Abstract
The subjective visual vertical (SVV) indicates perceived direction of gravity. Even in healthy human subjects, roll-angle dependent misestimations - roll over-compensation (A-effect, head-roll>60°&<135°) and under-compensation (E-effect, head-roll<60°) - occur. Previously, we demonstrated that, after prolonged roll-tilt, SVV estimates when upright are biased towards the preceding roll position, which indicates that perceived vertical (PV) is shifted by the prior tilt. Hypothetically, PV in any roll position could be biased towards the previous roll position. We asked whether such a "global" bias occurs or whether the bias is "local". The SVV of healthy human subjects (N=9) was measured in nine roll positions (-120° to +120°, steps=30°) after 5min of roll-tilt in one of two adaptation positions (±90°) and compared to control trials without adaptation. After adapting, adjustments were shifted significantly (p<0.05) towards the previous adaptation position for nearby roll-tilted positions (±30, ±60°) and upright only. We computationally simulated errors based on the sum of a monotonically increasing function (producing roll under-compensation) and a mixture of Gaussian functions (representing roll over-compensation centered around perceived vertical). In combination, the pattern of A- and E-effects could be generated. By shifting the function representing local over-compensation towards the adaptation position, the experimental post-adaptation data could be fitted successfully. We conclude that prolonged roll-tilt locally distorts PV rather than globally shifting it. Short-term adaptation of roll over-compensation may explain these shifts and could reflect the brain's strategy to optimize SVV-estimates around recent roll-positions. Thus, postural stability can be improved by visually-mediated compensatory responses at any sustained body-roll orientation.