Absence of representational momentum for rotating objects

In apparent-motion displays, the remembered final position of a moving object is often displaced in the motion direction. This forward displacement has been called representational momentum, and has led to the hypothesis that subjects automatically predict the object's future path, on the basis of an internalised representation of the object's physical properties. To test this hypothesis, we investigated whether representational momentum occurs not only for translating motion, but also for rotation. Observers saw three sequentially presented images (presentation duration and interstimulus interval of 250 ms each), implying rotation around a vertical axis through the object's centre of gravity. They were followed by a test image, which the subjects had to adjust to match the last (third) position. Stimuli were rendered from 3-D models of either a simple rotating cuboid or a complex rotating face. Responses were measured as displacement from the position of the last presented image. For simple objects, there was no significant effect: subjects reported the final object's position accurately. For complex objects, we found an unexpected backward displacement, incompatible with the representational-momentum hypothesis. We explore alternative explanations that do not rely exclusively on representational momentum, and that account for the dependence of the results on the complexity of the object