Title: Biases in belief and simulation in obsession-compulsion

Speaker: Tricia Seow (Senior Research Fellow, Max-Planck UCL Centre for Computational Psychiatry and Ageing Research)

Location: In person at 12 Queen Square & online via Zoom

Abstract: Compulsions and obsessions are core symptoms of obsessive-compulsive disorder (OCD). In this talk, I will focus on two mental processes that contribute to these symptoms: biases in belief and biases in mental simulation. First, I will discuss metacognitive biases—explicit confidence ratings—showing that overconfidence in decisions is linked to a transdiagnostic dimension of compulsivity, replicated across multiple cognitive domains. The rigidity of this overconfidence appears to arise from an impaired sensitivity to external feedback signals. Next, I will turn to the period preceding explicit belief reports—deliberation—and examine mental simulation, measured using magnetoencephalography (MEG) to track neural replay during decision-making. I will show that humans use replay to simulate options before decisions, and that in OCD, individuals with higher obsessional symptoms show exaggerated replay for salient outcomes, which is associated with increased avoidance behaviour. Together, these findings suggest that biases in implicit simulation and explicit belief may represent complementary mechanisms, contributing to obsessional and compulsive symptoms, respectively. I will close with a hierarchical framework that extends to broader forms of self-reflection in the clinic and offers a step toward a unified conceptualisation of metacognition across neurocognitive and clinical disciplines.

Title: LLMs as models of high level perceptual processing

Speaker: Adrien Doerig (Guest Professor, Freie Universität Berlin)

Location: In person at 12 Queen Square & online via Zoom

Abstract: The brain extracts far more from vision than just objects: it encodes relations, context, and meaning. I will show that embeddings from large language models (LLMs) provide quantitative models of this complexity, aligning closely with brain activity evoked by natural scenes. These embeddings capture functional selectivity across cortex, enable reconstruction of captions from brain data, and provide a more brain-like representational format than many state-of-the-art alternatives. I will then look beyond our results to recent findings and ongoing work in my lab, and speculate that LLM embeddings may offer a general framework for modeling high-level perceptual processing in the human brain.

Speaker Bluesky: https://bsky.app/profile/adriendoerig.bsky.social