Figure 1.
Schematic diagram showing a subject sitting on the rotating chair for simultaneous psychophysical and eye movement (ENG) assessment in the dark.
A: Threshold vestibular task. The subject carries a hand-held device with two buttons (left and right) whilst exposed to step acceleration rotations with an initial acceleration of 0.5°/s2, increasing by 0.5°/s2 every 3 s. The subject presses the appropriate button to indicate perceived direction (leftward vs. rightward) as soon as they were sure they were moving in a particular direction. Vestibulo-perceptual (VP) thresholds were measured by the time taken from chair acceleration onset to button press (button press) and converted to °/s when appropriate. The vestibulo-ocular (VO) threshold was measured as the point at which the slow-phase eye velocity curve left the baseline and did not return (nystagmus onset). B: Supra-threshold vestibular task. Subjects in the motorised rotating chair were exposed to a velocity step of 90°/s for 60 s, either leftwards or rightwards. They were instructed to turn the wheel at maximal speed on starting/stopping rotation (the point of maximal subjective and ocular angular velocity) and to slow the tachometer speed in proportion to their own perceived slowing of rotational velocity [10]. A representative raw trace and fitted exponential curve from the tachometer wheel in a normal subject is shown on the right. This allows for accurate measurement of the time constant (TC) of decay of the vestibular perceptual response. Vestibulo-ocular responses (not shown) were obtained using electronystagmography (ENG = EOG), and follow a similar exponential decay to the perceptual responses.
Figure 2.
Schematic illustration of the sensor fusion and signal detection model.
A: Sensor fusion as maximum-likelihood estimation: the two likelihood functions (red and blue distributions) for two sensors with slightly different sensor variability and gain are shown. The blue distribution is correctly centred at the signal, while the red, more variable distribution is centred closer to zero. The pink curve shows the combined distribution, which has a lower variability. B: Signal detection: the combined signal distribution (pink) and the noise distribution (centred at zero) are assumed to have equal variance. The threshold signal is reached, if the overlap between both distributions becomes small enough. The respective decision criterion is marked by the dashed line.
Figure 3.
VO and VP thresholds for patients and simulated model data.
Mean perceptual (top) and vestibulo-ocular (bottom) thresholds for patient (±SE) and simulated model data, and normal subjects. Thresholds are symmetrical in healthy subjects and lower for VOR than for perception. After acute unilateral vestibular neuritis, thresholds become asymmetric and bilaterally increased with higher thresholds on the affected (ipsilesional) side. After recovery, thresholds remain elevated, but become symmetric again. Simulated thresholds derived from the model show good agreement with patient data; both VO and VP thresholds are elevated and asymmetrical acutely, with a reduction in asymmetry at recovery.
Figure 4.
Grand average acute VO and VP responses during supra-threshold task.
Grand averages of slow phase eye velocity (vestibulo-ocular) and perceived angular velocity (perception, normalised) in response to 90°/s velocity steps, for normal controls (dotted line) and acute VN patients when accelerating towards the side of the lesion (ipsilesional, dashed line) and towards the healthy side (contralesional, solid line). Note symmetrical and shorter time constants for perceptual data despite grossly asymmetrical ocular responses in acute VN patients.
Figure 5.
Time constant and duration VO and VP supra-threshold responses at acute and recovery stages.
Mean (±SE) supra-threshold duration and time constants for perception (right panel) and vestibulo-ocular (left panel) responses in VN patients, acutely and at recovery. Grey horizontal bars show normative data (95% confidence interval for mean).
Figure 6.
Scatter plots showing correlation between VO/VP threshold and supra-threshold responses.
Correlation plots between duration of the response to the supra-threshold stimulus (90°/s velocity step, x axis) and vestibular thresholds (y axis) for the vestibulo-ocular (A) and vestibulo-perceptual systems (B). The plots show good correlation between the two vestibulo-ocular results but absence of correlation between the perceptual results.