Yesterday, I described a classic physics experiment in which what might have happened but didn’t could have a real, observable effect. What I’m going to talk about today is an extension of this. I’m a bit nervous about attempting this, because it’s complicated. But it’s more startling than the EPR experiment I mentioned in a previous comment because the effect is not simply faster than light, but explicitly backwards in time. If you can follow this, nothing else I say should bother you.
It is natural to want to know what is going on when a single photon passes into the system, seems to travel both routes at once, and thereby always comes out the same way. By which route does it really go? We already know that if we stick a detector into its path to absorb it, the effect disappears. But is there a way we could figure out which way it went without stopping it, and thereby see what’s going on?
Well, there is a sort of material, one that has a nonlinear response to the incoming wave, that is able to split a photon into two smaller photons. So we could put a splitter into each path, and then do our tests on one of the smaller photons while allowing the other to continue on its way. We have one photon going round by the two routes as normal, and one test photon coming out of one side or the other.
It will come as no surprise to discover that if we simply look at the test photon to see which side it came from, we do indeed see it on one side or the other, but the correlation disappears. Photons come out at D or E equally often. Well, we did mess with it, after all.
But if we change our mind and take the two test photon outputs and combine them using another half-silvered glass, in such a way – and this is important – that we can no longer tell which side it came from, we can restore wave interference at the original detectors again. (Although of a more subtle sort – the effect is scrambled and we need the results of the test observation to unscramble them).
So we have a way of turning the effect on and off. If we peek, the mysterious effect disappears. And if we close our eyes, it starts acting mysteriously again. This is called the Quantum Eraser experiment – we can erase the information in the test photon that breaks the original effect.
There is a reasonable way of explaining this. The decision is made when we make a measurement. If we peek, then the wave collapses to a single photon going down one definite route, and then making a random choice alone at the second half-silvered glass. If we don’t peek, then we still have waves travelling both routes and can still get interference.
But this is where we can upset the timing. Because we can lead our test photons away to a considerable distance, possibly store them in a ring of mirrors, and delay our choice until after the original photon has been recombined and observed. The future can affect the past.
We can choose to peek at which way the photon went after it has been detected. But if we don’t peek, we can still see wave interference at the original detectors.
It would appear that the choice between mysterious correlation or the photon taking a definite route one way or the other – whether it is a wave or a particle - extends to times even after photons have been detected. It’s a subtle effect, and often misunderstood.
I love this kind of stuff; unfortunately, I can only manage to get my head (nearly) around it when I’ve had exactly the right amount of caffeine - enough to get my braincells working on overdrive, but not so much that I’m actually seeing after-images of each word as I read across the screen!
Everybody does. In fact, it’s been said that if you think you understand quantum mechanics, then you’ve missed something.
This bit you don’t really need to know in that much detail. What I wanted to do was talk about some of the evidence, such as it is, for the Everett-Wheeler interpretation of quantum mechanics, so that I could go on to talk about multiple time-lines, histories cancelling, quantum time travel, and that time-travelling Higgs.
If you don’t have any background in the physics, it’s too easy to get the wrong idea.
We shall see.
I suppose that there is not enough time to do this in reality, but as a thought experiment, what happens if we did this:
Feed the test photons over such a long distance that we have minutes before they reach the recombiner. Observe the results at D and E, then activate or deactivate the combiner, contrary to the results from detectors at D and E. That is to say, we choose to peek or not peek after the observation.
(I accept that I could easily have not understood the experiment, and that the above is a ridiculous thing to say). Even if it’s not practical to do, doesn’t the fact of effect preceding cause give rise to the possibility of paradox? (I’m quite sure their would be no paradox if this experiment were possible, the universe is a slippery bugger).
Has this experiment actually been done? Also, what Andy said. And this needs diagrams. But it is awesome.
Andy,
Yes, you could. The experiment has actually been done by Scully et al. and found to work, although they only did it over about 50 metres and the gap was in the neighbourhood of a few nanoseconds, rather than minutes.
There are no paradoxes in this case. It’s only when observers get together again and compare notes that the interference pattern can be descrambled and seen. Since it can all be explained by the Everett-Wheeler interpretation (see next post), which doesn’t have any of this spooky FTL stuff, it can be explained without contradictions or backwards causation. The only thing that is ’caused’ retrospectively is the point at which the wave collapses into a particle, and that, as Everett said, has no observable consequences.
Rob,
I might have a go at adding a diagram for you at the weekend. No promises, mind.
Ah. What an excellent explanation. I think I see. There is a result, which can only be interpreted with reference to another result. In that sense the experiment I propose is impossible, as you can’t do anything with the result you have until you look at the delayed result.
When you do, miraculously the results always agree with each other, even though the first one couldn’t have know what was going to happen with the second.
The question I have is as follows:
Suppose we send several thousands of photons through the splitter.
Suppose also that the detector of the other photons is far away, and no decision has been yet made regarding the way they are going to be observed.
What will be the interference pattern of the photons which have already passed the slits and reached the detection surface?
According to the interpretation of the experiment, the pattern says what in the future will be done regarding the detection setup of the other photons?!
So something which has not yet happened determines what happens now!
And what happens now, predicts precisely what will happen at a far place in the future.
Surely the observer in the first place cannot (since he cannot move faster than the speed of light) influence the result at his place, but an observer at the other place can, be varying the observation setup, send a signal to the observer in the past, and do it instaneniouly!!
Does this mean, that communication from the future to past is possible and at a speed higher than the speed of light?
I understand it does not contradict causality, since the communication is between two space like (or light like) points, but ut is good to know that some extraterrestial in the future can warn us now that it might be not a great idea to visit him and save us all the trouble and time traveling to his planet!?
Does it all make sense?