Abstract
We examined the role of color in perceptual grouping, and how the salience of patterns defined by color is influenced by adaptation to the distribution of background colors. Backgrounds (7.8 by 5.8 deg) were formed from uniform hexagonal arrays (∼0.3 deg / element), with the color of each element drawn at random from a predefined color-luminance distribution. Targets were composed of six connected elements arranged to form a larger hexagon (∼1 deg). Targets differed in their mean color and/or in the magnitude of the color difference between successive elements. On each trial a target was presented at a random position on the background, with reaction times measured for detecting the location (left or right side). Targets defined by elements with the same color were highly salient and thus rapidly detected. However, reaction times increased as the mean target color approached the mean color of the background or as the color variance within the target increased. Performance was similar for luminance-varying or color-varying patterns once targets and backgrounds were scaled for equal subjective contrast. This suggests that regions of common color (but varying luminance) are no more readily grouped than regions of common luminance (but varying color), even though chromatic borders may be more reliable cues to object boundaries under natural lighting (e.g. because of shading or cast shadows). To examine adaptation effects, response times were compared after 3-min adaptation to a uniform field or to backgrounds with colors resampled from the distribution at ∼1 sec intervals, with 6-sec of readaptation preceding each trail. Prior adaptation to a biased background (e.g. with 4 times more L-M chromatic contrast than S chromatic contrast) reduced search times for targets with variance along the background (L-M) axis. Thus adaptation may facilitate grouping based on novel colors by reducing sensitivity to the camouflaging color variance carried by the background colors.