Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Dissociating Confidence Bias and Confidence Noise in Perceptual and Knowledge-Based Decisions

Matteo Lisi¹, Marc Pabst, Lottie Wood; ¹Royal Holloway University of London

Presenter: Matteo Lisi

People often misjudge how reliable their decisions are, leading to confidence errors. These errors can be due to systematic distortions (confidence bias) or random variability (confidence noise). In this study we use a dual-decision method, in which participants are required to use confidence in a prior decision to inform expectations about subsequent choices, to examine the importance of these two sources of confidence error in both perceptual and knowledge-based tasks. Across a reanalysis of published data and two new studies, we find that perceptual tasks elicit under-confidence relative to Bayesian optimal predictions, while knowledge-based tasks exhibit increased confidence noise. Additional conditions using calorie estimation tasks suggest that some domains blend perceptual and knowledge-based decision features. These findings provide novel insights into the computational structure of confidence, suggesting that different cognitive domains are subject to distinct metacognitive constraints.

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF