Toward a cognitive system algebra: Application to facial expression learning and imitation

In this paper, we try to demonstrate the capability of a very simple architecture to learn to recognize and reproduce facial expressions without the innate capability to recognize the facial expressions of others. In the first part, the main properties of an algebra useful to describe architectures devoted to the control of autonomous and embodied "intelligent" systems are described. Next, we propose a very simple architecture and study the conditions for a stable behavior learning. We show the solution relies on the importance of the interactions with another system/agent knowing already a set of emotional expressions. A condition for the learning stability of the proposed architecture is derived. The teacher agent must act as a mirror of the baby agent (and not as a classical teacher). In conclusion, we discuss the limitations of the proposed formalism and encourage people to imagine more powerful theoretical frameworks in order to compare and analyze the different "intelligent" systems that could be developed.