Joint visual-vestibular computation of head direction and reflexive eye movement

Several cognitive maps have been identified, but what sensory signals drive them and how these are combined are not well understood. One such map, the head-direction representation, is believed to be primarily driven by vestibular motion signals in mammals. Here, we combine in vivo imaging of neuronal activity, genetic perturbation of neuronal circuits, behavioral testing, and theoretical modeling to show that the representation of head direction in mouse is driven by not only vestibular but also visual motion signals: both are essential, and the latter, originating in direction-selective retinal ganglion cell activity, dominates at low speeds. We show that, correspondingly, visual perturbations alter navigational behavior that relies on head-direction computation. Finally, we find that head-direction representation and the slow phase of reflexive eye movement are tightly correlated, and we propose a theoretical model that elucidates their emergence from coupled visual and vestibular processes. Our results suggest that the brains estimate of head direction is built on an oculomotor reflex pathway driven by both visual and vestibular signals.