Search Termination Time in Visual Search is Determined by Expectation of Target Prevalence

Abstract

Reaction times for search termination decisions (“no” RTs) are shorter when targets are rarer (lower “target prevalence”). Is this “prevalence effect” a result of the observers' expectation of target prevalence or of repetition priming from repeated “no” responses? If number of repetitions determines “no” RTs, independent of target prevalence, evidence would favor a repetition priming account. In contrast, if implicit or explicit knowledge of the target prevalence determines “no” RTs, regardless of repetitions, evidence would favor an expectation account. In experiment 1, we conducted visual search tasks where prevalence varied across blocks, and analyzed correct rejection trials in time series. The main effect of number of repetitions was not statistically significant, but the main effect of target prevalence was statistically significant. In experiment 2, high-prevalence and low-prevalence trials were randomly mixed and the prevalence for the next trial was cued. The result showed that the “no” RTs in the low prevalence trials were still faster than those on high prevalence trials. These results suggest that the prevalence effect on the “no” RTs is the result of the expectation of target prevalence.