I A question about the delayed choice quantum eraser experiment

[eyeseefan]

This delayed choice quantum eraser experiment captured my attention and after deep thought about it, I find it really startling and totally punishing to my common sense. My question is: Can you predict the future, say, a human's mind, using the quantum eraser? I imagine a modified version of the initial quantum eraser experiment. In the initial experiment, there were detectors that are able to tell us "which-path" information and detectors that obscure the "which-path" information which act as quantum erasers.
Let's say the experiment is set up so that there is a switch, the switch that controls whether the photons will be directed at two detectors that we are able to differentiate(we are able to tell which-path) and whether the photons will be directed to a quantum eraser(no which-path information).
Now we shoot a collection of photons to the double slit and separate them into entangled particles, one of which(let's say particle A) will go to the screen, and the other(let's say B) to the detectors. Let's make the other entangled particle(particle B) travel a really long path so that maybe only 1 hour(a bit exaggerated) after the experiment(after the pattern is detected on the screen), this entangled particle will be able to reach the detectors controlled by the switch. During this "1 hour", we are totally free to control the switch, but an pattern is already formed on the screen. Let's say we see an interference pattern on the screen before this "1 hour", aren't we able to turn the switch during this "1 hour" to force the entangled particles arrive at the detectors which we are able to tell "which-path"? Isn't this a paradox? Or does it imply that free will doesn't exist and our human mind(whether we choose to turn the switch or not) is predictable?
I must have missed something in my argument. Can anyone help to point out?

 

[zonde]

I must have missed something in my argument. Can anyone help to point out?

Yes, you are missing the part about coincidence counting.

During this "1 hour", we are totally free to control the switch, but an pattern is already formed on the screen.

There is never interference pattern on the screen. Interference is observed only when part of the "clicks" in detectors is considered.
The experiment goes like this. You emit photon pairs with low intensity so that there is no problem telling apart the pairs by looking at the "clicking" time in detectors. Then you take clicks in one of the detectors from side B and take only matching (by time) clicks in detector at side A. And in this subset you can or can't see the interference pattern.

 

[georgir]

A surprisingly common mistake.
And it's a good reason I do not consider quantum eraser's interference pattern as equally interesting as the one in the usual double-slit experiment. In the normal experiment, the interference pattern is bright in places where neither slit's solo lump pattern is bright - i.e. photons often go to places you wouldn't expect them to. In the eraser setup, the interference is always less bright than the lump pattern - the lump pattern is effectively just decomposed into two complementary interference patterns - photons go to the same old boring places and you just found an interesting way to group them and find a pattern where there is none.