Experimental parameter estimation of a visuo-vestibular interaction model in humans

Visuo-vestibular interactions in monkeys can be accurately modelled using the classical Raphan and Cohen's model. This model is composed of direct vestibular and visual contributions to the vestibulo-ocular reflex (VOR) and of a velocity storage. We applied this model to humans and estimated its parameters in a series of experiments: yaw rotations at moderate (60°/s) and high velocities (240°/s), suppression of the VOR by a head-fixed wide-field visual stimulus, and optokinetic stimulation with measurements of optokinetic nystagmus (OKN) and optokinetic afternystagmus (OKAN). We found the velocity storage time constant to be 13 s, which decreased to 8 s during visual suppression. OKAN initial velocity was 12% of the OKN stimulus velocity. The gain of the direct visual pathway was 0.75 during both visual suppression and OKN; however, the visual input to the velocity storage was higher during visual suppression than during OKN. We could not estimate the time constant of the semicircular canals accurately. Finally, we inferred from high-velocity rotations that the velocity storage saturates around 20-30°/s. Our results indicate that the dynamics of visuo-vestibular interactions in humans is similar as in monkeys. The central integration of visual cues, however, is weaker in humans.