John Bell Workshop 2014

A reasonable thing that just might work

Viewing 9 posts - 1 through 9 (of 9 total)
  • Author
  • #1723
    Daniel Rohrlich

    In 1964, John Bell proved that quantum mechanics is “unreasonable” (to use Einstein’s term):  there are nonlocal bipartite quantum correlations.  But they are not the most nonlocal bipartite correlations consistent with relativistic causality (“no superluminal signalling”): also maximally nonlocal “superquantum” (or “PR-box”) correlations are consistent with relativistic causality.  I show that – unlike quantum correlations – these correlations do not have a classical limit consistent with relativistic causality.  The generalization of this result to all stronger-than-quantum nonlocal correlations is a derivation of Tsirelson’s bound – a theorem of quantum mechanics – from the three axioms of relativistic causality, nonlocality, and the existence of a classical limit.  But is it reasonable to derive (a part of) quantum mechanics from the unreasonable axiom of nonlocality?!  I consider replacing the nonlocality axiom with an equivalent axiom that even Bell and Einstein might have considered reasonable:  an axiom of local retrocausality.

    Rohrlich – Bell paper

    Robert Griffiths

    6 January 2015

    Dear Daniel,

    I only recently got around to looking at your post. Here are some comments, numbered to help me keep them straight.

    1. ‘Retrocausality’ means the future influencing the past, which sounds like science fiction. I think you should stick to ‘microscopic time-reversal symmetry’ as otherwise others will be as confused as I was.

    2. What I think you mean in Fig. 1(a) is Alice and Bob are PREPARING spin states and Claire is MEASURING them in a Bell basis. As used in ordinary laboratory physics preparation means to produce a state, so you know what it is after the preparation, and the outcome of a measurement tells you what the state was before you measured it. I advocate making the same distinction in foundations papers. If, with these definitions, in Fig. 1(b) Claire prepares and Alice and Bob measure, then I know what you are talking about; otherwise I am confused.

    3. I am happy to see that you think QM should be local and time-symmetric, as that is also my opinion. This is true in the consistent histories (CH) approach, which does not seem present in your references. In particular you might take a look at Found. Phys. 41 (2011) 705; arXiv:0908.2914, where I argue that QM contains a principle of Einstein locality. David Mermin was the referee and he tried very hard to find significant errors, but was unsuccessful. (I challenge you to do better!)

    4. The trouble with ABL and also Aharonov’s two-state formalism is that they employ MEASUREMENT, and since real measurements are thermodynamically irreversible, this makes it hard to use them to produce a theory with microscopic time reversal symmetry. In CH there is a microscopic description of what goes on in a simple measurement process, and you might find this interesting. Two chapters in my CONSISTENT QUANTUM THEORY explain it, but will give you a quick idea.

    All the best! Bob Griffiths

    Ken Wharton

    Hi Bob,

    On your point #1, I grant that Daniel is mainly using time-reversal symmetry as a *motivation*, but the consequence is indeed retrocausal. (As defined by Huw Price: an external intervention on a system that is correlated with a physical property of that system at an earlier time.) If that makes you think of bad science fiction stories, perhaps my explicit model in this workshop might make for a better framework for considering such things.

    In fact, Huw and I have a nice argument about how introducing time-symmetry *necessarily* forces one into a retrocausal story if one has any quantization at all. (Specifically, if the space of measurement outcomes is smaller than in the corresponding classical theory.) So you can have time-symmetry without retrocausality in classical physics, but not in quantum physics. See , and Huw’s earlier piece at .

    This result has now been improved/generalized by Matt Pusey and Matt Leifer:

    Cheers, Ken

    Lee Smolin

    Hi Ken,

    Do these interesting results imply that if one doesn’t want to admit retrocausality in quantum physics one must also deny time reversal symmetry? At what level must one give up time reversal symmetry? Must unitary time evolution or the Schroedinger equation be modified?



    Ruth Kastner

    We need to be clear about what we mean by ‘retrocausation’. Retrocausation in the sense that a pre-existing future event influences the past or the present is not at all a necessary inference from time symmetry. The slide into this inference is based on the assumption that time symmetric quantum process necessarily imply a block world, which is not the case. I present a formulation of the transactional interpretation in which the underlying quantum processes–specifically the emissions of quantum states– are time-symmetric, but the actualization of a specific spacetime event involves time-asymmetric collapse. Out of this, we get a growing universe, not a block universe. This is all perfectly harmonious with relativity, as it aligns with the ‘causal set’ growing universe picture as proposed by Sorkin et al. The time-asymmetric process arises from the response of absorbers and provides for a natural physical referent for the Born Rule. I discuss this in my Cambridge talk at the Conference on Retrocausation in Quantum Theory (link available at the Cambridge site or on youtube).

    Ken Wharton

    Dear Lee,

    I don’t think there’s just one way forward from the dilemma… or even merely a few. Both Everettians and Bohmians get around the argument via an underlying continuum even after measurement: it may look like a limited space of measurement outcomes, but in these cases the sub-classical “restriction” of measurement outcomes is just illusory. So here one can recover the classical symmetry-without-retrocausality.

    This loophole aside, our argument does go through for psi-epistemic cases (as now proved by Pusey and Leifer) and also for cases that are psi-ontic at the measurement itself (as in the case where a measurement of photon polarization means some ontological field is restricted to only oscillating along that particular axis, at the time of measurement; this is our main example in those papers.)

    But even in these cases, there are lots of ways to recover a forward-causal theory by breaking time-symmetry. One way, as Ruth essentially points out, is just to deny the block universe, essentially claiming that “the future” is less real than “the past” and/or “the present” in some objective sense. (Whether or not this claim can plausibly be placed on some objective basis, I have my doubts; I feel that the line between future/present/past is about as subjective as it is possible to be, but I know we disagree on this point!) Without a real future, one forbids any retrocausality by fiat (if the future isn’t real, it certainly can’t cause anything). And I would think it should be obvious that any such premise would break time-symmetry as well, in at least some important sense.

    More on this point in my next post…

    Best, Ken

    Ken Wharton

    Hi Ruth,

    I know we’ve debated the block universe viewpoint in person many times, to precious little mutual agreement. But it’s also the initial crucial point that has to be gotten straight before many of these discussions can even get off the ground. As you know, you’re not alone in your dislike of the block universe, but I’m concerned that that in many cases this dislike translates into a mental inflexibility to even *consider* arguments couched in a block-perspective. (And Daniel, as you told me you’re an “agnostic” on this issue, I can hardly resist a chance to proselytize my side of this debate in your own topic… 🙂

    One point I like to make is that for those few of us willing to rip up the foundations of QM and try to find some deeper story, the remaining pillar of physics should continue to be general relativity. (Sure, I suppose we could rip up both pillars at once, but that seems insanely reckless.) And GR is firmly built on a block-universe worldview. (One can perhaps(?) dispense of it in Hamiltonian GR, but not Lagrangian GR; what would a closed-timelike curve even *mean* without a block universe?)

    Even apart from the block-universe aspects of GR, it’s even more important to note that the GR we know and test is in 3+1 dimensions, not 3+2 dimensions (or 3N+1 dimensions, but that’s another diatribe). Given that we have only one time dimension, I’m concerned that certain phrases you use like “growing universe”, “actualization”, and “pre-existing… future” don’t have a mathematical framework in which they can be objectively defined.

    Take the former phrase. The word “growing” normally means a change of spatial size with respect to a change of time. But applied to time (which is what you’re doing), you would like it to mean a change of temporal size with respect to time. And this is no different from dt/dt=1. The concept is therefore a meaningless tautology, unless you have in mind two different time dimensions, so that a change in one of those dimensions can be measured in terms of the other. Any sort of “growing universe” has to be growing *with respect to something*, and it seems to me that any attempt to carefully define what one means by this “something” requires a second time/reference dimension.

    Now, I do think that you have two times in mind; the time of an event, and the time at which that (past or future) event is “discussed” or “considered”. So perhaps the question about the logical viability of non-block-universes comes down to whether this second time is objective or subjective. But again, I think that if it’s objective, you’re essentially talking about 2-dimensional time, and throwing everything we know about GR out the window. Furthermore, what could be more *subjective* than the “now” that I’m considering to be “now” right “now”? (In my experience, every time that I’ve ever experienced seemed like “now”, so I seriously doubt such an experience is any more special or objective than my notion of “here”.)

    Granted, for whatever reason, most people don’t *want* the future to be as “real” as the past. But trying to find a logical framework in which that desire comes true is so different from the framework in which we do physics that I think it’s a dangerous departure from everything we already know about the universe. And of course, once you are in such a two-time framework, what is meant by “time-symmetry” is far from clear or evident… which is presumably how you can claim that you have time-symmetry in one sense while defending a dramatic past/future time-asymmetry in another sense…?

    All the Best,


    Ruth Kastner

    I’d like to focus on Ken’s comment that “the line between future/present/past is about as subjective as it is possible to be.”

    Why would making a distinction between future, present and past be thought of as subjective in the extreme? Physics is an empirical science, and we can all corroborate that we all experience the present, and that’s all that we ever experience. Now, maybe that doesn’t qualify as ‘objective’ in the sense that we can’t point to ‘the present’ on a spacetime diagram. But the spacetime diagram is not demonstrably more than a useful map, and (as I point out in my book) the map is not the territory. If one takes the map as the territory, then one presumes a block world as an objective matter, and then one’s direct experience of a distinction between the present moment and other moments is seen as merely perspectival, or merely subjective. But that judgment is based on a presumption that there is a block world. So my basic point is that perhaps the idea that the future/present/past distinction is ‘merely subjective’ is based on the presumption that objectively there is a block world. But we don’t really know that to be the case.

    Also, to clarify my position, I don’t call the future ‘unreal,’ nor do I propose anything about the future ‘by fiat’. In my proposal (PTI), the future is real, but not actual. It’s a collection of real quantum possibilities, much like the yarn you use when knitting a garment. The extruded garment is the ‘fabric of spacetime’ including past events; the stitches on the needle are the ‘present.’ The Now does not ‘move’ anywhere in this growing universe picture. It naturally comes out of the direct-action theory where quantized energy transfers result asymmetrically from absorber responses, and those energy transfers from emitters to absorbers define temporal direction. It also naturally explains the Born Rule.

    So there’s nothing that need be ‘subjective’ about this. It’s a proposed interpretation of quantum theory that is grounded in well-established theories of field propagation. It could well describe what is actually happening, in an objective sense. As noted above, if one assumes there is a block world then the distinction between future/present/past is by definition merely perspectival and subjective. But that’s only because the block world is presumed to be objective truth, when it may not be. There’s nothing necessarily ‘subjective’ (meaning irreducibly perspectival and partial) about taking seriously that there may be a real physical distinction between future, present, and past.

    Ruth Kastner

    Ken, I posted that last response before I saw your second post.
    The causal set picture of a growing universe provides a mathematically rigorous formulation of the kind of growing universe model I have in mind. PTI just provides the dynamics for how events get ‘sprinkled’ into that growing spacetime from a quantum substratum (I have a paper on this on the arxiv, contribution to an edited volume by Ignazio Licata). I don’t think it would really be fair to describe the work of Sorkin, Rideout, Dowker etc. as lacking mathematical rigor. Dowker in particular has written about the local nature of ‘Now’ in this kind of picture.

    The basic field propagation is not ‘in time’, so there are not two different levels of time. It’s better to think of the symmetric field propagation as energy-symmetric. Spacetime intervals are established by actualized energy transfers–those involve positive energy only, and that’s how a temporal interval is created (energy being the generator of time translations).

    The current ‘framework in which we do physics’ has failed to provide a physical reason for the Born Rule or any consensus on what QM means, and has not solved the problem of quantum gravity. Sometimes we need to expand our conceptual framework to solve extant problems. Remember that Boltzmann deviated from his contemporary framework by daring to propose unobservable tiny objects called ‘atoms’.

    In any case, I do address many of the issues you raise in my 2012 book, and in more conceptual detail in my new book (early 2015). So perhaps your concerns about my proposal should best be handled by your referencing specific areas of my published papers or those of the causal set folks that you have questions about, and I’ll try to clarify.


Viewing 9 posts - 1 through 9 (of 9 total)
  • You must be logged in to reply to this topic.