Abstract
The relationships between large color differences are important for understanding the perceptual organization of color. We examined these in a novel phase matching task. Two juxtaposed chromatic gratings defined by different colors (e.g. one red-green and the other orange-cyan), appear more clearly aligned in one phase (e.g. red with orange) than the other (e.g. red with cyan). We used this effect to explore the pairings between different directions in color space. Stimuli were horizontal equiluminant gratings (0.5 c/deg), and subtended 2x8 deg on the mean gray background (20 cd/m2). Contrast varied sinusoidally over +/-80 nominal units along directions at 22.5-deg angles in the cone-opponent plane. A fixed reference axis (e.g. LvsM) was shown in the center of the screen, and was abutted by a test grating (e.g. SvsLM) displayed with opposite phases on the left and right sides. Observers chose which test phase appeared better aligned with the reference grating. The test angle randomly varied across trials with each of the 8 angles shown 10 times to estimate the “orthogonal” color direction at which the two phases were chosen equally. These nulls varied systematically with the reference angle but showed consistent individual differences, and for some observers exhibited plateau-like steps suggesting possible categorical biases. Each observer also selected (from a circle showing 36 color directions) the angles corresponding to their unique and binary hues. From these we estimated the perceived hue of the gratings and their nulls. Orthogonal differences in the perceptual space provided a poorer account of the nulls than orthogonal directions in the LvsM and SvsLM plane, yet for both deviations were evident. These deviations and the individual differences could in part reflect differences in the salience of the two hues within a grating, for example in terms of which hue is perceived as the “foreground” color.