Postural effects on arm movement variability are idiosyncratic and feedback-dependent

Reaching movements are subject to noise arising during the sensing, planning and execution phases of movement production, which contributes to movement variability. When vision of the moving hand is available, reaching variability appears to be strongly influenced by noise occurring during the specification and/or online updating of movement plans in visual coordinates. In contrast, when vision of the hand is unavailable, variability appears more dependent upon hand movement direction, suggesting a greater influence of execution noise. Given that execution noise acts in part at the muscular level, we hypothesized that reaching variability should depend not only on movement direction but initial arm posture as well. Moreover, given that the effects of execution noise are more apparent when movements are performed without vision of the hand, we reasoned that postural effects would be more evident when visual feedback was withheld. To test these hypotheses, subjects planned memory-guided reaching movements to three frontal plane targets, using either an "adducted" or "abducted" initial arm posture. Movements were then executed with and without hand vision. We found that the effects of initial arm posture on movement variability were idiosyncratic in both visual feedback conditions. In addition, without visual feedback, posture-dependent differences in variability varied with movement extent, growing abruptly larger in magnitude during the terminal phases of movement, and were moderately correlated with differences in mean endpoint positions. The results emphasize the role of factors other than noise (i.e. biomechanics and suboptimal sensorimotor integration) in constraining patterns of movement variability in 3D space.