Multi-rate State-dependent Primitives Underlie The Motor Adaptation and Unlearning to Motion-Depedent Force Perturbations

Abstract

The motor system can compensate for perturbations to the body and within the environ- ment through experience. Motor adaptation studies have suggested that this compensation takes place by developing and updating of an internal model of the body and environment. Previous research has examined the time-scales, learning primitives, and stability of the motor memory following adaptation to forces dependent on motion kinematics. However, computational models that simultaneously capture these aspects of motor adaptation are lacking. In this thesis, we propose a model that encompasses different features of adapta- tion to motion-dependent force-fields. We first trained human subjects in different force-field environments and measured the adaptation and subsequent unlearning. We then formu- lated a motor-adaptation model that takes into account both the motion-dependency and time-scales of motor memory, and investigated its ability to explain several characteristics of experimental finding, including the hysteresis between adaptation and unlearning, and motion-dependent adaptation asymmetries. We finally use the new model to predict the motor adaptation behavior under gradual introduction of the perturbation, as well as savings upon re-exposure to perturbation after a period of inactivity.