Unity and diversity of metacognition.

Despite many research efforts dedicated toward deciphering the functional architecture underlying metacognition, it is still unclear if there is a common metacognitive resource for different functional requirements. Here, using laboratory measures of metacognition across several domains in a large sample (N = 155), we examined whether metacognitive ability is determined by universal or modular processes, and whether "online" laboratory measures are related to "offline" self-report measures of real-world metacognition. Trial-by-trial ratings of confidence were collected in pairs of tasks tapping into the domains of visual perception and episodic memory, whereas in the attention-to-action domain, one task obtained trial-by-trial confidence ratings and the other signal-dependent measures of error awareness. Relationships between metacognitive efficiency scores across paradigms and domains were assessed using a combination of correlational and latent variable approaches. The results point to a mixture of domain-general (unity) and domain-specific (diversity) components. Specifically, Bayesian correlation estimates of metacognitive efficiency as well as confirmatory factor analysis of interdomain correlations suggested metacognition about perceptual judgments to be mostly domain-specific, whereas convergent indications for interrelations between metacognition in the domains of attention-to-action and memory implied the coexistence of partly specialized metacognitive subsystems. Notably, offline measures of metacognition represented online metacognitive bias rather than online metacognitive efficiency, underscoring prevalent skepticism whether self-report questionnaires provide a useful proxy in metacognition research, as they appear susceptible to potentially unreliable introspections and memory distortions. Overall, our results indicate a constitution of both universal and specialized parts for task-based metacognition. (PsycInfo Database Record (c) 2022 APA, all rights reserved).