Abstract
The standard model of color appearance supposes that we have black-white, red-green and yellow-blue opponent mechanisms which provide three dimensions of color appearance at each point in the visual field. Yet we can perceive the colors of both transparent filters and surfaces placed behind. The work of Blaser and colleagues shows that we can segregate and track such layers through attention even with dynamic changes in their spatial frequency, orientation and color features. How does unlayered, incoming chromatic information get divided into such layers and at what stage(s) of processing does this occur? The sum of the parts can exceed the whole. For instance, a red filter in front of a green surface may produce a nominally achromatic stimulus, for some given steady state of neutral adaptation. Yet this null stimulus can be associated with two colors in the layered representation. This suggests that incoming chromatic information is not simply partitioned point-by-point among various layers. Rather, segmentation mechanisms which integrate local color changes, to weigh evidence for the presence of a transparent filter, can create color, presumably through a filling-in process that coordinates among layers. fMRI results of Knoblauch and colleagues show that color scission is associated with activity in both frontal areas and in ventral occipital cortex. Such late processing agrees with the variety of information sources for transparency (e.g., motion, color, stereo). Little is known of how this high-level process may feed back on lower-level processes to, for instance, alter the chromatic sensitivities of lower-level processes, although there is every reason to think that attention plays a major role.