Abstract
Grouping principles such as proximity and similarity allow a complex scene to be parsed into a simpler description. Is grouping local, differentiating only those items to which these principles directly apply, or does it “spread” to other elements in a scene? We found that when a set of undifferentiated items is situated adjacent to items that are paired by Gestalt grouping factors (similarity, proximity, etc.), the undifferentiated items appear to be paired. We term this transitive grouping effect ‘second-order grouping.’ This study investigates how second-order grouping affects visual attention using a method adapted from Beck and Palmer (2002). Subjects searched through a row of mostly alternating squares and circles (target items) for the repeated pair of shapes. An irrelevant set of red and green cross-shaped distracters appeared above or below the targets. The distracters were either ungrouped or they were grouped pair-wise by color. When distracters were ungrouped, detection of the repetition of the target items took equally long whether the distracter items alternated in color or were all in one color. However, when the distracters were grouped, target repetition detection was substantially slower if the repeated targets aligned with two distracters of different colors than if they aligned with two distracters of the same color. Thus, grouping from the adjacent distracter row spreads to the target row. This effect was eliminated by inserting a line between the target and distracter rows, showing that the spread of grouping was not simply attributable to the proximity of the grouped distracters to the target items. We propose that Gestalt grouping factors are extended by transitivity, and such higher-order effects can strongly impact the allocation of attention during visual search.
Supported by NIH 071788 and NSF 0345525