Bounded practical social reasoning in the ESB framework

Reasoning about others, as performed by agents in order to coordinate their behaviours with those of others, commonly involves forming and updating beliefs about hidden system properties such as other agents’ mental states. In this paper we introduce the Expectation-Strategy-Behaviour (ESB) framework which provides a generic machinery for such practical social reasoning and can be easily coupled with deliberative, knowledge-based architectures such as BDI. We present a conceptual model of ESB, its formal semantics, and simple initial reasoning algorithms that illustrate how the principles of ESB can be used to implement bounded “social” rationality in multiagent designs. A case study is used to show how ESB substantially simplifies the design of agents that include social reasoning functionality.