Integration and accumulation of information across saccadic eye movements

Two experiments investigated the nature of information integration and accumulation across saccadic eye movements. In experiment 1, subjects viewed an array of colored letters while they fixated a central point; this array was erased upon initiation of a saccade to a target. Some time after the saccade, a cue was presented above or below one of the array locations and each subject attempted to report the color and the identity of the letter that had occupied the probed position. Subjects remembered 3-4 color + identity + position units across the saccade; information near the saccade target was remembered better than information appearing in other array locations. Probe delay had little effect on performance. Most errors were mislocations rather than misidentifications. Experiment 2 showed that memory for position and identity information was improved only slightly when subjects made two as opposed to one fixation on the letter array, suggesting that limited information accumulation occurs across multiple eye movements. The results are discussed in terms of a new theory of transsaccadic memory conceived within the theoretical framework for object perception proposed by Treisman (1988).