Humans can detect single photons
Quantum light source plus electron-multiplying CCD (EMCCD) made the experiment possible.
While it's been known that rod photoreceptor cells in the human eye can be actuated by single photons, it has never been determined -- until now -- that a single photon entering the eye could be perceived by a human. Researchers from the Research Institute of Molecular Pathology (Vienna, Austria) and Rockefeller University (New York, NY) have determined that humans can detect single photons incident on the cornea with a probability "significantly above chance."1 Previous studies had established that human subjects acclimated to the dark were capable only of reporting flashes of five to seven photons.
Spontaneous parametric downconversion light source
The group built a quantum light source that emits single photons via spontaneous parametric downconversion (SPDC), which produces correlated pairs of photons; the "signal" photon is sent to the eye, while the "idler" photon is sent to an electron-multiplying CCD (EMCCD) to determine when the photon arrives, and also whether or not multiple photons were actually generated (which leads to rejection of that particular data point). In addition, they tried a heavily filtered ordinary light source (a "Poissonian" light source), for which the ratio of multiple- to single-photon incidents are known; this resulted in additional data.
A modified version of what is called the "two-alternative forced-choice (2AFC)" protocol was used to ensure the human subjects had actually detected a photon (or not); in the procedure, the subjects are repeatedly asked to choose between two time intervals, one of which contains a single photon while the other one is a blank. A total of 30,767 trials were carried out and 2420 single-photon events passed the experimental criteria; the resulting averaged probability of correct response was 0.516±0.010 (P=0.0545). With the results from the Poissonian source factored in, the P number became a more significant 0.014.
Based on the experimental optical system and the human eye, the researchers estimated that in about 6% of all post-selected events an actual photon signal was generated by a rod cell -- thus the subjects were detecting actual single-photon eye signals with a higher probability than the numbers would indicate.
In addition to recording the ability of the human eye to register a single photon, the researchers found that the probability of doing so was enhanced when a second photon was flashed a few seconds earlier, as if one photon "primes" the system to register the next.
The work was led by Alipasha Vaziri, associate professor and head of the Laboratory of Neurotechnology and Biophysics at Rockefeller and an adjunct investigator at the Research Institute of Molecular Pathology. "The response that the photon generates survives all the way to the level of our awareness despite the ubiquitous background noise," he says. "Any manmade detector would need to be cooled and isolated from noise to behave the same way."
1. Jonathan N. Tinsley et al., Nature Communications (2016); doi: 10.1038/ncomms12172