Artificial Intelligence: the Contribution of Embodied Paradigms

This article has a double objective. The first is to present various proposals concerning an embodied approach to cognition make by the community of computer scientists and robotics. The second aim is to introduce a debate on their contributions and their limits, concerning the highly delicate questions of the construction of meaning, phenomenal consciousness or yet again the relations between mind, matter and organization. The first part of the article draws up a historical reminder of the initial aims of Artificial Intelligence, as well as the various orientations that have subsequently been adopted by this community. The next part positions the debate concerning the fundamental questions that Artificial Intelligence can, or cannot, study in order to reply to the hard questions in cognitive science, and in particular the interest or the limitations related to the use of an embodied approach in order to reply. The third part consists of detailing the embodied approach according to a structure in terms of families, defined by the domains or the various focal points of the neurosciences, psychology or biology. We provide a description of the principles on which each of them rests, and we identify the limits and the possibilities relative to the debate in question. The whole is synthesized by a conclusion which puts the presented research in perspective.