A person modeling a skill is an event. To reveal event properties used to guide the acquisition of ball bouncing skill, we compared changes in the movements of learners who either did or did not view a model in 6 practice sessions. At the end, the learners who viewed the model were more similar to him in respect to posture and hand-ball phasing. We attempted to isolate the motions of the model as information in patch-light displays, but patch placement did not allow apprehension of hand-ball contact. Learners who viewed the patch-light displays attempted to reproduce the model's posture, but not the hand-ball phasing. As reflected in mean number of bounces per trial, participants in the patch-light condition performed less well in the end than did those in either the fully lit or no demonstrationconditions. The results were interpreted to support a dynamical perception/action analysis of the information available in the demonstrations.