gaips_bea image1 image2 image3 image4 image5 gaips_ecute_beach_bar_banner gaips_ecute_train_incorrect_ticket_banner
Persuasive Social Agents using Social Power Dynamics


Abstract Social Power, the potential for social influence, is a pervasive social process in human-human interactions. Despite its acknowledged role in social interaction, little attentions have been paid to this phenomenon in human-agent interaction. One prominent example of social agents that have been of interest these days is the evolution of robotic agents. Further, with respect to robotic agents, limited studies have addressed social power in human-robot interactions. Recent advances on Social Robotics raise the question of whether a social robot can be used as a persuasive agent. To date, different attempts have been performed using several approaches to tackle this research question. In a nutshell, the objective of the present work is empowering intelligent agents with social power dynamics to develop more rational agents and hence more persuasive. On the one hand, proper decision making facing social power increases the rationality of agents and hence enhances their believability. On the other hand, possessing sources of social power boosts their persuasibility. In this text, we report the results of our recent advancements for this objective and draw suggestions for future directions. In sum, we first approached the problem by conceptualizing social power inspired by a theory proposed by French and Raven. Then we performed different user-studies to investigate how different bases of social power contribute to persuasiveness of social robots. In Study 1, within an adversarial setting, we programmed two robots with expert and reward social power. In particular, we programmed one robot to express expertise in its behaviour by giving information to the users. Also, as a specific instance of reward, we used social rewards (telling a joke by the other robot). In the second study, a single robot was used to persuade the user using two different strategies (reward and coercion) comparing to a control condition. Finally, in the last study (Study 3), the robot used one po
Year 2021
Keywords Intelligent Virtual Agents;Social Robotic Companions;
Authors Mojgan Hashemian
Publisher Instituto Superior Técnico
Pdf File \"pdf
BibTex bib icon or see it here down icon

@book { hashemian21, abstract = {Social Power, the potential for social influence, is a pervasive social process in human-human interactions. Despite its acknowledged role in social interaction, little attentions have been paid to this phenomenon in human-agent interaction. One prominent example of social agents that have been of interest these days is the evolution of robotic agents. Further, with respect to robotic agents, limited studies have addressed social power in human-robot interactions. Recent advances on Social Robotics raise the question of whether a social robot can be used as a persuasive agent. To date, different attempts have been performed using several approaches to tackle this research question. In a nutshell, the objective of the present work is empowering intelligent agents with social power dynamics to develop more rational agents and hence more persuasive. On the one hand, proper decision making facing social power increases the rationality of agents and hence enhances their believability. On the other hand, possessing sources of social power boosts their persuasibility. In this text, we report the results of our recent advancements for this objective and draw suggestions for future directions. In sum, we first approached the problem by conceptualizing social power inspired by a theory proposed by French and Raven. Then we performed different user-studies to investigate how different bases of social power contribute to persuasiveness of social robots. In Study 1, within an adversarial setting, we programmed two robots with expert and reward social power. In particular, we programmed one robot to express expertise in its behaviour by giving information to the users. Also, as a specific instance of reward, we used social rewards (telling a joke by the other robot). In the second study, a single robot was used to persuade the user using two different strategies (reward and coercion) comparing to a control condition. Finally, in the last study (Study 3), the robot used one po }, keywords = {Intelligent Virtual Agents;Social Robotic Companions;}, month = {June}, publisher = {Instituto Superior Técnico}, title = {Persuasive Social Agents using Social Power Dynamics}, year = {2021}, author = {Mojgan Hashemian} }

up icon hide this content