Estimation of neural noise spectrum in a postural control model

Abstract

A simple linear feedback control model representing a standing human is driven by neural and torque noise sources. A mathematical expression was derived for the neural noise spectrum as a function of spectra computed from two signals: the electromyogram (EMG) and the angle of the subject with respect to the vertical direction. Simulations of a stochastic postural control system were used to generate “EMG” and “angle” signals that were used in the theoretically derived neural noise spectrum. The comparison with the directly estimated neural noise spectrum showed that the mathematical expression yields estimates that have useful information about the spectral shape of the neural noise. In addition, the method also yields useful estimates of the neural noise spectral bandwidth.