Sean Carroll on free will

January 2, 2023 • 11:30 am

Below is a fairly new and short (7 minute) video by the Official Website Physicist® Sean Carroll on free will. As ever, he argues that we do have free will, but it’s a compatibilist form of free will. That is, he accept “physical determinism” as totally underlying our behavior (he means “the laws of physics, which can include purely indeterminate quantum mechanics”), but says that because we cannot predict the future or what we are going to do, the laws of physics aren’t useful in helping us understand or predict our behaviors. The word “determinism” seems to be playing a big role here, conflating prediction with reality, which is why I prefer to use the word “naturalism” now.

As Sean’s said before, his view of “free will” invokes a level different from that involving the laws physics: it’s the fact, as he says, that  “We can talk about human beings as agents making choices, while also agreeing that we don’t violate the laws of physics.” That is, we can’t use those laws to decide what we’ll choose in a restaurant. He argues that each human is a collection of desires, preferences, and values, which are useful in both discussing our behaviors and predicting them, but we’re also  “a collections of neurons and obey laws of physics.” Thus we get the compatibility between physics and “free will”, which of course is not “libertarian” I-could-have-chosen-differently free will.

In that sense, every organism also has free will, although some lack values.

Sean, then, sees his form of free will as an emergent property of neuronal organization and evolution that has given us brains that secrete our behaviors. But he also admits that if we were able to predict perfectly what we would do, then “free will would go away.”

It is that last sentence that lays bare what I see as the problems with Sean’s argument. That’s because one thing is for sure: over the coming decades and centuries, as we learn more about the brain, we will be able to use measures of physics independent of “values and desires” to predict more and more of what we do. Already brain-scan experiments using MRI and similar crude techniques can predict what we will choose (in very simple binary-choice experiments) seven to ten seconds in advance. Does this mean that some of our free will has been taken away? This is a kind of free will destined to disappear when we learn more about science.

I do agree with Sean that we talk as if we have free will, and that we act as if we could have done things differently from what we did. As he says, this is because, “given the actual information you have about yourself, you could have acted differently, because the information you have yourself is wildly incomplete”.  And that is true as well. But what is also true is that, at bottom, what we do does depends completely on the laws of physics, and our actions are “emergent” only in the sense that at bottom they rest on those laws. Any “emergence” of behavior isn’t based on some non-physical phenomenon like “will”; it is simply our inability to presently extrapolate from lower to higher levels. But there’s nothing new happening at those higher levels.

So if we’re talking about everyday paralance, I have no real problem with Sean’s conception of compatibilist free will. But I think he avoids the question that obsesses me, which I’d pose to Sean like this:

“Yes, Sean, we don’t know enough about our constituent particles and cells to make complete predictions about our behavior. But, on the physical level, isn’t it true that we could not have done other than what we did?”

I think he’d have to agree with me, because he sees no form of non-physical “will” that, given an exact rerun of physical circumstances, could somehow change the resulting behavior. Ergo I think Sean overly neglects libertarian free will, which, after all, is the form of free will that most people envision. Indeed, when I debate the issue with friends and acquaintances, they are astounded to hear that they could not have done otherwise, even if we feel we could have. Most people do seem to adhere to a form of nebulous, un-physical “will”. And if you tell them “well, given what you know you could have made a choice but you really couldn’t have,” that wouldn’t satisfy them.

Nor would it satisfy the many religionists who absolutely believe in libertarian free will. If you accept Jesus as your savior because that’s compatible with your feelings and desires, but your choice could actually be predicted if you had perfect knowledge about your body and the universe, I don’t think that Christians would say that this alone will bring you to God! For that turns every Christian into a Calvinist!

I also agree with Sean that “we have a responsibility for what’s going to happen next.” It’s a mistake to think that hard determinists like me don’t agree with the notion of responsibility. I just don’t agree with the concept of moral responsibility, for that form of responsibility rests on whether someone could actually have done otherwise, not on whether someone feels they could have done otherwise.

In the end, I think Sean is evading an important question—the one I raised above. Sure, we feel as if we could have chosen differently because we don’t have enough information to make an accurate prediction, but he doesn’t come to grips at all with the idea that given the laws of physics that underlie our behavior, there is no way we could have chosen differently. With complete information, everything is either predictable, or, if unpredictable, rests on quantum indeterminacy that has nothing to do with our will.

And that makes a ton of difference when you think about crime and punishment and when you take people to task for saying “they could have chosen otherwise”. Much of our legal system depends on an assumption of libertarian free will, not compatibilist free will. Certainly all retributive punishment does. And recognizing this fact can and should create big changes in both our judicial system and how harshly we judge other people.  Under hard determinism, people can be viewed as broken cars. When our car is broken, we don’t think it had a choice, but we do things like repairing it or, if it’s dangerous, taking it off the road. You don’t beat it with a sledgehammer for acting badly in line because its nature was to have a wonky carburetor and and broken transmission.

Recognizing the falsity of libertarian free will also leads to a lessening of self-rebuke. Telling yourself “If I had done X, Y wouldn’t have happened” is useful only in rewiring your brain so you wouldn’t do X again. It is not useful in beating yourself up for behaving in a way that you couldn’t have helped.

I wish Sean would take on the issue in all its fullness. Compatibilist free will is different in important ways from libertarian free will, and those differences have huge consequences. (For those who think that there’s no material difference, remember the surveys in which people who are asked whether we have moral responsibility in a deterministic universe mostly answer “no”.) Well, it’s time that they know that we do live in a deterministic universe. I wish Sean would tell people that we could not have behaved differently, even if we feel we could have!

Again, he’s right when he’s talking about everyday notions, but if at bottom libertarian free will is a total illusion, I wish Sean would say it straight out.

I’m not psychologizing Sean here, but I think the big love of compatibilist free will among philosophers comes from a fear of naturalism and a fear (expressed by Dan Dennett, among others), that if we abandon libertarian free will, as we should do explicitly, society will become totally immoral. In other words, the notion of compatibilism is there to keep us in line.

h/t: Barry

110 thoughts on “Sean Carroll on free will

  1. ” But he also admits that if we were able to predict perfectly what we would do, then “free will would go away.” ”

    That _is_ an interesting thought.

    And now I am thinking of other examples where the definition of “choose” is put to the test.

    Did Charlie Parker “choose” each note of his solos? Did he choose the sound? What else was there where it appears Parker “chose” any thing which we assume to identify as one thing?

    I find it simpler to conclude that no, he did not choose any apparently individual thing, and likewise, he did not choose any two things together.

  2. I would ask him this: We cannot measure the combination of velocity, gravity, friction, and bounce well enough to predict the outcome of a throw of dice, but does that mean that the dice have free will?

    1. Agree!

      I was thinking of an already-running internal combustion engine – where the illusion of “choice” will be traced back to the inventor of the engine, the dice, the tools used to make them, and so on.

      I was also trying to STOP COMMENTING ALREADY, but hey, it’s “free will”.

    2. No, dice don’t have free will because “free will” is a concept about human social interactions, so only applies to humans. (Other social concepts such as “morality” are similar. We hold a human “morally” responsible for killing another human, but wouldn’t hold a snake “morally” responsible for the same.)

      1. “Will” is, if I understand, dependent on “choice”.

        So are the dice an example of absence of “choice”?

        Are the two thrown dice “choosing” to land so the six faces up in the same way the dice thrower “chooses” two sixes by manually and visually turning them so six faces up on each?

        Degrees of freedom are in play, I think, but where are the “choices” in each scenario – without setting up an infinite regress?

        That is why “choice” is not parsimonious – the infinite regress.

        1. “Choice” is another socially-constructed concept that is used for decisions of highly-complicated calculating devices such as our brains (and, possibly, sufficiently advanced software). A lot of “human” terms are of this sort.

          1. I’m thinking of examples to compare :

            “Seeing” is what everyone/most humans agrees is going on when anyone “looks” at some thing. It is in reference to that by which we _assume_ what person A sees is what person B sees. But if person B is color blind, not exactly. If A has one eye, also not exactly. Details matter.

            Likewise, an optical sensor “sees” an object – but only in reference to what humans understand it is like to see.

            In detail – in the full explanation – the optical sensor, person A, and person B are largely bystanders to what is going on – they do not at each step exert a “will”, and they do not see equally at all.

            Thus “choice” and “will” are not uniform concepts, and are not parsimonious.

            I will TRY to stop commenting but this is very fun!… for me.

            1. Perhaps I can help with “will”. A person’s “will” is their intent to do something specific in the near or distant future.

              We typically choose what we will do. That choice sets our intent, and that intent motivates and directs our subsequent actions. For example, if we decide to have dinner at a restaurant, that chosen intent leads us to get in the car, drive to the restaurant, walk in, sit at the table, choose a meal, order it, eat it, and pay the cashier on the way out.

              Having completed that intention, we are free to decide what we will do next.

          2. Well, “choice” is a well-defined operation: choosing inputs two or more options, applies some criteria of comparative evaluation, and outputs a single choice. We observe it happening every time we walk into a restaurant. There are people reading a menu of the many things that they can order for dinner, and then deciding what they will order.

            We know it when we see it. Just like we know walking or bicycling when we see them happening. That these events are actually happening cannot be reasonably denied.

    3. No, no free will. When we throw dice, our possibilities of having exact data on velocity, rotation rate and exact nature of the material on where the dice touch down, air motion plus another host of physical properties that we would have to know with an unachievable precision, is beyond our possibilities to allow the prediction of the outcome. .It*s pure mechanics, but beyond our possibilities to predict the outcome accurately.

    4. Unfortunately that example is similar to asking “so if we separated a pair of dice, would they be lonely?”

      That’s a nonsense question because die don’t have the faculties that can cause feelings.

      Similarly, dice don’t have a “will” because: “As traditionally conceived, the will is the faculty of choice or decision, by which we determine which actions we shall perform.”

      Since dice don’t have minds, thoughts or desires asking if their “will” is free is nonsensical and not a “gotcha” question.

      People clearly have desires and make decisions based on attempting to fulfill those desires. Whether that can amount to “Free Will” needs to be aimed at agents that can be reasonably described as having a “Will.”

    1. Not with free will. I’m not sure why you’re asking, but all I can say is we don’t understand how it works, but it gives every sign of being a naturalistic process (i.e. we can take it away or alter it with drugs, bring it back, and trick people into thinking they’re doing something conscious when their brain is doing it.

      I sure wouldn’t explain it as a result of God’s actions, or of anything that doesn’t have to do with naturalism.

  3. I wish they could just do away with the term “free will”, which is so misleading, and is associated with so many clearly fictional concepts that lead people to feel morally justified in vendettas and cruel punishments, and just use something like “agency” or some similar term that makes it clear that we’re not just boulders rolling downhill, and that we can be “responsible” as PCC(E) points out, but that morality doesn’t apply to the past, but only to our desire to make better future choices. I really like how PCC(E) points out that we can learn from our mistakes to avoid future ones, but it’s pointless to perseverate and castigate (let alone flagellate) ourselves too much over the past, when we could no more do other than we did than a character in DVD can take a different course when you “rewind” the movie and play it again.

    1. Can we disconnect the word “free” from the word “will?” In other words, can we use the lone word “will” to label the subjective experience of stimulus-response in the organism cum environment?

  4. I apologize but I’m gripped by a brain storm :

    I was just thinking that the notion of choice, the word “choose” has been confusing to me. It is more sensible to me that, as a matter of experience, I am merely pushed around. Just because typing an “G” or a “1” on my keyboard fits the notion of “choosing” the G or 0 does not mean in granular detail I am choosing it. At every level I find myself wrapped up in choice upon choice : my fingertip, my computer, the electricity, the weather –

    So I find myself abandoning the notion of choice altogether.


    1. “You are not controlling the storm, and you are not lost in it. You are the storm.” -Sam Harris

      One of my favorite quoted from his book on Free Will. Your comment reminded me of it, so I shared it…through no free will of my own. ^_^

    2. “At every level I find myself wrapped up in choice upon choice : my fingertip, my computer, the electricity, the weather –”

      My belief is that intention and belief simply cannot be explained in terms of physics. Consider this example: Two people see an automobile accident, one pulls out his phone and dials 911 while the other rushes to the car to see if he can render first aid. Both have the intention of “helping an injured person”, and both must believe that the accident is real, not (say) a performance on a movie set. I would assert there is no way to bring the two actions into an equivalence relation without knowing the witnesses’ states of mind, and no way to interpret those states of mind without reference to concepts like “believe” and “intend”. At the levels of physics, chemistry and mechanics, the actions are simply different.

  5. Suppose the physical determinants of choice between either X or Y are essentially equal—so finely balanced, in other words, that a tiny signal or a tiny change in signal could determine the choice. Does such a situation fulfill the definition of a choice that could go either way? Furthermore, suppose that the final choice is determined by chance—by thermal noise in a single neuron. Does not that fulfill the definitions both of unpredictability, and of a choice that could have gone either way? I concede the point of comment #2, that unpredictability does not exactly equal free will. Or does it?

    1. A tiny signal is deterministic unless it reflects a quantum phenomenon. The problem with so many discussions of free will is that they conflate “predictability” with “conscious will”. Yes, choices could go either way IF a. quantum mechanics is purely indeterministic AND b. quantum fluctuations can ramify upwards to change behaviors. As far as we know, a. is correct to date but we know bupkes about (b)

      1. In my previous existence as a neuroscientist, I remember studying a paper about some neurons in which single miniature epsps (excitatory postsynaptic potentials) could trigger action potentials. Miniatures epsps are responses to the random release of a single vesicle of neurotransmitter at a synapse; in some cases they result from the random opening of a single calcium channel. I believe – but am not totally sure about this – that this random opening is a genuinely non-deterministic quantum affect. Perhaps a physicist can confirm or correct this.

  6. if at bottom libertarian free will is a total illusion, I wish Sean would say it straight out.

    First, as a compatibilist, I’ll say it straight out: libertarian free will does not exist. But:

    … recognizing this fact can and should create big changes in both our judicial system and how harshly we judge other people.

    Here I disagree, since I see our judicial system as being pragmatic at root (so it does not depend on an “assumption of libertarian free will”; such notions are, as i see it, a commentary about our judicial system).

    Jerry, your Hili Dialogs often discuss topical events, including punishment of offenders — everything from speeding car drivers hitting a herd of bison to Elizabeth Holmes of the Theranos fraud.

    And it seems to me that the appropriate punishments you suggest are usually pretty much in line with typical Democrat-voting-American attitudes. Which seems out of kilter with the idea that rejecting libertarian free will must entail major changes to the judicial system.

    1. The entire Anglo-American system of criminal jurisprudence presupposes the existence of libertarian free will. That’s why the blameworthiness (and, hence, punishment) of a defendant turns in large measure on the defendant’s mental state (mens rea) while committing the act that constitutes the crime — whether willfulness or knowledge or recklessness or negligence. It is also why in some cases (homicide, for example) it is considered an aggravating factor if the defendant acted with premeditation and calculation and a mitigating factor if the defendant acted with a substantially diminished mental capacity to appreciate the nature and wrongfulness of his or her conduct.

      1. “The entire Anglo-American system of criminal jurisprudence presupposes the existence of libertarian free will.”

        Isn’t “free will” for participants in the legal system defined _in_reference_ to some basic scenario that “free will” is similar to?

        “Choosing” right or left for instance – the left hand on the Bible, the right held up (do they do this anymore?), or replying when one’s name is called out. In _reference_ to that, the complexities can be built up.

        But the court room is not Nature. So I am comfortable that law itself is not under threat.

      2. You can straightforwardly re-interpret mens rea as being about susceptibility to deterrence (no libertarian free will needed). After all, the underlying point of a justice system is to deter bad behaviour, and that concept holds fine under compatibilism.

        1. Maybe where you live, Coel, but here in the U.S. the justice system seems be all about retribution, i.e., punishment. “If you hurt society, society will hurt you back”. Multiply harm. That’s what it’s for, and that’s what it does. People already know they don’t want to go to prison — if deterrence were the goal we wouldn’t have millions of incarcerated people.

      3. “The entire Anglo-American system of criminal jurisprudence presupposes the existence of libertarian free will.” then …”it is considered an aggravating factor if the defendant acted with premeditation and calculation.” I assume you are connecting these.
        But IMO “acting with premeditation” is not inconsistent with deterministic laws governing us/our “choices” and allowing for PCC’s “we couldn’t have done otherwise” under those conditions. We can still “feel” like we are making choices (moral or otherwise), even if our brain is in fact a “machine” writ broadly.
        Libertarian free will as most define it is an illusion, yet our justice system still stands.
        I really can’t find the fundamental conflict in all of this.

  7. Just because I’m a liberal doesn’t mean that I don’t want changes in the judicial system. In fact, I’m not sure what you mean because there are plenty of people like me who see that retributive punishment is flawed and needs reform. And surely you must recognize than when someone has a deterministic “excuse”, like a brain tumor or having been abused or being mentally ill, they are punished in very different ways from normal offenders. You recognize, right, that the former are assumed to have not made a free choice while the latter ones are assumed to have made a free choice?

    1. I’m not sure what you mean …

      I mean that, judging from the Hili Dialogs, I don’t see much difference between the punishments that a typical Democrat-voting American who believes in libertarian free will would call for and those that you would call for.

      1. But you’re judging from the Hili dialogues (you don’t mean just what the cat says, right?). Well, neither of us knows the data, but I’m sure you’ll find some Democrats who call for capital punishment, and there are lots of Democrats who think people are morally responsible and should be punished on that basis. How many of them would approve a 21-year sentence and a pretty cushy cell for Anders Breivik, the Norwegian guy who murdered dozens of people? It’s not just punishments, either: I’ll bet that lots of Democrats hold people responsible for the “crime” of being poor.

        1. I recently read The Tyranny of Merit : What’s Become of the Common Good? (by Michael Sandel), which directly addresses this issue and many related ones (some surprisingly so). It’s led me to usefully examine my attitudes on a wide range of issues around responsibility, fairness, talent, and more.

          Very highly recommend.

  8. Let us disabuse ourselves of the notion that there is some kind of interior homunculus that wills. This is a hindering notion lingering as a vestige of the discredited belief in the soul.

    1. You’ve convinced me. Now if you could just convince the interior homunculus in my skull — the one standing at the helm, with one hand on the steering wheel, the other on the throttle, tracking the visual display, listening to the speakers broadcasting from the ears, and monitoring the readouts from the other three physical senses. 🙂

    1. That’s not bad, either. So long as we don’t understand the physical basis of our desires, character, and actions, we have free will. The minute we figure something out, free will goes out the window–at least a bit.

  9. I think a good analogy is weather. It’s too complex to predict except in very broad strokes (kind of like Jerry’s binary choice example). People used to think it was driven by gods or prayers or ancestors, etc., but now we know that it’s a product of physical forces.
    It’s still common to say things like “Act of God” for these types of events – and some folks actually believe that literally – but most understand it’s not true.
    Sean points out that, for communication, it carries meaning to say “I chose” – but he recognizes that this was an event driven by forces.
    My guess is that this is semantics. I think Sean still sees value in keeping free will language around as long as our understanding/prediction of the mind’s workings is so limited. As our understanding deepens, I would bet Sean would be amenable to whatever new descriptions are accurate.

    1. It’s not just semantic if one fully absorbs (though Sean surely has) that we really had no choice. For a LOT of people think that if libertarian free will is an illusion, moral responsibility goes out the window, and everything is permitted. It’s only by people telling us that we have free will that, say compatibilist philosophers, society is kept in line.

      The good thing is that Carroll realizes that libertarian choice is “illusory”. The bad thing is that he avoids saying that explicitly, nor works out the consequences of naturalism. At least Sabine Hossenfelder admits that we’re more or less meat robots.

      1. Why I think this is semantic… another analogy :does the Google algorithm “choose” what to put in your search results? I think most people would agree that the answer is yes, while still understanding that, presented with the same options the Google algorithm would make the same “choice” every time. There seems to me to be room in the use in the word for both libertarian and deterministic world views to correctly use the word “choose”. Defining “choice” as, “if I replayed the tape of the universe I might make a different selection”, then your point stands, but I think Sean is using a broader definition of the word.

        Sean says explicitly that it is possible to predict choices the same way that Laplace’s daemon allows us to predict the universe – it’s a hypothetical that is beyond the range of our capability. If this isn’t an explicit acknowledgement the libertarian choice is “illusory”, then I don’t know what is. He adds that, could we ever get that level of prediction, then “free will” would go away. So it seems we all agree on the underlying truth of deterministic “free will”. Again, we can still talk about human “choices” the same we talk about Google making choices (you don’t even need emergence to do this).

        The moral responsibility argument is a separate, related topic about the consequences. I agree that arguing backwards, “I don’t like the consequences of the truth, so I’ll argue against it”, is really terrible logic – but that’s not what Sean is advocating (at least not in this clip).

  10. This seems to be entirely a semantic issue.

    – we/our behaviors are governed by deterministic laws

    – we can never use these laws to actually predict what we will do/”choose” due to their being too complex and the initial conditions requiring far too much information

    – for any apparent “choice,” we could not have “chosen differently”

    All of these statements can be true at once. If it floats your boat, call it “compatibilist.”

    What is the issue??

    I’ve pretty much removed “free will” and consciousness from my list of foundational problems needing explanation.

    1. It’s not semantic, as I argue above. One’s notions about agency play into all sorts of views, not the least how the law works. Why else is “not guilty by reason of insanity” a widely-accepted verdict? Because an “insane” person is said to lack the ability to have done other than what he/she did.

        1. Oh for crying out loud, you haven’t followed my writings on this at all. There are very good reasons to convict sane people of a crime: to keep them away from society, to reform them, and to act as a deterrent. Go read “free will” post on this site before you make remarks like that.

          1. I read most of your books, I thank you for your contribution to 10% of society that is a freethinking people. However, your claim that the insanity defense is an appropriate and legal defense for the “insane”….How about the “sane” college graduate who murdered 4 innocent kids in Iowa…could he be “insane” and walk free?

            1. I’m sorry but you keep getting things wrong. I did not say that the insanity defense was appropriate and legal, I said it was USED as a defense, and that implies that one can be excused from criminal punishment on grounds that one had no choice.

              Please stop making me correct you. Just go back and read some of the posts here on free will. You don’t seem to comprehend what I’m saying here, and it’s not hard to comprehend.

      1. Still semantics IMO, just goes to what “ability” means. A “sane” person (presumably possessing a mind functioning “normally” according to society) can “choose” to do other than they did only insofar has that person’s inputs and brain calculations would cause them to perceive making a choice. The “insane” person does something on account of their brain functioning differently than the sane person’s. Besides the outcome, they may also internally perceive the sense of “choice” differently (i.e., maybe not being aware of the possibility.). It’s still all a wet computer experiencing its environment, doing a calculation, and acting.
        (The question of “sentience” and our self-awareness and sense of choice are, IMO, nothing mysterious or requiring some sort of special sauce. Pinker makes this point well. But I think this is where people get hung up conceptually.)

        Whether or not this view has or should have an impact on jurisprudence, and what that impact should be, is a separate question. But this is a one way street: whether or not it does/should impact jurisprudence has *nothing to do* with what is actually happening in our physical minds. That is what it is, and has to be primary in this discussion.

  11. Re: Sean Carroll. He endorsed this chart claiming to show why sex is a spectrum which lacks any intellectual rigor. #1: if it’s a spectrum they don’t define the x axis for what makes someone more or less of either sex. And overall, it’s a mess, with random disorders put onto a line with impressive graphics, but without any coherent framework or argument for why disorders means that the definition of sex relying on which gamete our reproductive systems are organized around producing at any point in the lifecycle is at all deficient.
    IOW, whatever substantive Physics work he may have done, I question his intellectual honesty.

    1. I don’t question his intellectual honesty at all; he’s not a biologist and didn’t perhaps fully grasp that chart. And I did criticize him for making that tweet. I would take back that questioning of intellectual honesty which isn’t at all justified.

  12. I’d always had a problem with the notion of free will. In the last 10 years or so arguments from the host, Sam Harris, neuroscience etc. turned me into a true believer in no free will.
    NYC (FL)

  13. Any discussion of free will needs to address the issue that we have no model for how human consciousness works. One (strong) possibility is that it depends on physics that are simply beyond are comprehension. This isn’t so weird, as if you consider the possibility that our notions of time and space are artifacts of some more fundamental physical reality, it would follow that our psyches, which evolved in “our world’s” manifestation of time and space simply wouldn’t be able to grapple with a physics outside of these phenomenon. (Also, this physics would be unobservable by us.)
    So, what we perceive as “free will” may exist but just rely on the same inexplicable physics that’s outside of our notions of time (particularly).

  14. In *Fundamentals: Ten Keys to Reality,” Frank Wilczek discusses a relevant experiment:

    Using a technique known as transcranial magnetic stimulation (TMS), it is possible to stimulate the left or right brain motor centers in a subject’s brain, at the experimenter’s discretion. A properly sculpted TMS signal to the right motor center will cause a twitch of the left wrist, while a properly sculpted TMS signal to the left motor center will cause a twitch of the right wrist. Alvaro Pascual-Leone used this technique ingeniously in a simple experiment that has profound implications. He asked subjects, upon receiving a cue, to decide whether they wanted to twitch their right or their left wrist. Then they were instructed to act out their intention upon receiving an additional cue. The subjects were in a brain scanner, so the experimenter could watch their motor areas preparing the twitch. If they had decided to twitch their right wrist, their left motor area was active; if they decided to twitch their left wrist, their right motor area was active. It was possible, in this way, to predict what choice had been made before any motion occurred.

    Now comes a revealing twist. Occasionally Pascual-Leone would apply a TMS signal to contradict (and, it turns out, override) the subject’s choice. The subject’s twitch would then be the one that TMS imposed, rather than the one he or she originally chose. The remarkable thing is how the subjects explained what had happened. They did not report that some external force had possessed them. Rather, they said, “I changed my mind.”

  15. I agree that Sean’s analysis is flawed, but locate the flaw in a different place:

    As {Sean] says, this is because, “given the actual information you have about yourself, you could have acted differently, because the information you have yourself is wildly incomplete”.

    This gets it backward. Your information about yourself is wildly incomplete, but even in the best of all possible scientific positions, your ability to predict your next act would be limited because you have yet to decide what to do. You could never predict your next act, no matter how much molecule-by-molecule information you have. You have to decide it. And that’s why it’s correct to say you have free will, and could have done otherwise.

    Interestingly, this logic doesn’t in principle prevent someone else from predicting what you will do. But that just makes them a reporter; you’re still the decider. It doesn’t matter in principle whether the report comes earlier or later – although, in practice it does, because in practice only after-the-decision reports are reliable. Relatedly, note that Laplace’s Demon is not only practically impossible, but physically impossible.

    1. I am not getting either version. Neither the incompleteness of the information we act on, nor the time constraint on the decision making, have any relevance to our lack of libertarian volition. Without libertarian volition there is no free will.

      1. My point was not about time constraint, but was about the logic of self-reference. If you are offered the choice of coffee, tea, or neither, and you say “I’ll choose tea”, your statement is self-validating. No amount of molecular data could force you to another conclusion, no matter how quickly you could do the calculation.

        And “Without libertarian volition there is no free will” is an assertion in need of justification.

    2. Plus even with very incomplete information all you need to know is which switch Pascual-Leone has turned on (left vs. right TMS) then you could predict both your next act (right vs. left hand twitch) and your interpretation of the act (whether or not you changed your mind), all without any possibility that you could have done otherwise (because of the TMS).

      1. I don’t think a stimulus that bypasses an agents beliefs and desires to cause a muscle twitch, counts as an “act” in the relevant sense. If you tase me, my twitching and seizing is not under my control, but this doesn’t enlighten us about free will.

      2. I think the point is that one’s conscious awareness of having made a decision would not be any different with or without the TMS, but with the TMS a third-party observer could know that in fact no decision was made. That’s the sense in which there is no will (free or otherwise) involved in determining the action.

        1. The experiment does demonstrate that illusions-of-will can occur. Yet, plenty of experiments demonstrate that optical illusions occur. Few people take this to show that all visual experience is illusory, and that nothing we think we see is real.

  16. My view is that this *is* a semantic issue. Because it all depends on what one means by free will. If it means ‘could have done otherwise in those precise circumstances’ then compatibilists do not believe in free will. And I speak as a compatibilist. We do not believe in libertarian (or, as it’s also sometimes called, originalist or contra-causal) free will. Not believing in that is what defines us.

    If free will means acting in line with one’s own volitions then compatibilists do believe we have that sort of free will. And I think that Professor Ceiling Cat, though he wouldn’t use the *phrase* free will, agrees that we do sometimes do things we want to, and those are the acts we are *responsible* for. In this sense, free will and responsibility are synonyms.

    PCC notes that compatibilists often aren’t emphatic enough that one could never have done otherwise than one what did. If that is true, then compatibilists are being insufficiently compatibilist! Compatibilists are determinists (by definition) and this ought to inform the policies we advocate on reward and punishment, praise and blame. For the compatibilist, punishment and retribution should be ditched.

    1. Sorry, just to clarify that last point: the only justifications for punishment would be deterrence or incapacitation – not retribution.

      1. I disagree because when a choice is “free” is not often so easy. What if a father wants to spend money on a new car but has to put that money towards his kid’s college? Is that a “free choice”?

      1. Ah. Well, I think this confirms that this is a semantic issue. You call it choice, I call it will. But at least we agree it’s determined.

        1. This is a communication issue. Some of us want the additional precision and clarity regarding the universal lack of libertarian volition that is lost when the phrase “free will” is retained as a poetic substitute for free choice.

    2. I disagree with your statement that “compatibilists [who] aren’t emphatic enough that one could never have done otherwise than one what did … are being insufficiently compatibilist.” There is a substantive, non-verbal question about what should be held constant when considering whether you could have done otherwise. I would point out that only facts which hold independent of your decision should be held constant in this inquiry.

      Now, you might say: OK, but every fact prior to your decision, even microscopic details, holds independent of your decision. Now, that’s a scientific question about how causality works. And you would be wrong about that – as I explain on my website which is mine. (The essay has links at the bottom to other parts of a larger argument.)

  17. You’ve just read “All the Light We Cannot See”. Andrew Doerr writes that all we have in this world is “chance and physics.” He pretty much nails it right there.

  18. “. . .if at bottom libertarian free will is a total illusion, I wish Sean would say it straight out.”

    Proposed title for Sean Carroll’s next book: Why Convolution Is True. 😊

  19. I try to read each of the posts on this subject a couple of times, although it is far from my areas of expertise.
    It does seem more like philosophy than science.

    I am not a fatalist. Even if I was, there isn’t anything I could do about it.*

    It runs counter to my Dad’s philosophy, which has always been that a person’s success and happiness are largely determined by the series of choices they make throughout their lives.

    It also seems to rely on the assumption that randomness is not possible. That seems like a physics question to me.
    What I mean by randomness is that we start with the belief that if you constructed two universes, identical to the atom, they would develop over time in exactly the same way, with Max writing this sentence at exactly the same time from the origin.
    If there is even a little randomness, then the universes diverge progressively. At the same time, you have less confidence that a given set of inputs will yield the exact same result every time, or that a person will make the same decision every time they are presented with the exactly the same options.

    I enjoy good science fiction relating to time travel, where versions of this principal are explored in a variety of ways.

    *Emo Phillips

    1. What you say is true; even the Big Bang was rife with quantum effects. BUT the issue here is whether one’s WILL, independent of physical circumstances, can make you behave in two or more different ways given a starting condition. Note that I said “will” instead of “quantum indetermiacy”; they are not of course the same thing.

      1. Sabine Hossenfelder has some interesting commentary on quantum mechanics and “free will.” (You’d probably like her videos on this topic.) She allows for randomness in some neural calculations due to quantum fluctuations, at a level, and with what actual effect, no one knows for sure (but probably not very significant given the scale of neuronal activity). But in the end “determinism with some randomness thrown in” is irrelevant and not related to the sense of “choice” or “free will,” and does nothing to argue for compatibilism.
        IMO, quantum effects are another red herring in all these discussions.

      2. I am trying to understand the arguments. It seems to me that consciousness itself is sort of an electro-mechanical process, and subject to the same sorts of unpredictability as any other complex system.

        Perhaps, as an intellectual exercise, you had a time machine as it is conventionally imagined, where you can travel back on our timeline, but once there, interacting with the world has cascading results. Pick a time and place where someone performed an action, but observe passively. The question is whether the person would always perform the action, no matter how many times you went back to observe. If you watch the Zapruder film a million times, Kennedy will always be shot.
        My thought is that randomness might affect people’s choices as well. If you go back on our timeline to 1200 on 11/22/63, shots will almost certainly ring out half an hour later. If you went back to 1953 and waited a decade, the shooting might be slightly less likely. 1753, even more so.

        I do not have the background to discuss quantum anything, so I am stuck with analogy. The nature of “will” seems to be an issue here as well. Admittedly, I am just a guy in a big hat sitting by a fire, so my lines of reasoning may be unsound.

        1. Rewind the clock. Start the clock. Everything remains the same as it was. Can someone change the behavior they previously carried out at that instant? By what supernatural magic do they behave differently absent any changes to the past and revisited present, disregarding any unpredictability potentially introduced by quantum uncertainty? Will is, by definition, under the agent’s complete self-control.

    2. It can take awhile to grasp the concept that we are biological/metabolic machines lacking the magical, supernatural capability to transcend the laws of nature as required to actually have free will. Deities have free will, but deities are fictional. Scientists who are ontological naturalists will have a difficult time defending free will by any means other than redefining having free will as being almost indistinguishable from having free choice.

  20. “But he also admits that if we were able to predict perfectly what we would do, then “free will would go away.”

    It is that last sentence that lays bare what I see as the problems with Sean’s argument. “

    I agree with your analysis, Jerry.

    Carroll seems to build up a case for free will and then give it away at the last moment with that concession. I’ve never found the argument that free will can be based on our insufficient knowledge to be a good one. We may not know exactly what choice we’ll make, but as you say we DO know that it will be just as determined as the last one we made. And that right there is enough (for most people) to threaten the idea of free will.

    I think if Carroll had stuck to a case based more on emergence and the role of counterfactuals, it would have been better (to me). But he at least implies that counterfactuals are only useful in the case where we lack knowledge of “what will happen.” But that doesn’t seem right to me. Counterfactuals are useful and true even IF we know what would happen, just as they are useful to think about them regarding things we already know happened. The glass of water I placed in my freezer froze solid. I have knowledge about what happened and that is never changing. But it is still tells us something about the nature of water to say “but IF I had left the water in the cup on the counter, it would have remained liquid (and eventually evaporated).”

    (I’m not gonna stick around and argue for Free Will though…)

  21. Thanks to the arrow of time, what we might* have done in the past is moot speculation.
    * there’s no ‘could’ about it – we can’t unchoose a historic choice.

  22. In the US there are an extraordinary high number of people in prison under very bad living-conditions and the US has still high crime rates. In Europe we see the same thing; compare UK and Wales with Norway. Deterrence seems not to work as expected and might not be an important factor in respect to keeping crime-rates low.

    We know for ages that delayed punishments are not very effective.

  23. given the actual information you have about yourself, you could have acted differently, because the information you have yourself is wildly incomplete”. And that is true as well.

    I think better would be: ‘ …you could have IMAGINE acted differently with incomplete information

  24. Excellent outstanding comment/article of Jerry on footage of Sean! I could not agree more with Jerry!!👍

  25. I’ve thought about this a lot over more than two decades. Here’s my take: I agree with what, I understand, both you and Sean Carroll are saying and that is that, at its core, reality can be thought of as being understood at different levels of interpretation. The universe can be thought of as particles interacting according to the laws of quantum mechanics; at a higher level, atoms reacting to electromagnetic forces producing chemistry; higher still, we can think of celestial objects moving according to Newtonian mechanics/general relativity. All these levels are valid ways of understanding what is going on, but at some point, during the evolutionary history of life on our planet there was a key moment when animals first acquired the capacity for feeling, I think of this as the Nova Point, because I have speculated that this was the moment when the first animal to acquire subjective experiencing, or what scientists call minimal animal consciousness, an animal I’ve called Nova, appeared. This creature crossed what I’ve called the Awareness Horizon. The Nova Point initiated a new level of interpretation and arguably the only important one, because it was the beginning of consciousness, and the only level of interpretation where meaningful values can be said to inhere.
    So far, so good, for hard determinism because each level of interpretation supervenes on the one below it, and because there is no evidence of anything going on outside of this system hard determinism must be valid. However, this is where a “but” comes in. While it must be true that it could never actually be true that anyone could choose differently, it is also true that no one exists in isolation, and some decisions we make depend on the actions of other people, social institutions and culturally defined norms.
    No one can choose their genetic makeup, their upbringing, or the specific complexity of brain wiring during their development, and these influences taken together define the person someone is, so no one can be blamed for who they are, nor can they be blamed for what that kind of person does, even if they are a monstrously uncaring serial killer. The reason for incarcerating such a person must be to prevent their reoffending, not to punish them for their offending behaviour, because it can’t be their fault that they are who they are. (A very good reason for opposing capital punishment). However, even if none of us can, in truth, choose differently, many decisions are made according to their particular circumstances, the people around them, and the societal milieu in which they find themselves. If someone knows that they might be fined for speeding, they might slow down at a dangerous bend and prevent an accident. So, in this sense I agree with Carroll that as human beings, all we can do, is to act “as if” we could choose otherwise.
    I have published a podcast (one of a series of eight) that describe the animal Nova must have been and the implications of its evolution in terms of meaning in the universe. Overall, the series summarise a philosophical point of view, with the Awareness Horizon at its core, and which argues strongly for the importance of evolutionary theory in understanding what it is to be human. If anyone’s interested, here’s a link to the Nova Point Podcast:

  26. I can explain why your acquaintances “are astounded to hear that they could not have done otherwise”. It’s called “cognitive dissonance”.

    Suppose, you tell your child “I have two ice cream cones, chocolate and vanilla. You can choose either one, and I will have the other.” Then the child chooses the chocolate. Now, if you say to the child, “You could not have chosen the vanilla”, and your child is bright, she will say, “Either you were lying at first, or you are lying now. You told me I could choose the vanilla. Now you are saying I could not have chosen it! One of those must be false, because they directly contradict each other.”

    And the child is correct, of course.

    You’ve made a small error in your logic. You’ve equated what “can” happen with what “will” happen. You’ve assumed that if something “would not happen, then it “could not have happened”.

    To eliminate the dissonance, try replacing “could not” with “would not”. If you say to your child “You would not have chosen the vanilla”, she will accept that. She knows that she prefers chocolate, and that given the same options, she would always choose chocolate. So, “you would not have chosen vanilla” does not produce any cognitive dissonance.

    But she will not accept the notion that she “could not” have chosen the vanilla, because, after all, you just finished telling her that she could choose the vanilla, just before she chose the chocolate.

    And this is always true whenever choosing shows up in a causal chain. In order to make a choice, we must first believe that we actually have two things that we can choose between. From the two things we “can” choose, we select the single thing that we “will” choose.

    Thus, at the end of any choosing event, we will have the single thing that we “would” choose, and at least one other thing that we “could have” chosen, but didn’t.

    The logical blunder has unfortunately become ingrained in the common definition of determinism, as implying that we “could not have done otherwise”, when all that is truly implied is that we “would not have done otherwise”.

    When we say that we “could have chosen vanilla” we are implying that (a) we did not choose it and (b) that we only would have chosen it if circumstances were different. And this comports with the facts of the matter.

    She could have chosen vanilla, but she would not have chosen it. This statement is TRUE in both its parts.

    1. The discussion about free will sounds much like a scientist’s version of creationism, substituting physics for God. We can not mathematically model the system, we can not make predictions based on any known models, and we can never predict what will happen because it is too complicated.

      On a separate note, if we did known all the relevant information and could mathematically model and make predictions on future choices, should we lock people away for a crime that individual will commit before the actual illegal action is perpetrated?

      1. Ironically, praise/blame and reward/punishment are deterministic methods of behavior modification. They are a response to behavior that we find either beneficial (which we’d like to encourage) or harmful (which we’d like to discourage.

        If, as you suggest, we could eventually predict that a person would commit a crime, then we would intervene early to prevent it, and start rehab in advance.

        The only reason for locking someone away, either in a prison or in a secure medical facility, is to prevent them from harming someone or themselves.

        When considering “just deserts”, expand it to what treatment someone would justly deserve. Because the aim of Justice is to protect everyone’s rights, a just penalty would seek to (a) repair the harm to the victim if possible, (b) correct the offender’s future behavior if corrigible, (c) secure the offender to protect others until his behavior is corrected, and (d) do no more harm to the offender or his rights than is reasonably required to accomplish (a), (b), and (c).

        Morality, which seeks the best good and least harm for everyone, would oppose any unnecessary harm.

        1. Given the limitations of this form of communication, I ask sincerely if you are advocating for forced rehabilitation of individuals who have yet to commit a crime?

          1. Trew, You posed a scenario in which we had perfect knowledge of all future events. Are you now trying to pin me down to an answer that you’ve clearly taken outside of the context you suggested?

            By the way. We already force rehabilitation upon people before they have committed any crime. It is called “raising a child”.

            1. Sorry for the confusion. I stipulated that we know “all the relevant information and could mathematically model and make predictions on future choices”. I did not mean to imply perfect knowledge. Like any complex model, there is a degree of uncertainty. Since we are talking about physics, we will estimate a certainty level of 4 – 5 sigma for the equation. Given that level of confidence, should we use these predictions for rehabilitation purposes.

    2. OK, I couldn’t resist :

      ” Suppose, you tell your child “I have two ice cream cones, chocolate and vanilla. You can choose either one, and I will have the other.” Then the child chooses the chocolate. ”

      TL;DR : does a “either vanilla or chocolate and no other flavor ice cream” store (or equivalent) actually exist? Yes, I think that matters, as silly as it is.

      When does this choice ever occur in this way as put forth above in a human-to-human context? Where it is not a shell game where yet another choice subsumes the one to which our attention is directed – e.g. the moment the adult offered to go to a weird ice cream store?

      If we invent a theory with no observation to back it up, what does that say about the theory?

      I know I got in a spat before about courtrooms and Nature, so I’ll try to cool it.

      1. Thyroid, it is a simple example that makes a simple point. There’s nothing tricky about it. Your wife and daughter are in the car while you run into the ice cream store to buy a couple of cones. You can’t remember whether your daughter prefers chocolate or vanilla, so you buy one of each. Then you tell her that she can choose either one and you’ll take the other.

        Having told her that she can choose the vanilla, it creates cognitive dissonance when you turn around and insist that she could not have chosen the vanilla, because it is an obvious lie.

        She could have chosen the vanilla. But, since she preferred chocolate, she never would have chosen vanilla. The issue here is whether you understand the difference between “could have” and “would have”.

  27. The only way for free will to be free is for the will to exist outside the laws of physics. There’s really no way around this except by defining “free will” as something that it really isn’t—which is what the proponents seem to do.

    If we accept that the laws of physics apply to everything, then our behavior at any particular instant of time is determined by the actions of the already moving atoms and molecules involved—and is not up to us. One cannot suspend those ongoing molecular and atomic actions, “decide” during that miraculous suspension how we are going to act, and then turn those molecular and atomic movements back on again so we can execute the act that we decided during that miracle interval. That would be tantamount to saying that sentient beings have ability to suspend the laws of physics. That seems ridiculous to me.

    1. Norman, It is all a matter of how the atoms are organized. If you organize the atoms into an inanimate object, then its behavior will be governed by physical forces. Place a bowling ball on a slope and it will always roll downhill.

      But suppose we organize the atoms into a living organism, such as a squirrel. Place a squirrel on that same slope and he may go uphill, down, or any other direction where he hopes to find his next acorn. While still “affected” by gravity, he is no longer “governed” by it. Instead, he is governed by biological drives to survive, thrive, and reproduce.

      Evolve a living organism with an intelligent brain, capable of imagination, evaluation, and decision making, and its behavior will still be affected by gravity, and affected by biological drives, but its behavior will now be “governed” by calculation and choice. It gets to choose when, where, and how it will go about satisfying its biological drives.

      Now, if you want to reduce everything to the behavior of atoms, you could make a list of the position and trajectory of every atom at time 1, then make a second list of their position and trajectory at time 2, then make a third list of the differences between list 11 and list 2. You would have lots of data, but you would have no clue what to make of it. You would have plenty of raw data, but no useful information.

      So, we have more than the Physical sciences to help us explain what’s going on. We also have the Life sciences and the Social sciences.

      Physics can easily explain why a cup of water, poured on a slope, will flow downhill. But physics has no clue as to why a similar cup of water, heated, and mixed with a little coffee, suddenly jumps into an automobile and goes grocery shopping.

  28. Late to the party. I missed this discussion when Jerry first posted the article. Just came across the Big Think video myself today and sent a link to Jerry because I thought he’d be interested!

    Something that stood out to me is that he starts off by explaining that he doesn’t like the term “free will” and thinks it should be discarded because it carries too much baggage (my characterization)! That always used to be a key point, heck the key point, of the Incompatibilist position. Given this I really don’t see much left that an incompatibilist could argue with about Sean’s view. Well, except that he does make some comments about his understanding of the incompatibilist view that I don’t think are accurate, leaving little doubt that he would reject my claim.

    Another interesting thing is that he explains that in his view determinism is a category error with respect to free will and that all that matters is that there are laws, not what the laws are. Rather than the traditional (I thought) meanings of Compatabilism = free will is compatible with determinism and Incompatibilism = free will is incompatible with determinism, he says that determinism is the wrong metric. It should be “the laws of physics, whatever they may be.” I’ve noted that for quite some time now Jerry has also made this clarification when writing about free will.

    Sean goes on to say that if someone starts talking about determinism in a discussion about free will that he knows that they’ve missed the point. It happens that I completely agree with that and long since came to that view, but I hadn’t heard Sean say so before. I know that in an older article he wrote that “determinism is true, at least locally, and that is all that matters for the issue of free will” (my paraphrase). Of course, perhaps his views on the subject have evolved. Mine certainly have!

    Off to read through the comments.

  29. I think that a better argument against free will is Cosmic Evolution, beginning with the Big Bang and its precursors. Ideas of evolution incorporate ideas of randomness and unpredictability. Instead of calling myself a “hard” determinist, I now prefer to call myself a Cosmic Evolutionist.
    As to religious concepts of free will, those seem to be a big ball of worms. They seem to boil down to the god or gods giving you free will so you can screw up and be saved in the end.

    1. OK, but where’s the natural selection in that, and how is it distinguished from biological natural selection?

      1. Try reading some cosmology (perhaps you have). Books like “A Universe from Nothing” by bad boy Lawrence Krauss, or “Until the End of Time” by Brian Greene. There’s a lot of “stuff” in the universe and a lot of evolution still going on, beginning with stars, galaxies and planets, and continuing with the origin and evolution of life.

Leave a Reply