Understanding the Effectiveness of a Conversational Agent with an Alternative Empathy Orientation and Emotional Expressivity in Mitigating Verbal Abuse

The primary objective of this paper is to further deepen our understanding on effective responses of CAs in relation to verbal abuse by examining alternative empathy styles.
We examine whether alternative forms of empathy orientation (other-oriented, self-oriented) and emotional expressivity (no facial expression, fixed facial expression, varied facial expression) of voice-activated virtual assistants have different influences on users’ moral emotions found to reduce aggressive behaviors as well as whether they affect users’ perceptions of the agent’s capability.

We conducted a lab experiment involving ninety-eight participants and ten alternative forms of CAs, each of which implemented a different response style against verbal abuse.
Ninety eight participants were assigned to one of the three emotional expressivity conditions (no-facial expression, fixed-facial expression, varied-facial expression) and interacted with two alternative empathy orientation conditions (other-oriented, self-oriented) of agents.

The study results clearly show that other-oriented empathy response style is the most effective in positively influencing the user’s moral emotions tied to reducing the abusive behaviors as well as the perceptions regarding the agent’s capabilities.
In contrast, the agent’s emotional expressivity implemented through facial expressions had no signicant effect on the study outcomes.
Users valued the agent’s verbal contents and vocal characteristics over the agent’s facial expressions.
Participants showed a desire to interact with the agent that can have its emotion well reflected in various vocal features.