Balancing the upper body is pivotal for upright and efficient gait. While models have identified potentially useful characteristics of biarticular thigh muscles for postural control of the upper body, experimental evidence for their specific role is lacking. Based on theoretical findings, we hypothesised that biarticular muscle activity would increase strongly in response to upper-body perturbations. To test this hypothesis, we used a novel Angular Momentum Perturbator (AMP) that, in contrast to existing methods, perturbs the upper-body posture with only minimal effect on Centre of Mass (CoM) excursions. The impulse-like AMP torques applied to the trunk of subjects resulted in upper-body pitch deflections of up to 17° with only small CoM excursions below 2 cm. Biarticular thigh muscles (biceps femoris long head and rectus femoris) showed the strongest increase in muscular activity (mid- and long-latency reflexes, starting 100 ms after perturbation onset) of all eight measured leg muscles which highlights the importance of biarticular muscles for restoring upper-body balance. These insights could be used for improving technological aids like rehabilitation or assistive devices, and the effectiveness of physical training for fall prevention e.g. for elderly people.

Previous research has identified two major non-stepping strategies used to recover balance following mechanical perturbations: ankle and hip strategy [1, 2]. These strategies are selected depending on eg the perturbation magnitude, prior experience, and configuration of the support surface [2] in order to control the posture (upright trunk and leg orientation) and angular momentum [3, 4]. Following an external mechanical perturbation, both body posture and angular momentum depend, in part, on passive properties of the body, such as the amount and distribution of mass. Simple mechanical models, like the inverted pendulum (IP)[4, 5] or the double IP [6] suggest an approximately linear inverse relationship between the inertia of a perturbed body segment and the resultant acceleration and, presumably, also the segment deflection.