According to recent photon entanglement research, the state of entangled photons which is recorded before a later entanglement gate is activated or inactivated is determined by the gate’s state before its state is decided. The researchers generated two independent pairs of entangled photons, and measured the state of one photon from each of the pairs before sending the other photon from each pair down a 108m optical delay line. As the photons traveled down the line, a quantum random number generator decided if an entanglement gate at the end of the delay line would be in an entanglement state or a pass state. If it was set to an entanglement state, the entire 4 photon system would be entangled, influencing the states of the originally measured photon pair, despite the fact the measurement was already recorded and those photons destroyed. If the gate was set to a ‘pass’ state, the photon pairs would not be entangled, and the probability curve of the measured states of the two photon pairs would not be correlated.
The results show that, when the gate state was used to filter the state data from the recordings of the initial photon pairs, statically the states indicate that when the gate was to be in the ‘pass’ state, the photon pairs were not entangled, and when the gate state was to be in the entangle state, the photon pairs were entangled, despite the fact that their states were recorded before the gate state was decided.
Unfortunately, the statistics cannot be used to convey information into the past, so to speak, because the state of the gate is required to know which way the photon pairs should be correlated. Without the gate state information, their states just appear as noise. So causality is preserved despite the apparent time-decision paradox.
http://arxiv.org/ftp/arxiv/papers/1203/1203.4834.pdf