Speakers

Computational Psychiatry: towards the multiregional brain

In recent years it has been increasingly recognized that advances in neuroscience, in close interplay with clinical research, have the potential to help build a solid foundation for better diagnostic nosology, biomarkers and treatments of mental disorders. For instance, people afflicted with addiction continue to choose actions (e.g. taking cocaine) in spite of devastating consequences, thus the core problem is abnormal decision making. Can neural circuit dissection of decision making and reinforcement learning lead to new treatment of drug abuse? As another example, working memory is impaired in Schizophrenia and ADHD, can neuroscientific knowledge about the brain basis of working memory be translated into better pharmacology for working memory deficits? Computational psychiatry fosters collaborations between systems/theoretical neuroscientists and researchers in mental health. Here I will introduce three approaches of this nascent field that involve analysis of big data, normative theory and biologically-based neural circuit modeling. I will provide concrete examples that emphasize cross-level understanding from molecules to recurrent neural network dynamics to behavior, and frontier research on connectome-based large-scale modeling of the multiregional brain.

[Bio]

Xiao-Jing Wang is Distinguished Global Professor of Neural Science, director of the Swartz Center for Theoretical Neuroscience at New York University. Between 2012 and 2017 he served as the founding provost and vice president for research at the Shanghai campus of NYU.  Previously he was Professor at Yale University School of Medicine. Dr. Wang uses theory and mathematical models to investigate neural mechanisms of cognitive functions such as working memory, decision-making and executive control of flexible behavior, with a special interest in the prefrontal cortex. He also worked with clinical researchers to foster the nascent field of Computational Psychiatry. More recently, his group developed connectome-based modeling of large-scale multi-regional brain circuits for distributed dynamics and cognition. His lab’s publications have been cited more than 43,800 and his h-index is 99. Dr. Wang is a recipient of Alfred P. Sloan Research Fellowship, Guggenheim Fellowship, Swartz Prize for Theoretical and Computational Neuroscience Prize, Goldman-Rakic Prize for Outstanding Achievement in Cognitive Neuroscience, Highly Cited Researcher by Clarivate Analytics/Web of Science Group. He was elected to the Royal Academy of Belgium.

Bring in the social context into computational psychiatry

Evolutionally, the human brain is constantly shaped by the increasingly complex social structure of human society; as such, social factors are essential to our mental wellbeing. However, the social context is largely ignored in the field of computational psychiatry and in biological psychiatry in general, leaving an enormous gap between theory and clinical reality. In this talk, I will discuss recent attempts of modeling social behaviors and how incorporating specific computational theories of human interactions may create greater opportunities for bridging the gap between underlying mechanisms and clinical applications.

[Bio]

Dr. Xiaosi Gu is a tenured Associate Professor of Psychiatry and Neuroscience and Director of the Center for Computational Psychiatry at the Icahn School of Medicine at Mount Sinai (ISMMS) in New York City. Dr. Gu is one of the foremost researchers in the nascent area of computational psychiatry. Specifically, her research examines the neural computations underlying complex cognition in humans such as beliefs and social decision-making. She is currently leading multiple NIH and private foundation grants to study on how these processes might go awry across various psychiatric disorders, including depression, autism, personality disorders, and addiction.

Dr. Gu received her Ph.D. in Neuroscience at ISMMS and postdoctoral training at the Wellcome Trust Centre for Neuroimaging, University College London (UCL). Before re-joining ISMMS, Dr. Gu held faculty positions at the University of Texas (UT), Dallas and UT Southwestern Medical Center.  She has published widely in high impact scientific journals and has advised organizations worldwide including the Wellcome Trust and the Max Planck Society. Dr. Gu is an Editor-in-Chief for the newly established journal Computational Psychiatry.  She is also currently chairing the organizing committee of the 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 is a frequent speaker and panelist at mental health-related public events including a Tedx conference (2018) and the Global Partnerships in Brain Research Science Summit at the United Nations General Assembly (2022, 2023).

Computational Neuropharmacological Modeling in Psychiatric Disease

The complex hierarchical structure of the human brain poses significant challenges in directly observing how neuropharmacological interventions influence clinical electrophysiology. Computational modeling plays a pivotal role in bridging this gap by simulating the impacts of drugs and neural interactions across multiple brain function levels, from cellular receptors to entire brain networks. This talk will focus on the use of dynamic causal modeling (DCM) integrated with a canonical microcircuit model of N-methyl-D-aspartate (CMM-NMDA) to analyze and predict long-term changes in EEG signals following clozapine treatment in schizophrenia patients. We will examine how specific neurobiological changes at both cellular and network levels, such as alterations in membrane capacitance, receptor time constants, and synaptic connectivity, correlate with modifications in the EEG’s cross-spectral density (CSD). A crucial element of our approach involves virtual perturbation analysis, which assesses the impact of various neural properties on EEG dynamics. This study demonstrates how computational neuropharmacological modeling acts as a vital link between clinical practice and theoretical neuropharmacology. By enhancing our understanding of how antipsychotic medications affect brain electrophysiology, this method strengthens the connection between computational tools and clinical data. Such integration is essential for advancing psychiatric treatments and improving patient outcomes, highlighting the critical role of computational strategies in contemporary psychiatry.

[Bio]

Dr. Hae-Jeong Park is a Professor at the Graduate School of Medical Science, Brain Korea 21 Project, Department of Nuclear Medicine, Psychiatry, Yonsei University College of Medicine and Department of Cognitive Science, Yonsei University, Seoul, South Korea. Dr. Park received a B.S. degree in Electrical Engineering and M.S. and Ph.D. in Biomedical Engineering from Seoul National University. Dr. Park did his post-doc at the Laboratory of Neuroscience, Department of Psychiatry and Surgical Planning Laboratory, Brigham and Women's Hospital, Harvard Medical School, where he developed whole-brain tractography in 2002. He moved to Yonsei University College of Medicine in 2004 and conducted clinical and applied cognitive neurosciences by developing multimodal neuroimaging techniques. In 2012, Dr. Park spent his sabbatical at the Wellcome trust center for neuroimaging at University College London, UK. Dr. Park is the director of the Laboratory of Molecular Neuroimaging Technology and chair of the Center for Systems and Translational Brain Science, Institute of Human Complexity and Systems Science, Yonsei University, where he tries to interface neuroimaging methods with the investigation and treatment of psychiatric and neurological diseases through computational, nonlinear systems, and theoretical analysis of the brain. Dr. Park is also interested in neuroimaging and art and served as the director of the neuroart division in the media art performance in 2008 and 2009. He is the chairman of the Board of the Korean Society for Human Brain Mapping from 2023 to 2024.

Empirically testing Bayesian accounts of altered perception in the nonclinical continuum of psychosis

Conflicting Bayesian theories postulate aberrations in either top-down or bottom-up processing in attempts to understand the occurrence of altered perceptual experiences. The top-down theory predicts an overreliance on prior beliefs whereas the bottom-up theory predicts an overreliance on current sensory information.

In this talk, I will show how we empirically adjudicated between these models by mathematically quantifying subjective prior and likelihood precisions and the relative reliance on priors and sensory information. Across two datasets (discovery dataset n=363; validation dataset n=782) we showed that psychotic-like experiences were associated with an overweighting of sensory information relative to prior expectations, driven by decreased precision in priors. However, participants with greater psychotic-like experiences also encoded likelihood information with greater sensory noise.

We propose a revised model for perceptual inference in psychosis, whereby both prior and sensory representations are less precise when psychotic-like experiences increase, but their relative imprecision is such that it leads to an overreliance on likelihood information, in keeping with bottom-up theories.

[Bio]

Associate Professor Marta Garrido leads the Cognitive Neuroscience and Computational Psychiatry Laboratory and is the Director of the Cognitive Neuroscience Hub at the Melbourne School of Psychological Sciences. Marta is also a Research Program Lead at the Graeme Clark Institute for Biomedical Engineering, at The University of Melbourne. Marta initially trained in Engineering Physics at the University of Lisbon, and then did a PhD in Neuroscience at University College London. Marta’s team uses a combination of brain imaging techniques and computational modelling to understand how the brains of typical individuals and people with psychiatric conditions learn from experience and make decisions. To date, Marta has produced 80 peer-reviewed publications some in the world’s most prestigious journals including Science, Nature Reviews Neuroscience, Nature Neuroscience, eLife, Current Biology, and PNAS. Her work has been cited over 6000 times and she has secured A$23 million in competitive funding as a chief investigator. The quality of Marta’s work has been recognised by prestigious awards including the 2020 Paxinos-Watson Prize from the Australasian Neuroscience Society and the 2019 Aubrey Lewis Award from Biological Psychiatry Australia. Marta is a former DECRA fellow, the past Chair of the Organisation for Human Brain Mapping, Australian Chapter, and an advocate for Open Science.

Toward building a decision-making paradigm for dynamic and real-world addictive behaviors

In this talk, I will discuss the progression and recent developments in computational approaches designed to capture the dynamic aspects of human cognition and psychopathology, with a particular focus on addiction. Specifically, I will highlight the transition from traditional laboratory tasks and cognitive models to daily digital phenotyping aided by Bayesian adaptive testing and more immersive, naturalistic real-time tasks. I will also discuss the use of deep neural network models for analyzing data from these real-time tasks. This line of research may address some limitations of conventional approaches and shed light on the development of reliable phenotypes for psychiatric disorders. 

[Bio]

Woo-Young (Young) Ahn is an Associate Professor in the Department of Psychology at Seoul National University (September 2019 - Present; Assistant Professor from September 2017 to August 2019). He was previously an Assistant Professor in the Department of Psychology and an affiliated faculty at Translational Data Analytics at The Ohio State University (August 2015 – August 2017). He earned his B.S. in materials science & engineering in 2002 from Seoul National University and then went to Harvard University as a doctoral candidate for applied physics and received his S.M. in applied physics in 2003. Due to his interests in the human mind, he decided to change his major to clinical psychology so that he could study the human mind from multiple perspectives. He continued on to receive his M.A. in clinical psychology from Seoul National University in 2006, and his Ph.D. in clinical psychology from Indiana University, Bloomington in August 2012 co-advised by Jerome Busemeyer and Brian O’Donnell. He completed his (APA accredited) clinical psychology internship at the University of Illinois at Chicago (UIC) in June 2012. He worked then as a postdoc with Read Montague and Peter Dayan for two years at Virginia Tech Carilion Research Institute (VTCRI) and for a year at Virginia Commonwealth University Institute for Drug and Alcohol Studies.

Computational and Neuroimaging Approaches for Understanding Psychopathology using Reinforcement Learning

Humans constantly make decisions in their day-to-day lives and update behavior based on the feedback to maximize positive and reduce negative outcomes. This behavior called Reinforcement Learning is critical for survival and is impaired across several psychiatric illnesses. Reinforcement learning is not a unitary construct and is multifaceted. Different facets of RL are often impaired across several psychiatric disorders and are associated with different clinical phenotypes. Dr. Kumar will talk about how we can use computational models and multimodal imaging methods to identify functional and structural connectome of these RL facets. In addition, she will discuss ways to explain heterogeneity in psychopathology using RL behavior.

[Bio]

Dr. Kumar completed her bachelor’s in Instrumentation Engineering at University of Madras, India. She then completed her master’s in medical Imaging and PhD in Mental Health at University of Aberdeen, Scotland. After postdoctoral fellowships at University of Oxford and McLean Hospital/Harvard Medical School, Poornima Kumar is currently an Assistant Professor of Psychiatry in the Department of Psychiatry at Harvard Medical School and Director of the Computational Psychopathology (COMP) Group at McLean Hospital. Poornima Kumar’s research program focuses on understanding how we make reinforcement learning based decision-making and how this goes awry in individuals with psychopathology specifically, mood disorders. Dr. Kumar uses multimodal assessment (MRI, EEG, behavioral tasks), pharmacological manipulation and computational models to elucidate the neural, behavioral and molecular mechanisms of reinforcement learning.

Triangulating multimodal representations of affective experiences during naturalistic movie viewing

Emotions reflect coordinated, multi-system responses to events and situations relevant to survival and well-being. These responses ​emerge from appraisals of personal meaning that reference one’s goals, memories, internal body states, and beliefs about the world. Dysregulation of emotions is central to many brain and body-related disorders, making it of paramount importance to understand the neurobiological mechanisms that govern emotional experiences. Unfortunately, the field of emotion has struggled to reliably elicit and measure affective experiences, which has limited theoretical developments. One of the main focuses of our laboratory is to use a computational cognitive neuroscience framework to develop models of affective experiences. In this talk, I will present examples of how we can combine naturalistic elicitation of feelings with pattern-based neuroimaging analyses to develop brain-based models of affect. These models appear to generalize across individuals and can aid in revealing the temporal dynamics of individual affective experiences. We hope that this interdisciplinary work will aid in facilitating a more cumulative and extensible science of emotion.

[Bio]

Luke Chang, PhD is an Associate Professor of Psychological and Brain Sciences at Dartmouth College where he directs the Computational Social Affective Neuroscience Laboratory and co-directs the Consortium for Interacting Minds. He completed a BA in psychology at Reed College, an MA in psychology at the New School for Social Research, and a PhD in clinical psychology and cognitive neuroscience at the University of Arizona. In addition, Luke completed his predoctoral clinical internship training in behavioral medicine at the University of California Los Angeles and a postdoctoral fellowship at the University of Colorado Boulder in multivariate neuroimaging techniques. His research is funded by the NSF and NIH and is focused on understanding the neurobiological and computational mechanisms underlying emotions and social interactions. He has been recognized by the Association for Psychological Science with the Janet Taylor Spence Award for Transformative Early Career Contributions and is a strong advocate for improving methods and quantitative training and has developed several opensource software packages, summer training programs, and online books.