Abstract
Texture segregation and perceptual grouping go hand in hand but they have mostly been studied in separation, each using their own stimuli and tasks, in order to test specific models. For instance, models of texture segregation can be classified as region-based and edge-based, while in perceptual grouping two traditions of models are related to two stimulus types, with elements in random positions (e.g., in snake detection displays) or organized in a grid (e.g., dot lattices). The goal of this study was to provide a more integrated approach. We conducted two psychophysical experiments in which we presented stimuli consisting of oriented Gabors. Perceptual grouping of elements sharing similar orientations may lead to segregation from an ungrouped region with random orientations, or the two regions may be segregated based on local orientation differences at the edge between the regions. We manipulated (1) the degree of orientation similarity in one region by adding orientation jitter to the average orientation (0°–80° in 10° steps), (2) the positions of the Gabors, which were either quasi-random or in a regular grid, and (3) the edge between the two regions, which was either straight or curved. In Experiment 1 a two-interval forced-choice procedure was used with one interval containing only Gabors with random orientations and the other with one region consisting of Gabors in non-random orientations. This experiment required detection of the pattern with a non-random region. Experiment 2 required shape discrimination of the edge (straight or curved) in a yes/no task. We found (1) a gradual decline in performance with increasing orientation jitter in all conditions, (2) lower thresholds for detection of the region than discrimination of the edge, and (3) comparable performance levels for regular and quasi-random Gabor positions. These results are congruent with region-based texture segregation and perceptual grouping.