Effect of accuracy constraint on joint coordination during pointing movements.

Given the number of muscles and joints of the arm, more ways are available to produce an identical hand movement when pointing to a target than are strictly necessary. How the nervous system manages these abundant degrees of freedom was the focus of this study of pointing to targets of low and high indices of difficulty (ID). Two essential features of movement synergies were examined. The first reflects the preferred relations among the outputs of each movement element and was studied through principal component analysis. The second feature of synergy reflects the flexibility of those relationships evidenced by the use of multiple, goal-equivalent solutions to joint coordination. This second feature, which is the main focus of this report, was studied using the uncontrolled manifold approach. Motor abundance was defined operationally as the component of variance of joint combinations that left unchanged the value of important performance variables (goal-equivalent variability, GEV). This variance component was contrasted with the component of variance leading to a change in the value of these variables (non-goal-equivalent variability, NGEV). The difference between GEV and NGEV was evaluated with respect to the performance variables movement extent, movement direction, and path of the arm's center of mass. More than 90% of the variance of joint motions across the pointing trial were accounted for by one principal component, indicating a consistent temporal coupling among most joint motions in a single functional synergy. The flexible nature of this synergy was revealed by the variability analysis. All subjects had significantly higher GEV than NGEV for most of the movement path. Thus, variable patterns of joint coordination did not represent noise but the use of equivalent coordinative solutions related to stabilizing important performance variables. Higher GEV than NGEV was present regardless of the task's ID. One exception was at the time of peak velocity, leading to poorer control of movement extent than movement direction. Increasing the task's ID led to an overall reduction of joint configuraion variance, particularly GEV. These results support earlier work indicating that the use of goal-equivalent solutions to joint coordination is a common feature of the control of this and many other motor tasks. Functionally important performance variables appear to be controlled through flexible but task-specific coordination among the motor elements.