Persistent decrease in neuromotor noise by manipulating error tolerance

Abstract

Prior work of Sternad and colleagues proposed that variability in a skilled task that affords a solution manifold can be decomposed into three components: Tolerance, Noise and Covariation. With extensive practice, all three components decrease but noise remains the highest. The present study examined how signaling error tolerance can further decrease this noise component. Using a virtual throwing task, subjects performed a single-joint rotation about the elbow and released a virtual ball that traversed a concentric force field to hit a target. A successful throw resulted when the ball passed within a threshold around the target. A large threshold signaled large tolerance, which allowed for high noise. Conversely, a small threshold signaled small tolerance, which required low noise for task achievement. We hypothesized that (H1) signaling tolerance leads to lower noise level than under normal practice conditions, (H2) when signaled tolerance is decreased, subjects decrease their noise, and (H3) when signaled tolerance is increased, subjects increase their noise. æResults showed that subjects reduced noise to a lower noise level when tolerance was signaled. When the signaled tolerance was decreased during practice, subjects decreased their noise. When the tolerance was subsequently increased, however, subjects did not increase their noise as expected. Importantly, this decrease in noise persisted even after five days of practice with the increased tolerance. These results have implications for rehabilitation as signal detection and control processes in the sensorimotor system may be enhanced by manipulating visual feedback based on error tolerance.