So I've been reading a lot about Bell tests. I have a question that I can't seem to find an answer to. The tests indicate that a photon decides to be horizontally or vertically polarized at the moment of detection. What are they basing this on? To make such a counter-intuitive assumption seems like it would require a lot of data, but having read z dozen papers on Bell tests by Aspect and others, I don't see it. It seems more likely that the photon has a polarization to begin with but it changes due to the detection method, thus causing an artificial correlation. If this were true, it would mean reality does exist but we erase it when we try to detect it.