The configural advantage in object change detection persists across depth rotation

Abstract

Although traditionally there has been a debate over whether object recognition involves 3-D structural descriptions or 2-D views, most current approaches to object recognition include the representation of object structure in some form. An advantage for the processing of structural or configural information in objects has been recently demonstrated using a change detection task (Keane, Hayward, and Burke, 2003). We report two experiments that extend this finding and show that configural information dominates change detection performance regardless of an object's orientation. Experiment 1 demonstrated the advantage that configural information has over shape and part arrangement information in change detection across four different object rotations in depth. Experiment 2 showed that this advantage occurs for both categorical and coordinate configural changes. These results are consistent with the hypothesis that configural information is a critical feature of object representations and that this information is utilized effectively in object recognition across changes in viewpoint.