Perceiving Instability: How Expectations Bias Sensorimotor Processing in Balance Control

Maintaining balance requires rapid integration of sensory input with top-down sensorimotor predictions. Predictive coding frameworks posit that mismatches between the two (“prediction errors”) drive perceptual inference. Although these frameworks have helped explain perception in unisensory domains (e.g., vision, touch), it remains unclear whether such mechanisms govern perception in fast-acting sensorimotor systems like human balance. Using electroencephalography (EEG) and discrete postural perturbations, we independently manipulated expectations and sensory input. Participants were primed to expect small or large perturbations, with occasional violations (i.e., delivering smaller or larger-than-expected perturbations). Subjective perception of instability was shaped by expectation alone, regardless of actual sensory input (i.e., perturbation magnitude). Similarly, both pre-perturbation beta-band suppression and post-perturbation gamma increases were sensitive to expected perturbation magnitude. In contrast, both the balance N1 (a well-established stimulus-evoked cortical potential) and the objective postural responses of instability tracked actual perturbation magnitude alone. Crucially, neither perception nor early neural or behavioural markers encoded prediction errors. These findings point to two distinct systems: one shaped by top-down expectations and another driven solely by sensory input, independent of mismatch or prediction error. This challenges the central role of prediction error in fast sensorimotor systems like postural control and may inform conditions involving distorted bodily awareness.