Relativity & Quantum Theory: Is Locality Violated?

[LnGrrrR]

So... still with me? Do we agree that, in principle, a theory can violate Bell Locality (and hence not be consistent with relativity) yet still not support the sending of signals faster than light?

This is a question I'm certainly interested to hear the answer to from you folks, as it is one I have read about that seems counter-intuitive at first glance.

 

[ttn]

I know that you want to talk to Dr C. and I don't want to interrupt your discussion. But I wanted to make something clear. We know that signal locality is a less severe requirement than Bell locality. However, we should not forget where this "locality" requirement comes from: it comes from the idea that spacetime has a geometrical Minkowski structure ; from this, it follows that the laws of nature should not depend on how we LABEL the events in Minkowski space - which results that ALL physically relevant expressions should be writable in Lorentz-invariant form.

Patrick, very good and interesting post. This probably deserves its own thread... but I'll answer briefly here, and then participate if anyone wants to tear this off and give it its own thread.

Basically, I disagree about the hierarchical relation between "locality" and spacetime's Minkowski structure. You suggest that the only reason we *care* about requiring theories to be local is to enforce this underlying spacetime structure. If we knew directly from God that spacetime *did* have this structure, then I think you'd be right -- including being right about "signal locality" being a useless/irrelevant idea once you reject Bell Locality (which is much closer to the requirement you're after, that the dynamical laws respect the minkowski structure).

However, I do *not* think it's the case that God whispers this in our ear. That spacetime has a certain structure is an *inference* from more concrete experimental data, including such things as the measured invariance of the speed of light (e.g., the M-M experiment) and the failure of all attempts to send signals faster-than-light.

The point is, if we should someday encounter some empirical evidence that Bell Locality is violated (i.e., that it is impossible to formulate dynamical laws in a Lorentz invariant way) then we will simply have to accept that, despite the tons of evidence for this conclusion that was found up to this point, the conclusion turns out to have been premature and wrong -- Lorentz invariance is *not* the fundamental/final word in spacetime structure. ...which is, though surely *surprising*, not the end of the world (unless one has this crazy view that we know it is true a priori, from god).

This is basically nothing but our same old disagreement about whether or not locality is proved by experiment, whether or not MWI is a counterexample to my claims that nonlocality is proved by experiment. I take the outcomes of experiments ("naively" interpreted) as the rock-bottom givens. Theory, in my opinion, must always stand or fall with experimental data. So if (as I think) some experimental data is in conflict with the idea of Bell Locality (i.e., the fundamental principles of relativity theory) then it is so much the worse for locality/relativity. We must reject these ideas as wrong -- or at any rate non-universal. On the other hand, you seem to give this "untouchable rock-bottom" status to something very high up and abstract, namely the Lorentz invariance of dynamical laws. As I understand it your view is that this is sacred and untouchable, so if the experimental data "naively" appears to conflict with it, we need to find some way to reinterpret the experimental data so as to render it consistent with the sacred principle -- and hence is born this idea that, really, when Bob thinks his needle points left, he's deluded, and what's actually happening is that it's pointing both left and right at the same time for two different Bobs (or in two different universes or branches of the wf or whatever).

As a philosophical summary of all this, I'd say that I'm much more of an empiricist and you're much more of a rationalist. What's sacred to me is basic perceptual facts like Bob sees his needle go left; all the abstract stuff about Lorentz invariance and such is, if need be (i.e., if empirical data requires it), negotiable. What's sacred for you is the abstract principle, while all the nitty-gritty perceptual facts (Bob seeing the needle go left) are, if need be (i.e., if the principle requires it), negotiable.

And just to hint at what I think is wrong with your approach: as an empiricist (who doesn't believe we have *any* a priori knowledge, no revelations from God, etc.) I think that we only get to abstract principles by organization/interpretation of the closer-to-perception type data. So your whole approach strikes me as circular: you're willing to radically reinterpret something like Bob's perception of a hunk of aluminum in front of him, in order to "save" some abstract principle which (I would argue) we only believe in in the first place because we accepted as given such things as Bob's ability to correctly perceive bits of aluminum in front of him.

As such, it doesn't make, IMO, much sense to *even talk* about signal locality if the Minkowski geometry of spacetime is not respected.

I don't agree. Signal Locality is a summary of empirical facts. We know from experience that it isn't possible (by any mechanism studied so far) to send signals faster than light. It is not unreasonable to formulate this summary as a principle and hypothesize that it is general, i.e., to expect future theories to also respect it. But you're right that, without some kind of prior certainty about the underlying cause of signal locality (minkowski spacetime structure) we can't be *certain* of this extrapolation. And that is perhaps frustrating... but if one is an empiricist at least, that's just the way things are, the way science works.

There is not the remotest motivation to even think about signal locality if the Minkowski geometry is not there. Of course, there can be a kind of "conspiration" for a theory, which doesn't respect the Minkowski geometry, to nevertheless manifest requirements that ONLY MADE SENSE for that geometry, but there's absolutely no fundamental reason for this to happen.

I don't see how you could possibly know the latter. Minkowski geometry is one possible underlying cause for our in-practice inability to send signals FTL. But something like what you call a conspiracy is also a possible cause of this. The fact is, we just don't know for sure a priori. What the cause is of our inability to send signals FTL will have to be ultimately settled by experiment and (future) theories based on experiment. Your attitude seems to be that all the experiments prior to (say) 1950 constitute logically sufficient proof that spacetime has this minkowski geometry, and that this principle is therefore now untouchable, and that we therefore have to go to these ridiculous crazy MWI ideas in order to respect that principle. I completely disagree with that initial claim to certainty, though. It was known (or should have been known) all along that an "ether" type view was logically possible and consistent with all the empirical data (if inelegant). I'm not saying people should have believed in the ether, just that they shouldn't have foreclosed on it and claimed certainty about this so early. And, I think, that chicken is now coming home to roost, because the Bell/Aspect stuff (if you just accept all the experimental data at face value) proves that minkowski geometry is not the final word in spacetime structure.

It is a bit as if one discovered that, say, all international banc transfer amounts in the last 20 years, expressed in dollars, are prime numbers. Of course this is possible. But there's no reason to it.

Spoke like a true rationalist. =) If we *did* discover this, of *course* there'd be some reason, some cause, for it. It'd be too much of a coincidence to happen by coincidence. But just because no reason, no explanation, is apparent to us right now, doesn't mean we should just stipulate that it's impossible and then proceed to invent wild fantasies about how, really, when Joe Banker counts dollar bills, he is deluded into thinking there's 7 of them when really there's 10...

In the same way, a theory that doesn't respect the Minkowski geometry of spacetime (is not expressible in a Lorentz-invariant way) has absolutely no "reason" to respect information locality. It's of course possible. But it would be a very weird thing, as with the banc tranfers being prime numbers.

Bohmian Mechanics provides a nice counterexample of this claim. It violates Bell Locality (and the easiest way to formulate it is to have a preferred reference frame, contra Minkowski) and yet respects signal locality. Why? Because of the uncertainty in the initial locations of the particles -- uncertainty which turns out, as a *theorem* of Bohmian Mechanics, to be *absolute*. See

http://www.arxiv.org/abs/quant-ph/0308039

 

[ttn]

Oh, one other quick comment: all this debate about "signal locality" is interesting, but really beside the point. Signal Locality in fact plays no role whatever in Bell's argument for nonlocality. I only raised it (w/ Dr C) because I have too often been burned by people listening to the argument, only to play the switcheroo game at the end and saying: OK, but really there's no nonlocality here because you can't send a signal with it. So I wanted to make absolutely sure we agreed, up front, that there are two in-principle distinguishable questions -- whether you can send a signal FTL, and whether a given theory respects relativity's prohibition on superluminal causation.

The claim I am at pains to demonstrate (and it is not original to me, it is Bell's claim, which has been widely misunderstood) is that no theory respecting that latter condition (no causally local theory, no Bell Local theory) can be in agreement with the empirical data. Not just no "hidden variable theory" that is local, not just any "deterministic" theory that is local, but no theory *at all* that is local. No causally local theory *at all* can be in agreement with experiment. This is what I am convinced (by Bell) is true. And this claim has nothing whatever to do with signal locality -- except that it is perhaps a bit of a surprise that a Bell Non-Local theory can nevertheless prohibit superluminal signalling.

 

[vanesch]

However, I do *not* think it's the case that God whispers this in our ear. That spacetime has a certain structure is an *inference* from more concrete experimental data, including such things as the measured invariance of the speed of light (e.g., the M-M experiment) and the failure of all attempts to send signals faster-than-light.

Of course. Let's say that taking on this structure for spacetime gives us then IMMEDIATELY the reason for:
1) the constancy of the speed of light,
2) the failure to send signals faster than light
3) the transformation properties in special relativity
4) the EXPLANATION of why ALL LAWS OF NATURE seem to be able to be expressed as Lorentz invariant quantities.
...

In other words, this is ONE SINGLE principle, from which a WHOLE LOT of consequences can be derived, and which ALL have been verified experimentally.
So, if some reasoning, which also makes OTHER hypotheses, leads us to conclude, that after all, this structure of spacetime cannot be true, it takes a really convincing argument that it is THIS point, and not all the other hypotheses, that need to fail.

The point is, if we should someday encounter some empirical evidence that Bell Locality is violated (i.e., that it is impossible to formulate dynamical laws in a Lorentz invariant way) then we will simply have to accept that, despite the tons of evidence for this conclusion that was found up to this point, the conclusion turns out to have been premature and wrong -- Lorentz invariance is *not* the fundamental/final word in spacetime structure.

This would be true if Bell locality were a direct consequence of the spacetime structure of Minkowski space. But there are extra hypotheses needed to do so, and the most evident OTHER hypothesis is the denial of that other great principle: the superposition principle.

This is basically nothing but our same old disagreement about whether or not locality is proved by experiment, whether or not MWI is a counterexample to my claims that nonlocality is proved by experiment. I take the outcomes of experiments ("naively" interpreted) as the rock-bottom givens. Theory, in my opinion, must always stand or fall with experimental data. So if (as I think) some experimental data is in conflict with the idea of Bell Locality (i.e., the fundamental principles of relativity theory) then it is so much the worse for locality/relativity. We must reject these ideas as wrong -- or at any rate non-universal. On the other hand, you seem to give this "untouchable rock-bottom" status to something very high up and abstract, namely the Lorentz invariance of dynamical laws. As I understand it your view is that this is sacred and untouchable, so if the experimental data "naively" appears to conflict with it, we need to find some way to reinterpret the experimental data so as to render it consistent with the sacred principle -- and hence is born this idea that, really, when Bob thinks his needle points left, he's deluded, and what's actually happening is that it's pointing both left and right at the same time for two different Bobs (or in two different universes or branches of the wf or whatever).

Yes. For TWO reasons, not only one. We already found one great principle, which is the Minkowski structure of spacetime, as being able to explain naturally points 1,2,3 and 4 in my little list above. These points include A LOT of empirical evidence, and none EXPLICITLY against it (that is, a DIRECT derivation of a result based upon the spacetime structure being a Minkowski space, in contradiction with experiment ; say, the wrong life times of muons or so, or clocks not behaving as computed).
But we discovered also ANOTHER great principle, which is the superposition principle, which ALSO explained a lot of empirical results. It gave rise to all of quantum mechanics, in its "bare bones" applications, from atomic spectral lines, solid state stuff, ...
Applying "naively" the superposition principle to "remote Bob and his needle" would imply that indeed, Bob exists in two states. But somehow we don't want to see that, so we state that this shouldn't be so.
So now we MAKE THE ASSUMPTION that 1) the superposition principle DOES NOT APPLY TO remote Bob and his needle, and 2) the Minkowski geometry of spacetime and we arrive at a contradiction with a prediction of a theory based upon the superposition principle - and some experimental support for it: namely the violation of Bell locality.
So it seems that if, at a certain point, you DO NOT ALLOW FOR THE SUPERPOSITION PRINCIPLE anymore, and you assume the superposition principle for microscopic systems, that you run into problems with Minkowski spacetime, both theoretically and experimentally.
Conclusion: Minkowski space is dead. Is it ? Or is it in the *assumption of the non-application of the superposition principle* that the error resides ?

It sounds a bit weird that we conclude about the non-validity of two principles, namely the superposition principle, and the Minkowski structure of spacetime, which were otherwise empirically very successful, simply because at a certain point we REFUSE to apply the superposition principle, no ?

As a philosophical summary of all this, I'd say that I'm much more of an empiricist and you're much more of a rationalist. What's sacred to me is basic perceptual facts like Bob sees his needle go left; all the abstract stuff about Lorentz invariance and such is, if need be (i.e., if empirical data requires it), negotiable. What's sacred for you is the abstract principle, while all the nitty-gritty perceptual facts (Bob seeing the needle go left) are, if need be (i.e., if the principle requires it), negotiable.

That's a bit carricatural. Every great principle is negociable, ON THE CONDITION that we have a better, more encompassing principle to replace it. One that has MORE explanatory power.

And just to hint at what I think is wrong with your approach: as an empiricist (who doesn't believe we have *any* a priori knowledge, no revelations from God, etc.) I think that we only get to abstract principles by organization/interpretation of the closer-to-perception type data. So your whole approach strikes me as circular: you're willing to radically reinterpret something like Bob's perception of a hunk of aluminum in front of him, in order to "save" some abstract principle which (I would argue) we only believe in in the first place because we accepted as given such things as Bob's ability to correctly perceive bits of aluminum in front of him.

This is often the case, that, when things are better understood, we realize that our perception of things were not what we thought it was, but nevertheless at a certain point we had to go by there to arrive where we are now.

You could say the same of, say, the "atomic hypothesis", which was regarded as a very hypothetical idea in the 19th century, but of which one had to recon, one could deduce quite some observed facts. One could argue a bit like you do here: in order to make the atomic hypothesis (which includes that objects are made of tiny little things, with lots of empty space in between them), we'd need to consider that we are "deluded" in thinking that we have massive objects of continuous matter around us, while these are essentially "empty" pieces of space, with some tiny tiny matter points in them. And the funny thing is, that to even MAKE the atomic hypothesis, we have to use instruments that are EXACTLY MADE OF MASSIVE, CONTINUOUS MATTER.
Now, it might be that using the atomic hypothesis, you might have an explanation of why atomic matter *looks and feels* like massive, continuous matter, but isn't it strange that in order to "save this abstract principle of atomic matter" one has to deny the existence of the continuous matter which we used in the first place to arrive at this "atomic matter". We used "continuous" electrical wires, "continuous" pieces of metal, ...

I don't agree. Signal Locality is a summary of empirical facts. We know from experience that it isn't possible (by any mechanism studied so far) to send signals faster than light. It is not unreasonable to formulate this summary as a principle and hypothesize that it is general, i.e., to expect future theories to also respect it. But you're right that, without some kind of prior certainty about the underlying cause of signal locality (minkowski spacetime structure) we can't be *certain* of this extrapolation. And that is perhaps frustrating... but if one is an empiricist at least, that's just the way things are, the way science works.

But the great advancements of science are exactly when one realizes that a general principle encompasses a lot of empirical facts.

I don't see how you could possibly know the latter. Minkowski geometry is one possible underlying cause for our in-practice inability to send signals FTL. But something like what you call a conspiracy is also a possible cause of this. The fact is, we just don't know for sure a priori.

Yes, but when you get a lot of "conspiracies" that can find their explanation in a principle otherwise, one should really consider that principle, no ?

What the cause is of our inability to send signals FTL will have to be ultimately settled by experiment and (future) theories based on experiment. Your attitude seems to be that all the experiments prior to (say) 1950 constitute logically sufficient proof that spacetime has this minkowski geometry, and that this principle is therefore now untouchable, and that we therefore have to go to these ridiculous crazy MWI ideas in order to respect that principle.

First of all, if one takes the attitude that one can never formulate a principle because one day it might be falsified, one isn't going to make much progress!
I'm not stating that the Minkowski structure is there once and for all (hey, we already know it is not the case thanks to general relativity!) I'm saying that NOTHING has ever been found that contradicted it. The only "contradiction" that is found, is when you are using another principle and violate it at the same time: the superposition principle.
Entanglement is a consequence of the superposition principle ; and saying that Bob CANNOT be in two states is denying this. Now from this double standard, you derive that Minkowski spacetime must come into troubles. My answer is: apply RIGOROUSLY the superposition principle, and apply RIGOROUSLY the Minkowski spacetime structure, and you don't have the problems you're talking about.

I completely disagree with that initial claim to certainty, though. It was known (or should have been known) all along that an "ether" type view was logically possible and consistent with all the empirical data (if inelegant). I'm not saying people should have believed in the ether, just that they shouldn't have foreclosed on it and claimed certainty about this so early. And, I think, that chicken is now coming home to roost, because the Bell/Aspect stuff (if you just accept all the experimental data at face value) proves that minkowski geometry is not the final word in spacetime structure.

So, spacetime is not Minkowski, but behaves in almost all respects AS IF it were. The superposition principle is not valid, but things behave AS IF it were valid on microscales. We know that if spacetime IS Minkowski, and the superposition principle IS valid, that there is no problem and that all empirical data can be explained too, but this goes against our intuition. Hmmm...

Bohmian Mechanics provides a nice counterexample of this claim. It violates Bell Locality (and the easiest way to formulate it is to have a preferred reference frame, contra Minkowski) and yet respects signal locality.

Bohmian mechanics also violates a priori signal locality. It is only when we equip it (with much difficulty!) with Lorentz-invariant dynamics apart from the quantum potential that it doesn't. In other words, Bohmian mechanics has NO EXPLANATION for signal locality ; it has to be put in there by hand.

So does unitary quantum theory BTW. But the difference is that unitary QM can be made COMPLETELY Lorentz invariant. As such, Minkowski spacetime can be considered, and serves then as an explanation for it. In other words, quantum theory can be considered over Minkowski spacetime. In Bohmian mechanics, as its formulation does not allow for Minkowski spacetime to exist, and there are explicit violations of this formulation to be Lorentz invariant. But SOME expressions must be Lorentz invariant while others aren't. And if you do the mix in the right way - with no good reason as of WHY - you can get out signal locality.
So this framework has much less explanatory power than the combination of Minkowski spacetime and the superposition principle.

 

[DrChinese]

So... still with me? Do we agree that, in principle, a theory can violate Bell Locality (and hence not be consistent with relativity) yet still not support the sending of signals faster than light?

Yes, I guess I can see it conceptually. There could exist tachyon-like particles that do not otherwise interact with currently known particles except for their ability to synchronize chance events at space-like separated spaces. Were this the case, we would presumably need to extend/modify relativity to compensate.

 

[ttn]

In other words, this is ONE SINGLE principle, from which a WHOLE LOT of consequences can be derived, and which ALL have been verified experimentally.
So, if some reasoning, which also makes OTHER hypotheses, leads us to conclude, that after all, this structure of spacetime cannot be true, it takes a really convincing argument that it is THIS point, and not all the other hypotheses, that need to fail.

I agree. It takes a really convincing argument. But I think Bell gave such an argument.

Yes. For TWO reasons, not only one. We already found one great principle, which is the Minkowski structure of spacetime, as being able to explain naturally points 1,2,3 and 4 in my little list above. These points include A LOT of empirical evidence, and none EXPLICITLY against it (that is, a DIRECT derivation of a result based upon the spacetime structure being a Minkowski space, in contradiction with experiment ; say, the wrong life times of muons or so, or clocks not behaving as computed).
But we discovered also ANOTHER great principle, which is the superposition principle, which ALSO explained a lot of empirical results. It gave rise to all of quantum mechanics, in its "bare bones" applications, from atomic spectral lines, solid state stuff, ...

I don't agree at all that it's just "the superposition principle" which leads to an explanation for atomic spectral lines, solid state stuff, etc. Surely, at very least, you need the usual quantum measurement axioms (the collapse postulate, or some proto-version of it like Bohr's "quantum jumps" between stationary states). If you *just* take the superposition principle (and the unitary quantum dynamics) you get Schroedinger's cat paradox spiralling out of control in all directions -- *not* an explanation of why a certain atom emitted light at a certain frequency, but an absurd-looking claim that there's no definite fact of the matter about what the atom did or didn't do. You can only convert this into an explanation of spectral lines (etc) if you *completely* change the usual, ages-old scientific assumption that what you see is what you get.

Applying "naively" the superposition principle to "remote Bob and his needle" would imply that indeed, Bob exists in two states. But somehow we don't want to see that, so we state that this shouldn't be so.

It has nothing to do with "want". Or rather, the *reason* "we don't want to see that" (as a prediction of our theory) is that, in actual empirical fact, we *don't* see that! You make it sound like it's some kind of arbitrary, unsupportable, subjective whim that makes me want theory to conform itself to perceptual experience / data, rather than the other way round. Call it what you want; I call it *the fundamental axiom of science*.

So now we MAKE THE ASSUMPTION that 1) the superposition principle DOES NOT APPLY TO remote Bob and his needle, and 2) the Minkowski geometry of spacetime and we arrive at a contradiction with a prediction of a theory based upon the superposition principle - and some experimental support for it: namely the violation of Bell locality.
So it seems that if, at a certain point, you DO NOT ALLOW FOR THE SUPERPOSITION PRINCIPLE anymore, and you assume the superposition principle for microscopic systems, that you run into problems with Minkowski spacetime, both theoretically and experimentally.

I (and virtually everyone else, even the quantum founders that I have a very low esteem for!) *already* did not allow for the superposition principle. There's a *good empirical reason* why you need something like the collapse postulate -- not for super complex EPR type situations but for simple things like cats. The collapse postulate was introduced in the first place to make QM consistent with *experiment*. The idea of getting rid of the collapse in favor of unitary-only-dynamics is already, for me, off the table long before we get to any questions about nonlocality, etc.

It sounds a bit weird that we conclude about the non-validity of two principles, namely the superposition principle, and the Minkowski structure of spacetime, which were otherwise empirically very successful, simply because at a certain point we REFUSE to apply the superposition principle, no ?

If you put it that way it sounds weird, yes. But if you remember why people refused to apply the superposition principle all the time (universally) in the first place, there is nothing in the least surprising or weird here.

That's a bit carricatural. Every great principle is negociable, ON THE CONDITION that we have a better, more encompassing principle to replace it. One that has MORE explanatory power.

I strongly disagree with this. It's entirely possible to know that some theory is wrong, without knowing yet what better theory might replace it. This is part of what it means to be an empiricist. Somebody can propose something for which there is some good evidence -- even lots of good evidence -- and which is, say, the only currently-known way to explain a certain broad range of phenomena. But this is *not* sufficient to prove with absolute untouchable certainty that the theory is true. Sometimes new (surprising) data shows that the theory is wrong, and scientists have to go back to the drawing board. Psychologically this is of course difficult to accept. But physics isn't psychology. It's based on experimental data, not whatever-keeps-physicists-happy.

You could say the same of, say, the "atomic hypothesis", which was regarded as a very hypothetical idea in the 19th century, but of which one had to recon, one could deduce quite some observed facts. One could argue a bit like you do here: in order to make the atomic hypothesis (which includes that objects are made of tiny little things, with lots of empty space in between them), we'd need to consider that we are "deluded" in thinking that we have massive objects of continuous matter around us, while these are essentially "empty" pieces of space, with some tiny tiny matter points in them. And the funny thing is, that to even MAKE the atomic hypothesis, we have to use instruments that are EXACTLY MADE OF MASSIVE, CONTINUOUS MATTER.
Now, it might be that using the atomic hypothesis, you might have an explanation of why atomic matter *looks and feels* like massive, continuous matter, but isn't it strange that in order to "save this abstract principle of atomic matter" one has to deny the existence of the continuous matter which we used in the first place to arrive at this "atomic matter". We used "continuous" electrical wires, "continuous" pieces of metal, ...

This is not a parallel case at all. The atomic theory gives a detailed micro-picture of (stuff like) solid matter. It in *no way contradicts* what we observe with our eyes (unless your eyesight is about 10^10 more accurate than mine). Accepting the atomic theory does *not* mean believing that what you previously believed (about stuff like solid matter) was delusional.

But the great advancements of science are exactly when one realizes that a general principle encompasses a lot of empirical facts.

Sure, but even greater advancements happen when one discovers whether such general principles are true or false.

Yes, but when you get a lot of "conspiracies" that can find their explanation in a principle otherwise, one should really consider that principle, no ?

Sure. Unless it's CRAZY! Seriously, think about all the conspiracies *your* side would have us believe: it *looks* like there's only one world, like the cat is always either dead or alive, like the needle always points either left or right, etc., yet there is some weird conspiracy in the mind-matter relationship (which is where you put the Born rule) which just makes it "appear" this way to poor deluded souls; really, the world is entirely different, in literally every respect, from those appearances. I submit that this is the ULTIMATE CONSPIRACY, the exact equivalent of the "brain in vat" scenario that no empiricist/scientist should (or even can) take seriously.

First of all, if one takes the attitude that one can never formulate a principle because one day it might be falsified, one isn't going to make much progress!

One often makes progress by extrapolating something that might be general/universal simply to find out if it is -- e.g., P*V = const... is this a universal law, or will it break down at high pressures? Let's crank up the pressure and *look*. It's by such means that one eventually discovers the underlying causes of observed regularities, which is where the real progress in science lies.

Bohmian mechanics also violates a priori signal locality. It is only when we equip it (with much difficulty!) with Lorentz-invariant dynamics apart from the quantum potential that it doesn't. In other words, Bohmian mechanics has NO EXPLANATION for signal locality ; it has to be put in there by hand.

Huh? Bohmian Mechanics does not violate signal locality. (What do you mean by "a priori signal locality"?) It is just as signal-local as orthodox QM because it shares precisely the same empirical predictions as orthodox QM. I also don't understand the rest of what you said here. Bohmian mechanics does not have a Lorentz-invariant dynamics; if you equip it with one, you get some new theory which would be ruled out by EPR/Bell experiments. It's true that this new theory would also be signal local. But that's not interesting; all Bell Local theories will be signal local. The interesting point is the converse: not all signal local theories are Bell Local. That is, what's interesting is that the *original* non-loretnz-invariant-dynamics Bohm theory is *already* signal local.

I don't understand why you say BM has no explanation for signal locality. Yes, it does. The theory as a whole is the explanation, since signal locality is a consequence -- a prediction -- of the theory. (Of course this alone doesn't prove anything about the truth of the theory, since lots of other theories make the same prediction here; the virtue of Bohm's theory lies elsewhere, basically in its uniform treatment of all physical interactions, as contrasted with the orthodox view which postulates fundamental dynamical distinctions between subject and object, between measurement-interactions and normal-interactions, etc.)

So does unitary quantum theory BTW. But the difference is that unitary QM can be made COMPLETELY Lorentz invariant.

Sure, all sorts of crazy things can be made completely lorentz invariant. The problem is, "unitary quantum theory" is contradicted by every experiment that's ever been done (the Sch cat problem)... so the fact that it can be made lorentz invariant isn't quite enough to make me want to consider it as possibly-viable.

 

[DrChinese]

...and whether a given theory respects relativity's prohibition on superluminal causation.

That's the issue I have. What would qualify as superluminal causation? My definition is probably a lot different than yours. After all, what is causing what? Correlation is not the same as causation.

 

[ttn]

Yes, I guess I can see it conceptually. There could exist tachyon-like particles that do not otherwise interact with currently known particles except for their ability to synchronize chance events at space-like separated spaces. Were this the case, we would presumably need to extend/modify relativity to compensate.

Yes, that's right. The basic idea is that in such a theory (non-Bell-Local, but signal local) the relativity-violating effects have to be "washed out" somehow or other -- "hidden" or maybe rendered uncontrollable by something like noise, or intrinsic dynamical randomness, or ineliminable ignorance about the precise states of things prior to measurements, or some such.

OK, so I gather we're on the same page about the general point: it's possible for a theory to violate Bell Locality, yet still be signal local.

As I said before, "signal locality" doesn't really play a role in my argument here. It's just important to make clear from the beginning that the claim I'm arguing for involves something *different* from signal locality (namely Bell Locality). The claim is going to be that no Bell Local theory (that is, no theory that is relativistically causal, no theory that respects relativity's prohibition on superluminal causation) can agree with experiment (even though there will be theories that *do* agree with experiment in which, for whatever reason, the necessary superluminal causation cannot be harnessed by humans to send signals). OK?

So maybe here I'll just outline what the structure of the argument is going to look like and make sure we're both on the same page about that before scrutinizing the parts in more detail.

The argument will have two parts, both involving the typical EPR/Bell setup where Alice and Bob make space-like separated measurements of spin components on the two members of a pair in a singlet spin state.

Part One: It is an empirical fact that whenever Alice and Bob measure along the same axis, their results are *perfectly (anti-) correlated*. We then ask: how must a Bell Local theory work if it is going to successfully predict this empirical fact? The answer turns out to be: its state descriptions must include variables which, in effect, pre-assign measurement outcomes for all spin components. Such variables are often called "local hidden variables" or some such.

Part Two: This part is the better-known part, Bell's Theorem. There is a straightforward mathematical proof that the kind of theory we arrived at in Part One (a "local hidden variables" theory, or whatever you want to call it) cannot make the empirically-correct predictions for the general case (where Alice and Bob don't necessarily measure along the same axis).

The conclusion of the two parts put together is then just the conclusion I keep alluding to: no Bell Local theory can be in agreement with experiment. (If the logic there is not clear, I'd be happy to elaborate... but I think it's pretty obvious. In Part One we infer X from Bell Locality and a certain sub-class of the empirical data; in Part Two we infer that X is inconsistent with some other class of empirical data. And the only assumption other than empirical data we can possibly blame for this is Bell Locality.)

As an aside, note that none of this makes any reference whatsoever to Orthodox QM or Bohm's Theory or any other specific theory. We start with no assumptions whatever about the kinds of theories we'll consider, except for insisting on relativisitic causality / Bell Locality. We then ask: of that universe of Bell Local theories, are there any that are consistent with *both classes* of empirical data? And the answer turns out to be no. Some people might think "well there's got to be something wrong with the argument since Orthodox QM is a counterexample" -- this is really beside the point, but it is not a counterexample. Orthodox QM violates Bell Locality. But I think it's better to put off talking about specific theories (possible counterexamples, etc.) until after the *general* argument is grasped.

So... I'll stop there for now and wait for the daily sign-off. =) Are we on the same page with everything still? At this point, I won't be exactly sure where to go next unless you help me by telling me which part of the argument I've laid out you're suspicious of. For example, you might be totally cool already with Part Two, and so there'll be no point wasting time going through that in detail. Anyway, let me know what part of this you'd like to scrutinize and I'll do my best to flesh out that part of the argument.

 

[vanesch]

If you *just* take the superposition principle (and the unitary quantum dynamics) you get Schroedinger's cat paradox spiralling out of control in all directions -- *not* an explanation of why a certain atom emitted light at a certain frequency, but an absurd-looking claim that there's no definite fact of the matter about what the atom did or didn't do. You can only convert this into an explanation of spectral lines (etc) if you *completely* change the usual, ages-old scientific assumption that what you see is what you get.

If you *just* take the superposition principle, and apply it ALL THE WAY, you get a multitude of classically-looking, mostly non-interacting systems. This has not been realized immediately, you need decoherence for that. It is as if the following happened: imagine you have a classical theory of the universe, with its phase space and flow, and a point representing its actual state. From that point, you need to derive all what you observe and experience, your memories, everything you know about. You're happy, the world is a point in a big phase space.
Now, imagine a theory that gives you exactly that: the correct phase space, and that point. I assume that you would now say that this theory describes perfectly the universe as you know it.
And, now, imagine, that that theory ALSO gives a second phase space, which is unconnected to this first one, with another point in it.
Is this theory OBVIOUSLY WRONG ? I don't think so. As long as you have the phase space you wanted, with the point you wanted, whatever is "extra" doesn't matter. Whether or not to consider it is a matter of economy. If this second phase space is introduced "by hand" there's certainly no reason to consider it ; but if it FOLLOWS from a more general mathematical formulation, there's no reason not to consider it.
As a silly example, the two phase spaces could be solutions in the same way the two blades of a hyperbola are solutions to a geometrical equation in R^3. In what way would you feel "deluded" if there where this second phase space ? You have your classical phase space, and your point, and your flow, and you were able to deduce all your memories and so on from this structure. So if a physical theory gives you this, then it is reasonable to expect that ANY physical theory which generates a similar structure, will give you a comparable point where there is a creature like you, with memories like you and all that. And if that theory also generates OTHER structures, doesn't really matter, does it ?

And this is what happens when you apply unitary dynamics to a large system: approximate structures arise which look very much like the flow in a phase space, for each individual decohered term. It is as if each individual term "autogenerated" approximately its own classical phase space with flow. So ONE unitary flow in a complicated enough hilbert space gives rise to several, approximate, classical dynamics, each in their approximate phase spaces, and a specific vector in hilbert space corresponds to several points in each of these classical phase spaces. If you pick out ONE of these phase spaces, you'd have a point there, and that point would then correspond to what you would classically have done, and deduced from that point that there was somebody like you, with memories like you, and which would be evolving rather classically.
And another such phase space would look very similar, but different in a few aspects.
What's wrong with this view, so far ?
Strictly unitary dynamics over a hilbert space has given rise to SEVERAL classical phase spaces with their flows (at least approximately).
If that were the ONLY thing that unitary dynamics did, we could just as well stick to just ONE phase space, with its "derived" flow on it.
But, sometimes, unitary dynamics does something else: from one phase space, two are generated. Two almost identical phase spaces, with points in almost identical situations. This is due to the approximate character of the generated classical theories.
For a creature represented by a "point in the classical phase space before the split", a classically incomprehensible thing would happen, and because of the split, it would - being a classical creature - appear afterwards in two copies in the two classically-like phase spaces that result from this.
Of course there's no way, purely from this description, to tell you what would be the SUBJECTIVE experience of this creature.
And this is where the Born rule, the collapse, and all that, comes in: It will be subjectively now corresponding to one of the two resulting approximately classical phase spaces, with a measure given by the Born rule.

So what do we have ? You say that all of our perceptions and so result from a point in a classical phase space (that's what you do as a Bohmian, or as a classical physicist). Ok. So a theory that gives you, to good approximation, such a phase space with point will correspond entirely to all of your perceptions, right ? If it also generates OTHER such spaces is, somehow, none of your business, right ? You're not being "deluded" if that's so, right ? All you need is *A* classical phase space plus point.
Fine, I tell you that that's what you get out of unitary dynamics.

Next, it sometimes happens that the so generated approximate classical phase space "dedoubles", "splits", whatever, and that there is only a minor difference between the two resulting phase spaces (the minor difference being an "outcome of experiment" which is different in both cases). Ok, the only question that remains now is: if you were subjectively "living" the phase space before split, which one are you going to live after the split ? And here, the answer is given by the Born rule, of the two terms in the wavefunction that are at the origin of the two newly generated approximately classical phase spaces + point.

you can see the wavefunction as a "generator of classical phase spaces + point" and the unitary flow over Hilbert space as a "generator of classical flows" over all of these phase spaces.
With the occasional split.

 

[ttn]

What's wrong with this view, so far ?

I don't have much to say here that I haven't said before, and I don't want this to eclipse entirely what might actually be some progress (gasp) with Dr C.

But, for the record, I don't accept the parallel between branches of the wf and different phase spaces. No, I wouldn't object to a theory merely on the grounds that it somehow predicted some extra phase space appendage that didn't correspond to (nor contradict) anything so far observed. (...though that would be very weird and I don't see how it could possibly happen.) But the reason for this is simple: a different phase space would mean a bunch of particles which are located off in some other part of the universe, or which are located in the regular universe but which don't interact with the matter we're used to. Their existence would be something *extra*, on top of the stuff I already knew about. For example, if I say I've got a box with 10 marbles in it, but it turns out there are also some sneutrinos flying through, that doesn't contradict what I said about the marbles; it supplements it, yes, but it in no way undermines it; it in no way forces me to say I was *deluded* about the real existence of the marbles.

Other branches/worlds/universes in MWI just don't work that way. Actually they *would* work that way if you took the misnomer title "many worlds" seriously, but *that* version of the theory just doesn't work (isn't even well defined). ...which is why you uphold the particular version of MWI you do uphold, in which there's only one universe, and a complete state of the matter in it is given by a big giant wave function obeying unitary dynamics always. But then you put in this "consciousness token" -- at random -- into one or another of the branches. So in this (btw, solipsist) theory, what you are aware of is merely the branch in which you happen to find yourself, so to speak. You experience a living cat, but the *truth* about the cat (according to the theory) is that it is *not* really living, but is really in a massively entangled state including both living and dead (and never existed and quantum tunneled just now into santa claus and...) components. Your belief about the way things are, is *false*. You are *deluded*.

We can argue all day about whether this is really a delusion, or what, but the fact is there is a crucial difference between having veridical awareness of some fact (but lacking omniscience) and having a belief which doesn't correspond to the way things really are. I accept your phase space hypothetical because it falls under the former; I can't accept MWI because it requires me to put *all* my knowledge in the latter category. And I'm just not able to accept "maybe *everything* I know is false."

Maybe here's the best reason not to bother arguing about this anymore: if you're right, I don't even exist (or I only exist as a mindless hulk), and in either case you'll never convince me. I'm sure that sounds like a joke, but I actually mean it quite seriously. If you are right about MWI, none of the rest of us exist as conscious beings -- we are just mindless hulks who can't actually think. So what can it possibly matter what we think? i.e., why should you bother trying to convince us of anything? You'd have an easier time (and equal odds of success) trying to convince your kitchen sink. =)

 

[vanesch]

Huh? Bohmian Mechanics does not violate signal locality. (What do you mean by "a priori signal locality"?) It is just as signal-local as orthodox QM because it shares precisely the same empirical predictions as orthodox QM. I also don't understand the rest of what you said here. Bohmian mechanics does not have a Lorentz-invariant dynamics; if you equip it with one, you get some new theory which would be ruled out by EPR/Bell experiments.

What I simply meant was that if the Hamiltonian doesn't imply Lorentz invariance and signal locality, then, Bohmian mechanics, as well as non-relativistic quantum mechanics, is NOT signal local. For instance, in NR QM, if you have a term 1/(r2 - r1) in the hamiltonian, you can use this to signal instantaneously (in Bohmian mechanics as well as in NR QM). That's no surprise because it is the quantum version of a classical theory that also wasn't signal local (Coulomb electrostatics).
You only obtain signal locality in quantum theory by going to a Lorentz invariant dynamics (such as in QFT).
And here comes the crux: we can UNDERSTAND this requirement for Lorentz invariance because of the Minkowski structure of spacetime - but if we do so, then ALL of the theory has to be Lorentz invariant.
In Bohmian mechanics, in order to save signal locality we have to do the same thing: we have to introduce a Lorentz invariant dynamics (which you can do). But this time, because a PART of the theory is not Lorentz invariant, Minkowski spacetime cannot be the explanation. So there's no explanation for the fact that we had to write the part, corresponding to the dynamics, in a Lorentz invariant form.

Sure, all sorts of crazy things can be made completely lorentz invariant. The problem is, "unitary quantum theory" is contradicted by every experiment that's ever been done (the Sch cat problem)

Again, the Schroedinger cat is NOT contradicted by experiment. Schroedinger didn't know about decoherence effects, and thought that his cat would end up in a weird state "half dead, half alive". But, when you look upon the quantum theory and the wavefunction as "generator of approximately classical theories" (which is what happens in unitary QM), then the only thing that the "Schroedinger cat" setup tells you, is that there are now, to good approximation, two classical phase spaces, one in which the point indicates "dead cat" and another in which the point indicates "live cat", and everything that goes with it. "You" also being an aspect of that point in phase space, in one phase space, there is something that looks like "you" which sees a dead cat, and in the other phase space, there's something like "you" which sees a live cat.

All experiments with cats so far do indicate that the predictions are VALIDATED: there's always a "you" who sees a live cat, or who sees a dead cat.

The only problem with this "multitude of classical phase spaces, points and flows" being generated by unitary quantum theory, is: which one do *I* observe subjectively ? And that's answered by the Born rule, which is a NEW ingredient into something that looks very classical otherwise.

I don't find that so crazy.

 

[ttn]

That's the issue I have. What would qualify as superluminal causation? My definition is probably a lot different than yours. After all, what is causing what? Correlation is not the same as causation.

OK, a very good question. I'll just tell you what Bell wrote down as a mathematical condition he thought captured this.

First off, a theory must have some "state assignments". It must describe physical systems in some way. So, for OQM, we have wave functions only, for Bohm's theory we have wave functions plus particle positions, for classical physics we have particle positions and momenta, for electrodynamics we have electric and magnetic field configurations, etc.

Now here is Bell's "local causality criterion" (which I call Bell LOcality for short). The probability a theory assigns to a given event should be uniquely specified once a complete description of the physical state in the past light cone of the event is given. (What "uniquely specified" here means is that the probability, assigned by the theory to the event in question, won't change if you additionally specify some information that is not located in the past light cone.)

Example:

Suppose there's a charged particle sitting at the origin, and you want to know what it's going to do at time t. Well, suppose you specify the electric and magnetic field configuration inside a sphere of radius c*t surrounding the origin at t=0. According to the theory "Maxwellian Electrodynamics", this is a complete state description, and also according to that theory, once you specify this state description, the behavior of the charged particle is uniquely determined. Note in particular that if you were *also* to specify (say) the price of tea in china (which we assume is outside that sphere at t=0, i.e., spacelike separated) and ask "*now* what does the theory predict the particle will do?" you get *the same answer as before*. The theory's claims about what will happen do not change when that additional information is specified.

According to Bell (and I agree, it makes perfect sense) this is a way of defining what we mean by "locally causal." A theory which *violated* this -- a theory in which the probability assigned by the theory to some event *did* change when some spacelike separated info was given, even though a complete description of the state of the system in the past light cone has already been given -- could, I think obviously, be said to include a causal connection between these spacelike separated events.

So that's Bell Locality. If anyone wants to get this from the horse's mouth, Bell's article "La Nouvelle Cuisine" (reprinted in the 2nd edition of Speakable and Unspeakable, but not the 1st edition) is the best thing to read.

So... will you accept that Bell Locality is a prima facie reasonable way of defining what it means for a theory to respect relativity's prohibition on superluminal causation? Or more questions? Or maybe you already disagree with something?

 

[vanesch]

Other branches/worlds/universes in MWI just don't work that way. Actually they *would* work that way if you took the misnomer title "many worlds" seriously, but *that* version of the theory just doesn't work (isn't even well defined).

No, the problem with calling other branches "worlds" is that people then go "world counting" which runs in all sorts of problems.

...which is why you uphold the particular version of MWI you do uphold, in which there's only one universe, and a complete state of the matter in it is given by a big giant wave function obeying unitary dynamics always.

Yes, but what's the difference between this, and the GENERATED decohered terms which look like classical phase space evolutions ?

But then you put in this "consciousness token" -- at random -- into one or another of the branches. So in this (btw, solipsist) theory, what you are aware of is merely the branch in which you happen to find yourself, so to speak. You experience a living cat, but the *truth* about the cat (according to the theory) is that it is *not* really living, but is really in a massively entangled state including both living and dead (and never existed and quantum tunneled just now into santa claus and...) components.

It depends what you call "cat". If, from this massively entangled system, you deduce TWO classical phase spaces + flow, and you call "there's a cat living" a certain region in such a phase space, then, if the point in the phase space is in that region, then there's a cat living there. So "cat" is then a classical concept, and "a living cat" too. And our quantum system now generated two approximately classical systems, one in which the point is in the region "living cat" of his phase space, and the other in which the point is in the region "dead cat" of his phase space.

Your belief about the way things are, is *false*. You are *deluded*.

No, because your "belief" about your "classical things" is true, and your observations too, within the approximate classical phase space + flow you happen to live.

We can argue all day about whether this is really a delusion, or what, but the fact is there is a crucial difference between having veridical awareness of some fact (but lacking omniscience) and having a belief which doesn't correspond to the way things really are. I accept your phase space hypothetical because it falls under the former; I can't accept MWI because it requires me to put *all* my knowledge in the latter category.

But there's NO DIFFERENCE between both. If you see the wavefunction as a generator of approximately classical worlds, there's really no difference.

Maybe here's the best reason not to bother arguing about this anymore: if you're right, I don't even exist (or I only exist as a mindless hulk), and in either case you'll never convince me. I'm sure that sounds like a joke, but I actually mean it quite seriously. If you are right about MWI, none of the rest of us exist as conscious beings -- we are just mindless hulks who can't actually think.

I never said that MWI IMPLIES that others are mindless hulks. I argued at a certain point that in any case one cannot make any behavioural difference between a mindless hulk and a conscious being, so arguing about OTHER consciousnesses was to me, a waste of time and what counted only was what happened to ONE consciousness. But this is JUST AS WELL true in a classical world, or in Bohmian mechanics or in any other physical theory: if you have a strictly classical theory where you have a phase space and a point in there, what makes you make you experience the experiences you have, given that point in phase space ?
Now, this is the same with a branch in the WF. And given that through decoherence, what happens to a branch in the WF LOOKS A LOT like what happens to a point in phase space, I'd say that on the physics side, we're in almost identical situations.

So if you can live with your "approximate truth" about things in a classical phase space, I don't see what's suddenly so delusional if you understand that this classical phase space exists in an approximate form, AMONGST MANY OTHERS, in the unitary dynamics of quantum theory.

 

[ttn]

And here comes the crux: we can UNDERSTAND this requirement for Lorentz invariance because of the Minkowski structure of spacetime - but if we do so, then ALL of the theory has to be Lorentz invariant.
In Bohmian mechanics, in order to save signal locality we have to do the same thing: we have to introduce a Lorentz invariant dynamics (which you can do). But this time, because a PART of the theory is not Lorentz invariant, Minkowski spacetime cannot be the explanation. So there's no explanation for the fact that we had to write the part, corresponding to the dynamics, in a Lorentz invariant form.

Yup. What can I say, except: I too would be very worried about this if there were independent reason to believe in minkowski spacetime... rather than, as I think is actually the case, strong independent reason to *disbelieve* minkowski spacetime.

Again, the Schroedinger cat is NOT contradicted by experiment. Schroedinger didn't know about decoherence effects, and thought that his cat would end up in a weird state "half dead, half alive".

He was right about that. Decoherence doesn't change that conclusion. It only shows how hard it would be in practice to get those two branches of the wf to interfere again. But the cat is still both alive and dead.

But, when you look upon the quantum theory and the wavefunction as "generator of approximately classical theories" (which is what happens in unitary QM), then the only thing that the "Schroedinger cat" setup tells you, is that there are now, to good approximation, two classical phase spaces, one in which the point indicates "dead cat" and another in which the point indicates "live cat", and everything that goes with it. "You" also being an aspect of that point in phase space, in one phase space, there is something that looks like "you" which sees a dead cat, and in the other phase space, there's something like "you" which sees a live cat.

It *sounds* like you're suggesting that decoherence alone solves the measurement problem. That's not true, as I think you know. Decoherence has nothing to do with this -- what's really "solving" the problem for you here is your putting in a "consciousness token" according to the Born rule. (Decoherence merely makes this pseudo-well-defined by cleaning up the boundaries between the different possible branches.)

Also, in regard to your last sentence, I thought you thought that in the other "phase space" [sic], there *isn't* another "me". Maybe there's a mindless hulk that has the same shape as me, but it isn't conscious, right? So it doesn't "see a live cat" or anything else.

All experiments with cats so far do indicate that the predictions are VALIDATED: there's always a "you" who sees a live cat, or who sees a dead cat.

Sure, but your theory forces me to say that when I "see a live cat" (or whichever) it's a delusion. Ha! That makes you worse than those annoying people who think that Aspect et al made up their data and it's all a big conspiracy...

The only problem with this "multitude of classical phase spaces, points and flows" being generated by unitary quantum theory, is: which one do *I* observe subjectively ? And that's answered by the Born rule, which is a NEW ingredient into something that looks very classical otherwise.

I don't find that so crazy.

That just proves you're crazy. =)

 

[vanesch]

According to Bell (and I agree, it makes perfect sense) this is a way of defining what we mean by "locally causal." A theory which *violated* this -- a theory in which the probability assigned by the theory to some event *did* change when some spacelike separated info was given, even though a complete description of the state of the system in the past light cone has already been given -- could, I think obviously, be said to include a causal connection between these spacelike separated events.

Ok, silly counter example:

Imagine that I throw a dice at event E1. The probability for the dice to give "2" is equal to 1/6, say, when we take into account all we know in its past lightcone. But consider now event E2, a bit in the FUTURE lightcone of E1 where I look at the outcome of the dice which was thrown a bit before E1. The outcome is "4".
If I know this, then the probability of event E1 to give "2" is not 1/6 anymore, but "0". Nevertheless, this FUTURE event E2 didn't have any causal influence on E1, did it ? And I DID change the probability of outcome at E1 using this information from E2, which is outside of the PAST lightcone of E1.

 

[vanesch]

He was right about that. Decoherence doesn't change that conclusion. It only shows how hard it would be in practice to get those two branches of the wf to interfere again. But the cat is still both alive and dead.

But that's the whole point! The decoherence prohibits any "interaction between branches" (also called quantum interference), and from the moment you have that, you have an approximately classical system for each of the terms.

It *sounds* like you're suggesting that decoherence alone solves the measurement problem. That's not true, as I think you know. Decoherence has nothing to do with this -- what's really "solving" the problem for you here is your putting in a "consciousness token" according to the Born rule.

Yes, because the measurement problem addresses another issue: WHICH ONE of these emerging classical worlds will I perceive ?
It's like these stories about a photocopy of your body, and which one is "the original you" and so on.
THIS is something that is NOT answered by unitary dynamics, in the same way as it is not answered by the physics of the "body photocopying machine". My claim is that this issue is not part of physics per se anymore, but about the relationship with conscious perception.

(Decoherence merely makes this pseudo-well-defined by cleaning up the boundaries between the different possible branches.)

Yes, indeed.

Also, in regard to your last sentence, I thought you thought that in the other "phase space" [sic], there *isn't* another "me". Maybe there's a mindless hulk that has the same shape as me, but it isn't conscious, right?

I don't see why you should consider that those "other you's" aren't conscious. But I don't see why you should even consider the question: what do you care about *another* classical world where a similar construction as you is running around, whether it is conscious or not ?
No, the only thing is that you are the "original" you from before the split, and the other one is "the copy" - an assymetrical split which only makes sense from your subjective viewpoint, and which is completely symmetrical "from the outside".

Sure, but your theory forces me to say that when I "see a live cat" (or whichever) it's a delusion.

I don't see why you insist on saying that "what you see is a delusion". What you see is *classical* and you're aware of a *classical world*, and in that classical world THERE IS A LIVE CAT.
(there's also ANOTHER CLASSICAL WORLD, where ANOTHER YOU is seeing a dead cat, but it is a world that is now totally cut off from your world, so this is entirely equivalent as the "theory with two phase spaces" from the beginning, which you didn't object to).

 

[ttn]

But that's the whole point! The decoherence prohibits any "interaction between branches" (also called quantum interference), and from the moment you have that, you have an approximately classical system for each of the terms.

I'm not objecting to the "approximate classicalness" of "each term" -- I'm objecting to the fact that there are different terms (each of which represents something that is simultaneously happening to the same entities, and the different happenings are not consistent!).

No, the only thing is that you are the "original" you from before the split, and the other one is "the copy" - an assymetrical split which only makes sense from your subjective viewpoint, and which is completely symmetrical "from the outside".

I previously understood that in your version of MWI the "consciousness token" gets put in a branch according to the Born rule. So then there are empty (non-token-containing) branches, right? And the material copies of "you" in these other empty branches thus wouldn't be conscious, right?

I don't see why you insist on saying that "what you see is a delusion". What you see is *classical* and you're aware of a *classical world*, and in that classical world THERE IS A LIVE CAT.

But the *real* world is not much like that "classical world" you are aware of and which contains a live cat. Hence, delusion.

 

[vanesch]

I'm not objecting to the "approximate classicalness" of "each term" -- I'm objecting to the fact that there are different terms (each of which represents something that is simultaneously happening to the same entities, and the different happenings are not consistent!).

What I'm trying to point out is that there is maybe a subtlety in the concept of "the same entities". If an "entity" is "a state in a classical phase space", then these "same" entities are just "copies" if we have SEVERAL classical phase spaces. If "living cat" is a concept belonging to a classical phase space, then having generated two phase spaces means we now have two of these classical cats, one in each phase space. Of course, quantum-mechanically, it is "the same entity", but who says that what you intuitively call an entity (such as "living cat") is not a concept that only has meaning in a classical context ?
So - this was my point - if your theory GIVES RISE to several of these classical phase spaces, then you just have several of these entities around, and if ONE of these classical phase spaces corresponds to what you are classically used to, then that's good enough, no ? Whether or not they find their common origin in ANOTHER CONCEPT, which is a "quantum cat" is something else. As you've only seen CLASSICAL cats, you have no idea what is a quantum cat, and hence you cannot claim that it is silly to talk about "a quantum cat being live and dead": you only know about classical cats, and our theory gives us DIFFERENT classical cats, which are OR live OR dead. A quantum cat is then nothing else but a "generator of classical cats" in this respect.

I previously understood that in your version of MWI the "consciousness token" gets put in a branch according to the Born rule. So then there are empty (non-token-containing) branches, right? And the material copies of "you" in these other empty branches thus wouldn't be conscious, right?

This doesn't need to be the case: there was not THE consciousness token, there was MY consciousness token. Whether or not the others got "new ones", I left it out of the discussion, because it doesn't mean anything useful. If, in a classical world, I can already not find out whether another body is conscious or not (because behaviourally identical), why would I break my head over a COPY in a world that I cannot even behaviourally interact with ? Even classically, you have not to assume that "others" are conscious. One consciousness is enough to explain your subjective experiences, even purely classically. There's no need to assume that another consciousness exists apart from your own one - no matter whether we do quantum theory or classical theory. So I don't see the need to complicate the issue in the quantum context: I took the same viewpoint there.
If, however, classically, you assume that others are conscious "by analogy with yourself", well, you can do the same quantum-mechanically. Whatever behaves more or less "as if it is conscious" is then declared to be conscious. But this discussion doesn't matter, classically or quantum mechanically. The only thing that counts is what happens TO YOUR OWN conscious experience, if there are "copies" or "alternatives" - with WHICH ONE you are consciously associated ; because that WILL influence your subjective experiences. What eventually happens to others doesn't matter.

But the *real* world is not much like that "classical world" you are aware of and which contains a live cat. Hence, delusion.

Of course "delusion" in the sense that naive realism is not true, in that what we perceive with our senses is not ALL there is to the world. But not "delusion" in that there really IS a (part of) reality that corresponds to what you are aware of. That's not a big surprise, is it ?
You're bathing in a SEA of neutrinos and you've never seen them. There are more neutrinos around you than anything else but you're not seeing, feeling or hearing them. Are you deluded now ?

 

[ttn]

Ok, silly counter example:

Imagine that I throw a dice at event E1. The probability for the dice to give "2" is equal to 1/6, say, when we take into account all we know in its past lightcone. But consider now event E2, a bit in the FUTURE lightcone of E1 where I look at the outcome of the dice which was thrown a bit before E1. The outcome is "4".
If I know this, then the probability of event E1 to give "2" is not 1/6 anymore, but "0". Nevertheless, this FUTURE event E2 didn't have any causal influence on E1, did it ? And I DID change the probability of outcome at E1 using this information from E2, which is outside of the PAST lightcone of E1.

Good, I'm glad you raised this since I think it will allow me to clarify in an important way the definition of Bell Locality. No, I don't think this is a "counterexample." I think I must have used some phrase earlier like "outside the backwards light cone" that made you think the *forward* light cone was fair game for specifying this kind of additional information. But as your example makes clear, for any stochastic theory, conditionalizing on the causal effects of a given event will affect the probabilities assigned to those events. But this is all entirely beside the point, since the kind of information you're permitted to additionally specify (without changing the probability assigned to a given event, in a Bell Local theory) is information pertaining to a *space-like* separated event.

Here is Bell's statement: "A theory will be said to be locally causal [Bell Local] if the probabilities attached to values of local beables [i.e., whatever the theory is *about*, whatever it purports to provide a description of] in a space-time region 1 are unaltered by specification of values of local beables in a space-like separated region 2, when what happens in the backward light cone of 1 is already sufficiently specified, for example by a full specification of local beables in a space-time region 3 [that "fills" the backward light cone of 1 and shields 1 off from the backward light cone of 2]." This is from the article I mentioned before ("La Nouvelle Cuisine") and can be found on page 240 of the 2nd edition of Speakable & Unspeakable. You may also find the figures and some accompanying discussion in Section 2 of quant-ph/0601205.

I hope that clarifies.

 

[vanesch]

But this is all entirely beside the point, since the kind of information you're permitted to additionally specify (without changing the probability assigned to a given event, in a Bell Local theory) is information pertaining to a *space-like* separated event.

I know. But I wanted to illustrate, with a *SILLY* example, that "can influence the probability of something to happen" and "has a causal influence on" is NOT the same. The "naturalness" of Bell's statement comes from the tacit identification of both statements. We like to think that the only CAUSAL INFLUENCES on an event E "come from" its past light cone. And, if we are talked into accepting erroneously the identification of the above two statements, this translates into: "the PROBABILITY of something to happen at event E can only depend upon all we know in the past lightcone of this event." which is the essence of Bell's statement, of his definition of Bell locality. Of course Bell is thinking of "things happening at spacelike distance should not have a CAUSAL INFLUENCE on what's going on at event E" and he's translating this into "their knowledge should not alter the probability of what's happening at E". And of course he's aiming for spacelike distances.

But my silly example just wanted to show that KNOWLEDGE about something CAN alter probabilities of something to happen, without them influencing that something causally.
Indeed, although my KNOWLEDGE of the future event influences the probability of the current event (in a naive and trivial matter, even), you'll be hard-pressed to say that this FUTURE EVENT has a causal influence on the CURRENT event.

So it is not true, in general, that "knowledge about stuff alters the probability of outcomes at E" means that "stuff" has a causal influence on E.
And now we're home, because, IF IT ISN'T TRUE that KNOWLEDGE OF STUFF CHANGES PROBABILITIES OF EVENT E, then there is NO REASON to Bell's definition of locality. Indeed, if it is not in general true that the knowledge of some outcome means that there is a causal influence by this outcome onto my event, then (what we want) the requirement of only causal influences from the past lightcone does NOT imply that extra knowledge outside of this cone should not influence probabilities of my event at E.

 

[ttn]

What I'm trying to point out is that there is maybe a subtlety in the concept of "the same entities". If an "entity" is "a state in a classical phase space", then these "same" entities are just "copies" if we have SEVERAL classical phase spaces. If "living cat" is a concept belonging to a classical phase space, then having generated two phase spaces means we now have two of these classical cats, one in each phase space. Of course, quantum-mechanically, it is "the same entity", but who says that what you intuitively call an entity (such as "living cat") is not a concept that only has meaning in a classical context ?
So - this was my point - if your theory GIVES RISE to several of these classical phase spaces, then you just have several of these entities around, and if ONE of these classical phase spaces corresponds to what you are classically used to, then that's good enough, no ? Whether or not they find their common origin in ANOTHER CONCEPT, which is a "quantum cat" is something else. As you've only seen CLASSICAL cats, you have no idea what is a quantum cat, and hence you cannot claim that it is silly to talk about "a quantum cat being live and dead": you only know about classical cats, and our theory gives us DIFFERENT classical cats, which are OR live OR dead. A quantum cat is then nothing else but a "generator of classical cats" in this respect.

I get all of this. I just don't understand what you think any of it has to do with refuting my claim that, according to MWI, we're deluded when we look at a cat and perceive that it is alive. The truth is that there is, as you say, a "quantum cat" which is in a big entangled superposition of alive and dead. This simply does not match with my direct perceptual experience. So what I come to believe based on that experience does not *correspond* to the real (quantum) state of the cat. My belief is *false*. I am *deluded*.

I suppose you want to go back to this old argument that it's not really a delusion, but merely a true belief about some one *part* of the real world. And that takes us back to the old debate about whether the word "part" is really appropriate here. You think it is; I think it isn't. And I suppose this is what you meant when you said maybe there was a "subtlety in the concept of the same entities." Frankly, though, that kind of statement alone is enough to make me reject this whole theory as not serious. It's like when Bill Clinton starts saying "it depends on what the definition of the word 'is' is", you know (as I think Griffiths says in one of his texts in a slightly different context) you should hold onto your wallet. =)

This doesn't need to be the case: there was not THE consciousness token, there was MY consciousness token. Whether or not the others got "new ones", I left it out of the discussion, because it doesn't mean anything useful.

But then I don't think this version of the theory is coherent. If all the copies of you that the splittings generate all get consciousness tokens, then what is the meaning of the probability associated with the Born rule? There's now no one unique happening for those probabilities to be probabilities *of*. If your body splits into 10 copies and they all, with certainty, are conscious of the material surroundings in their branch, then what possible meaning can it have to associate some number like 37% with one or the other of the copies?

Also, a slightly different point for the benefit of lurkers, if you were to give each human being his own "consciousness token" obeying the Born rule, in a very short period of time, the odds are spectacularly good that no two consciousness tokens will inhabit the same branch. So what you consciously think are other sentient beings in the world, are in fact mere mindless hulks. (That is David Albert's term, and his point actually.) Call that one more delusion.

If, in a classical world, I can already not find out whether another body is conscious or not (because behaviourally identical), why would I break my head over a COPY in a world that I cannot even behaviourally interact with ? Even classically, you have not to assume that "others" are conscious. One consciousness is enough to explain your subjective experiences, even purely classically.

I'm sorry, but this is silly. There is good empirical evidence that other people are conscious, and you don't have to know Newtonian mechanics (not to mention later more advanced physics) to know this. Your point, that the non-consciousness of other people is consistent with classical physics, is about as relevant and interesting as the point that the Earth being flat is consistent with classical physics. Sure, but who cares, since we know it isn't true? The interesting point is that other people in fact *being* conscious *is* perfectly consistent with classical physics. If it weren't, since we know for sure that other people are conscious way before we get to advanced things like classical physics, we'd rationally have to *reject* classical physics. Happily, there's no need to do that, though, since there's no conflict.

With MWI on the other hand, we *do* have this conflict. If you accept MWI, you have to accept that what you erroneously took to be conscious-others are in fact mindless hulks. I call that good reason to reject MWI. It contradicts basic empirically grounded knowledge (and then, in order to escape this problem, spins a fantasy brain-in-vat scenario about how all of our earlier empirically grounded knowledge is delusional).

There's no need to assume that another consciousness exists apart from your own one - no matter whether we do quantum theory or classical theory.

Sure, this isn't an issue that really comes up in physics. But that doesn't mean it isn't 100% settled *prior* to doing physics (classical, quantum, or otherwise).

If, however, classically, you assume that others are conscious "by analogy with yourself", well, you can do the same quantum-mechanically. Whatever behaves more or less "as if it is conscious" is then declared to be conscious.

Ooh, interesting... so maybe the one last final thing we're deluded about is that we ourselves are conscious! (Please note how self-refuting such a claim would be.)

But this discussion doesn't matter, classically or quantum mechanically. The only thing that counts is what happens TO YOUR OWN conscious experience, ...

But see, to me, that is *not* "the only thing that counts. I actually believe that knowledge is hierarchical, and that we have to regard the more basic stuff as largely settled. If a scientific theory requires me to accept that *everything* I believed before (based on direct perception and low-level inference therefrom) is a delusion, I reject it.

Of course "delusion" in the sense that naive realism is not true, in that what we perceive with our senses is not ALL there is to the world. But not "delusion" in that there really IS a (part of) reality that corresponds to what you are aware of. That's not a big surprise, is it ?
You're bathing in a SEA of neutrinos and you've never seen them. There are more neutrinos around you than anything else but you're not seeing, feeling or hearing them. Are you deluded now ?

No, because I don't have any direct perceptual knowledge of those neutrinos (including their non-existence). Of course there are all sorts of facts I am not aware of. The point is, the ones I *am* aware of, I'm *aware* of. If I see a living cat, there might be neutrinos flying through it or Martians dancing jigs on Venus or who knows what else out there somewhere, but *there is a living cat*. MWI asks me to accept not that there's more facts out there in the world than that I perceive, but that what I perceive is a delusion. It really is different.

 

[ttn]

I know. But I wanted to illustrate, with a *SILLY* example, that "can influence the probability of something to happen" and "has a causal influence on" is NOT the same. The "naturalness" of Bell's statement comes from the tacit identification of both statements. We like to think that the only CAUSAL INFLUENCES on an event E "come from" its past light cone. And, if we are talked into accepting erroneously the identification of the above two statements, this translates into: "the PROBABILITY of something to happen at event E can only depend upon all we know in the past lightcone of this event." which is the essence of Bell's statement, of his definition of Bell locality.

No, I think you may have missed something crucial. The statement is *not* about "all we know in the past lightcone of this event." If it were merely that, it would be obviously possible to say "we didn't know enough", in which case the subjective probability of a given event *could* change, when some spacelike information is specified, but *without* any relativity-violating causal influence.

But "Bell Locality" is not a statement about subjective probabilities. It is a statement about a theory's *dynamics*. And the thing we have to specify before we can apply the criterion is not "all we know in the past lightcone of this event" but rather "all that *exists* in the past lightcone of this event." That is, Bell Locality requires a *complete specification of the actual state* in the past light cone.

How, you ask, could we ever in practice *know* whether or not we had specified the states *completely*? Not to worry -- this is what *theories* do for us. A theory (by definition, here) is something which proposes some kind of candidate complete description of states. We may then take the theory at its word and apply the criterion to find out if it is local. This of course doesn't mean the theory is or isn't *true*. That's a different question. But *given* some candidate for a "complete specification of states" (and a candidate for the dynamics, which allows us to calculate probabilities for events) -- i.e., given a proposed *theory* -- we can ask: "is the theory causally local?". That is what the Bell Locality criterion does.

Of course Bell is thinking of "things happening at spacelike distance should not have a CAUSAL INFLUENCE on what's going on at event E" and he's translating this into "their knowledge should not alter the probability of what's happening at E". And of course he's aiming for spacelike distances.

I don't know exactly what you mean by "their knowledge", but this has *nothing* to do with anybody's knowledge. The criterion applies to *theories* and the probabilities involved are simply a way of talking about the *dynamics* of the theories. What any person does or does not *know* is completely and totally irrelevant.

But my silly example just wanted to show that KNOWLEDGE about something CAN alter probabilities of something to happen, without them influencing that something causally.

If you have a *theory* whose *dynamics* has probabilities depending on somebody's knowledge, I would say that is a very strange theory indeed. But even this is no objection to Bell Locality, unless you want to insist that the probabilties depend on the knowledge that someone acquires at spacelike separation and that the assigned probabilities change when this is specified *and* you want to insist that such dependence is consistent with relativity's prohibition on superluminal causation!

So it is not true, in general, that "knowledge about stuff alters the probability of outcomes at E" means that "stuff" has a causal influence on E.

Absolutely right. But irrelevant since Bell Locality isn't about subjective probabilities; knowledge just doesn't enter into it one way or the other.

And now we're home, because, IF IT ISN'T TRUE that KNOWLEDGE OF STUFF CHANGES PROBABILITIES OF EVENT E, then there is NO REASON to Bell's definition of locality.

With all due respect (and it is considerable), you have not appreciated Bell's definition. He's already understood and eluded your worry.

 

[Hurkyl]

I just don't understand what you think any of it has to do with refuting my claim that, according to MWI, we're deluded when we look at a cat and perceive that it is alive. The truth is that there is, as you say, a "quantum cat" which is in a big entangled superposition of alive and dead. This simply does not match with my direct perceptual experience. So what I come to believe based on that experience does not *correspond* to the real (quantum) state of the cat. My belief is *false*. I am *deluded*.

There's (at least) one different interpretation: that when you have a quantum cat, we can say with certainty that it either looks alive or it looks dead: that's simply how quantum things look!

 

[ttn]

There's (at least) one different interpretation: that when you have a quantum cat, we can say with certainty that it either looks alive or it looks dead: that's simply how quantum things look!

Sure, but does that really help? So the way X looks is fundamentally in conflict with the way X is. The perceptual experience doesn't correspond to the facts. So the perceptual experience is delusional.

 

[vanesch]

But then I don't think this version of the theory is coherent. If all the copies of you that the splittings generate all get consciousness tokens, then what is the meaning of the probability associated with the Born rule? There's now no one unique happening for those probabilities to be probabilities *of*. If your body splits into 10 copies and they all, with certainty, are conscious of the material surroundings in their branch, then what possible meaning can it have to associate some number like 37% with one or the other of the copies?

YOUR experience. If there are 10 copies of your body, you might assume that ONE of these "is you" and the others are "copies". The probability for a body to be *your* experience, and not a copy's experience, is given by the Born rule. This doesn't say anything about conscious experiences those other copies might, or might not have.

Now, I have to point out that I'm "heretic" here with respect to most MWI-ers. Most MWI-ers have the hope of establishing, in one way or another, the Born rule "emerge" from an "uniform distribution", or world counting or whatever. What I think I've established is that this includes AT LEAST one extra postulate, and if the goal of that postulate is to allow us to say that FINALLY what we will be consciously aware of is given by the Born rule, we might as well take the shortcut and postulate INITIALLY this Born rule.
There are schemes in the making which hope to establish the "standard" MWI programme though. I'm thinking for instance about Robin Hanson's "mangled worlds" (it's on the arxiv), which, for short, establishes that these classical worlds only occur for a certain "time of stability" for those branches which are bigger than a certain lower hilbert norm (relative to the biggest ones). So if you postulate that you need a time of stability (say, of the order of at least a nanosecond) in order to be "eligible" as "classical world", and with this cutoff, you do "world counting", then he has indications that you arrive at something very close to the Born rule.
My point is simply that you don't have to go through all these pains. Just POSTULATE that what you experience, in the end, is a world drawn according to the Born rule, and that's good enough.

Also, a slightly different point for the benefit of lurkers, if you were to give each human being his own "consciousness token" obeying the Born rule, in a very short period of time, the odds are spectacularly good that no two consciousness tokens will inhabit the same branch. So what you consciously think are other sentient beings in the world, are in fact mere mindless hulks. (That is David Albert's term, and his point actually.) Call that one more delusion.

Well, if I were to give each human and each cat his own consciousness token (of which, I repeat, I don't see the utility...), then this token's "voyage" throughout the successive quantum states (using the Born rule) will indicate what they will experience ; in which successive classical worlds they will be. But there's no problem with the CREATION of NEW tokens in the other worlds, and for the "copies" of others in the worlds visited by the "original" consciousnesses.

I'm sorry, but this is silly. There is good empirical evidence that other people are conscious

Sorry, but that's impossible. There can only be BEHAVIOURAL empirical evidence, and behaviourally you cannot find out if there is, or if there isn't, a subjective experience "lived" by the BODY under study.

The interesting point is that other people in fact *being* conscious *is* perfectly consistent with classical physics. If it weren't, since we know for sure that other people are conscious way before we get to advanced things like classical physics, we'd rationally have to *reject* classical physics. Happily, there's no need to do that, though, since there's no conflict.

Apparently you didn't grasp the "hard problem of consciousness" then. No physical observation can distinguish between a physical process with which there goes a subjective experience, and an identical physical process which is not subjectively experienced. So consciousness of others will never be a falsification or a support for a physical, or scientific theory.

The only reason why we THINK that others are conscious is "by analogy to ourselves". How do you determine whether stones are conscious or not ?

With MWI on the other hand, we *do* have this conflict. If you accept MWI, you have to accept that what you erroneously took to be conscious-others are in fact mindless hulks.

Not at all. I prefer to think of them as "mindless hulks" in the same way I can ALREADY assume that others are "mindless hulks" ; quantum theory or not! But you can have it your way if you want to, and assign conscious experience to anything that classically breaths or not. This doesn't change, in the slightest bit, any physically observable fact by YOU.

For instance, you cannot know whether my body is conscious. Only *I* can know that. You can only perceive actions of my body, which could ultimately be explained by the physics of it ; by neurological and physiological processes. If you knew all that, you would be perfectly able to explain all my body does, says and writes on PF.

In the same way, I cannot know whether YOUR body is conscious.

Now, because I know that my body is conscious, and maybe you know that yours is, and because we see behavioural similarities, by ANALOGY and association, we take it that we're both conscious. But this will remain for ever a hypothesis. And from the moment physical processes are behaviourally totally different, and this basis of analogy disappears, WE HAVE NO MEANS at all to establish ever, whether this physical process is conscious or not. Hence your impossibility to establish whether stones are conscious. Unless by convention.

Ooh, interesting... so maybe the one last final thing we're deluded about is that we ourselves are conscious! (Please note how self-refuting such a claim would be.)

No, that will never be the case. It is the ONLY thing we know for sure.

No, because I don't have any direct perceptual knowledge of those neutrinos (including their non-existence). Of course there are all sorts of facts I am not aware of. The point is, the ones I *am* aware of, I'm *aware* of. If I see a living cat, there might be neutrinos flying through it or Martians dancing jigs on Venus or who knows what else out there somewhere, but *there is a living cat*. MWI asks me to accept not that there's more facts out there in the world than that I perceive, but that what I perceive is a delusion. It really is different.

No, MWI asks you to consider that, next to this living classical cat, there might be another world in which there is another classical dead cat.
To me, this is not so different from the neutrinos.

 

[DrChinese]

Part One: It is an empirical fact that whenever Alice and Bob measure along the same axis, their results are *perfectly (anti-) correlated*. We then ask: how must a Bell Local theory work if it is going to successfully predict this empirical fact? The answer turns out to be: its state descriptions must include variables which, in effect, pre-assign measurement outcomes for all spin components. Such variables are often called "local hidden variables" or some such.

Why does a Bell local theory need to be realistic to explain the correlations? You might want to prove that requirement. I know it seems like that is the only possibility to you, but it won't seem that way to everyone.

 

[LnGrrrR]

Vanesch,

This question is a bit off-the-wall, but I feel the need to ask it anyways.

The many-worlds question posits the idea that we're seeing only one of the 'outcomes' of waveform collapse, and other realities see the other results, correct?

Decoherence occurs though (and I'm describing this in the least scientific way I can) when there are too many 'waveform paths', so that forms function in a 'classical' sense.

Here's the question...is there a possibility of a universe where all the waveforms could 'sync up' in an object, thereby allowing a macroscopic object to behave how QM says it should?

 

[LnGrrrR]

Also, a slightly different point for the benefit of lurkers, if you were to give each human being his own "consciousness token" obeying the Born rule, in a very short period of time, the odds are spectacularly good that no two consciousness tokens will inhabit the same branch. So what you consciously think are other sentient beings in the world, are in fact mere mindless hulks. (That is David Albert's term, and his point actually.) Call that one more delusion.

I'm sorry, but this is silly. There is good empirical evidence that other people are conscious, and you don't have to know Newtonian mechanics (not to mention later more advanced physics) to know this. Your point, that the non-consciousness of other people is consistent with classical physics, is about as relevant and interesting as the point that the Earth being flat is consistent with classical physics. Sure, but who cares, since we know it isn't true? The interesting point is that other people in fact *being* conscious *is* perfectly consistent with classical physics. If it weren't, since we know for sure that other people are conscious way before we get to advanced things like classical physics, we'd rationally have to *reject* classical physics. Happily, there's no need to do that, though, since there's no conflict.

With MWI on the other hand, we *do* have this conflict. If you accept MWI, you have to accept that what you erroneously took to be conscious-others are in fact mindless hulks. I call that good reason to reject MWI. It contradicts basic empirically grounded knowledge (and then, in order to escape this problem, spins a fantasy brain-in-vat scenario about how all of our earlier empirically grounded knowledge is delusional).

I have a friend who takes this stance, and the question I usually ask him, "Well, how do you know if you're actually 'real' then?" His response tends to be 'just cause'.

 

[vanesch]

With all due respect (and it is considerable), you have not appreciated Bell's definition. He's already understood and eluded your worry.

Maybe. It is always difficult to respond to an "accusation" of not having understood something and being deluded into thinking one did. The only answer I can think of is by saying exactly what I understand of Bell.

The way I understand Bell is the following. I know you see it larger, but I think that there are hidden assumptions in the enlargement. Here it goes. I think that Bell considered (consciously or not), a DETERMINISTIC evolution and considered that all of our probabilistic considerations resulted from us not having full knowledge of the fully deterministic description of nature - whether this impossibility was practical or in principle.
Given a deterministic evolution, and a requirement of non-causality between spacelike separated events, which means that no decision taken at one event U (like, deciding in which direction to put your polarizer) can CHANGE the full state in the backward lightcone of an event E, and hence not change the deterministic outcome of all thinkable measurements we can do at E, the probability of our outcome at E cannot depend on anything EXTRA that was produced at U. Of course, there can be a dependence on the outcome at U, BUT ONLY IN SUCH A MEASURE THAT this U was influenced by the SAME "beables" as those that were IN ANY CASE going to determine what was going to be the result at E.
In other words, the only correlation that can be seen between the outcomes at U and the outcomes at E are those of "common origin" - and this for ALL POTENTIAL OUTCOMES OF ALL THINKABLE MEASUREMENTS. And this is clear if the "beables" in the past lightcone of E ALREADY ENTIRELY FIXED what was going to happen at E, in all possible and thinkable measurements we can decide about at E (and at U).

So the fundamental hypothesis is that, for all thinkable outcomes at U and at E, the outcomes are ALREADY DETERMINISTICALLY FIXED by what happened in their past lightcones. They only APPEAR stochastical to us because we lack certain information about this fully deterministic initial state back in the past lightcone of E and of U. And as such, IT IS PLAUSIBLE TO POSTULATE A PROBABILITY DISTRIBUTION over these initial deterministic states, from which ALL PROBABILITIES OF ACTUAL AND POTENTIAL MEASUREMENTS at U and E can be derived ; and the only correlations we can find between U and E are those generated by this postulated common probability distribution, which has to have in it, all the possible outcomes we COULD obtain potentially.

The essence of Bell's theorem is the existence of an overall probability distribution of actual and potential measurement outcomes at E and U - and to me this makes perfect sense ONLY in the case of a deterministic evolution but with partly unknown initial states. In fact, in this case, Bell makes really perfectly sense: given the fact that the outcome is ENTIRELY DETERMINED by the "hidden state" in the past lightcone of E, its "probability of giving a result" only depends on the probability we assigned to the states that will (deterministically, with 100% certainty) evolve in this result. All correlations with things which cannot influence this evolution (which are hence outside of the past lightcone) are then purely correlations due to "common origin", and which are hence generated by *the same probability distribution of those initial states*.

However, in the case of *stochastic* evolution, meaning, there's NOTHING IN NATURE, no hidden state or anything, which "determines" what's going to happen, but "it just happens" one way or another, there's no way to IMPOSE how this randomness should occur. Of course, signal locality imposes a certain constraint in the link between CHOICES MADE at an event U and RESULTS OBTAINED at an event E at spacelike distance from U, but there's no a priori way to REQUIRE any other constraint on the inherent randomness of nature. In the same way that in a strictly stochastic theory, the randomness of individual events is *unexplained*, in the same way *correlations* can be unexplained.
In a strictly stochastic theory, THERE NEEDS NOT TO BE a common probability distribution which describes all POTENTIAL outcomes at U and E: there only needs to be a distribution of the ACTUAL outcomes as a function of the decision taken to measure this or that. As there is, by definition, NO DYNAMICAL EXPLANATION for this randomness (because if there were, we would have a DETERMINISTIC theory!), there can be no requirement on this randomness, and there can be no speculation about what WOULD have happened if we did a measurement we didn't perform: that's only something we can reasonably think about if we know a DETERMINISTIC mechanism on how the randomness comes about.
So there's no reason, in this case, to impose the same requirements as in the case of a deterministic evolution of a prior existing common probability distribution of all potential and actual measurement outcomes (in other words, to impose Bell locality).

But of course the temptation is great to turn a stochastic theory into a deterministic one with a "hidden state" in some way, so that the APPARENT stochasticity is IN FACT resulting from an underlying deterministic evolution, but of which we can only (in principle or in practice) determine in a probabilistic way the initial state. In THIS case, of course, Bell makes sense again.

So I come back to my old claim that Bell locality is nothing else but "signal locality on a principal level" and "a requirement of fundamental determinism".

This is how *I* understand Bell's theorem.

Now, this reflection (the underlying assumption of fundamental determinism) is in fact no problem, nor for classical physics, nor for MWI style quantum theory, nor for Bohmian mechanics: they are ALL deterministic theories.
However, the reflection IS relevant to "algorithmic" interpretations of QM: if you take it that the QM "algorithm" is just calculating "probabilities of outcomes" as in a fundamentally STOCHASTICAL theory, then you cannot make the Bell assumption. The only thing you can conclude is that it won't have an underlying deterministic theory which is local in a fundamental sense.Again, maybe I missed its content.

 

[LnGrrrR]

Vanesch,

I'm a bit confused on your consciousness idea. Is it your idea that mindless hulks CAN exist in your MWI? Or can they not?

If they CAN exist, how do you know whether or not you are conscious?

 

[vanesch]

Vanesch,
I'm a bit confused on your consciousness idea. Is it your idea that mindless hulks CAN exist in your MWI? Or can they not?

It has nothing to do with MWI or anything. When I first wrote about it, I assumed some knowledge about "the hard problem of consciousness".

Here's a quote from the Internet Encyclopedia of Philosophy:
http://www.iep.utm.edu/c/consciou.htm

Joseph Levine (1983) coined the expression “the explanatory gap” to express a difficulty for any materialistic attempt to explain consciousness. Although not concerned to reject the metaphysics of materialism, Levine gives eloquent expression to the idea that there is a key gap in our ability to explain the connection between phenomenal properties and brain properties (see also Levine 1993, 2001). The basic problem is that it is, at least at present, very difficult for us to understand the relationship between brain properties and phenomenal properties in any explanatory satisfying way, especially given the fact that it seems possible for one to be present without the other. There is an odd kind of arbitrariness involved: Why or how does some particular brain process produce that particular taste or visual sensation? It is difficult to see any real explanatory connection between specific conscious states and brain states in a way that explains just how or why the former are identical with the latter. There is therefore an explanatory gap between the physical and mental. Levine argues that this difficulty in explaining consciousness is unique; that is, we do not have similar worries about other scientific identities, such as that “water is H2O” or that “heat is mean molecular kinetic energy.” There is “an important sense in which we can’t really understand how [materialism] could be true.” (2001: 68)

David Chalmers (1995) has articulated a similar worry by using the catchy phrase “the hard problem of consciousness,” which basically refers to the difficulty of explaining just how physical processes in the brain give rise to subjective conscious experiences. The “really hard problem is the problem of experience…How can we explain why there is something it is like to entertain a mental image, or to experience an emotion?” (1995: 201) Others have made similar points, as Chalmers acknowledges, but reference to the phrase “the hard problem” has now become commonplace in the literature. Unlike Levine, however, Chalmers is much more inclined to draw anti-materialist metaphysical conclusions from these and other considerations. Chalmers usefully distinguishes the hard problem of consciousness from what he calls the (relatively) “easy problems” of consciousness, such as the ability to discriminate and categorize stimuli, the ability of a cognitive system to access its own internal states, and the difference between wakefulness and sleep. The easy problems generally have more to do with the functions of consciousness, but Chalmers urges that solving them does not touch the hard problem of phenomenal consciousness. Most philosophers, according to Chalmers, are really only addressing the easy problems, perhaps merely with something like Block’s “access consciousness” in mind. Their theories ignore phenomenal consciousness.

In other words, physical theories NEVER say anything about what's called "phenomenal consciousness" in the above text, nor does neurobiology or any other scientific discipline.

As such, one could make it easy, and just say there's only one real consciousness, which is mine (yours). It's the simplest assumption. I can even "empirically prove it" I stick a pin in my leg, and I feel pain. so there's some consciousness involved here, because I feel it. I now stick a pin in YOUR leg, and I don't feel anything. Hence, there's no consciousness involved. QED

Ok, you'll tell me that you'll scream and shout and hit me in the face, but this is a PHYSICAL REACTION I could perfectly well analyse if I knew the physics of your body well enough. I could follow some nerve pulses going from your leg to your brain, I could follow the processing in your brain, resulting in nerve pulses going to your throat and lungs resulting in vibrating air which I hear, and pulses going to your arm and fist hitting me in my face. This would be "system analysis" in the same way I could analyse how programmable logic circuits would react to certain stimuli.

So, in a certain respect, it is simpler to consider that there is only one conscious being (me), and one subjective experience (mine) and all the rest is just physics. This has nothing per se to do with quantum theory or classical physics. It's the easiest way to reason. After all, I'm not aware of any OTHER subjective experience apart from mine.

But then, by analogy, I could accept the hypothesis that SIMILAR bodystates also relate to *other* conscious experiences. I can never know for sure of course, but it sounds acceptable, somehow, that other people might be also having a subjective experience of their own.
Well, then, repeat the explanation from their point of view !

Now, it also seems that I derive my conscious experience from a certain part of a physical world (called, my body state). And this is what I do, then: I try to explain from WHICH bodystate I'm deriving this subjective experience. In classical physics, it derives from certain dynamical states of matter (corresponding to body states). It's not because there's such a state somewhere that *I* experience this state. I only experience ONE such state. I don't know why, but it seems to be so. I experience the state of some matter which I call "my body", and I don't seem to experience the state of something very similar, which is "your body". I don't really have an explanation for this. I experience certain matter states, and not some other, similar, matter states. Don't know why.

Now, in MWI quantum physics, this is about the same. But now, I seem to experience not the FULL state of some material structure, but only ONE of its terms. So be it. I was there already in classical physics: I couldn't understand why I experienced certain matter states (body 1, also called "my body"), and not others which were very similar (body 2, also called "your body"). In classical physics, we then said that body 2 was maybe experienced by ANOTHER consciousness (namely you). Or maybe not. Who knows. So be it.
Well, I do the same in MWI: I seem to experience only ONE term of a body state. We now say that those other states are maybe experienced by OTHER consciousnesses ("copies of me"). Or maybe not. Who knows. So be it.

There's a difference, though. In classical physics, once I've accepted that I'm "related to a certain matter state called "my body"", I KEEP to this matter state, which evolves classically and continuously. So once the choice is made, because of the continuity of evolution in classical physics, I *stick* with the same state which is evolving. In MWI quantum theory, this is not true: sometimes the state from which I derive my subjective experiences SPLITS in several "states from which we can derive a conscious experience". And then, there's a choice to be made which will be my "new" matter state to which my subjective experiences "stick". THIS is the new thing, and this is then determined by the Born rule. Why ? Don't know. In the same way I already didn't understand why I experienced my body and not yours, I don't know why I experience THIS body state, and not that one.

But the important part in all this is this: to derive what *I* experience, it doesn't make any difference whether I postulate, or not, whether other similar states are experienced by other consciousnesses.


As to your question:

If they CAN exist, how do you know whether or not you are conscious?

Ask Descartes :-) It is the ONLY thing that is absolutely knowable.
All the rest is hypothesis.

 

[ttn]

Why does a Bell local theory need to be realistic to explain the correlations? You might want to prove that requirement. I know it seems like that is the only possibility to you, but it won't seem that way to everyone.

First a terminological point: I'd urge you not to use the term "realistic." I know you just meant it as a one-word summary of what I claimed in the earlier post empirically-adequate Bell Local theories had to be like. But the particular word "realist" has such a whole slew of other meanings, it's dangerous to use it here for this specialized purpose; you run the risk of forgetting later that you *didn't* mean one of those other possible meanings of the word. But... I won't lose sleep over terminology if you insist.

In regard to the substantive question (how to prove that a Bell Local theory which successfully accounts for the perfect anti-correlation of outcomes when Alice/Bob measure along the same axis, must possesses this particular claimed structure) would it be a turn-off if I just referred you to section 3 of

http://www.arxiv.org/abs/quant-ph/0601205

? The formal proof is there. I could try to explain it in less formal terms here instead, but that seems pointless when the more careful presentation already exists. Of course I'd be delighted to entertain questions/objections about that section of the paper.

 

[ttn]

However, in the case of *stochastic* evolution, meaning, there's NOTHING IN NATURE, no hidden state or anything, which "determines" what's going to happen, but "it just happens" one way or another, there's no way to IMPOSE how this randomness should occur. Of course, signal locality imposes a certain constraint in the link between CHOICES MADE at an event U and RESULTS OBTAINED at an event E at spacelike distance from U, but there's no a priori way to REQUIRE any other constraint on the inherent randomness of nature.

I think this whole issue/objection comes down to this. What should relativity require for a non-deterministic theory?

In the same way that in a strictly stochastic theory, the randomness of individual events is *unexplained*, in the same way *correlations* can be unexplained.

Yes, that's precisely the issue. Do we allow (as consistent with relativity) that irreducibly-random events at spacelike separations should nevertheless demonstrate persistent correlations? For example, Alice and Bob are very far apart and independently flip coins (and in a universe where the outcomes aren't based on any micro-hidden-details, but are genuinely irreducibly random) but magically always get the same outcomes (both H or both T).

Is such a thing consistent with relativity? I say "no" and am thus not at all bothered (but rather relieved) that such a scenario violates Bell's local causality requirement. But I am admittedly at a loss as to what to say next...

Maybe it would be useful to ask: could anyone think of a Lorentz invariant candidate toy theory that would predict the "both H or both T" example above? The collapse postulate of your "algorithmic" orthodox QM is surely not going to work here, right?

 

[Hurkyl]

Sure, but does that really help? So the way X looks is fundamentally in conflict with the way X is. The perceptual experience doesn't correspond to the facts. So the perceptual experience is delusional.

Why? What is the conflict?

Can you explain why there's a confict in a quantum cat looking as if it was alive or looking as if it was dead?

I rather suspect that you have simply developed a bias based on the fact you first learned that classical cats appear as if they are alive or as if they were dead, and so you're conditioned to believe that cats that appear as if they are alive or as if they were dead must be classical.

 

[Hurkyl]

Maybe it would be useful to ask: could anyone think of a Lorentz invariant candidate toy theory that would predict the "both H or both T" example above? The collapse postulate of your "algorithmic" orthodox QM is surely not going to work here, right?

It seems easy enough.

There are two purely random coins in the universe: A and B, and each may be flipped once. (For simplicity)

The joint distribution on the flips is given by:
P(HH) = 1/2
P(HT) = 0
P(TH) = 0
P(TT) = 1/2

I've not postulated a state of the universe, or any sort of collapse. I've simply posited the existence of a pair of random coins, and stated their joint probability distribution.

This is manifestly Lorentz-invarient, since it makes absolutely no reference to space-time whatsoever.


The only thing that's violated is, when the coins are flipped at space-like separated times, that the joint probability distribution violates the mathematical criterion for statistical independence.

 

[LnGrrrR]

Vanesch,

HEhe. I thought you might bring Descartes up...I'm a very firm believer in that statement. :)

Again, (and not trying to go too far off track) it depends on how we define consciousness. For instance, are you a complete materialist? Do you believe that every 'thought' that a person has involves a physical correlation? Is our thinking something that is 'on its own', or is merely a response to some other first cause that has so many variable it is considered to be practically 'unknown'?

For instance, I am at an ice cream stand, deciding what I want to get. Do I pick cherry because I have a 'mind' that says, "I want cherry" and then translates it into the physical act of choosing? Or is that act determined by genetic and environment, and my 'deciding' is only an illusion? Both answers have their pitfalls. :)

http://en.wikipedia.org/wiki/Philosophy_of_the_mind

Ps. Personally, I believe in the idea that physical objects arranged in a certain way can produce 'mental' properties.

The problem then becomes, "How do you know you're conscious?" If you say, "I know because I am thinking", then certainly, others around also seem to have 'consciousness' because they also think and respond to you.

 

[ttn]

It seems easy enough.

There are two purely random coins in the universe: A and B, and each may be flipped once. (For simplicity)

The joint distribution on the flips is given by:
P(HH) = 1/2
P(HT) = 0
P(TH) = 0
P(TT) = 1/2

I've not postulated a state of the universe, or any sort of collapse. I've simply posited the existence of a pair of random coins, and stated their joint probability distribution.

But you have simply asserted a summary of what is (taken to be) observed in the lab (so to speak). This is not a *theory*. *Of course* if you just say "X happens half the time" (etc.) there is no problem with locality or Lorentz invariance because you haven't actually *said* anything about anything. A theory (as I've defined it above in the context of defining Bell locality) is some candidate description of the physical world which includes some formalism that allows you to calculate the probabilities of possible happenings.

So I should reformulate the question: can anyone think of a *theory* which explains the coin-flipping results in a local or lorentz invariant way? Just repeating back what those results are, doesn't cut it.

 

[vanesch]

So I should reformulate the question: can anyone think of a *theory* which explains the coin-flipping results in a local or lorentz invariant way? Just repeating back what those results are, doesn't cut it.

I think you are here thinking of "a mechanism that can explain the randomness"... and as such should not be intrinsically random itself ; in other words, a deterministic mechanism.

What's wrong with say, have "stochastical" classical mechanics, where you have to add, at spacelike separated events, intrinsically random variables to the local equations of motion, but in such a way that these random variables are correlated ?

Of course, you can not find any DETERMINISTIC mechanism that can explain this correlation from an underlying "deterministic mechanics with lack of knowledge of the initial state" ; but if the intrinsically random variables are FUNDAMENTAL ? If you have no underlying mechanism, how are you going to require any statistical independence ?

As I tried to outline already a few times, the requirements you want to impose onto these intrinsic random variables are exactly such that they CAN be explained by an underlying deterministic theory where the randomness is a pure result from "ignorance of initial conditions".

 

[vanesch]

I would like to avoid too much delving into these philosophical issues for there own sake. I'm often forced into them here, because discussions around MWI usually start with misconceptions about those issues, which play a role in these views - and often lead to a "premature" rejection of an MWI view induced by conceptual errors in the philosophical view to adopt. I try hard to limit myself to what's relevant for the case at hand.

EDIT: oh, yes, and the reason why I "defend" the MWI view (or at least, fight the statement that it is a totally ridiculous view) is exactly the subject matter of this thread: the peaceful coexistence of relativity and Bell-type quantum phenomena with an ontological picture.

But you bring up of course a point which is of interest...

For instance, are you a complete materialist? Do you believe that every 'thought' that a person has involves a physical correlation?

I'm not a complete materialist, but something in between a property dualist http://en.wikipedia.org/wiki/Property_dualism and an epiphenomalist http://en.wikipedia.org/wiki/Epiphenomenalism.

Is our thinking something that is 'on its own', or is merely a response to some other first cause that has so many variable it is considered to be practically 'unknown'?

I think that our thinking is something that *finds it origin* in the physics, but that the (passive) subjective experience ITSELF is not purely part of physics as we know it. However, I don't think that this experience is active, in that it can intervene in the physical world. I think there is a physical phenomenon which makes my body utter the words "give me some cherry" ; but that the subjective experience that goes with it is generated by this body physics. My subjective experience passively undergoes what my body physics is doing.
This is why I think that behavioralism is missing the issue, because the (physical, observational) behaviour of a body is no indication at all for a subjective experience (qualia) that go with it or not.

For instance, I am at an ice cream stand, deciding what I want to get. Do I pick cherry because I have a 'mind' that says, "I want cherry" and then translates it into the physical act of choosing? Or is that act determined by genetic and environment, and my 'deciding' is only an illusion? Both answers have their pitfalls. :)

Yes, so I think it is the physics of my body that generates this "desire of cherry" and it are purely physical phenomena which make me "pick it", and I have the illusion of having made a decision.

This part is somehow necessary to "leave the physical ontology alone", and to allow for the split between a "purely physical ontology which has its physical laws" and then a "rule that determines how I derive my (passive) subjective experiences from that".

The problem then becomes, "How do you know you're conscious?" If you say, "I know because I am thinking", then certainly, others around also seem to have 'consciousness' because they also think and respond to you.

Well, I think we misread Descartes into "thinking as a problem solving intellectual activity". This is behavioural. I think that the "thinking" of Descartes meant "I have subjective experiences, qualia".

To illustrate this: prove me that a stone has no qualia. Prove me that it doesn't "hurt" a stone when I hit chips of it. You can't. At best you'll REDEFINE in an anthropocentric way what it means, to have subjective experiences, and then prove that there's no remote way in which there could be SIMILAR processes in a stone than in a human body. Sure, it must "feel" totally different to "be a stone" than to be "Joe Sixpack"

Again, I need a certain, limited, form of dualism, in order to specify things such that there is a physical ontology FROM WHICH my subjective experience is derived, and in such a way that this subjective world has no "influence" on the physical ontology. But there's still a lot of "wiggle room" to adapt it to your taste.

In fact, as much as I'm a fundamental reductionist concerning the physical ontology (in the Weinberg sense: that nature's ontology must correspond to a mathematical object, which describes _everything_ of that ontology), I might here incline to anti-reductionists, to say that subjective experiences might be an emerging phenomenon from certain physical structures, which is NOT described by the ontology itself (although all the ontology says about the physical structure is still entirely true - so there's no contradiction).

 

[Hurkyl]

But you have simply asserted a summary of what is (taken to be) observed in the lab (so to speak). This is not a *theory*.

No, this is my theory. In my theory, the probability distribution is a fundamental physical constant.

And besides, this is exactly what you were asking for: you were asking a theory about "irreducibly-random events", and further clarified by stating "a universe where the outcomes aren't based on any micro-hidden-details, but are genuinely irreducibly random".

So, in fact, the very conditions you've put forth require that there are no other physical variables that affect the outcomes of the coin flips -- the only property these coins have is their joint probability distribution!



This seems, to me, to completely fulfill the requirements you set forth. If you disagree, it would help greatly if you could put forth any theory that had irreducibly-random events in a universe where the outcomes aren't based on any micro-hidden-details, but are genuinely irreducibly random

So I should reformulate the question: can anyone think of a *theory* which explains the coin-flipping results in a local or lorentz invariant way? Just repeating back what those results are, doesn't cut it.

But if you're looking for any theory, we don't even have to be hypothetical. We could even take ordinary classical mechanics!

I can, right now, create two devices that will display either "H" or "T" when a button is pressed. I can give these to Alice and Bob, and they can press the buttons in whatever way they want, and when they compare notes, they will find they both got the same sequences of "H"s and "T"s.

 

[LnGrrrR]

Vanesch,

Yes, it's certainly interesting how QM can intrude (or at the least, walk hand in hand) with some forms of philosophy.

I myself lean towards property dualism. :)

I can not prove the qualia of a stone...but I also can't prove an invisible spaghetti monster lives on the planet Uranus. So I choose to assume that neither happens. :)

Thanks for the replies though. This is a 'pet topic' of mine.

 

[ttn]

No, this is my theory. In my theory, the probability distribution is a fundamental physical constant.

The probability distribution *of what*? You'll maybe say something like "the outcomes"... but what *are* these? What are they *made of*? What's actually *going on* in this scenario?

Is it really such a myserious surprise that there is a difference between saying "X happens" and providing a *theory* about X?

And besides, this is exactly what you were asking for: you were asking a theory about "irreducibly-random events", and further clarified by stating "a universe where the outcomes aren't based on any micro-hidden-details, but are genuinely irreducibly random".

A theory can provide some account of beables involved in the production of some "observable", and still include irreducibly random events. I'm not asking for a deterministic theory -- just some kind of theory with some kind of state descriptions and some kind of dynamics.

So, in fact, the very conditions you've put forth require that there are no other physical variables that affect the outcomes of the coin flips -- the only property these coins have is their joint probability distribution!

Maybe the problem is that the example is too silly. But the main point here is that my conditions *don't* "require that there are no other physical variables that affect the outcomes..." Indeed, I don't see how you can propose a theory without talking about, well, *some* variables. An example would be orthodox QM: the collapse postulate involves irreducible randomness, yet it is part of a theory which proposes a definite state description and definite dynamics. So what you need is an example of something like that -- but something that doesn't violate Bell Locality.

This seems, to me, to completely fulfill the requirements you set forth. If you disagree, it would help greatly if you could put forth any theory that had irreducibly-random events in a universe where the outcomes aren't based on any micro-hidden-details, but are genuinely irreducibly random.

Orthodox QM w/ collapse postulate. Only, make the collapse occur along future-light-cone out from the measurement event so it doesn't violate Bell Locality. That would be a local theory w/ irreducible randomness. (But of course it isn't empirically viable.)

I can, right now, create two devices that will display either "H" or "T" when a button is pressed. I can give these to Alice and Bob, and they can press the buttons in whatever way they want, and when they compare notes, they will find they both got the same sequences of "H"s and "T"s.

Sure you can do this. But what does this have to do with Bell Locality? Perhaps you've forgotten that Bell Locality speaks not of subjective probabilities but of probabilities assigned by a theory. And it doesn't count to say "my theory is that there's nothing actually going on" -- i.e., "my theory is that there is no theory."

 

[ttn]

I think you are here thinking of "a mechanism that can explain the randomness"... and as such should not be intrinsically random itself ; in other words, a deterministic mechanism.

You're half right. I'm thinking of what might as well be called "a mechanism that can explain"... but the point is not to explain the randomness (in terms of an underlying deterministic mechanism) but just to explain the results. This is nothing special. It's called having a theory about what's going on (as opposed to just shrugging and saying "this particle detector over here ended up firing at a certain moment; oh well; I guess that's just an irreducible inexplicable fact").

What's wrong with say, have "stochastical" classical mechanics, where you have to add, at spacelike separated events, intrinsically random variables to the local equations of motion, but in such a way that these random variables are correlated ?

I have no objection to injecting random variables into the dynamics. What I object to is picking a random number and then simultaneously injecting it into the dynamics of two spatially separated regions.

Of course, you can not find any DETERMINISTIC mechanism that can explain this correlation from an underlying "deterministic mechanics with lack of knowledge of the initial state" ; but if the intrinsically random variables are FUNDAMENTAL ? If you have no underlying mechanism, how are you going to require any statistical independence ?

Maybe this is what's causing the disagreement here. You seem to think that "irreducible randomness" means there's "no underlying mechanism." I'm willing to accept candidate descriptions of underlying mechanisms (i.e., *theories*!) which involve irreducible randomness.

For example, I think Orthodox QM is perfectly well a "theory" in the sense I'm insisting on. It provides an account of what's going on behind the scenes to produce something like a detector firing. And what I don't like about that theory is *not* that half of its dynamics isn't deterministic. The "problem" (in this context) is that this half of the dynamics is nonlocal. It says that an irreducibly random event at one place has an instantaneous effect at other places.

Do people think that the whole notion of "causal influence" is absent/meaningless in a non-deterministic theory? Maybe that's the source of the trouble here...

 

[selfAdjoint]

The "problem" (in this context) is that this half of the dynamics is nonlocal. It says that an irreducibly random event at one place has an instantaneous effect at other places.

Do people think that the whole notion of "causal influence" is absent/meaningless in a non-deterministic theory? Maybe that's the source of the trouble here...

The problem is that "orthodox quantum mechanics" doesn't say that, and does have a perfectly good relativistic notion of causality. It says that you only see the correlation after the fact, and that the correlation isn't because the one event caused the other, because there is no causality between spacelike separated events. It all hangs together and gives correct predictions, but you have to take to heart the old saying "Correlation is not causation."

 

[ttn]

The problem is that "orthodox quantum mechanics" doesn't say that, and does have a perfectly good relativistic notion of causality. It says that you only see the correlation after the fact, and that the correlation isn't because the one event caused the other, because there is no causality between spacelike separated events. It all hangs together and gives correct predictions, but you have to take to heart the old saying "Correlation is not causation."

In orthodox QM, the state description of one object (or, in one region) *changes* as a result of a measurement on some other object (in some other region). One can of course always say "Oh, you shouldn't take those state descriptions literally." Well, OK. But not taking the descriptions as literal does not change the fact that *if* those descriptions are literally correct ("complete") descriptions of the real world, then the real world contains relativity-violating causal influences. That's just what the theory *says*, and frankly I find it ridiculous that so many otherwise reasonable people are able to get lulled into the cognitive trap of simply denying the literal truth of the theory as a way of avoiding this implication. The fact is, orthodox QM provides a candidate description of the world and dynamics for it, and this combination -- this theory -- is nonlocal (in the specific sense of violating Bell Locality).

A given person may or may not be troubled by this fact, depending on whether or not he thinks this theory is true. (If you don't think OQM is true, then there's no reason to worry about its conflicting with relativity.) But then, you have to face the obvious next question: what theory *is* true? Anybody who is not interested in asking and answering this question is, in my opinion, no physicist.

As to "correlation is not causation", the point of this old saying is that you can't infer, from the mere observed correlation of A and B, that A causes B. But this has precisely nothing to do with what we're talking about here. Yes, it's notoriously difficult to validly infer causal relationships from observation. But the whole beauty of Bell Locality is that we don't have to try to do this, because the criterion isn't *about* observed correlations (or "subjective probabilities" as I keep saying). It's about the predicitons of *theories*. And it is notoriously UN-difficult -- notoriously *easy* -- to validly infer causal relationships from *theories*, because (by virtue of what it means to be a theory in the first place) theories simply *tell* us what causal relationships exist.

This is an absolutely fundamental point to this discussion, and there's no point talking about anything else until this is clear. Nobody is saying that you can look at some empirical fact (like Alice's and Bob's outcomes are correlated in a certain way) and infer that there is some spooky superluminal causation going on. The claim is that you can look at a *candidate theory* (like OQM or Bohmian Mechanics or whatever) and infer that, according to this theory, there is or is not spooky superluminal causation going on. To be confused about this point, is the same about being confused about the distinction between Bell Locality and Signal Locality. (A violation of the latter *can* be directly inferred from observed correlations; a violation of the former cannot.)

 

[selfAdjoint]

In orthodox QM, the state description of one object (or, in one region) *changes* as a result of a measurement on some other object (in some other region).

Orthodoxly the state function doesn't exists in any region of spacetime. What exists is a measurement, the result of a preparation and an action. Those are events in spacetime. And the time relations of spacelike events are indeterminate. So what you're saying here is violating both QM and relativity.

 

[ttn]

Orthodoxly the state function doesn't exists in any region of spacetime. What exists is a measurement, the result of a preparation and an action. Those are events in spacetime. And the time relations of spacelike events are indeterminate. So what you're saying here is violating both QM and relativity.

This point has already been covered. There are simply two possible version of "orthodox QM". One (perhaps closer to von Neumann's ideas, although Bohr's insistence on the completeness doctrine makes me think this is close to Bohr's views too) is that the wave function provides a literal and complete description of physical states, with two different dynamical laws depending on whether or not a measurement is being made. The other possible view is to take the whole quantum formalism as an empty black-box algorithm for predicting measurement outcomes. This latter may or may not constitute a theory (in the relevant sense) depending on whether or not its advocate is claiming *ignorance* about real goings-on at the sub-microscopic level, or, rather, is claiming that there are no such goings on (i.e., that *the only thing that exists are readings on measurement devices* -- a view that to me is too preposterous to take seriously since it contradicts pretty much everything discovered by scientists in the last 200 years).

 

[vanesch]

I have no objection to injecting random variables into the dynamics. What I object to is picking a random number and then simultaneously injecting it into the dynamics of two spatially separated regions.

Why ?

Maybe this is what's causing the disagreement here. You seem to think that "irreducible randomness" means there's "no underlying mechanism." I'm willing to accept candidate descriptions of underlying mechanisms (i.e., *theories*!) which involve irreducible randomness.

Yes, but you place extra limits on how this irreducible randomness can be applied. For instance, what's indeed wrong with using the SAME (or correlated) *intrinsic random numbers" at two spatially separated events ? If these numbers have no ontology attached to them, and are fundamental quantities, I don't see why this should be forbidden. Them not having a *mechanism* in them, there's no "causality" involved in this. It "just happens that way". This is the essence of an intrinsically fundamental stochastical theory, no ? "things just happen this way".

For example, I think Orthodox QM is perfectly well a "theory" in the sense I'm insisting on. It provides an account of what's going on behind the scenes to produce something like a detector firing. And what I don't like about that theory is *not* that half of its dynamics isn't deterministic. The "problem" (in this context) is that this half of the dynamics is nonlocal. It says that an irreducibly random event at one place has an instantaneous effect at other places.

Well, this is only true in the case that one assigns some reality to the concept of the wavefunction ; not if it is an "algorithm to calculate probabilities", right ?

Do people think that the whole notion of "causal influence" is absent/meaningless in a non-deterministic theory? Maybe that's the source of the trouble here...

I think that there is a total absense of the notion of causal influence in THE STOCHASTICAL ELEMENT of an *intrinsically* non-deterministic theory. That doesn't mean that the theory as a whole does not have elements of causality to it: the "deterministic part" (the evolution equations of the ontologically postulated objects) does have such a thing of course. But the "random variables" that are supposed to describe the intrinsic randomness of the whole don't have - a priori - to obey any kind of rules, no ?

 

[Hurkyl]

The probability distribution *of what*? You'll maybe say something like "the outcomes"... but what *are* these? What are they *made of*? What's actually *going on* in this scenario?

In my theory, the outcomes "H" and "T" are fundamental things. They cannot be analyzed any further. The thing I call a "probability distribution" is a fundamental quantity.

The term "probability" is justified for the following reasion:

Suppose we had many pairs of coins whose "joint probability distribution" factors into the "probability distributions" on the pairs. (Where, again, P(HH) = P(TT) = 1/2 and P(TH) = P(HT) = 0 for the two coins in a pair)

Then, this "probability distribution" gives a value of nearly one to the set of outcomes where roughly half of the pairs of coins flip to HH and rest flip to TT.

So, my "probability distribution" really does give probabilities -- it satisfies the frequentist interpretation of probabilities.

Is it really such a myserious surprise that there is a difference between saying "X happens" and providing a *theory* about X?
...
Maybe the problem is that the example is too silly.

But yet, there is a theory that says nothing more than "X happens". The class of theories is very broad, and includes lots of dumb, ridiculous, uninteresting, unrealistic, and impractical things.

But yet, they're all still theories.

I'm not asking for a deterministic theory -- just some kind of theory with some kind of state descriptions and some kind of dynamics.

Fine -- flesh the rest out however you want. Let's make the coins Special Relativistic point particles, and each can be in one of three fundamental states: "H, T, and unflipped". The initially start out as "unflipped", and via an interaction called "flipping" can transition to "H" or "T". The transition is nondeterminsitic, and is governed by the joint probability distribution P(TT)=P(HH) = 1/2, P(TH) = P(HT) = 0. This joint probability distribution is a fundamental constant of the theory.


More details just obscures the point -- the theory is simple and clear. It doesn't have messy details to work through and understand, and it's manifestly Lorentz invariant.

If you really needed me to, I'm sure I could flesh this out to work with many coins and additional interactions, such as ways to revert a coin back to the "unflipped state" and a pairwise interaction on coins called "entanglement", but that would just obscure what's going on, and I don't think I'd be doing anything more than a nondeterministic variation of classical mechanics without the axiom that spatially separated probabilities are statistically independent.

Sure you can do this. But what does this have to do with Bell Locality?

It has to do with what you had asked in the paragraph I quoted.




Incidentally, another way to go is to assert that observations are random variables. And not the silly stuff we learned as kids: I'm saying that observations are like the random variables defined in mathematical statistics.

Observations are probabilitiy distributions on a space of outcomes, nothing more. In particular, observations never actually "take on" the value of a particular outcome.

 

[selfAdjoint]

This point has already been covered. There are simply two possible version of "orthodox QM". One (perhaps closer to von Neumann's ideas, although Bohr's insistence on the completeness doctrine makes me think this is close to Bohr's views too) is that the wave function provides a literal and complete description of physical states

You say you have been over it, but you keep coming back to the same issues. What does "literal and complete" mean? A map can provide a complete description of a town; every street, every address, is there; but the map is still not the town, and the space of quantum states is still not in spacetime.