This paper explores the frailty of particular notions of ‘actant’ and ‘affect’ for an understanding of the emergent socialities that cross virtual and actual worlds. It uses work on robots and avatars to explore a humanly grounded theory of sociality. It discusses the virtual character of selves and social relations, and how forms of presence apparent in robotics and virtual worlds both enhance and augment our understanding of specifically human forms of sociality. It suggests that critiques of subject-object dualisms do not depend on a rejection of the distinctiveness of anthropos.