Disentangling sensory contributions to postural control regulation through sample entropy and neural modeling: A...

Postural control relies on the integration of visual, vestibular, proprioceptive, and auditory inputs to maintain stability. While previous studies have explored the effects of individual sensory modalities, the combined influence of multisensory disruptions on postural predictability remains unclear. We preliminary assessed postural behavior during quiet standing under manipulated sensory conditions in healthy participants. Eight adults stood barefoot on a Wii Balance Board, wearing a mixed reality headset with headphones and receiving vibrotactile stimulation to the Achilles tendons. Experimental conditions combined vibrotactile, static, or moving auditory stimuli across three visual states (Eyes Open, Eyes Closed, Blurred Vision). Center of pressure (CoP) in anterior-posterior (AP) and medio-lateral (ML) planes was collected and analyzed for signal predictability using Sample Entropy. The same experimental procedures were simulated by a biologically inspired neural mass model to investigate multisensory integration in postural control at the neural level. Behaviorally, proprioceptive perturbation via Achilles tendon vibration significantly increased Sample Entropy in both spatial planes, indicating reduced system predictability, while blurred vision decreased only in the AP plane, suggesting less flexible postural dynamics. The neural model replicated key behavioral trends and also revealed distinct central mechanisms underlying sensory interactions: while auditory inputs had minimal behavioral effects, at the simulated neural level, they increased Sample Entropy, especially under dynamic conditions. This revealed subtle modulation of predictability from auditory cues, which was not detectable at the behavioral level. These preliminary results highlight the modality-specific contributions to postural control and demonstrate the utility of Sample Entropy as measure of predictability, with our computational modeling uncovering, for the first time to our knowledge, neural integration processes related to multisensory integration and postural control predictability.