Embeddedness of Local Gravity in Perception & Action

Abstract

Understanding and making predictions about the physical forces determining object motion could greatly facilitate the perception of dynamic events. To be optimal, trajectory judgments of a falling object should be consistent with the physical laws that govern object motion, however previous studies show that beliefs often violate Newtonian laws. The aim of this study was to determine whether human observers integrate Newtonian predictions of object motion with noisy sensory information to reduce ambiguity in the perception of dynamic events. When a ball rolls off a surface, the trajectory of its descent is determined by a relatively constant horizontal velocity and a downward acceleration caused by gravity. For this experiment, we simulated this dynamic event in virtual reality: participants viewed a marble-sized ball fall from a tabletop under two different horizontal velocities and seven different downward accelerations (one matching Earth’s gravity, three lower, and three higher, spanning the range from Mercury’s gravity to Jupiter’s gravity). On each trial, participants indicated the perceived trajectory of the falling ball by adjusting the curvature of a parabolic arc. Each participant’s internal prediction about local gravity in this virtual environment was assessed using trials where the ball rolled up to the edge of the tabletop, then disappeared, and the participant indicated the expected continuation of the trajectory (as opposed to the perceived trajectory indicated in other trials). Although participants’ internal predictions were biased to underestimate the true gravity of Earth (which may serve a practical role for guiding interceptive actions), their perceptual judgments reflected a combination of this prediction with the available visual information. Furthermore, the combination process appears to be consistent with maximum likelihood estimation, as the influence of the internal prediction increased proportionally with the amount of sensory noise, which we varied by manipulating the amount of time the ball’s descent was visible.