Abstract
While many visual search tasks require looking for stationary objects (e.g., finding mustard in the refrigerator), some search tasks are more complex, such as when security guards monitor multiple closed-circuit camera feeds. We utilized Multi-element Asynchronous Dynamic (MAD; Kunar & Watson, 2011) search displays to investigate search efficiency in dynamic settings. In Experiments 1a and 1b, participants searched categorically for target animals in displays with stationary, slowly blinking, and moving objects. We manipulated the rate at which targets possessed the various dynamic features (between-subjects) to determine if the relative occurrence of such features or the dynamic features themselves would have a dominant effect on search behavior (e.g., errors, RTs). Stimuli were photographs of animals in Experiment 1a (amid varying visual set sizes), and line drawings of similar categories (with a fixed set size) in Experiment 1b. We found that both dynamic features and their prevalence within target categories affected behavior, albeit somewhat inconsistently between experiments. In Experiment 2, participants completed blocks of MAD and fully static search (with trial-by-trial accuracy feedback); we examined search termination behavior surrounding missed targets (stimuli and set size were similar to Experiment 1b). Participants elicited higher miss rates in MAD search (compared to static search), but there were no significant differences in hit RTs between search displays or dynamic features. Misses in MAD search appear to be due largely to premature search termination errors, while misses in static search did not show this same early quitting trend. Together, we found that both specific dynamic features and their associations with targets affected search performance. Dynamic displays also resulted in more premature terminations compared to less complex displays. These results may inform training programs for operators that monitor complex displays, by emphasizing cognitive control to carefully rule out distractors and maintain an effective quitting threshold.
Meeting abstract presented at VSS 2016