Abstract
PURPOSE: We studied the mental representation of shape using a novel psychophysical task. BACKGROUND: Previous studies (Barenholtz and Feldman, VSS 2001) using a simple probe comparison task (i.e., classifying two spatially separated circular probes as “same” or “different” in size) have shown a systematic effect of contour curvature: responses are slower when the intervening contour between the two probes is more curved compared to when it is straighter. In the current study we sought to use this effect as a tool to study the representation of shape more generally. We speculated that response latency in this task might be affected not only by curvature but by whatever factors, including more global cues, influence the psychological representation of segments of the contour. METHODS: In the experiments, pairs of probes were centered around numerous points (24 per shape) on the contour of a complete shape, including points of particular interest such as high curvature points and potential part boundaries. Shapes studied included the circle (to get a baseline noise level for the task), ellipse, square, rectangle, and both halves of the well-known “divided egg” of Attneave, in which the same contour segment is given different figure-ground assignments. RESULTS: A visually informative way of plotting the reaction times is along the contour itself, in the outward normal direction. The resulting line plot is a “psychologically distorted” version of the shape itself—a window onto the mental representation of shape. The results confirm the influence of curvature on reaction time (corroborating its central importance in shape representation), confirm the psychological reality of the distinction between convex and concave curvature, and show several other intriguing effects of global shape organization.