Abstract
When humans observe the actions of others, they can often accurately anticipate the outcome of those actions on the basis of information available in the actor's movement. This ability was investigated within the context of blocking a penalty kick in soccer, where extreme time constraints force the keeper to anticipate kick direction on the basis of information in the kicker's movements before foot-to-ball contact. In this study, we focused on the contribution of movement information that is local to a particular body segment versus information that is distributed across multiple segments. Using a motion capture system, we recorded the joint locations of experienced soccer players taking penalty kicks. The motion data were used to create animations of a point-light kicker approaching and kicking a ball. The animations ended at foot-to-ball contact. Subjects viewed the animations from the keepers' perspective and made judgments of kick direction (left/right). Previously, Diaz et al. (2010) found that judgments were highly correlated with two sources of local information (hip rotation and location of ball contact), and two sources of distributed information identified using principal components analysis. This experiment provided a more direct test of the contributions of these four sources of information. Subjects were trained for 360 trials using animations with veridical movement information and trial-by-trial feedback. Following training, 180 normal stimuli were intermixed with 72 catch trials in which the usefulness of the information sources identified by Diaz (2010) was systematically manipulated, and feedback was withheld. Subjects performed at chance levels when only hip information was reliable, and were able to judge kick direction when ball-contact information was made unreliable but distributed was left intact. In sum, the results are consistent with the use of distributed information, or some yet to be identified source of local information.