Models for Automated Motor Skills Training

Abstract

Motor control tasks are characterized by large inter-subject variability which for efficiency necessitates an individualized approach to training. Because of the storage and memory capacity of the computer, research directed toward automated optimizing of individualized motor skills training is possible. Two studies that examined various training models in motor learning are presented. These models include the linear adaptive model, a learner-centered model, a shifting-difficulty model, and a fixed-difficulty model. Each of these training models was evaluated using a two-dimensional pursuit tracking task with an isometric controller. Feedback in terms of level of task difficulty and level of tracking performance was provided during training. The transfer task involved one seven-minute tracking session in which task difficulty shifted to one of three levels after each minute of tracking. Results for each training model are discussed in terms of training time required to reach the exit criterion and in terms of root-mean-square tracking error exhibited during the transfer task.