid then one could infer
P therefore N~P
That is, it would follow that every true proposition was (metaphysically) necessary (whether or not determinism is true). This is the classical form of the Fatalist argument.

This is not just logically niggling. Getting straight the difference between Fatalism and Incompatiblism is essential to thinking clearly on either topic.

I would also have thought that clarity on this score was essential to sorting out "folk intuitions" about freedom. We can be sure that Fatalism has considerable intuitive appeal to "folks". We can be sure, not because we've given questionnaires to a few dozen undergraduates, but because the idea recurs in human history from earliest times and has on some occasions played a central role in historical events (cf. martin luther and all that).

[Jonathan Weinberg]

It is perhaps worth noting here, in support of Brendan's last comment, that ordinary English just does not parcel out its modalities very cleanly. E.g., people all the time say things like, "If we don't get to the store tonight, it will be impossible to cook tuna surprise for dinner" -- but when they say that, they clearly aren't committed to the silly claim
~S --> N(~T)

The original response still holds, that unless the formalism reveals, say, a real ambiguity in the experimental materials, or an alternative interpretation of the data, then all the formal stuff is basically irrelevant.

[Terrance Tomkow]

"Ultimately, I think that it would be more helpful for Terrance to criticize the actual studies we ran rather than picking one sentence out of two lengthy papers to suggest that we are modally sloppy. "

Note that the "one sentence" I am "picking on" is the *first* sentence of both your abstract and your paper. And if you still don't see what is wrong with describing compatiblism as the thesis that:
if Determinism then N~(Free Will)
you need to join Brendan and go back to that modal logic class.

"if Terrance is convinced that our studies produce meaningless results, perhaps he could suggest what studies we should have run--since he obviously feels he has better insight into how this sort of thing should be carried out (preferably without condescending)."

I'll do my best. The experimental methodology seems to me hopelessly naive. To give one example: If you ask a large number of people "If you discovered your parents didn't love you, would you be unhappy?" a sizable percentage will answer "No.". If you ask them to explain their answer they will say "Because I know my parents *do* love me!" . Trying to make sense of these sorts apparently "illogical" responses won Khaneman a Nobel prize. There is a large and mostly respectable empirical literature in psychology and economics on the vagaries of "common sense" logic and it should warn anyone off thinking that asking logically complex questions such as yours is likely to produce perspicuous results.

At the very least I suggest that you should ask a control question which tests whether or not your subjects believe that determinism is *possible*. E.g. you might ask them if they thought it was possible that a computer could predict their choices. Otherwise their responses are as meaningless as yours might be to the question "If triangles had four sides would pigs fly?". I predict that you will find that many of those your study counts as compatibilist will insist that no one could possibly predict *their* behavior.

Another obvious shortcoming of your method is that you seem to assume that only incompatiblism could lead your subjects to deny that choices were free. Some of your subjects may be fatalists and would deny that any choice was free whether or not the scenario was deterministic. In that case your results overcounting the number of incompatibilists.

[Justin]

Right now, there are two candidates for how to parse the natural language sentence (S):

(S) "if determinism is true, then we cannot have free will"

Terrance's way: P then (N~Q)
Brendan's way: N(P then ~Q)

Brendan's way strikes me as the much (much!) more natural parsing of (S). On Terrance's parsing, (S) asserts that if determinism is true in the actual world, then there's no possible world -- not even indeterministic worlds! -- where we have free will. But surely, that's not what (S) naturally means. Surely, someone who asserts (S) can consistently hold the following: determinism is true in the actual world and so we don't actually have free will, but there are other possible worlds where determinism is false and we do have free will. This claim is inconsistent with (S) if Terrance's parsing is right, but not if Brendan's is. So Brendan's gets things right and Terrance's gets things wrong.

Worse, on Terrance's parsing, (S) is true in the following scenario: there are non-actual possible worlds in which both determinism is true and we have free will, but the actual world happens to be a place where determinism is false. But surely, the natural reading of (S) *shouldn't* be consistent with this scenario. On the most natural reading, you can't consistently say "if determinism is true then we cannot have free will, but, you know, it's possible that there is determinism and free will" -- that is, on the natural reading of (S) you can't consistently first assert (S) and then also assert that there are possible worlds where determinism is true and we have free will. But, if Terrance's parsing of (S) were right, we *should* be able to consistently say this. So much the worse for Terrance's parsing. Brendan's way gets things right -- if N(P then ~Q), then not possibly (P & Q).

As Brendan suggests, his parsing is lgically equivalent to N~(P & Q), since (P then ~Q) is logically equivalent to ~(P & Q). Terrance may have deeper methodological concerns about the work, but I doubt his worries about modal sloppiness are worth taking too seriously.

[Eddy Nahmias]

Boy, there are a lot more interesting things to talk about in our paper than this potential modal fallacy! The sentence Terrance is picking on comes from the abstract not the paper. The first sentence of the paper could also be read in the improper way he is suggesting, but there's no reason to read it that way, and obviously no incompatibilist nor we think that is the right way to read it. More importantly, none of our arguments or studies indicate we are attributing this modal view to incompatibilists or that we hold it.

The more relevant discussion comes in footnote 4 where we cite an important claim defended by (incompatibilist) Fritz Warfield:

4. See Warfield (2000) for an explanation of why the proper incompatibilist view 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” (169).

This point is too often missed in the debate, and we suggest it makes the incompatibilist view more metaphysically contentious and hence take on the burden of proof. This itself is a contentious claim (see pp. 2-4 of paper) and one I would like to see discussed.

I hope we can get off this issue and on to more relevant ones.

[tnadelhoffer]

Terrance says:
"At the very least I suggest that you should ask a control question which tests whether or not your subjects believe that determinism is *possible*. E.g. you might ask them if they thought it was possible that a computer could predict their choices. Otherwise their responses are as meaningless as yours might be to the question "If triangles had four sides would pigs fly?". I predict that you will find that many of those your study counts as compatibilist will insist that no one could possibly predict *their* behavior."

We did ask a control question. And after students answered yes/no to whether they thought the world we live in is like the one in the scenario, we reminded them that for present purposes they were to assume for the sake of argument that the world really is the way the scenario says it is. We also had a manipulation check on the back to make sure that participants were picking up on the salient issue. We threw out responses from all those who missed the manipulation check.

In any event, I wanted to thank Justin, who has succinctly shown that perhaps Terrance needs to worry about his own modal sloppiness--at least as far as the free will debate is concerned. More importantly, I now wonder whether Terrance actually read the paper. It seems like perhaps he read the abstract (and the first sentence of the paper--which is *not* the same as the sentence in the abstract!) and decided that our project was not worthwhile. Minimally, it is evident that he has not read our other paper that is forthcoming in Philosophical Psychology--where we address a number of the methodological issues that arose during the course of our investigation.

Terrance also made the following point (yet again in a tone that suggests that he has a chip on his shoulder):

"Another obvious shortcoming of your method is that you seem to assume that only incompatiblism could lead your subjects to deny that choices were free. Some of your subjects may be fatalists and would deny that any choice was free whether or not the scenario was deterministic. In that case your results overcounting the number of incompatibilists."

This remark suggests that Terrance has entirely missed the point of the studies. After all, our goal was to see whether ordinary people would judge that an agent is free and morally responsible in a determinisitic scenario. Many incompatibilists talk as if we should predict that people will judge that an agent in a deterministic scenario is unfree and not morally responsible. We tested (and seemingly falsified) this prediction. Terrance responds by pointing out that we may have been double-dipping when it comes to the responses by fatalists. Even if this were true, wouldn't this just provide more evidence for our conclusion--namely, that if P is a fair characterization of incompatibilism, incompatibilists ought to give P up since it is false. If we really were inappropriately counting fatalistic responses twice--which I will concede for the sake of argument--this would mean that there were even fewer incompatibilists among the participants than our results suggest. Of course, this would strengthen (and not weaken!) our argument about the incompatibilist prediction P.

[Brendan]

Eddy,

Warfield seems right about what incompatibilists are committed to.
I have been wondering what exactly what sort of modal claims materialists are committed to. It seems in classic modal arguments, the structure is something like

(1) M then N(M)
(2) P~M
(3) ~M

What seems interesting is that we seem to want (3) to be modal, and the only way of making it so that seems to make sense is,

(3*) N~M

But this assumes,

(0)~M then N~M

What I wonder about is whether or not materialism is committed to the following (which is just (0) AND (1)):

(M)If materialism is true, then it is necessarily true, and if materialism is false, then it is necessarily false.

Because if the above translation of a modal argument is correct, it seems that (M) is what materialists have to work with. (M) seems like a very strong claim, and I am not sure it is what I think of when I think materialism is true.
My intuition is that materialism is necessarily true, but that is something that has to be proven. I certainly do not think materialism is necessarily true qua materialism.

Sorry if this seems a little off topic.

[Terrance Tomkow]

"If we really were inappropriately counting fatalistic responses twice--which I will concede for the sake of argument--this would mean that there were even fewer incompatibilists among the participants than our results suggest. Of course, this would strengthen (and not weaken!) our argument about the incompatibilist prediction P. "

I *told* you I was going to be helpful. I shall expect a gracious footnote.