Delayed Choice Quantum Eraser Question

[JPBenowitz]

I've heard we can rule out retro-causal effects if we think of the photon as both being a wave and a particle simultaneously. However I understand wave-particle duality as being the sum of all possible paths taken by a particle to a detector whilst being impossible to predict individual paths. So how exactly can we rule out retro-causality?

 

[soothsayer]

Retro-causality being the idea that the effect generated the cause? That is that a particle being measured at point A caused it to go through path A?

Consider the two slit experiment: Unless we directly measure which slit the particle goes through, each particle behaves like a wave and will leave an interference pattern. That pattern isn't collapsed simply by looking at it. We can either say: we observed a particle going through slit A subsequently ending up at point A, or we can say that the particle was not observed, maintained its wavelike nature, and went through both slits "at the same time" to generate an interference pattern. There's no violation of cause and effect.

Let me know if I have misunderstood you.

 

[alexepascual]

I think it is not possible to talk about a topic like this if the problem is not presented in more detail and explaining in which context this may have some meaning.
From what I know, this may refer to the experiment carried by Birgit Dopfer under Anton Zeilinger's supervision. Some people have thought that a modification of that experiment could be used to transmit messages "into the past". The main proponent of that idea has been John Cramer (the creator of the transactional interpretation) who labeled this messaging back in time as "retrocausality". Most of the physics community thinks (for various reasons) that this could not be done. Cramer has designed a modified version of Dopfer's experiment and has been working for years on it but until now it appears he has not been able to get meaningful results.
With respect to the double slit experiment you mention, as far as I know, retrocausality would not have anything to do with it. The experiments where retrocausality may be be thoght to be possible typically involve two particles that are entangled to each other.

 

[alexepascual]

JPBenowitz: The other context in which some people (not phycisists as far as I know) have mentioned (including in this forum) that something like this could happen is in the EPR experiment. In that case, even though the outcomes at each end are correlated to each other, they are still random. So you can't use this to send messages. Even if you did, in which case you could be sending messages faster than light, that would not necessarily imply that you can send them "back in time"
Another example (more similar to a double slit experiment) is Wheelers delayed choice experiment, which he exemplified by thinking about light from a distant galaxy going both ways around a gravitational lense and being able to decide, many years after the light has gone through the lens, if it went one way or both ways. In this case, even though it is interesting and paradoxical, the photon is not detected in any way on it's path, so we can't talk about retrocausality either. In other words, we can't use this method to influence macroscopic events in the past.

 

[JPBenowitz]

I think it is not possible to talk about a topic like this if the problem is not presented in more detail and explaining in which context this may have some meaning.
From what I know, this may refer to the experiment carried by Birgit Dopfer under Anton Zeilinger's supervision. Some people have thought that a modification of that experiment could be used to transmit messages "into the past". The main proponent of that idea has been John Cramer (the creator of the transactional interpretation) who labeled this messaging back in time as "retrocausality". Most of the physics community thinks (for various reasons) that this could not be done. Cramer has designed a modified version of Dopfer's experiment and has been working for years on it but until now it appears he has not been able to get meaningful results.
With respect to the double slit experiment you mention, as far as I know, retrocausality would not have anything to do with it. The experiments where retrocausality may be be thoght to be possible typically involve two particles that are entangled to each other.

Sorry I didn't post the experiment it is this one: http://arxiv.org/abs/quant-ph/9903047

 

[alexepascual]

Thanks for posting the link to the experiment.
It would be good to know:
1) How you define (carefully and precisely) "retrocausality"
2) How you think this experiment could be used to implement it.
I must say that one of the arguments that has frequently been used to deny the possibility of retrocausality or time travel is the contradiction we face when we can after the fact provoke some event in the past that would imply a change in the present when the present is something we are already experiencing. If there is a way around this with the many worlds interpretation I don't know.
There is also an argument that has been presented as a proof that faster-than-light communication is not possible by (if I recall correctly) Eberhard. But some people have considered that proof as circular reasoning. Even though this proof relates to faster-than-light communication and not to "retrocausality" I think there might be a connection, depending on how such retrocausality is proposed to be achieved.

 

[rkastner]

I address delayed choice and quantum eraser/delayed choice experiments in terms of the Transactional Interpretation in my new book:

http://www.cambridge.org/us/knowledge/discountpromotion/?site_locale=en_US&code=L2TIQM

 

[San K]

Sorry I didn't post the experiment it is this one: http://arxiv.org/abs/quant-ph/9903047

DCQE (delayed choice quantum eraser) does not prove Retro causality. The interference patterns are seen only after comparing both the entangled photons and then filtering to get interference pattern. The past is not changed.

As an example let's say we are able to decipher some ancient language. We now have more information about the past, however the past is not changed.

Another example if we get more information about some distant galaxy or big-bang etc, we have more knowledge about the past, however we have not changed the past.

 

[rkastner]

I agree -- claims of explicit retrocausation in q.eraser are misleading and based on ignoring the correlation between each idler and signal photon. As the idler's state is collapsed at a particular detector, that influences the signal photon's detection probabilities; and vice versa. I explain this in detail in Chapter 5 of my book.

RK

DCQE (delayed choice quantum eraser) does not prove Retro causality. The interference patterns are seen only after comparing both the entangled photons and then filtering to get interference pattern. The past is not changed.

As an example let's say we are able to decipher some ancient language. We now have more information about the past, however the past is not changed.

Another example if we get more information about some distant galaxy or big-bang etc, we have more knowledge about the past, however we have not changed the past.

 

[alexepascual]

I wonder Ruth how you would define retrocausation. I understand you don't believe it is possible (I don't either), but even if we don't believe in it, we can make clear what we mean by it. Otherwise, how could we deny something we can't define?

 

[rkastner]

Hi Alex, swamped right now but will try to give a short answer with clarification to follow if you need it.

In TI a real event corresponds to an actualized transaction. Once that has occurred, it can't be changed by anything in the future.

Claims such as a 'future choice affecting a past measurement result' are spurious in my view because a measurement result is an actualized transaction. There is sub-empirical retrocausality in that advanced confirmation waves play a role in setting up a set of incipient transaction, only one of which can be actualized. But once actualized, there's nothing about it that can be changed. The past is open to 'revision' only if it is indeterminate, as in delayed choice experiments.

 

[DrChinese]

In TI a real event corresponds to an actualized transaction. Once that has occurred, it can't be changed by anything in the future.

Claims such as a 'future choice affecting a past measurement result' are spurious in my view because a measurement result is an actualized transaction. There is sub-empirical retrocausality in that advanced confirmation waves play a role in setting up a set of incipient transaction, only one of which can be actualized. But once actualized, there's nothing about it that can be changed. The past is open to 'revision' only if it is indeterminate, as in delayed choice experiments.

Clearly, if you move your time pointer so that all elements are in the past, there does not seem to be a retrocausal* element (how could there be if everything relevant happened in the past). However, at the time you are doing certain measurements, there are both future events and past events that are part of the equation explaining results you see now.

*Retrocausal being treated here as a label for this interpretation, you call also refer to it as Time Symmetric. Whether or not anything from the FUTURE is "causing" anything in the past is debatable, as you mentioned. I would also debate the idea that anything in the PAST is "causing" the specific measurement, regardless of interpretation. We don't know, and there is no indication, that ANYTHING causes specific outcomes.

 

[rkastner]

Thanks Dr. Chinese. However there are certain distinctions in my proposal (PTI) that your discussion doesn't appear to take into account. One of these is that in PTI we do not have a 'block world'. Rather, the future is only possibilities (absorbers are fundamentally quantum field currents, which are treated in PTI as non-actualized entities). Also, in PTI, elements of the past can indeed remain indeterminate until a confirmation wave is generated that retrocausally contributes to creating a set of incipient transactions, of which only one is actualized. As you note, there is no determinate 'cause' of the actualization of anyone of these. Their probabilistic weights are given by the Born Rule. Indeed the set of weighted incipient transactions is precisely the physical system to which the Born Rule refers (i.e. the set of incipient transactions corresponds to the density matrix for the measurement system in von Neumann's "Process 1"-- in which a pure state transforms to a mixed state upon 'measurement' -- TI defines what prompts this transformation). Detail of this are given in Chapter 3 of my new book, available here: http://www.cambridge.org/us/knowledge/discountpromotion/?site_locale=en_US&code=L2TIQM

Further details are of course presented in the book concerning this proposed ontology.

 

[alexepascual]

*Retrocausal being treated here as a label for this interpretation, you call also refer to it as Time Symmetric. Whether or not anything from the FUTURE is "causing" anything in the past is debatable, as you mentioned. I would also debate the idea that anything in the PAST is "causing" the specific measurement, regardless of interpretation. We don't know, and there is no indication, that ANYTHING causes specific outcomes.

DrChinese: If I am not mistaken, when you wrote this you identified TI with retrocausality. This may be correct if you stick to Cramer's ideas. But it looks (I haven't read her book yet) that Ruth Kastner has an interpretation which is based on Cramer's interpretation but denies retrocausality. Maybe she can correct me if I am wrong. My question above was not so much directed at judging retorcausality as something that happens or that does not happen, but just to get a definition of what the concept of retrocausality implies. The concept does indeed exist (I understand that it could mean different things to different people). So I am not asking about the "territory" but about the "map".
With respect to your observations in this quote, I notice that you wrote PAST and FUTURE in capital letters. In a Newtonian universe, with absolute time and the concept of instantaineity valid in all reference frames, I think PAST and FUTURE are fairly clear and don't need much explanation. On the other hand, in a Minkowsky-type universe, I think it is useful to clarify these concepts when they are used.
With respect to your second observation (that something in the PAST does not "cause" something in the FUTURE) I would disagree. First, in classical physics, events in the PAST do cause events in the future. Second, in quantum mechanics, when you have a superposition of states and you make a measurement, after that point, you will get repeatedly the same value, and this state could be said to have some cause in the macroscopic world. For example, if I get eigenvalue "a" I stay in the lab, if I get "b" I go home. I hope I didn't misinterpret what you said.

 

[alexepascual]

Thanks Dr. Chinese. However there are certain distinctions in my proposal (PTI) that your discussion doesn't appear to take into account. One of these is that in PTI we do not have a 'block world'. Rather, the future is only possibilities (absorbers are fundamentally quantum field currents, which are treated in PTI as non-actualized entities). Also, in PTI, elements of the past can indeed remain indeterminate until a confirmation wave is generated that retrocausally contributes to creating a set of incipient transactions, of which only one is actualized. As you note, there is no determinate 'cause' of the actualization of anyone of these. Their probabilistic weights are given by the Born Rule. Indeed the set of weighted incipient transactions is precisely the physical system to which the Born Rule refers (i.e. the set of incipient transactions corresponds to the density matrix for the measurement system in von Neumann's "Process 1"-- in which a pure state transforms to a mixed state upon 'measurement' -- TI defines what prompts this transformation). Detail of this are given in Chapter 3 of my new book, available here: http://www.cambridge.org/us/knowledge/discountpromotion/?site_locale=en_US&code=L2TIQM

Further details are of course presented in the book concerning this proposed ontology.

Ruth, I was trying to get a definition for retrocausality that would be based on observations at the macroscopic level and not dependent on interpretation. Let's start with plain causality. So, let's say we have two locations in space and define events at particular times in each location. (Lets call these events A and B) For some reason, we think that event A causes event B. It could be because we repeat the experiment in similar experimental situations and with the same relative position in space-time and every time we get A we also get B. But that may not be enough as it would just imply a correlation. (like in EPR) We could specify that we need to have some variable that we (human beings) can "freely" manipulate in order to get A. Or that we have some random number generator which generates the number without the possibility of B influencing A. Now, let's say that we have two events where we agree that A "causes" B. How do these events need to be located in space-time for this to be defined as retrocausality?

 

[rkastner]

Re Alex's question: " let's say that we have two events where we agree that A "causes" B. How do these events need to be located in space-time for this to be defined as retrocausality?"

In my view you could never say that with certainty at the quantum level, but (for the purpose of definitions) at the classical level, A would be in the future light cone of B for that to be called 'retrocausality'.

 

[alexepascual]

Re Alex's question: " let's say that we have two events where we agree that A "causes" B. How do these events need to be located in space-time for this to be defined as retrocausality?"

In my view you could never say that with certainty at the quantum level, but (for the purpose of definitions) at the classical level, A would be in the future light cone of B for that to be called 'retrocausality'.

I agree with that. And I think that even in quantum mechanics, to talk about retrocausality, we would have to look at the results of measurements, which always consist of macroscopic (therefore calssical) states. When we look at things like the wavefunction, or the two types of waves in the Transactional Interpretation, things are not so clear because these waves are subject to interpretation. But the results of measurements are much more clear, how we got a certain value is a different story, but at least we can look at results of measurements, compare, make calculations and get some conclusions.
Ruth, I understand you are busy with other things, but when you have some time, I would like to know if you agree with the idea that faster-than-light communication (again regardless of the possibility of this phenomenom being real or not) would imply retro-causality. In my opinion it would not.

 

[rkastner]

Alex, I think the ability to send a signal superluminally wouldn't imply retrocausality as we've defined it here, because there is really no well-defined direction of the signal in a causal sense.

 

[San K]

As the idler's state is collapsed at a particular detector, that influences the signal photon's detection probabilities; and vice versa. I explain this in detail in Chapter 5 of my book.

RK

Good point rkastner. May I add that:

The idler's state during collapse is not controllable/pre-determinable

 

[rkastner]

San K: indeed.

 

[alexepascual]

Even though you can't decide the particular value you'll get for the observable you are measuring on the idler photon, you can decide which observable to measure. Don't you think that that would affect the statistics on the other side? Even if this was possible, given the fact that both measurement events are space-like, this sould not imply retrocausality, but it would imply faster-than-light signaling. I know most people think that in Dopfer's experiment coincidence counting is the only way to get an interference pattern, but I am not convinced yet.

 

[alexepascual]

Ruth, If after making a measurement, or having some quantum state recorded in a stable way (decoherence or "collapse" in the language of CI) you have a definite state, then the set of all these outcomes in the past, present and future could be mapped on a diagram, constituting what could be called "a block model". Wouldn't it ? This is if we think that there is only one outcome after measurement, which all one-world interpretations (not only TI) propose.

 

[alexepascual]

Alex, I think the ability to send a signal superluminally wouldn't imply retrocausality as we've defined it here, because there is really no well-defined direction of the signal in a causal sense.

I agree. It happens that I have read some other places (I don't remember if it was on this forum or some article) people saying the opposite. So I was interested in knowing your opinion.

 

[rkastner]

Even though you can't decide the particular value you'll get for the observable you are measuring on the idler photon, you can decide which observable to measure. Don't you think that that would affect the statistics on the other side? Even if this was possible, given the fact that both measurement events are space-like, this sould not imply retrocausality, but it would imply faster-than-light signaling. I know most people think that in Dopfer's experiment coincidence counting is the only way to get an interference pattern, but I am not convinced yet.

As far as I know, every time someone claims to have devised an experiment that could do this, there is a problem with it; e.g. a contribution has been disregarded as negligible that really isn't. You always need a coincidence count in order to see such statistics.

Regarding sending a signal into one's own past: in order to do this using FTL signalling you have to collaborate with someone in a different inertial frame, both parties using FTL signalling. Wiki has a pretty good discussion of this here: http://en.wikipedia.org/wiki/Minkowski_diagram (under 'speed of light as a limit'). The problem is that each observer just sends a FTL signal to the other observer based on their observation of an event, but they have have no way of knowing whether that event was a 'signal' or not.

 

[rkastner]

Ruth, If after making a measurement, or having some quantum state recorded in a stable way (decoherence or "collapse" in the language of CI) you have a definite state, then the set of all these outcomes in the past, present and future could be mapped on a diagram, constituting what could be called "a block model". Wouldn't it ? This is if we think that there is only one outcome after measurement, which all one-world interpretations (not only TI) propose.

You can only put the determinate events on a ST diagram after the events have been actualized in spacetime. In my interpretation (PTI) the future really does not exist yet except as indeterminate possibilities and those are not spacetime entities. Since future events don't exist in spacetime you can't put them on a spacetime diagram. Similarly, any past situations that are not actualized (such as which path a photon took in a delayed choice experiment whose choice has not yet been made) cannot be put on a ST diagram either. This is an important issue and I discuss it in detail in Chapter 8 of my book.

 

[alexepascual]

You can only put the determinate events on a ST diagram after the events have been actualized in spacetime. In my interpretation (PTI) the future really does not exist yet except as indeterminate possibilities and those are not spacetime entities. Since future events don't exist in spacetime you can't put them on a spacetime diagram. Similarly, any past situations that are not actualized (such as which path a photon took in a delayed choice experiment whose choice has not yet been made) cannot be put on a ST diagram either. This is an important issue and I discuss it in detail in Chapter 8 of my book.

Ruth, I think this would make a very interesting conversation. I will give you my point of view perhaps tomorrow because I have some personal problems to take care now. I'll also reply to your other post at that time. Thanks, --Alex--

 

[alexepascual]

You can only put the determinate events on a ST diagram after the events have been actualized in spacetime. In my interpretation (PTI) the future really does not exist yet except as indeterminate possibilities and those are not spacetime entities. Since future events don't exist in spacetime you can't put them on a spacetime diagram. Similarly, any past situations that are not actualized (such as which path a photon took in a delayed choice experiment whose choice has not yet been made) cannot be put on a ST diagram either. This is an important issue and I discuss it in detail in Chapter 8 of my book.

Ruth, If you think that when a measurement is made and you get one of the possible outcomes, that represents an actualization and that is the only outcome that gets to be part of "reality" (if you don't have objections to using that word). We could imagine a universe composed of all those events. Even if there is no way for us to know what the events in the future will be, we know that each act of measurement will produce only one value for the observable being measured. In other words, we have two alternatives in representing the future 1) We look at "one history" or chain of events or 2) we can look at a branching scheme where all the possible outcomes of each measurement are included and each leads to different chain of events that in turn branch again when new "measurements" are made. I put "measurements" in quotes because actually we don't need a measurement carried by a physicsist in the lab but just an interaction between the quantum state with a macroscopic object that "collapses" the wave function. If you believe that a measurement produces only one eigenvalue, I don't see why we could not imagine a future that consists of all these unique autcomes, even if we can't predict now what the value will be.
We can also imagine that the weather this year will show some kind of definite evolution which we can't predict either. But we can imagine that at the end of the year we'll be able to put the actuall data in a diagram. We usually imagine a single future, not a series of possible ones. Of course if can consider the different alternatives and their probability in order to make some decision, such as what crop to plant now if that's what we do. But usually have the belief that the future will be one. I think that's the usual way to look at things. Of course there is the many-worlds interpretation where all possible paths of evolution count and are "real" and the consistent histories where we consider different histories but only one is part of reality. I'll appreciate your opinion on this and it would be nice to hear from Dr. Chinese.

 

[alexepascual]

Alex, I think the ability to send a signal superluminally wouldn't imply retrocausality as we've defined it here, because there is really no well-defined direction of the signal in a causal sense.

I already expresed my agreement with this, but I would like to add that part of the difficulty here may reside in the fact that most people use the term causality without having a very clear definition. In classical physics, of course for something to be a "cause" it has to be in the past light cone of what is considered to be the "effect", but if someone talks about "retrocausality", then we are not considering that order in time as the main characteristic of causality. We usually think that between the cause and the effect there is some "mechanism" and usually we think in terms of locality or in other words "no action at a distance". But if we can't look at the "mechanism", then maybe we can find some useful definition of causality in terms of the type of correlation between cause and effect. This is something that I have not investigated by looking for articles on it, but I can imagine that it would also apply to analysis of causal connections in medicine, biology and other subjects. Probably one of the elements that we would consider is that the "effect" would show an increase in frequency when we use our "freedom of choice" or some quantum random number generator that we assume is not influenced by the "effect". Tweaking the "effect" by some means should not influence the statistics of the "cause". I know I have expressed this in a not very clear way. It is because I am pondering this as I write. Perhaps you or one of our friends in the forum can throw some more light on this.

 

[rkastner]

Causality is a very slippery notion indeed. I do address this in Chapter 7 of my book.

 

[alexepascual]

Causality is a very slippery notion indeed. I do address this in Chapter 7 of my book.

I won't be able for the moment to get a hold of your book. Maybe you could give us your point of view summarized in a few sentences.
Thanks,
--Alex--

 

[rkastner]

you can find it here: http://transactionalinterpretation.org/

Scroll down a bit

 

[DrChinese]

you can find it here: http://transactionalinterpretation.org/

Scroll down a bit

I liked the PowerPoint presentation, Ruth. Nice job of delineating some of the differences between TI and CI.

 

[alexepascual]

you can find it here: http://transactionalinterpretation.org/
Scroll down a bit

Thanks Ruth, I'll take a look at the material in your website and let you know if I have questions or comments on it.

 

[rkastner]

Thanks Dr. Chinese! :)

 

[alexepascual]

Causality is a very slippery notion indeed. I do address this in Chapter 7 of my book.

Ruth, I have now read chapter 7. In it you say:
"As noted in Chapter 1, traditional approaches to measurement in quantum theory
inevitably end up needing to invoke an ‘observing consciousness’ in order to ‘collapse’
the wave function (or state vector) and bring about a determinate outcome, necessitating
speculative forays into psycho-physical parallelism"
I would like to understand if this means you see psycho-physical parallelism as a problematic concept and if this is so, why you think there is a problem with it.
Thanks.

 

[rkastner]

I view psycho-physical parallelism as problematic in the context of interpreting physical theory because it is so speculative, and because it does not appear to be based on any sort of physical theory. As Kent has noted, "we don't have a good theory of mind." In Chapter 1 I quote from Kent (2010): "“...the fact that we don't have a good theory of mind, even in classical physics, doesn't give us a free pass to conclude anything we please. That way lies scientific ruin: any physical theory is consistent with any observations if we can bridge any discrepancy by tacking on arbitrary assumptions about the link between mind states and physics.” (A. Kent, 2010, from Many Worlds? Everett, Quantum Theory, and Reality , p.21)