Abstract

In many discrimination tasks, better performance is obtained for stimuli that vary only in brightness than for stimuli that vary only in color. Such results are sometimes interpreted as indicating either different processing mechanisms for luminance and chromatic information, or poorer processing of chromatic information by the same mechanisms. However, drawing such conclusions is questionable without an appropriate metric for comparing luminance and chromatic stimuli. One appropriate metric is the performance of a photon-noise-limited ideal observer operating at the level of the receptor photopigments. This ideal observer uses all the information available at the photopigments, and hence allows one to determine the efficiency of the subsequent neural machinery. Following a brief description of the method forcomputing ideal-observer performance, the neural efficiency of processing luminance and chromatic stimuli will be compared for a wide range of discrimination tasks. It will be shown that although efficiencies for luminance and chromatic stimuli are sometimes different, there are a number of tasks in which they are the same. The implications of these results for understanding neural mechanisms and for understanding the perceptual processing of "natural" images will be discussed.

© 1990 Optical Society of America