Abstract
Smooth pursuit eye movements can use cues to an impending collision between a “launcher” and a target to predict a target’s future path of motion (Badler et al., 2010; Wang et al., VSS 2021). The predicted path can be based on knowledge of Newtonian mechanics, but simpler preferences to extrapolate the path of the launcher may also contribute. To disentangle these factors, the relationship between the paths of the launcher and target was varied. A launcher moved from a randomly chosen corner of the display and collided with a central-located stationary target at one of two collision points. The direction of post-collision target motion was either consistent with (Newtonian condition) or different from the Newtonian prediction. In the non-Newtonian conditions, the angle between the launcher and target motions was either smaller (shallow angle condition) or larger (sharp angle condition) than the corresponding angle in the Newtonian path. Different path types (Newtonian; non-Newtonian shallow; non-Newtonian sharp) were tested in separate blocks to allow opportunity for learning the relationship between launcher and target motion. Anticipatory pursuit in four of seven subjects followed the Newtonian prediction in all conditions, showing that prior beliefs about physics received more weight than recent experience. Anticipatory pursuit in the other three subjects was biased toward the Newtonian paths in the sharp angle conditions. In the shallow condition, anticipatory pursuit was either negligible or biased to extrapolate the launcher path. These results provide further support for the role of knowledge of Newtonian physics in anticipatory pursuit, while also revealing sensitivity to other cues that may be weighted differently by different individuals. Other cues include preferences to favor motion extrapolation or memory for recent collisions that do not necessarily follow Newtonian principles.