Eye-tracking as a lens into meta-reasoning

Abstract

Central questions in metacognitive research are how reliably people’s confidence reflect success and how mental effort is regulated during cognitive tasks. Ackerman’s Bird’s-Eye View of Cue Integration methodology exposed confidence biases in a geometric problem-solving task. We replicated one experiment, incorporating eye-tracking to expose additional confidence predictors and insights into mental effort regulation. Fifty-four university students performed the task wearing eye-tracking glasses. The original confidence biases were replicated. Notably, the best predictive model for confidence included eye-tracking measures. Mean pupil diameter and number of visits to stimuli uniquely predicted success and confidence, highlighting the role of mental effort investment and attentional shifts. Individual-level analyses revealed that while the eye-tracking measures identified individuals with higher success, they did not reliably predict confidence or resolution (confidence associated with success). These findings underscore the potential of eye-tracking to provide valuable insights into metacognitive monitoring and mental effort regulation, and suggest eye-tracking potential utility when eliciting confidence ratings is impractical.