Location probability learning in visual search of targets with low prevalence

Visual search is a daily activity, yet people often miss targets when they occur infrequently, leading to costly errors, such as overlooking rare cancerous tumors during routine screenings. Previous research on this low-prevalence effect (LPE) has focused on situations where the target can appear anywhere in the visual field, raising questions about how the LPE interacts with other statistical learning mechanisms. This study examines how people integrate a search target's spatial probability with its prevalence rate. In two experiments, participants searched for a target present either 50% (high prevalence) or 10% (low prevalence) of the time. When present, the target disproportionately appeared in one visual quadrant, allowing participants to learn its spatial probability. We tested whether participants could learn this spatial probability and if learning was affected by the target's prevalence rate. Results showed that participants prioritized high-probability spatial locations under both prevalence conditions. When tested in a neutral phase with an even spatial distribution and a 50% target prevalence, the two experiments showed equivalent probability learning, indicating that neither short-term repetition (stronger in high prevalence) nor spaced learning (more likely in low prevalence) was significant. Miss rates were higher during low-prevalence training, but the LPE was smaller in high-probability target locations. These findings suggest that prevalence rates and spatial probability have independent effects, and the LPE may be reduced in high-probability locations.