# Nonlocality: correlation vs causation

• Demystifier
In summary, the conversation discusses the idea that quantum mechanics may not be truly nonlocal, as the only nonlocal aspect is the correlations between particles rather than actual causation. The Bohmian interpretation is often criticized for being too nonlocal, but it is argued that there is no substantial difference between correlation and causation. The argument is made that both correlation and causation have the same form and that Bohmian interpretation is not more or less nonlocal than the standard correlation interpretation. There is also discussion about the difference between perfect correlation and causation, with the conclusion that there is no significant difference between the two.
Demystifier
Gold Member
It is often said that QM is not truly nonlocal, because the "only" thing which is nonlocal are the correlations; there is no true nonlocal causation involved.

At the same time, the Bohmian interpretation is accused for being "too" nonlocal, by involving a true nonlocal causation.

But what exactly the difference between correlation and causation is? Here I want to argue that there is no substantial difference at all. If my argument is correct, then Bohmian interpretation is not more (nor less) nonlocal than the standard correlation interpretation.

Here is the argument. For simplicity, consider the case of perfect correlation. A perfect correlation always has the form

whenever system A has property P1, the system B has property P2

For example, whenever the left particle has spin up, the right particle has spin down.

But Bohmian nonlocality also has this form: Whenever the left particle has that position, the right particle on the other position has this velocity.

So where is the difference between Bohmian nonlocality and correlation nonlocality? What is the difference between causation and perfect correlation? I don't see any substantial difference. Do you?

Demystifier said:
So where is the difference between Bohmian nonlocality and correlation nonlocality? What is the difference between causation and perfect correlation? I don't see any substantial difference. Do you?

You are probably right (since there is just one wavefunction). I think it would be hard to make the counter argument. Either way, it is interesting that things are arranged "just so" that FTL signaling is not possible. Which is one of the caveats also mentioned regarding the issue. Not that I consider that an argument one way or the other.

Just my 2 cents (although with inflation, I might should bump that to 3 cents).

DrChinese said:
You are probably right (since there is just one wavefunction). I think it would be hard to make the counter argument.
So we agree, good!

DrChinese said:
Either way, it is interesting that things are arranged "just so" that FTL signaling is not possible.
Which (as you know, but I just want to be more explicit) is true for both interpretations.

Demystifier said:
It is often said that QM is not truly nonlocal, because the "only" thing which is nonlocal are the correlations; there is no true nonlocal causation involved.

At the same time, the Bohmian interpretation is accused for being "too" nonlocal, by involving a true nonlocal causation.

But what exactly the difference between correlation and causation is? Here I want to argue that there is no substantial difference at all. If my argument is correct, then Bohmian interpretation is not more (nor less) nonlocal than the standard correlation interpretation.

Here is the argument. For simplicity, consider the case of perfect correlation. A perfect correlation always has the form

whenever system A has property P1, the system B has property P2

For example, whenever the left particle has spin up, the right particle has spin down.

But Bohmian nonlocality also has this form: Whenever the left particle has that position, the right particle on the other position has this velocity.

So where is the difference between Bohmian nonlocality and correlation nonlocality? What is the difference between causation and perfect correlation? I don't see any substantial difference. Do you?
I see your point(s). So then can we consider Bohmian and standand qm as either both describing a causal relationship (via instantaneous action at a distance) between spatially separated events, OR as both describing a correlation (but not necessarily a causal relationship) between spatially separated events?

My current choice is the latter. And this is good for the general acceptability of Bohmian qm, isn't it?

The thing about assuming nonlocality is that it's quite possibly an unwarranted leap. That is, we KNOW that there's a correlation there. The problem is that, because of the requirements for a general and explicitly local and realistic account of entanglement, LR models which would 'explain' the correlations in terms of a common cause are ruled out. Complicating the interpretation of this is the fact that the violation of Bell inequalities might simply be due to formal considerations, as opposed to assuming that it's due to the existence of nonlocal 'transmissions' or 'influences'.

My current choice is that the nonviability of (at least Bell-like) LR models of entanglement is explained formally. Which means that, even if our universe is evolving solely in accordance with the principle of local causality, an explicit LR model of entanglement is impossible. Which means that the nonviability of explicit LR models of entanglement doesn't imply that the correlations can't be 'understood', nonformally, in terms of common causes and standard applicable optics principles and conservation laws, or that the assumption of nonlocality is required.

Also, there is the overriding correlation between the angular difference of the analyzers and the coincidental photon flux. These correlations are not perfect, but do correspond to standard optical expectations. In fact, it's an LR model which produces a perfect (linear), and incorrect, correlation in this context.

Last edited:
Demystifier said:
It is often said that QM is not truly nonlocal, because the "only" thing which is nonlocal are the correlations; there is no true nonlocal causation involved.
...
What is the difference between causation and perfect correlation? I don't see any substantial difference. Do you?

In what I think you mean with "perfect correlation" it sounds like it's the same as causation, however, as I see it the central difference is that of what is inferrable and what is not.

From the inference perspective I see it like this:

Correlations in the past can be inferred from history, future correlations can be rationally EXPECTED from this (not deduced or perfectly predicted). Causations can never be strictly deduced, it is an realist abstraction.

When the expectation of correlation is confident enough it effectively is a causal contraint on the observers action. This is clear when you take the gaming perspective: The choices ofa player is determined by his expectations, not by the "truth". Expectations rules, even then they are wrong. (See stock market and economical systems).

So from my biased perspective - which takes the extreme opposite to those of a realist and thinks that an intrinsic theory can be constructed only but inferrable building blocks - there is really no such thing as perfect inferrable causation nor inferrable perfect correlation, all there is are imperfect correlations, which indirectly, via rational expectations, produces an imperfect EXPECTATION of "correlation" which then guides the action of the observer. This is in my view how causality emerges as effective constraints, just like effective laws emerge in a game due to equilibrium between gamers action choices.

Instead what replaces "deductive causality" (essential to a realist) in my understanding is rational expectations (which are more inductive to nature). Still one can save locality, if we consider that local rational expectations depend only on local information. Ie the rational expectations of an observer, depends only upon information in her possesion.

Yet another angle:

From a purely descriptive view, all there ever is are correlations. Descriptions always refer to the past, the history. You don't "describe the future" in the strict sense.

From the decision theoretic view, you have EXPECTATIONS on the future, but the sole purposes of these expectations is that they determined the obsevers current actions. So expectations of the future shouldn't be tested right or wrong against in retrospect as beeing right or wrong with actual recods, it should be tested against the action of the observer encoding the expectation, just analogies to the saying that "actions revels your thoughts". The same way do I think the action of a physical system, reveals it's encode expectations of the future. An expectation can be rational, yet wrong. This is why this is not a descriptive problem like when you describe an actual history in retrospect.

So about QM prefect correlations, I think in fact is just EXPECTATIONS of correlations. The unitary evolution is an EXPECTED evoltuion (at leat in my interpretation). This expectation also hold accurate in retrospect as long as the system is isolated. The problem of coruse is - how does the observer KNOW it's isolated? Well it doesn't. At least not until in retrospect. It's rather "assumed to be so" otherwise QM predictions doesn't hold ;)

/Fredrik

ThomasT said:
I see your point(s). So then can we consider Bohmian and standand qm as either both describing a causal relationship (via instantaneous action at a distance) between spatially separated events, OR as both describing a correlation (but not necessarily a causal relationship) between spatially separated events?
Both views make sense. In fact, I don't see a real difference between these two views.

ThomasT said:
My current choice is the latter. And this is good for the general acceptability of Bohmian qm, isn't it?
I think so.

Last edited:
ThomasT said:
So then can we consider Bohmian and standand qm as either both describing a causal relationship (via instantaneous action at a distance) between spatially separated events, OR as both describing a correlation (but not necessarily a causal relationship) between spatially separated events?

Demystifier said:
Both views make sense. In fact, I don't see a real difference between these two views.
I don't think both views make sense. Causality involves the indexing of the temporal evolutions of systems. "Instantantaneous action at a distance" is equivalent to "simultaneous action at a distance", where a causal relationship between the distant events is, by definition, disallowed. "Instantaneous transmissions" aren't so much spooky as simply silly (which is maybe what Einstein was getting at). "Instantaneous transmissions" is a contradiction in terms, an oxymoron with absolutely no physical meaning.

I think it would be a good move to stop referring to the formalisms of standard and Bohmian qm as 'nonlocal'. They're simply nonrealistic in certain respects -- while Bohmian qm still retains a more realistic flavor than standard qm regarding certain other features of its construction.

Currently, there's not much more that can be said about the entanglement correlations that give rise to notions of nonlocality than that they're correlations that, as such, are based on some sort of relationship between the quantum realm disturbances underlying the measurements. Exactly what this relationship is will remain a mystery so long as our qualitative apprehension of these disturbances is based on inferences from more or less random probes of the underlying reality. There are four distinct possibilities wrt where the relationship is produced, 1) via emission processes, 2) in flight, 3) via filtration processes, and 4) via detection processes. The most reasonable consideration given the current state of affairs is 1).

In any case, discarding silly notions of nonlocality benefits the case for Bohmian qm, while their retention carries with it a bit too much semantic baggage, imho.

What do you think?

Last edited:
ThomasT, I would summarize these ideas in the following way:
If standard QM is only about correlations, then Bohmian QM is only about perfect correlations. When viewed in that way, Bohmian QM is as (non)local as standard QM.

Would you agree?

ThomasT said:
I think it would be a good move to stop referring to the formalisms of standard and Bohmian qm as 'nonlocal'. They're simply nonrealistic in certain respects ...
Can you specify in what respect Bohmian QM is nonrealistic?

But I must say, there is SOME sense in which "standard" QM is more local than Bohmian QM. But this has nothing to do with correlation vs. causation. Instead, it has to do with the positivistic philosophy according to which only observed phenomenon is a phenomenon.

To explain this, consider Alice measuring the particle on the left and Bob measuring the particle on the right, and let the wave function be |up>|down>+|down>|up>.

Let Alice find the left particle in the state up at time t. The crucial question is the following: Does it mean that the right particle is in the state down at t?

If it does, then it is a correlation-nonlocality, which is not any "weaker" than nonlocality involved in Bohmian QM.

But "standard" QM, or at least one version of it, denies that it means that the right particle is in the state down at t. Even if Bob measures it and finds that it is, it is of no relevance to Alice because at time t she couldn't possibly know what Bob has measured at t. Of course, if she trusts the THEORY called QM, she could CALCULATE that the state measured by Bob should be down. But calculation is not a measurement. For Alice, the state of the right particle at time t is not observed at time t. She can observe it only later, after a time t + Delta t, where Delta t is time needed for a signal to come from Bob to Alice. So for her, the right particle is in the state down only later, which saves locality.

In this way, one can save even correlation-locality. But the price is very big. Not only that unobserved phenomenons are not physical, but even phenomenons observed by someone else are not physical. This is a logically consistent way of thinking, but is that really what physics should be about?

Last edited:
Demystifier said:
But what exactly the difference between correlation and causation is? Here I want to argue that there is no substantial difference at all.

I'm not arguing against your specific claim about differences in the meaning of nonlocality between Bohmian QM and any other. But I can't subscribe to the general postulate above. There is a specific counter example I can give, though it may not be totally without controversy.

According to the Maxwell equations (in vector form) there is a perfect correlation between the E and B fields. From a superficial viewpoint one might be led to believe that changes in one field cause changes in the other and vice versa. But the all mighty originators of the theory (Faraday, Maxwell, Heavyside and Fitzgerald) didn't state that to be the case. Recently (relatively) Jefimenko has shown mathematically that there is no such causal relationship. That is especially clearly shown in that wave equations for each field are of the same exact form and yet contain no term of the other field - perfect correlation with no causal relationship.

Maybe this should be rephrased to indicate that while there is certainly a definite causal relationship involving the two parameters (with at least one additional parameter), the causal relationship is not directly between the two, whereas the correlation is.

Last edited:
Demystifier said:
ThomasT, I would summarize these ideas in the following way:
If standard QM is only about correlations, then Bohmian QM is only about perfect correlations. When viewed in that way, Bohmian QM is as (non)local as standard QM.

Would you agree?
If Bohmian qm reproduces the standard qm predictions, then it's about the same correlations that standard qm is. Isn't it? Keeping in mind that the correlation in, eg., optical Bell tests, is between the angular difference of the analyzers and the rate of coincidental detection, and that this correlation is not perfect, then Bohmian qm is not only about perfect correlations. Unless I've misunderstood something about Bohmian qm, wrt which I have only a cursory familiarity. In which case I welcome any correction(s) to my way of thinking about this that you might want to suggest.

Regarding the correlation between spacelike separated detection events, A and B, with analyzers aligned, I agree with the way PhilDSP put it: "... the causal relationship is not directly between the two, whereas the correlation is."

I should add that there's currently simply no way of ascertaining if there's a direct causal link between A and B. Assuming that there is, and that it involves instantaneous action at a distance, is simply, prima facie, an untenable position due to semantic contradictions.

But it can be said without contradiction that, eg., in the case of optical Bell tests with analyzers aligned, the appropriately paired detection events, A and B, are perfectly correlated. In which case, we would look to an additional parameter, ie., the relationship between the underlying disturbances, to 'explain' (at least informally) the perfect correlation.

Bottom line is that both standard qm and Bohmian qm are essentially acausal or noncausal probability calculuses, so we shouldn't be referring to them as being either local or nonlocal. They're essentially neither, even though they both contain arguably realistic (and therefore, arguably, local and nonlocal) elements in their constructions and in the 'physical' models that they incorporate and are associated with certain experimental preparations.

So, yes, one could argue that there's "... SOME sense in which standard QM is more local than Bohmian QM." But I was taught to not think of quantum states as real physical states. I read "|up>|down>+|down>|up>" as meaning that, for appropriately paired sets of detection attributes, if A registers a detection, then B will register no detection, and vice versa. It doesn't refer to the underlying disturbances.

As well, one might argue that there's SOME sense in which Bohmian qm is more 'realistic' than standard qm. However, while they're both based on the assumption of the existence of a fundamental quantum of action, they're not essentially dynamical theories based on fundamental dynamical laws, and ultimately they both amount to convoluted probability calculuses which produce the same probabilistic predictions, so referring to them as either local or nonlocal theories is a misnomer, imho.

EDIT: So, I think I agree with your original premise, for somewhat different reasons, that the "Bohmian interpretation is not more (nor less) nonlocal than the standard correlation interpretation."

EDIT: On second thought, what you're arguing is that Bohmian qm and standard qm are both nonlocal, for essentially the same reason. Is that correct? If so, then since I think that using the term 'nonlocal' to describe either is something of a misnomer, then I guess I disagree with your original premise.

Last edited:
Fra said:
From a purely descriptive view, all there ever is are correlations. Descriptions always refer to the past, the history. You don't "describe the future" in the strict sense.
I used to have a signature that said "physics doesn't predict the future, it predicts the past before it happens." I see we are on the same page there!
So about QM prefect correlations, I think in fact is just EXPECTATIONS of correlations. The unitary evolution is an EXPECTED evoltuion (at leat in my interpretation). This expectation also hold accurate in retrospect as long as the system is isolated. The problem of coruse is - how does the observer KNOW it's isolated? Well it doesn't. At least not until in retrospect. It's rather "assumed to be so" otherwise QM predictions doesn't hold ;)
And, we actually know the system is never isolated in science, because science always requires that we look at it. A system that is never coupled to an analyzing agent is not subject to analysis, so the whole idea of a "closed system" was always a toy model.

PhilDSP said:
Maybe this should be rephrased to indicate that while there is certainly a definite causal relationship involving the two parameters (with at least one additional parameter), the causal relationship is not directly between the two, whereas the correlation is.
I'm not sure you are really disagreeing with Demystifier. You are saying you can have correlation without causation, but if I understand him correctly, he is saying that causation is a kind of myth, at least at the elementary level where you have things like single-particle Bohmian trajectories. I won't put words in his mouth, I'll just say it from my perspective: correlation is in the empirical data, causation is in the stories we tell about the data. What seems to be the law of physics here is that we can never find correlations in data that cannot be weaved into the causation story-- no correlations can cross the light cone, for example. So if physics protects itself from having correlations that can't be woven into the causation story, it means there's really no difference between correlation and causation that isn't purely a function of the story being told. You couldn't for example say "that correlation can't be causative because it crosses the light cone", expressly because correlations don't.

An example I point to is the reversibility of the equations of physics. In most elementary interactions, there is no arrow of time, so the difference between a cause and an effect is arbitrary. That difference doesn't come from the elementary physical interactions, it comes from the way we weave those pieces into a coherent description of what is happening. Correlation is empirical and objective, causation is rational and subjective. Even if I tell a story where a person shoots another with a gun, someone else could tell the story that the shooter was compelled to the act by the necessity of the fact that the other person got shot. It's just a different story, not likely to hold up in a court of morality, but perfectly good physics.

Demystifier said:
But I must say, there is SOME sense in which "standard" QM is more local than Bohmian QM. But this has nothing to do with correlation vs. causation. Instead, it has to do with the positivistic philosophy according to which only observed phenomenon is a phenomenon.
Another way to say this is, a "reference frame" is a completely local thing. In special relativity, it is popular to talk about a global reference frame with a grid of clocks and so on, but that's actually not a reference frame, it's a coordinate system. A big message of relativity is that physics should be coordinate-independent, which is just another way of saying that observers are local creatures, and physics is about what observers do.

Nevertheless, relativity, which is viewed as a kind of "arm" of physics, is about cobbling together what different observers do, into a unified "objective" whole that is coordinate independent, but nonlocal. So I think the nonlocality of quantum mechanics is not so much from the positivism of doing measurements, but from the rationalism of using relativity. Without some form of relativity, be it Galilean or Lorentzian, physics is very incomplete, and quantum mechanics is no exception.
In this way, one can save even correlation-locality. But the price is very big. Not only that unobserved phenomenons are not physical, but even phenomenons observed by someone else are not physical. This is a logically consistent way of thinking, but is that really what physics should be about?
Exactly, it is not what we usually think of physics as being about, because we usually think of physics + relativity, or else it has little predictive power. How can we predict local events without some global constructions to reference?

Ken G said:
... I'll just say it from my perspective: correlation is in the empirical data, causation is in the stories we tell about the data. What seems to be the law of physics here is that we can never find correlations in data that cannot be weaved into the causation story-- no correlations can cross the light cone, for example. So if physics protects itself from having correlations that can't be woven into the causation story, it means there's really no difference between correlation and causation that isn't purely a function of the story being told.
That is a clarifying way of putting it, I think. So, our causal story can have the underlying disturbances causally 'influencing' each other directly, or it can have them related in such a way that a common global parameter yields the observed (nonperfect as well as perfect) correlations. I find the latter to be a more reasonable assumptive approach. The question then has to do with where the relationship between the underlying disturbances is produced. Any thoughts as to the most reasonable assumptive approach to this question?

EDIT: By the way, I notice that you refer to the 'nonlocality' of qm. What exactly are you referring to? Does it involve instantaneous action at a distance, or ftl transmissions, or is it something else?

I'll offer a third possibility: stuff just happens, and the concept of "causation" is a template that our intelligence holds to what happens. We definitely get a lot of value out of that concept, it must be one of the first things a formative intelligence learns, but does reality itself have the slightest clue what this concept is? I don't think so, I think it's a simplification of how things actually work. It's a story that captures some crucial element or it wouldn't work so well, but there are probably illusory elements to it, related to the illusory elements of an "arrow of time."

To get an idea of what I mean here, take the following flight of fancy. Imagine that time actually goes in the opposite direction from what we perceive (I'm not suggesting this would make any more sense then our current thinking, I'm trying to cast doubt on our current thinking by making it look equally implausible). In this flight of fancy, the "truth" is that effects compel causes-- if something happens, then it must hold that a cause pops up some time later (which we call earlier). A man gets shot, so someone has to shoot him. Who has the gun? You do, it must have been you. So you must have gotten the gun somewhere. Where is there a gun, such that you're getting it is consistent with all the facts? In your drawer at home, that's the gun you know how to use. And so on-- the story just plays out in reverse. Where is the violation of physics in that story? Entropy doesn't increase, it decreases, we had it wrong all this time. We don't recall the past, we notice the evidence that is compelling it to happen that is stored in our brains.

If we cannot even tell a cause from an effect without bringing in all kinds of sociocultural elements of how we think, then can causation really be a physical principle? And I think this gets us to Demystifier's intent, if I understand it correctly: since preserving causation seems to be the main payoff one gets from all the rest of the unnecessary Bohmian overhead, how justified can that overhead really be if causation is not actually a physical principle at all?

Last edited:
DrChinese said:
Just my 2 cents (although with inflation, I might should bump that to 3 cents).

Depends upon which year you take as the base...;)

if it is to be the year this phrase/idiom was invented/got popular ...then it would be much for than 3 cents...maybe 3 dollars...;)

Ken G said:
I'll offer a third possibility: stuff just happens, and the concept of "causation" is a template that our intelligence holds to what happens. We definitely get a lot of value out of that concept, it must be one of the first things a formative intelligence learns, but does reality itself have the slightest clue what this concept is? I don't think so, I think it's a simplification of how things actually work. It's a story that captures some crucial element or it wouldn't work so well, but there are probably illusory elements to it, related to the illusory elements of an "arrow of time."

To get an idea of what I mean here, take the following flight of fancy. Imagine that time actually goes in the opposite direction from what we perceive (I'm not suggesting this would make any more sense then our current thinking, I'm trying to cast doubt on our current thinking by making it look equally implausible). In this flight of fancy, the "truth" is that effects compel causes-- if something happens, then it must hold that a cause pops up some time later (which we call earlier). A man gets shot, so someone has to shoot him. Who has the gun? You do, it must have been you. So you must have gotten the gun somewhere. Where is there a gun, such that you're getting it is consistent with all the facts? In your drawer at home, that's the gun you know how to use. And so on-- the story just plays out in reverse. Where is the violation of physics in that story? Entropy doesn't increase, it decreases, we had it wrong all this time. We don't recall the past, we notice the evidence of the future that is stored in our brains.

If we cannot even tell a cause from an effect without bringing in all kinds of sociocultural elements of how we think, then can causation really be a physical principle? And I think this gets us to Demystifier's intent, if I understand it correctly: since preserving causation seems to be the main payoff one gets from all the rest of the unnecessary Bohmian overhead, how justified can that overhead really be if causation is not actually a physical principle at all? I don't mean to steal his steam here, I just so completely agree with that that I'm just blurting it out.
I think of a causal (or deterministic) account of some set of more or less obviously related events as a temporal indexing of the evolution of those events. By definition, in any particular cause-effect relationship, the cause is associated with a lower number in the quantitative index than the effect.

Along with this comes the notion of necessary and sufficient conditions for a particular event to manifest.

I don't think that the observed arrows of time are in any way illusory, and that it's one of the problematic aspects of modern physics that a fundamental dynamical law accounting for the observed arrows of time isn't yet a part of physics.

The fact that the fundamental equations of motion are time-symmetric isn't in any way indicative of the possibility of time reversal in the real world. They are after all simply equations of motion, not fundamental dynamical laws of nature.

Anyway, I think that Demystifier's main intent is to say that the nonlocality of Bohmian qm is essentially the same as, and indeed is grounded in, the nonlocality of standard qm, so that this is not a reason to dismiss Bohmian qm.

Last edited:
Ken G said:
An example I point to is the reversibility of the equations of physics. In most elementary interactions, there is no arrow of time, so the difference between a cause and an effect is arbitrary. That difference doesn't come from the elementary physical interactions, it comes from the way we weave those pieces into a coherent description of what is happening. Correlation is empirical and objective, causation is rational and subjective. Even if I tell a story where a person shoots another with a gun, someone else could tell the story that the shooter was compelled to the act by the necessity of the fact that the other person got shot. It's just a different story, not likely to hold up in a court of morality, but perfectly good physics.

Wouldn't you acknowledge though that with the laws of thermodynamics as we know them, being as ubiquitous as they are, that in fact, there is always an arrow of time, and it only points one way (i.e. entropy only increases or stays constant). I'm not sure if you can argue that nature could take place backwards... Could you explain what you mean here.

jfy4 said:
Wouldn't you acknowledge though that with the laws of thermodynamics as we know them, being as ubiquitous as they are, that in fact, there is always an arrow of time, and it only points one way (i.e. entropy only increases or stays constant). I'm not sure if you can argue that nature could take place backwards... Could you explain what you mean here.
The second law of thermodynamics says that irreversible things cause entropy to increase, but that already presupposes a "sign" to the arrow of time. In reality, thermodynamics does not actually give a sign to time, because if we reverse the sign, we simply say that entropy decreases. All the theorem really says is that you cannot have one process with entropy increasing going on in one closed box, and another process with entropy decreasing going on alongside it in another closed box. The sign of the arrow comes from the way we experience time, not from thermodynamics, and we might be fooling ourselves. (I'm not saying I think the arrow really goes the other way, I'm saying I don't think the sign of the arrow comes from what we are studying, I think it comes from how we think.)

Last edited:
Ken G said:
The second law of thermodynamics says that irreversible things cause entropy to increase, but that already presupposes a "sign" to the arrow of time. In reality, thermodynamics does not actually give a sign to time, because if we reverse the sign, we simply say that entropy decreases. All the theorem really says is that you cannot have one process with entropy increasing going on in one closed box, and another process with entropy decreasing going on alongside it in another closed box. The sign of the arrow comes from the way we experience time, not from thermodynamics, and we might be fooling ourselves. (I'm not saying I think the arrow really goes the other way, I'm saying I don't think the sign of the arrow comes from what we are studying, I think it comes from how we think.)

Curious,

To make sure I understand you right, are you saying it would be possible to interpret a muon (an elementary particle) decay as not taking place in say, the decay into an electron and two neutrinos, and instead that the existence of an electron and two neutrinos prompted the decay of a muon?

Exactly. Indeed, it is often said that antiparticles can be viewed as regular particles going backward in time, and matter/antimatter annihilation is just a single particle turning around in the time dimension. At the most elementary levels, the time reversibility of the basic equations seems to allow all these interpretations. (There are some weak decays that don't time reverse, but CPT symmetry may be maintained, so you have to also switch left and right, along with the antiparticles going backward in time. The technicalities are beyond my knowledge.)

ThomasT said:
I don't think that the observed arrows of time are in any way illusory, and that it's one of the problematic aspects of modern physics that a fundamental dynamical law accounting for the observed arrows of time isn't yet a part of physics.

I agree with you, however, what about the importance of the Lorentzian signature in say, QFT, and GR. If one uses a Euclidean signature, as far as I know, the results are not consistent with observation, regardless of interpretation...

I would interpret this as that whatever is happening with time is certainly unique.

ThomasT said:
I don't think that the observed arrows of time are in any way illusory, and that it's one of the problematic aspects of modern physics that a fundamental dynamical law accounting for the observed arrows of time isn't yet a part of physics.
It really depends on what one means by "illusory." The term can have the connotation of "a mistake", like thinking a mirage is water, or it can just mean "developed in the mind", like the way a movie looks like motion but is really just a sequence of still shots. I'm using it in the latter way. I think the sign of the "arrow of time" (to me, the arrow itself is simply the fact that if entropy increases in one closed box, it will in another too, or if it decreases in one closed box, it will in another too) comes from our minds, so it is that type of "illusion." But it's not a mistake that needs fixing-- it is fine for us to perceive the world through that filter, unavoidable really, and I agree with you that incorporating this perception into our laws may leave some unfinished work.

But getting back to the sign of the arrow, let me ask you this. If you see a movie that shows a thousand coins of random orientation, and the floor shakes, and they all flip to "heads", what do you conclude?

A natural conclusion would be that you are seeing a movie run backward, because it would be vastly unlikely for that to happen if the movie were running forward. But if you dig deeper, you realize that the only reason you can reach that conclusion is that you have made certain assumptions about what is likely. You have assumed that it is not improbable that someone has intentionally placed all those coins as heads, then shaken the floor, so you are imagining that you are looking at an open system. But an open system doesn't have all the information available to you, so if you are told that you are looking at a closed system, a bunch of coins that have been in that room for a billion years, flipping around with no interference from anyone, then you have all the information about that movie other than the direction of the arrow of time. And now you find an interesting thing-- you cannot tell which way time is going! That's because it is just as unlikely to have a random bunch of coins flip all to heads, as it is to find a room filled with all heads in the first place. So either way you run the movie, you know you are looking at a very special event that must have been selected from a very long film, with no way to know which way time is going.

In other words, the sign of the arrow does not by itself stem from statistical mechanics, that only appears when we can make certain additional plausibility assumptions about the constraints on what we are seeing. Those constraints are special to our situation, and we shove a lot of what is happening in those open systems under the rug to tell the stories we tell. That's why the sign of the arrow is a construct of our processing, not something that is actually in nature, and that's why our arrow has a sign even though the laws of elementary interactions are time reversible.

jfy4 said:
I agree with you, however, what about the importance of the Lorentzian signature in say, QFT, and GR. If one uses a Euclidean signature, as far as I know, the results are not consistent with observation, regardless of interpretation...

I would interpret this as that whatever is happening with time is certainly unique.
Yes, but note the Lorentzian signature applies to the square of the time, so although it signals a unique importance to proper time, and it imposes a time ordering for proper time, it does not impose a sign to that ordering.

jfy4 said:
I agree with you, however, what about the importance of the Lorentzian signature in say, QFT, and GR. If one uses a Euclidean signature, as far as I know, the results are not consistent with observation, regardless of interpretation...
I'm not knowledgeable enough about those theories to comment on your question. But these are after all simply mathematical constructions. Isn't our experience, ie. qualitatively apprehensible reality, the evolution of configurations of ponderable objects in 3D Euclidian space? Is there some reason to think that laws/principles governing the reality underlying our sensory experience are somehow essentially different than the laws governing the reality of our sensory experience? Or is it more reasonable to assume that there are fundamental dynamical laws (maybe even just one single, say, wave mechanical law) that the apparently unique organizing principles seen at various behavioral scales have evolved from?

jfy4 said:
I would interpret this as that whatever is happening with time is certainly unique.
I don't know what you mean by this. Time is our indexing of configurational change. Do you mean that any 'snapshot' of the evolution of our universe is a unique configuration? I suspect you mean something else.

Ken G said:
At the most elementary levels, the time reversibility of the basic equations seems to allow all these interpretations.
The basic equations of motion allow "all these interpretations" only in the sense that their time symmetry is logically independent from the question of time reversibility in nature.

Ken G said:
(I'm not saying I think the arrow really goes the other way, I'm saying I don't think the sign of the arrow comes from what we are studying, I think it comes from how we think.)
On the contrary, how we think (which would include the basic equations of motion, ctc's, etc.) includes all sorts of exotic creations whereby just about anything is possible, whereas what we're studying, ie., our objective experience, suggests that some artifacts of otherwise quite useful mathematical constructions are most reasonably thought of as not possible.

Ken G said:
Yes, but note the Lorentzian signature applies to the square of the time, so although it signals a unique importance to proper time, and it imposes a time ordering for proper time, it does not impose a sign to that ordering.
Our ordering/indexing of unique configurations is the arrow of time. Isn't it? What would the imposition of a sign to the ordering of our experience tell us that we don't already know from the ordering itself?

Ken G said:
In other words, the sign of the arrow does not by itself stem from statistical mechanics, that only appears when we can make certain additional plausibility assumptions about the constraints on what we are seeing. Those constraints are special to our situation, and we shove a lot of what is happening in those open systems under the rug to tell the stories we tell. That's why the sign of the arrow is a construct of our processing, not something that is actually in nature, and that's why our arrow has a sign even though the laws of elementary interactions are time reversible.
I think I understand your view. It's interesting, but I disagree with it. Of course it's reasonable to infer that there is a deeper reality underlying our sensory experience. I also think it's reasonable to assume that the deep reality and our sensory reality have evolved from the same fundamental underlying dynamical laws/principles. This entails that the arrow of time that's evident via our experiential indexing of configurations/events corresponds to the evolution of nature.

If we view certain sequences of events, like elementary interactions, in isolation, then we might say that they're time reversible. But that's an artificial view, because they're not isolated. They're part of the evolution of nature, which has an objectively documented order which we call the arrow of time.

I don't see any compelling reason to think that the way we process sensory data, and therefore our objective record of the evolution of nature, doesn't correspond to the actual evolution of nature.

ThomasT said:
On the contrary, how we think (which would include the basic equations of motion, ctc's, etc.) includes all sorts of exotic creations whereby just about anything is possible, whereas what we're studying, ie., our objective experience, suggests that some artifacts of otherwise quite useful mathematical constructions are most reasonably thought of as not possible.
Let me clarify what I mean by "the way we think." You are right that our minds are capable of "flights of fancy," such as my imagining a different sign to the arrow of time. But by "the way we think", I mean the thought patterns that have been reinforced as successful by our situation. You seem to come from the thesis that if "the way we think" is reinforced by imagining a certain sign to the arrow of time, then there must be a good reason for that, and that good reason must be essentially that it is true in some deeper or more absolute sense. That is exactly the hypothesis I am calling into question. Although I do see it is implausible that we somehow got the sign backward from "the truth", what I am suggesting is that the very idea that there is a "truth" to the sign of the arrow of time, in some absolute sense, is what we should be skeptical of. Just because we find advantage in our circumstances to assign a sign to the arrow, does not necessarily imply that there is any particular meaning to that sign, beyond the simple fact that we find advantage in imagining it. The equations of physics paint a very different picture, hinting to our role in creating a kind of "fiction" about the nature of time.

In other words, fictions can still be useful to us, and I'm sure you can see some ready examples of this with minimal thought-- though which are the truths and which are the fictions is a matter that generates a significant amount of disagreement among people. For example, some view the idea that we have a soul that transcends our physical form is a kind of fiction, while others may get even more radical and assert that even the concept of an identity is a fiction-- those who would call into question even Descartes' seemingly unassailable "I think therefore I am." Some fictions are just more useful than others, but the pure empicist says only what you measure is real, and you never measure a sign to the arrow of time, not without passing it through an interpretive filter.

Last edited:
ThomasT said:
Our ordering/indexing of unique configurations is the arrow of time. Isn't it? What would the imposition of a sign to the ordering of our experience tell us that we don't already know from the ordering itself?
I should have said "sequencing" not "ordering". You are right that "order" already implies a sign. I see it as essential to the concept of proper time that it have a sequence, but not that it have an order.

ThomasT said:
Of course it's reasonable to infer that there is a deeper reality underlying our sensory experience. I also think it's reasonable to assume that the deep reality and our sensory reality have evolved from the same fundamental underlying dynamical laws/principles.
This is the crux of the matter, and I suspected you were holding this view. I don't say it's right or wrong, but I suggest an alternative possibility-- the entire idea that the universe "obeys laws" as its raison d'etre is an artificial construct of how we think about the universe. There's no such thing as a "fundamental underlying principle", because it is a category error: principles are thoughts, they are templates that the thinking brain lays against the reality to simplify it. The reality itself can't actually function that way. I'd say the same thing about numbers, to give an example-- numbers don't exist in the reality, they exist in how we think about the reality.
If we view certain sequences of events, like elementary interactions, in isolation, then we might say that they're time reversible. But that's an artificial view, because they're not isolated. They're part of the evolution of nature, which has an objectively documented order which we call the arrow of time.
That is certainly one interpretation of the meaning of "artificial", but I would like to make the opposite argument. I see the elementary processes as what cannot be artifiical because they are elementary enough for us to conceive them in a fairly pure way, whereas the global constraints that act upon them and condition them as part of larger systems are so complex and intractable that we must simplify them and interpret them in ways that fit inside our heads, which is the source of artificialities like the sign of the arrow of time.
I don't see any compelling reason to think that the way we process sensory data, and therefore our objective record of the evolution of nature, doesn't correspond to the actual evolution of nature.
The reason to suspect that is our experience of nature is highly limited, and much of what our brain does is to throw out what nature is doing, more so than notice it. Survival demands that we ignore a spectacular fraction of the sensory inputs around us, and even if it didn't, we only have our five senses and our hugely aggregated sums of processes that contribute to those senses. Yours was a common attitude before quantum mechanics and relativity, but both of those theories seriously questioned just how close our simplified thought processes about reality are to the actualities we can test in more sophisticated experiments than those we experience in the environment our brains evolved under.

@KenG

You are saying we cannot distinguish between cause and effect precisely because we cannot measure an arrow of time, correct?

If that is so, consider the following:

Consider a model of physics that places the causal relationship between a muon and an electron and two nutrinos as "backwards". That is, an electron and two neutrinos cause a muon. While this causal relationship satisfies and accommodates what is observed, as a condition statement: if there is an electron and two neutrinos, then there will be a muon, is false (you can just have an electron and two neutrinos that have nothing to do with each other). That is, this physical model only accommodates the causal relationship, while, the classical physical model predicts it. That is: if there is a muon, then there will be an electron and two neutrinos (there are two decay modes but you can see that is irrelevant here).

So I guess while it's possible to interpret effect as being the cause and visa versa, it would not be a strong physical theory in some cases... Which it sounds like you have acknowledged judging from your posts, I just have to see if I have a handle on this idea.

Ken G said:
I should have said "sequencing" not "ordering". You are right that "order" already implies a sign. I see it as essential to the concept of proper time that it have a sequence, but not that it have an order.
"Sequencing" and "ordering" mean the same thing.

• Quantum Physics
Replies
58
Views
868
• Quantum Physics
Replies
6
Views
2K
• Quantum Physics
Replies
18
Views
2K
• Quantum Physics
Replies
47
Views
4K
• Quantum Interpretations and Foundations
Replies
139
Views
6K
• Quantum Interpretations and Foundations
Replies
54
Views
4K
• Quantum Interpretations and Foundations
Replies
226
Views
19K
• Quantum Interpretations and Foundations
Replies
37
Views
2K
• Quantum Physics
Replies
36
Views
8K
• General Discussion
Replies
190
Views
10K