Nonverbal leakage in robots

摘要: Human communication involves a number of nonverbal cues that are seemingly unintentional, unconscious, and automatic-both in their production perception-and convey rich information on the emotional state intentions an individual. One family such is called "nonverbal leakage." In this paper, we explore whether people can read leakage cues-particularly gaze cues-in humanlike robots make inferences robots' intentions, physical design robot affects these inferences. We designed cue for Geminoid-a highly android-and Robovie-a with stylized, abstract features-that allowed to "leak" what they might have mind. controlled laboratory experiment, asked participants play game guessing either evaluated how affected participants' task performance. found did, fact, lead better performance, from which infer led attributions mental states intentionality. Our results implications design, particularly designing expression intentionality, our understanding respond human social when enacted by robots.