Heterophenomenological methods aim to track human first-person experiences from the third-person perspective of empirical observers. We argue in this paper for the importance of including heterophenomenological methods in HRI research in general, and explore a specific version thereof—the analysis of ethical deliberations. Ethical deliberation and decision making is influenced by the presence of other social agents. We report on a study that uses this connection to explore how humans experience the sociality of robotic ‘social’ agents. We proceed from the overarching hypothesis that attention to the cognitive process of perspective taking can reveal what is unique in our experience of the sociality of social interactions with robots. As we illustrate, by using a mixed methods model with qualitative extensions of quantitative behavioural research, we can gain empirical insights on phenomenological accessible (but not salient) processes of human social cognition. A total of 158 participants were randomized to one of three different conditions: facilitation by a human (n = 66), a robot (n = 74), or a computer (n = 18). All participants were then presented with three different imagined actionable scenarios pertaining to ethical decisions. Explicit justifications elicited during the facilitation phase were thematically analysed and quantified. The robot condition produced as many ethical decisions as the human condition, but significantly more justifications of ethical decisions involving mental role switching (perspective-taking). This unexpected result can be explained, we suggest, by a unique aspect of our sociality experiences with robots that we describe using the new theoretical construct of “sociomorphing.”