Cokely and Feltz on the Folk Containing Both Compatibilists and Incompatibilists

I'm surprised that this has not been mentioned (to my knowledge) at the Garden yet: the Experimental Philosophy blog has a post introducing new research by Cokely and Feltz on free will and folk intuitions.  They describe their main thesis thus:

"Our main finding is that there seem to be groups of people who express compatibilist or incompatibilist intuitions--hence, the folk are not compatibilists or incompatibilists. Moreover, contrary to Nahmias et al's tentative position, our data suggest the biggest group of folk are incompatibilists. Section 3 reports these new findings (sections 1 and 2 are background)."

Their conclusions seem intuitive to me.  But I need to read it again to digest how these new results square with the earlier findings by Knobe, Nahmias, and their colleagues.  I'm also very curious to see how these results might fit with the research that Nahmias, with others, is currently conducting.

The paper is here.
The Experimental Philosophy post is here.

Enjoy.

Comments

[Kip Werking]

I should add: at the end of their article Cokely and Feltz suggest that, if we identifies two different groups (compatibilists and incompatibilists, speculated to be about 25% and 75% of the population respectively), then we can give these groups personality tests, or other tests, and correlate their compatibilism/incompatibilist with other values.

I find this sort of research absolutely fascinating. And it suggests a good way to test error theories. For example, if the average SAT score of compatibilists is about 300 points higher than for incompatibilists, that doesn't bode well for incompatibilism. Of course, most values, like the color of one's hair, probably have no correlation whatsoever with the compatibilism answer. But others might.

Here's a question for other Gardeners: what values, if any, would you like to see tested to see if it correlates with compatibilism/incompatibilism? Here are my suggestions:

1. intelligence
2. depression / healthy mood
3. sensitivity to anger (the irrational influence of affect)
4. religious belief
5. creativity

[Kevin Timpe]

Kip,

I'm perplexed. Are you suggesting that the SAT is a good measure of intelligence? And are you suggesting that compatibilism correlates with higher intelligence?

Then again, maybe I'm just misunderstanding because I'm one of the stupid incompatibilists....

[Fritz Warfield]

Does anyone know where these researchers, or others they are following, got the idea that "complete causation" is a good way to express the thesis of determinism?

[Kip Werking]

Kevin:

1. I am not suggesting that compatibilists are more intelligent. That was just an example of the sort of claim that would be interesting to test. I'm an incompatibilist too, and so I wouldn't want that to be true either.

2. I am not an expert on intelligence. But I imagine that the SAT would be at least a rough measure of intelligence. The advantage of the SAT is that most people have already taken it. But if anyone objects, feel free to substitute the intelligence measure of your choice.

[tnadelhoffer]

Kip,

Thanks for posting a link--it was on my list of things to do today! A minor point: I think the paper's primary is Feltz rather than Cokely.

Kevin,

Adam and Edward are working on some new studies that suggest that it is the opposite--namely, compatibilists don't think things through as carefully as incompatibilists (or something along those lines). I, for one, look forward to seeing the results of the studies!

Fritz,

They used the vignettes adopted by Nichols and Knobe in their recent paper on free will. Indeed, as far as I can tell, that is the primary weakness of the paper. After all, I think the original Nichols and Knobe studies were problematic--which similarly weakens the results of the follow-up paper by Adam and Edward. By my lights, the original studies had two central flaws:

First, in the original studies that I ran with Eddy, Jason, and Steve, we probed people's intuitions about both free will and moral responsibility with vignettes involving bad, good, or morally neutral behavior on the part of the agent in the story. We found that people's intuitions about free will seemed were relatively consistent across the cases. Yet when Shaun and Joshua ran their own studies, they no longer asked participants about free will at all. They merely probed people's intuitions concerning moral responsibility and determinism. And while this might be a good way of testing something like Fischer's semi-compatibilism, it is not a good way of testing compatibilism with respect to free will and determinism. As far as I can tell, neither Shaun and Joshua nor Adam and Edward are justified in concluding anything at all about their participants' intuitions about free will. After all, they weren't asked about free will, they were asked about moral responsibility. In order to correct for this shortcoming, Adam, Edward, and I are in the process of re-running the Nichols and Knobe studies--except this time some participants will be asked about free will, some will be asked about moral responsibility, and some will be asked about both. Only then can any real conclusions be drawn concerning folk intuitions and compatibilism. I suspect we’ll find that people are more willing to give up on free will than they are moral responsibility—but I have obviously been wrong about these things before!

A second shortcoming of the original Nichols and Knobe studies is that they had participants focus on moral responsibility without disambiguating for them the difference between desert-entailing responsibility and forward-looking responsibility. After all, even the non-retributivistic free will skeptic might think there will still be some notion of responsibility in play--namely, the kind of responsibility that is undergirded by compatibilist control. But this makes it really hard to conclude anything about people's intuitions about free will when the only data we have concerns whether they thought the agent in the vignette was responsible. Keep in mind that participants were asked "Is it possible that P is fully morally responsible for x-ing"? First, it is unclear why they are asked whether they think it possible rather than whether they think the agent actually is responsible given the conditions of the scenario. Second, as I have already pointed out, there are two natural ways of reading "fully morally responsible"--one which might require libertarian free will and one which does not.

Of course, these are all criticisms of the earlier work by Shaun and Joshua rather than the follow-up paper by Adam and Edward. Indeed, for me, the paper by Adam and Edward helps bring into focus some of the shortcomings with Nichols and Knobe’s earlier studies. Hopefully, our next round of studies will further resolve the underlying confusions and difficulties discussed above.

[Eddy Nahmias]

I'll try to post more substantial comments later, but for now...

The last sentences participants read in the Nichols and Knobe survey are: "The key difference, then, is that in Universe A every decision is completely caused by what happened before the decision--given the past, each decision *has to happen* the way it does. By contrast, in Universe B, decisions are not completely caused by the past, and each human decision *does *not* have to happen* the way that it does."
I find this description of determinism and its implications problematic. (Fritz's concerns about describing determinism in terms of complete causation are also legitimate--I'd love to hear any suggestions from him or others about how to present determinism in a way non-philosophers can understand--if it can't be done, then it seems problematic to claim that non-philosophers find incompatibilism intuitive, no?).

The reason I find these sentences problematic is that, even as a compatibilist, I would be inclined to answer that Universe B is most like ours and I would also be hesitant to say that people can be morally responsible (or free, were I asked about free will) in Universe A. This is because I don't think determinism (or complete causation) entails that decisions *have to happen* the way they do and I don't think that, in our universe (whether it's deterministic or not), decisions *have to happen* the way they do. And if they did *have to happen* the way they do, I might balk at ascribing responsibility (depending on how the modal claims are cashed out).

One key issue in the free will debate is whether determinism entails that we cannot do otherwise *in the sense* relevant to freedom and responsibility (and to me, ordinary people's intuitions *are* relevant to understanding what sense is relevant). Finding out that ordinary people think we live in a universe where decisions do not *have to happen* the way they do or that responsibility requires some sort of ability to do otherwise is not surprising.

Feltz and Cokely have found some interesting differences and it would not surprise me if there are differences between individuals in their intuitions about free will, determinism, and moral responsibility or that these differences correlate with personality traits (even intelligence!).

I really do wish the Garden community--at least those who see *any* value of any sort in trying to figure out non-philosophers' intuitions--would help those of us trying to do the empirical work find ways of probing these intuitions in as fair and accurate a way as possible. Ideas?

[Adam Feltz]

Thanks for all the great comments ya'll.

Kip,

We too are interested in what others think the relevant individual differences might be. We actually have some data on this, but we welcome more input. For example, we haven't found that intuitions (at least about the Nichols and Knobe set-up) are correlated with any of the Big Five personality traits. However, as Thomas mentioned, we have found a significant correlation between high Cognitive Reflection Task scores and incompatibilist responses, and low Cognitive Reflection Tasks scores and compatibilist responses. We are currently in the process of replicating this study.

Eddie and Fritz,

I'd be happy to describe determinism a different way—I'm not committed to Nichols and Knobe's description. I'm all ears!

In fact, one of our interests is finding ways to define groups of people who express compatibilist and incompatibilist intuitions. After all, even in Nahmias et al.'s original studies, there were some people who expressed incompatibilist intuitions. Once we find ways to define these groups, we hope to use more fine-grained methods (e.g. Protocol analysis) to determine what accounts for the differences in those groups. So, having a good way to describe determinism would help in that project.

Thomas,

You make an interesting tentative prediction. You say that you think that people are more likely to give up on free will than moral responsibility. In that case, it seems that our study would under-report the number of incompatibilists. That is, we already have only about 25% of people giving matched compatibilist responses in the moral responsibility set-up. If these people would be willing to give up on free will, then the percentage should be even smaller. Interesting!

[Fritz Warfield]

Eddy,

The loose dilemma that one faces is presumably something like this:

On the one hand, descriptions using phrases like "fully caused by" and the like aren't obviously and uncontroversially useful for assessing one's commitments regarding determinism.

But on the other hand, the natural way to impress upon students the more precise meaning of determinism is to give them a good semester long course on the metaphysics of free will -- and by the time *that* course is over, one will surely have reasonable doubts about whether one is really tapping into folk intuition any more. [A similar issue faces Nadelhofer and others who sensibly want to make sure subjects are seeing the distinction between various notions of responsibility as discussed in his remark above].

I call this a loose dilemma because there is obviously plenty of potentially habitable space between the horns. Whether one can find a stable point between the horns from which to run a study is the open question. I have no positive suggestion to make for how to do this. But I report the following non-folk intution: there's a better chance of pulling this off for the "more ordinary and familiar to the folk" notions of moral responsibility than there is for the higher-philosophical-entry-cost notions of "determinism" and "metaphysical freedom".

I'm taking some of these issues up in a paper I hope to some day share with you all.

[Joshua Knobe]

Eddy,

Just wanted to respond to your remarks about our study. It seems like the key issue here is about how to interpret the expression:

(1) Each decision *has to* happen the way it does.

The standard view in linguistics is that this sort of expression does *not* mean anything like:

(2) In all possible worlds, the decision happens the way it actually does.

Rather, the expression means something like:

(3) In all of the possible worlds that are picked out as relevant by the present conversational context, the decision happens the way it actually does.

Hence, when one is talking about the laws of nature and the past and then uses an expression like (1), no listener would think that one means to make the ridiculous claim that, given the past and the laws of nature, it is actually a necessary truth that the decision will happen the way it actually does. Instead, the only legitimate interpretation will be that the decision happens the way it actually does in all of the worlds that are picked out as relevant by the conversational context -- namely, those worlds that have the same past and laws of nature as the actual world.

Does that sound like a reasonable interpretation to you?

[Paul Torek]

Joshua,

that does not sound reasonable to me. (I'm just the self-declared president of Eddy's fan club. I'm confident that Eddy's latest post will increase the membership.) The narrowed scope of the modal operator is prejudicial. In most deterministic theories that have actually been entertained by scientists, what is causally necessary is not the given decision, but the biconditional, "past events <--> this decision". Consider how comfortable you'd feel with the sentence, "given that you're making this decision, the past events had to have happened the way they did" - and you will get an idea how uncomfortable I feel with the vice-versa.

Eddy and Fritz,

How about using the words "causation" and "non-probabilistic" to get at determinism?

[Eddy Nahmias]

Hi Joshua, I do have concerns about the way one should properly read the modal claims about decisions given the way you and Shaun write the scenarios. While you are right that reading (1) as (2) seems too strong, I also think that the "relevant conversational context" of (3) is not clear enough to imply just what determinism legitimately implies and no more. (If anyone is interested in the details, I discuss these modal issues in more depth in Jason Turner and my response to Nichols in Mind Nov. 2006 and Journal of Cognition and Culture issue 2006.)

However, these concerns were not the ones I was trying to raise in my earlier post. There, I was more concerned with the pragmatic implications of the wording of the scenario. Consider the very last sentence the participants read: "By contrast, in Universe B, decisions are not completely caused by the past, and each human decision *does *not* have to happen* the way that it does." They are then immediately asked which universe (A or B) is "most like ours."

Again, it seems likely (to me) that many people would answer that universe B is most like ours because everyone knows that human decisions don't *have to happen* the way they actually happen. I also worry that once a participant has said that universe B is the one most like ours, then asking them about agents' moral responsibility in universe A (the one *unlike* ours) may be problematic.

Testing for my interpretations seems feasible. We could counterbalance the order of the similarity question and the MR questions to see if that has any effect. We could reword the scenarios to try to avoid any contrastive descriptions that might lead to "fatalistic" readings of the modality and see if that had any effect. We could try to describe the "has to happen" language differently and see what happens. (I take it that my FSU group's original studies did use different language and, of course, we got different results--results that did not vary across positive, negative, and neutral actions--and my current GSU group's studies are also trying to use different modal language--roughly, "once specific [neural or psychological] states occur, they will definitely cause the agent to make the decision [to X]."

I should emphasize that I'm not sure about the best (or most "fair"--given the concerns of the debate) way to describe the relevant modal claims or ask the relevant questions. But, as I've protested (enough methinks!), I believe that many people have been too quick to conclude that, "Well, that Nahmias, Nadelhoffer, Morris, Turner data was interesting, I suppose, but it's nice to see that it's been shown to be a quirk of affect or emotion (or somethhing) and that, after all, people's ordinary intuitions *are* incompatibilist."

I have no doubt that people's intuitions about responsibilty (probably free will too) are influenced by the emotional salience of cases (as Joshua and Shaun's experiments demonstrate), and perhaps by personality traits, and certainly by the way determinism is presented (e.g., if it is presented in a reductionistic way). But whether most people are pre-theoretical incompatibilists remains, at best, an open question.

[Kip Werking]

1. The group differences that interest me most are free willist versus non-realist differences. Incompatibilism includes both optimistic libertarians and pessimistic non-realists and mixing these two groups, which are probably very different, could cloud the results (where as all compatibilists seem to also think that humans have free will). I don't have any guesses about what the differences between compatibilists and incompatibilists might be. But I would expect some remarkable differences between free willists and non-realists---if only because non-realists represent such a strange and minority view (I don't have any data but I am guessing the percentage of Americans who would say "human beings don't have free will" is less than 3%).

2. The question seems to be whether "had to happen" has a compatibilist reading such that an event in a deterministic world might not "have to happen"---despite the fact that, by definition, there is only one unique future for any given state of the deterministic world. If there is only one future, then it certainly seems to me like the event has to happen. To say it didn't have to happen is to imply that the future could have been different---but this is exactly what determinism denies.

I'm not opposed to compatibilist analyses of all terms. For example, I think the compatibilist sense of "control" is so important that, just because a world is deterministic, does *not* mean that nothing controls anything. Thermostats control things, even in deterministic world (but they do not do so in a buck stopping way; humans seem to feel the control we have is buck stopping in a way that thermostat-control is not).

So, at best, "had to happen" is a gray area. I can understand Eddy's concern: if an event "had to happen", then does that imply that the actor is compulsive? Under duress? I suppose the folk could make this mistake. But I also wonder how fair this is to the incompatibilist, because the sense in which things, in a deterministic world, have to happen is crystal clear to me.

But I'm not sure we have to decide any of these questions. If "had to happen" is troublesome, why not just use "if ... then would happen" language. So "this world is deterministic, so whenever the same conditions obtain, then the same following event X would always follow." This seems to capture determinism while also avoiding Eddy's worry (or so I hope).

Note that: I bet you could obtain huge difference in results just by framing the exact same scenario differently. For example, if you repeated, over and over, the one-unique-future-ness of the situation, I bet you could obtain much stronger incompatibilist results. If, instead, you emphasized the powers and capacities that humans in a deterministic world still have, I bet you could obtain very compatibilist results. Just by emphasizing different things, in the same scenario, could made huge difference in outcome. And this makes one wonder how relevant these results could be.

[Randy Clarke]

Kip:

Here's conclusive evidence that SAT scores are nothing like a measure of IQ. SAT scores standardly rise by hundreds of points as one ages from, say, 12 to 17. IQ doesn't.

On characterizing determinism:

I'm troubled by definitions of determinism that employ a modal term of unclear or ambiguous scope. Why not just give the subjects the sort of definition we use? Something like: given determinism, the laws are such that the way the world is at any given point in time completely determines the way it is at any later time. For example, if S did A at t, then it's impossible for the world to have been exactly as it was at some earlier time long before S existed and to have had it's actual laws, and for S to have not done A at t. Not the easiest thing to understand, but if subjects aren't able to understand that, why should we be so interested in what they think about the compatibility question?

[Tamler Sommers]

Just came across this article in Nature (7137: 583-700) describing the study that Kip asked for.

Apparantly there is a strong correlation (adjusted r2 .99, p-value < .01) between different groups in the debate and the following traits:

LIBERTARIANS: (1) Mysterious (2) Obscure. (3) Panicky.

COMPATIBILISTS: (1) Wishy-washy. (2) Evasive and/or neurotic. (3) Untrustworthy (practices acts of wretched subterfuge).

REVISIONISTS. (1) Pony-tailed. (2) Masochistic (shows unhealthy interest in philosophy of language).

FREE WILL SKEPTICS. (1) Overabundance of intellectual courage and rugged good looks. (2) Solid jump-shot from inside 18 feet. (3) Takes jokes too far.

I hasten to add that this study tells us nothing about the truth of each position.

[Joe]

Half way through reading Tamler's post, I thought that the 'study' he was talking about was Strawson's "Freedom and Resentment."

[john a]

This may be a silly concern, but I am wondering if there is a correlation between what one finds re the intuitions (beliefs) of the 'folk' and the 'professionals' who are testing them? If x% of the folk are incompatibalists is there a similar % of incompatibalists among professionals?

[Mark]

I'm still waiting for someone to discuss how to deal with the various cultural biases implicit in any of these results. If these surveys had been taken by those living during the period of the Dutch reformation, for instance, they would have revealed a very heavy slant toward Compatibilism.

[Eddy Nahmias]

Mark, well, we can't survey people from the Dutch reformation. We can take demographic information on current participants, and we are doing that on my latest round of surveys (not much, but age, gender, and religious info including degree of belief in God and affiliation). Others, such as Joshua Knobe are trying to get data from people with different cultural backgrounds. The historical work would be very interesting.

John, well, I tried to do a sort of head count of professionals here: http://gfp.typepad.com/the_garden_of_forking_pat/2006/10/counting_heads.html

In fact, it was inspired by the van Inwagen article we are reading for our next paper discussion. I suggested (1) that about 2/3 of professional philosophers working on the free will debate are compatibilists, and (2) I speculated that about 2/3 of professional philosophers in general are compatibilists. And personally I think (3) that the data so far suggests about 2/3 of non-philosophers have compatibilist intuitions. But claim 1 is debatable in large part because of how to define "working on the FW debate." Claim 2 was entirely speculative, but I hope someone tries to do the empirical survey (at an APA or online). Claim 3 is contentious, but I am about as sure that it's right as I am that compatibilism is true ...

[Mark]

Eddy,

The reason I brought up the Dutch reformation was to motivate my worry that this project will inevitably shift from yielding conclusions about persons to yeilding conclusions about cultures, religions, etc. In other words, I fear that this project cannot amount to more than a study of the popularity of various belief systems and world views, and I have seen little evidence from the experimental philosophy camp to mitigate this fear.

Surely the Dutch reformation was comprised of a broad variety of personalities, income levels, ages, genders, etc., and yet the majority of the persons in that period in that geographic region were confirmed Compatibilists. Is there a reason that explains this phenomenon? Of course, and it happens to be religious in nature: during this period theological determinism was embraced from the bottom of the society to the top.

[Elli]

Could anybody tell me where the paper disappeared to? The Link is dead, which is unfortunate.