Retinal motion, stereo, and extra-retinal information from pursuit eye movements have all been implicated in catching balls. Land & MacLeod (Nature Neuroscience, 2000) have also showed that prediction of the bounce point is important for intercepting the ball with the bat in cricket. We show here that prediction is a pervasive aspect of performance in catching balls. We recorded eye, head, and hand movements while subjects caught balls thrown with a bounce. As in cricket, subjects initially fixate the hands of the thrower, then saccade to the anticipated bounce point, and then pursue the ball until it is close to the hands. However, ability to pursue the ball depends on experience with the ball’s dynamic properties. When the ball was unexpectedly replaced with a bouncier ball, subjects were unable to track the ball, and instead made a series of saccades. Within 2 or 3 trials, subjects were once again able to accurately pursue the ball. When subjects threw the ball, they displayed a different pattern of movements, first making a saccade to a point on the floor beyond of the bounce point, and then to the catcher’s hands, with both saccades preceding the ball by several hundred msec. When observing another thrower and catcher, subjects made a saccade somewhere near the anticipated bounce point (less accurately than for catching), then made a saccade to the hands in anticipation of the catch. The observer's head movements from thrower towards the catcher often begin as much as half a sec before the ball leaves the throwers hands. All these observations suggest that observers position their bodies in anticipation of expected events, in order to gather critical information. Thus, vision is proactive (cf Flanagan & Johansson, Nature, 2003). In addition, they suggest that observers maintain an internal model of the dynamic properties of the world, and rapidly update this model when errors occur.

Supported by NIH grants EY05729 and P41 RR09283.