Before, during and after you disappear: aspects of timing and dynamic updating of the real-time action simulation of human motions.

The detailed dynamics of action simulation was investigated using the occluder paradigm: a point light actor (PLA) was shown, then briefly occluded from view, during which period action simulation was generated. Following occlusion, the PLA reappeared, either a progression of the motion as it should be post-occlusion or temporally shifted earlier/later. Participants made judgements on whether the reappearing PLA was too early or too late to be a correct continuation (Experiments 1 and 3) or whether it was a veridical continuation or not (Experiment 2). Over three experiments we asked how action simulation is affected by motion information before, during and after occlusion. Reducing motion presented before occlusion retained the accuracy of action simulation judgements. Presenting 4 frames (67 ms) of PLA motion during the occluder duration dynamically updates or altogether regenerates the action simulation. Reducing the duration of the test motion after the occluder decreases judgement precision, which we interpret as a limitation in the process of postdictive motion judgments. Overall, this is further evidence that the action simulation process is remarkably adapted to making human motion predictions.