Abstract
Previous work has shown that when observers are looking for a fixed target that is sufficiently different from the distractors, RTs increase logarithmically with set size. These results have been modeled in terms of a parallel evidence accumulation process where items in the display are compared to a target template in mind, and the time it takes to compute that comparison depends on the featural similarity between target and distractors. Here, we examine the role target and distractor uncertainty play in RT in efficient search, as a function of target-distractor similarity. In Experiment 1, we fixed the distractor color across the experiment while varying the target color on each trial. For each participant a random color in CIE space was defined as the distractor color. The target was randomly picked from one of three possible color angular distances away from the distractor (15–45; 45–75; 75–105). The results showed that RTs were within the efficient search range and increased logarithmically as a function of set size and with target-distractor similarity, for all three color separations. This suggests an equivalence between processing displays with a fixed target and with fixed distractors. In Experiment 2, both target and distractor colors were randomly selected on each trial (with the constraint that the current colors were at least 90 degrees different from the preceding trial colors). When target-distractor similarity is low, RTs decreased with set size, whereas they increased logarithmically when target-distractor similarity was relatively high. This is a qualitatively different pattern of results from Experiment 1 and highlights the importance of prior knowledge in efficient search. The results are discussed in the context of current models of visual search and computational models that propose RTs decrease with set size in oddball tasks because the target becomes more conspicuous as the number of distractors increases.