Abstract
Does the magical number “four” characterize our visual working memory (VWM) capacity for all kinds of objects, or is the capacity of VWM inversely related to the perceptual complexity of those objects? To find out how perceptual complexity affects VWM and whether the observed performance limitation in change detection reflected a limitation in the perceptual or memory process, we measured the slope of visual search RT as a function of display set size to independently quantify the informational load and used a change detection task to measure VWM capacity for six types of stimuli of different complexity: colors, letters, polygons, squiggles, cubes, and faces. We found that the estimated capacity decreased for more complex stimuli, suggesting that perceptual complexity was an important factor in determining VWM capacity. However, the considerable correlation between perceptual complexity and VWM capacity declined significantly if subjects were allowed to view the sample memory display longer. Although Alvarez and Cavanagh (2004) found that informational load was an excellent predictor of change detection performance, with a correlation coefficient of .996 between visual search slope and estimated VWM capacity, our findings show that informational load only accounted for 32% of the variance in VWM capacity with a memory display duration of 3 seconds or longer, indicating that informational load was a better predictor of the estimated VWM capacity at shorter durations. Because informational load should have a constant influence on the hypothetical VWM capacity, the increased correlation suggests that at shorter presentation durations, performance was additionally limited by perception. We conclude that when encoding limitations are minimized, perceptual complexity affects, but does not determine, VWM capacity.
NSF 0345525 and Harvard College Research Program