Metacognitive insight improves unfamiliar face matching through effective resource allocation

Little, Zoë, Towler, Alice, Persson, Annabel, White, David
Cognitive Research: Principles and Implications · Springer Nature · 2026-08-28 · DOI: 10.1186/s41235-026-00754-6

Abstract

Metacognitive insight—the ability to monitor one’s own performance—enables efficient allocation of limited resources, but whether this mechanism underlies performance on perceptual tasks has never been directly tested. Here we asked whether insight into face matching performance drives time-allocation decisions to improve accuracy. Participants ( n = 74) completed a two-phase task: an initial evaluation in which they saw face pairs for two seconds and selected up to half for extended review, followed by a 30-s review of all pairs. Accuracy in the review phase improved substantially more for selected than not-selected pairs (23% vs. 6%), showing that participants identified their most error-prone decisions and that additional time on those decisions was especially beneficial. This improvement was most salient for non-matching face pairs, suggesting that the well-documented dissociation between match and non-match performance has a metacognitive basis. These results provide the first mechanistic link between metacognitive insight and face matching accuracy: people who know which decisions to spend time on can use that knowledge to make fewer errors, particularly the costly error of failing to detect identity imposters. This finding has direct implications for applied settings—such as passport checking at international borders—where time pressure is inherent and non-match errors carry the greatest consequence. Matching unfamiliar faces is a surprisingly error-prone task, even under optimal conditions. In real-world settings like passport checking, people must do this task quickly. Crucially, time pressure in these settings is usually global: an examiner must process many cases within a fixed period, but can choose to spend longer on some cases than others. To perform well, they need enough insight into their own decisions to know which cases require more time. We show, for the first time, that people have exactly this kind of insight. Participants were able to identify the face pairs they were least likely to have correctly categorised, chose to review those pairs for longer, and improved their accuracy on them substantially more than on pairs they did not select. This benefit was largest for non-matching pairs, where correct identifications are most consequential in security settings to prevent identity imposters from going undetected. These findings establish a mechanistic link between metacognitive insight and face matching accuracy, and suggest that supporting effective resource allocation, such as through training or interface design, could reduce errors in applied settings.

Read at publisher →

Similar articles