Vision and single-photon detection in the retina
In a dark room, after half an hour of adaptation, your eye is a photon detector close to the physical limit. A single rod cell responds to a single photon, and you can see a flash containing only a handful. In 2016 an experiment using a true single-photon source suggested people can sometimes detect even one photon, slightly but measurably above chance. The photochemistry is among the fastest reactions in biology; the perception claim is subtle and statistical. This chapter separates the two.
On the plate: how often a flash is seen, as its brightness changes.
Rhodopsin and amplification
A photon absorbed by rhodopsin isomerizes its retinal chromophore from 11-cis to all-trans in about 200 femtoseconds, as Robert Schoenlein, Charles Shank and colleagues measured in 1991, one of the fastest reactions known, with a quantum yield about 0.67. The excited state’s shape is set by quantum chemistry: a trivial effect in Lambert’s sense, but an exquisitely tuned one. The activated rhodopsin then triggers a biochemical cascade: one molecule activates hundreds of transducin proteins, each turning on an enzyme that destroys thousands of cGMP molecules, closing ion channels and producing a current pulse of about a picoampere. Denis Baylor, Trevor Lamb and King-Wai Yau recorded these single-photon responses in toad rods in 1979. Established
Counting photons at the threshold
In 1942 Selig Hecht, Simon Shlaer and Maurice Pirenne flashed dim lights at observers and recorded how often they saw them. Because photons arrive randomly, the number absorbed follows a Poisson distribution; if seeing requires at least absorptions, the frequency-of-seeing curve has a shape fixed by alone. Their curves fit between 5 and 7. Set the threshold 6 photons and compare the curves: the higher , the steeper. The limit is not the rods, which respond to single photons, but downstream processing that must reject false alarms from spontaneous rhodopsin activations, about one per rod every few minutes. Fred Rieke and Baylor reviewed this in 1998. Established
One photon?
Classical light sources cannot deliver exactly one photon. In 2016 Jonathan Tinsley, Alipasha Vaziri and colleagues used a heralded single-photon source, spontaneous parametric down-conversion, in a two-alternative forced-choice design: in each trial a photon went into one of two time windows, and observers guessed which. Across more than 30,000 trials, observers were right about 51.6% of the time on single-photon trials, slightly above the 50% of guessing, with higher accuracy when the previous trial had also succeeded. With a correct rate of 0.516 and 2500 analysed trials, the plate computes how many standard errors above chance the result is. Small effects need many trials, and independent replication. Established for the photochemistry; Debated, leaning supported, for single-photon perception.
What you now have
Rhodopsin’s 200 fs isomerization, the amplification cascade in rods, single-photon rod currents, Poisson frequency-of-seeing analysis, the role of false alarms in setting the perceptual threshold, and the statistics of the 2016 single-photon experiment.
· · ·
Three questions to carry with you.
And one to work out. In a two-choice task at chance, the standard error of the proportion correct is . What is it for 2,500 trials?
Sources: S. Hecht, S. Shlaer and M. H. Pirenne, J. Gen. Physiol. 25, 819 (1942). D. A. Baylor, T. D. Lamb and K.-W. Yau, J. Physiol. 288, 613 (1979). R. W. Schoenlein et al., Science 254, 412 (1991). F. Rieke and D. A. Baylor, Rev. Mod. Phys. 70, 1027 (1998). J. N. Tinsley et al., Nat. Commun. 7, 12172 (2016).