Behavioral and Computational Mechanisms of Self-localization and Path Integration in Humans

The aim of this thesis is to study and model how multi-sensory integration determines the body representation in the brain and how it relates to higher level concepts such as ownership or agency. For this, we combine innovative experimental techniques and state-of-the-art statistical modeling work.

The work produced during this thesis can be separated in two parts. The first part involves a high-level (behavioral) account of sensory integration and is supported by experiments specifically designed to probe the corresponding hypothesis (e.g. can certain types of body illusions be explained by normative models of sensory integration? or how sensory integration determines the perceived boundaries of the body?). The second part involves a low level (neuronal) account of sensory integration as it establishes the links between normative models of sensory integration and synaptic plasticity using techniques from statistics and machine learning.

Keywords: Multi-sensory Integration, Body Representation, Bayesian Inference, Generative Models, Free Energy, Spike-timing Dependent Plasticity