Blog Coordinator

X-Phi Grad Programs

« Moral Psychology (Vol. 4): Free Will and Moral Responsibility | Main | Science, Religion, and Culture »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Manuel Vargas

Hi Josh-

This sounds really interesting. I look forward to reading the paper, but I'm wondering about something say above re: the "neither can be tossed out as an error" claim. As you know, a number of us are are well-disposed to (and have even defended!) the claim that the folk display mixed commitments on the compatibility question, and that it is no mere verbal dispute. And, it sounds like your work does a nice job of articulating some of the more precise contours of folk diversity of commitments.

However, by claiming that neither set of commitments are an error do you mean "neither set of commitments reflects an error in assessing folk beliefs" (i.e., the folk really do have mixed commitments) or do you mean "the folk aren't in error" (i.e., there is no mistake—conceptual or metaphysical—going on when the folk articulate what looks like mixed commitments).

I take it that one can think that folk commitments display both compatibilist and incompatibilist strands, that this reflects a variety of possibilities about how our "concepts" work (whether multiple folk concepts, conceptual fragmentation, varied weights in a prototype concepts, etc.), and that it is a still further question about what (all things considered) we should think free will comes to, i.e., the metaphysical question. Is there something about this sort of picture you take your work to be rejecting or overturning?

Okay, fine, I should go read the paper.

Josh May

Hi Manuel!

You're quite right to press me on this. I only meant to be directly addressing the first option you mention---the conceptual rather than metaphysical issue. Saying "performance error" instead of just "error" would be a bit better, but I avoided that mostly just to avoid technical terms (it's also somewhat narrow). My aim in this paper is just to understand our ordinary thinking about free will. So by "error" I meant only to be talking about whether certain intuitions should be discounted as not reflecting our ordinary thinking. Murray and Nahmias, for example, want to toss out apparently incompatibilist intuitions because they think it's a mistake to treat them as incompatibilist, not that it's a mistake to have incompatibilist intuitions.

I don't tackle the sort of project you've engaged in, which attempts to go beyond the given concept of freedom (or responsibility) and revise it. But I do happen to think that my project could inform such a revisionist one, with some added premises of course.

In fact, in working on this paper, I did explicitly look to see if you (or others) very clearly made similar claims that I make about the ordinary concept. I couldn't find anything explicit, but I did get the sense that you (and perhaps Nichols) may well think the ordinary concept is mixed in roughly the way I think it is, even if you haven't said it exactly in print. But I could be way off.

Manuel Vargas

Hi Josh-

I see. That's very helpful—thanks.

If you were interested, you might take a look at Chapter 1 of /Building Better Beings/, as I do explicitly argue for a mixed view story there. I don't adopt a "cluster" account but I think the view there is mostly consistent with your more nuanced account, and perhaps complementary w/r/t providing an account of why philosophers have tended to avoid mixed accounts.

Marcus Arvan

Josh: interesting project - though I worry about why our *commitments* regarding free will should have anything to do with the fact of whether we actually have it (can't we be committed to things that are just false?).

Anyway, just a quick thought on the way you set up the dilemma. You write, "But now suppose determinism is false so that you have a robust sense of options. But then which action you perform seems partly a matter of luck."

I don't think the falsity of determinism implies that which action you perform is partly a matter of luck. Certain libertarians about free will (of which I am one) take free will to be a *primitive* feature of the world--a genuinely uncaused cause comprised by mentality itself. I've heard that some think this is incoherent, but I've never understood that argument. Electrons behave the way they do because, well, their behavior is a *primitive* law of nature. Why can't libertarian free will be similar? Nature has to have primitives, so why can't thought/free will ex nihilo be one?

Josh May

Hi Marcus. Thanks for your comments! I totally agree that facts about the ordinary concept (or our "commitments") needn't automatically lead to conclusions about the phenomenon itself. That's why I think we'd need extra premises to get to any conclusions about whether we have or lack free will. But I do think that debates about the phenomenon itself (freedom) sometimes depend on, or are affected by, facts about how we think about it. For example, I think many philosophers try to navigate out of the dialectical stalemates by trying to discount intuitions on the opposite side.

On luck: I agree that the book isn't closed on the luck argument. I didn't want to really defend that argument here but rather to use it to illustrate the "mystery." I think the problem can be illustrated in other ways, such as by data like mine, but prior to that data I think it can be raised by these sorts of arguments. Even if the consequence and luck arguments are ultimately unsound, the intuitive problem is still visible, I think.

I do happen to think the luck argument is pretty powerful and that it poses a serious problem for incompatibilism. But that's a very different battle indeed!

Florian Cova

Thanks for this fascinating paper,

(Though I kind of worry that if you had asked Nahmias and Murray's bypassing questions, they could have tracked participants' answers in those case, and thus explained your results in a deflationary way. Did you do something like that in a control study?)

When reading your paper, I was constantly reminded of Fischer and Ravizza's distinction between two kinds of control: guidance control (actual causal control on one's action) and regulative control (ability to make things happen differently). Fischer and Ravizza argue that each kind of control is independently sufficient for moral responsibility. It felt like you were recapturing some of this idea. So, I think you might be interested in looking at this distinction if by chance you did not know it.

Josh May

Thanks for this, Florian! I didn't really have those two kinds of control in mind, but they do fit very closely my liberty and ensurance. It probably would have been best to cite Fischer and Ravizza at least. One reason for avoiding their terminology, though, is that they think of these as both a kind of control. But I was trying to leave that issue open. Some might think of liberty (roughly, the having of options) as just a matter of control, but I think others would balk at that. I think many incompatibilists, for example, think that having options is a further condition beyond control (even though other incompatibilists definitely think of it as at least affecting control). Moreover, I think some fans of the luck argument would likewise resists the idea that my "liberty" has anything to do with enhanced control!

On bypassing questions: I avoided asking Murray & Nahmias's questions for several reasons. One is that I think some of their questions are problematic. Another reason is that I think their whole methodology is a bit suspect or not as informative as I'd hope. Finally, there are other critiques (e.g. Rose and Nichols's data). I figured this was enough to avoid testing for bypassing confusions.

Eddy Nahmias

Josh, interesting paper and results. Obviously, I don't share your view that Murray and my 'whole methodology is a bit suspect', but I won't quibble here with your discussion of our work. I will push Florian's question, though.

Start with an assumption we probably agree on: the ordinary understanding of free will is closely tied to the notion of choice. An agent who can't make choices lacks free will (or perhaps: An agent who doesn't have choices lacks free will, though there may be complications with that version). An agent who is not able to reason or act on her reasons can't really make choices properly.

So, suppose participant Pete reads a scenario and thinks it implies that "In this universe, an agent's reasons have no effect on what she does" (a version of a bypassing statement I'm using now, but I'm happy to hear if you or anyone else has some better ideas). It would be likely that Pete would also think the agent in the scenario cannot really make choices (note that this fits in some ways with Rose and Nichols' findings on the existence of decisions). And hence, Pete would also think the agent lacks free will.

It's possible everything's lumped together for Pete, so we can't tell whether his bypassing judgment is driving his judgments about choice and free will or vice versa or what. Careful experimentation might help us sort this out (Murray and my results was trying to be careful!). But the worry is that the bypassing judgment seems to me to be misreading what determinism implies. It also seems to suggest Pete is interpreting determinism to preclude "ensurance" (or control by one's psychological states).

So, even if Pete read your scenario in the bottom left (Ensurance with no Liberty), it's not clear he's assuming Ensurance. He may be making a bypassing judgment, which is, I take it, one reason Florian was asking if you'd asked any bypassing questions. (I am a bit surprised that the use of "will" instead of "must" didn't have larger effects, but it's Nichols and Knobe's Universe A *vs.* Universe B that leads to the highest bypassing judgments. In any case, changing that word in your scenarios is not really a test of the bypassing hypothesis.)

One way to put this is that I think compatibilists should certainly not give up on the idea that free will requires choice. They should offer a plausible analysis of making (and even having) choices in a deterministic universe. And we should see (test) whether ordinary people find that analysis to be a wretched subterfuge, or instead, as I think (at this point), it fits just fine with most people's "theory-lite" understanding of choice. But people's understanding of choice certainly requires that agent's choices can be based on their reasoning and reasons.

Josh May

Eddy, thanks so much for the comments. It's great to hear some of your thoughts about these results. I should say, I was a big fan of the bypassing hypothesis initially. I've only recently lost my way!

I take it the thrust of your comment is that there is a way for your bypassing theory to explain my results that is friendly to your claim that (roughly) only compatibilism is really "intuitive." I agree. I didn't mean to suggest that I've found results that are incompatible with the hypothesis (though perhaps Rose & Nichols have). Rather, I was content to not definitively rule it out because I thought there are enough critiques that the burden is now back on you and Dylan to defend the hypothesis further.

That's just by way of saying why I didn't include bypassing questions. But that doesn't mean it shouldn't be addressed in further work, which sounds like your main concern. However, I'd like to hear more about the alternative explanation. If I read you correctly, the idea is roughly that IF some of my participants misinterpreted determinism as involving bypassing, then you can explain my results (while still hanging on to your theory). That seems exactly right. But I think it would be ideal to see some independent reason to suspect people did this with my cases. Your previous explanation (about these sort of cases) was nicely specific: strong modal terms like "must" (vs. "will") increase bypassing judgments. Then we can test the hypothesis rather directly by altering the problematic wording. I can't discern a similar new hypothesis about my sort of cases. But maybe your view is that a good number of people will just make bypassing judgments about deterministic universes no matter how they're described.

If that's the move, then it seems that the only way to test it is to ask a battery of bypassing questions, as you an Dylan do. That seems like it could work if done carefully, as you suggest. My only worry about it is that it's so difficult to ask the right sorts of questions. There are e.g. the worries raised by me, Rose and Nichols, and Josh Shepherd. But of course difficult doesn't mean impossible! (BTW, about the bypassing question you propose in your comment, wouldn't Shepherd's worry apply to it?)

Eddy Nahmias

Well, you've already got half the participants offering compatibilist responses in the deterministic ensurance scenarios (which, interestingly, is a good bit lower than we got with the rollback concrete case, which may have something to do with the different questions we asked). And my own rough estimate of the proportion of "pure incompatibilists" before philosophical training is 15-30% (not too different than the proportion among philosophers)--I don't think only compatibilism is intuitive, just that it's not a wretched subterfuge or changing the subject, etc.

So, I'd predict that leaves about 20-35% of your participants who might have read the rollback cases with a bypassing interpretation (with either the 'must' or 'will' language, though it looks like you got a higher mean on the 'will' version, but you don't report the % for that one). I'd predict that about that proportion would agree to statements like "In Universe 49, Jill's reasons have no effect on what she ends up doing." (We got 35% agreeing to bypassing questions in the concrete rollback case.) And I think it's accurate to say that such participants are not interpreting the scenarios to allow "ensurance", as you call it. Why would they read the scenarios that way? I suspect some of them are reading them fatalistically, such that, because the same thing [must or will] happen every time, it will happen that way no matter what Jill's reasons are or how she reasons. Maybe the phrase you added, "even though she knows it's wrong," primes some of them to think this way? I'd like to get better information about this issue. It is indeed difficult to ask the right sorts of questions!

I'm not sure how to read Shepherd's worry. If people are reading the scenarios to mean that agents' reasons can have no effect on their actions in the sort of way a bad offense can have no effect on a great defense, I think that is still suggesting a sort of fatalism or bypassing. The offense can't make a difference no matter what it does. An addict's reasoning has causal effects, but he's going to end up taking the heroin eventually one way or another. Free will requires that agents (and their reasoning) can be difference-makers. But tell me if I'm missing his point.

Josh Shepherd

Eddy: I dunno whether I'd equate that with fatalism (although I feel the pull of what you're saying). Just talking out loud here, as I'm on my way out the door: the thought wasn't about difference-making, but rather about causal importance, in a loose folksy way. If you like Gunnar's Explanatory Salience Model, then talk of determinism makes the causal impact of desires/beliefs/etc. less salient, and thereby less significant as an explanation of what happened. So one could deny that desires/beliefs had any effect - which would mean 'had much of an effect' or 'had an important effect' or 'had an explanatorily significant effect' - without having any explicit commitment to the truth or falsity of fatalism.

Or so I just suggested.

Paul Torek

Nice paper, Josh! I think it would be interesting to follow up on your Experiment 2, to see if some participants infer "must happen" when all you say is "will happen". This isn't quite the same as Eddy's Bypassing hypothesis, though it may be similar. Eddy's idea goes further than "must happen" to specify "must, independent of the thoughts desires and reasons of the agent" whereas I am simply wondering what the participants infer, or don't infer, about a general modal claim. Of course, like Eddy, I would want to know how belief in the stronger modal claim (if some have it) affects participants' free will judgments.

Josh May

Hi Paul. Good point. People might be making the inference to "must" from "will." I was primarily trying to test Eddy and Dylan's prediction that "must" (presumably rather than "will") increases bypassing judgments in such cases (and thus removing it will lead to different results). But it does seem like it's only problematic for my theory if the inference to the "must" leads to an increase in bypassing judgments. After all, if people interpret "must" without bypassing, then it seems they are properly understanding the scenario as involving determinism in a way that works through the actor's psychology.

So, it seems to me that, if we test and find people are inferring "must" from "will," then we are just back to Eddy's worry that bypassing may be occurring. I'd then love to find a different way of trying to remove the tendency toward bypassing and see if it changes responses. In other words, while Eddy and Dylan's strategy is to track bypassing judgments to see if they mediate freedom/responsibility intuitions, I'm hoping to remove the culprit itself to see if it changes freedom/responsibility responses. But, as I said to Eddy, one might just think bypassing will occur no matter how the vignette is framed. If that's right, then there would be no way to carry out my strategy for testing the bypassing hypothesis regarding these cases.

Eddy Nahmias

Two quick points. First, Dylan and I did try to manipulate bypassing responses in our second study reported in the PPR paper, as Josh suggests should be done here. We explained that in Universe A and in Universe C (our rollback one), even though everything is completely caused by prior events going back to the beginning of the universe (indeed, we used 'had to' and 'must' language in both), people's mental states still were part of the causal chain and affected what they did. Not surprisingly, bypassing responses went down. And as our theory predicts, MR and FW judgments went up.

Second, while I think a sort of fatalistic reading of determinism--my actions will occur no matter what I choose or try to do--causes some people's bypassing judgments (and 'must' and 'had to' language may suggest fatalism to people), I think we need to do more work to figure out how much of the effects are driven by that sort of thinking, or reductionistic thinking, or interpretations of 'external' sufficient cause competing with internal causes, or the simplest explanation (that is consistent with the Rose/Nichols view), that something about the scenarios leads people to think agents don't even make decisions at all, leading them to no FW judgments and bypassing judgments. I take that response to be a mistaken interpretation of determinism, but in any case, if it's being driven by, say, the description of Universe A in contrast to B (the one where it explicitly says human decisions occur), then it is not an informative response about the consequences of determinism. It's likely a complex story and we've just started to unravel it.

The comments to this entry are closed.