Virtual signals of head rotation induce gravity-dependent inferences of linear acceleration

Considerable debate exists regarding whether electrical vestibular stimuli encoded by vestibular afferents induce a net signal of linear acceleration, rotation or a combination of the two. This debate exists because an isolated signal of head rotation encoded by the vestibular afferents can cause perceptions of both linear and angular motion. We recorded participants' perceptions in different orientations relative to gravity and predicted their responses by modelling the effect of electrical vestibular stimuli on vestibular afferents and a current model of central vestibular processing. We show that, even if electrical vestibular stimuli are encoded as a net signal of head rotation, participants perceive both linear acceleration and rotation motions, provided the electrical stimulation-induced rotational vector has a component orthogonal to gravity. The emergence of a perception of linear acceleration from a single rotational input signal clarifies the origins of the neural mechanisms underlying electrical vestibular stimulation.

Electrical vestibular stimulation (EVS) is an increasingly popular biomedical tool for generating sensations of virtual motion in humans, for which the mechanism of action is a topic of considerable debate. Contention surrounds whether the evoked vestibular afferent activity encodes a signal of net rotation and/or linear acceleration. Central processing of vestibular self-motion signals occurs through an internal representation of gravity that can lead to inferred linear accelerations in absence of a true inertial acceleration. Applying this model to virtual signals of rotation evoked by EVS, we predict that EVS will induce behaviours attributed to both angular and linear motion, depending on the head orientation relative to gravity. To demonstrate this, 18 subjects indicated their perceived motion during sinusoidal EVS when in one of four head/body positions orienting the gravitational vector parallel or orthogonal to the EVS rotation vector. During stimulation, participants selected one simulated movement from seven that corresponded best to what they perceived. Participants' responses in each orientation were predicted by a model combining the influence of EVS on vestibular afferents with known mechanisms of vestibular processing. When the EVS rotation vector had a component orthogonal to gravity, human perceptual responses were consistent with a non-zero central estimate of interaural or superior-inferior linear acceleration. The emergence of a perception of linear acceleration from a single rotational input signal clarifies the origins of the neural mechanisms underlying EVS, which has important implications for its use in human biomedical or sensory augmentation applications.