An objective measure of the relative strength of perceptual grouping cues using object-based attention

Abstract

The principles of perceptual grouping (i.e., common motion, color similarity, proximity, collinearity, etc.) were described nearly a century ago by the Gestalt psychologists. Despite abundant qualitative and phenomenological evidence, a quantitative measure of the relative strength of these grouping cues has eluded researchers. We report data from a new method that provides an objective, quantitative measure of grouping strength. Stimuli were a modified version of the Egly et al. (JEP: General, 1994) double-rectangle display to which we added orthogonal occluders (cf. Moore et al., Psych. Sci., 1998) creating four discrete surface patches which could be grouped two different ways. Each trial began with a linking display that biased grouping in one direction. For example, during color similarity linking, two patches were green while the remaining two were blue-green. An attention cue then appeared at one patch followed by a target (a convex or concave change in the contour of one of the four patches) at one of three locations: the cued location (70%), the uncued patch in the cued object (15%), or the patch in the uncued object equidistant to the cue (15%). Discrimination response times were faster to uncued targets on the cued object than to those on the uncued object (the same-object advantage) for each grouping principle we tested. Furthermore, by manipulating an intrinsic parameter of each grouping principle (e.g., chromatic difference for color similarity; motion phase for common motion), we were able to parametrically manipulate the strength of grouping. Data obtained with this method constrain predictions about which grouping principles will dominate when two or more cues conflict. This method provides a means for testing a variety of hypotheses concerning perceptual organization.