Characterizing and Modeling Throwing Strategies in Trained and Untrained Humans

Abstract

How can humans control the performance of a goal-oriented task in order to achieve the best possible outcome? Although research in the domains of biomechanics, neuroscience, and physics has uncovered many mechanisms involved in human movement, the manner in which humans optimize physical performance is not well understood. In this study, the example of accurate throwing was addressed to investigate whether humans select throwing strategies in accordance with how well these strategies minimize the amplification of error from release point to landing position. Furthermore, trained individuals were compared with untrained participants to determine whether throwing-specific practice improves the ability to choose throwing strategies associated with lower error propagation. A theoretical model predicting strategy choice was adapted from a previous study (44), and experiments were performed to measure the velocity and angle combinations selected by the subjects, as well as broader underarm and overarm strategies. Results show that overall, the distributions of subject velocity-angle combinations do not follow the optimal strategy predicted by the model. Additionally, the experiment does not reveal clear trends predicting how humans select underarm and overarm strategies. The findings suggest that for trained and untrained individuals alike, minimization of error amplification is not a significant factor in the accurate performance of throws. Thus, error propagation does not appear to constitute a major underlying principle in human motor planning of accurate arm movements such as those involved in throwing.