When visual search target is rare: overweighting or under weighting the probability?

Abstract

The prevalence of items in visual search may have substantial performance consequences. In laboratory visual search tasks in which the target is rare, viewers are likely to miss the target. A dual-threshold model proposed by Wolfe and Van Wert (2010) assumes that in the low prevalence condition, viewers shift their criteria resulting in more miss errors. However, from the prospective of prospect theory (Kahneman & Tversky, 1979), decision makers tend to overweight small probability. To explore how viewers subjectively weight the probability in the low prevalence visual search task, we compared viewers’ criteria with the optimal criteria by presenting different probability descriptions for a fixed prevalence rate. The data from this experiment indicated that target presence had an effect on viewers’ accuracy and response times but not probability descriptions. Viewers’ criteria under different probability descriptions were higher than optimal. These results are in accordance with the dual-threshold model assumption that viewers respond “target absent” more frequently than optimal, leading to more miss errors in the low prevalence condition.