I in an other’s eye

This paper presents a model of how the fundamental cognitive machinery of self emerges as an accident of sociality, reflecting Buber’s assertion of the primacy of I–Thou relationships. This stands in contrast with the standard ‘I first’ model of theory of mind, which suggests that we understand others’ thought processes by imagining ourselves in their heads. However, this standard model tacitly assumes that understanding oneself is in some way easy, counter to experience in knowledge elicitation, where experts find it hard to reflect on and externalise tacit thought processes. Furthermore, it is hard to create convincing evolutionary accounts for the spontaneous emergence of self. The paper argues that the reflexive understanding of self is both more plausible phylogenically as an evolutionary development and fully consonant ontogenically with research on childhood cognitive development. This reflexive understanding has practical implications for efforts to create artificial agents or robots that are in some sense conscious, and may also inform discussions of the ethical and spiritual implications of advances in artificial intelligence.