Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Could Have Done Otherwise - What does it mean exactly?

  1. Jul 20, 2006 #1
    The phrase “Could Have Done Otherwise” (CHDO) often comes up in philosophical discussions on the existence or non-existence of “free will”. Many libertarians would claim “free will” means that, no matter what one does, if the act was a free will act then one could have done otherwise (CHDO) than what one actually did. This seems to be the libertarian definition of free will.

    Firstly - Would any libertarians out there disagree with this? If yes, I would welcome your comments as to what you believe is the “correct” libertarian definition of free will.

    Now to my question. What, exactly, are we to understand from the expression CHDO?

    Does it mean CHDO in the counterfactual sense? In other words, does it mean :

    A : CHDO = “If circumstances had been different, then it follows that I could have wanted to do something different, and it also follows that I could have done something different”

    Or does it mean CHDO in some non-counterfactual (ie factual) sense, in the sense of either B or C below :

    B : CHDO = “If circumstances had been exactly the same, then I could still have wanted to do something different, and it follows that I could have done something different”

    or even :

    C : CHDO = “If circumstances had been exactly the same and I had not wanted to do something different, then I could still have done something different”

    Or does it mean something else entirely?

    My understanding of libertarian free will is that the libertarian would reject both (A) and (C) above as explanations of the meaning of the phrase CHDO. Why? Because (A) is completely compatible with a deterministic account (in other words, we do not need to posit libertarian free will in order to claim counterfactual ability to have done otherwise under determinism); and because (C) is irrational – what rational reason could I have for suggesting that I could have done something different if I had not wanted to do something different?

    This seems to leave us with (B) above as the meaning of CHDO according to a libertarian.

    I welcome comments or criticism at this stage (before developing the idea further)

    Best Regards
    Last edited: Jul 20, 2006
  2. jcsd
  3. Jul 20, 2006 #2
    Well it seems that you tie in the concept of free will which is really one of physics with politics.
  4. Jul 20, 2006 #3

    There are two meanings of libertarian - one is philosophical and one is political. This is a philosophy thread...... I wasn't referring to political libertarians (I guess I should have made that clear).

    See here for one discussion of libertarian free will in the philosophical as opposed to political sense :


    Best Regards
  5. Jul 20, 2006 #4
    Some people claim to have difficulty understanding Alternative Possibilities. Here is a formal definition.
    Let S1 be the state of the universe at time T1, and L the laws of the universe. According to indeterminism, there is (in general) a set of more than one future states of the universe {S21, S22, S23...} , which are all possible in that they do not contradict the conjunction of S1 and L.

    <poss>(L AND S1 AND S21)

    <poss>(L AND S1 AND S22)

    <poss>(L AND S1 AND S23)

    According to determinism , the set {S2..} has only one member ... there is only one state which is compatible with the conjunction of S1 and L (being the only possibility, it is also actual and ncessary).

    [nec](L AND S1 AND S21)

    NOT(L AND S1 AND S22)

    In standard single-world metaphysics, only one possibillity can actually occur. The other members of the set are "alternative possibilities". Thus, if S22 is the possibility that occurs, S21 and S23 are the AP's in that case.
  6. Jul 20, 2006 #5
    A bit simplistic, but basically right. We often have competing desires
    and aims, the weightings or evaluations we attach to them
    are themselves subject to evolution over time.

    Too simplisitc. It assumes you only have one "want" in mind at a time.
  7. Jul 20, 2006 #6
    Hi Tournesol

    Many thanks for this post – very interesting.

    One small point : Do you consider the "laws of the universe" as you describe above to be prescriptive or descriptive laws?

    All three interpretations (A, B and C) of CHDO that I have suggested would seem to be compatible with the above analysis of alternative possibilities – would you agree?

    A is compatible with both determinism and indeterminism, whereas B and C are compatible only with some form of indeterminism – would you agree?

    Do you have any views on which of A, B or C (if any) a (philosophical) libertarian would accept as a correct interpretation of CHDO?

    Best Regards
  8. Jul 20, 2006 #7
    I would be happy for you to suggest a better description if you consider this too simplistic.

    If you are suggesting that the “weightings or evaluations” change between the two circumstances in question, then this would not conform to the condition that circumstances are exactly the same in both cases.

    “Circumstances exactly the same” means, in effect, that if in the original case we have state S1 and laws L (to use your terminology from post #4), then in the second case we must once again have state S1 and laws L – there is no possibility for “weightings or evaluations” to change between the two cases (since the weightings and evaluations are part of either S1 or L).

    Interpretation B of CHDO entails that S1 and L are identical in both cases (the original case and the “could have done otherwise” case). As far as I can see, the only way that this could give rise to a different outcome (following your analysis of alternative possibilities) is if the outcome is a result of some form of indeterminism in the process which leads from antecedent states (the circumstances defined by S1 and L) to "what we want".

    Best Regards
    Last edited: Jul 20, 2006
  9. Jul 20, 2006 #8
    A self-forming action isn't "just random" -- instead it is connected
    to a persons character. However the connection works
    forwards, not backwards; the action forms future character.

    The argument that libertarianism must be false because
    it cannot "explain" people's actions is only valid if there
    is a realistic alternative. But there isn't -- and there
    isn't for reasons of human psyhcology, not the way
    the universe works. People just can't give totally
    bulletproof chains of explanations for their
    actions that go all the way to the BB. Sooner
    or later the chain of explanations peters out in somethign
    that is just assumed. something that might as well be indeterminsitic.
  10. Jul 20, 2006 #9
    I mean the "real" laws, or an ideal description thereof.

    No. I am assuming that some set of circumstances has
    occurred and is fixed. The A interpretation amounts
    to the claim that the history of the universe could have been
    different. That is different from the claim that more than one thing could
    have happened under a different set of circumstance.

    of course, A can't be translated into a power psosessed by
    an agent.

    B is the closest
    Last edited: Jul 20, 2006
  11. Jul 20, 2006 #10
    It could conform to the external circumstances being the same.

    Then talk about S1' which is the state of everything outside my CNS.

    Naturalistic FW is based on indeterminism, but to avoid the Burridan's
    Ass problem it needs to operate over some span of time.
  12. Jul 20, 2006 #11


    User Avatar
    Science Advisor

    Suppose we have 5 events, A, B, C, D and E where each event can & will cause one of the other events once it is triggered (an event triggers another in a deterministic, non-random manner, albeit possibly through a complex process). It's a cycle. Once one of A, B, C, D and E is triggered this cycle follows a path of events, maybe A, B, C, A, D, A, E, B ... etc.
    Clearly, this cycle is autonomous. Like an agent with free will the cycle alone determines its own behavior. On the other hand the path the cycle takes is completely deterministic. Is this enough to say the cycle has free will?
    What if each of A, B, C, D and E were to randomly trigger eachother? Then the cycle is both autonomous and non-deterministic. Does it have free will then?
  13. Jul 21, 2006 #12
  14. Jul 22, 2006 #13


    User Avatar
    Science Advisor
    Homework Helper
    Gold Member

    Hi MF. I have no comment on your OP directly. I'm curious though what you feel are the implicit assumptions you're making. Specifically, it seems to me you're only concerning yourself with weakly emergent phenomena and you've ruled out strongly emergent ones, if in fact anything like strong emergence can exist.

    In short, the idea might be that strongly emergent phenomena which evolves within some mechanism or system, might causally affect a subset of that system, however that causal affect might not be deterministic. It seems to me this might more closely match the perception we have of "free will".

    For a definition of "weak emergence" I'll quote the guy everyone else seems to be quoting, Bedau.
    And for "strong emergence" I'll quote http://consc.net/papers/emergence.pdf" [Broken].
    *where Chalmers states "not deducible even in principal" I take to mean "not onticaly determinable"

    If we assume only weakly emergent phenomena exist, then we can model any of those phenomena as if there are local causal actions acting on each 'microstate'. Those causal actions might be deterministic or random. It seems to me, this is one of the implicit assumptions you're making in your OP.

    On the other hand, if we assume there is such a thing as strong emergence and strong downward causation, I believe what's being suggested is that some emergent phenomena can have a causal affect on some set of microstates which is neither deterministic nor uncaused.

    Let's look at computationalism for a second: IMHO, computationalism can only result in weakly emergent phenomena. The actions of any computational device can be determined from the "microdynamics" of the system as Bedau would put it. For example, a switch in a conventional digital computer acts in a completely deterministic way and is only affected by the local affects of electrical potentials that act on it. There is no downward causation which could possibly change the state of a switch, it is only affected by the local microdynamics.

    If one suggests consciousness arises from the act of computation, and thus it is "strongly emergent" then I'd have to say that such a claim is a trivial one in the sense that it is not needed, even if it is real, to explain the action of the computer or system. Why invoke "strong emergence" when the actions of a computer can easily be explained by the actions of the switches or other parts of the device? Certainly those subjective experiences have no causal impact on any portion of any computational device. They can't. No part of the computer is susceptible to influences beyond the local ones acting on each portion of the device. The only benefit (if you can call it a benefit) gained by suggesting consciousness arises from the act of computation is an explanation for subjective feelings, but the system has no additional capability or features which arise from this allegedly strongly emergent phenomenon of subjective experience. Thus the need for 'downward causation' for any strongly emergent view of consciousness.

    If downward causation is conceived of as being a feature of consciousness then, we might also suggest that there are influences created by the phenomena which are not deterministic, but they do result from a cause. Thus, for "CHDO", we might suggest that there is another definition which isn't covered by the OP because of the implicit assumptions made. We might suggest that CHDO arises from a strongly emergent phenomenon which causes a change in microstates through downward causation, but that this change in microstates, although caused, is not ontically determinable. I think this kind of "free will" is much closer to what we actually experience, and also provides for such things as "moral responsibility".

    I think another alternative is to suggest downward causation will create only a single, very specific and ontically deterministic change in the system or mechanism. In other words, if there is such a thing as downward causation, one might suggest that given some state of a mechanism or system, that state results in a deterministic change of state to some portion of that system or mechanism. Such a definition of downward causation seems a bit awkward though as it almost seems like it isn't downward causation at that point since this says that a set of microstates will result in some deterministic microdynamic. I haven't thought about alternative enough though. At any rate, I wonder if any of this gives you some ideas about what the implicit assumptions are in your OP.
    Last edited by a moderator: May 2, 2017
  15. Jul 22, 2006 #14
    (Note : Self Forming Action / Self Forming Will are terms borrowed from Kane’s treatment in The Oxford Handbook of Free Will)

    This is inconsistent, or incomplete. You claim SFAs (self forming actions) are not random, but at the same time that SFAs form character and not the reverse. In other words (according to your account), character supervenes on SFAs, SFAs do not supervene on character. What, then, forms SFAs (if SFAs are not random)? In other words, SFAs are either regular (have regular relationships with some kind of antecedent states) or irregular (no relationships with any antecedent states). If irregular then SFAs would be random.

    Nobody has claimed (so far) in this thread that libertarianism is false. I am simply trying to understand what exactly a libertarian means when he says he could have done otherwise.

    If I were to argue against the concept of libertarian free will, then I would argue not that it is false, but that (if fully explicated, which not many people apart from your good self seems prepared to do) it either leads simply to arbitrary behaviour, or else the very notion is either inconsistent or incoherent.

    You claim that there is no realistic alternative to libertarianism. I disagree. That “people just can't give totally bulletproof chains of explanations for their actions that go all the way to the BB” is a reflection of the limits of our epistemology (what we can know about the world), it is not necessarily a reflection of lack of determinism. Chaos (the mathematical kind) and the HUP already place limits on what we can know about the world (in other words, even if the world were totally deterministic we still would not be able to know everything because of the limits to our knowledge imposed by chaos and HUP). I agree with your comment that “sooner or later the chain of explanations peters out in somethign that is just assumed. something that might as well be indeterminsitic”, and indeed the Big Bang itself might have been a quantum-indeterministic event, and the world may operate according to quantum-stochastic (but not totally random) principles rather than deterministic principles. If the libertarian simply wishes to equate his concept of “free will” with stochastic behaviour as opposed to deterministic behaviour then I have no problem with this – but there is no more autonomy or “will” involved in stochasticity than there is involved in determinism, and stochasticity entails much less control over one’s actions than does determinism.


    Thank you.

    Okay, I understand. You are taking the weightings and evaluations to be internal to the “will”, and not included in the external circumstances. The problem with this is that the weightings and evaluations are part of the antecedent states which “determine” the will (or the outcome of the will), hence must be considered part of the circumstances.

    We need to compare apples with apples.

    What determinism says is that given two perfectly identical worlds (identical not only in the physical, but also in all other senses, the “soul” (if such exists) and “will” sense) the subsequent evolution of these worlds will also be identical. In other words, under determinism the “will” has no power to choose that one world will be any different to the other – all is determined.

    What the libertarian concept of CHDO seems to be saying is that, given these same two perfectly identical worlds, the subsequent evolution of these worlds will not necessarily be identical, that the “will” has some kind of power to “choose” that one world will be different to the other.

    In other words, CHDO assumes that all circumstances (not only those external to the will) are identical in both cases. This includes any weightings or evaluations that the will may “use” in deciding what to do. In other words, anything which has any causal role in determining the outcome of the will must be included in the fixed antecedent “circumstances”.

    How does this avoid the problem? Surely Burridan’s Ass had all the time in the world to decide which way to go?

    By “autonomous” here I assume you mean “acting with free will”.

    As you said, the path is deterministic. In what sense therefore could you claim that this series was autonomous? It’s about as autonomous as a computer program.

    Why should random behaviour result in autonomy? If my computer program is deterministic and has no autonomy, how does it suddenly become autonomous simply by virtue of introducing some random element into its calculations? Random elements will simply result (if they result in anything at all) in random or stochastic behaviour, there is not necessarily any autonomy involved.

    The problem with the AP idea is that, to have any value at all, an AP must be under the control of the will – the will must be able to select which AP will occur. But if APs occur according to some kind of random process then the will has no more control (in fact it has even less control) over the eventual outcome than if the process had been entirely deterministic.

    Best Regards
    Last edited: Jul 22, 2006
  16. Jul 23, 2006 #15
    Hi Q_Goest

    I’m not assuming anything specific about emergent phenomena, except that I am assuming that however the world operates, determinism is either true or false in that world.

    If not deterministic then it seems to me the evolution of the system must be either random (showing no regularities whatsoever) or stochastic (showing probabilistic regularities).

    Why can we not apply the same 3 alternative notions of CHDO to such downward causation? In such a case, the “circumstances” includes the higher-level configurations that you refer to. In other words, given a particular state of the world, including the higher-level configurations of any downward-causation phenomena in that world, are there any alternative possibilities available, or is the future determined?

    Think of it in terms of comparing two parallel and identical worlds, where all the circumstances including higher-level configurations of any downward-causation phenomena, are precisely replicated in both worlds. Determinism would say that both worlds will necessarily evolve with time in an identical fashion, whereas indeterminism would say that the worlds would not necessarily evolve with time in an identical fashion. This would be true whether downward-causation phenomena were present or absent.

    We need to be careful not to confuse determinable (what you refer to as deducible) with deterministic. A system can be deterministic without being determinable. Whether “causal impacts” are deducible (= determinable) in principle or not is not necessarily relevant to the question of whether the universe is deterministic or not. A 100% deterministic world may be not deducible (ie not predictable or not determinable) by virtue of (for example) Heisenberg’s Uncertainty Principle, and/or mathematical chaos, and/or self-referentiality.

    In summary, I believe that in any given system, whether it includes downward-causation phenomena or not, determinism will be either true or false. If determinism is false, then we say that the system is indeterministic (where here we take indeterministic to include stochastic), and this gives rise to “alternative possibilities” or “could have done otherwise”. If we have two parallel and identical worlds, then if determinism is true these worlds will evolve identically in all respects (no alternative possibilities, and no could have done otherwise), whereas if determinism is false then these worlds will not necessarily evolve identically in all respects (alternative possibilities exist and so too does could have done otherwise).

    Best Regards
    Last edited: Jul 23, 2006
  17. Jul 23, 2006 #16


    User Avatar
    Science Advisor
    Homework Helper
    Gold Member

    Hi Movingfinger. I think you're still clutching to a very specific model of the world that may or may not be correct. People often think in terms of any given phenomena being deterministic, stochastic or random, but this model may only be applicable to weak emergence or "upward causation".

    Campbell seems to recognize the same problems you're having with CHDO. I'd suggest reading http://www.lehigh.edu/~mhb0/physicalemergence.pdf" [Broken]for more on this. They also quote J. Kim "Mind in a Physical World" MIT Press, 1998 which it seems from reading Campbell, is applicable to what you're searching for. It seems their arguments follow the same path I've already provided above.

    What I believe is being suggested is that downward causation, created by a mental property (M-property in the literature), can cause a change in a physical property (P-property) which can not be explained, nor predicted, even in principal. They would appear random, and in affect they would have the capability of producing two different, parallel worlds. But would you call such an event random or even stochastic if it is caused by some mental property? You could simply suggest it is random, but I'd argue that such a conclusion doesn't suffice because any change in state which is caused can't be random. On the other hand, it can't be deterministic either since there can be more than one possible outcome.

    I realize this too is only a possible model of the world, just as there can be others and I don't know how it fits in with the "libertarian definition of free will". If you're only looking to understand the libertarian perspective, I think one can only recognize that such perspectives may contain implicit assumptions which may or may not be valid. However, strong emergence and downward causation has some very interesting aspects to it that may leave you head scratching. As I've mentioned previously, the concept has obviously already been considered. Suggesting strongly emergent phenomena arise from computationalism is "superfluous" and "causally impotent" which I believe is what Alexander is suggesting when he says:

    That seems like one of the most damning condemnations of computationalism I've ever heard.
    Last edited by a moderator: May 2, 2017
  18. Jul 23, 2006 #17


    User Avatar
    Science Advisor

    Not at all, actually. By autonomous i meant that, from a strictly physical viewpoint, the cycle alone is responsible for it's behavior, which is the case. There's no notion of a choice in here, especially since the path the cycle will take is deterministic. That the cycle doesn't have a "choice", and yet it is autonomous is what i was proposing.

    It would be as autonomous, possibly, as a human.

    I was calling it autonomous before introducing randomness into the calculations, you misunderstood that.
  19. Jul 23, 2006 #18
    Hi Q-Goest

    Thank you for the reference - I shall read it with great interest.

    It seems that you are perhaps confusing determinability (whether something can be explained/predicted, even in principle) with determinism (whether, given antecedent states of the world, there is one and only one possible outcome). A world may be 100% deterministic (ie determinism is true in that world) without being determinable (even in principle). Limits to determinability in our own world include chaos, Heisenberg's uncertainty principle, and the in-principle indeterminability of self-referential systems - but none of these in-principle limits to determinability implies that determinism is necessarily false in our world.

    "Appearing random" is an epistemic property, it does not follow from a random appearance that the world is (ontically) random, neither does it follow that more than one outcome is possible.

    It depends on whether the cause (the mental property) is itself caused or uncaused. Is the so-called mental event supposed itself to be caused, or uncaused?

    If the mental event is itself uncaused (ie spontaneous) then presumably it's behaviour is either random or stochastic - if not, what else could it be?

    If the mental event is itself caused, then move the question back to the cause of the mental event.

    Of course it can. If X is a random event, and X causes Y, then Y is also random (since it depends on X). Thus Y is caused but is nevertheless random (because the causal chain originates in a random event)

    How do you know there can be more than one possible outcome? For this to be true, you must assume determinism is false in the first place.

    I have already looked into downward cauation some time back, and found nothing significant which (imho) would have any real bearing on free will. But I will take a look at the Campbell & Bickhard paper and get back to you.

    Best Regards
    Last edited by a moderator: May 2, 2017
  20. Jul 23, 2006 #19
    OK, I see what you mean. However I think that most libertarians would claim that any system which has no "choice" in the way that it acts is not autonomous. It comes down to the definition of autonomous, which a libertarian would claim entails ability to choose, whereas you do not. (Here I am also assuming that choice is defined in the libertarian sense of free will choice).

    I tend to agree - because I do not believe either a human or the system described is autonomous (in the libertarian sense of the word)

    Best Regards
  21. Jul 24, 2006 #20
    They are not random in the way that making decisions according to
    a die is random. Such decisions have no causes and no consequences either.
    However, is still true that SFA's have no *prior* causes.
    There is more than one sense of "random" here.

    It's not that black-and-white.

    Not in the way a die-throw is random. They have a relationship
    to character, but a forwad-firing one.

    The "arbitrary behaviour" complaint is just the same point again. Libertarian FW is only
    arntirary in the senses of not being able to give totally bulletproof chains of explanations for actions that go all
    the way to the BB.

    No, but so what ? You seem to have confuse causes with reasons. The point of insisting that
    actions must have causes is to ensure that they have reasons; rationality is the payoff,
    causality is a means to the end. Crazy, irrational actions have casues as much as any other. Causality
    is not part of a version of FW worth wanting, raionallity is.

    The control lies in the SIS's selection of proposal from the RIG.

    No. By "circumstances" I mean both what is temporally prior, and what is "outside the head".

    No, libertaraian CHDO deals only with identical external circumstances.
    The private thoughts in your head can vary.

    Its decision will be disconneted from its prior state of mind.

    If "autonomy" means that the output of a systems cannot be predicted from
    a complete specification of its input , then random behaviour *does* mean

    If autonomy involves rationallity (to a realistic extent and under appropriate circumstances)
    then a SIS will be needed as well as a RIG!

    The SIS can filter the output of the RIG.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook