Abstract:

The notion of “interdependency” is crucial in all sciences, and lies at the core of our highly interconnected modern world. A popular approach to assess the interdependencies of particular systems is to represent them as networks of pairwise interactions, whose properties have been studied with great success in many disciplines. However, these networks miss collective properties of groups of variables that cannot be observed in any subgroup, which are technically known as "statistical synergy.” In this talk we discuss implications of recent advances in the theory of synergistic effects, and how they might change some of our intuitions about interdependency, complexity, and the human brain. Our talk uses the brain and the literature of computational neuroscience as guides in our exploration, and information theory as a preferred toolkit to quantify multivariate statistical properties. We first focus on structural (i.e. synchronous) interdependencies, and show examples of multi-agent systems whose interdependencies can only be seen in the whole but not in any sub-parts. We then provide a critical discussion on the framework of Integrated Information Theory (IIT), which provides a novel way to understand information dynamics. Finally, we explore how these findings provide a quantitative framework to assess causal emergence. We illustrate the presented methods in various scenarios, including cellular automata, Baroque music scores, and the human brain.

Bio:

Fernando Rosas received the B.A. degree in music composition, the B.Sc. degree in mathematics, and the M.S. and Ph.D. degrees in engineering sciences from the Pontificia Universidad Catolica de Chile. He is currently a Research Associate at the Centre for Psychedelic Research at the Department of Medicine, Research Associate at the Data Science Institute, and an Associate Fellow at the Centre for Complexity Science, Imperial College London. Previously, he worked as a postdoctoral researcher at the Department of Electrical Engineering of KU Leuven, Research Fellow at the Department of Electrical Engineering of National Taiwan University, and Marie Curie Research Fellow at the Department of Mathematics and Department of Electronic Engineering at Imperial College London. His research interests lie in the interface between computational neuroscience, complexity science, and information and communication theory.

Fernando Rosas, Research Associate

Department of Brain Sciences  

Imperial College London