CAGE Round Table: Behavioural Game Theory

In this round table discussion we show how recently experimental economics has become a bridge for interdisciplinary collaborations across economics, biology, neuro science, psychology, and other disciplines of the social and natural sciences, and hopefully also in the near future with a link to the humanities. The point of departure is economic modelling through game theory. These abstract situations are then transformed into feasible experiments in which humans act according to a given set of rules in the lab or in the field. The resulting behavior is compared with the rational solutions. In case of divergence between theory and behavior new descriptive bounded rational models are derived. At the same time of playing the games the actors are linked with all possible apparatus, like eye trackers, fMRI, skin conductance etc. to obtain biological data during the decision process. The new data is useful to understand procedural or emotional factors (not) visible in decision making. We will use the Beauty contest game to exemplify these modern developments, some of which have been discussed long time ago in the humanities, like in short stories or philosophy. We conclude with a poem, “The Rational Man as a Tuning Fork”.

Research on how players coordinate in common interest games has generally tested specific theories, such as Level-k or cognitive hierarchy theory, team reasoning, and strong Stackelberg reasoning, all of which predict it. We used protocol analysis, which seeks to determine from the bottom up how players reason and has the potential to discover approaches never imagined by researchers. Our 16 players managed to coordinate 84% of the time in nine dyadic common interest games. Frequently used reasons for strategy choice were maximax and vicarious maximax (assuming the co-player will use maximax, these two often being used together), followed by team reasoning and avoid-the-worst. To a much lesser extent there were instances of Level-1 and Level-2 reasoning, equality-seeking, vicarious avoid-the-worst and relative payoff maximization. The biggest surprise was that many players used multiple reasoning processes, even within the same game.

One starting point for a theory of social interaction is that participants ask: “what would we agree?” and behave accordingly. Explicit communication will be necessary only to the extent that agreement cannot reliably be established. But with no external enforcement, and where people cannot necessarily trust the good will of others, what kinds of agreement can be made? I outline a theory, virtual bargaining (developed with Tigran Melkonyan, Jennifer Misyak & Hossam Zeitoun), which allows a broader set of ‘equilibria’ than Nash and selects between them via a simulated bargaining process. This viewpoint can be viewed a variant of team reasoning, where the team’s preferences are generated by the preferences of its members through the mechanism of bargaining.