Abstract
The programming of saccadic eye movements to complex visual stimuli is influenced not only by stimulus properties, but also by cognitive processes. It has been shown that saccade responses to two-target stimuli are influenced by previous target displacements. In light of recent evidence of object-centered coding of visual stimuli, we examined whether recent position of targets in object-centered coordinates could also influence saccade programming. We measured performance of saccades to an “object” consisting of two stimuli with a small, fixed horizontal separation. Eye movements of 3 subjects were measured using the scleral search coil system. Trials began with the appearance of a central fixation point. 500–800 ms later two annuli (width 1.5() appeared, side by side with a separation of 5(. The position of the pair was centered at −10,−5,0,5,or 10( horiz. and +/−6( vert. On each trial, one of the annuli was notched. Subjects were instructed to make a saccade to the notched annulus as quickly and accurately as possible. The probability of the change of the object-centered location of the notched annulus (left or right) from trial to trial was manipulated for separate blocks of trials, while the location of the pair on the display varied randomly. As a control, one-target trials were also tested. For all 3 subjects, saccade endpoint depended significantly on the object-centered location of the notched annulus when held constant across a block of trials. Saccade latencies were virtually identical to those of one-target saccades. This object-centered dependence was much less strong when the object-centered location of the notched annulus alternated from one side to the other from trial to trial or was completely random. These results suggest that saccade programming is facilitated by priming operating in object-centered coordinates.