Abstract
The axis of maximum variability in settings of unique white lies near the blue/yellow diagonal in the MacLeod-Boynton chromaticity diagram (Beer and MacLeod, VSS 2005). This axis coincides with the locus of natural daylight illuminants. A possible reason for this oblique orientation is color constancy. An observer may be willing to accept as white any chromaticity that could plausibly be interpreted as a white reflectance spectrum under a natural illuminant, and natural illuminants vary mainly along this blue/yellow direction. Here we extend this work to points in chromaticity space other than unique white. Observers were asked to memorize the color of a particular stimulus, presented in a dark surround, at the beginning of an experimental session. Observers fixated a single-pixel central fixation dot, and the stimulus was presented parafoveally, rotating through twelve positions around the fixation dot in order to avoid adaptation of one particular retinal location. Observers were then required to set matches to this remembered chromaticity for the rest of the session. The matching stimulus also rotated around a centrally presented fixation dot, and observers adjusted its chromaticity within an isoluminant plane using a tracker ball. Each observer set 35 matches to each remembered color. The distributions of matches show a blue/yellow diagonal elongation. This elongation is not attributable to nonuniformity of the chromaticity diagram, since it is more pronounced in absolute color judgments than in side-by-side-discrimination. A possible explanation for our results is color constancy: observers remember an "object" with a particular reflectance spectrum, and accept as color matches to this object the chromaticities which that reflectance spectrum yields under a range of plausible natural illuminants.
Meeting abstract presented at VSS 2012