1 Introduction
When optimising the behavior of a group, one of the difficulties is to solve the conflicts between the personal interest and the group interest. In order to optimize in this context, it is necessary to state social laws, which lead the group to choose a satisfying solution. The goal is to find an order on the preferences for each agent, so that the global behavior of the group will be satisfying. We consider a system made of agents, each one having a set of objectives to reach. Each agent uses a Markov Decision Process (MDP) to find an optimal policy to solve his problem. The outcomes of objectives of an agent can have some effects on the accomplishment of other agent objectives. This problem is directly related to collective intelligence (COIN), which deals with effects of individuals actions on the global welfare, just by using local utilities without defining the global desired behavior [11].