Reasoning about self and others: communicating agents in a modal action logic

We propose an approach to reasoning about conversation protocols within the framework of a logic-based agent language. We show how to embed a theory of communicative actions in the framework of a modal logic of action and beliefs, to specify software agents that, situated in a multi-agent environment, can interact with one another by a speech act based communication mechanism. Agents have their own local beliefs on the world and on the other agents mental state. Complex communicative behaviors can be specified as conversation protocols, and agents can reason on the belief dynamics caused by communications, before committing to a given interaction.