Decision Making in Artificial Social Conditions

Advances in network science may be highly applicable in wide range of research areas for better understanding human behavior and predicting its outcomes. The article presents results of behavioral economics experiment aimed at investigating decision making strategies and their transformation in iterative Prisoner’s Dilemma (IPD). During the experiment, participants were involved in artificial social interaction and experimentally induced socialization. Data on participants’ behavior and its evolution in IPD were classified according to win-stay, lose-shift, and other strategies. The process of strategy evolution and shifting was traced from the point of reference sample interaction, sequences of actions, and then generalized at the level of neural network. We found that information on a behavioral strategy and a number of cooperative and non-cooperative trials increased the efficiency of the model for predicting individual behavior. Results based on a neural network employed as a classifier for recognizing implicitly expressed strategies in various social conditions were also discussed.