Theory on framework issues

Saturday, July 7, 2012

10.1.2. Reflexive prediction, determinism, and the impossibility of free will



Will you—or won’t you—take an M & M candy in the next minute after it’s placed in front of you? In this thought experiment, we assume that I’m able to predict your action with absolute accuracy based on physics. I inform you that I predict you’ll take the candy, but you’re a rebellious sort, and just to prove you’re no automaton, you decide on defiance. Your proneness to disconfirm predictions deliberately seems to render impossible any prediction of your choice, provided you’re informed of it in advance. But remember I can predict your choice. 

    At least one of the following assumptions concerning the physical possibilities must be false to remove the contradiction:

    1. That it’s physically possible that you be informed of a deterministic prediction when you’re indifferent to the alternatives
  
    2. That it’s physically possible to predict your behavior precisely
  
    3. That it’s physically possible for you to defy the prediction

A.   Because you don't want to defy it

B.   Because you won't be able to defy it despite wanting to

The possibility of conveying a precise prediction (No. 1)
It might be claimed that physics precludes that you could ever be informed of a determinate prediction when you’re indifferent to the alternatives, since you’ll be moved to falsify it. But if perfect prediction is possible, conveying its substance should be easy: no barriers are apparent.

The possibility of predicting precisely (No. 2)
We may try denying the second assumption, that precise prediction of your behavior is possible. It might be said that human organisms aren’t the kind of systems that can be predicted precisely. But again, one must wonder why not. Organisms are but clumps of matter; they can be described as systems in biological terms, but otherwise they are no less subject to purely physical description than other purely physical systems. No specific barrier limits prediction of biological systems when they’re conceived as physical systems.

The possibility of defying the prediction: You won’t want to defy it (No. 3A)
The possibility of defying the prediction must be questioned. Perhaps you won’t choose to be negative, your character and personality notwithstanding. But the truth of the prediction doesn’t affect any of the things that we use to attribute a reactant tendency to infer what you want to do; its truth affords no cause for bringing other motives into play. A true prediction confronts you no differently from a false prediction: whether the prediction happens to be true or false has in itself no bearing on how it’s received.

The possibility of defying the prediction: You can’t defy the prediction despite wanting to (No. 3B)
Having excluded the other possibilities, we should conclude that you will confirm the prediction despite not wanting to confirm it. This result is due to your taking of the candy being a physical act, which must conform to physical prediction, whereas having an attitude of wanting is identified by psychological criteria and does not necessarily conform to any specific physical criteria.

A note on quantum indeterminacy
Although the ground is well traveled by others, I should say a couple of words about quantum mechanics in its bearing on whether absolutely true predictions really are physically possible. First, the relevant macroscopic states are essentially deterministic.  Second, quantum uncertainty doesn’t explain why in the thought experiment you would want determinately according to your character, for which expectation I argue.

Conclusion
The thought experiment serves as an “intuition pump” (Daniel Dennett) concerning the incongruence between mentalistic and physical concepts. Compatibilism is the doctrine that the world can be deterministic despite your ability to exercise “volition.” Compatibilism thus implies that in describing our “wants” we describe a real cause of our behavior.

Mental concepts, such as "wanting," refer only fictitiously. They serve fallibly to enhance foresight about our behavior and that of other people. Fatally for compatibilism, the thought experiment accentuates that fallibility.

29 comments:

  1. I'd stand by no.2, or a variant there of:

    "It would be impossible to take every measurement necessary to predict your behaviour without interfering with the behaviour".

    I'd say this was not a phenomenom exclusive to organic life. If I were to measure the temperature of my afternoon cup of tea, with a mercury thermometer, some heat would be transfered from my tea to the thermometer, and thus the recorded temperature would be lower than the pre-recorded temperature, though in this case, not by a significant amount.

    I take your point about macroscopic states being essentialky deterministic, however this plasters (not papers) over the difference between a perfect predicter, and a predicter with a 1 in a 1000 error rate.

    I am inclined to agree that "wanting" preceeding "acting" may well be a fiction.

    I am not sure why character would not, in principle, be reducible to positions, momenta (?) etc of atoms, electrons and such. The same, in principle, for an act of wanting.

    ReplyDelete
  2. "If I were to measure the temperature of my afternoon cup of tea, with a mercury thermometer, some heat would be transfered from my tea to the thermometer, and thus the recorded temperature would be lower than the pre-recorded temperature, though in this case, not by a significant amount."

    But bracketing quantum effects (which I think are irrelevant to the thought experiment) the change induced by measurement is itself predictable. You can adjust for it--in principle, down to the quantum level. And measurements in principle can be reduced in their degree of interference to become arbitrarily small (or, in reality, quantum small).


    What I find strange about this objection (outside a quantum context) is that measurement interactions are singled out. Why? They're just ordinary physical interactions, as predictable as the interactions going on when no measurement is made by a human. Perhaps you can clarify why singling out human-measurement interactions seems compelling.

    "I take your point about macroscopic states being essentially deterministic, however this plasters (not papers) over the difference between a perfect predicter, and a predicter with a 1 in a 1000 error rate."

    Yes, I sort of felt guilty writing that, but I don't think it matters to the thought experiment whether prediction is perfect or "practically" perfect, since we are talking about effects the thought experiment would allow reproduction of.

    "I am inclined to agree that "wanting" preceding "acting" may well be a fiction."

    Good. I had actually expected that would disturb you.

    "I am not sure why character would not, in principle, be reducible to positions, momenta (?) etc of atoms, electrons and such. The same, in principle, for an act of wanting."

    Because character and wanting are concepts that describe a functional relationship rather than a physical relationship. Using as example a nonpsychological functional relationship might clarify the reasoning. A heart (I think--consider it a stipulated definition if I'm wrong) is identified as the organ responsible for circulating the blood. The heart need not have any specific physical description to count as being a heart. A particular heart has a specific physical description, of course, but given a static physical description, you can't necessarily say the organ is a heart. The test of heartness is in the circulation of blood; the test for character and wanting is also something functional, although it's harder to say what that is.

    ReplyDelete
  3. "Why are measurement interactions singled out?"
    An interesting question, I'd guess because when dealing with atoms and such the effects of the observer on the system are so much more obvious than the effect of using a thermometer to measure the temperature of a cup of tea.
    I'll try to clarify my line of thought and see where that takes things. I tend to start thinking in practical terms, as you've probably noted I'm having trouble keeping the perfect predictor as a thought experiment (also, I've noted, spelling it correctly).

    If consciousness could be reduced to spatial relations of neurotransmitters, ions etc, then to predict a decision perfectly you would need perfect knowledge of the position (and movement) of every single ion, every single atom that made up ion gates in the cell membranes of the neurones, every single neurotransmitter etc, and probably a complete knowledge of everything within the same light cone. To do so you would probably do something along the lines of pinging a photon at every single one (atom, ion etc ), in doing so some energy would have to be imparted to every single atom, so whilst the individual effect would be small, it would occur on a vast scale and the effect might be enough to, for example, send an ion to a side of a membrane, and...chaos theory magic....have a much larger effect.

    "Good. I had actually expected that would disturb you"

    It does! All the arguments I can think of against it are either not supported by the evidence, or involve redefining "want" unfairly. I still think compatibilism can be reformulated; you can do as you will, but your will is created after the act, but you cannot control what you will.

    In terms of defining character as a function, as with the heart example I think we would need to define further terms, like blood, pump etc, otherwise, and I may be wrong, the definition would lack enough precision (to be useful).

    ReplyDelete
  4. "I still think compatibilism can be reformulated; you can do as you will, but your will is created after the act, but you cannot control what you will."

    I don't see what makes that a reformulation rather than an arbitrary redefinition.

    "In terms of defining character as a function, as with the heart example I think we would need to define further terms, like blood, pump etc, otherwise, and I may be wrong, the definition would lack enough precision (to be useful)."

    I'm concerned with the concept of willing and wanting as functional; they're the concepts relevant to compatibilism. What does character have to do with it? I act consistent with my character seems no more a matter of free will than a piece of metal acts consistent with its magnetism.

    To say whether "character" is functional, I'd need a definition, as different theories employ the concept differently. I think the concepts of "want" and "believe" are philosophically important because I think they're innate concepts, but the concept "character" is invented.

    ReplyDelete
  5. Joseph,

    Why you referred to character--I now realize--you probably meant the "character" in my "Perhaps you won’t choose to be negative, your character and personality notwithstanding." In this use of character, it is a functional concept. The character trait of being inclined to disconfirm predictions about oneself would serve the function of asserting one's autonomy or something on that plane. So it isn't that character is itself necessarily a functional concept but character traits typically are.

    ReplyDelete
  6. Why I did not think of it as an arbitary redefinition was basically the principle that the hypothesis must preserve the results. Thus if volition has been found experimentally to follow an unconscious state (?), then all theories of will must be reconstrued to reflect these results, or perish. The experiment itself, as I've understood it, falsifies the hypothesis that conscious volition preceeds action. If volition is by definition conscious, then the hypothesis that volition preceeds action is falsified.

    To break english grammar and phrase things in a japanese style:

    the to be willed thing is done.

    ReplyDelete
  7. Yes, that is why I referred to character, I still think a character could be defined as both a pattern of events (matter-energy) in space-time, as well as a functional description, something like Russell's Monism. The idea of the thing and the material thing are two sides of a coin.

    ReplyDelete
  8. Good argument that your redefinition isn't arbitrary; but it only takes account of the Libet results. If you accept that wanting is defined functionally, then it won't invariably correspond to a specific physical description. Libet is more colorful, but ordinary hypnosis makes the point. When a suggestion is given to a good subject that she will move her hand so against her will, she ends up with a movement that doesn't correspond to what she ended up wanting.

    Russell thought that some neutral substance could be described in two ways, but Russell's point was that the same thing was being described. Mind and matter, that is, were each conceived as substances. But a functional relation isn't a substance and can't be conceived as one. The physical expression of a functional relationship would be an infinite disjunction.

    ReplyDelete
  9. I think at worst my definition is ad hoc, I'd have to plead guilty to that.
    I think I'll have to bow to your knowledge on matters psychological, as my knowledge is limited to things like "stress is a candidate for the main cause of iFLUTD". May I ask a few questions before going on?

    My understanding of hypnosis is that some are more susceptible than others, and that an element of crowd pleasing, authority pleasing is involved?

    ReplyDelete
  10. There are indeed wide differences in susceptibility to hypnosis, but differential desire to please crowds and authority doesn't have much to do with it. I think differential susceptibility is the result of different tendencies to experience dissociation.

    A Chinese psychologist provides a good (if poorly written) paragraph:

    "(c) An Explanation of Hypnosis
    When a person is hypnotized, it might be that his/her imagination is dissociated and sends the imagined content back to the sensory cortex, resulting in dreams or hallucinations; or some of his/her senses are dissociated, resulting in hypnotic anesthesia; or his/her motor function is dissociated, resulting in immobility; or his/her reason is dissociated and he/she obeys the hypnotist’s orders; or his/her thought is dissociated and not controlled by reason, hence strives to straighten out his/her body between two chairs. A command can also be planted into the hypnotized mind and acted out accordingly long after the session of hypnosis, as follows: A person obeys the orders of reason in normal state, but when hypnotized, his/her reason is replaced by the hypnotist's command to make decisions or believes, and he/she will be very uneasy if he/she does not do things as decided or his/her belief is contradicted. Hypnotherapy is also based on this principle." (http://myweb.ncku.edu.tw/~ydtsai/mindbody/)

    ReplyDelete
  11. I'm having some trouble checking hypnotism, google results are a mire of self-help, religious institutes attacking hypnotism and pop psychology.
    If all your senses are "tricked" by dissociation could you be said to be acting according to your (fraudulent) perceptions. Similar to any trick. I.e. If I were told I had choose to kill a puppy, or not, but if I decided not to a million babies would be killed, and was shown false evidence that this was both possible and highly probable,and I choose to kill the puppy, you would say I did so of my own free will. Is this a fair analogy? The hypnotist's subject has free will, but acts on false evidence?

    ReplyDelete
  12. That's only one form of dissociation. The hypnotist can suggest that the subject arm will move despite the subject's not willing it. He needn't provide a reason this will happen. The quote in my previous comment is a good summary of the various forms of dissociation that the hypnotist can induce.

    ReplyDelete
  13. I thought about responding case by case, which I can if you like, but I thought it might be better to reduce it to one question:

    Libertarian Free will and compatibilism (as concepts) seem to be alive and healthy despite hypnotism having being known for quite a while. How have they escaped execution, so far (solely with regard to hypnotism)?

    ReplyDelete
  14. I don't think libertarian free will has any serious intellectual credibility. That's why I don't bother with libertarianism and direct my arguments against compatibilitism, which I think is today dominant.

    What's novel in my argument in 10.1.1 and 10.1.2 is that rather than make a subtle argument about what origination really is (I do that in 10.1 to 10.3) I instead argue that mentalistic terms are fictitious. Not that they're empirically misplaced (the usual interpretation of Libet) but that mentalistic terms, being functional in nature, don't correspond to physical states. Thus, explaining conduct by volition is not to offer any kind of truly causal explanation. I'm saying more than just that mentalistic terms are mere approximations. I maintain that a conceptual fissure between volition and physical causation is the only way out of the M & M paradox.

    Hypnosis is an example of where mentalistic explanation fails. But an example of the failure of a mentalistic doesn't by itself prove compatibilism is false because compatibilism is the claim that volition is consistent with determinism, not that behavioral determinism requires volition. Volition may fail under compatibilism; what's precluded that volition is conceptually incompatible with determinism. Once I introduce the argument, however, I think the phenomenon of hypnosis can help a reader grasp how determination diverging from volition is possible.

    Also, most psychologists don't know much about hypnosis--they seem afraid of it. So, I think probably viewer philosophers of mind do.

    ReplyDelete
  15. I agree that explaining conduct by volition is not to give a causal account.
    I still struggle with the idea that volition only have a functional explanation, I think the situation is akin to the example you gave, a heart. A heart can be described functionally, in many different ways, but it can be defined physically too. A functional description does not exclude a physical one. More specifically I am thinking of the experiments where after intensive mapping an image could be formed from measuring levels of activity in the brain of a variety of test subjects. Though, as I understand, no two subjects pattern of activity was indentical, a physical account of the brain's visualisation could be given for each individual.

    For some of the examples of why hypnosis may rule out compatibilism could I not say:

    Well I can shackle a persons hands and then he/she cannot move them in spite of his/her volition.
    Alternatively I could inject a muscle relaxant, or cut the appropriate nerve etc etc, then the subject could not move in spite of his/her volition.

    ReplyDelete
  16. "More specifically I am thinking of the experiments where after intensive mapping an image could be formed from measuring levels of activity in the brain of a variety of test subjects. Though, as I understand, no two subjects pattern of activity was indentical, a physical account of the brain's visualisation could be given for each individual."

    The concepts involved here aren't functional--or at least not obviously so. You expose a person visually to an image and record the brain pattern. You judge whether the person has been exposed to the stimulus physically, and your record the result. But consider instead whether the person believes that a given vase has a particular design. We judge belief by, among other indicia, whether the person is apt under appropriate circumstances to utter elements of an indefinitely large set of propositions affirming that the vase has the particular pattern. The physical states that count has having a belief are heterogeneous because they're grouped together by their function of predicting overt behavior from expressive behavior.

    My argument isn't that it's logically impossible that functional concepts map onto physical concepts. Rather, that it's logically possible that they don't, and they don't in the case of psychological concepts like belief and desire--those in particular.

    "For some of the examples of why hypnosis may rule out compatibilism could I not say:

    "Well I can shackle a persons hands and then he/she cannot move them in spite of his/her volition.
    Alternatively I could inject a muscle relaxant, or cut the appropriate nerve etc etc, then the subject could not move in spite of his/her volition."

    Yes, which is why I said hypnosis is a model for hard determinism rather than proof of it and why I didn't rely on it primarily. But there is a difference between those measures and hypnosis: they're qualitatively different from a person's ordinary state. We all enter states of mild hypnosis and are subject to mild hypnotic effects on a daily basis. So, if compatibilism doesn't apply to hypnotic states, it makes it more likely that its because compatibilism is wrong.

    ReplyDelete
  17. I am still having trouble with the issue of functional descriptions, I know you've given the example that the physical structures that lead to beliefs are heterogenous, but then so are hearts (for example that of a worm, compared to a chelonian, compared to a crocodile, compared to a mammal, or if we really open our definition the heart of a tree, a community, or an idea).

    Your clarification of why you choose to give the example of hypnotism is most helpful, I read about how even something like reading a book had elements of entering a hypnotic state, but was unsure how much weight to give such statements. The internet is rife with the stools of various artiodactyls. Perhaps you wouldn't mind expanding on this a little. Thankyou.

    ReplyDelete
  18. "I know you've given the example that the physical structures that lead to beliefs are heterogenous, but then so are hearts (for example that of a worm, compared to a chelonian, compared to a crocodile, compared to a mammal..."

    That's a correct point. It isn't, in fact, that the heterogeneous expressions of a belief is what makes belief a functional concept. It is, rather, that the same structure can serve different functions. But that's hard to show concretely for beliefs, and their heterogenous expressions (I hope) makes their multifunctionality more transparent.

    But let's go to hearts, since for hearts it seems easy to show how hearts aren't defined physically (which is to say, structurally). A heart is a heart by virtue of its role in a process, the circulation of blood to maintain life. The heart doesn't necessarily change structurally the instant the organism dies, but it ceases to be a heart. So, you can't determine that the organism has a heart by looking at a snapshot, because that doesn't tell you what role, if any, that blob of muscle tissue plays.

    The short answer is that one way functional concepts differ from physical concepts (assuming all physical concepts are structural—which might serve as a good definition of materialism) is that functional concepts describe properties that are expressed only over time, never instantaneously.

    ReplyDelete
  19. But the heart of a cadaver, or one in formalin, is still, generally considered a heart no? Just as a bedroom is still a bedroom if I am not using it?

    ReplyDelete
  20. So is a Valentine's Day heart. I wouldn't actually say that the "heart" of a cadaver is considered a heart. It's called a "heart" but it's really a different sense of the term. (The cadaver itself can be referred to as the dead person, but that doesn't mean they're essentially the same.) In any event, the fact that you can define a structural concept of the heart doesn't affect the conclusion that a functional concept doesn't correspond to a physical configuration. (In the case of "belief," there's no obvious structural entity to substitute, anyway.)

    "Your bedroom" is a functional concept defined over a longer period of time than when you're actually sleeping there. But it's a more conventional concept--as opposed to "heart," which is more a biological natural kind.

    ReplyDelete
  21. It seems that if you definition includes a verb that describes the action of the heart (beating, pumping, etc), the description must be functional. Otherwise you cannot be sure if an action is occurring.

    If the definition is a structure that could hypothetically perform a function then we can give a physical description.

    For example, a heart is a configuration of striated cardiac muscle, connective tissue and purkinje fibres, that possess the ability to pump the blood around the circulatory system, in the appropriate direction.

    We could apply a similar definition to memories, from what I understand, and beliefs don't seem a great stretch from memories.

    ReplyDelete
  22. Yes to all, but it's important to remember that the existence of a delimited structure responsible for a function isn't general to functional concepts. The "heart" example is useful because of the fact of a clear underlying structure, in that the example shows clearly that a functional view is compatible with a physicalist ontology. But not all functional concepts have matching structural descriptions. While there are functional "hearts" and structural "hearts," there is no structural "circulation of the blood."

    In the case of memories, concrete interactions of with the organism's environment must leave some structural record. There's a physical memory trace, which probably shouldn't be referred to as "the memory" but which serves as its discrete basis. There's no reason to think the same regarding beliefs.

    ReplyDelete
  23. Couldn't you in theory, describe a circulatory system and the liquid circulating, and describe each circulating molecule/protein/erythrocyte with a position and momentum? (Going purely classically).

    ReplyDelete
  24. Yes. The circulation example may be still too concrete to make my point clearly. Consider another: the process of organic evolution itself. What is the matching structural definition?

    ReplyDelete
  25. This has been something I've considered numerous times in the last month. The problem seems to be something like this:
    If asked to describe a travelling car you could describe each of the atoms (or go down to deeper levels if you wanted) and their relationships to each other etc. You could then assign them qualities such as kinetic energy, momentum, acceleration etc, which are all measures relating to past and future reference frames.

    The difficultly with describing evolution in a static reference frame then becomes clear. How do we define an "evolutionionary direction". The sorts of ideas that sprang to mind are you coud describe a certain structure decreasing, or increasing, in size, such as the human appendix and the human brain respectively. You could describe it in terms of allele frequency within an interbreeding population.

    A way around this would be to give up desribing a "direction" as such but just give a physical description of a number of time frames were what we refer to as evolution is occuring, and say something like "this is an example of a definition of evolution". To attempt to give an equivalent example for the car, it would be like saying "the car (consisting of....huge explanation) was travelling at 20mph at time A, 40mph at time B, time A preceeded time B" this is acceleration, though not an exclusive definition.

    ReplyDelete
  26. My question is based on a statement from Determinism in Wikipedia: Determinism rarely requires that perfect prediction be practically possible – only prediction in theory.

    This is a serious issue to me: A "deterministic" (theoretically) prediction can not be confirmed in practical terms, in the real world.
    So, what is this definition of "deterministic" good for ?
    If we can not (practically) confirm the results, in what do we need to believe ?
    In a theoretical result ?

    That makes no sense: Determinism does not require perfect prediction.
    To me the term "Determinism" should request for absolutely perfect prediction.
    And we all know, we have absolutely perfect prediction in almost nothing.

    But, if "Determinism" means we do not care about the quality of the prediction, then OK, everything is Deterministic, but I see not point in play like that.

    ReplyDelete
    Replies
    1. What do you have against believing in theoretical results? All of science consists of them.

      Delete
  27. Interesting that 'human organisms' replace 'human beings' in your dismissal of (2). Considered "as meat", we certainly look uncannily like meat. Conceived "as physical systems", we must be determinable. Ah, but conceived as hydrogen atoms and as oxygen atoms, we get ourselves in a pickle explaining the qualities of water. The trick is not to reduce too far, and in the case of human beings, don't we relevantly involve ourselves in stuff which isn't readily reducible to physics? Stuff such as, say, "meaning".

    To defend free will from such determinism, it is necessary to establish that not everything collapses directly into physics (and really - by what nifty bootstrapping trick can physics make a rule about that being compulsory?). It would be nice to delineate precisely what such irreducibles might be in the human case, but that's work for a full-time philosopher, or many of them. I'm content to accept that there's a semantic layer of reality which human beings access and from which some of our causes can issue, uncaused - a layer that articulates with but does not collapse into bare nomological physics).

    ReplyDelete
  28. You have no grounds for stating that the relevant microscopic states are deterministic.

    QM effects can be amplified to the macroscopic level.

    Human behaviour cannot in general be predicted.




    ReplyDelete

Blog Archive

About Me

Joshua Tree, California 92252-2141, United States
SUPPLIER OF LEGAL THEORIES. Attorneys' ghostwriter of legal briefs and motion papers, serving all U.S. jurisdictions. Former Appellate/Law & Motion Attorney at large Los Angeles law firm; J.D. (University of Denver); American Jurisprudence Award in Contract Law; Ph.D. (Psychology); B.A. (The Johns Hopkins University). E-MAIL: srdiamond@gmail.com Phone: 760.974.9279 Some other legal-brief writers research thoroughly and analyze penetratingly, but I bring another two merits. The first is succinctness. I spurn the unreadable verbosity and stupefying impertinence of ordinary briefs to perform feats of concision and uphold strict relevance to the issues. The second is high polish, achieved by allotting more time to each project than competitors afford. Succinct style and polished language — manifested in my legal-writing blog, Disputed Issues — reverse the common limitations besetting brief writers: lack of skill for concision and lack of time for perfection.