Modeling human learning on visualmotor tasks