Investigating the computations underlying complex motor skill learning

Abstract

In order for a person to learn a new skill from scratch such as riding a bike or playing the piano, their brain must generate a new motor controller (a policy which maps one’s goal and current state to movements) that can perform this task, a process known as de novo learning. Despite the important role that de novo learning plays in acquiring motor skills, very little is understood about this learning process as the motor learning community has largely focused on investigating how existing skills are recalibrated, a process known as adaptation. In the present project, I designed an experimental paradigm inspired by control theoretic principles which can be used to assess how people learn new skills de novo. I then used this paradigm to investigate how people learn continuous movement skills as well as how motor habits form during learning.