Home › Forums › 2015 International Workshop on Quantum Foundations › Retrocausal theories › Are retrocausal accounts of nonlocality conspiratorial? A toy model. › Reply To: Are retrocausal accounts of nonlocality conspiratorial? A toy model.
Hi Travis and Dustin,
Thanks for the interesting discussion! I thought I’m chime in with a few points of my own…
Travis, on your point #2, you never mentioned another aspect of locality (really, a pre-requisite!): restricting models to those that have exclusively spacetime-local beables. To me, *that’s* the important thing that retrocausality offers, and I know this is important to you as well. Retrocausality is not the only path to such beables, as you well know, but one of the few paths still open. There are plenty of models that can be rejected on the grounds of having nonlocal beables, even if one considers retrocausality.
Once you’ve restricted models to this subset, I see further distinctions between models with superluminal-causation and zig-zag causality. You seem to be worried that the mere option of retrocausation blurs these accounts together in a way that can’t be properly distinguished. But differences include: 1) the different places in space-time where such a causal influence could in principle be blocked, 2) A different relationship with Lorentz invariance, 3) in an EPR-geometry, the ability to treat Alice and Bob on an equal footing (rather than having one of them influencing the other, they could both mutually cause the past hidden variables).
One last point: in that last sentence in (2) it almost sounded like you were asking retrocausal models to “rule out” certain *phenomena* as generally impossible. Obviously, a specific model can do this, but is it really so bad that the general concept of retrocausal theories doesn’t obviously (a priori) rule out any particular phenomena? (Sure, the “initial conditions + inside-lightcone-dynamics = future” framework rules out some phenomena, but this is not a plus, this is a terrible minus, because some of the ruled-out phenomena actually occur!)
Moving to the conspiracy issue (3), I have pretty strong feelings about this because (as you’ve both already discussed) there’s a widespread and unfair confusion between superdeterministic and retrocausal explanations already. Here what you’re ignoring in your China/Boston example is the influence of intermediate and external confounding factors.
Look at it from the forward-time perspective. Suppose I mailed identical letters to Boston and China, and then claimed that I am a past-common cause that might explain the correlation in *next year’s* Chinese tea price with *next year’s* Boston mosquitoes. This model could be ruled out on conspiratorial grounds having nothing to do with whether there was a possible past-common-cause. There are just too many confounding factors and too little past influence to explain any extensive or predictable correlation.
In your retrocausal example, the confounding factors are even more severe. Sure, there might be a tiny future-common-cause, but no non-conspiratorial model could use it to explain such large correlations in the past, for basically the same reason as the previous paragraph.
Now, sure, if there are *no* confounding factors, if two photons are beamed directly towards the same detector from different sources, retrocausal accounts generally call for correlations between the past hidden variables of those separate photons. But hopefully this is evidently more plausible (far less conspiratorial) than your example.
Your biggest point, (4), is of course the real outstanding problem. (Well, unless you throw in the problem that hardly anyone is willing to take retrocausality seriously in the first place!) But maybe part of this latter problem is the problem you mention: if doesn’t seem like there’s a way to explain the special status of future boundary conditions in a supposedly-universal theory, fewer people are going to consider it as a live option.
I think Dustin is right, though, that’s when analyzing simple few-particle toy models, it’s reasonable (in the framework of such models) to treat external interactions with much larger entities as effective boundary conditions on the smaller subsystem. But you’re worried about how such models scale up, and think that such stories smack of the usual measurement problem. And you’ve been writing long and detailed explications of this problem to me and Rod, while I haven’t properly been making the case that there might be an eventual solution. So I’ll try to write something up in the next day or so, to this effect, and see if it helps. At the very least, it might help refresh some of the key points in my own mind, since I’ve kind of put these issues on the backburner in my own research. So — more soon!