Christoph Jung and Christian Gerber

Abstract Resources as a Modelling Device in the Bounded Optimal Society



The bounded rational agent is nowadays a widely agreed and well understood paradigm in Cognitive Science, especially in Artificial Intelligence (AI). Yet, the generalization of theories and their practical implications, such as of bounded optimality, to societies of agents has not yet been investigated systematically. We address these aspects from the perspective of Distributed Artificial Intelligence (DAI) by developing a hierarchical resource adaption scheme. This scheme links the micro-level of a hybrid and layered agent architecture to the macro-level of the corresponding society. Each of the social and individual-agent decision stages integrates the complementary benefits of simple (one-step) and complex (about a sequence of steps) decision making. Both mechanisms employ the representational device of abstract resources as a unified representation of quantitative environmental and architectural constraints. Besides their practical aspects in optimising agent applications, we propose abstract resources as a modelling tool for social (on the macro-level) as well as the cognitive sciences (the micro-level).