Are the moral responses attributed to robots sensitive to their emotional expressiveness? A psychological study with QTrobot

This study examined whether human perception of humanoid robots’ moral responses could be influenced by robots’ perceived capacity to express emotions, and why people usually attributed them more utilitarian choices by assessing moral thinking across perspectives (robots vs. human agents). Two experimental conditions were designed. QTrobot was shown in avideo presenting two classic sacrificial dilemmas (the Trolley and Footbridge dilemmas). In the condition with emotion, QTrobot expressed emotions while presenting the dilemmas, whereas in the condition without emotion it did not. Participants were required to predict the robot’s action in each dilemma and to complete the Oxford Utilitarianism Scale (OUS) from its perspective in order to measure perceived moral thinking. These data were compared to participants’ own responses on the same tasks (i.e. OUS and dilemmas). Participants attributed more utilitarian responses to QTrobot (as measured through dilemmas and the OUS) than to themselves, regardless of condition. They also perceived QTrobot’s moral thinking as more utilitarian, based not only on stronger endorsement of impartiality, but also on stronger tolerance of causing harm to one person to benefit the greater good. These results deepen our understanding of how people perceive the morality of robots compared to humans, and highlight promising avenues for future research.