Abstract
Motion plays an important role in object recognition at both the behavioural and neural levels. For example, studies have shown that observers extrapolate to unfamiliar views of objects rotating in depth when the motion is smooth and predictable. Using a combined psychophysics and fMRI study, we tested whether the smoothness of rotation affected performance and neural responses. Sixteen observers performed a same-different discrimination task in a 3T scanner at the Max Planck Institute. They were presented with a probe-test stimulus sequence, and judged whether both depicted the same or different objects. In blocks of trials, the probe stimulus was either a static image of an object, a smooth animation of a rotating object, or a scrambled animation of an object in which the frames of a smooth animation were randomized. Importantly, both motion blocks presented the same set of views. Within a block, the test stimulus was an image which depicted the object from unfamiliar views that preceded (pre condition) or continued the observed rotation trajectory (post condition). The blocks were optimized to counterbalance for history effects. Observers responded more quickly in the post than pre condition with smooth animations but responded equally fast for these conditions in scrambled and static blocks. Whole-brain group analyses showed that parietal regions were more active in smooth than scrambled blocks, frontal regions were more active for smooth than static blocks, and medial temporal regions were more active in both motion blocks relative to static blocks. These regions are known to process dynamic stimuli. Preliminary analyses of the time courses within these regions show different patterns of activation between pre and post conditions across the different blocks. Overall, the results highlight the importance of smooth motion, and suggest that a rotating object is more than the sum of its views.