Metacognition refers to the capacity to reflect on our own cognition and mental states, and can be usefully quantified by the extent to which subjective appraisals (such as confidence) track behavioural performance. I will outline how we can derive metrics of metacognition from confidence rating data within Bayesian generative models, and use these metrics both to understand the structure of metacognition across distinct tasks, and to isolate metacognitive distortions that map onto transdiagnostic symptom clusters in large-N datasets. I will then present recent work aimed at understanding the hierarchical structure of metacognition, ranging from local confidence in individual decisions to global estimates of skills and abilities. Combining brain imaging and computational modeling we have uncovered distinct neurocognitive underpinnings of global and local metacognition, and identified selective differences in global metacognition that relate to ageing and mental health. I will close by outlining a theoretical framework in which interactions between these levels underpins the formation of self-beliefs.

Prof. Stephen Fleming
Department of Experimental Psychology and
Wellcome Centre for Human Neuroimaging
University College London




Stephen Fleming – How local and global metacognition shape mental health