Limb Position Drift and Its Implications for Movement and Position Control

Open Access
Brown, Liana Elizabeth
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
December 03, 2002
Committee Members:
  • Rick Owen Gilmore, Committee Member
  • Cathleen M Moore, Committee Member
  • David A. Rosenbaum, Committee Chair
  • Robert L Sainburg, Committee Chair
  • human performance
  • motor control
  • movement planning
  • drift
Perception of limb position is dependent on visual, proprioceptive, and haptic information. Limb position information is important for movement planning and production, therefore changes in limb position can illuminate how vision and proprioception are used to control movement. Psychophysical research suggests that, without vision, perception of limb position drifts over time. Is this position drift due to a decay of proprioceptive information? If so, this decay should have consequences for movement performance. The experiments reported here address this hypothesis. Limb position cumulative drift was observed in repetitive shape-drawing and reaching tasks. After visual feedback was removed, drift accumulated over time in a systematic pattern. Drift increased quickly early in the series of movements, but then plateaued. The direction of drift accumulation varied widely between participants, but did not vary with the initial position of the limb in the workspace. Instantaneous drift distance, a measure of the movement-to-movement changes in limb position did not change over a series of movements. Instantaneous drift direction, however, was fairly consistent early in the series but then became less consistent, allowing drift to plateau. Both shape form and reaching trajectory were preserved as position drifted. An analysis of movement kinetics at the shoulder and elbow showed that movement distance and direction were preserved because shoulder torque was modified. A forward rigid-body simulation showed that if these torque modifications had not been made, movement direction would not have been preserved. We hypothesized a posture controller that allowed small position errors to accumulate to a fixed threshold. Experimental manipulations of movement speed led to predictable changes in error that affected drift accumulation rates but not drift plateau level, confirming our hypothesis. Together, the results suggested that proprioception remains a reliable source of limb position information, but that it is used differently by separate movement and position controllers. This hypothesis was favored over explanations for drift in terms of a spatial attractor, a growing state estimation error, and differential salience of static and dynamic limb position information.