Abstract
In a series of studies, we tested how many object features and locations could be retained across saccades in 6 human subjects. Visual targets were either circular gabor patches or luminance disks. The Saccade Task consisted of briefly presenting a random number of targets (as many as 15). Each target's spatial position and orientation or luminance was varied randomly. Then, subjects saccaded to a different location and were briefly presented with a probe. The probe's orientation or luminance was systematically varied relative to the pre-saccadic target at the same location (the 80% detection amount determined from preliminary one-target trials). Subjects reported how the probe's visual feature differed from the original target. We compared the performance in this task to a Fixation Task which was identical except subjects maintained eye-fixation throughout the trial. The magnetic search coil technique was used for precise monitoring of eye movements. Results showed that up to 6 targets the subjects' accuracy in the Saccade Task was the same as in the Fixation Task. For trials with more than 6 targets, performance in both tasks declined but the Saccade Task declined at a faster rate than the Fixation Task. This decline was not observed when the test target was attentionally cued - showing that these were not low-level effects. Moreover, subjects' performance was poorer with larger saccade amplitudes than smaller saccades. These findings suggest that limits of transsaccadic memory depend on the number of objects it can retain and the size of the saccade.
This research was supported by the Natural Science and Engineering Research Council (NSERC).