Human Vestibular Signals Generated by Natural Locomotion

摘要

Sensory systems are believed to take advantage of the properties of natural stimuli. Natural images, for example, follow normality and a power-law which are reflected in the dynamics of visual cells. In order to better understand the vestibular system we examined natural human motion. We measured torso and head angular velocities of human subjects who walked, jogged, and climbed a staircase. Angular velocity distributions of the head and torso were fit well by Cauchy distributions, while power spectral densities did not follow a power law. We found that neither a power law nor a two-line-segment fit were sufficient to fit power spectral densities of angular velocity. Increases in power at the gait frequency and its harmonics are not well fit by lines. Differences between torso and head motion show a more evenly distributed reduction of angular velocities, presumably by the neck, in the semicircular canal frame of reference. Coherence between torso and head angular velocity did not show a linear relationship over all frequencies, but did suggest a linear relationship at the fundamental gait frequency and its harmonics. Reduction in angular velocity between the torso and head was then modeled by an adaptive linear filter. Results were mixed and depended on subject, condition, and axis. Qualitatively, predictions of angular velocity were good, capturing both the amplitude and periodicity of the actual head velocity. Finally, initial results were replicated while normalizing gait cycles using linear length normalization. Natural walking and running conditions were compared to treadmill walking and running. Subjects showed significantly different peak velocities during natural and treadmill conditions despite similar movement speeds. Coherence was also different between natural and treadmill conditions. These results provide evidence that natural and treadmill locomotion are treated differently, possibly due to the lack of visual input during treadmill locomotion. Subjects also walked with their heads turned to either the left or right, separating direction of motion and direction of the head. Angular velocity during these conditions show that head direction is not important for stabilizing the head, suggesting that efference copies play a role in head stabilization.