As one of the most fundamental models in modern economic theory, the Nash bargaining solution (Nash 1950) has been developed in the past six decades into a high sophisticated theory with varieties of models, extensions, and applications. Despite mathematical simplicity and elegance, it has been found that such a purely numerical theory of bargaining is difficult to model bargaining among autonomous software agents.
From Artificial Intelligence (AI) point of view, bargaining is a rivalry between intelligent agents. Logical reasoning must play an essential role in a bargaining process. A bargaining theory should be able to model both logical reasoning behind bargaining. This talk gives a brief introduction to a logical theory of bargaining. Instead of abstracting a bargaining problem as a numerical game, we describe bargaining terms and their constraints in logical language. A bargaining problem is then modelled as a vector of sets of ordered bargaining terms in terms of bargainers’ preferences. A solution to a bargaining problem is defined as a logically consistent collection of bargaining terms that maximize individual demand and minimize the imbalance of individual satisfaction. Depending on the representation of bargainers’ preferences, the characterization of a bargaining solution can be in pure logical axioms or consists of a combination of logical and game-theoretic axioms. The logical theory of bargaining seemingly has offered a new methodology for bargaining analysis therefore could be also applied to the disciplines other than computer science, such as economics, social science and political science.
About the speaker: Dongmo Zhang is an associate professor at the University of Western Sydney. His research interests include belief revision, reasoning about action, bargaining theory, e-trading and multiagent systems. He is the leader of jackaroo trading agent team.