Tech Reports


Modeling Arguments in the Dictator Game

Ricardo L. Parreira Duarte


In this project we use an approach to modelling reasoning in a simple scenario from experimental economics, the Dictator game, using preferences over social values to provide transparent justification of actions. This approach does not require estimation of utilities and weights for different factors, instead it will consider how argumentation based on preferences relating to personal values of the subjects can affect the choices made by the dictator, these choices allow a different perspective and they can be used to analyse the game. Using this model we can explain the behaviour of subjects in such experiments, and in particular, gain insight into the framing effect observed by some experimenters.

The Dictator game is a very simple game, or, more an economical experiment. There are two players involved: First is the dictator, who chooses how small sum of money is divided, and the other player just receives what the dictator has left. This player does not have choice but to accept whatever it was given by the dictator.

If only the economic well being of the Dictator was considered, the Dictator would have kept all the money, and there would be no reason for this experiment to exist in economics. But according to some experiments the Dictator often gives money away; this suggests that around 70% of dictators give a non-zero sum away.

This experiment can convince us that the majority of people do not act according to with their economic self-interest, in this way it is possible to believe that other reasons might affect the dictator choices, such as fairness, where the dictator does not appear selfish (i.e, they might care about what the person conducting the experiment thinks of them), etc.

[Full Paper]