Your eye can notice a photon: 30,767 trials confirm it

It took 30,767 trials to measure whether the human eye can detect a photon. And a photon is the smallest amount of light the universe hands out: below that there is nothing. In July 2016, three volunteers narrowly beat chance and proved that this minimum could reach their consciousness: they chose correctly 51.6% of the time. The team led by Alipasha Vaziri, the physicist heading the laboratory, considered it proven: a single photon can be noticed.

The experiment was split between the laboratories of Rockefeller University, in New York, and the IMP institute in Vienna. Its design demanded an unglamorous kind of patience. Each volunteer spent some forty minutes adapting to total darkness before starting. Then they entered a dark booth and placed their eye before a quantum source built to deliver photons one at a time.

The wait in the dark had its chemical reason. During those forty minutes, the rods of the retina recharge their pigment—rhodopsin—and multiply the eye's sensitivity thousands of times. Their neighbors the cones, which handle color by day, are out of the game from the start: to a cone, a stray photon is nothing. The whole experiment rests on the rods, the night-shift cells.

Each trial contained two intervals marked by sounds. In one of them, the stimulus of a single photon reached the eye. The volunteer had to choose in which one they believed they had perceived light, and then state how confident they were in their answer. They could not say they had seen nothing, nor skip the trial: they had to pick one of the two alternatives. By forcing a choice between the two every time, the format separated a faint perception, whatever it was, from a mere subjective claim.

The decisive word in the setup was the source. Dimming a lamp is not enough: ordinary weak light arrives in bursts that cannot be governed, sometimes two photons in a row, sometimes none, and that proves nothing. The team turned to a trick of quantum optics: a special crystal that, on receiving a photon, splits it into a pair of twins. One of the two flies toward a detector, which announces its arrival; the other shoots off toward the volunteer's pupil. When the detector sings, the laboratory knows that at that instant a photon, and only one, was traveling toward the eye.

The final percentage looks disappointing for a few seconds, and it deserves a slow look. A coin would score, on average, 50% correct. The volunteers managed 51.6%. The difference is barely 1.6 points, too small for a person to notice their own performance during the session. Yet when the test is repeated 30,767 times, that margin leaves a statistical footprint that does not fit a string of answers thrown out at random.

The honesty of the result lies in that tiny margin. Nobody left the booth claiming to have seen clear flashes: the decisions were made blindly, one after another, many with the feeling of guessing. And yet, once tens of thousands of choices had piled up, the correct answers appeared more often than luck could explain. Something coming from that single photon was taking part in the conscious decision.

There was also a second signal. When the volunteers got it right, they tended to report more confidence than when they got it wrong. That link between accuracy and confidence matters, because the percentage could have climbed on a fluctuation without those answering being able to tell one trial from another. That is not what happened: the correct decisions came, on the whole, with a slightly firmer sense of having perceived the flash. That is the signature of a genuine perception, however faint: confidence grew when something had truly come in through the pupil, not when the volunteer simply wanted to be right.

The finding closed a question that had spent decades shrinking. In 1942, the vision physiologist Selig Hecht cornered the human threshold in his Columbia laboratory with ever-fainter flashes, and his team calculated that between five and seven absorbed photons were enough to trigger a visual experience. Those tests had brought perception down to a handful of photons, but they could not isolate with certainty the effect of a single one.

The next piece arrived in 1979, when the physiologist Denis Baylor isolated a single rod and recorded its electrical response to one photon: the cell, on its own, reacted to the minimum amount of light. The rest of the journey still had to be crossed, because a tiny current in one cell did not prove that the human brain ever got the news.

The 2016 study connected those two ends: the source delivered the lone photon and the volunteer answered from consciousness. There was no need to describe a shape, a color or a direction; it was enough to locate the right interval more often than chance can explain.

The study came out in Nature Communications and, as records go, it belongs to a rare family. Records of eyesight usually measure who reads the smallest line on the chart or picks out a lighthouse farther down the coast. This one measured the floor of the building: the point below which physics has nothing left to hand out.

Everyday scale makes clear the size of the wager. An ordinary candle emits on the order of 10^17 photons per second: a one followed by seventeen zeros, a hundred quadrillion photons every second. Similar bursts are streaming right now past the eye reading this screen. The experiment set all that abundance aside, put the candle away, closed the room and asked whether a single one of those specks, separated from the swarm and aimed at the pupil, left a mark on a person.

It did. And with that, the human eye was inscribed at the absolute physical edge of light, competing with the best detectors engineering knows how to build: it can notice a photon, and no device in the world will ever be able to boast of noticing less, because less than a photon does not exist.