Don't Lose Sight of the Forest: Why the Big Picture of Social Intelligence is Essential

This paper argues that “social intelligence,” a hot topic in this community in recent years, is still not properly understood, and that this has important implications for multiagent systems. The term shows up repeatedly in the intelligent agents and distributed artificial intelligence literature over many years, but in most cases the authors refer to just limited sub-components of social intelligence, such as norms, or trust. It is argued here that these (and other) facets of social intelligence are inextricably linked, and that we should only attempt to model them in isolation if we know what the implications of ignoring the other facets will be on model outcomes. We do not necessarily need to implement a “big picture” of social intelligence for every agent system, but having a conceptual model of this big picture will help us to understand the implications of what we do choose to model. This paper sketches a possible conceptual model of social intelligence, and begins to untangle the implications for different facets of social intelligence that are commonly scrutinised by the multi-agent systems community. The community is challenged to go further in developing this model and understanding of how its various facets make up the whole.