Investigations of binocular visual direction have concentrated mainly on stationary objects, Eye positions were generally not measured and binocular fixation was assumed to be perfect, During the viewing of stationary objects, vergence errors are not negligible but small, During the viewing of moving objects, however, errors in binocular fixation are much larger, Existing rules for binocular visual direction were examined under the latter, more demanding viewing conditions, Eye movements were measured objectively by the scleral coil technique, Subjects viewed a large stereogram in which the half-images oscillated in counterphase. The stereogram contained two square random-dot patterns placed side by side with a gap in between, A vertical line, visible only to one eye, oscillated in the gap, Subjects were asked to adjust the amplitude of line motion until the line was perceived to be stationary, In so doing, they set amplitudes equal to the amplitudes of half-image motion if the gap between the patterns was narrow, They set amplitudes significantly smaller in wider gaps, Subjects made considerable fixational errors in following the oscillations of the line and the random-dot patterns, The results of the settings and of the retinal errors together refute existing rules for binocular visual direction of monocular objects, Perceived directions of monocular objects cannot be specified by geometrical rules that include only the positions of the objects and of the two eyes, The results suggest that perceived directions of monocular objects are captured by the binocular visual directions of adjacent binocular objects, Capture of binocular visual direction was found to be effective for gaps as wide as 8 deg between the binocular objects, The phenomenon of binocular capture has negative consequences for the general use of nonius lines as indicators of eye position. (C) 1997 Elsevier Science Ltd.
