Publications

Abstract

Previous research has demonstrated that discriminability of target and background elements has powerful effects on visual search performance. Typically, discriminability has been manipulated between blocks or subjects, allowing subjects to anticipate the discriminability in advance of each trial, and the results, therefore, have been consistent with a wide range of models. Two experiments are reported in which the number of distractors similar to the target was varied from trial to trial, preventing such anticipation. In Experiment 1, subjects' response times increased as target-confusable distractors were added, for both "yes" and "no" responses. In Experiment 2, subjects searched for either of two targets. Distractors similar to one target slowed down detection of either that target or the other target, to similar degrees. The results indicate that decision noise, rather than feature-specific inhibition, is the major source of the latency effects. The results do not support models proposing that decisions in speeded search are based upon information integrated across all positions in the display. The data also would require (perhaps implausible) modification of the independent-channels model. They are broadly consistent, however, with models hypothesizing a parallel search followed by some slow serial check-ing, such as that developed by Hoffman (1978).