Christian List on free will

For many people, the idea of free will is bound up with the notion of “could have done otherwise”. By their lights, if only one future is possible for a person — that is, if the person cannot do otherwise — then free will is an illusion.

Philosopher Christian List — author of the recent book Why Free Will is Real — proposes an interesting species of free will based on the claim that while physics may be deterministic, behaviors at the agent level are not. Agents can do otherwise, according to List, and this is enough to ground free will even if physics is deterministic.

I think List is mistaken, but I’ll save my criticisms for the comment thread.

Readers can find List’s argument in this paper:

Free Will, Determinism, and the Possibility of Doing Otherwise

See you in the comment thread.

756 thoughts on “Christian List on free will

  1. Thanks for posting this. I’m in LA and don’t want to read it on my phone, but i’ll have a look when i get home. Looks like it might be interesting.

  2. colewd,

    Can you start off with a working definition of free will we can use?

    Here’s what List is aiming at. From the paper:

    Free will requires that (at the time of interest) more than one alternative course of action is possible for the agent.

  3. “1. The problem
    A standard argument against the compatibility of free will and determinism is the
    following.
    2
    Premise 1: A necessary condition for someone’s action to count as free is
    that the agent can do otherwise.
    Premise 2: Determinism implies that the agent cannot do otherwise.
    Conclusion: Either there are no free actions, or determinism is false (or both).”

    Scenario 1:
    Jesus is about to be captured and killed but he escapes because it is not his time to die, as per the MANY prophecies…
    The opposers of Jesus had all the available options to capture and kill Jesus, like stone him etc and yet there were not able to exercise their free will because it was restricted by a predetermined prophecy that was suppose to meet the criteria such as date, time, kind of death etc…

  4. List’s argument is best understood in reference to two of his figures (reproduced below).

  5. The small dots in Figure 1 represent possible physical states of the agent. (Technically they include the state of the environment too, but let’s ignore that for the sake of simplicity.) The rectangles represent possible psychological states of the agent. Some rectangles contain more than one small dot, which means that a single psychological state can be realized by multiple physical states.

    Determinism is assumed to be true, so each physical state leads to one (and only one) successor state. This means that two branches can never leave a single small dot. (Two branches can converge on a single dot, however.)

    Now take a step back and look at the agent in terms of his or her psychological states (the large dots in Figure 2). At this level, it’s no longer true that each state leads to one and only one successor state, because psychological states may have multiple physical realizations. That’s why you’ll see more than one branch leaving a large dot.

    List takes this to mean that at the psychological level, determinism is false. Therefore, by his reasoning, an agent can do otherwise at the psychological level, and free will is therefore true at that level.

    I think he’s confused, but I’ll withhold my criticism until readers have had some time to digest List’s argument.

  6. keiths,

    Doesn’t that assume that a psychological state can correspond to many physical states simultaneously?

  7. dazz,

    Doesn’t that assume that a psychological state can correspond to many physical states simultaneously?

    Either that, or List is confusing epistemic possibility with metaphysical possibility. His argument fails either way.

  8. To elaborate: two subjects may be in the same psychological states but different physical states. Their next psychological state is still fully determined by the next physical one, because a physical state always corresponds to a single psychological state under that model

  9. dazz,

    Right. The next physical state is determined, and therefore so is the next psychological state.

    If you only know the psychological state and not the exact physical state, then it appears that more than one future state is possible. But that’s only due to your lack of knowledge. It’s an epistemic possibility, not a metaphysical one.

  10. keiths:

    Either that, or List is confusing epistemic possibility with metaphysical possibility.His argument fails either way.

    I read him as dealing with the metaphysics by combining NRP with the naturalistic ontological attitude and then applying that combination to the science of psychology.

    Why do you think those arguments fail?

    (ETA: Deleted a misdirected reply)

  11. dazz:
    To elaborate: two subjects may be in the same psychological states but different physical states. Their next psychological state is still fully determined by the next physical one, because a physical state always corresponds to a single psychological state under that model

    You are mixing determination in physics explanation with determination in psychological explanation. He rejects that mixing via NRP and the NOA.

  12. Bruce,

    I read him as dealing with the metaphysics by combining NRP with the naturalistic ontological attitude and then applying that combination to the science of psychology.

    Why do you think those arguments fail?

    List believes that each physical state corresponds to exactly one psychological state, and that each physical state has exactly one successor. Given those two premises, it’s guaranteed that the future is fixed at both the physical and the psychological levels.

    You can only get indeterminism at the psychological level by ignoring the supervenience.

  13. BruceS: You are mixing determination in physics explanation with determination in psychological explanation.He rejects that mixing via NRP and the NOA.

    I have no idea what any of that means, but I suspected it couldn’t be that easy. Thanks, Bruce. I’ll read List’s paper, but it will probably go over my head

  14. dazz,

    Don’t give up so easily. You were actually right.

    Now let’s see if I can convince Bruce of that. 🙂

  15. Bruce,

    To elaborate on what I said above, let me add a couple of points:

    1) The naturalistic ontological attitude requires us to accept the reality of psychological states, but it doesn’t require us to deny the physical states upon which those psychological states supervene.

    2) Even if non-reductive physicalism is true, the supervenience relationship still holds.

    Let s1 and s2 be physical states, with s2 as the unique successor of s1. Let p1 and p2 be psychological states that supervene on (at least) s1 and s2, respectively.

    If the current physical state is s1, then the current psychological state is p1. The next physical state must be s2, and therefore the next psychological state must be p2.

    The future is determined at both levels.

  16. List writes:

    A central concept of any version of decision or game theory, whether we take the original versions of von Neumann and Morgenstern, Nash, and Savage or their latest, psychologically more advanced incarnations, is an agent’s set of possible actions or strategies. The decision-theoretic or game-theoretic explanation of many social phenomena relies crucially on the assumption that the agents’ action-or-strategy sets contain more than one option. Sometimes the addition or removal of options can make a significant difference to what the agents are predicted to do even if these options are not ultimately chosen. Unless we accept that there is at least a thin, technical sense in which such options could have been chosen, it is hard to make sense of those effects.

    This seems wrong to me. The additional options have an effect simply by virtue of being considered. It’s not necessary for them to be metaphysically possible.

  17. Accepting indeterminism is not much more attractive either.
    Although some current physical theories, such as quantum mechanics, seem to support it, it is unclear whether a future grand unified theory of physics will retain this indeterminism, and there is little consensus on the extent to which the quantum
    indeterminacies permitted by current physical theories are “amplified” to a macroscopic level where they could affect human beings and other organisms. But even if quantum indeterminacies were relevant to biological systems, a further argument would be needed to dispel the worry that they introduce just randomness into the world, not free will.”

    Can anyone tell me what Christian List is talking about here?
    What future grand unified theory of physics is he talking about? It can’t be quantum gravity, can it?

  18. J-Mac,

    Can anyone tell me what Christian List is talking about here?

    He’s not talking about a specific theory. He’s talking about whatever unified theory actually prevails.

    Anyway, do you understand his point about how quantum indeterminism, if it actually exists, isn’t a suitable basis for free will?

  19. keiths: He’s not talking about a specific theory. He’s talking about whatever unified theory actually prevails.

    Really? How do you think the description of gravity in terms of quantum mechanics is going to affect how we view determinism? Give me a few examples…

    keiths: Anyway, do you understand his point about how quantum indeterminism, if it actually exists, isn’t a suitable basis for free will?

    No, I don’t! What’s worse, I don’t think he does either. But feel free to enlighten me…

  20. J-Mac,

    Really? How do you think the description of gravity in terms of quantum mechanics is going to affect how we view determinism? Give me a few examples…

    Not sure why you’re asking me. It’s List’s claim, not mine.

    keiths:

    Anyway, do you understand his point about how quantum indeterminism, if it actually exists, isn’t a suitable basis for free will?

    J-Mac:

    No, I don’t! What’s worse, I don’t think he does either. But feel free to enlighten me…

    Quantum indeterminism doesn’t rescue free will because it doesn’t reflect the nature, desires, deliberations, etc. of the agent. It’s just randomness.

    You can’t turn an unfree decision making process into a free one simply by adding indeterminism.

    It’s actually determinism of the right kind that rescues free will, not indeterminism.

  21. dazz: I have no idea what any of that means, but I suspected it couldn’t be that easy. Thanks, Bruce. I’ll read List’s paper, but it will probably go over my head

    Here is my take on the whole paper, glossing over many details.
    His argument depends two things that are not captured in the diagram:
    1. Non-Reductive Physicalism: That is, you cannot replace all the sciences by physics alone. NRP has been thrashed out many times at TSZ, so I am not proposing to revisit it. I am just saying he takes it as a premise.

    2. The Naturalistic Ontological Attitude. That says that when it comes to deciding what it real, we should simply take science at face value. If our best science says something is real, then it is, or at least that is our best bet.

    3. 1 and 2 together say we should take what each separate science says as real; each has valid things to say about the domain of reality that science studies. In particular, we should take psychology seriously and separately from physics.

    4. Our best psychology is indeterministic and there is every reason to believe it will remain so. The best it can do is predict probabilities of various outcomes.

    5. These psychological probabilities yield real possibilities (not just limits on our knowledge) because of 3.

    6. Free will is a concept that must be understood at the level of psychology. But psychology says there are real alternative possibilities. Hence real alternative possibility is part of free well.

    As best I understand it, your view (and Keith’s and Coyne’s) implicitly deny 1 or 2 or both. Which is fine. He is not arguing for those in that paper, although he cites separate paper for them IIRC.

    I was just looking for more explicit acknowledgement that that was the part of his argument that was being questioned.

  22. keiths: If the current physical state is s1, then the current psychological state is p1. The next physical state must be s2, and therefore the next psychological state must be p2.

    That’s the detail I was looking for.

    However, I don’t think you are taking NOA as seriously has he wants to take it. I think he would deny that any story about necessity and possibility at the physical level applies to the psychological level, because of NOA and NRP. I think he would say that you are assuming the reductive approach to modality that he wants to deny through NRP and NOA.

    Anyway, I will post my concerns with the argument separately.

  23. Since it involves free will, I’ll post this here.

    An odd statement from William J Murray in an OP at UD:

    Non-materialists (people that are not materialists) insist that epistemological validity requires that some sort of external world exists independent of mind that can cause universal or near universal mental states in observers . It seems no one has figured out that if one insists that non-mental, independent external commodities can cause mental states, thoughts and experiences, they have just given up free will and have become an “in principle” materialist, consigning themselves to existence as caused automatons.

  24. Bruce,

    That’s the detail I was looking for.

    However, I don’t think you are taking NOA as seriously has he wants to take it. I think he could deny the any story about necessity and possibility at the physical level applies to the psychological level, because of NOA and NRP. I think he would say that you are assuming the reductive approach to modality that he wants to deny through NRP and NOA.

    I think List’s argument fails even if one assumes NRP and the NOA.

    The key ingredient is supervenience, which List accepts. If you look at my argument, it depends on nothing more than

    a) the supervenience relation; and
    b) determinism.

    Determinism gives you the (unique) physical state sequence, and supervenience gives you the (unique) psychological state sequence. Reductive physicalism is not required, nor is the denial of the NOA.

  25. Here is my problem with List’s argument.

    Even assuming NRP and NOA, I don’t think we can use science to argue for a metaphysical stance on the nature of modality.

    It is fair to say that science needs regularity and talk of counterfactuals and talk of probabilities. But science says says nothing about the metaphysics of the modality of the laws codifying regularity or of the counterfactual possibilities used in modeling and explaining. Similarly, science says nothing about the philosophical interpretation of probability.

    So when List attempts to get metaphysical possibility from science alone, I think his argument fails.

    For example, consider determinism in science alone. Determinism is just a mathematical relationship. Given the state at one time, and the state change equations, one can calculate the state at another time. Further, at least for Newtonian microstate physics, the state change equations are symmetric in time. So if I know the state for t=0, I can calculate the state for any other time coordinate. If t=0 corresponds to “now”, it includes the agents state, and so the agent’s state (partly) determines the initial state of the universe (ie some t<0).

    When many people see the word “determines”, they immediately read in extra claims about metaphysical necessity of the past necessitating the future. But metaphysical claims about necessity and time are not part of the science. For (Newtonian) science, time is just a variable or a coordinate on an axis, and “determines” is only about mathematical calculation.

    Extra added philosophical bonus: Quine and Putnam used an argument that resembles List’s to argue for Platonist existence of mathematical entities (just sets for Quine): That is, since math was essential for science, and science was our best guide to reality, math entities must be real. Or so both claimed at one time.

  26. keiths:

    The key ingredient is supervenience, which List accepts.If you look at my argument,

    Determinism gives you the (unique) physical state sequence, and supervenience gives you the (unique) psychological state sequence.Reductive physicalism is not required, nor is the denial of the NOA.

    Some philosophers do not agree that supervenience works that way, and the different views tend to be correlated with the views on NRP, at least as far as I have bothered to understand the arguments. So my guess is that List would reject your view of what supervenience entails because of his NRP.

    (Aside: I thought you lived on the West Coast of the US. My excuse for posting this early is insomnia and living in Eastern time zone. )

    https://plato.stanford.edu/entries/supervenience/#SupeRedu

  27. Bruce,

    I thought you lived on the West Coast of the US.

    Yes, I’m in California. I’ve been staying up late these days.

    Some philosophers do not agree that supervenience works that way, and the different views tend to be correlated with the views on NRP, at least as far as I have bothered to understand the arguments. So my guess is that List would reject your view of the what supervenience entails because of his NRP.

    Well, he specifies what supervenience means to him:

    What matters is that an agential state, while supervening on (being fully determined by) the underlying physical state of the world, is more coarse-grained than that physical state.

    So by dint of his own characterization of supervenience, a fixed physical state sequence determines a corresponding fixed psychological state sequence. That’s all I need to make my case.

  28. An observation:

    In List’s toy example, you can actually infer the physical state sequence by observing the psychological state sequence, as long as you know the supervenience relation.

    Every branch in the sequence gives you more information about the underlying physical state. By t=5, you’ve pinned down the entire physical state sequence, including all future states.

  29. keiths:
    An observation:

    OK, I understand your take on his paper.

    These days, I try to avoid supervenience and stick with “all science must be constrained by fundamental physics.” Exactly what “constrains” means I leave to emerge from ongoing scientific practice; in particular what scientists consider acceptable explanations. Yes, that’s somewhat vague, but I can live with it

  30. Bruce,

    OK, I understand your take on his paper.

    Do you agree with dazz and me about List’s error, given his own definition of supervenience?

  31. Thought experiment: Assume one stance or another, and describe how the world would be different if the opposite were true.

    Then do the reverse.

  32. petrushka,

    Thought experiment: Assume one stance or another, and describe how the world would be different if the opposite were true.

    If List’s position were correct, it would imply that systems can occupy more than one physical state at a time, as dazz pointed out.

    If dazz and I are right, systems can occupy only one physical state at a time (neglecting QM, of course).

  33. BruceS: Extra added philosophical bonus: Quine and Putnam used an argument that resembles List’s to argue for Platonist existence of mathematical entities (just sets for Quine): That is, since math was essential for science, and science was our best guide to reality, math entities must be real. Or so both claimed at one time.

    Classic map-territory conflation?

  34. keiths:
    Bruce,

    Do you agree with dazz and me about List’s error, given his own definition of supervenience?

    Geez, I think you have exchanged enough posts with me by this stage to know how I would answer that question.

    ETA: Remember when we were discussing Graziano’s illusionism ? I had the wrong view of his thinking then. Alan F pulled a Woody Allen Mcluhan maneuver on me, and got Graziano to email a clarification of his thinking.

    So maybe there is a way to run your concerns past List? I tried some searches for his email, but failed. However, you have much better Google skills than I do.

  35. dazz: Classic map-territory conflation?

    Could be.

    I was influenced by some recent reading of a Sellars comment on Quine’s original view, where Sellars says, in my very rough paraphrase: When science talks about theoretical entities such as Higgs Bosons, then the whole theory explains how we make causal contact with the entity. That is, how to spend billions building LHC measuring tools that detect it and also how to do the related statistical inferences that need to be done as part of that detection.

    But scientific theories say nothing about how to make causal contact with abstract entities like mathematical sets.

    So it is wrong to apply NOA (or Quine’s version of it) to mathematical sets.

  36. Bruce,

    I tried some searches for his email, but failed. However, you have much better Google skills than I do.

    I found it. I’ll PM it to you.

  37. Bruce,

    Geez, I think you have exchanged enough posts with me by this stage to know how I would answer that question.

    I thought you might have changed your mind after I showed you List’s statement regarding supervenience. With that in hand, it’s just a matter of deductive logic to show that the psychological state sequence is fixed.

  38. Walto had mentioned Peter van Inwagen as a defender of agent causation, so I picked up a book of van Inwagen’s called Thinking About Free Will.

    It contains this remarkable admission:

    In the second part of this paper, I will defend the conclusion that the concept of agent causation is of no use to the philosopher who wants to maintain that free will and indeterminism are compatible. But I will not try to show that the concept of agent causation is incoherent or that the real existence of agent causation should be rejected for scientific reasons. I will assume — for the sake of argument — that agent causation is possible, and that it in fact exists. I will, however, present an argument for the conclusion that free will and indeterminism are incompatible even if our acts or their causal antecedents are products of agent causation . I see no way to respond to this argument. I conclude that free will remains a mystery — that is, that free will undeniably exists and that there is a strong and unanswered prima facie case for its impossibility.

    The book came out in 2017, but the paper in question was published in 2000. I don’t (yet) know if van Inwagen’s position has changed during the interim.

    So to summarize:
    1. He thinks that free will is incompatible with determinism.
    2. He thinks that free will is incompatible with indeterminism.
    3. He concludes that free will appears, on the face of it, to be impossible.
    4. He nevertheless believes in free will (and considers it undeniable) for some reason.
    5. He thinks agent causation does nothing to demonstrate the compatibility of free will and indeterminism.

    I’ll comment further when I’ve read the chapter.

  39. I’ve finished the chapter, but I never did figure out exactly why van Inwagen thinks that free will is “undeniable” despite its apparent impossibility (according to van Inwagen’s own argument). The closest he comes to revealing it seems to be here:

    The problem of free will in its broadest outlines is this. Free will seems to be incompatible both with determinism and indeterminism. Free will seems, therefore, to be impossible. But free will also seems to exist. The impossible therefore seems to exist. A solution to the problem of free will would be a way to resolve this apparent contradiction.

    There really is no contradiction here. It’s quite possible for it to seem that we have free will even if we really don’t. Looks to me like van Inwagen is trusting his intuition over his reasoning in this case.

    Props to him for being honest about the situation and his motivations, however. He writes:

    I confess I believe there is something wrong with the argument [that free will is impossible]. (I expect I believe this because I fervently hope that there is something wrong with it.) But it seems clear to me that if there is, as I hope and believe, something wrong with the argument, its flaw is not that it overlooks the possibility that my actions have their root in agent causation.

  40. I should note that van Inwagen is a Christian, as this may have something to do with his hopes for a “could have done otherwise” form of free will. There are Christian determinists (mostly Calvinists), but most seem to be libertarians. Christian theology fits better with libertarianism than with compatibilism.

  41. Bruce,

    Here’s a deductive logic puzzle on free will for you…

    Looks pretty involved. I’ll have to save it for tomorrow. The bed is calling.

  42. keiths:
    Bruce,

    Looks pretty involved.I’ll have to save it for tomorrow.The bed is calling.

    I’ve always liked the BTO song Blue Collar which extols the virtues of being up at 4:00 AM. But I’ve wondered whether that means getting up at 4 AM (like me) or going to bed around then, as it seems you do.

    I suspect the latter.

    https://www.youtube.com/watch?v=hnmhBul6svk

  43. BruceS: Could be.

    I was influenced by some recent reading of a Sellars comment on Quine’s original view, where Sellars says, in my very rough paraphrase:When science talks about theoretical entities such as Higgs Bosons, then the whole theory explains how we make causal contact with the entity.That is, how to spend billions building LHC measuring tools that detect it and also how to do the related statistical inferences that need to be done as part of that detection.

    But scientific theories say nothing about how to make causal contact with abstract entities like mathematical sets.

    So it is wrong to apply NOA (or Quine’s version of it) to mathematical sets.

    For the curious, Sellars makes this argument in Naturalism and Ontology p. 13 (Chapter 1, paragraphs 32-34), repeated at the end of “Behaviorism, Language, and Meaning” (see here).

  44. keiths:
    petrushka,

    If List’s position were correct, it would imply that systems can occupy more than one physical state at a time, as dazz pointed out.

    If dazz and I are right, systems can occupy only one physical state at a time (neglecting QM, of course).

    Schroedinger’s Will.

    But I had something a bit simpler in mind. For example, does the answer to the problem affect how we apply rewards and sanctions for behavior?

  45. If List is still at the London School of Econ., his email address can be found at their site. I’m heading east today and will try to have a look at his paper tomorrow.

    The van Inwagen book I read on free will is his older Essay on Free Will. I remember finding a mistake in that. Dunno if his views have changed since then.

  46. keiths: Quantum indeterminism doesn’t rescue free will because it doesn’t reflect the nature, desires, deliberations, etc. of the agent. It’s just randomness.

    How do you know that? Got proof? I do …

  47. “while physics may be deterministic”

    Promoting this lie is dishonest when you know by now that physics is in fact not deterministic.

    Remember this?

    “Schrödinger equation doesn’t tell you anything about determinism one way of the other just as Newton’s equations do not tell you either. Experimentally, we KNOW that a system described by the Schrödinger equation is nondeterministic. Here’s a thought experiment that disproves determinism:
    We have a double slit experiment with single photon emissions and the target area separated in 10 different sections labeled 0 to 9. Once a section is hit, it stays on (cannot detect multiple hits) Determine the output sequence? Is it 012…9? Is it 8754219036? What is it? Even if you have “many worlds”, the output is still not determined in either of those worlds.”

    It’s not going away…

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.