Abstract
Purpose:
At the scale of single receptors, luminance and color are indistinguishable in the retinal image, yet the visual system is capable of reconstructing fine spatial detail and rich color experience from the external world. The mechanisms by which percepts of color are derived from individual cones and facilitated by the postreceptoral circuitry remain unclear. Our goal was to investigate the contribution of individual cones to color perception and how it depends on the spectral topography of the cone mosaic.
Methods:
The cone mosaic at 1.5 deg eccentricity was spectrally classified with adaptive optics(AO) imaging and densitometry. Hue percepts were recorded by individually stimulating cones in this region with AO-corrected, cone-sized and retinally-stabilized stimuli at 543nm. The stimuli appeared on a neutral white background. While an entire gamut of hue choices was available, pilot experiments revealed that responses were categorized under only red, green, white and not-seen under these stimulus conditions. Color responses were analyzed for purity, dependence on local cone-type environment and spatial clustering.
Results:
Purity, defined as the number of responses in the most frequently named color category divided by the total number of trials, was 0.73 on average across cones and departed consistently from random behavior. White was reported 50% of the time across L & M cones. In cases when a color other than white was reported, L-cones signaled red while M-cones signaled green in 90% of trials. Color responses depended significantly on the surrounding cone environment. For L-cones, the proportion of ‘red’ decreased and ‘white’ increased with increasing M+S cones in the surround. For M-cones, the proportion of ‘green’ decreased and ‘white’ increased with increasing L+S cones in the surround. Overall, sub-mosaics of same color responses, consisting of more than one cone, emerged.
Conclusions:
Color sensations elicited from individual cones were repeatable and showed a strong dependence on cone-type. White reporting cones were situated in a mixed cone-type neighborhood, while color (red/green) reporting cones were situated amidst their same cone-types. Overall, these findings were inconsistent with models of single cone-center opponent midget ganglion cells at the fovea. It rather suggests spatial pooling via cone coupling, correlations between neighboring ganglion cell outputs or convergence elsewhere in the visual pathway.