Abstract
In a fixed-target efficient search task, reaction times (RTs) have been shown to increase logarithmically with set size. The magnitude of the logarithmic slope is determined by the similarity between the target and distractors. These findings are attributed to a parallel, unlimited capacity, resolution limited architecture and have been observed in discrimination tasks where participants report a visual attribute of the target. Here we ask if the same results can be observed in a target detection task, where the target was absent 50% of the time. In a series of experiments, participants completed both a discrimination and detection efficient search task. Search task was manipulated within-subject, whereas target-distractor similarity was manipulated between experiments. The experimental set up minimized the contribution of crowding by displaying items with inter-item spacing that satisfied Bouma’s law, reduced eccentricity effects by applying a cortical magnification factor that increased the size of items with eccentricity, and minimized eye-movements by briefly displaying search arrays (94 ms). In the target-present conditions of both tasks, logarithmic increases in RT performance with set size were found. These results are consistent with a parallel, unlimited capacity, resolution limited architecture. Surprisingly, RTs were overall faster in the discrimination task compared to the detection task, and search efficiency was determined by an interaction between target-distractor similarity and task. Also unexpected, the target-absent conditions did not mirror the target-present functions, showing either a decrease with set size (at low similarity) or a flat function (at high similarity). The results from the target absent condition are discussed in terms of differences in termination rule between the two tasks.