Take the 2-minute tour ×
Physics Stack Exchange is a question and answer site for active researchers, academics and students of physics. It's 100% free, no registration required.

In his remarkable works (1,2 and their recent development 3) Thomas Breuer proves by diagonalization the phenomenon that the observer cannot distinguish all phase space states of a system where he is contained, a theorem paralleling the famous incompleteness theorem by Gödel.

This is applicable for both classical and quantum case, but for a quantum system the result is stronger. He thus calls the phenomenon subjective decoherence in case of quantum mechanics. He underlines that for quantum mechanics the phenomenon bears statistical character.

In (3) he gives the following illustration: enter image description here

More fundamentally, Breuer concludes that neither deterministic, nor probabilistic universally valid theories are possible: no theory can predict the future of the system where the observer is properly included.

Non-self-predictability implies that, even in a deterministic theory, accurate predictions about subsystems of the universe may at most be feasible for some observer, namely for one outside the system whose behaviour is to be predicted. But there will be no observer able to predict everything accurately. Every deterministic theory must admit the existence of unpredictable events when a predictor applies it to himself. Even in classical mechanics with a deterministic time evolution we have this kind of unpredictability.

I believe the conclusions of non-self-predictability and non-self-measurability are correct. In this chapter, they will follow in a more general framework from the fact that no observer can obtain or store information sufficient to distinguish all states of a system in which he or she is contained.

This is mentioned in Stanford Encyclopedia of Philosophy along a similar result by Marisa Dalla Chiara.

But recently I have encountered the essay by Scott Aaronson "The ghost in the quantum Turing machine" where he discussed the problem of free will. In his essay he references the famous argument by Peter van Inwagen who argued that neither determinism, nor randomness are compatible with free will (the part about determinism is known as the consequence argument, the full argument can be found in his monograph "Metaphysics"). Since Peter van Inwagen does not consider other theories than probabilistic and deterministic, he concludes that free will is impossible (quite a surprising conclusion for a Catholic professior).

In section 3.3 of his essay, Aaronson introduces an idea of "freebits". A freebit in his words, is simply a qubit for which the most complete physical description possible involves Knightian uncertainty.

Scott Aaronson introduces freebits as follows:

Thus, by the freebit picture, I mean the picture of the world according to which

  • (i) due to Knightian uncertainty about the universe’s initial quantum state $|\psi\rangle$, at least some of the qubits found in nature are regarded as freebits, and
  • (ii) the presence of these freebits makes predicting certain future events—possibly including some human decisions—physically impossible, even probabilistically and even with arbitrarily- advanced future technology.

The Knightian uncertainty is a term borrowed from economics to describe systems whose states have uncertain probability. A formal way of dealing with such variables is described by Dempster-Shafer theory, and in the appendix Aaronson gives a formal description of freebits.

In this light it is worth mentioning the idea that Breuer's self-referential uncertainty can also be traced to the initial state of the universe. In his article Ignorance of the Own Past Breuer proves that for an observer the past of a system where he is properly contained (such as the universe) is uncertain due to self-reference. The bad side in this argument is that the proof relies on the explicit assumption of determinism.

It should be noted though that the idea that free will may have self-referential nature is rejected by Aaronson on the grounds that in his view it amounts to solipsism. As such, he strives to ascribe free will to all animated beings, a position which seems to me unjustified. At best it looks as an attempt to artificially stretch a physical theory to fit certain philosophical belief.

On the other hand, it seems, subjective decoherence perfectly fits the definition of freebits. Since not all states of a quantum system (represented by a wave function, that is, probability) can be discriminated from the inside, such system seems to possess Knightian uncertainty, and as such, the freebits.

It should be noted that a theory dealing with uncertain probability, such as Knightian uncertainty (some call such theories "possibilistic" contrary the "probabilistic" theories, you can learn more about such theories in the comprehensive monograph on Generalized Information Theory by George Klir), would overcome the objections rised by both Breuer and Peter van Inwagen. Thus a universally-valid possiblistic (rather than probabilistic) theory would not be impossible even if Breuer's argument is correct. It also will not contradict the existence of free will as understood by van Inwagen. A drawback of such theory would be even smaller predictive power than that of a probabilistic theory.

That said, I wonder whether the idea of freebits, and Dempster-Shafer theory in general, is applicable to the description of the phenomenon of subjective decoherence.

share|improve this question
a tangential comment about free will and incompatibility with randomness. In most arguments of this type i have seen, it is implied that randmness is sth external and as a consequence one does not have power over it thus no free will to do othervise. In a sense this randmness represents an external law. But of course by the same token randomness can represent an internal law (or in other words) the external law is made of many, evolving internal laws which each person carries (and can vary) and this can be (part of) free will. –  Nikos M. Aug 23 '14 at 3:18
@Nikos M. indeed, in the argument randomness is understood as some probability distribution. –  Anixx Aug 23 '14 at 6:39

2 Answers 2

Some thoughts about Breuer (1995). Not really an answer, but too long to be a comment.

Breuer concludes that ... (1) no theory can predict the future of the system where the observer is properly included.

Breuer proves ... that (2) the observer cannot distinguish all phase space states of a system where he is contained.

How can one conclude (1) from (2)? Breuer's (1995) framework remains agnostic about the the possibility that an internal observer unable to distinguish all states of a system is nevertheless able to exactly measure/uniquely distinguish the particular state of the system now. In which case he is able to predict the future perfectly. At best, one may say in repeated experiments where the system changes over all possible states, some of those states are indistinguishable from each other, hence there be no consistent prediction about the future under those states. But this need not be a problem for self-measurements alone. For example, any external observer with finitely many states cannot distinguish all states of a system of sufficient complexity (sufficiently many possible states), given a fixed inference map $\theta$.

Another point. The key assumption that Breuer uses to prove (2) is the assumption of proper inclusion : enter image description here
Where $|_A$ describes a surjective map from the states of the observed system $O$ to the states of the observing apparatus $A$. The assumption says there exist different states of the system that map to the same state of the apparatus, which is almost what the conclusion (2) says. But in the case $S_A$ the set of states of the apparatus is infinite, there's no a priori reason why the assumption must hold. I think Breuer fails to give a reasonable account on this.

share|improve this answer

Free-bits have nothing to do with it.

We deal with van Inwagen by the following argument.

Determinism does not rule out free will, because free will within a system is necessarily deterministic. Otherwise it would be totally useless and would not be able to observe anything, because it would not be able to influence the apparatus used for observation. In other words, if I have free will and I decide to flip a coin, then my choice to flip the coin must necessarily determine my subsequent observation in which I see myself flip a coin and then observe the outcome. I must be able to move my eyes in the direction of the coin to see what happens.

For free will to exist within a system defined by the boundaries of traditional luminal communication and causality, does not require Knightian uncertainty leading up to the entrance of the observer onto the scene. No free-bits are necessary, because Breuer frees us from needing free-bits by pointing out that free will coming onto the scene causes its own Knightian uncertainties to arise, without the need for their pre-existence. I find this a more parsimonious approach than one that requires previously-existing free-bits, and it's one that Occam would prefer as well, I think.

We still need some Knightian uncertainty because if the observer could see the exact probability of every outcome, including their own, free will would not exist. Therefore Breuer's example provides the driver with an ongoing "blind spot" that does not require pre-existing free-bits. Rather this "blind spot" is conjured into being by the very presence of the observer and disappears when observation ceases.

Another consequence of Breuer is that if anyone should suggest that free will acts super-luminally and non-locally from outside the observable system via Bohmian mechanics -- as well it might -- such a scenario would not save us from Gödel because it requires the external observer to be entangled with the observed, making certain states (like Bell states) impossible to discriminate.

And this is good for free will because the unknowability of certain states frees the observer to choose based purely on free will rather than the previous state. It's not so much that the unknowability of certain states is a consequence of a previous state of things where there was Knightian uncertainty, as much as it's a consequence of the uncertainty principle.

So whether or not you want to accept Bohmian mechanics and spooky action at a distance, and I feel most physicists reject it, Breuer shows that such an external vantage point (or lack thereof) has no impact on the question of free will.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.