Learning and Memory Formation of Arm Movements

Abstract

Learning a motor task is characterized by a gradual transition from a high demand on attention to the task becoming automatic and non-attentive. Studies that have recorded limb movements during learning of a motor task have shown that this increase in automaticity of movements is acocmpanied by key kinematic features: (1) stiffness of the limbs decrease (Milner and Cloutier, 1993), as evidenced by a decreased co-activation of the muscles and an increased compliance in response to a perturbation. (2) Movements become smoother (Hreljac, 1993), as evidenced by a reduction in a cost function that scales with the jerkiness of the movement (second derivative of velocity). (3) Motion of the joints becomes decoupled (Vereijken et al. 1992), as evidenced by a reduction in the cross-correlation between patterns of joint rotations. The central hypothesis is that the kinematic features result from the formation of motor memory: the content of motor memory is called an "internal model" of the task. Formation of the internal model allows the nervous system to reduce the dependence of the motor program on the visual and proprioceptive feedback. This leads to a reduction in the attention requirements of the task and increases the reliance of motor output on an internal model that predicts motor patterns that should be produced in order to execute a desired movement. Although this description of an internal model brings to mind learning of complex motor skills, it is equally valid for simple tasks. This can be illustrated by an example: if one is asked to rapidly pick up an empty bottle of milke that has been painted white, the arm exhibits a flailing like motion. This is an indication that in programming the motor output to the muscles, the nervous system predicts and attempts to compensate for the mechanical dynamics of the perceived full bottle. In a control theory framework, the internal model (IM) is an association from a desired trafjectory for the hand to a set of muscle torques (Shadmehr and Mussa-Ivaldi, 1994a). Since in principle this map is unique for the objects and tools which we have learned to interact with, "motor memory" may be thought to contain, at least in part, a colleciton of IMs where visual information serves as an indentifiying cue that allows for binding of an appropriate association, ie., recall. We learn these IMs with experience (Gordon et al., 1992) and they are an integral part of our ability to interact with the objects and systems in our environment. Yet, we know little about the neural substrate of motor memory or the processes that culminate in its formation. The objective of this chapter is to reviewas well as present some new results on psychophysics of learning to make arm movements, and the put these results in perspective of what we know about memory systems of the brain.