I The Extended Wigner's Friend Scenarios

[ChadGPT]

TL;DR Summary

Does Wigner change what the friend saw simply by entering the lab and asking? And how is it possible for Wigner to erase the memory of the friend simply by performing an interference measurement on the lab? And what is an example of a clear case that violates LF inequalities?

Brukner, Renner, and Cavalcanti recently won the Paul Ehrenfest award for their work deriving new no-go theorems involving an Extended Wigner's Friend Scenario, but I'm having trouble understanding certain aspects of how this thought experiment works. I'm trying to understand it conceptually. Here are my questions:

1) Brukner seems to say in some of his lectures that when Wigner opens the lab and asks the friend what he saw, that this can actually change what the friend saw. For instance, say the friend measures the particle along a certain axis and gets "spin up." Then, afterwards, Wigner opens the lab and asks the friend what he saw. This means that Wigner's wave function which includes the lab+friend+particle is projected into a definite state, with equal probability of finding that the friend says that he measured "spin up" and "spin down." So, sometimes, Wigner will ask the friend what he saw and the friend will say he remembers measuring "spin down" even though he actually originally measured "spin up." Since this particle is entangled with the particle that is being measured in the other lab, this change produced by Wigner opening the lab can affect what the friend in the other lab sees. Am I getting this right?

2) Also, instead of opening the lab and asking the friend what he measured, Wigner has the alternative option to perform a reverse unitary measurement on the entire lab+friend+particle, which has the effect of coherently erasing the friend's memory and returning the entire lab+friend+particle to a superposition. Wigner can then proceed to measure the particle directly, and may measure it "spin up" even if the friend measured it "spin down" originally. Is this right? And is it correct that just performing a certain kind of measurement from outside of the lab Wigner can actually cause the friend's memory to be erased? This is surprising to me.

3) Finally, I'm struggling to conceptualize a case in which the Local Friendliness assumptions are clearly violated. For instance, let's say Friend A measures the particle as "spin up", and Friend B measures the particle as "Spin down." This has to be the case due to the entanglement, right? At least when the two friends measure along the same axis. So then what can the Wigner's do to violate the inequalities? Perhaps Wigner A asks Friend A what was measured and Friend A says "Spin down," (changing his result) and then Wigner B performs a reverse unitary transformation on Lab B, erasing Friend B's memory, and then measures Particle B to be "spin up." The results of the two Wigners would then contradict the results of the two friends, but would still be consistent with the entangled state. Is something like that what is going on? Or something else.

Thank you in advance for your help and patience. :)

 

[Nugatory]

Brukner, Renner, and Cavalcanti … for their work

Links? ArXiv is ok if the formal publications are behind a paywall.

 

[ChadGPT]

Brukner's Original Paper: https://arxiv.org/abs/1507.05255
Brukner's second paper: https://arxiv.org/abs/1804.00749
Cavalcanti's Paper: https://arxiv.org/abs/1907.05607

 

[PeterDonis]

Brukner seems to say in some of his lectures that when Wigner opens the lab and asks the friend what he saw, that this can actually change what the friend saw.

Yes, but there is an unstated assumption: in order to "actually change what his friend saw", Wigner has to be able to reverse decoherence.

In lab experiments which have been called "Wigner's friend" experiments, this crucial part is eliminated by doing the experiments on qubits, and maintaining their coherence; so the analogue of "Wigner opens the lab and asks the friend what he saw, and this can actually change what the friend saw" is just an ordinary unitary operation on a qubit, which has been commonplace for some time now, and involves no reversal of decoherence because there is no decoherence to begin with.

But in a "Wigner's friend" experiment with humans instead of qubits, "what the friend saw" does involve decoherence, and Wigner would have to be able to reverse that decoherence to change what the friend saw. Of course this is not even remotely possible in practice; but the physicists who are making these claims are relying on the implicit claim of QM as it currently stands that it somehow is possible in principle. I don't think this crucial aspect is emphasized anywhere near enough.

Wigner has the alternative option to perform a reverse unitary measurement on the entire lab+friend+particle

Which, again, would require Wigner to be able to reverse decoherence. Note that "the entire lab+friend+particle" is going to include an astronomical number of degrees of freedom (something like $10^{30}$ or more at an absolute bare minimum), and all of them would have to be coherently controlled in order to make this "reverse unitary measurement". And this also assumes that this whole shebang is perfectly isolated from the rest of the universe; one stray photon escaping to the outside and it all falls apart. Again, I don't think this crucial aspect is emphasized anywhere near enough.

let's say Friend A measures the particle as "spin up", and Friend B measures the particle as "Spin down." This has to be the case due to the entanglement, right?

No. The two friends could measure two different particles as spin up and spin down, and if the particles were entangled, such measurements would have to show particular correlations.

But for the two friends to get opposite results measuring the spin on the same particle, again, the decoherence that took place with Friend A's measurement would have to be reversed before Friend B made his measurement.

 

[ChadGPT]

Thank you for your reply, that helps clarify a few things. I think I wasn't clear the way I worded the final question though. I didn't mean to suggest that Friend A and Friend B measure the same particle. In the EWFS, Friend A and Friend B are in two space-like separated labs, which each contain one particle of an entangled pair. So they are measuring two different particles, but these particles are entangled, much like in the Bell experiments.

One thing I'm still not clear on is how to visualize a case in which a clear violation of the LF inequalities occurs. In the experiment, the two Friend's in the two labs each measure their particle at a time prior to the two Wigners making their free choice on how to measure. The Wigner's can 1) open the lab and ask the friend what was measured, or 2) perform a reverse unitary measurement on the entire lab+friend+particle, reverting it back into a superposition state, and then Wigner can himself measure the spin of the particle in question.

The specific LF assumption that the theorists hone in on is what they call Absoluteness of Events (AoE). This is the idea that once something has been measured, it becomes an absolute fact about the state of the objective universe. They then state that these LF inequalities, which rely on AoE (and also Locality and no-superdeterminism) are violated in several cases. But they never explicitly say what these cases might look like.

I can imagine a case in which, say, Friend A measures spin up and Friend B measures spin down, then both Wigners choose to perform the unitary and revert the labs back into a superposition, and then measure the particles, resulting in Wigner A measuring spin down and Wigner B measuring spin up. This would violate AoE because if AoE were true, then Wigner A should get the same measurement result as Friend A did originally, and the same for Wigner B and Friend B's original measurement.

Is this possibly close to what they are suggesting happens?

 

[ChadGPT]

For clarification, here are some diagrams of how the EWFS experiment is setup in the Cavalcanti version:

 

[PeterDonis]

In the EWFS, Friend A and Friend B are in two space-like separated labs, which each contain one particle of an entangled pair. So they are measuring two different particles, but these particles are entangled, much like in the Bell experiments.

That just means that the measurement results Friend A and Friend B get must be consistent with the same entangled state of the particles. For example, if the two particles are in the singlet state, and both friends measure their spins along the same axis, the two results must be opposite. If Friend A and Friend B make their measurements and do find opposite results, then in order for Wigner to "undo" this, he would have to be able to undo the decoherence of both friends. Just one wouldn't be enough, because of the entanglement; undoing, say, just Friend B and their particle would just put things back to before Friend B made his measurement, and if Friend A's result is still the same, Friend B's still has to be opposite. There's no way to have the two particles in the singlet state and have Wigner somehow make both friends' results come out the same.

I can imagine a case in which, say, Friend A measures spin up and Friend B measures spin down, then both Wigners choose to perform the unitary and revert the labs back into a superposition, and then measure the particles, resulting in Wigner A measuring spin down and Wigner B measuring spin up.

This would only be possible if Wigner reversed the decoherence of both friends. If that is assumed to be possible, then yes, Wigner could do this. But that's a huge assumption.

 

[ChadGPT]

yes, Wigner could do this.

Okay then, that would explain why they are saying this is a no-go for assumptions like "Absoluteness of Events (AoE)". If Particle A is measured by Friend A to be "spin up", then assuming that AoE is true, this should mean this is an objective fact about the particle that should be true for all observers. Yet, if Wigner A measures the lab+friend+particle along the "Deutsch Basis," this projects the lab+friend+particle back into a superposition state. Thus, when Wigner A enters the lab and measures the particle, it will again be perfectly random whether Wigner A measures it as either "spin up" or "spin down," meaning that the "spin up" result is not a definite objective fact, but just a relative fact, since Wigner might measure it as "Spin Down" even though the Friend originally measured the same particle as "spin up."

But that's a huge assumption.

But it's technically allowed by quantum mechanics. Perhaps using quantum computers and advanced Ai in place of human beings it is even possible in the not too distance future to test. Assuming we have the tech to control all of the free variables in the isolated lab, the result of performing a unitary interference measurement would give us one of two results: 1) Every time the interference measurement is performed we always get the outcome consistent with the lab+friend+particle in a superposition state, confirming that the lab+friend+particle is in a superposition state, and thus erasing the friend's memory and undoing the decoherence, or 2) When Wigner performs the interference measurement he gets either possible outcome 50% of the time, meaning that the lab+friend+particle is in a mixed state, and not in a pure entangled state, such that the friend's memory is not erased, and Wigner is not able to undo the decoherence.

Without being able to actually perform the experiment we will not know whether 1) or 2) will result. However, 1) allows that the friend's memory is erased and the decoherence is undone by the measurement from outside the lab, but 2) is not predicted by the current quantum formalism, and would thus suggest "new physics" according to Brukner. Wigner, however, seems to have predicted that 2) will result. As Cavalcanti writes:

Importantly, Wigner’s proposed solution is an operationally distinct theory. He predicts, via his argument, that the quantum state |Ψ1⟩SF would give wrong predictions for Wigner— e.g. no interference effects could be observed. This is an empirical matter, which cannot be decided by argumentation alone. I shall leave this empirical question for future experiments to decide, perhaps with human-level AI agents running in sufficiently large quantum computers.
If Wigner’s conjecture turns out to be incorrect, this would in particular imply that Local Friendliness inequalities can be violated with conscious observers.

By "operationally distinct theory" I think Cavalanti mean's that Quantum Mechanics would need to be updated, if 2) is confirmed experimentally. Right?

I mean, it would mean that, even though the superobserver has absolutely no information regarding anything that is happening inside the lab, and he can exclude all known effects caused by conventional decoherence, nevertheless the state is not in a superposition. Quantum Mechanics does not predict this result. Something would be need to added to reflect why this is the case. Otherwise, quantum mechanics just gives the super observer wrong predictions.

So it seems we are left with two not-great options. Both 1) and 2) are disconcerting in one way or another.

 

[PeterDonis]

If Particle A is measured by Friend A to be "spin up", then assuming that AoE is true, this should mean this is an objective fact about the particle that should be true for all observers.

Yes, but that assumption requires that decoherence cannot be reversed, because decoherence is (part of) what makes the observation an "objective fact" that all observers must agree on.

If decoherence can be reversed, then it is impossible for AoE to be true.

it's technically allowed by quantum mechanics.

Sort of. QM, as we have actually tested it experimentally, includes the AoE, because in order to test it, we have to assume that the measurement results we are comparing with the theoretical predictions are objective facts.

It's true that you can wave your hands and write down mathematical expressions that describe things like Wigner undoing Friend A's and Friend B's measurement results and then having them find different ones. But then what about Wigner? How does Wigner know that he's done this? Is what Wigner does objective fact? Or is there some super-Wigner that undoes what Wigner did? And so on up an infinite chain of super-super-Wigners. That's the logical conclusion of this type of speculation. And it leads nowhere; we could never know anything. The whole viewpoint undermines itself.

In order to have a workable theory of physics at all, all this has to bottom out somewhere in something that's objective. And once you realize that, it's obvious where that bottoming out should be: in the measurements we actually make and the results we actually observe.

even though the superobserver has absolutely no information regarding anything that is happening inside the lab, and he can exclude all known effects caused by conventional decoherence, nevertheless the state is not in a superposition.

If this were the case, leaving aside all the concerns I expressed above, yes, this would mean QM as it stands would have to be modified, since it does not predict such a thing.

However, it seems to me that a much simpler way out is to simply not believe that such "superobservers" can exist, because it is impossible to maintain coherence over the number of degrees of freedom that would be required for a "superobserver" to perform the operations that are attributed to them. Strictly speaking, it would not be impossible in principle, just as, according to our best current understanding, it is not impossible in principle that the second law of thermodynamics could be violated. But the expected time for such a thing to happen is many, many orders of magnitude longer than the lifetime of the universe.

 

[ChadGPT]

However, it seems to me that a much simpler way out is to simply not believe that such "superobservers" can exist, because it is impossible to maintain coherence over the number of degrees of freedom that would be required for a "superobserver" to perform the operations that are attributed to them.

I agree that it does not seem to be possible using human beings, but I find the arguments that one could use quantum computers and advanced AI to perform the experiment convincing. Quantum computers already use methods to perfectly isolate systems in order to prevent decoherence, so that part does not seem to be a problem. And when we are talking about an AI instead of a human being, it seems feasible to be able to control all of its degrees of freedom by another AI outside of the isolated system.

The theorists involved all seem to agree that this should indeed be a possible experiment to perform, perhaps within the next 10 to 20 years. If this is done, then it will be interesting to see the result. Either QM as it stands would have to be modified, or we have to accept that AoE is false and decoherence can be reversed.

Thanks for the back and forth PeterDonis, this helped me understand a little better. If anyone else would like to give an opinion on all of this, I'm interested to hear what other people think.

 

[PeterDonis]

I find the arguments that one could use quantum computers and advanced AI to perform the experiment convincing.

I don't. A quantum computer and advanced AI that was capable of the same kind of experience as a human would have to have as many degrees of freedom as a human. That's many, many, many orders of magnitude too many for...

Quantum computers already use methods to perfectly isolate systems in order to prevent decoherence

...this to work.

The theorists involved all seem to agree that this should indeed be a possible experiment to perform, perhaps within the next 10 to 20 years.

Can you give some specific references? I strongly suspect that the experiments they are considering still involve a number of degrees of freedom that is many, many, many orders of magnitude smaller than what would be required to do this for a human, or any other being of roughly equivalent complexity.

when we are talking about an AI instead of a human being, it seems feasible to be able to control all of its degrees of freedom by another AI outside of the isolated system.

This doesn't seem at all feasible to me. "AI" is not magic; it doesn't somehow evade the fact that a huge number of degrees of freedom ($10^{30}$ or more) cannot be feasibly controlled. We humans don't control all of those degrees of freedom in ourselves; not even close; not even close by many, many, many orders of magnitude. An AI with that many degrees of freedom wouldn't be able to control them all, or even come close, by the same huge, huge, huge gap.

 

[Demystifier]

Wigner-type thought experiments require control over a large number of degrees of freedom. However, they do not necessarily require a time-reversal of them. Instead, such experiments can also be based on quantum measurements in a basis of a Hilbert space of a system with a large number of degrees of freedom. In fact, these two kinds of Wigner-type experiments are equivalent: either you perform a measurement in a basis of such a big system, or you perform a time reversal so that you can perform an equivalent measurement in a basis of a small system. This tradeoff and equivalence is explained in https://arxiv.org/abs/2308.16220 Sec. II.E.

 

[PeterDonis]

Wigner-type taught experiments require control over a large number of degrees of freedom. However, they do not necessarily require a time-reversal of them.

No, but they do require the ability to perform unitary operations on them that maintain quantum coherence. That's where the issue is.

 

[Morbert]

I agree that it does not seem to be possible using human beings, but I find the arguments that one could use quantum computers and advanced AI to perform the experiment convincing.

Roland Omnes attempted to quantify the number of degrees of freedom a superobserver apparatus would need to have in order to make these kinds of measurements on macroscopic objects. The estimate he arrived at was $10^{10^{18}}$. For comparison, the total number of protons within the visible horizon is approximately $10^{80}$.

 

[Demystifier]

The review https://arxiv.org/abs/2308.16220 argues that, among the many extended Wigner-type thought experiments considered recently, the most convincing conclusions are those of the local friendliness (LF) no-go theorem. This theorem is very similar to the Bell theorem, except for the following differences:

1. The Bell theorem assumes both parameter independence (the choice of apparatus settings on the Alice side do not influence the measurement outcomes on the Bob side) and outcome independence (the measurement outcomes on the Alice side do not influence the measurement outcomes on the Bob side). By contrast, the LF theorem assumes only the parameter independence. In that sense the assumptions of the LF theorem are weaker, which makes the theorem stronger. This is a technical difference between the two theorems.

2. The Bell theorem implicitly assumes that the measurement outcomes objectively exist, but does not associate such outcomes with conscious observers. By contrast, in the spirit of other Wigner-friend type experiments, the LF theorem associates the measurement outcomes with conscious observers. This is a conceptual philosophical difference, which does not influence the logical/mathematical aspects of the theorem.

Then the two theorems proceed in a similar way. By taking the locality assumption (1.), the realism assumption (2.) and the no-superdeterminism assumption (the measurement outcomes in the past are not correlated with the choices of apparatus settings in the future), one derives a contradiction with quantum mechanics.

We emphasize that locality (1.) is one of the assumptions of the theorem, so one way to avoid the conclusion that objective outcomes don't exist is to accept non-locality. In that context one might argue that locality is not really an issue because another theorem, the Frauchiger-Renner one, arrives at similar conclusions without assuming locality. However, the review cited above points out that a locality assumption is implicit in the Frauchiger-Renner theorem because the theorem rests on an assumption that timings of the choices are relevant, while such an assumption is difficult to justify without some kind of a locality assumption.

 

[Sambuco]

2. The Bell theorem implicitly assumes that the measurement outcomes objectively exist, but does not associate such outcomes with conscious observers. By contrast, in the spirit of other Wigner-friend type experiments, the LF theorem associates the measurement outcomes with conscious observers. This is a conceptual philosophical difference, which does not influence the logical/mathematical aspects of the theorem.

I think the inclusion of conscious observers is done to avoid some issues related to the measurement problem. More precisely, there is consensus that conscious observers are always on the classical side of the Heisenberg cut, which ensures that when Wigner's friend makes a measurement, it is accepted that an event occurred, at least relative to Wigner's friend.

Then the two theorems proceed in a similar way. By taking the locality assumption (1.), the realism assumption (2.) and the no-superdeterminism assumption (the measurement outcomes in the past are not correlated with the choices of apparatus settings in the future), one derives a contradiction with quantum mechanics.

When you say that the assumptions of the theorem lead to a contradiction with quantum mechanics, are you assuming what is often called "universality"? That is, that all observers can model their own measurement as an entirely unitary process. I mention this because I think one way to respond to the theorem is to assume an orthodox (textbook) interpretation where, once Wigner's friend carries out a measurement, there is an objetive system's state-update ("collapse"), so Wigner is not allowed to model the "large" system (system + friend) as a unitarily evolving quantum system.

Lucas.

 

[Sambuco]

Okay then, that would explain why they are saying this is a no-go for assumptions like "Absoluteness of Events (AoE)". If Particle A is measured by Friend A to be "spin up", then assuming that AoE is true, this should mean this is an objective fact about the particle that should be true for all observers. Yet, if Wigner A measures the lab+friend+particle along the "Deutsch Basis," this projects the lab+friend+particle back into a superposition state. Thus, when Wigner A enters the lab and measures the particle, it will again be perfectly random whether Wigner A measures it as either "spin up" or "spin down," meaning that the "spin up" result is not a definite objective fact, but just a relative fact, since Wigner might measure it as "Spin Down" even though the Friend originally measured the same particle as "spin up."

That's not what "Absoluteness of Observed Events" (AOE) means. Even if Wigner could undone the measurement performed by his friend, this measurement has an observer-independent (absolute) result. In other words, AOE means that the outcome of the measurement his friend made in the lab is something objetive.

To clarify this point, I refer to a quote from the original paper about local-friendliness (https://www.nature.com/articles/s41567-020-0990-x):

the assumption of AOE only entails assigning truth values to propositions about observed outcomes. In particular, Alice’s measurement outcome Ax (which in our notation corresponds to the value of a when she performs the measurement labelled by x) for x ≠ 1 has a value only when she performs that measurement. However, A1 is different in that it has a value even when x ≠ 1, because it is encoded in c, which is actually measured by Charlie in every run. All this is in keeping with Peres’ dictum ‘unperformed experiments have no results’27; AOE is the assumption that performed experiments have observer-independent (that is, absolute) results.

Lucas.

 

[Sambuco]

If decoherence can be reversed, then it is impossible for AoE to be true.

That's not entirely true. There are interpretations where decoherence can be reversed (at least in principle) for which AOE still holds, such as Bohmian mechanics.

It's true that you can wave your hands and write down mathematical expressions that describe things like Wigner undoing Friend A's and Friend B's measurement results and then having them find different ones. But then what about Wigner? How does Wigner know that he's done this? Is what Wigner does objective fact? Or is there some super-Wigner that undoes what Wigner did? And so on up an infinite chain of super-super-Wigners. That's the logical conclusion of this type of speculation. And it leads nowhere; we could never know anything. The whole viewpoint undermines itself.

It's worth noting that this isn't what AOE means. That a super-observer can reverse a measurement performed by his/her friend is not the same as saying that the result of that measurement is not absolute (see #17).

QM, as we have actually tested it experimentally, includes the AoE, because in order to test it, we have to assume that the measurement results we are comparing with the theoretical predictions are objective facts.

Not exactly. In fact, there is an ongoing debate in the community about this. For example, Carlo Rovelli claims that his original formulation of relational quantum mechanics (RQM) is sufficient to explain our experience and way of doing science (https://arxiv.org/pdf/2410.20012), while Emily Adlam claims that we need some kind of additional postulate that guarantees intersubjectivity, and proposed what is known as "cross-perspective links" (https://arxiv.org/abs/2502.06991).

Lucas.

 

[PeterDonis]

There are interpretations where decoherence can be reversed (at least in principle) for which AOE still holds, such as Bohmian mechanics.

Do you have a reference for this? I'm not aware of this being a feature of Bohmian mechanics.

That a super-observer can reverse a measurement performed by his/her friend is not the same as saying that the result of that measurement is not absolute (see #17).

This is just adopting a meaning of "absolute" that makes the term useless. If the Friend measures particle A to be spin up, and then Wigner reverses the measurement by performing some quantum operation on the system particle A + Friend + Friend's lab (and whatever else is involved in the measurement the Friend made), then after Wigner has performed the operation, the Friend no longer remembers measuring particle A as spin up, he longer believes particle A was measured as spin up, and all other features of any other part of the quantum system that was involved in the measurement are no longer as they were when the Friend completed his measurement and before Wigner performed his operation. Any meaning of the word "absolute" that allows this to happen to something "absolute" is pointless. "Absolute" is supposed to mean that no future events can change what happened--but the whole point of the Wigner's Friend thought experiment is that Wigner's operation changes what happened. Everything in the universe is as if the Friend never made his measurement at all.

As I've said, I personally believe that the best way to interpret the thought experiment is as a reductio ad absurdum of the claim that the kind of quantum operation attributed to Wigner in the thought experiment is possible at all. Gerrymandering the meaning of the word "absolute" doesn't change that.

 

[Sambuco]

Do you have a reference for this? I'm not aware of this being a feature of Bohmian mechanics.

About what? About reversing decoherence? If so, in Bohmian mechanics there is a global wavefunction that never collapses, so it is compatible with the (usually implicit) assumption of "universality", that is that every observer can model his/her system under measurement as a quantum mechanical system.

This is just adopting a meaning of "absolute" that makes the term useless. If the Friend measures particle A to be spin up, and then Wigner reverses the measurement by performing some quantum operation on the system particle A + Friend + Friend's lab (and whatever else is involved in the measurement the Friend made), then after Wigner has performed the operation, the Friend no longer remembers measuring particle A as spin up, he longer believes particle A was measured as spin up, and all other features of any other part of the quantum system that was involved in the measurement are no longer as they were when the Friend completed his measurement and before Wigner performed his operation. Any meaning of the word "absolute" that allows this to happen to something "absolute" is pointless. "Absolute" is supposed to mean that no future events can change what happened--but the whole point of the Wigner's Friend thought experiment is that Wigner's operation changes what happened. Everything in the universe is as if the Friend never made his measurement at all.

Absoluteness of Observed Events (AOE) is a technical term defined as in the paper by Cavalcanti's group (https://www.nature.com/articles/s41567-020-0990-x). You're using "absolute" with a different meaning. I'm not denying what you said about the implications of these kinds of thought experiments; I'm just saying that, in this context, AOE has a different and precise meaning.

As I've said, I personally believe that the best way to interpret the thought experiment is as a reductio ad absurdum of the claim that the kind of quantum operation attributed to Wigner in the thought experiment is possible at all. Gerrymandering the meaning of the word "absolute" doesn't change that.

Even though I favour the idea that these theorems (particularly the local friendliness one) slightly suggest that observed events are not absolute, but relative, I agree with you that one possible way out is simply to deny the possibility that Wigner could undo his friend's measurement, even in principle. Some consider this possibility compatible only with physical collapse, as in the Girardi-Rimini-Weber interpretation, but it is also possible that our current understanding of quantum theory is incomplete. For example, the little-known Montevideo interpretation claims to solve these problems by considering real clocks. I quote a paper by Gambini and Pullin. (https://www.mdpi.com/2218-1997/6/12/236):

All of the recent criticisms based on Wigner’s friend, like that of Frauchiger and Renner [53,54] to quantum mechanics, stem from the assumption that the unitary evolution predicted by ordinary quantum mechanics with a classical time is correct with infinite precision. The quantum decoherence due to the use of real clocks eliminates all of the problems mentioned before. Events can happen without violating the causal evolution described by the master equation, with probabilities that coincide exactly with those that were predicted by the textbook interpretations.

Lucas.

 

[Demystifier]

I think the inclusion of conscious observers is done to avoid some issues related to the measurement problem. More precisely, there is consensus that conscious observers are always on the classical side of the Heisenberg cut, which ensures that when Wigner's friend makes a measurement, it is accepted that an event occurred, at least relative to Wigner's friend.


When you say that the assumptions of the theorem lead to a contradiction with quantum mechanics, are you assuming what is often called "universality"? That is, that all observers can model their own measurement as an entirely unitary process. I mention this because I think one way to respond to the theorem is to assume an orthodox (textbook) interpretation where, once Wigner's friend carries out a measurement, there is an objetive system's state-update ("collapse"), so Wigner is not allowed to model the "large" system (system + friend) as a unitarily evolving quantum system.

Lucas.

Yes, universality is a crucial assumption in all those Wigner-friend thought experiments.

 

[Demystifier]

Do you have a reference for this? I'm not aware of this being a feature of Bohmian mechanics.

Of course it is. Clearly, the wave function can be reversed for one or few particles in any interpretation of QM, it must be so because this possibility is an experimental fact. But Bohmian interpretation (as well as all other interpretations which deny a true collapse) says that the laws of physics for macroscopic objects don't differ from those of microscopic objects, hence the wave function can be reversed even for macroscopic objects.

 

[PeterDonis]

in Bohmian mechanics there is a global wavefunction that never collapses

Bohmian interpretation (as well as all other interpretations which deny a true collapse) says that the laws of physics for macroscopic objects don't differ from those of microscopic objects, hence the wave function can be reversed even for macroscopic objects.

But the wave function in Bohmian mechanics does not describe the actual, physical state of the system. That is described by the unobservable particle motions. The wave function is just part of the dynamical law for the unobservable particle motions. So reversing the wave function by itself is not enough to undo a measurement; you would also have to reverse all of the unobservable particle motions. But Bohmian mechanics says that's impossible--you can't even observe the unobservable particle motions, much less magically reverse them.

 

[Demystifier]

But the wave function in Bohmian mechanics does not describe the actual, physical state of the system. That is described by the unobservable particle motions. The wave function is just part of the dynamical law for the unobservable particle motions. So reversing the wave function by itself is not enough to undo a measurement; you would also have to reverse all of the unobservable particle motions. But Bohmian mechanics says that's impossible--you can't even observe the unobservable particle motions, much less magically reverse them.

Yes, but that's not needed for the extended Wigner friend experiment. The experiment only involves a reversion (or other kinds of manipulation) on the wave function.

 

[Sambuco]

But the wave function in Bohmian mechanics does not describe the actual, physical state of the system. That is described by the unobservable particle motions. The wave function is just part of the dynamical law for the unobservable particle motions. So reversing the wave function by itself is not enough to undo a measurement; you would also have to reverse all of the unobservable particle motions. But Bohmian mechanics says that's impossible--you can't even observe the unobservable particle motions, much less magically reverse them.

Good point!
Perhaps @Demystifier can answer better, but if we reverse decoherence by making the wave function evolve "as if" it were going backward-in-time, doesn't that imply that the particles travel the same paths they traveled before, but now in the opposite direction?

Lucas.

 

[Demystifier]

Good point!
Perhaps @Demystifier can answer better, but if we reverse decoherence by making the wave function evolve "as if" it were going backward-in-time, doesn't that imply that the particles travel the same paths they traveled before, but now in the opposite direction?

Lucas.

That's a good question. It would be so if the reversal of the wave function was ideal, meaning that it is evolved with the Hamiltonian $-H$. But no reversal of the wave function is ideal in that sense, it's impossible to have the Hamiltonian $-H$ (for instance, the kinetic part of the Hamiltonian is always positive). Even for very simple systems, say one coherent photon, the reversal does not work that way. You can, for instance, reflect the photon on the mirror, which is a kind on reversal, but it is not the ideal reversion in that sense.

But the ideal reversion is not what is needed in the Wigner-friend experiments. You do not need to return the photon in the exact same state as before. In such experiments you are typically interested only in the state of polarization of the photon, so you must return the photon to the same state of polarization as before, while the spatial part of the "wave function" can be different. Thus the Bohmian positions and velocities will not return to the same exact values. Likewise, in the thought experiments with Wigner friend, you want to return her brain to the same memory state as before, but there are many different states of the Wigner friend which correspond to the same memory state of her brain. We may not completely understand how brain memory works, but for our purpose it is not completely wrong to compare it with classical computer memory, and clearly two computer chips in the same memory state are not in the exact same quantum state.

 

[PeterDonis]

that's not needed for the extended Wigner friend experiment

Yes, it is, since the unobservable particle motions determine the results of all experiments in Bohmian mechanics. So to get the results you want, you can't just reverse the wave function.

in the thought experiments with Wigner friend, you want to return her brain to the same memory state as before, but there are many different states of the Wigner friend which correspond to the same memory state of her brain

This doesn't change anything I said before. Sure, a given memory state of the friend corresponds to some large subspace of the Hilbert space of the friend's brain. But the subspaces corresponding to different memory states are orthogonal, and Wigner's measurement has to involve interference between different macroscopically distinguishable states of the friend and her laboratory, and hence is done in a basis consisting of states which are superpositions of the different subspaces corresponding to different memory states. And hence Wigner's measurement has to involve either reversing decoherence in the friend/laboratory, or maintaining the friend/laboratory in a state of quantum coherence and not letting decoherence happen there at all (meaning the friend never actually observes anything, because the state of her brain never gets into a single subspace corresponding to a single memory state).

 

[Demystifier]

Yes, it is, since the unobservable particle motions determine the results of all experiments in Bohmian mechanics. So to get the results you want, you can't just reverse the wave function.

I disagree. In other interpretations of QM the results of all experiments are random, not determined by anything. By not controlling and not reversing Bohmian hidden variables, you effectively make the results of all experiments random, just as in other interpretations. By contrast, if you insist that Bohmian hidden variables must be reversed as well, than you are not merely considering a Bohmian interpretation of the same experiment, but an entirely different experiment, which would be even more difficult to perform than the original one.

 

[gentzen]

That's not entirely true. There are interpretations where decoherence can be reversed (at least in principle) for which AOE still holds, such as Bohmian mechanics.

Do you have a reference for this? I'm not aware of this being a feature of Bohmian mechanics.

Of course it is. Clearly, the wave function can be reversed for one or few particles in any interpretation of QM, ..., hence the wave function can be reversed even for macroscopic objects.

It certainly was not obvious to me. Overall, I found PeterDonis' arguments slightly more convincing than yours, but was still unsure. But then came

That's a good question. It would be so if the reversal of the wave function was ideal, meaning that it is evolved with the Hamiltonian $-H$. But no reversal of the wave function is ideal in that sense, it's impossible to have the Hamiltonian $-H$ (for instance, the kinetic part of the Hamiltonian is always positive). Even for very simple systems, say one coherent photon, the reversal does not work that way. You can, for instance, reflect the photon on the mirror, which is a kind on reversal, but it is not the ideal reversion in that sense.

This starts to get ridiculous: Even if you could "exactly" prescribe $H(t)$, it would still remain completely open (and unclear) what you could to do to reverse the wave function.

At this point, I suggest that you either find a really convincing argument, or else concede the point to PeterDonis (for the moment).

 

[Demystifier]

At this point, I suggest that you either find a really convincing argument, or else concede the point to PeterDonis (for the moment).

I choose the former, I describe a system where the quantum state is reversed in practice, not just in theory: the spin precession of electron in a magnetic field. For the magnetic field in the z-direction, the interaction Hamiltonian is
$$H_{\rm int}=-\gamma B S_z$$
where $\gamma$ is the gyromagnetic ratio and $S_z$ is the spin operator. The spin precession is described by the unitary operator
$$U(t)=e^{-iH_{\rm int}t}$$
(in units $\hbar=1$). Clearly, this evolution can be reversed by reversing the sign of $B$. However, the Hamiltonian above is not the full Hamiltonian (there is also the free kinetic part of the Hamiltonian), so it does not completely reverse the wave function of the electron. It only reverses the spin precession.

I hope the rest of what I said is now clearer too. If not, please ask what specifically isn't clear.

 

[gentzen]

I hope the rest of what I said is now clearer too. If not, please ask what specifically isn't clear.

It is unclear to me why an example with a single electron should convince me that decoherence can be reversed in Bohmian mechanics.

 

[Demystifier]

It is unclear to me why an example with a single electron should convince me that decoherence can be reversed in Bohmian mechanics.

Because decoherence refers to the wave function (not to Bohmian particle positions and velocities), which in Bohmian mechanics is the same as in other interpretations. Hence, decoherence can be reversed in Bohmian mechanics if and only if it can be reversed in standard QM.

 

[gentzen]

Because decoherence refers to the wave function (not to Bohmian particle positions and velocities), which in Bohmian mechanics is the same as in other interpretations. Hence, decoherence can be reversed in Bohmian mechanics if and only if it can be reversed in standard QM.

Let me remind you that this is a discussion about AoE and it relation to decoherence:

Yes, but that assumption requires that decoherence cannot be reversed, because decoherence is (part of) what makes the observation an "objective fact" that all observers must agree on.

If decoherence can be reversed, then it is impossible for AoE to be true.

Since AoE is true FAPP, decoherence cannot be reversed in standard QM, FAPP. And depending on how you expand AOE, it may simply be true, not just true FAPP:

The specific LF assumption that the theorists hone in on is what they call Absoluteness of Events (AoE). This is the idea that once something has been measured, it becomes an absolute fact about the state of the objective universe.

That's not what "Absoluteness of Observed Events" (AOE) means.

In addition, you seem to claim that AoE would remain true in Bohmian mechanics, even if decoherence could be reversed. Here I am unsure. This seem to depend on how you interpret the de Broglie-Bohm theory. Or maybe AoE is true in that theory, indepedent of how you interpret it. I am currently not convinced, but open for good arguments.

 

[Sambuco]

It certainly was not obvious to me. Overall, I found PeterDonis' arguments slightly more convincing than yours, but was still unsure.

I'm not entirely sure what the objection is. If it is about the idea that decoherence can be reversed (at least in principle), that is something that is assumed in all the papers we are discussing about Wigner's friend thought experiments. The core idea behind these kind of thought experiments is that Wigner has complete control of his friend's lab, including the system and the friend, in the same way that the experimenter has complete control over what happens to an electron (to its wave function, of course) during a Stern-Gerlach experiment.

Lucas.

 

[gentzen]

I'm not entirely sure what the objection is. If it is about the idea that decoherence can be reversed (at least in principle), that is something that is assumed in all the papers we are discussing about Wigner's friend thought experiments.

Neither Schrödinger, nor Wigner made that assumption. For both, sufficient personal testimony is available to known which points they wanted to make with their thought experiments.

Some later papers seem to make such assumptions. But neither do those papers succeed in justifying those assumptions, nor are their conclusions generally accepted.

I'm on record since a long time that you must justify the possibility of such a reversal (and that macroscopic superpositions might not be unplausible in some situations where such a reversal is possible):
20 Mar 2017 TK: historical defence of Copenhagen

But I disagree that the quoted passage is technical. If he adheres to this passage, then Heisenberg cannot claim that Schrödinger’s cat would be both alive and dead, or that the moon would not be there if nobody watches.

Others, like Christopher A. Fuchs and Asher Peres in "Quantum Theory Needs No Interpretation", are apparently less sure whether (neo-Copenhagen) quantum theory is so clear about that fact. Hence they try to weasel out by claiming: “If Erwin has performed no observation, then there is no reason he cannot reverse Cathy’s digestion and memories. Of course, for that he would need complete control of all the microscopic degrees of freedom of Cathy and her laboratory, but that is a practical problem, not a fundamental one.”

This is non-sense, because the description of the experiment given previously was complete enough to rule out any possibility for Erwin to reverse the situation. Note the relevance of “… a consistent interpretation of QM as applied to what we do in a physical laboratory and how practitioners experience QM in that context.” If Erwin had access to a time machine enabling him to realistically reverse the situation, then it might turn out that Cathy and Erwin indeed lived multiple times through both situations (and experienced real macroscopic superpositions), as depicted in movies like “Back to the Future”.

 

[Sambuco]

Since AoE is true FAPP, decoherence cannot be reversed in standard QM

When you say this, you are referring to the following post

If decoherence can be reversed, then it is impossible for AoE to be true.

As I explained in post #17 and #18, that's not true, because the term AOE does not have the meaning Peter was giving it.

In addition, you seem to claim that AoE would remain true in Bohmian mechanics, even if decoherence could be reversed. Here I am unsure. This seem to depend on how you interpret the de Broglie-Bohm theory. Or maybe AoE is true in that theory, indepedent of how you interpret it. I am currently not convinced, but open for good arguments.

The fact that AOE is true in Bohmian mechanics is mentioned in the original paper by Cavalcanti's group (https://www.nature.com/articles/s41567-020-0990-x). They explicitly say: "Bohmian mechanics violates L but not the other assumptions." The other assumptions are AOE and no-superdeterminism.

We have to take into account that this kind of Extended Wigner's Friend Scenario (EWFS) assumes that quantum mechanics is "universal," meaning that any observer can model their system under observation as a quantum mechanical system, regardless of its composition. This is, let's say, the zeroth assumption.

Lucas.

 

[Demystifier]

Let me remind you that this is a discussion about AoE and it relation to decoherence:

Since AoE is true FAPP, decoherence cannot be reversed in standard QM, FAPP. And depending on how you expand AOE, it may simply be true, not just true FAPP:



In addition, you seem to claim that AoE would remain true in Bohmian mechanics, even if decoherence could be reversed. Here I am unsure. This seem to depend on how you interpret the de Broglie-Bohm theory. Or maybe AoE is true in that theory, indepedent of how you interpret it. I am currently not convinced, but open for good arguments.

I suspect we have different understandings of what AoE means. Can you explain it on some examples unrelated to quantum physics?

 

[gentzen]

I suspect we have different understandings of what AoE means. Can you explain it on some examples unrelated to quantum physics?

Let me repeat my old example cited in my reply to Sambuco:

If Erwin had access to a time machine enabling him to realistically reverse the situation, then it might turn out that Cathy and Erwin indeed lived multiple times through both situations (and experienced real macroscopic superpositions), as depicted in movies like “Back to the Future”.

If you have a time machine, then "observed events" are not necessarily "absolute". They may have happened for one person, but nevertheless not have happened for another person, even if those two persons should happen to talk with each other.

 

[PeterDonis]

In other interpretations of QM the results of all experiments are random, not determined by anything.

No, that's not correct. It's correct for collapse interpretations, but not for, e.g., the MWI.

By not controlling and not reversing Bohmian hidden variables, you effectively make the results of all experiments random

And since in actual fact, we can't control the Bohmian hidden variables, the actual results of quantum experiments are random in Bohmian mechanics. The only difference is that in Bohmian mechanics, unlike collapse interpretations, the randomness has the standard ignorance interpretation: we don't know the exact initial conditions. The underlying dynamics is fully deterministic. But since we can never know the exact initial conditions, the fact that the underlying dynamics is fully deterministic doesn't help us at all if we want to do things like run a Wigner's friend experiment with a human friend. We still can't control the Bohmian hidden variables.

 

[PeterDonis]

Clearly, this evolution can be reversed by reversing the sign of $B$.

Not just that, no. Pick a time when the electron has just emerged from your apparatus, and reverse the sign of $B$. Does the electron go back through, reversing its previous path? Of course not.

What will be the case is that if you reverse the sign of $B$ for the next electron, it will go the opposite way. But to call that "reversing the evolution" seems to me to be a gross misuse of language.

 

[Sambuco]

If Erwin had access to a time machine enabling him to realistically reverse the situation, then it might turn out that Cathy and Erwin indeed lived multiple times through both situations (and experienced real macroscopic superpositions), as depicted in movies like “Back to the Future”.

If I understood the example correctly, Erwin plays the role of Wigner, right? In that case, Wigner is on the classical side of the Heisenberg cut, so any reversal of decoherence does not affect him (at least in principle).

(and that macroscopic superpositions might not be unplausible in some situations where such a reversal is possible

That's a good point! I'm not aware of any publications on this issue.

If you have a time machine, then "observed events" are not necessarily "absolute". They may have happened for one person, but nevertheless not have happened for another person, even if those two persons should happen to talk with each other.

But that's not the AOE assumption in local friendliness theorem. AOE means that a probability can be assigned to the ocurrence of an experiment by Wigner's friend (see eq. (3) in the Methods section in Cavalcanti's paper). In other words, if AOE is false, it means that Wigner cannot assume that an event occurred inside his friend's lab.

Lucas.

 

[Demystifier]

Not just that, no. Pick a time when the electron has just emerged from your apparatus, and reverse the sign of $B$. Does the electron go back through, reversing its previous path? Of course not.

What do you mean by electron path? It's not even defined, except in the Bohmian interpretation. Since the thought experiment is defined in terms which do not depend on the choice of interpretation, the reversing can only refer to the wave function, not to the electron path. The Bohmian interpretation comes only afterwards; if the wave function is manipulated in such and such way, the Bohmian interpretation can say what happens with the electron path. And you are right, the path will not be reversed even if the wave function will. But that doesn't mean that the experiment cannot be performed. It only means that the experiment is devised so that only the wave function is reversed, nothing more and nothing less.

 

[PeterDonis]

What do you mean by electron path?

I didn't use the word "path", nor does my argument depend on the electron having a particular path. My point was simply that there is a time when the electron has left the region of space where the magnetic field is present--and reversing the sign of $B$ does not magically cause the electron to go back through the field after it has already emerged. So reversing the sign of $B$, by itself, does not reverse the evolution. Something else also has to be reversed. What that something is, will depend on which interpretation of QM you adopt; but there will always have to be something.

the reversing can only refer to the wave function, not to the electron path

No, you didn't say the wave function was reversed. You said the sign of $B$ was reversed. My point is that that alone is not enough to reverse the evolution. Indeed, "reverse the evolution" implies that you are somehow reversing the entire experiment--starting with the end state of the original experiment, and arranging things so the system returns to the initial state of the original experiment. Reversing the sign of $B$ in itself certainly doesn't do that.