Robots that refuse to admit losing – A case study in game playing using self-serving bias in the humanoid robot MARC

Abstract

The research presented in this paper is part of a wider study investigating the role cognitive bias plays in developing long-term companionship between a robot and human. In this paper we discuss how the self-serving cognitive bias can play a role in robot-human interaction. One of the robots used in this study called MARC (See Fig. 1) was given a series of self-serving trait behaviours such as denying own faults for failures, blaming on others and bragging. Such fallible behaviours were compared to the robot’s non-biased friendly behaviours. In the current paper, we present comparisons of two case studies using the selfserving bias and a non-biased algorithm. It is hoped that such humanlike fallible characteristics can help in developing a more natural and believable companionship between Robots and Humans. The results of the current experiments show that the participants initially warmed to the robot with the self-serving traits.