Learning Social Calculus with Genetic Programing

Physical or simulated agents sharing an environment with humans must evaluate the impact of their own and other agents’ actions in the specific social and cultural context. It is desirable that this social calculus aligns itself with the models developed in sociology and psychology however, it needs to be expressed in an operational, algorithmic form, suitable for implementation. While we can develop the framework of social calculus based on psychological theories of human behavior, the actual form of the algorithms can only be acquired from the knowledge of the specific culture. In this paper we consider social calculus based on culturesanctioned social values (CSSMs). A critical component of this model is the set of action-impact functions (AIFs), which describe how the actions of the agents change the CSSMs in specific settings. We describe a technique to evolve the AIFs using genetic programming based on a limited set of data pairs which can be obtained by surveying humans immersed in the specific culture. We describe the proposed model through a scenario involving a group of soldiers and a robot acting on a peacekeeping mission.