Purchase this article with an account.
Andrew Stockman; Cone signals feeding into luminance can exhibit large phase delays and sign reversals: the effect of an inhibitory surround network?. Invest. Ophthalmol. Vis. Sci. 2013;54(15):3710.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The luminance flicker pathway is typically conceived of as a simple additive combination of M- and L-cone signals. As we reported before (ARVO, 2001), this simple picture is complicated by the existence of multiple cone signals with different delays and signs, which can be revealed by most conditions of chromatic adaptation. Here, we extend these measurements and model the results in terms of an inhibitory surround network.
Cone delays were measured by a flicker photometric cancellation technique. Human subjects varied the relative delay between two superimposed flickering lights (and their relative modulation) in order to minimize or null the perception of flicker. One flickering light was either an M-cone or an L-cone isolating stimulus, and the other was a “standard” light that was of the same wavelength as the background. Measurements were made as a function of temporal frequency from 2.5 to about 25 Hz on a series of adapting backgrounds with wavelengths between 410 and 658 nm and at several background intensities.
Under most conditions, sizeable delays or advances by as much as 33 ms have to be introduced between the M- or L-cone flickering light and the standard in order to null the perception of flicker, particularly when M-cone flicker is used. In general, the results are consistent with the L- and M-cone signals being made up of faster components combined with delayed ones of the same or opposite sign. For M-cone signals, the delayed components can be larger than the fast, so that they predominate.
These results challenge the idea that the luminance channel is made up of only additive M- and L-cone inputs. However, these complexities may simply reflect the fact that the inhibitory surround of each cone has both L- and M-cone inputs, and that there are typically more L-cone inputs than M. We model the surround as a network of direct and indirect inhibitory lateral connections across which each step in the network inverts and delays the signal. Thus, the effective sign and overall delay of the slow signal depend upon the relative strengths of the inhibitory feedback at each step and upon the number of steps.
This PDF is available to Subscribers Only