Abstract
Several new experiments demonstrate robust effects of stimulus composition on search time, equating stimulus parts and connectedness. Observers were instructed to search for an oddball target in an array of distractors (identical to each other, and different than a target). Both targets and distractors were composite objects, made of two parts. For example, in one trial, each distractor is a black square sitting directly above the top of a white square, and the target is a white square above a black square. An array of such stimuli can be collectively rotated 90, 180, or 270 degrees, preserving an oddball target. Other example stimuli include variations on one simple shape attached to another. The results are clear: observers are much faster to locate targets when the composites are vertically oriented, and when composites are mirror symmetrical across a vertical axis. Composite stimuli produce a number of interesting effects, including an initial statistical impression of irregularity present or absent, even before an oddball can be localized. The observed effects of part-configuration go beyond symmetry and complexity. These findings are discussed as possible evidence for mid-level configuration detectors.