In Aspect 2002 - Bell's Theorem, the Naive view of an Experimentalist, he suggests conditions necessary for coming up with situations/experiments that can conflict with Bell's inequality. I'm interested in explanations for the two conditions he line up:
Moreover, in situations usually described by quantum mechanics, it does not often happen that there is a conflict with Bell’s inequalities. More precisely, for situations in which one looks for correlations between two separated subsystems (that may have interacted in the past), we can point out two conditions necessary to have the possibility of a conflict with Bell’s inequalities:
• 1) The two separated subsystems must be in an entangled state, non-factorizable
• 2) For each subsystem, it must be possible to choose the measured quantity among at least two non-commuting observables (such as polarization measurements along directions a and a’, neither parallel nor perpendicular)
I know what a Bell-state is and I know what non-commuting observables are which are usually used in the Bell-type experiments (polarizer angles that are not orthogonal or parallell) but I'm trying to go deeper towards the "root cause" as it were. For example, for 2) above, for a CHSH-type experiment you actually measure 4 correlations between two non-commuting polarizer angles for detector A and two angles for detector B and sum them in the inequality statement - why is the equality satisfied if only one or two of the 4 pairs are summed? And why don't you get any conflict if the angles are not chosen in a specific range? I know the equations and I've seen the plots, but I'm interested in the causes, if it's possible to drill down more..