Abstract
Many models of memory predict that similar stimuli will interfere with each other impairing performance in memory tasks. Yet, contrary to this expectation, some data in change detection tasks has found a benefit when the display colors are similar (all green for example) as compared with when they are dissimilar (each a different color category) (Lin & Luck, 2009). We investigated two possible explanations for this effect. First, it could be that when display colors are all within category, subjects are prompted to encode the colors in more specific detail (e.g. ‘light green’ etc.). As foils are always within category, a more specific encoding in the similar condition would be more likely to result in detection of a change. A second possibility is that when display colors are very similar subjects can use chunking or ensemble encoding strategies, reducing the set size of the display or increasing the information available to support their response. We replicated the original study, and compared performance across different within-category displays that varied in how distinguishable the colors were from each other. Consistent with both hypotheses, the replicated effect was only evident on ‘change’ trials; participants were more likely to respond ‘same’ on a change trial in the heterogenous condition. The benefit of similarity varied, however, as a function of how many distractors were close to or indistinguishable from the test item, and dropped off completely as display colors became highly distinguishable. Performance was highest when the foil was the furthest color from the test color, and when the foil was further from the test color than either of the two distractors. These results suggest that the performance benefit is due to chunking of very close or indistinguishable display colors, and the contrast between the foil and gist information of the display colors.