Talk by Prof David Rudrauf, UNIGE
11.10.18 - October 15, 2018 - RLC D1 661, 10:00
The Projective Consciousness Model: integrating perception, imagination, emotion and action in a global model of the embodied mind, perspective for strong AI and autonomous robotics
The Projective Consciousness Model (PCM) (Rudrauf et al, 2017) is an attempt to unify psychology computationally, with the broadest possible explanatory power about a multiplicity of phenomena and behaviours, from perception, imagination, appraisal, emotion,social cognition, motivation, and action. The PCM advances previous formulations of active inference by featuring an explicit psychological and cybernetic model of the form, structure and dynamics of conscious experience, formulated as a formal Field of Consciousness, integrating 3D projective geometry and the Free Energy principle (Friston, 2010). The model integrates counterfactual or multi-perspectival first-person perspectives with affective dynamics, for the global optimisation of action outcomes. The PCM offers an explicit, formal, computable and integrative basis for testing hypotheses about normal and pathological psychological mechanisms quantitatively. The principles of the model will be explained, and will be illustrated with applications: to perception, focusing on visual illusions; to artificial agent simulations of resilient navigation in the context of affective stressors based on imaginary projections in the future; to social perspective taking in the context of modelling joint attention and Autism Spectrum Disorders, using both artificial agent simulations and initial implementation in collective robots; to the generation of complex facial expressions as a function of affective states. We will discuss perspectives for strong AI and autonomous robotics.
David Rudrauf is a Psychologist and Neuroscientist and currently an Associate Professor at the University of Geneva, FAPSE, and a member of the Swiss Centre for Affective Sciences and Centre Universitaire d’Informatique. He directs the Laboratory of Multimodal