No Rage Against the Machine: How Computer Agents Mitigate Human Emotional Processes in Electronic Negotiations

Abstract

With the proliferation of information technology and artificial intelligence in society, human users have started to engage in social interactions with computer agents. In this study, we conducted a laboratory experiment in which neurophysiological measurements were used to investigate the effect of computer agents on the affective processes and behavior of human negotiators. Participants engaged in alternating-offer bargaining over the partition of a pie with either human or computer counterparts and under different levels of urgency to reach an agreement. Overall, our data show that the subjects claimed significantly higher proportions for themselves when they made opening offers to computer agents than when bargaining with human counterparts, regardless of the degree of urgency in the negotiation. However, when the subjects responded to computer-issued offers the picture was more complex. Whereas under high-level urgency, the subjects were more likely to accept offers made by computer agents than by human counterparts, we observed the opposite effect for low-level urgency, where they were less likely to accept the offers of computer agents. In combination, these behavioral patterns lead to the use of computer agents yielding an increase in economic efficiency. Further, the subjects exhibited less emotionally charged behavior when facing computer agents than when facing human counterparts, as the intensity of affective processes was lower and the relationship between arousal and offer acceptance was observable only when the counterparts were human. The results of our study shed light on the potential benefits and intricacies of employing computer agents in electronic negotiations.