Another paper on “folk intuitions” about free will: Nahmias et al.

February 21, 2014 • 7:38 am

To complement the paper of Sarkissian et al., which I wrote about the other day, I’ll present as briefly as I can the results of an earlier paper on beliefs about free will by Eddy Nahmias et al. (references to both papers are at bottom, free download on this one).

In contrast to the results of Sarkissian et al., Nahmias et al. conclude that the “average person” (in this case, students “drawn from an Honors student colloquium and several introductory philosophy classes at Florida State University”) were compatibilists about free will. In other words, given a hypothetical “deterministic” universe in which the future was completely determined by the laws of nature acting on the present situation, students still believed that in many concrete situations requiring “moral” judgement, individuals retained free will and moral responsibility for their actions.

Nahmias et al. posed three sets of questions to the students.


Students were given a deterministic scenario and asked two questions about it. Here’s the scenario:

Screen shot 2014-02-21 at 6.47.40 AM

As with the Sarkissian et al. paper, there is no quantum indeterminacy in this scenario, which almost certainly means that one cannot deduce the future state of the universe from the present one, but I don’t think that would affect the results, and at any rate it would be hard to explain to undergraduates the idea of pure indeterminacy.

The students were first asked if the scenario given above was possible. The majority of the students said “no” for various reasons (including quantum indeterminacy!), but also for nondeterministic reasons, like “the computer could never acquire that much information.” Like the students in the Sarkissian et al. study (the latter from four countries), then, these students were not deterministic.

Then the crucial question about free will:

Regardless of how you answered question 1, imagine such a supercomputer actually did exist and actually could predict the future, including Jeremy’s robbing the bank (and assume Jeremy does not know about the prediction):
Do you think that, when Jeremy robs the bank, he acts of his own free will?

76% of the students said “yes,” indicating a compatibilist view of free will. Given the deterministic scenario, it’s clear that either this is genuine compatibilist free will exercised in a deterministic universe, or else the students believed in libertarian free will despite the deterministic scenario! That would understand an inability to comprehend true determinism.

To test whether the students accepted free will only because Jeremy did something bad, Nahmias et al. also asked them if Jeremy had free will in this deterministic universe if either b) went jogging (a “neutral” action) or c) saved a child from a burning building (a “praiseworthy” action). They were also asked if Jeremy had moral responsibility in the bank-robbing and saving-child situations.

In all cases the results were “yes”, with more than 60% of the students agreeing that Jeremy had both free will and moral responsibility. Here are the results given in bar charts:

Screen shot 2014-02-21 at 6.59.14 AM

As the authors note, as have other philosophers like Dan Dennett, judgements of moral responsibility are closely aligned with those of free will.


In this study, the authors wanted to see if the respondents thought that Jeremy could have acted otherwise in this situation. They call this the “ability to choose otherwise” (ACO), and this is what many see as a libertarian notion of free will. The authors describe the question:

In these cases, participants were asked—again, imagining the scenario were actual—whether or not Jeremy could have chosen not to rob the bank (case 6), whether he could have chosen not to save the child (case 7), or whether he could have chosen not to go jogging.

The bar graph gives the ACO (“could have chosen otherwise” figures compared to those already given for judgement about whether Jeremy had free will:

Screen shot 2014-02-21 at 7.12.26 AM

The authors summarize these data:

In the blameworthy variation, participants’ judgments of Jeremy’s ability to choose otherwise (ACO) did in fact track the judgments of free will and responsibility we collected, with 67% responding that Jeremy could have chosen not to rob the bank. However, in the praiseworthy case, judgments of ACO were significantly different from judgments of his free will and responsibility: Whereas a large majority of participants had judged that Jeremy is free and responsible for saving the child, a majority (62%) answered ‘‘no’’ to the question: ‘‘Do you think he could have chosen not to save the child?’’ Finally, in the morally neutral case, judgments of ACO were also significantly different from judgments of free will—again, whereas a large majority had judged that Jeremy goes jogging of his own free will, a majority (57%) answered ‘‘no’’ to the question: ‘‘Do you think he could have chosen not to go jogging?’’

I have two comments here.  I’m puzzled that despite the presentation of an explicitly deterministic scenario for human action, 67% of the students still concluded that Jeremy could have chosen not to rob the bank. While that could superficially be seen as compatibilism, it also seems to be a compatibilism based largely on an acceptance of libertarian free will, so that perhaps the students don’t understand the real conflict between libertarianism and determinism.

Second, the notion of “choosing otherwise” may mean different things in a praiseworthy versus a blameworthy situation. In the bank-robbing situation, it may mean that the students really did think Jeremy had a choice. In the “save-a-child” situtation, it may mean that it would be unthinkable for Jeremy not to save the child, so “no choice” is a sign of moral duty, not freedom of will.


The authors proffered a third scenario because of the possibility that [they] “did not make the deterministic nature of the scenario salient enough to the participants.” (They were worried that the “supercomputer” example was not clear enough in mandating determinism.) They thus described a third scenario corresponding to determinism based on genes and environment:

Imagine there is a world where the beliefs and values of every person are caused completely by the combination of one’s genes and one’s environment. For instance, one day in this world, two identical twins, named Fred and Barney, are born to a mother who puts them up for adoption. Fred is adopted by the Jerksons and Barney is adopted by the Kindersons. In Fred’s case, his genes and his upbringing by the selfish Jerkson family have caused him to value money above all else and to believe it is OK to acquire money however you can. In Barney’s case, his (identical) genes and his upbringing by the kindly Kinderson family have caused him to value honesty above all else and to believe one should always respect others’ property. Both Fred and Barney are intelligent individuals who are capable of deliberating about what they do.

One day Fred and Barney each happen to find a wallet containing $1000 and the identification of the owner (neither man knows the owner). Each man is sure there is nobody else around. After deliberation, Fred Jerkson, because of his beliefs and values, keeps the money. After deliberation, Barney Kinderson, because of his beliefs and values, returns the wallet to its owner.

Given that, in this world, one’s genes and environment completely cause one’s beliefs and values, it is true that if Fred had been adopted by the Kindersons, he would have had the beliefs and values that would have caused him to return the wallet; and if Barney had been adopted by the Jerksons, he would have had the beliefs and values that would have caused him to keep the wallet.

The result were these: 76% of the participants judged that Barney returned the wallet and Fred kept it of their own free will.  That result is similar to the figures from the Jeremy scenario. Further, 60% of the participants judged Fred blameworthy for keeping the wallet and 64% of the participants found Barney praiseworthy for returning it. These views were concordant 90% of the time. Finally, 76% of the participants judged that both Fred and Barney “could have done otherwise.”

Again, this evinces a superficial compatibilism, but I am a bit worried about the last result.  Clearly, in a deterministic universe—one in which Fred and Barney’s actions were completely determined by their genes and environment—each could have made only one decision. Either the students do not understand what “could have done otherwise” means, or they have a very sophisticated notion, à la Dennett, of what it does mean, which is that at any given moment either decision was not possible in identical circumstances, but in slightly different circumstances a different decision was possible.

Although the authors note that the students’ replies indicate that they were compatibilist, I am worried that the students still don’t fully comprehend what determinism really means, something that I think philosophers need to clarify when asking such questions. I simply don’t think they’re sophisticated enough to comport “I could have behaved otherwise” with accepting a purely deterministic world. To the authors’ credit, though, they too worry about this.

I won’t summarize the authors’ discussion, but it’s a very good summary of the state of the art, with all the proper caveats and possible objections to their results. On the whole, I liked the paper.

I am of course a “hard incompatibilist”, but the subject of these papers was not to judge whether compatibilism or incompatibilism is the philosophically proper stance. Rather, Nahmias et al. and Sarkissian et al. had identical tasks: are most people compatibilists or incompatibilists? The former says “compatibilists”; the latter “incompatibilists.” How do we reconcile these conflicting results?

As Sarkissian et al. note, perhaps the students tend to be incompatibilists when presented with a scenario asking them to choose “world views”—the vast majority of students in their four-country samples were not determinists and did accept free will—while students tend to be compatibilists when presented, as did Nahmias et al., with more concrete moral dilemmas. This disparity deserves further exploration. But I also think that philosophers who are physical determinists (while accepting some quantum indeterminacy) need to work harder to convey that view to the public. After all, most secular philosophers dealing with this issue are deterministics. The difficulty of limning determinism might be evinced in some of the counterintutive results of the Nahmias et al. paper.


Nahmias, E., S. Morris, T. Nadelhoffer, and J. turner. 2006. Surveying freedom: Folk intuitions about free will and moral responsibility. Philosohical Psychology 18:561-584.

Sarkissian, H., A. Chatterjee, F. De Brigard, J. Knobe, N. S., and S. Sirker. 2010. Is belief in free will a cultural universal? Mind & Language 25:346-358.

68 thoughts on “Another paper on “folk intuitions” about free will: Nahmias et al.

  1. Rather, Nahmias et al. and Sarkissian et al. had identical tasks: are most people compatibilists or incompatibilists? The former says “compatibilists”; the latter “incompatibilists.” How do we reconcile these conflicting results?

    Perhaps it is best to conclude that most people think superficially about the topic and are confused when faced with philosophical paradoxes.

    1. Thinking back to my own college experience, many of these students are looking to complete the task as fast as they can, collect their compensation (or get out of their extra writing assignment given for not “volunteering” for a study), and get to the party.

    2. I’m not sure superficiality or confusion are the right adjectives or right explanation. This could be a side-effect of context-based thinking (compartmentalization). You give a person a general/theoretical question, and they use certain pathways in their brain to think it through. You give them a concrete, visceral question, and they use different pathways. They may be trying just as hard to solve the problem in both cases – giving both questions all their due consideration, without being lazy about it – but they use different cognitive methods to do the solving in the two cases.

      In any event, a very interesting set of studies.

      As side comments, I don’t find the “no” answer to study 1 question 1 to be particularly unusual – the folks who answer “no” are likely just interpreting “possible” in the normal, probabilistic or empirical sense of the word, rather than the “absolute philosophical sense” of possibility. IOW the were answering “is the computer possible” question the same way most of us would answer the question “if you step outside a second story windown, is it possible that you float instead of fall.” That is also philosophically possible…just very unexpected. My second side comment is that I actually find the computer story much easier to think about (in terms of determinancy) than the twin story. I don’t think case study 2 actually ‘fixed’ any potential problem the students may have had in understanding determinancy. But that’s a personal preference and the studies are certainly better off for having asked *both* versions, rather than just one or the other.

      1. “They may be trying just as hard to solve the problem in both cases – giving both questions all their due consideration, without being lazy about it – but they use different cognitive methods to do the solving in the two cases.”

        I think you’re probably right. A number of similar studies have been done on ethical decisions, that reveal people tend to switch theories in different situations.

        I tend to think of philosophy as a two-part process: a descriptive process that requires understanding how people speak and reason about philosophical questions, and a prescriptive process that helps us improve clarity, avoid logical inconsistencies and resolve apparent paradoxes. These studies contribute to the first part.

        1. And the descriptive process is useful because how does one get from here to there if one doesn’t know where “here” is. I think this is the greatest lesson of the 19th century: Darwin, Nietzsche, Hegel (after a fashion), Comte, Marx and others investigated origins of things.

      2. The vagueness of “could have done otherwise” does seem to be a glaring problem in every case presented. “Could have done otherwise” in what sense? The philosophical determinism/free will sense of “CHDO” is not the most common way people use that phrase. Could have noticed other factors to consider? Other physically possible scenarios seem plausible? Could have thought about the problem in a different way to come to a different decision?

        1. Yes the phrase is ambiguous – Whether you “Could have done otherwise” depends on who is observing and at what time. “He could have jumped that high, because he has done it before” versus, taking account his failure in the actual circumstances: “clearly he wasn’t able to jump that high on the day so he couldn’t have done so”. These two different meanings create a confusing equivocation, that makes many discussions involving this phrase meaningless.

    1. +1

      After reading the End of Faith in high school, I began noticing that nearly every TV show, movie and comic book features a steadfast hero who sticks to his beliefs no matter what and overcomes the logical and calculating villain via brute force. Afterward, the hero or a secondary character often reasserts the moral of the story (usually “your destiny is yours to make”, “trying to improve/change human nature is wrong” or something along those lines) without ever meeting the villain’s argument (usually a staw-man argument against religion, free will or another of society’s sacred cows). This is the case even in stories that aren’t explicitly religious or even about religion.

      I think this trend in storytelling is very harmful to our species. Children are inundated with heroes who defend the status quo, or who rebel against a cartoonish regime which represents the fear of the moment having taken over society in the near future, but how many heroes really buck the status quo and ask kids to question the traditions and beliefs of their society?

      1. That seems a rather sweeping generalization, so I can’t say I fully believe it. What kind of works make that endorsement, and aren’t there a good number of works that don’t take that attitude?

  2. Such a supercomputer could never make that prediction. Its inputs would be variables, not attributes, and they could never be measured to absolute accuracy. That means that Jeremy’s actual behaviour will diverge more and more from the predictions as he grows up. If there are any feedback mechanisms, like those, for example, on an evolutionary fitness landscape, and there probably would be, that divergence could be huge. It is not necessary to invoke quantum indeterminacy.

    1. Maybe we are the denizens of some simulation run on a cosmic supercomputer, what difference would that make? If you have the computing power to predict my decisions, why should that imply that those decisions are not my own? If I predict that my chess opponent is going to castle rather than move his knight, and he does so, does that make my opponent’s decision any less his own?

      In actuality decisions become more easily predictable the closer they are to a rational solution to the problem to which those decisions are addressed. Randomness and eccentricity only act to make decisions less appropriate to the problems that one is trying to solve.

      1. My point is that no computer, or anything else, can measure a variable exactly. Even tiny rounding errors comprise noise. Furthermore, errors always accumulate. A simple example is the random walk. For N steps, this on average will end up sqrt(N) from the origin, which tends to infinity as N tends to infinity. Of course, proportionately to the total distance travelled, the error averages 1/sqrt(N), which tends to zero as N tends to infinity, and which is where we get the idea that errors cancel out. The proposed supercomputer is a fantasy. I am very happy to use limits in maths. I am comfortable with taking extreme but plausible cases in less defined situations. But to compose an impossible situation and argue from that is indefensible. It is one example of where philosophy is simply disreputable. I realise that it might have been constructed simply to test the students, but the answer in the impossible case thus proposed is trivially obvious: Jeremy has no free will.

        1. Imagine that our universe is just a single instance in a massive multiverse, where deterministic universes constantly bubble up out of an infinite quantum vacuum. Now in that case the initial conditions for the universe we happen to be in would have been repeated infinitely, somewhere in the multiverse. So everything that happens in our universe would be a replay of countless other universes. Would we then have free will in the universe we are in? Would our shadow selves in the first universe ever created in such a multiverse have free will? Is it the case that once you run a deterministic system once, that free will then disappears from any system when you run it again from the same initial configuration? How is running a simulation of a universe virtually (on a “supercomputer”), different to running it actually?

    2. Whether the computer could have precise enough inputs is somewhat irrelevant. The claim is that the computer can predict everything about the future state of the universe. That necessarily includes the computer’s own future state. So in effect Case 1 postulates a computer than can solve the halting problem. Since Turing proved that no such computer can exist, that’s sufficient to conclude that Case 1 is not possible.

      This is a logical result of computability theory, akin to Gödel’s theorem, and has nothing to do with whether the universe is physically deterministic. So if students answer No when asked if Case 1 is possible, we can’t reliably infer that they’re not determinists; they might just be invoking Turing’s theorem to rule out the possibility of such a computational oracle.

      1. The idea is that the supercomputer is a Laplace’s demon that is outside the universe, in another dimension or whatever, so that that objection would not apply. I imagine most students would twig that. And it’s a thought experiment so whether it is possible in practice or not, doesn’t really matter. Certainly some deterministic systems can be simulated on computers, would that then have any effect on the extent of agency in those systems? It would be odd to claim that impracticality of simulation was the reason we had free will, I reckon.

        1. It specifically says “we build a supercomputer”, implying (at least to me) that the computer is embedded in the universe it’s trying to predict.

          But wherever you choose to locate the computer, the fact that it communicates its predictions to us means it’s causally linked to our universe. Its predictions become part of the causal chain that determines future events within the universe. So it’s still subject to the halting problem even if you hide it in hyperspace.

          The point is not that this somehow gives us free will; I didn’t claim anything remotely like that. My point is that we can’t reliably infer anything about the students’ beliefs in determinism based on a scenario that is demonstrably impossible regardless of whether determinism is true.

          1. TBH, I think this is a bit of a quibble. Physicists often create thought experiments that aren’t actually possible. I’m sure that Maxwell didn’t really imagine it to be possible to have a demon that opened and closed a door, whenever a molecule was nearby. And the authors make the thought experiment status very clear. I would certainly have understood what they had in mind, if I were one of the students :).

            1. It stops being a thought experiment the moment they ask “Do you think this is actually possible?”

              And perhaps it’s a quibble, but Jerry specifically asked us not to go easy on Nahmias just because we like his conclusions.

          2. My problem with this type of thought experiment is that the way it is phrased is biased and acts to lead the responses. For instance, if instead, you said something like:

            Your friend dislikes any sweets, except dark chocolate. You offer him a choice between dark and milk chocolate. If you can write a computer program that can predict with 100% accuracy which type of chocolate your friend will choose, would you say he freely chose between the chocolate types?

            That is essentially the same question, but I’d bet the % results would be very different.

      2. I think that whether one regards the imprecision of the inputs or the failure to solve the halting problem as prohibiting the proposed supercomputer depends on one’s viewpoint. My background is physics, and I am concerned with physical reality. Yours appears to be maths. Where I hope we can agree is that proposing an impossible model and then trying to reason from it is not even not even wrong: it is not even interesting.

  3. I’ve been slowly going through a very recent one:

    Haven’t finished, but looks unusually solid.
    Full citation is:

    Nadelhoffer T., Shepard J., Nahmias E., Sripada C. & Ross L.T. (2014). The free will inventory: Measuring beliefs about agency and responsibility, Consciousness and Cognition, 25 27-41. DOI: 10.1016/j.concog.2014.01.006

    (Long read Alert!)

  4. A comment on all these studies: we would not expect people’s answers to such exercises to be consistent and coherent, indeed we’d expect them not to be.

    That’s because people do not adopt a coherent metaphysics, and then reason from it to produce an answer.

    Rather, they produce their answer from their day-to-day attitudes, and their metaphysics is a rationalisation of those attitudes, and thus need not be consistent.

    Why do people answer “yes” to the “morally responsible” question? They are not reasoning from their metaphysics, they are simply going by their *feeling* of whether such a person would be morally responsible.

    Such feelings have been programmed into us by evolution for entirely pragmatic reasons (since evolution is entirely pragmatic), and the metaphysics is largely irrelevant.

    Thus I think that the whole basis of such investigations: “given such and such a metaphysics, how do you feel about morality in situation A?” is missing the point. People are going to answer the latter according to their ingrained feelings, regardless of the first half of the question.

    This is also why I disagree with Jerry that adopting determinism would lead to a radically different justice system. The justice system is as it is for largely pragmatic reasons. To first order, all that would change would be our commentary about it.

    1. This is also why I disagree with Jerry that adopting determinism would lead to a radically different justice system. The justice system is as it is for largely pragmatic reasons. To first order, all that would change would be our commentary about it.

      Time will tell, but it also depends on geography. My impression of the American justice system, for example, is that pragmatism often takes a backseat to penalty.

      The eye-for-an-eye mentality could be one of the things ( belief in supernatural intervention being another ) that would be diminished if hard determinism was largely accepted.

      But yeah, so far it’s anyone’s guess, I guess…

      1. In America, the justice system actually works outside the justice system. How many rapes, child, spousal, and sexual abuses go unnoticed due largely to the stigma of shame and that no one would actually believe the accuser? There are an awful lot of victims who are punished for life, regardless of what happens to the perpetrator.

        Discourse is a real simple ingredient for mitigating punishment. When people talk to one another that helps both victims and assailants.

        1. Depending on the crime, being a victim is a punishment for life. More focus and research on how to best handle these emotions rationally would be progress imo.

          I completely agree that communication is crucial.

      2. There are certain “crimes” for which the punishment is indeed disproportionately severe. The use if recreational marijuana is one. But that has nothing to do with free will. I’d advocate for its decriminalization whether or not we had libertarian or compatibilist free will.

        With much more serious offenses it’s hard to untangle retribution from deterrence. Consider this case in which the author decries the sentence for being to lenient, and many commenters agree.

        One could argue, as Jerry or Sam might, given their statements about retributive justice, that as long as the Schaibles are denied access to their children, further punishment is not necessary. The problem was the danger in which they placed their children. Prison time would be superfluous.

        On the other hand, one could also argue that prison time, and lots of it, would be necessary, not just to satisfy our sense of justice, but to provide a credible deterrent.

        So Coel may be correct that the justice system wouldn’t change much. Punishments that satisfy our thirst for retribution may map pretty closely onto what would serve as a credible deterrent.

        I am not claiming to have resolved anything here; just food for thought.

        1. So Coel may be correct that the justice system wouldn’t change much. Punishments that satisfy our thirst for retribution may map pretty closely onto what would serve as a credible deterrent.

          Which is exactly what one would expect if our morality and “thirst for retribution” had been pragmatically fine-tuned by evolution to do the job of providing sufficient deterrence.

  5. I wonder whether there is merit to the idea that, even assuming that moral culpability cannot truly exist in a deterministic universe, maintaining the notion that we are responsible for our actions acts as a very strong motivator of “good” behavior–i.e., makes its probability much more likely and is thus worth retaining in the same sense that imprisonment is worth retaining as a deterrent even if it has no value as a “punishment” in a deterministic universe.

    1. Theists have floated your exact argument in defense of religion. I.e., if it keeps people in line, there’s merit to in even if its wrong and a lie.

      If it’s not a good argument for theism, it’s not a good argument for free will. (Though I suppose one could consistently defend the opposite conclusion too – that it’s a good argument for both.)

    2. … moral culpability cannot truly exist in a deterministic universe …

      Here you assume that the only “true” form of morality is a dualistic/theological one, one that is incoherent and nonsensical when examined.

      Surely the “true” form of morality is the one that does actually exist, namely the compatibilistic version of morality.

  6. Jeremy does not act of his own free will. The thought experiment clearly states that. In any of the thought experiments, those who really think “could have done otherwise” are almost always naive-dualists, i.e., they may know free will is an illusion but they want it to be true. The general public is not thinking deeply about this.
    I think by 2150 people will regard the thought experiment of Jeremy’s robbery as pointless as others now consider the philosophical question of: would you switch the train tracks to prevent many deaths versus one. (It makes no sense since no such situation will ever exist, except as fiction.)
    The more that physicists, engineers, computer scientists, and mathematicians try to simulate complex systems the more we are learning that we are very likely and unavoidable limited with regard to simulating real systems. That being said, small miracles could happen in science between now and then, e.g., quantum computing but those advances still do not guarantee successful prediction capabilities for computers.

  7. I personally would argue the 100% accurate predictive supercomputer is impossible because its predictions would have an effect on the future it is attempting to predict. Thus with some predictions, either it can make a prediction and keep that particular scenario from happening, or not make the prediction and the scenario does happen. It cannot simultaneously make and not make a prediction in order to get the outcome right all the time.

      1. Well, take the bank robbery example. If the prediction were widely known, the robber might well choose not to rob the bank on the high likelihood of being caught. The bank operators might choose to close that day to avoid being robbed. Now the computer can make the prediction which will become false, or falsely not make the prediction that will supposedly come true without influence.

        1. You assume that the computer’s prediction becomes widely known. You also assume that the people will manage to act as you described. It doesn’t have to be that way.
          Imagine the bank operator is on drugs, forgets about the prediction and opens the bank on that day anyway.

    1. I think the real problem with it is the amount of resources, including time and energy, that it would require to do this. Pace, Ben Goren.

      No, a computer like that will never actually exist. But that is beside the point in these surveys. We are asked to suspend disbelief for the sake of argument in all sorts of other contexts. That’s all that’s happening here.

      1. But respondents were specifically asked if they thought Case 1 was possible, and inferences about belief in determinism were drawn from those answers. If the question is flawed and proposes a scenario that is known (by at least some students) to be impossible for reasons unrelated to determinism, then that calls those inferences into question.

  8. What I gleaned from this study is something that I’ve been aware of for a long time: a majority of people simply don’t know how to think!

  9. “I won’t summarize the authors’ discussion, but it’s a very good summary of the state of the art, with all the proper caveats and possible objections to their results. On the whole, I liked the paper.”

    OK good. So can we agree that “compatibilism” has some evidence going for it, even if we may want to revise our opinion of this research at a later time? Compatibilism is not merely a response to determinism that philosophers concocted to give themselves jobs, etc. etc. There is some prima facie evidence that can be found for the view.

        1. Relevant to the discussion perhaps, but not “evidence that can be found for the view”, unless you mean the existence of such a view in people’s minds, which is a fairly trite point to want to argue about.

          1. Coel is right about this. If the issue concerns how people understand the meaning of “free will”, then evidence about what they believe is relevant.

            1. Your original comment wasn’t very clear about whether it was dealing with the truth-status of compatibilism or the do-many-people-actually-think-this-way-status of compatibilism.

              1. Jerry has repeatedly claimed that “the only reason people defend compatibilism is that philosophers have concocted a face saving device to protect people from determinism, etc. etc.” This claim is false if there is independent evidence that people understand the word “free will” in the way the compatibilists’ claim.

  10. Thanks for the discussion of our studies, Jerry et al. Consider: If a coin lands heads in a deterministic universe, could it have landed tails? If a dog barely drops a frisbee in a deterministic universe, could it have caught it? If a U.S. hockey player barely misses a shot in a deterministic universe, could she have made it (or was it inevitable or fated since the Big Bang that Canada would kick our asses twice)?

    The answer to these questions is not obvious or uncontroversial. I think commonsense favors ‘yes’ answers to all of these questions, but perhaps if we think hard enough about what determinism means, many people will think the answer should be ‘no’ or ‘no, not really, not unless something had been slightly different going all the way back to the Big Bang.’ Because analyzing the relevant counterfactuals (and nearness of ‘possible worlds’) is ugly under the assumption of determinism, I like to think the actual universe is indeterministic.

    But the indeterminism is not essential to free will–the sort of control over our choices and actions required to deserve praise and blame. And the sort of choice and alternatives we need for such free will can be understood and analyzed in the same sort of ways that we can understand and analyze the coin’s landing tails, the dog’s catching the frisbee, and the hockey player’s scoring, even though they actually didn’t and even if determinism is true. At least that’s the sort of view I think compatibilists can advance without giving up the idea that free will requires alternatives/ability to do otherwise. And that’s the sort of view I think is consistent with the “theory-lite” understanding of these issues that most ordinary people seem to have.

    Hence, when Jerry states, “Clearly, in a deterministic universe—one in which Fred and Barney’s actions were completely determined by their genes and environment—each could have made only one decision,” I think it is *not* so clear. It depends on how one understands the ‘could’. But lest there be any confusion, let me make clear that:

    1. I understand that determinism does entail that they could, *holding fixed all prior conditions and the laws*, only make one decision. And some of my more recent studies have shown that even when people understand that to be true, many still say they had a choice and could have done otherwise. These people are not confused, any more than a scientist is confused when she says that the weather patterns could have been different, and (as a good scientist?) she also believes determinism is true.

    2. I think dualism is false, and I understand that many people think dualism is true. I just don’t think many people are committed to dualism’s being required for free will. (I also think many people are “conceptual dualists” who are theory-lite about the mind-body relation too and would be fine with a non-reductive form of physicalism; some recent studies I’ve done to test Sam Harris’ claims about the folk view of free will suggest that most people are comfortable with the idea that our choices can be perfectly predicted based on prior neural activity. More on those studies if you want.)

    3. I do NOT think that answering the question of what most people think about free will, choice, and responsibility (or their relationship with determinism or naturalism) answers many of the most important questions in these debates. But it helps us find the target of our metaphysical and scientific debates and it may help us avoid talking past one another. E.g., Coyne (and Harris) and I agree that dualism is false, that there is no contracausal free will, and that the modern mind sciences show we are mistaken about many of ideas about how the mind works. But because they think most people are committed to dualistic libertarian free will, they think we have no free will, whereas I think most people think of the concept as closely related to rational self-control, so telling people they lack free will is telling them they lack more than they in fact do lack (and I also think the science does not show that we lack conscious or rational self-control, of a naturalistic, not dualistic, variety: see

    Sorry, I didn’t mean to go on so long. But thanks again for the discussion!

    1. But… why do you like to think that the universe is indeterministic? Just for ease in calculating the counterfactuals? Surely, the pseudo random nature of events, given classical chaos, serves just as well to present agents with situations they have to cope with in real time.

      It seems to me that this hankering after indeterministic causes to events must hide some predilection to believe that somehow we could be more free in an indeterministic universe than in a deterministic one. And, surely when you look at this rationally, that isn’t the case. How can randomness make any decision more “down to us” whether that randomness is in the universe or in our brains?

  11. As I wrote last time, the self-contradictory answers are probably best explained by the fact that most people aren’t particularly good at thinking.

    That being said, I do not understand why Dennett’s view of could have done otherwise is somehow considered so otherworldly and “sophisticated”. Prof. Ceiling Cat writes that clearly, in a deterministic universe each could have made only one decision, but well, even in a completely random or supernatural universe every decision only happens exactly once.

    Thus “could have done otherwise if they had, at that moment, had a different preference inside their brains” is the only sensible meaning for “could have done otherwise”; there is no other that works. It is possible that the intuitions of even the rather confused participants were good enough to get this because this is, after all, also how we evaluate people’s actions on a daily basis. I could have tried to become a programmer instead of a biologist if (subjunctive) I had had that preference, but looking back there would only ever be one outcome regardless of which it would have been.

    1. Yes. I’ve made this kind of point too: “Could have done otherwise” only makes sense when you are referring back to the situation before you made a decision, when the options are still open. it makes no sense to interpret the phrase, from your perspective after you made the decision, as some incompatibilists appear to do.

  12. I’m sure we can all think of times in our lives when something that had us stumped finally comes into focus, and we have what I call a “T.H. Huxley moment” and say, if to no one but ourselves, what Huxley did after grasping the central concept of the Origin of Species, “How extremely stupid of me not to have thought of that!” When the phrase “could have done otherwise” (CHDO) is brought up (by many parties in this discussion/debate) both sides seem unaware of (or just do not think it is important enough to point out) other applications of CHDO to human cognition and our ability to learn.

    If incompatibilists (hard or otherwise) and compatiblists (e.g. Daniel Dennett) share the goal of persuading the mass of humanity that contra-causal, libertarian “free will” is an illusion, there are concepts related to CHDO that need to be acknowledged, and either included or excluded, as appropriate, in our discussions. Speaking only for myself, when, after being stumped by something, say a bit of algebra, I finally see the way to the solution, I (sometimes literally) smack my forehead with my palm and berate myself for not seeing the solution sooner. Part of my exasperation with myself is a feeling that I “could have done otherwise.” This feeling is, by its very nature, a subjective one, but anyone insisting they have never felt anything like that must never have experienced one of the most profound joys we humans can experience, that of understanding something new about the cosmos.

    Outside of a “philosophical” discussion like this, CHDO is usually invoked with the benefit of hindsight, referring to a retrospective realization that other options were open to us at some juncture, but we just were not aware of them at that time. In the universe we inhabit, governed by the physical laws that it is, as individuals, and as a biological species with a particular evolutionary heritage, there are sets of things that we can do (in circumstances not prohibited by a deterministic universe), and sets of things we cannot (ever) do. Our evolution has equipped at least some of us with the cognitive capacity to come up with the math to describe how “matter tells space how to curve, and space tells matter how to move” (as Lawrence Krauss put it, if I recall). Now such a feat as Einstein’s is way above my glass ceiling, and I could NEVER do otherwise than fail most spectacularly if I tried to pull of the same feat “from scratch” as it were.

    On the other hand, I got straight “D’s” in 9th grade algebra (circa 1980) and the old fart of a teacher I had advised me to not take geometry when I got to high school the following year because I would fail. At that time I also thought algebra, let alone geometry, trig, or the calculus, would ALWAYS AND FOREVER be beyond my ken, but I was wrong. The “math light” came on for me in my early 20’s and I eventually went on to successfully complete a year of undergraduate-level calculus and a semester of differential equations, among other math-intensive courses. My apologies if I seem strident…but that old fart of a teacher was WRONG. He was wrong in just about the biggest way is is possible to be wrong because he was empirically wrong. It was not that I was constitutionally incapable of ever being able to grasp algebra, he was just a lousy teacher (this is but a single chapter of a longer story I won’t go into here).

    As a Gnu Atheist®, I know how much fun it can be to be a Socratic “gadfly,” flitting to and fro deflating others vacuous sacred cows, like theistically motivated ideas of “free will.” However, if we actually want to help others by persuading them to change their mind, to learn, and apply, a new and better understanding of themselves and the nature of the universe, loudly decrying the notion that one “could have done otherwise” is a delusion, by itself, not going to work well. We cannot just assume they will understand that not being able “to do otherwise” PREVIOUSLY, does not mean that they cannot EVER learn to “do otherwise” going forward. Putting it more bluntly, it takes a lot of nerve to lecture folks on how their assumptions and intuitions about the world are illusory at best and delusional at worst, and yet not follow our own advice by checking our own assumptions about what subtleties or caveats we need to mention or not mention. Just to be deliberately provocative, I will go so far as to say that assuming that “When I/we say X is true, they know that I/we don’t need to point out that X′ is false,” is a failure of our moral responsibility to be intellectually honest.

    For those that dismiss the concept of being “self-aware,” consider that skeptical, rational thinking, and the scientific method specifically, REQUIRES that we be “self-aware,” in the sense of continually asking ourselves if there is anything we have overlooked, are there confounding variables we have not compensated for, or most succinctly, always asking ourselves, how can we be sure we are not wrong?

  13. I think the simple explanation here is that most people are dualists, and that a disclaimer isn’t enough to make them reason otherwise.

    You could try polling people, ‘Given that Jesus wasn’t God’s son, would he be a charlatan?’, as a sort of control.

    1. I think that is a reasonable possibility that isn’t ruled out by these experiments. The fact that the subjects of both questions are human beings may be encouraging the respondents to use their intuitions of human capabilities, rather than think about the question, as you suggest… What would be the result if you asked the students the same kind of questions about deterministic programs, such as chess engines? Say: “Could the program have played another move than the one that it did?”. I rather suspect that you might get a higher percentage of no responses to a question *phrased like that*, but who knows?

      Anyway, IMOP questions involving “could have done otherwise” are equivocal and you can subtley phrase the questions to favour one side of the equivocation or the other. It would seem hard to get round that. I’ll bet, for instance, that if you said to a chess player “could the computer have castled, rather than moving his knight? the answer would be yes, if castling was a legal (and decent) move in the position where the computer moved its knight.

  14. In the 1990s I was writing genetics simulations for educational materials, and I had occasion to use random number generator functions. After all, deterministic or not, the biological world is routinely treated as having large, significant, and apparently random processes, such as meiotic segregation, fertilization, etc. These are not trivial. Genetic drift, for example, is a random process that can have huge evolutionary consequences, not to mention providing us with exquisite tools for reconstructing the past. Anyway, the computer science people have written random number generators that we can use. Now of course these are deterministic; but they produce results that are to most of us indistinguishable from random numbers. Most of the readers of this blog understand this, probably better than I do.

    Given this, does anybody know if there are any known biological circuits that have the property of generating apparently random results? I am thinking of a specific possible example in my own experience, the behavior of cultured muscle cells, which form long multinucleated fibers that twitch from time to time, without any nervous input at all.

    What would be the adaptive advantage of a system of neurons, for example, that produced a random output? I can think of one, and that is that there could be situations in life when a decision – any decision – is better than none.

  15. It’s silly to pose questionnaires that the “experimenters” regard as presupposing the answer that is alleged to be sought “empirically”. The students’ answers showed they gave the experimenters the benefit of the doubt (i.e, in judging them non-silly and non-question-begging)–but I think they may have given the experimenters too much credit.

Leave a Reply