Electroretinograms evoked by sinusoidal excitation of human cones.

The amplitude and phase of the fundamental Fourier component of the human electroretinogram (e.r.g.) were recorded with a synchronous detection method under conditions in which each of the three species of cones can be assumed most sensitive in turn. Weber-Fechner behaviour is well established at, or more distal in the retina than, the source of these voltages. Results over the frequency range 7-50 Hz exclude a diffusion model of human flicker perception. The e.r.g. phase vs. frequency plot found with a 'red' test differs from that obtained with a 'green'. The shapes of the e.r.g. field sensitivity action spectra agree with those of the subject's IIj(mu) (j = 3, 4 and 5) mechanisms of Stiles and with in situ measurements of the absorbance spectra of human cone pigments. Threshold phase with each test was independent of background wave-length but, consistent with the results in 3, the phase of the response to the 'red' test (25 Hz) differed significantly from that to the 'green'. If these differences resulted from the absorption of test photons of different colours at different points along the outer segment (independent of cone spectral sensitivity), they would be as clear on dichromats as on trichromats. Results on a protanope are inconsistent with this prediction. We infer that differences in phase are due to the different kinetics of different cone mechanisms and that the e.r.g.s studied here are evoked by exciting only the most sensitive cone mechanism, even though dark-adaptation studies prove that at e.r.g. threshold the test is well above psychophysical threshold for all three cone species. If the inference in 5 is correct, studies of sensitivity across the retina suggest that the spatial distribution of long-, middle- and short-wave-sensitive cones in the human retina differ remarkably.