# Mermin on Spooky action at a distance

1. Oct 14, 2008

### Niles

Mermin on "Spooky action at a distance"

Hi all.

I've read Mermin's article on EPR and non-locality: www.physics.iitm.ac.in/~arvind/ph350/mermin.pdf[/URL]

I don't understand what he write on page 12: "Did the particle at A have its 3-color prior to the measurement of the 3-color of the particle at B? The answer cannot be yes, because, prior to the measurement of the 3-color at B, it is altogether possible that the roll of the dice at B or the whim of the B-operator will result in the 2-color or the 1-color being
measured at B instead. Barring the most paranoid of conspiracy theories, “prior to the measurement of the 3-color at B” is indistinguishable from “prior to the measurement of the 2- (or 1-) color at B”. If the 3-color already existed, so also must the 2- and 1-colors have existed. But instruction sets (which consist of a specification of the 1-, 2-, and 3-colors) do not exist."

If our detector is set to position 3, it measures if our particle has that property "3". We know that the particles do not carry instruction sets, but surely the particles know what state they are in. So if e.g. particles with property "2" are sent out and the detectors are both at position 3, then we get red at both detectors.

But where does his explanation come into this?

Last edited by a moderator: Apr 23, 2017
2. Oct 14, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

There are a couple of things going on. Clearly, if we ask the same question of both particles, we get the same answer.

The problem is that we might ask different questions, and the results are clearly affected - in some way - by the specific 2 different questions we ask. In the example, we would expect that we would get the same answer at *least* 1 in 3 times. But in practice, it is only 1 in 4. The implication is that particles take on properties in the context of how they are observed.

You mention that particles do not carry instruction sets. Yet you also mention that they must know what state they are in. And yet, that state is not determined prior to a measurement. So the idea that "particles with property "2" are sent out and the detectors are both at position 3" does not fit with the experimental facts. Unless, of course, they are in non-local communication or causal contact.

3. Oct 15, 2008

### ueit

Re: Mermin on "Spooky action at a distance"

There is nothing "paranoid" about the assumption that the "color" of the particle is related to the way this color is measured. For example, it might be that the detector itself produces a long range field that influence the colors of the entangled particles as they are "produced" at the source. If the detector is set to 3, you have a field; for a 2 setting you have a different field. Different fields, different colors.

Just ask the same question for a classical theory, like gravity. Say an object approaches the solar system from far away. Does the trajectory of that object (even when still far away) depend on the planets' configuration or not?

4. Oct 15, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

Of course, there are severe limits on this "field" explanation: it would need to be non-local, for one. And there appears to be no other evidence for such a field other than to explain entanglement.

Now, please note that Quantum Mechanics already fully accounts for the observed phenomena.

5. Oct 15, 2008

### ueit

Re: Mermin on "Spooky action at a distance"

No, that's not true. You make a confusion between locality (which requires that no influence can propagate faster than light) and independent evolution of two distant systems , like the source of entangled particles and the detector (which requires that no influence exists whatsoever). Two systems that interact through a local field are still not independent of each other. As an example take classical electrodynamics or general relativity.

Such a field would explain not only entanglement but also all other non-intuitive aspects of QM. What evidence other than this would you expect to find for such a field?

What do you mean by "Quantum Mechanics" and "fully accounts"? Do you have a speciffic interpretation in mind? Do you agree that the spot produced by a single particle in a double-slit experiment is an "observed phenomena"?

6. Oct 15, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

You know perfectly well that the settings of the detectors can be changed mid-flight without affecting the results in any way. If such a hypothetical field existed, it would need to transmit "something" to the other particle, and vice versa, so that they could yield "answers" that are consistent with the predictions of QM. So yes, there are severe restrictions and one of those is that any such mechanism must contain a non-local component.

7. Oct 15, 2008

### RandallB

Re: Mermin on "Spooky action at a distance"

Here I can only assume you are using the interdependence of Super Determinism and that is well understood to be non-local.
The fields of ‘classical electrodynamics’ must be assumed to be continuous meaning the effect caused by a field at different points is instantaneous which is just as non-local as Newtonian instantaneous gravity.

General Relativity has the same non-local problem with continuous gravitational fields – the main difference in GR is we usually do not call that problem a local vs. non-local issue; rather a “dependent vs. independent background” issue (see Smolin; Perimeter Institute). Here again there is nothing to refute the conclusion that GR requires an “independent background” which for the discussion here means non-local; as in unable to resolve things like entanglement or other non-intuitive aspects of QM.

But I do disagree with this statement;
Niels Bohr nor QM claim a “Full Accounting” of observed phenomena.
QM starts from an assumption of HUP that specifically rejects considering or describing individual particle behaviors during the measurement process.
The claim for QM is that no realistic description of behaviors during the measurement process can produce such a more complete full accounting than that given by HUP/QM.

8. Oct 16, 2008

### ueit

Re: Mermin on "Spooky action at a distance"

If by "results" you mean statistical results then I agree. We don't know if individual results are changed because QM doesn't say much about them.

Both detectors produce a field. Their combined field has some value at the location of the source. The particle spin, on each axis, is a function of the local value of the field in that place. No non-locality required.

If you chose to change the detectors' settings then you have to provide some device that is capable of doing that (say an electric engine). That device itself produces a field at the particle source location so you will get different particle spins than in the case of a static detector. In order to find out what the spins are you need to calculate the field for each experimental setup.

There are restrictions, indeed, but non-locality is not one of them. The only way to impose such a restriction is to make the assumption that QM is fundamentally indeterministic. In this case you can argue that the same field could corresponds to different detector settings so a local field is excluded. However I see no need to make such an assumption. I choose determinism+ locality instead non-determinism+ non-locality combo.

Last edited: Oct 16, 2008
9. Oct 16, 2008

### ueit

Re: Mermin on "Spooky action at a distance"

I don't think that there is a "interdependence of Super Determinism" as opposed to other types of interdependence. I'm trying to avoid the explicit use of the word "Super Determinism" in order not to deviate the discussion on pure philosophical notions like free will. So, if you want to use the term, I'm OK but let's discuss the assumption of a long-range local field that I've proposed. By "local field" I mean that a perturbation in that field propagates at a finite speed (c) and only the value of the field at the point where the particle is located exerts an effect upon that particle.

Why should a continuous field be necessarily non-local?

Again, why is a continuous field be necessarily non-local? AFAIK GR also has a local character. A massive body "feels" the effect of the space curvature at its location and a perturbation of the curvature travels at a finite speed, c.

10. Oct 16, 2008

### Niles

Re: Mermin on "Spooky action at a distance"

First I want to thank you all for participating.

My question is a lot more fundemental. This is how I have understood Mermin's article so far:

Two particles in a singlet spin-state are fired at the two detectors, which can measure the direction of spin at angles of 0, 120 and 240 degrees. Mermin then accounts for how why half the time the same color flashes, and this I understand.

Now look at this:
This is what I cannot understand. Let's look at the explanation Mermin gives with his "N-colors":

"Did the particle at A have its 3-color prior to the measurement of the 3-color of the particle at B? The answer cannot be yes, because, prior to the measurement of the 3-color at B, it is altogether possible that the roll of the dice at B or the whim of the B-operator will result in the 2-color or the 1-color being measured at B instead."

If we use spin-½ instead of N-colors, then spin in e.g. the z-direction is measured of particle B. But particle A cannot have a pre-determined spin in the z-direction, because if we chose to measure spin-½ in the x or y-direction at detector B, then ....

This where I am stuck. I cannot see how we can rule out the possibilty of particle A having a pre-determined spin-½ in a particular direction, because we are able to measure spin-½ of particle B in three different directions.

Thanks for enlightening me.

Last edited: Oct 16, 2008
11. Oct 16, 2008

### RandallB

Re: Mermin on "Spooky action at a distance"

The nature of the fields are defined by the theories that use fields in their descriptions. It is not like a “field” has ever been directly physically observed only; their indirect effects as described by the theories that define the fields.

In order for GR to be “local in character” it needs to demonstrate a “dependent background”. And most if not satisfied with the Smolin point that GR is fundamentally a independent background description, certainly have not been able to demonstrate a dependent background for it.
It is not something you can just assume without foundation.
No more than you can assume the principles of Super Determinism as a local description by not mentioning the word “determinism”.

Only if you can demonstrate how General Relativity and the fields defined by GR can describe reality within a completely “dependent background” could I accept the opinion that GR is “local in character”.

12. Oct 16, 2008

### Peter Morgan

Re: Mermin on "Spooky action at a distance"

Concerning Ueit's resort to fields, I refer you to my J. Phys. A paper "Bell inequalities for random fields", last pre-print at http://arxiv.org/abs/cond-mat/?0403692 (and thence a link to the published paper). It's not enough to use classical fields, probability also has to be introduced; mixing probability with classical fields results in a need for a mathematics called random fields that have quite intricate properties, which are rather similar to the mathematics of quantum fields. See also my just posted http://arXiv.org/abs/0810.2545 for an attempt at a broader discussion. The standard idea in Physics is that the papers of Bell's that I cite in "Bell inequalities for random fields" rule out classical fields; indeed the argument is not easy to gainsay definitively.

The question of non-locality is very delicate; for a random field model at equilibrium -- in the coarse-grained sense that the statistics of detector events are time-invariant even though in a fine-grained sense the detectors may be switching between different thermodynamic states thousands of times each second -- the global properties of the whole apparatus determine the detailed properties of the equilibrium, just as Bohr says they should. Hence, in a random field perspective it is more a question of holism at equilibrium -- which is reasonable in a classical Physics perspective -- than a question of nonlocality -- which is essentially not so reasonable.

It also should be noted that quantum fluctuations have to be introduced explicitly into random field models to give accounts for experiments in which quantum effects are seen. The import of the need for quantum theory is that the mean field approximation for quantum fluctuations is not always enough.

I note that Dr. Chinese says that "the settings of the detectors can be changed mid-flight without affecting the results in any way"; if there are no particles, however, only fields, there is no such thing as "mid-flight". The detector events are caused by the particular way that the "preparation device" drives the field, which has to be carefully tuned to obtain statistics that are incompatible with a particle property model. Provided the rapid random switching of the detectors is done in a way that does not change the essential right-left symmetries of the overall experiment, there is no reason to think that the statistics should change. If the statistics do change, then the random switching, a priori, does not preserve the symmetries of the apparatus.

I started writing this before Niles latest comment was posted. Seeing it's principal point, I note that in a field perspective, no event can possibly happen without a macroscopic detector being present. "The implication is that particles take on properties in the context of how they are observed", that is, only when a detector -- which is a delicately tuned macroscopic object that makes thermodynamic transitions repeatedly over time -- is placed in the right place. No detector, no events.

I hope this is useful to you, Niles, however this post is more comprehensible in terms of the discussion on fields between Ueit and Dr. Chinese. Thinking in terms of classical particles and their properties gets you into something of a mess, the thinking of the last 70 years pretty much shows (unless you want or are willing to go truly nonlocal with de Broglie-Bohm or similar models), so: go to fields. Sadly, that has to be random fields, not the more straightforward continuous fields. Good luck.

Last edited by a moderator: Apr 23, 2017
13. Oct 16, 2008

### Peter Morgan

Re: Mermin on "Spooky action at a distance"

Regarding superdeterminism, I point out that quantum field theory is superdeterministic in the sense that a quantum state that determines the probabilities of various measurements now also determines the probabilities of whatever measurements we might make in the past.

If we restrict to probabilistic classical models, in other words to random fields, the state of the random field is no more superdeterministic than a quantum field state. Only probabilities in the past are determined by models in a random field formalism, not individual events. Indeed, we can present a random field as a commutative quantum field.

Most Physicists take quantum field theory to be the real thing, while nonrelativistic finite-dimensional Hilbert spaces are just approximations; but perhaps quantum field theory is unacceptable because our knowledge of the quantum state now partially determines how we suppose the world was, probably, in the past?

14. Oct 16, 2008

### Niles

Re: Mermin on "Spooky action at a distance"

Although I really appreciate your posts, my question is a lot more fundemental. I simply do not understand (litteraly!) his explanations.

15. Oct 16, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

Yes, if you stick to angles that are perpendicular for spin 1/2, then the problem is not apparent. But that is not what makes Bell's Theorem work, and same for Mermin's version. I might recommend we stick to the photon (spin 1) version because it is easier to see. But it would also work for spin 1/2, as that is what Bell used.

Bell noted that at certain detector angles, you would be observing a mix of the degrees of freedom. At 3 angles spread across 2 degrees of freedom, he discovered that it would not be internally consistent as to a "simultaneous realistic" answer for all 3 settings. So that is the problem: you cannot have a logically consistent set of answers that match experiment.

So again, let's look at Mermin's 3 measurement positions. A measurement of one particle at position 1 guarantees that the OTHER particle's answer at position 2 or 3 will be "as if" it had already been measured at position 1. What's so weird about that? Well, that implies that there might be some kind of spooky action at a distance. And how does that manifest itself? It shows up as a statistical relationship (.25) that is lower than the lower bound should "logically" be (.33).

16. Oct 16, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

Not sure if that will make things any easier, but I have a web page that is devoted to explaning Bell using Mermin's ideas:

Bell's Theorem with Easy Math

I hope that helps too. Keep your questions coming...

17. Oct 16, 2008

### JesseM

Re: Mermin on "Spooky action at a distance"

If it helps, I came up with another analogy similar to Mermin's on this thread:
I also showed how this example could be applied to a different Bell inequality in post #8 of this thread if you're interested.

18. Oct 16, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

JesseM's example above is good too.

Returning to your question above, consider this... suppose you measure an electron at 0 degrees (call that x-axis), and follow it with another measurement at 30 degrees. The second measurement would be a mix of the x and y axes presumably. In other words, we had earlier established that there is an element of reality (by EPR's definition) associated with any possible angle setting because there are "perfect" correlations between entangled particles at any identical angle setting.

So if we are measuring a mix of axes, what does that mean? In your example, you picked the extremes where there is no connection between measurement results (axes are separated by 90 degrees, for spin 1/2 particles). For a spin 1 photon, that would be a separation of 45 degrees to get analogous results instead of 90 degrees. But the question you need to ask is: if you measure in between, a mix of 2 degrees of freedom (i.e. 2 axes), what happens then? That is what Bell asked. He realized that QM's predictions were internally inconsistent IF there were simultaneous reality to all possible observable angle settings. That led to Bell's Theorem. Mermin's presentation just makes the math easier to see. Or maybe not... LOL.

19. Oct 17, 2008

### Niles

Re: Mermin on "Spooky action at a distance"

JesseM's example is very good.

Mermin's explanation of the math is not too bad, I think. So now I fully agree that there is no way that pre-determined instruction sets can explain the results that we get. Also I fully understand Mermin's example, where he uses spin-½ to explain the results.

So is the conslusion: We can show that there are no instruction sets, because then the same color (I am using Mermin's example) woud flash 5/9 of all runs, whereas they only flash 1/2 of all runs. So no instruction sets.

Also, one way of building the device is to use spin-½. This configuration also accounts for the data. So since there are no instruction sets, the particles do not have pre-determined spin?

20. Oct 17, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

That is safe to say. Even in non-local theories, the results are contextual and therefore not predetermined. The results are somehow shaped by the specific observation settings chosen, which is done at a later time.

21. Oct 17, 2008

### Niles

Re: Mermin on "Spooky action at a distance"

Great, I understand the words now and the math. Thanks for that, everyone.

22. Oct 17, 2008

### Autochthon

Re: Mermin on "Spooky action at a distance"

This is exactly what I've been puzzling over lately (the perspective of the observation effects the result). I somehow miss the connection between Bell's Inequality and this conclusion. I'd always thought that Bell himself concluded that the inequality did no more than exclude locality. By contextual do you mean you are restricting the result to a single outcome rather than the family of outcomes encompassed by the possible perspectives? Why might not the full set of possible observations describe the object?

23. Oct 17, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

I think Bell envisioned a non-local theory as the likely outcome originally. But there has been a lot happen since then. GHZ is one example. The Weihs (and Aspect as well, which Bel knew about) experiment too. But there has been no really non-local model to provide any reasonable description which is non-contextual (i.e. realistic). They all talk about a mixed system in which the measuring apparatus influences the results. Well, that is contextual and not "realistic" in the EPR sense.

A realistic theory, on the other hand, will be able to provide answers for the question: what are the correlation probabilities for THREE simultaneous measurement positions. But we already know there are no such non-negative solutions. So I have no issue with a non-local theory, I just assert that a non-local realistic theory is not yet presented for out consideration that can meet this criterion.

The dBB-type solutions claim to be deterministic, but I am not sure that is really a fair characterization consistent with experiement. There must be a non-local signal (between Alice and Bob) in such a theory to account for the fact that the measurement settings can be changed mid-flight. I would not call that "pre-determined", quite the opposite.

24. Oct 21, 2008

### Ken G

Re: Mermin on "Spooky action at a distance"

Can you expound on this point? It sounds like you mean we have three entangled particles, and want the correlations among three measurements. Certainly quantum mechanics allows us to calculate those probabilistic correlations for a given set of observational settings. So what do you mean that there are no non-negative solutions?
I think you are raising an important issue about what "pre-determined" should mean. The most natural interpretation is what you are using, which sounds like, it is pre-determined if you can know in advance (in principle, even if our science cannot actually get these answers), and at the same time, what will be the answer to any set of questions you can put to the system, whether or not those questions are ever actually posed to the system. But there is a more flexible version of pre-determined which might just assert that we cannot know (even in principle) simultaneously the answer to all the possible questions, but we can know in advance (in principle) the answer to any one given set of questions. Knowing the answer (in advance) to those questions could preclude knowing the answer to certain other questions, but that can still be a version of "determinism", albeit a much more subtle one.

In other words, we must ask what is actually being "determined" in a "deterministic" reality. Does it have to be all things to all people? Or can we allow reality to only be able to determine the answer to questions that are actually posed. By which I mean, it is quite possible (and seems likely) that the posing of questions is not a purely hypothetical exercise, it is itself part of the reality we are trying to understand. We should not conclude reality cannot be deterministic simply because we have not yet determined what questions are going to be posed, that is not reality's fault. It can only determine what answers must appear to whatever questions are actually posed, as the latter is part of reality too, and reality cannot be self-consistently required to address unreal events that do not in fact occur.

25. Oct 21, 2008

### DrChinese

Re: Mermin on "Spooky action at a distance"

1. Einstein said the moon was there even if we are not observing it. EPR says that observable attributes that can be predicted with certainty without disturbing a system have an element of reality. I think these can be combined reasonably to conclude that a photon - per Einstein - has a definite polarization at all possible angles even if they cannot be known simultaneously. So let's just say (to be specific) that we want to assert that a photon has a definite polarization value at 3 angle settings. We know we can used an entangled photon pair to learn 2 of the 3 possible values. But Bell's Theorem shows us that there is an internal inconsistency in that argument (that there are 3), because the quantum mechanical predictions lead to predictions of negative probabilities at some angle settings.

Unfortunately, there is currently no 3 particle entanglement scenarios in which the so-called "perfect correlations" possible. There is 3 particle entanglement but different statistics apply.

So what I am saying is that any theory which claims to be "realistic" or "hidden variable" or similar should come out of the closet and tell us what the statistics are for one of these "forbidden" angle combinations: say A=0, B=120 and C=240 degrees for example. I would like someone to explain a scenario in which the correlation is 25% between any 2 of these 3 settings while at the same time the perfect correlations hold as well.

The internal (realistic) inconsistency is: IF the AB Correlation percentage is .25 (matching experiment and QM predictions), and the BC Correlation percentage is .25, THEN the AC Correlation percentage would be .56 (.75 * .75); YET all 3 should be the identical (.25) due to considerations of rotational invariance (and since they are all 120 degrees apart).

2. Some of the hypothetical solutions have asserted that the observer must be added into the system. I think the problem with the "pre-determined" scenario is that the results appear clearly contextual, yet the measurement apparati (essentially the observer or at least the observer's choice of settings) can be changed mid-flight. So this precludes predetermination, I would say.

In my opinion, this causes problems even for non-local theories. Because now we have the apparati (which now becomes the variable) in some kind of mutual communication. How is this possible, when such communication apparently cancels out to zero conveniently for everything else in the universe for this measurement?

In other words, I think it is pretty amazing that even trying to come up with an explicit mechanism which is either non-local or contextual (or both!), it isn't easy to construct a plausible scenario. The reason you have trouble with these mechanisms is that they seem to quickly run afoul of everything else we know about physics! You end up with new theory widgets which ONLY seem to apply to entangled particles.

I can picture another way out of things (this is merely speculation mind you!!!): relax the requirement that the past cannot be influenced by the future. If you drop that requirement, then it seems natural to me that a particle's history includes its future. There would be communication lines between entangled particles that way, and there would be the opportunity for all histories to potentially interfere with each other (as they appear to do). Not that this makes any more sense than any other explanation, but to me it is one more possibility to consider. So in this scenario, a measurement of Alice retro-influences Bob. This allows for the correct statistics, is fully contextual (and non-realistic) and is also local and time-symmetric. But you pay a strange price for it!!

The only other thing that seems to make any consistent sense is simply the mathematical formalism, which is what seems so empty - at some level anyway - to many.