Learning to stand with delays alters sensorimotor control but does not cause instability when returning to natural balance.

To maintain a bipedal posture, humans must compensate for inherent sensorimotor delays from neural conduction times and electromechanical delays. Ageing and certain neurological disorders increase these delays, so it is crucial that we adapt our control of balance to compensate for the uncertainty associated with acting on sensory information from the past. Although humans can adapt to imposed delays of 400 ms, the mechanisms underlying the adaptation process remain unknown because gross balance instability or errors are absent when returning to balancing without delays. To investigate this, we used a robotic balance simulator to impose delays of 250 ms while participants balanced upright. We characterized and modelled the adjustments in motor commands required to adapt to the addition and removal of delays. Following 20 min of adaptation, participants successfully maintained their balance with the imposed delay. When the delay was abruptly removed, participants remained upright with minimal changes in their whole-body oscillations, but we observed transient (5-20 s) spectral power increases between 1 and 2 Hz in the net ankle torque and lower limb muscle activity. Our computational model revealed that increased sensorimotor gains led to spectral changes in the balance motor commands. Our results indicate that increased sensorimotor gains are necessary to adapt balance control to longer delays and that these gains remained transiently elevated after the removal of the delays without resulting in postural instability. This highlights the remarkable adaptability of human balance control, revealing that the nervous system can flexibly adjust sensorimotor strategies to maintain balance under changing conditions. KEY POINTS: The human nervous system can adapt to sensorimotor delays, allowing us to maintain balance even though there are delays between sensed stimuli and our corrective motor actions. While balancing on a robotic simulator, participants exposed to a 250 ms delay between their self-generated motor commands and resulting whole-body motion exhibited initial difficulty maintaining balance and increased muscle (co)activation but adapted within minutes of exposure. Despite no postural instability following the abrupt removal of the 250 ms delay, participants exhibited transient (5-20 s) increases in leg muscle activation and ankle torque power (1-2 Hz). These changes in the neuromuscular control of balance after delay removal suggest increased sensitivity to sensory feedback, as supported by a computational model representing key physiological features of balance control. By revealing how the brain adapts when facing rapidly changing environments, our results highlight the flexibility of the neural control of balance to ensure robust bipedal stability.