Abstract
With each visual fixation, a vast amount of information enters the visual system, only part of which will be stored durably in visual short-term memory (VSTM). We present four experiments testing the hypothesis that saccadic eye movements decide upon the content of VSTM, and determine naturally what we remember and what we forget. We studied whether saccades, planned and executed after the disappearance of a visual stimulus array, influence memory performance even if the saccade is irrelevant to the memory task. We flashed briefly (100 ms) an array of oriented Gabor patches and—after a variable delay—prompted saccades to a random (now empty) location in the array. Subsequently, a response cue instructed observers to report the orientation of one randomly selected stimulus in the array, probing memory either at the same (congruent) or at a different location than the saccade target (incongruent). In all experiments, memory performance was markedly better when the probed location and the saccade target coincided. This effect of saccades on visual memory was time-dependent; it was strongest right after array offset, still present after 800 ms, and no longer significant after 1600 ms (Experiment 1). In contrast, the performance difference between congruent and incongruent trials remained constant across a large range of intervals between saccade cue and the onset of the response cue (Experiment 2). Masking the memory array up to 800 ms after its offset—but simultaneously with the onset of the saccade cue—reduced performance, yet a congruency effect remained (Experiment 3). Finally, we show that this impact of saccades on VSTM is inevitable even when congruent trials are far less likely than incongruent ones (Experiment 4). These results demonstrate that saccades inadvertently determine the content of VSTM, imposing a natural bottleneck on the transition of fragile sensory traces to durable memories.
Meeting abstract presented at VSS 2016