Prediction precedes control in motor learning

Review from Nature Reviews Neuroscience 4, 163 (2003)


Predict and control

Rachel Jones

According to work by Flanagan et al., described in Current Biology, we learn to predict the consequences of our actions before we learn to control them — under certain circumstances, at least. These experimental results were predicted by several recent theoretical models of motor learning, and should help us to understand more about motor learning and motor control.

Motor control can be considered in two parts: control, or the process of generating motor commands to produce a desired outcome; and prediction, which is the internal generation of expected sensory consequences from a set of motor commands. Flanagan and colleagues used a task in which subjects had to manipulate an object along a straight line, while the load on the object was varied during the trial. Over repeated trials, the subjects learned to compensate for the load so that they could produce a straight trajectory.To compare prediction with control, the authors looked at two measures of performance. The hand trajectory was used to measure how quickly subjects learned to control the movement, whereas prediction was measured by looking at changes in grip force. In early trials, grip force was changed reflexively as the hand path (and therefore the load force) was perturbed, but subjects quickly learned to alter their grip force predictively. By contrast, it took many trials for them to learn to control the load.

Recent theoretical models of motor control have included separate components for prediction and control, and some have proposed that the 'predictor' is used to train the 'controller'. The experimental finding that subjects learn to predict the behaviour of a manipulated object before they learn to control it is consistent with this idea.

References and links


Flanagan, J. R. et al. Prediction precedes control in motor learning. Curr. Biol. 13, 146-150 (2003).