Jan Pohl, Arvid Kappas & Kristina Nikolovska
While the necessity of a concept of “self” for understanding human behavior remains subject to debate, it evidently has significance in everyday life: Lay individuals ascribe selves to humans but also to animals and technical systems, shaping their interactions accordingly. The literature suggests that there are distal behavioral cues eliciting this attribution of selfhood and they may be as minimal as simple movement perceived as causal. In a series of experiments, we aimed to identify which types of behavioral cues, both non-social and social, may increase selfhood-attributions to other agents. We compared behavior of non-humanoid robots suggesting either the presence (C+) or absence (C-) of behavioral cues for one of the characteristics of interest. Results showed a consistent pattern of increased selfhood-attribution towards robots exhibiting any one of the examined minimal characteristics. Furthermore, most perceived sentient characteristics of the robot were triggered by any single characteristic’s cue, showing that people go way beyond the information given when attributing selfhood. We propose a Pars Pro Toto account of selfhood attribution with one model focusing on the internal over-generalization of the characteristics’ representations as they may prime each other, allowing for the activation of characteristics that are missing distal cues. Another model explores the potential semantic overlap between concepts highlighting the possibility that the investigated characteristics might be less organized and defined than we have considered. To explore this account, we are investigating whether selfhood-attribution differs between a (simple) robot exhibiting only a single behavioral cue and a complex robot suggesting the presence of multiple cues. This will provide further insides into participants’ attribution towards non-humanoid agents and stress whether selfhood is attributed gradually with increasing behavioral evidence or binary starting with a single cue
