A choice with consequences
In the world of quantum mechanics, light acts as both a particle (a photon) and a wave. This is known as wave-particle duality. A classic experiment to show this involves a Mach-Zehnder interferometer. A single photon enters the device and hits a beam splitter. This is like a half-silvered mirror that gives the photon a 50/50 chance of taking one of two paths. If detectors are placed on each path, the photon is always detected on one path or the other—it behaves like a particle that made a choice.
However, if the two paths are brought back together with a second beam splitter before the detectors, something strange happens. The photon interferes with itself, as if it traveled both paths at once like a wave. The fact is that the experimental setup determines the outcome. An arrangement designed to see which path the photon took shows particle behavior. An arrangement designed to see interference shows wave behavior. In 1978, physicist John Archibald Wheeler proposed a thought experiment that took this concept to its logical and baffling limit. Wheeler asked: what if the choice of which experiment to perform is made after the photon has already passed the first beam splitter and committed to its path (or paths)?
Maryland's moment in quantum history
Wheeler's thought experiment became a real experiment here at the University of Maryland. In the early 1980s, a team including Carroll Alley, Oleg Jakubowicz, and William Wickes built a working version. Their setup used a 4-meter-long interferometer and single photons produced by attenuating a 100-picosecond green laser pulse. After the photon entered the device and was in its superposition state (traveling both paths at once), a high-speed Pockels cell—an electro-optic switch, would randomly decide whether to insert the second beam splitter or not. The decision was made in the nanoseconds while the photon was in transit, after it had passed the first fork in the road.
The results were exactly as quantum mechanics predicted. When the second beam splitter was inserted at the last moment, an interference pattern appeared, showing the photon had behaved like a wave and traveled both paths. When it was left out, the detectors fired independently, showing the photon had behaved like a particle and traveled only one path. The measurement choice made in the present appeared to determine the reality of what the photon did in its past. These experiments do not mean we can send information back in time and violate causality. Instead, they challenge our classical intuition about reality, suggesting that a particle's past is not a fixed story until it is measured. As Wheeler himself put it, "It is wrong to attribute a tangibility to the photon in all its travel."