Abstract
An object rotating in depth presents a coherent sequence of views but each view is seen only briefly. This scenario raises two questions addressed in the present study: (1) Do observers encode the dynamics of the motion sequence or only the discrete views that comprise the sequence? (2) If motion dynamics are encoded, how do they affect recognition? Observers viewed a movie of an object rotating in depth (study) followed by a static frame showing a single view (test) and judged whether study and test showed the same or different objects. We also manipulated the diagnosticity of the shape information present in the stimulus set and where the test view fell along the rotation trajectory implied by the movie (preceding, early, middle, late, following — all equally spaced). We found that: (1) test frames that followed the trajectory of the motion sequence were better recognized than test frames that preceded the motion sequence; (2) test frames that were actually seen in the sequence were better recognized than following test frames; (3) test frames from the end of the sequence were better recognized than frames from the middle or early part of the sequence. We conclude that observers are biased to encode or remember views that occur near the end of an implied motion trajectory. Moreover, observers use dynamic information to predict unseen views of an object that are consistent with the implied direction of motion. Interestingly, a similar pattern across test conditions was observed for both same and different object trials, suggesting that information encoded about object dynamics is not limited to the particular object being viewed, but instead generalizes to visually-similar objects.