Investigating the Contribution of Instance-Reliant Learning in Visuomotor Adaptation and Its Generalization

Abstract

Motor adaptation has been of great interest in the past two decades as it reflects how movement skills are acquired and consolidated by the nervous system. In our recent studies, instance-reliant learning is considered as an essential component of visuomotor adaptation, since it plays a unique role in fast and automatized control of movement output. The goal of this dissertation is to investigate the nature of instance-reliant learning on two aspects: to determine the differential contributions of algorithmic learning and instance-reliant learning to visuomotor adaptation; and to determine the nature of movement instance involved in visuomotor adaptation and its generalization across different situations that involve magnitude, workspace, and limb configuration. Experimental results show that both algorithmic and instance-reliant learnings are positively associated with the improvements in the subsequent performance, which is compatible with our expectation. However, compared to algorithmic learning, which has been intensively studied before, instance-reliant learning exhibits different characteristics in terms of both visuomotor adaptation and its generalization. In Experiment 1 and 2, we found that algorithmic and instance-reliant learning led to substantial improvements in movement errors; but the learning rate in the subsequent test was only sensitive to algorithmic learning. In Experiment 3, 4, and 5, the movement instances associated with the reaching performance were magnitude, workspace, and limb configuration specific, although it could still generalize to a certain degree. Thus, the distinct contributions of instance-reliant learning to motor adaptation are elucidated in this dissertation. We expect that findings from this dissertation would prove valuable for developing rehabilitation strategies for patients who suffer from neuromotor impairments.