Evolution and Dynamics of Perceptual Confidence: From Perceptual Learning to Global Confidence Formation

This thesis examines the evolution and dynamics of perceptual confidence, defined as the subjective estimate of being correct in perceptual judgments and a central component of metacognition. Across three projects, it investigates how confidence is computed, how it changes through experience, and how it integrates into global self-evaluations. The first project shows that during perceptual learning without external feedback, confidence undergoes systematic reconfiguration. Learners display reduced confidence efficiency but increased confidence noise and boost, indicating adaptive changes in the weighting of decision-level information. The second project explores how local confidence judgments combine into a global sense of performance. Salient experiences, such as monetary incentives, are shown to enhance or disrupt metacognitive integration depending on their timing and relevance. The third project demonstrates that forming global confidence judgments requires greater cognitive effort than local ones, highlighting an inherent cost in metacognitive evaluation. Together, these findings establish confidence as a dynamic and context-sensitive process rather than a static measure of accuracy, advancing current models of metacognition and clarifying how humans monitor, adapt, and learn from their own cognitive processes.