Symposium

Paul Fletcher is Bernard Wolfe Professor of Health Neuroscience, Wellcome Trust Investigator and a Fellow of Clare College, Cambridge. He is an honorary Consultant Psychiatrist with Cambridgeshire and Peterborough NHS Trust.Paul trained in Medicine and then Psychiatry, practising full-time psychiatry in London before becoming involved in neuroimaging research exploring the brain basis of memory in health and disease. He took a PhD in Cognitive Neuroscience and spent a year studying neuroanatomy in Dusseldorf before moving to Cambridge where he is currently based. He currently divides his time between research, teaching and clinical work in psychiatry. He is director of studies for preclinical medicine at Clare College, Cambridge University.Paul uses combinations of pharmacology, neuroimaging and behavioural studies in healthy and clinical populations, with the aim of understanding the basis of perception, learning and decision-making in the human brain, as well as how this may be altered under conditions of illness or drug disturbance. His research draws on the principle that the mind/brain is a model of its world. It constructs reality in order to predict that world as a means of interacting with it efficiently, fruitfully and safely. In doing so, it is prone to changing or even losing contact with the reality of the world. Through understanding this, and the many ways in which we are all prone to losing touch with reality, the aim is to understand mental illness and mental distress more deeply.

ABSTRACT

Prediction, Perception and Unshared Realities: Understanding Psychosis Through Computational Psychiatry

In this presentation I will try to outline the principles of computational psychiatry approaches to understanding the complex symptoms and experiences that characterise mental illness. I will focus on one of the most complex and challenging areas from my own clinical and research work: psychosis – the experience of an unshared reality. How can we understand the bizarre beliefs and vivid perceptions that characterise psychosis in terms that can encompass both the profound subjective changes experienced by a person constructing a new reality and the accompanying neurobiological changes?Computational psychiatry has emerged from attempts to bridge such an explanatory gap and I will show how simple principles - based on the idea of the brain attempting to model the world through prediction-based inference – offer a powerful framework to support these attempts to link symptoms to neural processes. At the same time, it is important to recognise the shortcomings of such a framework and I will explore the areas where computational psychiatry fails as well as where it succeeds. 

Peggy Seriès is a Full Professor in Computational Psychiatry at the University of Edinburgh. With a background in artificial intelligence and biomathematics, Peggy Seriès did her PhD in computational neuroscience in Paris with Yves Frégnac and Jean Lorenceau and postdocs with Alexandre Pouget at University of Rochester, Peter Latham at the Gatsby Computational Neuroscience Unit in London and Eero Simoncelli at NYU. She was then working on models of the visual cortex, questions related to population coding, information transmission and Bayesian theories of the brain. 

She joined the School of Informatics at the University of Edinburgh in 2006 where she was recently promoted as a Full Professor in Computational Psychiatry. 

Peggy's current interests centre on computational models of cognition, with a particular emphasis on learning and decision making and their application for the understanding of mental illness.  Her research projects aim at a better understanding of behavioural differences in schizophrenia, depression, anxiety and autism and their underlying neurobiological mechanisms. In 2020, she has edited the first accessible textbook in the emerging domain of Computational Psychiatry ("Computational Psychiatry: a Primer", MIT Press). 

ABSTRACT

Ten Years of Bayesian Theories of Autism: What Have We Learned?

Predictive coding and the Bayesian brain have become dominant theories of  perception and cognition. In both theories, top-down signals (predictions or priors) are combined with bottom-up ones (prediction errors or likelihoods), weighted by their precision. This framework has also been proposed to provide an explanation for the diverse autistic symptoms, with an imbalance in the relative precisions of the two signals being at the root of the condition (for a review see e.g. Palmer et al., 2017). This hypothesis has now been studied for nearly 10 years. I will describe the different variants of the theories and review the literature, including work from my lab, to discuss the extent to which they are now backed by quantitative evidence. I will also discuss limitations of current work and guidelines for future research.

Eran Eldar is a faculty member of the Department of Psychology and Department of Cognitive & Brain Sciences at the Hebrew University of Jerusalem, where he heads the laboratory for computational cognition. Prior to that, he completed a MD at Tel-Aviv University, a PhD in Neuroscience at Princeton University, and a postdoctoral fellowship at the Max Planck UCL Centre for Computational Psychiatry in London. His research aims to understand the computations people perform to learn and make decisions, and how these computations influence and are influenced by emotions, moods, and personality. In particular, Eran has been using this computational approach to investigate what emotions and moods are, how they help guide people’s expectations and behavior, and how their malfunction might lead to mental health disorders.

ABSTRACT

Emotions as Computations

To understand psychopathologies of emotions and moods, we must first ask how emotions and moods guide adaptive behavior. This question, however, has been notoriously difficult to answer despite decades of ongoing debates. In this talk, I will show how recent progress in the computational modeling of learning and decision making may resolve this ongoing controversy. I will propose that distinct emotions and moods mediate distinct computations that serve to evaluate states, actions, or uncertain prospects. This mapping of emotions to computations integrates a wide range of research on the causes and consequences of different emotions, and suggest that emotions are integral to how we evaluate outcomes (pleasure & pain), learn to predict them (happiness & sadness), use them to inform our (frustration & content) and others’ (anger & gratitude) actions, and plan in order to realize (desire & hope) or avoid (fear & anxiety) uncertain outcomes. Finally, this mapping implicates key computational factors that may give rise to excessive emotions and moods of different types, and thus, to different psychopathologies of emotion.

Dr. Xiaosi Gu is an Associate Professor of Psychiatry and Neuroscience and Director of the Center for Computational Psychiatry at the Icahn School of Medicine at Mount Sinai in New York City. She is one of the foremost researchers in the nascent area of computational psychiatry. Specifically, her research examines how the human brain computes information during decision-making as we pursue rewards or engage in social interactions across the lifespan. She is currently leading multiple NIH and foundation grants to study on how the social brain might go awry across various psychiatric disorders, including autism spectrum disorder, mood disorders, addiction, etc.

Dr. Gu received her Ph.D. in Neuroscience at the Icahn School of Medicine at Mount Sinai and postdoctoral training in computational psychiatry at the Wellcome Trust Centre for Neuroimaging, University College London (UCL). Before re-joining Mount Sinai, Dr. Gu held faculty positions at the University of Texas, Dallas and UT Southwestern Medical Center. She has published widely in high impact scientific journals and served as a scientific advisor for nonprofit organizations across the world such as the Wellcome Trust. Dr. Gu is also an Editor-in-Chief for the new journal Computational Psychiatry. She is currently serving on the organizing committee of the Inaugural Computational Psychiatry Conference, which evolved from the UCL Computational Psychiatry Course she established in 2014.

Beyond her scientific work, Dr. Gu is an avid advocate for raising public awareness in mental health. She frequently speaks at mental health-related events including a 2018 Tedx conference and a recent Global Partnerships in Brain Research Science Summit at the 2022 UN General Assembly.

When she is not studying the brain, she is either chasing tennis balls or chasing her 6-year-old ☺

ABSTRACT

A Social Neuroscience Approach Towards Computational Psychiatry

Computational psychiatry (CP), a nascent field with close connections to biological psychiatry and computational neuroscience, has made important contributions to mental health research in terms of bringing computational frameworks and methods. However, the explanatory power ofCP remains limited, as it has thus far almost exclusively focused on basic perceptual inference and reinforcement learning processes. In this talk, I will argue that modeling social behaviors represents an important future direction of CP, as human social relationships can be both a major drive of psychiatric disorders as well as an important basis for treatment and intervention. Using norm adaptation and social controllability as examples, I will present our recent findings of how impaired social computations manifest in disorders that are not traditionally considered as “social disorders" such as addiction, OCD, and delusion. I will conclude by discussing the therapeutic implications of this work.

Matteo Colombo is an Associate Professor in the Tilburg Center for Logic, Ethics, and Philosophy of Science, and in the Department of Philosophy at Tilburg University, The Netherlands. Much of his work is in the foundations of computational neuroscience, philosophy of science, philosophy of mind, and moral psychology. He is especially interested in the places where these areas overlap. 

ABSTRACT

Does Neural Miscomputation Explain Mental Disorders? 

Computational psychiatrists often use a concept of miscomputation in their explanations of mental illnesses. Underlying these explanations are the ideas that brains are computing systems and that brain malfunction explains mental illness. In this talk, I want to put into better focus the ideas of a computing system and of miscomputation to explore how it could matter that miscomputation explains mental illness.