Perception of occluded motion depends on post-occlusion motion information

Moving objects frequently become occluded from view, yet we often experience a compelling sense of their continued motion and position during occlusion. Despite this subjective impression, recent behavioral and neuroimaging evidence suggests that representations of moving objects during occlusion may be limited. Here, across four experiments, we investigated 1) the accuracy of object localization during dynamic occlusion, 2) the motion information which may support successful position judgments, and 3) the relationship between gaze behavior and localization performance. In Experiment 1a, participants estimated the position of a moving object at unpredictable time points during occlusion. Position estimates were highly inaccurate, with errors increasing as objects travelled further behind the occluder. Increasing object speed exacerbated these errors in Experiment 1b, whereas decreasing speed improved performance. With decreased speed, there was also an increase in pre-occlusion motion information, which may aid object localization during dynamic occlusion. In Experiment 2, we found post-occlusion motion information could support accurate judgments of object reappearance. The necessity of post-occlusion information was further supported in Experiment 3, when removing post-occlusion motion information while testing a predictable intercept location resulted in inaccurate estimations. Across experiments, gaze position predicted behavioral errors, with eye movement strategies varying according to task demands. Together, these findings suggest that the position of moving objects is poorly represented during dynamic occlusion, despite our compelling subjective experience of their occluded location. Instead, accurate judgments depend critically on information acquired after the object re-emerges, consistent with a postdictive reconstruction of occluded object position.