Abstract
Visual sensory memory is often characterized as an "icon" - a precise snapshot of the visual scene with a high capacity but short duration. To test the assumption that visual sensory memory is a copy of the visual field, we examined the capacity of sensory memory for visual stimuli of varying complexity. Previous studies have shown that the capacity of sensory memory for relatively simple visual features, like color and orientation, is far greater than visual working memory capacity for those features. If sensory memory is truly a snapshot, akin to viewing the visual scene itself, then visual sensory memory capacity should likewise be larger than working memory capacity regardless of stimulus type. In support of this idea, in Experiment 1 we measured sensory and working memory capacity for visual stimuli that varied in shape and found that, as for simple features, sensory memory capacity for shapes is higher than working memory capacity. However, in Experiment 2 we found that sensory memory capacity for face stimuli was identical to working memory capacity for faces, such that only about one single face could be reported from either iconic or working memory. In Experiment 3 we found that this failure to store more than one face was not simply a limitation in stimulus encoding, as presenting the face stimuli one-at-a-time rather than simultaneously did not increase memory capacity for the faces. Taken together these results indicate that visual sensory memory has a high capacity for simple but not complex visual stimuli, suggesting that this memory system is not simply a snapshot of the visual field that can be accessed by attention. Instead, we propose that the binding of visual features into objects, such as is required for faces, can be accomplished during viewing, but not solely from features in the sensory store.