Constructivism and developmental psychology within AI and cognitive robotics

During the late ‘90s and the beginning of the new millennium the field of artificial intelligence (AI) got some refreshing ideas from developmental psychology for attacking some old problems in the field. Constructivist AI (as it was dubbed although the name doesn’t seem to have taken off) offered a viable alternative to the dominant and well established approaches like symbolic or connectionist AI.

My talk will be organized around an artificial agent architecture called Petitagé (Stojanov 1997, 1998, 2001, 2004; Stojanov et al 2006) and based on the notion of scheme as proposed by Jean Piaget. I will particularly talk about the process of internalization of the environment by the agent, and how the agent can use this “environment model” in achieving goals. I will discuss this process of environment internalization in the context of the classical AI problem of learning environment representation.

After discussing recent extensions of the original architecture, I will conclude the lecture by brief discussion of the state of the art in the constructivist AI and cognitive robotics.

Click here to view the presentation. (in french)

(If you are having trouble viewing, see technical advice)