MULTIPLE PAIRED FORWARD AND INVERSE MODELS FOR MOTOR CONTROL

Humans demonstrate a remarkable ability to generate accurate and appro priate motor behavior under many different and often uncertain environ mental conditions. In this paper, we propose a modular approach to suc h motor learning and control. We review the behavioral evidence and be nefits of modularity, and propose a new architecture based on multiple pairs of inverse (controller) and forward (predictor) models. Within each pair, the inverse and forward models are tightly coupled both dur ing their acquisition, through motor learning, and use, during which t he forward models determine the contribution of each inverse model's o utput to the final motor command. This architecture can simultaneously learn the multiple inverse models necessary for control as well as ho w to select the inverse models appropriate for a given environment. Fi nally, we describe specific predictions of the model, which can be tes ted experimentally. (C) 1998 Elsevier Science Ltd. All rights reserved .