Visual motion does not bias gravity-referenced vestibular coding in the primate cerebellar nodulus and uvula.

Visual motion is known to influence perceptions of tilt, verticality, and translation, suggesting that optic flow is combined with vestibular cues to estimate orientation relative to gravity. The cerebellar nodulus and ventral uvula (NU) are a prime candidate to perform this computation because this region uniquely receives convergent semicircular canal, otolith, and proprioceptive inputs, and in non-primate species full-field visual motion robustly modulates NU activity. Here, we tested whether visual roll motion, known to bias perceived orientation relative to gravity, alters the internal gravity-referenced transformation used by NU neurons to encode vestibular self-motion. To test this, we recorded single-unit activity from NU Purkinje cells in rhesus macaques during whole-body translations in darkness, either without visual stimulation or after prolonged full-field optokinetic roll motion. We hypothesized that visual motion simulating head tilt would bias the NU's internal gravity estimate, leading to altered translation-evoked responses. Contrary to this prediction, visual motion had no effect on either baseline firing rates or vestibular responses. Moreover, a computational model predicting visually induced shifts in neural tuning was not supported by the data. These results show that visual roll motion, although known to influence perceived orientation, does not bias gravity-referenced vestibular coding in the primate NU. This specialization may preserve a fast, body-anchored gravity estimate for postural and reflexive motor control, delegating visual-vestibular integration for perception to downstream circuits.