Abstract
Visual short-term memory allows humans to recall what they have just seen when the sensory input has disappeared from view, and many psychophysical studies have revealed its fragile nature. Here, we show that saccadic eye movements, planned and executed after the disappearance of a visual stimulus array, strongly bias memory performance in favor of the targets of saccades. Observers fixated the center of an array of test locations, highlighted by landmarks evenly arranged on an imaginary circle. After 500 ms of fixation, a memory array—consisting of Gabor patches with varying degrees of tilt—appeared inside the landmarks for 100 ms and then disappeared again. Another 400 ms later, a central movement cue instructed observers to make a saccade to the indicated location. Following saccade execution, and a total of 1200 ms after the disappearance of the memory array, a response cue highlighted a landmark previously occupied by a memory item, prompting participants to report its remembered orientation (clockwise vs. counterclockwise relative to vertical). Despite the fact that saccades went as often to the test location as to any other location in the array, memory performance was markedly better when the saccade had targeted the test location. Modeling memory performance as a function of the magnitude of tilt of the test item revealed that items at non-target locations were forgotten more often (irrespective of the magnitude of tilt), but that remembered items were retained in memory with a similar degree of fidelity (as measured by the just noticeable difference), irrespective of whether the saccade had targeted its location. These results reveal a strong impact of saccadic eye movements on visual short-term memory and highlight the crucial role of action for which parts of a scene we remember and which we forget.
Meeting abstract presented at VSS 2014