Robots are both anthropomorphized and dehumanized when harmed intentionally

Abstract The harm-made mind phenomenon implies that witnessing intentional harm towards agents with ambiguous minds, such as robots, leads to augmented mind perception in these agents. We conducted two replications of previous work on this effect and extended it by testing if robots that detect and simulate emotions elicit a stronger harm-made mind effect than robots that do not. Additionally, we explored if someone is perceived as less prosocial when harming a robot compared to treating it kindly. The harm made mind-effect was replicated: participants attributed a higher capacity to experience pain to the robot when it was harmed, compared to when it was not harmed. We did not find evidence that this effect was influenced by the robot’s ability to detect and simulate emotions. There were significant but conflicting direct and indirect effects of harm on the perception of mind in the robot: while harm had a positive indirect effect on mind perception in the robot through the perceived capacity for pain, the direct effect of harm on mind perception was negative. This suggests that robots are both anthropomorphized and dehumanized when harmed intentionally. Additionally, the results showed that someone is perceived as less prosocial when harming a robot compared to treating it kindly.