Using argumentation schemes to find motives and intentions of a rational agent

Abstract

Because motives and intentions are internal, and not directly observable by another agent, it has always been a problem to find a pathway of reasoning linking them to externally observable evidence. This paper proposes an argumentation-based method that one can use to support or attack hypotheses about the motives or intentions of an intelligent autonomous agent based on verifiable evidence. The method is based on a dialectical argumentation approach along with a commitment-based theory of mind. It is implemented using the Carneades Argumentation System, which already has 106 programmed schemes available to it and has an argument search tool. The method uses schemes, notably ones for abductive reasoning, argument from action to intention, argument from action to motive, and some new ones.