Abstract
When we see other people moving, we often see them briefly disappearing from view behind objects, or otherwise obscured from sight. Previous research shows that individuals generate real-time action simulations that aid the prediction of the future course of actions (Graf et al., 2007). Last year (Parkinson et al., VSS 2010) we showed that action simulations directly aid the perception of visually degraded human motions. Here we further investigated detailed aspects of timing in action simulation using newly developed versions of our occluder paradigm: A point light actor (PLA) representation of human motion performing a simple action, such as a basketball shot, was presented then briefly (500 ms) occluded from view, during which the participant's visuo-motor system automatically generated a real-time action simulation. Following the occlusion period, the PLA reappeared in motion as either a natural progression of the motion as it would have continued during occlusion (thus temporally congruent with the action simulation), or shifted earlier or later in time. Participants judged whether the reappearing test motion was a correct or incorrect motion continuation, testing the accuracy of their action simulations.
We show that inserting just four frames (67 ms) of PLA motion within the occluder dynamically updates the action simulation, accordingly affecting subsequent motion judgements. Further, we show that the duration of motion presented before the occluder makes little difference to judgement accuracy, with accurate motion predictions made from remarkably little perceived motion (∼50 ms). However, judgement accuracy is detrimentally affected by reducing the duration of the test motion (from 500 ms to 50 ms), suggesting that a certain minimum of visual motion information is required to compare to the internal action simulation. Thus we provide further evidence that action simulation is a dynamic, real-time process, remarkably adapted to making predictions about human biological motion from relatively impoverished information.