Social Identity [Tajfel, 1972] and Self-Categorization [Turner et al., 1987, 1994] theories postulate that one’s identity can both be personal and social. A personal identity is when one relates to others in an interpersonal manner, dependent personality traits and close personal relationships with other, while social identity relates to how a person perceives oneself in relation to others according to their shared features: if a social identity becomes salient, people are more likely to see themselves and others as interchangeable components of a larger social unit rather than unique individuals. This process of social identification often leads to bias in rational decision directing people to cooperate more with members of their in-group when the social group’s identity is salient, even if, from the individual’s perspective, the rational decision would be not to do so.
Accordingly to this theory, the psychological process of social identification constitutes a basis for intragroup cooperation: individuals are inclined to favour other individuals with similar identities (e.g. belonging to the same social groups) or blame others of different social identities for the problems in a group. This can be extended to team commitments: social identity will influence the decision of agents when considering following their own interests or commit (or keep the commitment) to their team goals.
On Project INVITE we believe that in order to improve the social intelligence of the autonomous agents and offer a more believable experience, these aspects should be incorporated into the agent’s identity. The model we propose, DIMA (Dynamic Identity Model for Agents), allows agents to have their identity associated with several social identities and have their behaviour influenced by the one that has more salience according to the social context. The influence on behaviour results on a social bias integrated on the decision to cooperate or defect in a social dilemma situation, such as the ones the INVITE’s research tool can provide.