Center of mass feedback can reproduce inverse dynamics joint torques during perturbed standing

Electromyographic (EMG) activity from muscles throughout the body during reactive balance have been reproduced using delayed feedback of center of mass (CoM) motion [1]. However, it has not been demonstrated whether such a task-level neural control scheme is biomechanically feasible in multijoint balance control. In this study, we investigated whether delayed CoM feedback could explain the inverse dynamics (ID) joint torques measured during reactive balance responses to forward and backward ramp-and-hold support surface translations in two healthy subjects (S1: male, 34y; S2: male, 24y) using a 3D musculoskeletal model with five degrees of freedom and 43 muscles per leg. We constrained the computed muscle activations to be a linear combination of the measured horizontal CoM position, velocity and acceleration at a fixed delay of 130ms, plus a constant baseline activation. We selected feedback gains, i.e. weights on CoM kinematic signals, and baseline muscle activity values that minimized a cost function consisting of a weighted sum of squared muscle activations and the squared difference between computed muscle torques and ID torques over the time course of the active response, i.e. starting 130ms after the onset of the perturbation. We found that ID torques were well-described based on computed muscle activations generated by delayed feedback of the CoM motion for all 19 trials across both subjects (Fig. 1). The root mean square difference between muscle and ID torques averaged over all trials and degrees of freedom was 1.87  0.70 Nm (mean  SD). Although the computed muscle activations reproduced important features of the EMG, they did not exhibit the initial burst in muscle activity observed experimentally and attributed to delayed CoM acceleration. This may be due to the fact that the torques generated by these initial burst, while important, are relatively small as has been observed in animals with large fiber peripheral neuropathy [2]. Our results suggest that CoM feedback is a plausible and biomechanically viable low-dimensional model for the control of balance.status: publishe