Proximal versus distal control of two-joint planar reaching movements in the presence of neuromuscular noise

Abstract

Determining how the human nervous system contends with neuro-motor noise is vital to understanding how humans achieve accurate goal-directed movements. Experimentally, people learning skilled tasks tend to reduce variability in distal joint movements more than in proximal joint movements. This suggests that they might be imposing greater control over distal joints than proximal joints. However, the reasons for this remain unclear, largely because it is not experimentally possible to directly manipulate either the noise or the control at each joint independently. Therefore, this study used a 2 degree-of-freedom torque driven arm model to determine how different combinations of noise and/or control independently applied at each joint affected the reaching accuracy and the total work required to make the movement. Signal-dependent noise was simultaneously and independently added to the shoulder and elbow torques to induce endpoint errors during planar reaching. Feedback control was then applied, independently and jointly, at each joint to reduce endpoint error due to the added neuromuscular noise. Movement direction and the inertia distribution along the arm were varied to quantify how these biomechanical variations affected the system performance. Endpoint error and total net work were computed as dependent measures. When each joint was independently subjected to noise in the absence of control, endpoint errors were more sensitive to distal (elbow) noise than to proximal (shoulder) noise for nearly all combinations of reaching direction and inertia ratio. The effects of distal noise on endpoint errors were more pronounced when inertia was distributed more toward the forearm. In contrast, the total net work decreased as mass was shifted to the upper arm for reaching movements in all directions. When noise was present at both joints and joint control was implemented, controlling the distal joint alone reduced endpoint errors more than controlling the proximal joint alone for nearly all combinations of reaching direction and inertia ratio. Applying control only at the distal joint was more effective at reducing endpoint errors when more of the mass was more proximally distributed. Likewise, controlling the distal joint alone required less total net work than controlling the proximal joint alone for nearly all combinations of reaching distance and inertia ratio. It is more efficient to reduce endpoint error and energetic cost by selectively applying control to reduce variability in the distal joint than the proximal joint. The reasons for this arise from the biomechanical configuration of the arm itself.

Fig. 1

(a) A 2-dof torque driven arm model used to simulate three different reaching movements: anterior (ANT), left diagonal (LD), and right diagonal (RD). (b) Total ranges of motion for each joint for each reaching movement. (c) Model of the PID controller.

Fig. 2

(a) Nominal torque profiles for the shoulder and elbow joint during anterior eaching without the influence of noise. (b) Shoulder joint signal-dependent noise profile. (c) Elbow joint signal-dependent noise profile.

Fig. 3

The biomechanical properties of the arm were manipulated by varying the inertia distribution along the arm

Fig. 4

Endpoint errors for reaching movements in the (a) anterior direction, (b) left diagonal direction, and (c) right diagonal direction when noise was independently applied at the elbow and shoulder joint. Error bars indicate 6 standard deviations.

Fig. 5

Total net work for reaching movements in the (a) anterior direction, (b) left diagonal, and (c) right diagonal when noise was independently applied at the elbow and shoulder joint. Error bars indicate 6 standard deviations.

Fig. 6

Endpoint errors due to the change in the control scheme for reaching movements in the (a) anterior direction, (b) left diagonal, and (c) right diagonal when noise was present at both joints. Error bars indicate 6 standard deviations.

Fig. 7

Total net work due to the change in the control scheme for reaching movements in the (a) anterior direction, (b) left diagonal, and (c) right diagonal when noise was present at both joints. Error bars indicate 6 standard deviations.