The Full-Libet experiment

I need a better name for it.

The Experiment: A Laplacian demon with exhaustive knowledge of the wave function of the universe hands you a sheet of paper that specifies whether you will be standing or not standing 30 seconds from now. He repeats the experiment ten times. You are entirely aware of everything the machine is trying to do, and you are hostile to its intentions. Will his predictions be necessarily true, that is, true no matter what you think about the matter?

Sean Carroll, for all his attempts to juggle clarity and nuance, is committed to saying yes. Einstein just said yes flat out. I’m constitutionally incapable of taking a “yes” as anything other than a reductio ad absurdum of any premise logically connected to it.

A failure of the Full-Libet limits the domain of physical law to those motions not initiated by intelligence. Given the infinitesimally small percentage of motions in the universe that fit this description, this is in one sense an insignificant qualification.





  1. February 14, 2017 at 9:28 pm

    This has nothing to do with free will. It is easy to specify a computer program that will read sentences saying, “you will output X” or “you will output Y” and output X if and only if the sentence says Y, and Y if and only if the sentence says X.

    The program is 100% deterministic, and it is impossible to write down on the paper what the program will output.

    • February 14, 2017 at 9:35 pm

      I knew you would respond. Hahaha

      I believe that someone can predict my motions when they tell me about it first and give me time to think about whether I’ll co-operate.

      So do you think the full libet is possible or not?

      • February 14, 2017 at 9:39 pm

        The scenario that you described? In an unlimited and general way, no, it is not possible, for the same reason you cannot write down what the program will do.

        In a limited way and in certain circumstances, yes. E.g. if I plan to carry out what I read.

      • February 15, 2017 at 6:47 am

        Can you flesh out the distinction you are making in the context of the question at hand? What is the difference between the limited and unlimited ways of a Laplacian demon giving someone a sheet of paper that tells them whether they will be standing or not standing in 30 seconds?

      • February 15, 2017 at 7:10 am

        I will simplify the description of the experiment in a way which should make the truth of this matter much more obvious.

        1. Let there be a falling ball, which is acting according to the physical laws of the universe. [This represents the human being.]

        2. Let there be a language defined as follows [this represents the English language and the paper]:
        a. hitting the ball upwards with a stick is defined to mean, “the ball will continue to fall.”
        b. waving the stick around without hitting the ball is defined to mean, “the ball will turn around and fly upwards.”

        Now for the experiment:

        3. Let there be a Laplacian demon with exhaustive knowledge of the wave function of the universe.

        Can the demon predict the behavior of the ball described in 1, using the language described in 2?


        No. The ball has “hostile” intentions relative to the demon here, namely because the language has causal interactions with the ball in such a way that a consistent wave function includes either “the demon predicts the ball will fall, but it will fly upwards,” or “the demon predicts the ball will rise, but it will fall.” Descriptions where the ball is correctly predicted are logically inconsistent: no wave function can imply them.

        Absolutely none of this, of course, shows that the behavior of the ball escapes the laws of physics. And of course, the demon may know what the ball will do; it simply cannot describe that using the described language.

        In the case of the human being, I said there was a limited possibility of the experiment, simply meaning that different human beings in different circumstances and times will respond differently. People do not consistently have hostile intentions to such an experiment, so it is quite possible on some occasions that I will be able to predict your response.

        But in general, the experiment corresponds to the above experiment with the ball, and does not show in any way that intelligent actions escape the laws of physics. Of course it doesn’t show that they don’t, but I did not say that: I said the experiment is irrelevant, and it is.

    • February 14, 2017 at 9:44 pm

      If there is a comprehensive wave-function of the universe, and the paper specifies what it entails, it would be impossible, by definition, for there to be anything on the paper other than what would happen, and thus any such countervailing program would be inconsistent with the universe having a deterministic history; you are illicitly assuming that a choice of programs in the universe is possible that is arbitrary in the sense of independent of the deterministic history of that universe.

      • February 14, 2017 at 9:48 pm

        I don’t think you understand the consequences of your own argument here.

        If there is a “comprehensive wave-function of the universe”, then the paper cannot specify what it entails, because the paper is included in the wave function, and the situation where the paper specifies what happens is inconsistent. Therefore the paper does not specify what the wave function entails, period.

        And that is why this whole experiment has nothing whatsoever to do with free will or determinism vs indeterminism.

      • February 14, 2017 at 10:14 pm

        Your response here seems to make literally the same illicit assumption: that it is possible, given that the paper identifies the state given by the wave-function, that we can arbitrarily have, in that universe, a computer program of the kind that you identify; and then, for whatever reasons, having assumed a contradiction, you somehow try to pull out of that an inconsistency to surprise me, as if one could not see the inconsistency immediately, or as if I had somehow forgotten that the scenario assumptions came first and that the inconsistency in question was generated solely by the addition of your arbitrary assumption that such a computer program could exist in such a scenario, or as if I had not just explicitly pointed out to you that your assumption was inconsistent with the assumptions of the scenario.

        I suppose your point could have been, despite your twice not having mentioned it, simply that we can make the assumption that arbitrary choice of programs is possible in a given universe because we know it somehow to be true and that therefore the scenario itself is impossible, which is the only other coherent possibility; but then it would be obviously false to claim that it has nothing to do with determinism.

      • February 15, 2017 at 7:15 am

        You are quite right that I made the “same assumption”; as I said I was drawing out the implications of your argument.

        The point isn’t whether such a program is possible, or such a paper is possible. The point is that you cannot have both the program and the paper. You can have the paper and a different program; or you can have the program but a paper which is wrong, or no paper.

        We actually do have human beings, and that tells us the wave function of the universe cannot include papers that describe what they are going to do. And notably, it doesn’t include such papers.

      • Carl said,

        February 15, 2017 at 12:02 pm

        Yes, this is the key. Any view presumes that the demon exists either inside the deterministic universe or outside. If inside, there’s no reason to think that the demon’s calculations are correct. It’s just acting according to brute laws, after all. If outside, then there is an outside, QED.

        I realized by reading the computer scientist Scott Aaronson that computer science presumes the existence of free will. If there is no free will, then the input into the machine is fixed, and if the input into the machine is fixed, then all of the algorithmic complexity hierarchies computer scientists spend their time describing are meaningless.

        For that matter, if there is no free will, science collapses as well because science learns something when a free agent performs and experiment that results in an observation that confirms or disconfirms an hypothesis. No free agent and the knowledge that science supposedly produces also disappears.

        All that said, my view of free will is fairly Kantian. We will never find inside the universe something acting outside of causal determinism because free will is not a thing in the universe. Free will can be thought of like this: Suppose that before the Big Bang, you preincarnated soul were asked “Should the universe where you rob be created or the universe where you don’t rob be created?” Your soul makes it choice and then the universe is created so that everything in it runs on mechanistic grounds.

        Based on this the Libet experiments are completely pointless. All they are measuring is the disconnect between conscious awareness of choice and subconscious beginning of choice. But the real choice was not made at the time you consciously become aware or unconsciously begin the mechanism of choosing. Rather, the real choice was made 14 BYA when the universe was created.

  2. February 15, 2017 at 9:33 am

    “A failure of the Full-Libet limits the domain of physical law to those motions not initiated by intelligence.”

    CS Lewis says something funny related to this in an essay in God in the Dock (although it was about miracles, not free will): the person who thinks that since the physics equations predict such-and-such, such-and-such will have to happen is like the person who thinks that, since two and two make four, then if I have two dollars in my drawer and add two dollars, then the next day there must be four dollars there. This person is forgetting of course that a thief might break into the house; and more generally doesn’t realize that the equations predict that such-and-such will happen … provided there is no interference.

  3. theofloinn said,

    February 15, 2017 at 10:18 pm

    All discussion of probabilities — whether something is certain, likely, unlikely, or impossible — must be assessed in the context of a particular model. There are no probabilities Pr(X) independent of the assumptions of the model being used Pr(X|M).

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: