Daniele Marinazzo
Speaker of Workshop 4
Will talk about: Information transfer in the brain: insights from a unified approach
Daniele Marinazzo obtained his PhD in Physics from the University of Bari, Italy, in 2007. He is a professor in the department of Data Analysis of the Faculty of Psychology and Pedagogical Sciences of the University of Gent, Belgium. He studies new techniques for the analysis of physiological data rooted in theoretical physics and machine learning, with a particular focus on retrieving brain connectivity from the analysis of neuroimaging data. His experimental work is dedicated to the connection between biophysics and function in the cortex.
Measuring directed interactions in the brain in terms of information transfer is a promising approach, mathematically treatable and amenable to encompass several methods. I will present two results obtained in this framework.I will first show how implementing simple dynamical models on different architectures will reveal the limited capacity of nodes to process the input information. For a given range ofthe parameters, the information flow pattern is characterized by exponential distribution ofthe incoming information and a fat‐tailed distribution of the outgoing information, as asignature of the law of diminishing marginal returns.A similar behavior is observed when dynamical models are implemented on the humanconnectome structural matrix and in EEG recordings. This suggests that overall brain effective connectivity networks may also be considered in thelight of the law of diminishing marginal returns.I will then propose a formal expansion of the transfer entropy to put in evidence irreduciblesets of variables which provide information for the future state of each assigned target.Multiplets characterized by a large contribution to the expansion are associated toinformational circuits present in the system, with an informational character (synergetic orredundant) which can be associated to the sign of the contribution.This approach allows an efficient and reliable reconstruction of directed networks andreveals specific patterns of informative multiplets in different physiological states.
