According to quantum mechanics, light can be either a graceful rippling wave or a hail of bulletlike particles, depending on how you look at it. Now, an experiment shows that an observer can make the choice retroactively, after light has entered a measuring apparatus. The result shows that reality is truly in the eye of the beholder.
A single dollop of light, or photon, must be described by a flowing quantum wave that gives the probability of finding it at any particular place and time. At the same time, the photon acts a bit like an indivisible bullet: When observed with a particle detector, it produces a distinct signal, like a pebble pinging off a car door. And things get weirder. The quantum wave can split in two and recombine, like ripples flowing around a stump in a pond, to create striking "interference" effects that determine which way the recombined wave flows. On the other hand, it's simply impossible to split a photon at a fork in the road. If there is no way to eventually put the pieces back together, the photon acts like a particle and goes one way or the other.
Even weirder still, the choice to allow the waves to recombine or not can be made even after the photon passes the fork where it should have split--or not. Famed physicist John Archibald Wheeler realized that nearly 30 years ago and dreamed up an experiment to prove the point. Now Jean-François Roch of the Ecole Normale Supérieure de Cachan in France and colleagues have performed the experiment. The researchers shot photons one by one at a half-silvered mirror, or "beam splitter," to cleave the quantum wave describing each photon. After traveling different distances, the two halves sloshed back together at a second beam splitter 50 meters away, which could recombine them. The experimenters could randomly switch this second beam splitter on and off electronically well after the photon had passed the first one.
If the second splitter was on, interference between the two pieces directed the recombined wave of probability toward one or the other of two detectors, depending on the difference in the path lengths. If the second beam splitter was turned off so the waves couldn't recombine, then the photon took one path or the other with 50-50 probability, and equal numbers of photons reached detectors. The results, reported this week in Science, prove that the photon does not decide whether to behave like a particle or a wave when it hits the first beam splitter, Roch says. Rather, the experimenter decides only later, when he decides whether to put in the second beam splitter. In a sense, at that moment, he chooses his reality.
Others had tried to perform Wheeler's experiment but had lacked the single-photon source and other elements to really do it right, says Arthur Zajonc, an experimenter at Amherst College in Massachusetts. "This is the experiment you wanted to do, but it was too hard," he says. The experiment will likely become a classic cited in textbooks, Zajonc says: "It's going to be seen as a kind of a landmark."