1. The spatial organization of the vestibuloocular reflex (VOR) was studied in six rhesus monkeys by applying fast, shortlasting, passive head and body tilts immediately after constant-velocity rotation (+/-90 degrees/s) about an earth-vertical axis. Two alternative hypotheses were investigated regarding the reference frame used for coding angular motion. 1) If the vestibular system is organized in head-centered coordinates, postrotatory eye velocity would decay invariably along the direction of applied head angular acceleration. 2) Alternatively, if the vestibular system codes angular motion in inertial, gravity-centered coordinates, postrotatory eye velocity would decay along the direction of gravity. 2. Horizontal VOR was studied with the monkeys upright. Pitch (roll) tilts away from upright elicited a transient vertical (torsional) VOR and shortened the time constant of the horizontal postrotatory slow phase velocity. In addition, an orthogonal torsional (after pitch tilts) or vertical (after roll tilts) response gradually built up. As a result, the eye velocity vector transiently deviated in the roll (pitch) plane and then gradually rotated in the same direction as gravity in the pitch (roll) head plane until the orthogonal component reached a peak value. Subsequently, the residual postrotatory eye velocity decayed along a line parallel to gravity. 3. The time constant of the horizontal postrotatory response was maximal in upright position (21.5 +/- 5.7 s, mean +/- SD) and minimal after tilts to prone (3.8 +/- 0.7 s), supine (4.5 +/- 1.2 s), and ear-down (5.2 +/- 1.6 s) positions. A similar dependence on head orientation relative to gravity characterized the dynamics of the resultant eye velocity vector in the pitch and roll planes. 4. Torsional VOR was studied with the monkeys in supine or prone position. Pitch (yawl tilts from the supine or prone position toward upright (ear-down) position elicited a transient vertical (horizontal) VOR and shortened the time constant of the torsional postrotatory response while a horizontal (vertical) orthogonal component slowly built up. As a result the eye velocity vector gradually rotated in the pitch (yaw) plane until the orthogonal component reached a peak value. Subsequently residual postrotatory eye velocity decayed along a line parallel to gravity. 5. The time constant of the torsional postrotatory response in supine/prone positions was 16.5 +/- 6.8 s. After tilts from supine/ prone positions toward upright position, time constants decreased and were minimal after tilts to upright position (2.7 +/- 0.7 s). After tilts toward upside-down position, time constants increased and were longest at the largest tilt angle tested (+/- 60 degrees head-down from supine/prone positions). After tilts in the yaw plane, time constants decreased as animals were tilted toward ear-down positions where torsional VOR time constants became minimal (5.3 +/- 2.2 s). 6. Vertical VOR was examined in ear-down positions. Roll (yawl tilts from ear-down toward upright (supine/prone) position shortened the time constant of the vertical postrotatory slow phase velocity while an orthogonal horizontal (after roll tilts) or torsional (after yaw tilts) component slowly built up. As a result the eye velocity vector gradually rotated in the roll (yaw) plane away from the pitch head axis until residual postrotatory velocity decayed along a line approximately parallel to gravity. aximal in supine/prone or ear-down positions. For torsional and vertical VOR the period of modulation was half that of the tilt angle after tilts in the yaw plane and took intermediate values after tilts in the pitch and roll planes.
