Users' Perception of Empathic Expressions by an Advanced Intelligent System

Abstract

The goal of this study was to examine user\textquoteright s perception of expressions of empathy by an autonomous system. In a survey eight different components of empathy identified in literature studies and prior tests (Expressing own feelings, Expressing to know what the other feels, Helping, Showing interest, Taking the others perspective, Displaying regard, Situational understanding, and Agreement) were compared to neutral expressions. Differences in participants evaluations were found across the components of empathy as well as individual differences were revealed. Expressions of cognitive empathy (Showing interest, situational understanding) and expressions of empathy of assistance (helping) were perceived positively by participants. However, expressions of affective empathy (expressing own feelings, expressing to know what the other feels) received mainly negative ratings. Cluster analysis revealed individual differences especially for items relating to affective empathy. Whereas one group of participants identified in the cluster analysis rated expressions of affective empathy negatively, a second group of participants rated these expressions positively. Furthermore, large differences across participants also existed for taking the other's perspective, a component of cognitive empathy. Integrating expressions of empathy in human-machine interaction is a sensitive issue and designers must carefully choose what components of empathy are adequate depending on the situational circumstances and the targeted user group.