Speaker: Stefano Panzeri (University Medical Center Hamburg-Eppendorf)
Title: Machine learning for reconstructing, understanding and intervening on neural interactions
Host: Martin Giese (Bernstein Center for Computational Neuroscience, CIN & HIH)
Location: Lecture Hall HNO (Hals-Nasen-Ohren-Klinik), Elfriede-Aulhorn-Str. 5, Tübingen and online. Please contact email@example.com for the zoom login details.
Afterwards, there will be a Get Together at the HNO Lecture Hall with drinks and Butterbrezeln for everybody to enjoy.
Abstract: Interactions between neurons are at the core of almost all brain computations and functions. Yet, quantifying and understanding their specific contribution to behaviors such as perception and decision-making have proven difficult. I will present analytical methods, employing principles of information theory, machine learning and biophysical modelling, which we have developed to address these questions. I will discuss how these methods can be used to measure properties of interaction among neurons, even from signals without cellular resolution such as EEGs. Further, I will discuss their use to quantify the contribution of interactions among neurons to perception and decision-making, and to design theoretically interventions on neural interaction that could prove causally their function or could improve or restore functions.