Contrast Polarity Affects Performance after Figure-Ground Coding

Abstract

Gilchrist et al (1997 Journal of Experimental Psychology: Human Perception and Performance23 464 – 480) proposed that some aspects of grouping are relatively insensitive to variations in contrast polarity between the elements to be grouped. We assessed the contrast-polarity sensitivity of grouping in a visual search experiment. Display elements were corner-brackets arranged at the vertices of regular polygons (see Donnelly et al, 1991 Journal of Experimental Psychology: Human Perception and Performance17 561 – 570), either aligned with polygon sides (strong-grouping condition), rotated through 20° (weak-grouping condition), or rotated through 180° (open condition). The background was grey; on same-contrast-polarity trials, elements were either all white or all black; on opposite-polarity trials, each element was white and black. The task was to detect a target element rotated 180° with respect to the others. With weak grouping present, opposite contrast polarity slowed reaction times dramatically: they were as slow as those to open displays. A second experiment in which display elements were pacmen showed that the contrast-polarity effect on performance is modulated by figure - ground relations: the dramatic effect of contrast polarity in the weak-grouping condition disappeared, presumably because search focused on the uniform grey illusory surface. These results suggest that grouping operates automatically to produce figure - ground coding of displays, but that contrast polarity differences within a figural surface affect the output of these codes to systems concerned with perceptual discriminations.