Dynamic visual–vestibular integration during goal directed human locomotion

Normal visual input plays a very dominant role during locomotion. Functionally, it can assist the central nervous system to overcome a destabilizing effect of abnormal or perturbed vestibular information. However, a recent study has shown a directional effect of transmastoidal galvanic vestibular stimulation (GVS) on gait trajectory when visual information is unreliable. The purpose of this study was to investigate how inputs from the visual and vestibular systems are weighted to optimize locomotor performance under impoverished visual conditions during goal directed locomotion. For unimodal stimulation, the visual input was manipulated using displacing prisms that caused 14° horizontal displacement of perceived target location to the right or left. In addition, GVS (0.8 mAmp) was applied to manipulate vestibular system information during bimodal stimulation conditions. Two bimodal stimulation conditions were defined by the polarity of the galvanic current (anode on congruent and incongruent sides of prismatic deviation). The center of mass (CoM) displacement, head and trunk yaw angles and trunk roll angles were computed to analyze the global output as well as segmental coordination, as the participants walked towards the target. Although the performance was primarily guided by visual information, both congruent and incongruent GVS significantly altered CoM displacement. Similarly, the basic pattern of segmental responses during steering was maintained; however, the magnitude of the responses was altered. Spatio-temporal analysis demonstrated that during bimodal stimulation, the effect of GVS on global output tapered off as the participants approached the target. Results suggest a dynamic visual–vestibular interaction in which the gain of the vestibular input is initially upregulated in the presence of insufficient or impoverished visual information. However, there is a gradual habituation and the visual information, although insufficient, primarily dominates during goal directed locomotion. The experimental trajectories resembled mathematically simulated trajectories with a decaying GVS gain as opposed to a constant gain, further supporting the dynamic nature of sensory integration.