Abstract
This paper presents an approach based on cooperative self-organization for artificial systems, in order to tackle openness and dynamics. In this work, cooperation is used as a local criterion enabling parts of the system – the cooperative agents – to reorganize as to autonomously modify their interactions and then the global function. The difficulty in defining cooperation and the means to reestablish a cooperative state within a non cooperative system are underlined and analyzed. Two cooperative agents' behaviors are expounded, at the boundary between altruism and selfishness. This approach is also illustrated by modeling, from a local viewpoint, a classical constraint satisfaction or optimization problems.