Rachael E. Jack Title: Understanding dynamic facial expression communication across cultures

Philippe G. Schyns Title: Dynamic Algorithmic Networks of Visual Categorizations

Rachael E. Jack Abstract: Understanding how humans use facial expressions to communicate social messages has remained a central question for over a century. However, this is empirically challenging due to their complexity, and the use of traditional, theory-driven approaches and Western-centric biases that have restricted understanding. New data-driven methods and advances in information-theoretic analysis tools now alleviate these constraints, giving real traction and delivering novel insights. Here, I will showcase one such approach that can objectively and precisely model dynamic facial expressions within and across cultures and characterize the specific social information they transmit. We show that four, not six, core expressive patterns are cross-cultural, that facial expressions transmit distinct social information over time in an evolving, broad-to-specific hierarchical structure, and can multiplex complex information. Our work challenges longstanding beliefs of cultural universality and forms the basis of a new theoretical framework. Finally, we show direct transference of our dynamic facial expression models to the digital economy including generative signalling models for digital agents.  

Bio: Rachael E. Jack is a Professor of Computational Social Cognition, Director of the FaceSyntax laboratory, and Head of the Centre for Social, Affective & Cognitive Neuroscience (cSCAN) at the University of Glasgow.

Jack’s research has produced significant advances in understanding how facial expressions dynamically communicate social information, particularly within and across cultures, using an interdisciplinary approach that combines social perception, visual psychophysics, data-driven methods, modern 3D computer graphics, and computational modelling. Jack’s work has challenged dominant theories in the field, leading to a new theoretical framework of facial expression communication that now informs the design of socially interactive virtual agents in Artificial Intelligence.

 Jack is Associate Editor at Psychological Science, Journal of Experimental Psychology: General, Journal of Experimental Social Psychology, and Affective Science, on the Editorial Boards of Emotion, Journal of Personality and Social Psychology: Attitudes and Social Cognition, and Behavior Research Methods, and is an ERC Advanced Grant panel member. Jack is also Fellow of the Association for Psychological Science (APS), recipient of the American Psychological Association (APA) New Investigator award, the Social and Affective Neuroscience Society (SANS) Innovation award, the British Psychological Society (BPS) Spearman Medal, Association for Psychological Science (APS) Rising Star award, and International Society for Research on Emotion (ISRE) Young Researcher Spotlight. Jack also serves on the committees/boards for the conferences of the Society for Affective Sciences, IEEE Automatic Face & Gesture Recognition, ACM Intelligent Virtual Agents, and the Vision Sciences Society, is Secretary of the Association for Psychological Science (APS) and Chair of the APS Global Engagement Committee.  

Jack’s team is currently funded by the European Research Council (ERC).  

Philippe G. Schyns Abstract: In cognitive neuroscience, a pivotal remaining challenge is the translation of brain activity into comprehensible information processing. Though sophisticated tools measure brain activity with exceptional spatial and temporal resolution, across varying scales of observation, how do we interpret this activity as a process that computes information? To address this critical question, I will introduce a framework that leverages generative models of visual categories to enable interpretation of the information that brain networks represent and transmit, and the computations that underlie perception. Using this approach, we can extract meaningful insights into information processing from dynamic brain activity and its Deep Neural Network models, thereby pushing the frontiers of brain imaging and computational neuroscience.

Bio: Philippe G. Schyns is Professor of Visual Cognition at University of Glasgow, Dean of Research Technology, former Director and founder of the Institute of Neuroscience and Psychology and the Centre for Cognitive Neuroimaging. After he received his Ph.D. from Brown University in 1992 (supervisors J. Anderson, H. Bulthoff, G. Murphy), he worked at the Centre for Biological and Computational Learning, MIT (with Dr. Poggio) in 1993-1994, the ATR Research Laboratories, in 1994-1996. He joined University of Glasgow in 1995 and was promoted to full professor in 1998. He is a Royal Society Wolfson Fellow and a Fellow of the Royal Society of Edinburgh.

He has co-authored over 125 papers over a three-period career.  Period 1. He made a key discovery that scene-specific information can bypass object information in recognition. This bootstrapped the field of rapid scene categorization in human and computational vision. For this, he invented the “hybrid image” methodology that multiplexes the Low Spatial Frequencies of e.g. a city with the High Spatial Frequencies of e.g. a highway, leading to mutually exclusive perceptions when the same hybrid image is presented for a short vs. a long duration. Period 2. As humans can perform multiple face, object and scene categorization tasks from the same stimulus (e.g. outdoor, city, New York), he proposed a unifying framework (diagnostic recognition) that incorporates the task of the observer to understand the strategic use of visual information. This research led to the development of the “Bubbles” methodology, which samples visual information to ascertain the relevant manifold of the categorization task considered. Further developments led to model the stimulus itself with generative models. Period 3. He studied the representation, communication and computation of task-relevant feature manifolds in dynamic neuroimaging (MEG) using perception and categorization tasks and their Deep Neural Network models. His results and methods have been awarded the Neuroimage Prize, 2008, the Social and Affective Neuroscience Society Innovation Award, 2015 and the Rank Prize Lecture, European Conference on Visual Perception,2019.