Learned non-rigid object motion is a view-invariant cue to recognizing novel objects

Abstract

Read online

Previous studies have shown that observers encode object motion when they learn novel dynamic objects and subsequently use that characteristic object motion to recognize those objects. These studies have shown that reversing the direction of an object’s characteristic motion, for instance, can impair recognition performance. In the current study, we used direction reversal to investigate whether characteristic object motion is encoded in a view-invariant or view-dependent manner. Observers in our study learned novel objects that had a characteristic motion from the same viewpoint. In Experiment 1, we found a significant contribution of characteristic non-rigid motion across different recognition tasks when there were no changes to the familiar viewpoint; that is, recognition sensitivity was impaired when the direction was reversed. In subsequent experiments, we tested the recognition of non-rigidly deforming (Experiment 2) and rigidly rotating (Experiment 3) objects. Recognition performance was affected by changes to the familiar viewpoint for both experiments. Importantly, we found a significant contribution of characteristic non-rigid motion to performance. The magnitude of this benefit was the same across all viewpoint changes. By comparison, we did not find a significant contribution of characteristic rigid motion to performance. Thus, we argue that characteristic motion constitutes a non-obligatory source of information that is separable from view-based contributions to object recognition.

Keywords

• motion
• spatiotemporal signature
• non-rigid motion
• reversal effect
• view-dependency
• rigid motion