Abstract
Introduction. Color can be used to detect surface boundaries and to identify objects. Context (background color and contrast patterns) affects both the detectability and appearance of colors. Do the same mechanisms that mediate color detection also mediate color appearance? To investigate this question, we measured effects of various contexts on color detection and appearance. Here we expand on results first reported at VSS 03. Methods. Stimuli were presented on a calibrated RGB monitor with 14-bit intensity resolution for each channel. In both detection and appearance experiments, blurred test spots (1.5 deg in diameter, 3 deg eccentricity) were presented for 200 ms against either uniform or bipartite backgrounds. Each half of the bipartite field was either uniform or contained dynamic spatially pink noise. Background and test spot colors were selected from a set that differed in L+M or S-cone intensity. Observers fixated a central marker. The detection experiment task was a 2-IFC. Multiple interleaved staircases controlled spot intensity. In the appearance experiment, left and right test spots flashed simultaneously. Subjects adjusted spot color on one side to match a fixed spot color on the other. Results. S-cone spot detection was affected by S-cone background changes but not L+M background changes. Similarly, L+M spot detection was affected L+M background changes but not S-cone background changes. Matches of S-cone tests were affected by mean S-cone background but not S-cone noise or L+M background changes. L+M cone tests were affected by mean L+M background but not by L+M noise or S-cone background changes. Effects of mean background color on detection and appearance can be explained by a common mechanism. Background noise, on the other hand, had a large effect on detection but no effect on super-threshold appearance. This result suggests a dissociation between mechanisms that mediate color detectability and those that mediate color appearance.