Chromatic temporal integration and retinal eccentricity: psychophysics, neurometric analysis and cortical pooling.

Psychophysical chromatic sensitivity deteriorates in peripheral retina, even after appropriate size scaling of targets. This decrease is more marked for stimuli targeted at the long- (L) to middle-wavelength (M) cone opponent system than for stimuli targeted at short-wavelength (S) pathways. Foveal chromatic mechanisms integrate over several hundred milliseconds for pulse detection. If the time course for integration were shorter in the periphery, this might account for sensitivity loss. Psychophysical chromatic temporal integration (critical duration) for human observers was estimated as a function of eccentricity. Critical duration decreased by a factor of 2 (from approximately 200 to approximately 100 ms) from the fovea to 20 degrees eccentricity. This partly (but not completely) accounts for the decrease in /L-M/ sensitivity in the periphery, but almost completely accounts for the decrease in S-cone sensitivity. Some loss of /L-M/I sensitivity thus has a cortical locus. In a physiological analysis, we consider how the /L-M/ cone parvocellular pathway integrates chromatic signals. Neurometric contrast sensitivities of individual retinal ganglion cells decreased with the square-root of stimulus duration (as expected from Poisson statistics of ganglion cell firing). In contrast, psychophysical data followed an inverse linear relationship (Bloch's law). Models of cortical pooling mechanisms incorporating uncertainty as to stimulus onset and duration can at least partially account for this discrepancy.