Jan Skerswetat¹ (), Rushil Kumar¹, Andrew Browne¹, Cherlyn J Ng¹; ¹University of California, Irvine

Visual metacognitive confidence, the ability to evaluate one's own performance, is underutilized in clinical vision assessments. To address this, the Angular Indication Measurement Metacognition (AIM-MetaCognition) paradigm was developed to jointly assess visual functions and metacognitive confidence. As proof-of-concept, AIM-MetaCognition assessed visual acuity (VA) in twenty healthy adults. Participants reported gap orientations of Landolt-Cs (4×4 grid, randomized) and rated confidence per cell. Stimulus sizes were log-scaled and adaptively adjusted (threshold ±2σ). Hence, VA and confidence were estimated on continuous scales. Reported gap-orientation errors and confidence ratings were fitted separately to psychometric functions to estimate VA derived from confidence and performance. Metacognitive calibration was assessed using linear regression. A Confidence-Signal-Detection (CSD) model estimated the confidence scaling factor (k), which quantifies how sensory evidence translated to confidence. AIM-Metacognition was validated in several experiments assessing its repeatability, and effects of binocular versus monocular viewing, simulated vision loss via Gaussian blur, task instructions, predictive versus retrospective confidence judgements. Additional experiments adaptively estimated VA on confidence alone and confidence-weighted performance. Cross-visual function generalizability was also demonstrated with AIM-Metacognition. Bland-Altman analyses revealed no significant test-retest bias for psychometric or metacognitive parameters. Confidence ratings were bimodally distributed with comparable proportions of low- and high- judgements. RM-ANOVA showed significant main effects of viewing condition and measurement type on VA(p < .001): binocular VA outperformed monocular VA, and Confidence-VA was worse than Performance-VA(p < .01). Binocular viewing produced greater CSD-k miscalibration than monocular (p = .03). Neither task instruction(Confidence versus Certainty) nor the method of determining VA (based on performance, confidence, or both) affected VA(p>.05). AIM-Metacognition offers a reliable, self-administered, and adaptive method for assessing visual functions and confidence. By capturing metacognition, it provides new insights for evaluating vision for clinical and basic vision science.

Acknowledgements: J.S. co-inventor of AIM (Angular Indication Measurement) method, owned by Northeastern University, Boston. J.S. is a shareholder and co-founder of PerZeption Inc., which has an exclusive licensing agreement for AIM with Northeastern University