Abstract
BACKGROUND: Consider two visual search paradigms: Chun and Jiang's “Contextual Cueing” and Wolfe's “Repeated Search”. In contextual cueing experiments, observers are presented with a mixture of repeated and novel visual search displays. The usual finding from these experiments is that, as the number of repetitions increases, RTs to find the targets in repeated displays become markedly faster than RTs for unrepeated displays. In repeated search experiments, repetition is more dramatic. The same display is unchanged for hundreds of trials. Observers are asked about the presence or absence of specific targets in the unchanging display. In contrast to the contextual cueing experiments, search does not become more efficient over time in repeated search nor is search more efficient than in unrepeated displays. METHODS: We explore this apparent contradiction by focusing on two critical differences between these designs. First, in a standard contextual cueing experiment, the target location on one trial is never occupied by a distractor item on another trial, whereas in a repeated search task the same items change roles from target to distractor from trial to trial. Second, in a contextual cueing experiment there is only one possible target location, whereas in a repeated search task there are several. Here, we manipulate two aspects of the contextual cueing design: 1) a distractor item could occupy a target location and 2) a target item could appear in up to four unique positions (which were never occupied by distractors). RESULTS: When targets and distractors can appear in the same locations on different trials, then we find no evidence for contextual cueing. However, if the targets are restricted to unique locations, we can obtain contextual cueing at multiple locations. Initially, the multiple location effect is weaker than a single location case but is strengthened after extended training.
National Institute of Mental Health