Contributions to the modelling of auditory hallucinations, social robotics and multiagent systems

Résumé

The Thesis covers three diverse lines of work that have been tackled with the central endeavor of modeling and understanding the phenomena under consideration. Firstly, the Thesis works on the problem of finding brain connectivity biomarkers of auditory hallucinations, a rather frequent phenomena that can be related some pathologies, but which is also present in healthy population. We apply machine learning techniques to assess the significance of effective brain connections extracted by either dynamical causal modeling or Granger causality. Secondly, the Thesis deals with the usefulness of social robotics strorytelling as a therapeutic tools for children at risk of exclussion. The Thesis reports on the observations gathered in several therapeutic sessions carried out in Spain and Bulgaria, under the supervision of tutors and caregivers. Thirdly, the Thesis deals with the spatio-temporal dynamic modeling of social agents trying to explain the phenomena of opinion survival of the social minorities. The Thesis proposes a eco-social model endowed with spatial mobility of the agents. Such mobility and the spatial perception of the agents are found to be strong mechanisms explaining opinion propagation and survival.