Abstract
A central goal in the study of human object perception is to determine the code used by the visual system for the representation of globally-integrated shapes. The visual system determines the shape of closed contours and can find an object among a set of distractors very rapidly if it contains a unique cue. We use search speed and asymmetries in performance to determine elements of the basic code for shape in human vision. Observers can often detect the shape of objects rapidly and efficiently when presented in a set of distractors. In some cases efficient detection is not found when the roles of the two shapes are reversed (search asymmetry). Kristjansson and Tse (2001) argued curvature discontinuities (CDs) are critical local cues to shape, supporting rapid visual search with minimal distractor interference when present in the target but absent in the distractors. However, studies using Radial Frequency contours, without CDs, have suggested the polar angle between two points of maximum curvature plays an elementary role. Three visual search experiments will be presented in which performance within-observers is contrasted for patterns differing in curvature, CD, corner numerosity and internal polar angle. The results show that patterns without CDs were readily discriminable, but patterns with the same internal polar angles and extreme differences in curvature and corner numerosity were not, leading to the conclusion that internal polar angle is a primary discriminable shape feature; accounting for both ‘pop-out’ and search asymmetry data when closed-contour targets are employed. The results argue against important roles for corner numerosity and curvature but support the conclusion that polar angles are labelled cues to shape in the human visual system and therefore a critical element in the code for object shape.
Meeting abstract presented at VSS 2015