# What's the problem with (non-causal) nonlocality?

Tags:
1. Jun 5, 2015

### greypilgrim

Hi,

In all the discussions about EPR, Bell's inequality and interpretations of QM locality seems to be a property that nobody likes to drop light-heartedly. This is somehow understandable since SR is an extremely successful theory.

But SR only says that we cannot transmit information faster than the the speed of light. The no-communication theorem states that we cannot communicate by performing local operations on an entangeld state, so this kind of nonlocality is strictly non-causal.

So why is nonlocality still something suspicious even though it's completely consistent with SR?

Last edited: Jun 5, 2015
2. Jun 5, 2015

### DrChinese

It's a matter of preference as best I can tell. There are a fair number of adherents to Bohmian Mechanics, for example.

3. Jun 5, 2015

### Staff: Mentor

Being consistent with SR is a necessary but not a sufficient condition for everyone to be happy with an idea.

We have the experimental results that show that non-local correlations exist, and it's a great relief to know that I'm not about to have an "everything I thought I knew about physics is wrong because SR has been falsified" experience. But I cannot even begin imagine a mechanism that would produce such correlations - there is simply no other other example anywhere in human experience of consistent experimentally repeatable (as opposed to random coincidence, like hemlines and the stock market) correlation without causality. Of course I don't need to imagine such a mechanism to use QM - that's what the "shut up and calculate" interpretation is about, and when nature tells me to shut up and calculate I will. But I don't have to like it.

So to some extent this a question of personal taste and preference... de gustibus non est disputandum.

4. Jun 5, 2015

### Demystifier

No, that's not the only thing SRT says. SRT does not only say that we (the macroscopic humans) cannot transmit information faster than light, but also that nature itself (at microscopic level) cannot do that. On the other hand, non-local interpretation of QM requires that nature can send information faster than light at the microscopic level, even if we, the macroscopic observers, cannot do that.

5. Jun 5, 2015

### greypilgrim

Bell says that QM cannot be both local and realistic. Which experiment can specifically only be explained by assuming non-locality, and rules out a local, but non-realistic interpretation?

Yes, but afaik it still only addresses information that can cause an effect.

Does this information cause an effect if it's not measured?

6. Jun 5, 2015

### Demystifier

Yes, by assumption it causes non-local EPR correlations, and these are measured.

7. Jun 5, 2015

### greypilgrim

Yes, but I'm asking about an effect if they are NOT measured. As soon as we measure, we (the macroscopic humans) are in the picture again and it's not only nature at a microscopic level.

8. Jun 5, 2015

### Staff: Mentor

I very carefully used the phrase "non-local correlations", as that is the experimental fact: the results of spacelike-separated measurements are correlated. This fact can be explained by rejecting locality or by rejecting realism.

9. Jun 5, 2015

### Demystifier

There is such an effect, but its exact nature depends on the exact theory of such non-local dynamics you assume. Of course, if you do not measure it, then you cannot see it.

10. Jun 5, 2015

### greypilgrim

Ok. But what are you trying to tell about QM using a phrasing this careful? Bertlmann's socks are non-locally correlated as well, but purely classically.

11. Jun 5, 2015

### Staff: Mentor

There's no conflict with relativity in the experimental results from either quantum entanglement or from observations of Bertlmann's socks. In both cases I have two sheets of paper, each containing one observer's results. I place the two sheets side by side and I see perfect correlation. At this point, I could stop worrying, be happy, and just shut up and calculate....

...Or I could ask myself what mechanism lies behind the perfect correlation. That's easy with the socks - even though the observations of Bertlemann's left foot and right foot were spacelike separated, the event "Bertlmann put on one pink sock and one blue sock this morning" is in the past light cone of both observation events so there's an obvious and natural mechanism behind the correlation. But Bell's theorem shows that there is no equivalent mechanism for the quantum entanglement correlation.

As DrChinese and I said in the first two replies in this thread, whether one finds this situation satisfactory or not is a matter of taste.

12. Jun 6, 2015

### Ilja

It is not an experiment, it is common sense.

On the one hand, if you accept hidden causal effects, you have to go back from the spacetime interpretation to the Lorentz interpretation. Not a big deal, only a change in the interpretation.

On the other hand, you would have to reject as realism, as causality, but gain nothing. Because what remains from causality if you give up realism is the absense of correlations - but this absence of correlations you don't have to give up anyway, even if you accept hidden causal influences.

Thus, realism + causality + realistic interpretations of QM + some problems of GR quantization disappearing (the "problem of time" - the incompatibility of the GR concepts of time with QM time - would disappear if you introduce a hidden preferred time, which could play the role of quantum time) against nothing at all.

13. Jun 6, 2015

### Ilja

No. Think about the meaning of realism. In fact, realism (in the extremely weak sense of realism used in Bell's theorem) reduces to the idea of the existence of realistic explanations for observed correlations. One can say that "realism" specifies only what is the meaning of a "realistic explanation", and, then postulates that a realistic explanation always has to exist.

If you follow this definition of realism, then giving up realism means giving up the idea that explanations exist, and leave the Bell correlations unexplained.

14. Jun 6, 2015

### stevendaryl

Staff Emeritus
According to some authors, realism is not actually an additional assumption of Bell's theorem (as applied to EPR), but is a derivable consequence of EPR's perfect correlations. In E, P, and R wrote in their 1935 paper: (as found on Dr. Chinese' web site) http://www.drchinese.com/David/EPR.pdf

I think that the way I would put it is this:

If without disturbing a system S we can predict with certainty the outcome q of a measurement on S, then there is an objective property of S corresponding to q.​

I'm not sure whether this should be considered an assumption of "realism", or not. It's extremely weak; it doesn't even assume that systems necessarily have objective properties, in general. It only assumes that IF something about a system is predictable with absolute certainty, then that something corresponds to an objective, pre-existing property of the system. In Laudisa's paper here: http://arxiv.org/ftp/arxiv/papers/0811/0811.2862.pdf, this principle is called "Property Definiteness", and he says about it:

So, as Laudisa says, it might be a mistake to think of Bell's proof as suggesting that either locality or realism must be given up in the face of QM. The only realism that is assumed is Property Definiteness. So if someone hopes to escape from Bell's proof by rejecting realism, rather than locality, he needs to think long and hard about what it means to reject Property Definiteness.

So in a twin-photon EPR experiment, Alice observes that her photon passes a polarizing filter oriented at angle $A$. Assuming that Bob has not yet measured his photon's polarization, Alice can predict with certainty that Bob's photon will also pass a polarizing filter at angle $A$. The assumption of Property Definiteness would associate this with Bob's photon. What is the alternative to property definiteness? The conclusion, that Bob's photon will pass through a filter at angle $A$, seems like an objective fact about the universe. It seems like to me, the only leeway that we have is whether to associate this fact with Bob's photon, or alternatively, it's just a fact about the universe as a whole, not particularly about Bob's photon. But the second alternative seems like a rejection of locality, not a rejection of realism: If we assume that the fact that Bob's photon will pass his filter (set at angle A, by assumption) is a fact about the universe at large, then it means that the result of Bob's local experiment depends on nonlocal facts, which seems like a rejection of locality.

So, to me, Bell's theorem, plus its violation by QM, leads to a different set of alternatives than "nonlocal or nonrealistic. I think there are two ways of escape the "nonlocal" conclusion: (1) Reject (contrary to experience) the assumption that a measurement produces a single outcome (this is the MWI approach), or (2) reject the assumption that Alice and Bob could potentially choose an arbitrary angle for their filters. To explain this: If Alice tested her photon at angle A, she concludes that there is an "element of reality" associated with Bob's photon that determines its potential to pass through a filter at angle A. At this point, Alice could reason: Since I could have measured my photon at any other angle, A', then it must be that there are elements of reality associated with every possible angle. But maybe Alice is wrong about this. Maybe she was somehow predestined to set her filter at angle A, and so there wasn't a possibility of her choosing a different angle. So the EPR argument would not imply that there were elements of reality at EVERY angle, only at the ones that are actually measured. The Bell's theorem would not go through.

So, I think that the alternatives are not "nonlocal or nonrealistic". I think that the alternatives are:
1. Nonlocal
2. Indefinite outcomes (more than one outcome, a la MWI)
3. Superdeterminism (detectors are not freely choosable)

15. Jun 7, 2015

### Ilja

I fully agree that it is extremely weak. I also tend to shift the argument toward excluding realism and focussing on causality, and Reichenbach's principle of common cause. Essentially, this is the same - if we observe a 100% correlation, then or one is the cause of the other, (this is what Einstein causality then excludes) or there has to be a common cause (this is the element of reality).

If one relies on common cause, one avoids the danger of discussing the various philosophical theories of realism which have nothing to do with the point. Of course, behind this is the same weak form of realism - a causal influence has to exist in reality to be a causal influence and not simply a correlation, one cannot name something a "causal influence" without having an idea that there is some real influence. But a focus on causality avoids the discussion of various philosophical realisms, and also allows to make two important points:

1.) Rejecting causality does not save Einstein causality - and what remains from Einstein causality without causality (the impossibility to send signals) is not questioned anyway.
2.) Giving up causality, giving up the search for common causes of observed correlations, endangers science in general. If observed correlations could be left without causal explanations, and commented with a "so what", what remains of science?