An unobserved quantum entity is said to exist in a "coherent superposition" of all the possible "states" permitted by its "wave function." But as soon as an observer makes a measurement capable of distinguishing between these states the wave function "collapses", and the entity is forced into a single state.
Yet even this deliberately abstract language contains some misleading implications. One is that measurement requires direct physical intervention. Physicists often explain the uncertainty principle in this way:in measuring the position of a quantum entity, one inevitably blocks it off its course, losing information about its direction and about its phase, the relative position of its crests and troughs.
Most experiments do in fact involve intrusive measurements. For example, blocking one path or the other or moving detectors close to the slits obviously disturbs the photons passage in the two-slit experiment as does placing a detector along one route of the delayed-choice experiment. But an experiment done last year by Mandel's team at the University of Rochester shows that a photon can be forced to switch from wavelike to particlelike behaviour by something much more subtle than direct intervention.
The experiment relies on a parametric down-converter an unusual lens that splits a photon of a given energy into two photons whose energy is half as great. Although the device was developed in the 1960s, the Rochester group pioneered its use in tests of quantum mechanics. In the experiment, a laser fires light at a beam splitter. Reflected photons are directed to one down - converter, and transmitted photons go to another down-converter. Each down-converter splits any photon impinging on it into two lower-frequency photons one called the signal and the other called the idler. The two down-converters are arranged so that the two idler beams merge into a single beam. Mirrors steer the overlapping idlers to one detector and the two signal beams to a separate detector.
This design does not permit an observer to tell which way any single photon went after encountering the beam splitter. Each photon therefore goes both right and left at the beam splitter, like a wave, and passes through both down-converters, producing two signal wavelets and two idler wavelets. The signal wavelets generate an interference pattern at their detector. The pattern is revealed by gradually lengthening the distance that signals from one down - converter must go to reach the detector. The rate of detection then rises and falls as the crests and troughs of the interference wavelets shift in relation to each other, go in and out of phase.
Now comes the odd part. The signal photons and the idler photons, once emitted by the down-converters, never again cross paths; they proceed to their respective detectors independently of each other. Nevertheless, simply by blocking the path of one set of idler photons, the researchers destroy the interference pattern of the signal photons. What has changed?
The answer is that the observer's potential knowledge has changed. He can now determine which route the signal photons took to their detector by comparing their arrival times with those of the remaining, unblocked idlers. The original photon can no longer go both ways at the beam splitter, like a wave, but must either bounce off or pass through like a particle.
