Abstract
Low prevalence search targets such as tumors in radiographs, result in disturbingly high miss rates relative to high prevalence targets (Wolfe, Horowitz, & Kenner, 2005). This suggests that a method to mitigate low prevalence misses could literally save lives. Recent work (Peltier & Becker, 2017), has suggested several individual difference measures that predict low prevalence miss rates. Building on this work, we examined metacognition’s contribution to target detection. Metacognition is the ability to introspect about one’s own thoughts, perceptions, and performance (Maniscalco, McCurdy, Odegaard, & Lau, 2017). We hypothesized that individuals with higher metacognitive estimates should more effectively monitoring their search performance, producing fewer low prevalence misses. Twenty-four participants completed a task designed to estimate metacognition (meta d’; Maniscalco & Lau, 2012), and a visual search task that manipulated target prevalence. To estimate metacognition, participants indicated which of two successive screens contained a tilted target gabor and rated their confidence. Performance was titrated to 75% accuracy. Participants then searched arrays of 24 pseudo-randomly placed objects and located targets from six potential target categories; distractors were drawn from 47 non-target categories. Overall target prevalence was held at 50% to control for motor errors. High (45%) and low (5%) prevalence trials were interleaved by manipulating the prevalence of individual target categories (counterbalanced across observers). Detection accuracy was predicted using hierarchical linear mixed effect models with metacognitive estimates, and target prevalence conditions as predictors, and with participant, trial, and target category as random effects. As hypothesized, higher metacognitive estimates result in higher low-prevalence detection accuracy (p=.008). This suggests that individuals with higher metacognitive estimates are less affected by low target prevalence, likely due to an improved ability to monitor and regulate their task performance. Futhermore, metacognitive estimates may serve as an effective way to prescreen searchers to minimize low prevalence misses.