If you Betray your Teammates, do you Think you Can be Spotted?

Rizzo, P., Leung, A., Haigh, K., Jemmali, C., & El-Nasr, M. S. If you betray your teammates, do you think you can be spotted?.


Behavior Modeling, Deceptive Communication, Online games


Detecting betrayers and liars in virtual environments is a topic of interest to many organizations and has spurred research for years. Here we address this problem by testing how well emotional Active Indicators (unobtrusive, deliberately introduced stimuli) trigger behaviors that distinguish betrayers. We focus on a theoretical framework about mental states that result from betrayal and that may affect subsequent behavior. To this aim, we developed an online chatbased game where participants are given a choice to betray their team by providing information to an opponent team. We embedded many automatically deployed active indicators in the game. Then we used statistical and machine learning techniques to develop models to discriminate between betrayers (people who chose to betray), non-betrayers (people who chose not to betray), and controls (people who were not given a choice to betray) based on the behavioral responses to stimuli. We also looked at the influence of demographics, personality and other factors on players’ choice to betray and their behaviors. Results show that betrayers engaged in chatting more than other groups, which suggests that they may use deceptive communication strategies analogous to those described in previous work. In addition to discussing results, in this paper we are also presenting the use of games as a method to investigate and deeply examine deceptive behavior in a controllable manner.