The ability of human beings to shift their vergence between multiple disparity stimuli was investigated. The stimulus was a stereogram consisting of a single bar (1 x 0.3 degree) projected in the center of a larger circular pattern (28 degrees dia) of dots (0.3 degree dia). In the initial condition, the subjects perceived a single bar in the center of a cluster of dots, all lying in a single depth plane. The subjects were instructed to fixate either the bar or one of the dots close to the bar. Stepwise changes of target-vergence of either the bar, or of all dots, or of both configurations in opposite directions, were imposed in a random sequence. Under these conditions, ocular vergence was controlled by the target-vergence of the selected target at all times, even if this implied a loss of binocular fusion for the non-selected target. The effect of target selection per se, without concomitant changes in retinal position of the selected target, was studied in two experienced subjects with stabilized image techniques. The subjects viewed a configuration containing short vertical bars placed at either side of a long vertical bar. The short bars on different sides of the long bar had opposite (crossed or uncrossed) disparities. After stabilization of the configuration while the subject fixated the long bar, subjects attempted to fixate the short bars alternatingly. Due to the stabilization the ensuing eye movements did not affect the retinal positions, and thus the disparity, of the bars. Attempted fixation of a parafoveally viewed, short bar induced vergence responses in the appropriate direction.(ABSTRACT TRUNCATED AT 250 WORDS)