Abstract
In conjunction search, a target is presented among two kinds of distractors, each sharing one of two target features with the target, usually in different dimensions. In most cases, conjunction search is more difficult than feature search and is associated with response times that increase linearly with set size. Traditionally, the difficulty in conjunction search tasks has been attributed to an inability to bind features from different dimensions into coherent object representations in peripheral vision. Here, we tested an alternative peripheral-vision informed hypothesis that focuses on the role of peripheral pooling regions. We propose that conjunction search becomes inefficient only when distractors of different types fall within the same pooling region. Because source information is partially lost, it is difficult for the visual system to distinguish pooling regions containing only distractors from those containing the target, as in both cases, a pooling region may contain both target features. Experiment 1 used a traditional rectangular search grid and replicated the traditional linear search RT patterns. Experiments 2 and 3 used the same stimuli and manipulated the search displays to minimize the likelihood of different types of distractors falling inside the same pooling regions by presenting stimuli along three concentric ring circles with sufficiently large inter-item separations. In Experiment 2, items were spatially intermixed and in Experiment 3, items were spatially segregated so that only one type of distractor was ever present in each quadrant of the display. In Experiments 2 and 3, search proceeded in parallel, demonstrating that conjunction search difficulty scales with the degree to which multiple search items occupy common pooling regions. These results suggest that the difficulty of conjunction search may not arise from a general inability of the visual system to bind features together but rather may be a consequence of pooling-region mediated processing in peripheral vision.