Identifying Learning in a Visuomotor Skill

—We develop a framework for model selection, using structured and unstructured linear time invariant control models, to describe human motor behavior for individual trials of a compensatory tracking experiment. We develop new metrics for quantifying performance and learning across trials using state space model parameters. Changes in metric parameter values obtained from the control model estimates are used to characterize learning. Classification of participants based on error performance provides a means to test how it relates to a set of model parameters that describe both performance and learning. We highlight the many benefits of using unstructured models, including the ability to quantify the evolution of the human control policy across trials.