Decision as choice of potential intentions

Abstract

Multi-agent systems research has been drawing innovative insights from other disciplines such as philosophy and cognitive science. While cognitive studies have shown that over 95% of human decisions conform to the recognition-primed decision making RPD model, there still lacks formal descriptions of RPD from agent perspective. The objective of this study is to establish a formal connection between agent mental attitudes and the meta-cognition concepts as used in RPD. We adopt the intentional attitudes as introduced in the SharedPlans theory to model the RPD process and its cognitive activities. In particular, situation recognition is framed as the formation of potential intentions, mental simulation and expectancy monitoring are modeled as two processes for intention evolution, and recognition diagnosis is linked to the notion of intention reconsideration. Such a mentalistic modeling of recognition suggests that the RPD process offers an ideal framework that can systematically integrate intention formation, intention evaluation, and intention reconsideration---three key activities that BDI agents typically implement. This study also indicates that agent-based modeling is a viable approach for formally studying meta-cognition concepts and team cognition concepts.