Abstract
Changes in the mean or variance of a color ensemble can be readily detected, but how sensitive are observers to the actual number or set of colors defining the ensemble? Discriminating changes in the number becomes difficult when the set exceeds four or more equiluminant hues (Retter et al. VSS 2019). Here we extended these results to examine palettes that also vary in lightness. Stimuli were textures defined by a grid of 10x10 square elements, with the color of each element chosen randomly from the palette. The palettes consisted of colors that evenly sampled angles along different planes in cardinal axis color space (LM, S, and achromatic). For example, in 3- or 4-color palettes the elements differed by 120 or 90 degrees, respectively. In the LM vs S plane elements varied only in hue at constant contrast, while in achromatic planes the sets varied in both lightness and saturation for constant chromatic axis. Four grids were displayed on a monitor and observers judged which one contained an additional color or a rotation with the same number. Accuracy and response times degraded as set size increased beyond three colors, and was similar for the color-luminance and chromatic planes. However, rotations of the elements were more readily discriminable in palettes with lightness variations, an advantage that held over large changes in the maximum luminance contrast, potentially reflecting sensitivity to features such as the lightness levels of the textures. Our results suggest that numerosity judgments within the domain of color are very restricted, and that the different attributes of color (hue, saturation and lightness) exhibit different properties in texture perception, which may be related to the different mechanisms that have been found to mediate the discrimination of pure achromatic or chromatic textures (Chubb et al. 1993; Ortiz et al. 2022).