Classifying perceptual grouping cues using interindividual differences

Visual perceptual grouping is driven by a diverse set of known cues such as proximity, similarity, and common fate. The set of cues may be composed of different classes, such as 'extrinsic' cues (common region, connectedness) that depend on items' common relationships to third-party elements of a scene (e.g., a box that surrounds two dots groups those dots), and 'intrinsic' cues (similarity, proximity) that depend on features of the grouped items (e.g., two red dots group together if surrounded by green dots). However, these classifications heretofore relied solely on qualitative judgment and reasoning (e.g., Palmer, 1999). In this study, we exploited interindividual differences to probe relationships among grouping cues. In a 'Repetition Discrimination Task' (RDT; Palmer & Beck, 2007), participants (N=127) searched a row of alternating circles and squares for repeated items (two adjacent circles or squares) and identified the shape that was repeated. On each trial the row was grouped into pairs by one of seven cues (color similarity, proximity, common motion, temporal synchrony, common region, element connectedness, and induced perceptual grouping). Repetitions that span groups (between-groups) are found slower (slower reaction time, RT), on average, than those within-groups, an effect that scales with perceived grouping strength. For each participant and each grouping cue, we calculated an index of grouping strength (between-group minus within-group RT), and correlated effects across cues. Consistent with the existence of distinct cue classes, we found that effects of two 'extrinsic' cues (common region, element connectedness) and two 'intrinsic' cues (similarity and proximity) were the most highly correlated with one another. Hierarchical clustering analysis also suggested a broad distinction between extrinsic and intrinsic cues. Our approach suggests that grouping cues are indeed clustered into types that can be revealed through individual differences, which may be useful in guiding the construction of models of perceptual grouping.