Humanoid robots are perceived as an evolutionary threat

Abstract

Studying how we perceive humanoid robots will provide insights for a deeper understanding of human-robot interaction. Our ideas about humanoid robots are mainly informed by science fiction, and humanoid robots are generally described as an evolutionary threat in science fiction that has not been tested. The preparedness model has emphasized that the fear module is automatically activated by evolutionary threats, and its underlying neural circuit is centered on the amygdala. We hypothesized that if humanoid robots are perceived as an evolutionary threat even though humanoid robots are manmade, modern objects, we would expect to observe a monocular advantage for humanoid robots and an amygdala response to unconsciously presented humanoid robots that were previously only evident in evolutionary threats. Here, we observed a monocular advantage for the perception of humanoid robots the same as an evolutionary threat (i.e., snakes). Our neuroimaging analysis indicated that unconscious presentation of humanoid robot vs. human images led to significant amygdala activation. Despite a positive humanoid robot-related association had been established by associative learning (as evidenced by results of successfully weakening the negative implicit attitude to humanoid robots and enhancing functional connectivity between the amygdala and hippocampus), the amygdala could still automatically and quickly detect humanoid robots. Our results reveal that processing of information about humanoid robots displays automaticity with regard to recruitment of visual pathway and amygdala activation. Our findings that humans apparently perceive humanoid robots as an evolutionary threat may help inform redefinition of human-robot interaction and robot ethics.