Afferent-like high-pass dynamics helped maintain robust VOR gains and minimize VOR phase lags at high head rotation frequencies.

(A) Example traces of the evoked eye movements during 1, 5, and 10 Hz virtual sinusoidal rotations, normalized to the peak evoked response (i.e., response evoked by the irregular mapping at 20 Hz). Note that the vestibular prosthesis stimulation induced contralaterally versus ipsilaterally directed VOR eye movements in response to increasing versus decreasing stimulation rate relative to the baseline, respectively. Dashed line indicated the simulated head velocity, scaled to facilitate phase response comparison. (B) The normalized VOR gains across natural frequency range (0.2–20 Hz), using the same normalization reference as in (A). The shaded areas around each trace indicate ± SEM. Light green shading indicates the higher frequencies (5–20 Hz) where the differences between each mapping are more pronounced. The superimposed dashed lines are the log-linear fit over the high frequency range. (Top inset) The VOR gain prior to normalization. (Bottom inset) The log-linear slope of the fit. (C) The phase response of the VOR. (Inset) The mean absolute phase at the high frequencies for each mapping. Red, blue, and gray refer to the irregular, regular, and static mappings, respectively. Data underlying this figure can be found at https://doi.org/10.5281/zenodo.6338639.