Computational motor adaptation-a kindergarten skill

Abstract

We investigated computational aspects of motor adaptation in 43 children (age range 6-17) who reached while holding the end effector of a lightweight robot. The robot applied an unpredictable, noisy, viscous force field to the hand of each subject. Children adapted to the force field in ostensibly the same manner as adults, reducing their reaching error with practice and forming a model of the approximate mean field gain. However, the children showed more variability in their movement trajectories. A simple ARX learning algorithm proposed previously for adults, adequately captured the children's performance. These results suggest that the computational algorithms for motor adaptation are established early in development, but operate in a context of increased neuromuscular variability.