Classical universe and experimental choices

[A. Neumaier]

Is a classical universe (with a deterministic dynamics) compatible with assuming that experimental choices can be made arbitrarily?

This question arises from the following statement of Bell:

'It has been assumed that the settings of instruments are in some sense free variables ...' For me this means that the values of such variables have implications only in their future light cones. They are in no sense a record of, and do not give information about what has gone before.

that I had cited (with a reference) in https://www.physicsforums.com/posts/5694545/, and commented with:

In a deterministic universe, this assumption is obviously violated. For it means that these values are independent of the values of all observables prior to the moment the settings are made. Indeed, if they would depend on the latter, fixing one of the values provides a nontrivial relation on the prior variables in their past light cone and hence provides information about the latter. Thus they give information about what was before, in direct contradiction to Bell's assumptions.

There was also a related statement by ueit in the same thread,

Once an initial condition (positions/velocities) is chosen everything becomes fixed. When the particles will be emitted by the source, what spin will they have, if and when they will be measured, what settings the detectors will have and what will be the result is a function of that particular initial condition. There is no need to use a "fine-tuned" initial condition. If you want the experimenters to "choose" a different setting you cannot implement that without changing the initial condition, and that change will also have an effect on the source and on the entangled particles themselves.

 

[A. Neumaier]

In this thread I want to continue part of the discussion initiated in a previous, now closed thread, namely that part that is independent of EPR-experiments and associated notions of superdeterminism. (The conflation with these seems to have been the reason the previous thread was closed.).

Thus this thread is not about EPR, Bell experiments, superdeterminism, or conspiracy arguments. Neither is it about the question whether such a universe replicates all predictions of quantum mechanics. It is solely about clarifying the conditions that one can or cannot assume in addition to the assumption of a deterministic universe.

I hope that in this way the discussion can be conducted without fear of getting it closed. Please do not derail the thread by using the words superdeterminism or EPR.

 

[atyy]

Is a classical universe (with a deterministic dynamics) compatible with assuming that experimental choices can be made arbitrarily?

Yes. Free will in Bell doesn't mean it's usual nonsensical meaning. The idea is that determinism can give rise to probability after coarse graining - the same as classical kinetic theory is consistent with Newtonian mechanics. In this case, free will just means that it is reasonable to model the measurement settings as independent. It is about the same as saying that we should be able to throw widely separated dice and model the joint probabilities as independent.

 

[atyy]

To add, superdeterminism is the idea that although we believe the dice to be independent, they actually are not (which is a logical possibility, just one that is not very useful for us to make predictions with).

 

[A. Neumaier]

Free will in Bell doesn't mean it's usual nonsensical meaning.

So you claim that it doesn't mean what Bell says it means, when he says, ''They [...] do not give information about what has gone before''?

 

[Haelfix]

If you had a super observer who had access to all the positions and momentum of every particle in this hypothetical world with perfect accuracy, then if you played the movie you could stop it, reverse it, play it again and all 'choices' would be repeated.

On the other hand with respect to the classical experimentalist (who arranges himself for a minute timeframe to make a measurement before all his/her electrons spiral into their nucleus'), he simply see's an open system and traces over the environmental degrees of freedom and has the almost perfect illusion of having free choice.

Of course, you could make two comments about this.
1) There is no superobserver, and its unclear if the finite speed of information makes every system open.
2) This is a very different proposition than if that same experimentalist (when measuring say the value of Newtons constant), also had it arranged so that all of his measurements would 'trick' him...

 

[stevendaryl]

The question can be phrased in terms of how much we need to know in order to make a prediction.

To make it concrete, let's assume that an experimenter, Alice as usual, is planning to make a yes/no choice at some time 1 hour in the future. Then even if we assume that Alice is made up of particles obeying deterministic laws, there are three obstacles to predicting her choice:

1. Her choice may depend on parts of the universe that are not currently accessible. As far as we know, there could be a lightspeed message heading toward Alice saying "Urgent! You must choose 'yes'! I will explain later. Bob."
2. Her choice may be hypersensitive to initial conditions. It could be that in order to predict Alice's choice, we would have to know the current positions and momenta of every particle in her body (and the surrounding neighborhood of a distance of 1 light-hour) to a ridiculous degree of accuracy.
3. Even if we do know the relevant state information to the relevant degree of accuracy, the computational time required to predict Alice's choice might take more than an hour.

If any of these three is the case, then for practical purposes Alice's choice is undetermined.

 

[N88]

The question can be phrased in terms of how much we need to know in order to make a prediction.

To make it concrete, let's assume that an experimenter, Alice as usual, is planning to make a yes/no choice at some time 1 hour in the future. Then even if we assume that Alice is made up of particles obeying deterministic laws, there are three obstacles to predicting her choice:

1. Her choice may depend on parts of the universe that are not currently accessible. As far as we know, there could be a lightspeed message heading toward Alice saying "Urgent! You must choose 'yes'! I will explain later. Bob."
2. Her choice may be hypersensitive to initial conditions. It could be that in order to predict Alice's choice, we would have to know the current positions and momenta of every particle in her body (and the surrounding neighborhood of a distance of 1 light-hour) to a ridiculous degree of accuracy.
3. Even if we do know the relevant state information to the relevant degree of accuracy, the computational time required to predict Alice's choice might take more than an hour.

If any of these three is the case, then for practical purposes Alice's choice is undetermined.

With respect and HTH; I'm concerned here re the already entangled semantics.

"The question can be phrased in terms of how much we need to know in order to make a prediction" -- but we need to know next to nothing to make a prediction. Check "your stars" in the paper today to see how others "predict".

Next. It seems to me that, in the GIVEN case: Alice's choice IS truly undetermined. For Alice is PLANNING to make a yes/no choice.

I'm assuming that you meant that Alice is planning to make (say) a YES choice. But if I need to assume this early in the discussion … then, with respect to engaging here, I'm sure you can predict (with certainty) what I am planning to say.

 

[stevendaryl]

With respect and HTH; I'm concerned here re the already entangled semantics.

"The question can be phrased in terms of how much we need to know in order to make a prediction" -- but we need to know next to nothing to make a prediction. Check "your stars" in the paper today to see how others "predict".

I meant a prediction that is guaranteed to be accurate.

Next. It seems to me that, in the GIVEN case: Alice's choice IS truly undetermined. For Alice is PLANNING to make a yes/no choice.

I'm assuming that you meant that Alice is planning to make (say) a YES choice. But if I need to assume this early in the discussion … then, with respect to engaging here, I'm sure you can predict (with certainty) what I am planning to say.

Sorry, I don't understand what you mean. We know ahead of time that Alice will either say "yes" or "no". We're trying to predict which one she will say. We're assuming for the sake of argument that Alice herself is describable by deterministic physics.

 

[N88]

Is a classical universe (with a deterministic dynamics) compatible with assuming that experimental choices can be made arbitrarily? …

In my view: Yes. Reason: Because our choices are made in a mutnauq world where the slightest nudge can produce a different eciohc; like something from my mischievous childhood.

Moreover, in my 'classical' world, having no need to accept that Bell proves "nonlocality" to be true, I have no need to resort to extreme 'out-there' explanations. (Well, as I see it: that at least is my choice.)

 

[N88]

I meant a prediction that is guaranteed to be accurate.

With respect, but to me: This is semantics in the extreme and a nice validation of my concern!

What would be a synonym for "a prediction that is guaranteed to be accurate"? What might be a helpful example of such?

Sorry, I don't understand what you mean. We know ahead of time that Alice will either say "yes" or "no". We're trying to predict which one she will say. We're assuming for the sake of argument that Alice herself is describable by deterministic physics.

Again, your "clarification" makes my point: I predict with certainty that any REAL Alice (animal, robotic, ...) lives in a world where h ≠ 0.

Thus, as d'Espagnat was fond of saying: REALITY is VEILED.

But note: Neither this veiled reality, nor h ≠ 0, stops me from choosing to think of (and study) the world "in a classical way". Even Einstein “argued that [spacelike-separated] correlations can be made intelligible only by completing the quantum mechanical account in a classical way,” Bell (2004:86).

PS: Given the OP's request, I've avoided using 'superdeterminism' or 'EPR'. I will likewise avoid 'Bell' in future.

 

[stevendaryl]

With respect, but to me: This is semantics in the extreme and a nice validation of my concern!

What would be a synonym for "a prediction that is guaranteed to be accurate"? What might be a helpful example of such?

Well, you can look up charts that show the dates for the full moon for the next 100 years, or the dates at which Venus and Mars and the Sun will be in alignment. So celestial mechanics is susceptible to pretty precise predictions.

Again, your "clarification" makes my point: I predict with certainty that any REAL Alice (animal, robotic, ...) lives in a world where h ≠ 0.

Thus, as d'Espagnat was fond of saying: REALITY is VEILED.

But note: Neither this veiled reality, nor h ≠ 0, stops me from choosing to think of (and study) the world "in a classical way". Even Einstein “argued that [spacelike-separated] correlations can be made intelligible only by completing the quantum mechanical account in a classical way,” Bell (2004:86).

I don't understand what you are talking about.

 

[N88]

...
I don't understand what you are talking about.

I was attempting to make (perhaps poorly) the general point that, in describing Alice, there is no need to assume that h = 0. In terms of the OP: It is solely about clarifying the conditions that one can or cannot assume in addition to the assumption of a deterministic universe. I wanted to be clear about my view: we can assume h ≠ 0.

 

[ueit]

Is a classical universe (with a deterministic dynamics) compatible with assuming that experimental choices can be made arbitrarily?

The problem is a little more complicated than it seems.

The short answer would be "NO". In a deterministic universe the concept of "choice" is meaningless.

However, in order to establish if it is justified to assume experimenter's freedom in a certain experiment one needs to answer the following question:

1. Does the theory that is tested in that experiment allow for the initial conditions to be chosen in such a way so that you can change the experimenter's "decision" without changing the rest of the experiment?

If the answer is "yes" then we can assume the experimenter is "free". We should expect that for a large number of experimental runs the "experimental choices" and the state of the system under observation should be statistically independent parameters. Newtonian mechanics of the rigid body is a theory where the experimenter can be assumed to be free. In this case, the system can be modeled as a group of billiard balls and the experimenter as another group of billiard balls. You can certainly imagine the experiment starting over and over with the system in the same state and the experimenter in a different state so that the experimenter measures the same system in different ways. One is then justified into combining the results of those experimental runs to get a consistent picture of the physical system.

If the answer is "no", obviously, the experimenter's freedom cannot be assumed. If you perform a delicate measurement of gravity or electromagnetism the experimenter is not free to move around the lab because the mass of his body directly affects the results. The initial state of the system includes the magnitude of the gravitational field in the vicinity of the system and that depends on the position of the experimenter. In this case one is not justified in combining the results of multiple experimental runs because each time the physical system is perturbed in a different way.

 

[ueit]

The question can be phrased in terms of how much we need to know in order to make a prediction.

To make it concrete, let's assume that an experimenter, Alice as usual, is planning to make a yes/no choice at some time 1 hour in the future. Then even if we assume that Alice is made up of particles obeying deterministic laws, there are three obstacles to predicting her choice:

1. Her choice may depend on parts of the universe that are not currently accessible. As far as we know, there could be a lightspeed message heading toward Alice saying "Urgent! You must choose 'yes'! I will explain later. Bob."
2. Her choice may be hypersensitive to initial conditions. It could be that in order to predict Alice's choice, we would have to know the current positions and momenta of every particle in her body (and the surrounding neighborhood of a distance of 1 light-hour) to a ridiculous degree of accuracy.
3. Even if we do know the relevant state information to the relevant degree of accuracy, the computational time required to predict Alice's choice might take more than an hour.

If any of these three is the case, then for practical purposes Alice's choice is undetermined.

I don't see any relevance of the "predictability" or "computation time" here. The only question is if the experimenter/system interaction can be eliminated. If no long range forces are present (like in a system of billiard balls) then there is no reason to suspect a dependence of the system on the experimenter's "choice" (other than that of a fine-tuning of the initial conditions). If the interaction cannot be eliminated (like in the case of field theories or even mechanical systems that are directly connected) then freedom cannot be assumed. Think of a system of 10 interconected pendulums. This is as unpredictable in practice as you can get. Now, if the experimenter's hand is physically connected to the pendulums we know for a fact that his choices will not be "free" in regards to this experiment. The fact that we cannot predict/compute anything is of no importance.

 

[stevendaryl]

I don't see any relevance of the "predictability" or "computation time" here. The only question is if the experimenter/system interaction can be eliminated. If no long range forces are present (like in a system of billiard balls) then there is no reason to suspect a dependence of the system on the experimenter's "choice" (other than that of a fine-tuning of the initial conditions). If the interaction cannot be eliminated (like in the case of field theories or even mechanical systems that are directly connected) then freedom cannot be assumed. Think of a system of 10 interconected pendulums. This is as unpredictable in practice as you can get. Now, if the experimenter's hand is physically connected to the pendulums we know for a fact that his choices will not be "free" in regards to this experiment. The fact that we cannot predict/compute anything is of no importance.

It depends on what your purpose is in modeling the choice as free. The purpose might be exactly predictability. The phrase "free" or "not free" doesn't have any physical (as opposed to philosophical) meaning independent of its role in modeling.

 

[ueit]

It depends on what your purpose is in modeling the choice as free. The purpose might be exactly predictability. The phrase "free" or "not free" doesn't have any physical (as opposed to philosophical) meaning independent of its role in modeling.

As far I can understand, the OP question relates to experiments, it is not a generic philosophical question about what it means to have free choice. And what matters in an experiment is if the "choices" made by the experimenter can or cannot perturb the system under observation. If the system is not perturbed, the evolution of the system up to the point of measurement can be assumed to be uncorrelated to whatever choice the experimenter makes, so the choice can be considered as "free" even in a deterministic setup.

From this point of view I do not see how replacing a simple decision device with a more convoluted one changes anything.

 

[A. Neumaier]

in such a way so that you can change the experimenter's "decision" without changing the rest of the experiment?

More is permitted - in such a way so that you can change the experimenter's "decision" without significantly changing the macroscopic view of the rest of the experiment!

The problem is a nontrivial one. Given the limited information we have about an experimental setting and the available choices, the question is whether there exist initial conditions of the universe such that for every possible choice, this information is reproducible to the accuracy with which it is known. Then that state would be a possible state of the universe, and hence the choice would be a valid choice. But the future would then also have to predicted from that state.

All this is independent of any issue in computational complexity, since the latter is only about our powers to duplicate Nature, not about what actually happens according to the deterministic dynamics.