Scholarpedia article on Bell's Theorem

ttn

I don't think you (correctly) understand what is going on in this passage you quote from the EPR paper. Now, in your defense, it is admittedly cryptic. Einstein thought and said so too, and we should take that seriously. In other words, we should not take the text of the EPR paper too seriously. When Einstein says it flubs and obscures the argument, we should listen to him, and hence rely on *other texts* to try to understand *his* views.

That said, here is what I think is the correct way to understand the passage you quoted. Actually, the just-following part (which you have also recently quoted) is highly relevant, so let me include the whole passage uninterrupted:

Now look carefully at the sentences after the one you had quoted. What are they about? They are about ... LOCALITY. He doesn't exactly say that, and it could surely have been made a bit clearer, but look at what he actually wrote. It amounts to this: "to say that whether P or Q of the distant particle exists depends on which of those quantities I *actually* measure over here, would mean that what's real over there depends on what I freely choose to do here." In other words: that would violate locality!

Thus, I think one should understand the infamous last sentence -- "No reasonable definition of reality could be expected to permit this" -- as a statement about locality. He is trying to express that *non-locality* should be considered unreasonable! But it would have been much clearer, and certainly better for the historical development on these issues, if he had just written: "locality requires this" instead of injecting simultaneously an assessment of how reasonable or unreasonable it is to consider rejecting locality. In any case, setting aside the question of whether or not one should believe locality (clearly E, P and R would all have believed it, but the point here is that that doesn't really matter) the idea being expressed is just what I keep saying: you'd have to deny locality (or the QM-predicted perfect correlations) to deny that (here) Q and P have pre-existing definite values.

DrChinese

Of course I agree. As you say (and I thought I said), 2 values are sufficient for EPR, and more are implied. But 2 was not sufficient for Bell though. He had to have at least 3, one of which is counterfactual. Hopefully, you don't question that.

Look, there is no need for you to spend time with me on this. I am not really arguing with you so much as laying out some of the counter-reasoning* to readers who are following this thread.

*Which is the standard view of most physicists.

billschnieder

I think there is some very unfortunate choice of words being used here is adding to the confusion. EPR never suggested that outcomes of measurements *exist* prior to measurement. Such a statement commits a modal fallacy and can be rejected outright. They said elements of reality corresponding to *definite* predictions *exist*, not that the outcomes themselves *exist* prior to the measurement. Therefore to say
is unfortunate. Maybe what was meant here is that the spin components have *definite* values (cf. EPR's "predict with certainty"). Which does not mean the same thing as they *exist*.

This is particularly important when you start comparing experimental outcomes, which ALL *exist*, with inequalities involving predictions which although all simultaneously valid, can not simultaneously *exist*. Failure to understand this distinction is at the root of many unnecessary paradoxes.

ttn

Well, at least one of the 2 was already counterfactual. So I still fail to see why you think anything important was added by going from 2 to 3. (It's also relevant that there are Bell type inequalities with only 2 settings on each side. Again: 2, 3, whatever. There is no issue here. Barking up the wrong tree.)


But that is exactly why I consider it worth my time to spend time with you on this. You do such a perfect job of expressing "the standard view of most physicists", and I think a lot of good might come from the audience seeing Bell's unorthodox views (or at least my best attempt to channel them) pitted up against these standard views in open discussion.

camboy

Hi ttn,

Sorry to butt in. I'm thoroughly enjoying this debate, and I've digested your article and enjoyed it. For me you win on points (and possibly even a technical knockout - sorry Dr. C.).
I agree - this debate is well worth having.

Now if you go further along this path the ultimate pit of hell into which you can descend is to argue with the very wonderful Lubos Motl. His views on nonlocality are here, and - not unexpectedly - they're pretty scathing and in direct contradiction to the conlusions of your article. Now, of course it's not worth arguing with him on his blog site, since he mostly just deletes comments that disagree with him. Nevertheless, is there anything substantive in his arguments that you can see?

ttn

Thanks, and thanks for saying so. It's nice to know that there are people watching and getting something out of this!

I have read some of Motl's comments on (e.g.) Bohm's theory before. Hadn't seen this particular post though. I skimmed it, only reading carefully the part where he purports to explain "misconceptions about nonlocality in QM". Basically everything he says is standard stale white bread orthodoxy: a rather dilute mixture of vague anti-realism, positivism, the (erroneous) identification of causal influences with signaling, and vitriol exuded toward the whole issue. It's of course telling that he doesn't even mention Bell. I'm sure (despite the in-passing reference to Bertlmann's Socks) that he's never actually read that paper, or any of Bell's other papers. Otherwise, don't you think he'd try to actually say what's wrong with *Bell's argument* -- instead of just talking tediously about what he feels is the right way to understand orthodox QM?

Note also that the essence of his attempt to argue that (you know, contrary to what the crackpot morons like me think) orthodox QM is perfectly local, is actually just a proof that the marginal distribution of outcomes on one system is independent of what might (or might not) be measured on an entangled system. So if I were going to discuss the issue with him, perhaps I would start by asking whether he thinks that the de Broglie - Bohm pilot wave theory is also local since, of course, it is also true in that theory that the marginal on one side is independent of what's measured on the other side. Presumably he'd so "no, obviously it's not, you &*#@ing moron" and then we could start discussing how to define "locality" so it captures the idea of "causal influences only coming from the past light cone" rather than merely this constraint on the marginals (which is roughly equivalent to a prohibition on signaling) .... and then a miracle occurs .... and then he becomes convinced that, yes, actually, if you formulate a precise notion of "locality" in *that* sense, then yes, not only Bohmian mechanics but also ordinary QM is nonlocal and indeed it can be proved that *any* theory sharing QM's empirical predictions will have to be nonlocal.

Somewhat more seriously, though, if someone were going to try to engage him on this issue, I'd say just recommend that he read "Bertlmann's Socks" or "La Nouvelle Cuisine" and challenge him to explain what's wrong with Bell's argument.

camboy

Hi ttn,

Thanks for the response. I esssentially agree with you..
To be fair, I just searched for the word 'Bell' in the article, and it comes up with:

"Entanglement isn't any sign of a nonlocality. Bell's inequalities guaranteed that the experimentally observed correlations can't be explained by a local realist theory. But in a striking contrast with the popular scientific literature, the wrong assumption isn't locality; it's realism. Locality is just a property of relativistic and similar theories, whether they're quantum or classical. And indeed, it holds. The validity of locality was one of the key results of Einstein's special relativistic revolution of 1905, a revolution that can't be undone anymore."

"On the contrary, realism is an assumption behind all classical theories, whether they're relativistic or not. And it's been shown invalid in the 1920s because classical physics has been shown wrong. Only probabilities of actual measurements may be predicted by physics. This is what the quantum revolution of the mid 1920s is all about. The new picture of the world is "local, non-realist". Everyone who suggests that it's "nonlocal, realist" apparently misunderstands both major revolutions of the 20th century physics, quantum mechanics and relativity."

Hardly an in-depth discussion, but still..
I'm sure he will be utterly delighted if one of us were to suggest that, or to point out what is wrong with his arguments. Perhaps he might even buy you a present to show his gratitude? I love it when people buy me presents.

ttn

Oh yeah, I stopped reading at the end of the "misconceptions..." section and didn't even notice that the next section was about QFT rather than one of the other weirdo thing he talks about.

So, Dr C will be pleased -- here's a real live regular physicist who thinks we get to choose whether to reject "locality" or "realism". But, IMHO, the opinion of this particular regular physicist is completely worthless since he has never actually looked into these issues but is instead just repeating what he read in textbooks written by others who had never actually looked into it...


I seem to recall that several years ago another Bohm-fan (and PhysicsForums participant) posted some comments on one of Motl's blog posts, and his (Motl's) responses made me think it wasn't worth trying to discuss any of this stuff with him. I can deal with ignorance and I can deal with hostility, but the combination tends to be unfruitful and unpleasant to argue with.

zonde

Are you sure that EPR was arguing for particles possessing some definite properties?
It seems possible to me that EPR was attacking Heisenberg's "uncertainty as measurement disturbance" view. So in that case the goal of EPR argument would be to show incompatibility between QM formalism and hidden variables which it does.

If we would think that QM formalism is compatible with hidden variables then argument with one value would give nothing.

harrylin

No I would not say that it is a measurable attribute of the orange: the exact way this measurement is done influences the outcome, and this may be not exactly reproducible so that it is a property of both the orange and the detection instrument.

harrylin

I think that DrC is here simply acknowledging an objection that has been raised in the literature and which is rather incompatible with your reply: A single EPRB experimental data point corresponds to a measured property at the two locations, along the 2 directions a and b. Not 1, and certainly not 3. Your reply seems to deny that fact; and why do you call such a measurement fact "counterfactual"?

ttn

You are confusing two very different things:

1. A Bell-experiment in which a pair of particles is sent out, and then the polarization along one direction is measured on each of the two particles. So, there's an "a" and a "b" involved, sure, but they refer to the polarization measurements on two separate particles.

2. The EPR argument, in which a hypothetical measurements along any axis on one particle, establishes (according to the EPR argument, i.e., assuming locality) that the corresponding property (i.e., the polarization along that same axis) is an "element of reality" for the distant particle. Here there's really only one angle in the mix: measure particle 1 along "a" and infer that particle 2 has a definite value along "a". But, the argument goes, you could measure along "b" instead, in which case you'd establish that particle 2 has a definite value along "b". Or if you prefer, "c". My main point before was that none of these measurements need be actually performed. It's really a thought experiment. So even if you only run the argument for "a" there is still a "counter-factual" element. Certainly if you run the argument for both "a" and "b" at least one is "counter-factual" in the sense that, if you make any actual measurements on particle 1 at all, you can only actually measure along "a" or "b" but not both.

But your comments suggest that you don't appreciate yet that the EPR argument is something distinct from just running the Bell experiment.

DrChinese

No offense taken. If ttn has converted you, then I too think the debate is worthwhile. He makes a lot of good points (wrong of course LOL) and deserves a voice. I always enjoy his forays here.


Now, you say this, but I don't think you really mean it. I can show you any number of quotes of Aspect, Zeilinger, et al (those that work in the field every day) that completely contradict everything you are saying. For example, from one of THE watershed experimental papers:

http://arxiv.org/abs/quant-ph/9810080
Gregor Weihs, Thomas Jennewein, Christoph Simon, Harald Weinfurter, Anton Zeilinger, 1998:

"After Bell’s discovery that EPR’s implication to explain the correlations using hidden parameters would contradict the predictions of quantum physics, a number of experimental tests have been performed [3–5]. All recent experiments confirm the predictions of quantum mechanics. Yet, from a strictly logical point of view, they don’t succeed in ruling out a local realistic explanation completely, because of two essential loopholes."

I could post these all day long, and you know it. Note the choice of words: "hidden parameters" (not local hidden parameters) and "local realistic" (not local). As I say, most folks in the field soundly reject realism as defined by EPR and Bell. None of these gents have changed their published views since (that I have seen - perhaps you will correct me on this point, because I suspect they all know your work).

My point to everyone again being: if you change definitions, as EPR says, you change your conclusions. If I adopted ttn's definitions, I too would arrive at his conclusion. There is a good reason why the scientific field rejects ttn's analysis: like me, they reject the premise. As a result, there is really nothing for me to defend here. Each is free to choose their starting point.

harrylin

Oops - indeed, I mixed up the angles a and b and angle settings a, b and c! Giving the same symbols to different things should be avoided.
I highly appreciate that clarification. However, my comment still stands for both detectors measuring along the same angle.
So, when you measure both along the same direction, then surely your measurement outcomes for these two measurements are factual; not one of these results is counterfactual. However, for Bell's argument, an additional assumption must be made about a third measurement result. Thus I suspect that you and DrC mean the same thing but use different ways to say it! As a reminder, I commented on his sentence:
"As you say (and I thought I said), 2 values are sufficient for EPR, and more are implied. But 2 was not sufficient for Bell though. He had to have at least 3, one of which is counterfactual. Hopefully, you don't question that."
You may be very right about that: I consider the Bell experiment as an attempt to test a variant of the EPR argument...

ttn

I know what you mean, but there's no contradiction. I never said "everybody who works in/on foundations agrees with me". Rather I said that the opinions of those who *don't* work in/on (and indeed who despise) foundations don't really matter that much because they are just repeating what they were taught in kindergarten.

You are of course absolutely right that there are big famous serious people who have thought seriously about Bell and who hold orthodox views that conflict with my/Bell's views. Zeilinger is indeed a prominent example. Mermin another. But none of that proves I'm wrong. I just think these guys have missed the EPR argument (and Bell's post-1964 papers!) so they don't understand the implications of Bell's theorem correctly at all.

You for example quote from on of Z's papers:

"After Bell’s discovery that EPR’s implication to explain the correlations using hidden parameters would contradict the predictions of quantum physics..."

Well, that's just not a correct statement of what Bell discovered -- at least if you ask Bell.


That I do not agree with at all. It is *not* an issue of definitions. It's an issue of whether you ignore the EPR argument or not.

ttn

Yes, obviously nobody thinks that the outcome of a real, actually-performed measurement is somehow not factual. (Well, maybe MWI people think that...)



No, that is not true. Bell's derivation talks about what some candidate local theory will predict for various experiments. It is not literally, directly about actually-performed measurements (such that there's some assumption about the reality of some mysterious third counterfactual doodad). It's about what a local theory will predict. All of this business about counterfactuals is a total red herring.


It is completely wrong to think of the Bell experiments as an attempt to test the EPR argument. The EPR argument is ... AN ARGUMENT. It's not a theory, or a prediction, or any other kind of thing that it makes sense to "test". One doesn't test arguments, one *makes* them. One judges whether they are good arguments or bad arguments. I think the EPR argument is a good (I mean valid) argument. So did Bell. So did Einstein. But no experiment is ever going to be able to show one way or the other that the argument is or isn't valid. The most it can do is show that one of the premises or the conclusion is true or false.

DrChinese

We sorta agree here, although your interpretation of the conclusion is wrong. The conclusion is not: The ability to predict with certainty implies non-locality. This is not stated anywhere.

The correct conclusions of EPR are: QM is incomplete if realism (elements of reality independent of simultaneous observation) is assumed. If realism is not assumed, then reality here is dependent on the choice of an observer there. To other readers: It is a bit difficult to parse, but it is in these words of the final paragraphs of EPR - you can draw your own conclusions if you think I am misinterpreting (bold added):

"Starting then with the assumption that the
wave function does give a complete description
of the physical reality, we arrived at the con-
clusion that two physical quantities, with non-
commuting operators, can have simultaneous
reality. Thus the negation of (1) leads to the
negation of the only other alternative (2). We
are thus forced to conclude that the quantum-
mechanical description of the physical reality
given by wave functions is not complete.

"One could object to this conclusion on the
grounds that our criterion of reality is not suf-
ficiently restrictive. Indeed, one would not ar-
rive at our conclusion if one insisted that two
or more physical quantities can be regarded
as simultaneous elements of reality only when
they can be simultaneously measured or pre-
dicted. On this point of view, since either one
or the other, but not both simultaneously, of
the quantities P and Q can be predicted, they
are not simultaneously real. This makes the
reality of P and Q depend upon the process
of measurement carried out on the first system
in any way. No reasonable definition of reality
could be expected to permit this."

Historically, there has never been much question about the correctness of this argument - certainly not from the Copenhagen side of the fence. Their view was that the realism is an invalid assumption, and we live in an observer dependent world - so no problem-o! Both of these are consistent with EPR, as I presented above. The question to the local realistic side has always been whether there is a deeper level of reality. So far, it has not been found to exist in a local setting despite the explicit EPR belief that it would be found.