Proceedings of the Annual Meeting of the Cognitive Science Society

People vary in their abilities to form self-knowledge (metacognitive monitoring) and use self-knowledge to guide behaviors (metacognitive control). To assess the test-retest reliability of metacognition, we tested participants with a perceptual task four times over a month. We quantified metacognitive monitoring by computing metacognitive efficiency and bias metrics within the hmetad model (O'Neill et al., 2026), and devised novel analogous measures of metacognitive control bias and sensitivity using logistic regression. Our preliminary results (N=157, half the planned sample) showed moderate test-retest reliability of metacognitive bias (r = .67 [.60 .72]). By developing a hierarchical Bayesian model of the longitudinal structure, we observed robust (albeit still low to moderate) test-retest reliability of metacognitive efficiency (r = .45 [.22 .66]). Moreover, test-retest reliabilities of both control bias (r = .86 [.82 .89]) and sensitivity (r = .77 [.68 .84]) were good. Our results provide methodological foundations for individual differences research on metacognition.