1. In this study we measured the loop gain of postural reflexes in standing human subjects. Reflex activity is conventionally described in terms of the muscle activation arising from a perturbation, but in this study the ability of the evoked muscle activity to correct the perturbation was also measured, and the behavior of the entire feedback loop is described. 2. A weak continuous random perturbation was applied at waist level to standing subjects. The effects of the perturbation on body sway and soleus electromyogram (EMG) were identified by crosscorrelation. and spectral analysis was used to estimate the open-loop reflex transmission characteristics (i.e., sway to EMG). Under the same conditions, activity in the leg muscles was evoked by galvanic vestibular stimulation with the use of a continuous randomly varying current. The effects on soleus EMG and the subsequent body sway were identified by cross-correlation. This allowed calculation of the open-loop muscle and load behavior (i.e., EMG to sway). From these open-loop reflex and muscle and load transfer functions, the loop gain and phase were calculated. 3. In addition to the gain of the feedback loop, the study describes the transmission characteristics of reflex responses in the leg muscles associated with body sway and the effects of excluding visual and proprioceptive contributions to the response; the transfer function of human soleus with a stimulus that preserves the normal recruitment of motoneurons, including the effects of different load conditions on the muscle; and the transmission characteristics of vestibular pathways that evoke responses in the leg muscles during standing in situations that might modify the reflexes. 4. When standing, the loop gain of reflex feedback is approximately unity and is unchanged by eye closure and stability of support. Reflex transmission introduced a marked phase advance, and this served to offset most of the phase lag introduced by muscle and load. The residual phase lag could explain the frequency of tremor observed during standing (6-8 Hz). 5. The gain of the feedback loop (similar to 1) is higher than suggested by both previous estimates and theoretical considerations, but is still insufficient to explain the stability of normal human standing. This implies that, although sensory information is used to control posture, it does not do so exclusively through a negative feedback control process. The experimental findings are consistent with a reflex response based on a feed-forward process, and this would result in prediction of the response necessary to counteract a postural disturbance.
