The utility of different object properties in change detection

Previous research has shown that changes to the configuration of an object's parts are better detected than changes to the shape/arrangement of those parts. This finding suggests that configural, rather than shape, information plays a critical role in object change detection. The current study investigated configural and shape changes in greater detail to determine what aspects of these two types of object properties, if any, were more or less important for change detection. Specifically we investigated configural changes in terms of the orientation of the part change and shape changes in terms of the non-accidental properties of the part change. Using a one-shot change detection task with a single object display, we manipulated: (i) the orientation of a configuration change (0°, 90° or 180°) and (ii) both the number and the type of non-accidental object properties (NAPs) involved in each shape change (3 NAPs were manipulated in total: curvature of axis, curvature of edges, and constancy of size). We found that changes to the curvature of the axis were better detected than changes to either the curvature of edges or to the constancy of size. Detection accuracy was better when there were more NAPs involved in a change. Configural changes involving 180° were more accurately detected than changes involving either 0° or 90°. These results suggest that the axes and basic layout of parts is critical information in change detection. Implications for theories of object recognition are discussed.