Hauptinhalt

Fachbereichskolloquium: Prof. Dr. Chris Mathys

“Hierarchical Gaussian filters as a window into belief up-dates”

Veranstaltungsdaten

Prof. Dr. Chris Mathys, Aarhus University, Denmark
„Hierarchical Gaussian filters as a window into belief updates”

An agent navigating a changing environment needs to process new information continually in order to make adaptive decisions. To process information opti-mally, the agent needs to update not only the content of its beliefs and the pre-dictions resulting from them, but also its uncertainty about them. Hierarchical Gaussian filters (HGFs) are a class of computational models which has success-fully been used to model human belief updating in various experimental set-tings. In this talk, I will first give a brief overview of the theory behind HGFs. I will then present a series of recent human behavioural and neuroimaging stud-ies which reveal consistent patterns in the signatures of structural elements of belief updating in both healthy and psychiatric patient populations. I will end by illustrating the use of HGFs in the context of active inference, a comprehen-sive behavioural modelling framework.

Der Link für die Teilnahme lautet: https://uni-marburg.de/vMZXf