A Theory of Social Agency for Human-Robot Interaction

Motivated by inconsistent, underspecified, or otherwise problematic theories and usages of social agency in the HRI literature, and leveraging philosophical work on moral agency, we present a theory of social agency wherein a social agent (a thing with social agency) is any agent capable of social action at some level of abstraction. Like previous theorists, we conceptualize agency as determined by the criteria of interactivity, autonomy, and adaptability. We use the concept of face from politeness theory to define social action as any action that threatens or affirms the face of a social patient. With these definitions in mind, we specify and examine the levels of abstraction most relevant to HRI research, compare notions of social agency and the surrounding concepts at each, and suggest new conventions for discussing social agency in our field.