Learning from demonstrations: the role of visual search during observational learning from video and point-light...

In this study, we examined the visual search strategies used during observation of video and point-light display models. We also assessed the relative effectiveness of video and point-light models in facilitating the learning of task outcomes and movement patterns. Twenty-one female novice soccer players were divided equally into video, point-light display and no-model (control) groups. Participants chipped a soccer ball onto a target area from which radial and variable error scores were taken. Kinematic data were also recorded using an opto-electrical system. Both a pre- and post-test were performed, interspersed with three periods of acquisition and observation of the model. A retention test was completed 2 days after the post-test. There was a significant main effect for test period for outcome accuracy and variability, but observation of a model did not facilitate outcome-based learning. Participants observing the models acquired a global movement pattern that was closer to that of the model than the controls, although they did not acquire the local relations in the movement pattern, evidenced by joint range of motion and angle-angle plots. There were no significant differences in learning between the point-light display and video groups. The point-light display model group used a more selective visual search pattern than the video model group, while both groups became more selective with successive trials and observation periods. The results are discussed in the context of Newell's hierarchy of coordination and control and Scully and Newell's visual perception perspective.