Proceedings of the Annual Meeting of the Cognitive Science Society

Virtual agents have quietly entered our life in diverse everydaydomains. Human-Agent-Interaction can evoke any reaction,from complete rejection to great interest. But do humans im-plicitly regard virtual agents as pure machines, or beings on ananthropomorphic level? We asked participants to train an erro-neous virtual agent on a cognitive task and to reward or punishit. The agent showed human-like emotional facial reactions forthe experimental but not for the control group. We expectedparticipants from the experimental group to give less harmfulreinforcement and show more hesitation before punishing. Ad-ditionally, we hypothesised that participants with higher em-pathy show more compassion towards the agent and thereforewould give more positive reinforcement and feel worse whenpunishing. The results indicate that the agent’s expression ofemotionality is not the relevant factor for showing compassiontowards it. Conversely, human empathy seems to be an impor-tant factor causing compassion for virtual agents.