Affective Embodied Agents and Their Effect on Decision Making

Abstract

Embodied agents, such as avatars and social robots, are increasingly incorporating a capacity to enact affective states and recognize the mood of their interlocutor. This influences how users perceive these technologies and how they interact with them. We report on an experiment aimed at assessing perceived empathy and fairness among individuals interacting with avatars and robots when compared to playing against a computer or a fellow human being. Twenty-one individuals were asked to play the ultimatum game, playing the role of a responder against another person, a computer, an avatar and a robot for a total of 32 games (8 per condition). We hypothesize that affective expressions by avatars and robots influence the emotional state of the users, leading them to irrational behavior by rejecting unfair proposals. We monitored galvanic skin response and heart rate of the players in the period when the offer was made by the proposer until the decision was announced by the responder. Our results show that most fair offers were accepted while most unfair offers were rejected. However, participants rejected more very unfair offers made by people and computers than by the avatars or robots.

Can robots recover a service using interactional justice as employees do? A literature review-based assessment

“…Servbots can also appear honest and trustful (IJ dimension: respect ) through human-like features (Cominelli et al 2021 ; Roesler et al 2021 ), cute faces (Pinney et al 2022 ), extroverted voice pitch (Niculescu et al 2013 ), and small social behaviors (Sakurai et al 2020 ). ServBots' capacity to display emotional expressions also increases customers' perception of trustfulness (Acosta-Mitjans et al 2019 ).…”

Unleashing Fairness: How a Group Norm-Aware Agent Shakes Up the Ultimatum Game

“…Several studies have investigated human physiological responses in human-robot ultimatum games. For instance, Mitjans et al examined the behavioral and physiological responses of participants who acted as proposers in an ultimatum game involving a human, a computer, a virtual avatar, and a robot as respondents [38]. They observed a noticeable increase in GSR values a few seconds after proposing an offer in some game settings with a social robot.…”

Humans Forgo Reward to Instill Fairness into AI

“…Our experiments expand on prior work that uses the ultimatum game to investigate how and when people change their perception of fairness. For example, there are many studies that show that humans' valuation of fairness changes when interacting with AI compared to human counterparts (Acosta-Mitjans et al 2019;Di Dio et al 2019;de Melo and Gratch 2015;Nishio et al 2012;Swiderska, Krumhuber, and Kappas 2019;Sandoval et al 2015;Tulk and Wiese 2018). The majority of these studies indicate that humans are more likely to accept unfair offers made by AI compared to human participants (Chen, Zhao, and Lai 2018;Moretti and di Pellegrino 2010;Sanfey et al 2003;van 't Wout et al 2006).…”