Human sensorimotor learning: adaptation, skill, and beyond

Recent studies of upper limb movements have provided insights into the computations, mechanisms, and taxonomy of human sensorimotor learning. Motor tasks differ with respect to how they weight different learning processes. These include adaptation, an internal-model based process that reduces sensory-prediction errors in order to return performance to pre-perturbation levels, use-dependent plasticity, and operant reinforcement. Visuomotor rotation and force-field tasks impose systematic errors and thereby emphasize adaptation. In skill learning tasks, which for the most part do not involve a perturbation, improved performance is manifest as reduced motor variability and probably depends less on adaptation and more on success-based exploration. Explicit awareness and declarative memory contribute, to varying degrees, to motor learning. The modularity of motor learning processes maps, at least to some extent, onto distinct brain structures.

Introduction

Sensorimotor learning refers to improvement, through practice, in the performance of sensory-guided motor behavior. Here we will focus primarily on learning studies of the hand and arm in humans. Based on our own scientific leanings and limited space for this review, we chose to neglect learning with eyes and legs. It is worthwhile to admit to this effector chauvinism as it raises a question that almost never gets explicitly mentioned in the field of motor learning: how to choose which animal, body part, or task to study? Reductionism applies in motor control as much as in as the rest of science; we need reduced systems in order to build up from the simple to the complex. Sherrington's studies of reflexes across single joints in cats and dogs were predicated on just this kind of reasoning [1]. Thus the sheer richness and variety of learned real-world motor behaviors has been reduced to a small repertoire of laboratory-based learning tasks using different body parts. The crucial question is how interchangeable are these tasks with respect to general insights? It is fairly typical to read a paper where the methods section is task-specific but the discussion assumes the results are about motor learning in general. Thus we would argue that there is a tendency in the field to prematurely lump, and that we should be splitting instead. We focus on goal-directed arm movements because, in our view, they represent an intermediate level of behavior that embodies both low-level motor execution and higher-level cognition.

Psychophysical studies, in which learning is recorded through quantitative movement analysis, reveal regularities and performance patterns at the behavioral level, which suggest organizational principles for learning. Computational modeling offers normative principles, such as optimal Bayesian estimation and minimization of costs, to explain and predict behavioral data. Lesions in patients and stimulation techniques, such as transcranial magnetic (TMS) and direct current stimulation (tDCS), can be used to test the causal role of anatomical structures.

This review, necessarily selective, will describe recent noteworthy studies of goal-directed arm movements, and is organized around the principles of modularity and hierarchy. The text is structured on the premise that motor learning (as a blanket term) consists of multiple component processes, each of which has been studied with particular experimental paradigms. We have divided the sections into what we view as roughly separable components of learning. The order of the sections proceeds from adaptation, to skills, and then to the role of explicit cognitive processes.