A three-dimensional analysis of vergence movements at various levels of elevation.

Earlier studies have shown that eye positions, recorded in subjects scanning a distant visual scene with the head in a stable position, have only two degrees of freedom (Listing's law). Due to cyclovergence, this law is modified in near-vision. Two previous quantitative studies have documented that the sign of the torsional vergence component depends systematically on elevation: when fixating a nearby target, the eyes show intorsion in up gaze, extorsion in down gaze and no cyclotorsion at some intermediate elevation level (to be denoted as the null elevation). Both studies found a linear cyclovergence/elevation relation, but disagreed on the amount of cyclotorsion. A further uncertainty is how this phenomenon develops dynamically when the binocular fixation point shifts from a far to a near position. Therefore, we have investigated the dynamic coupling between the horizontal and torsional components of vergence in human subjects who were instructed to refixate a light target after it stepped in depth. The target steps were presented at various vertical and horizontal directions relative to the straight-ahead axis of the cyclopean eye. We found that the quantitative relations among horizontal vergence, torsional vergence and elevation were intermediate between those found in the two earlier near-vision studies and that they correspond reasonably to the predictions of a model by Mok and co-workers. The cyclotorsion vergence component had about the same latency and dynamics as the horizontal component. When refixations were studied at different elevations, the torsional vergence component changed from incyclotorsion in up gaze to excyclotorsion in down gaze. In agreement with expectations derived from two quantitative models, the null elevation of cyclovergence was near the binocular primary position. Furthermore, we found no consistent additional dependence on the horizontal direction of the refixation trajectory relative to the midsaggital plane. Other experiments showed that the cyclotorsional changes accompanying convergence were not critically dependent upon the visual conditions. Quantitatively similar similar cyclotorsional components were found even in convergent refixations executed in full darkness towards the location of a remembered (flashed) target. We conclude that visual feedback is unlikely to be very important in controlling cyclovergence in these various conditions.