Aspect Experiment

In 1965, John S. Bell proved a celebrated theorem [1] which essentially states that no theory belonging to the class of “objective local theories” (OLT') can reproduce the experimental predictions of quantum mechanics for a situation in which two correlated particles are detected at mutually distant stations (► Bell' Theorem). A few years later Clauser et al. [2] extended the theorem so as to make possible an experiment which would in principle unambiguously discriminate between the predictions of the class of OLT' and those of quantum mechanics, and the first experiment of this type was carried out by Freedman and Clauser [3] in 1972. This experiment, and (with one exception) others performed in the next few years confirmed the predictions of quantum mechanics. However, they did not definitively rule out the class of OLT', because of a number of “loopholes” (► Loopholes in Experiments). Of these various loopholes, probably the most worrying was the “locality loophole”: a crucial ingredient in the definition of an OLT is the postulate that the outcome of a measurement at (e.g.) station 2 cannot depend on the nature of the measurement at the distant station 1 (i.e., on the experimenter' choice of which of two or more mutually incompatible measurements to perform). If the space-time interval between the “event” of the choice of measurement at station 1 and that of the outcome of the measurement at station 2 were spacelike, then violation of the postulate under the conditions of the experiment would imply, at least prima facie, a violation of the principles of special relativity, so that most physicists would have a great deal of confidence in the postulate. Unfortunately, in the experiments mentioned, the choice of which variable to measure was made in setting up the apparatus (polarizers, etc.) in a particular configuration, a process which obviously precedes the actual measurements by a time of the order of hours; since the spatial separation between the stations was only of the order of a few meters, it is clear that the events of choice at 1 and measurement at 2 fail to meet the condition of spacelike separation by many orders of magnitude, and the possibility is left open that information concerning the setting (choice) at station 1 has been transmitted (subluminally) to station 2 and affected the outcome of the measurement there. While such a hypothesis certainly seems bizarre within the framework of currently accepted physics, the question of the viability or not of the class of OLT' is so fundamental an issue that one cannot afford to neglect it completely.

In this situation it becomes highly desirable, as emphasized by Bell in his original paper, to perform an experiment in which the choice of what to measure at station 1 is made “at the last moment”, so that there is no time for information about this choice to be transmitted (subluminally or luminally) to station 2 before the outcome of the measurement there is realized. Of course, whether or not this condition is fulfilled in any given experiment depends crucially on exactly at what stage the “realization” of a specific outcome is taken to occur, and this question immediately gets us into the fundamental problem of measurement in quantum mechanics (► Measurement Theory); however, most discussions of the incompatibility of OLT' and quantum theory in the literature have been content to assume that the realization occurs no later than the first irreversible processes taking place in the macroscopic measuring device.(For example, in a typical photomultiplier it is assumed to take place when the photon hits the cathode and ejects the first electron, since in practice any processes taking place thereafter are irreversible). Although this assumption is certainly questionable, for the sake of definiteness it will be made until further notice.