Resolute Choice in interaction: a qualitative experiment

The purpose of this paper is that of extending the model of Resolute Choice (McClennen 1990) to a situation of interaction and comparing its performance with the Sophisticated-subgame perfect equilibrium model in an experiment. A non-cooperative game in which two players with different preference orderings over outcomes move sequentially is adopted as a framework to compare the two models. I consider those combinations of the players' preference structures which generate the different plans and find those game situations where either one or two outcomes Pareto-dominant over Sophisticated Choice exist. Two definitions of Resolute Choice are therefore tested, which allow to discriminate choice between two different Pareto dominant outcomes. In the experiment three games with the same structure but different payoffs are played. The design allows preliminary group discussion among the players about the decisions to be taken, which is taped and transcribed. The results show support for Resolute Choice as Pareto dominance, while the ability of Resolute Choice as Nash bargaining to explain behaviour is quite limited. The subjects' motivations are very useful in interpreting the results. They show that choice for a Pareto dominant outcome is mainly driven by the idea of Pareto optimality itself. Motivations differ slightly according to which strategy is chosen to reach one of the Pareto dominant outcomes. A result to be noted is the relevance of the different payoffs of the games in motivating choice. The method used in the experiment to elicit the subjects' responses is the strategy method. A direct consequence is that the results are all in terms of strategies chosen by subjects. In view of this, an alternative way to look at the experiment results has been tried, which consists in a simulation of the outcomes of the games that would have resulted from direct interaction among the players. The results have then been compared to the ones from the experiment.