Object-scene relationships vary the magnitude of target prevalence effects in visual search

Efficiency of visual search in real-world tasks is affected by several factors, including scene context and target prevalence. Observers are more efficient at detecting target objects in congruent locations, and less efficient at detecting rare targets. Although target prevalence and placement often covary, previous research has investigated context and prevalence effects independently. We conducted 2 experiments to explore the potential interaction between scene context and target prevalence effects. In Experiment 1, we varied target prevalence (high, low) and context (congruent, incongruent), and, for congruent contexts, target location (typical, atypical). Experiment 2 focused on the interaction between target prevalence (high, low) and location (typical, atypical) for congruent contexts, and recorded observers' eye movements to examine search strategies. Observers were poorer at detecting low versus high prevalence targets; however, prevalence effects were significantly reduced for targets in typical, congruent locations compared with atypical or incongruent locations. Eye movement analyses in Experiment 2 revealed this was related to observers dwelling disproportionately on the most typical target locations within a scene. This suggests that a byproduct of contextual guidance within scenes is that placing targets in unexpected or atypical locations will further increase miss rates for uncommon targets, which has implications for real-world situations in which rare targets appear in unexpected locations. Although prevalence effects are robust, our results suggest potential for mitigating the negative consequences of low prevalence through targeted training that teaches observers where to focus their search. (PsycINFO Database Record