Cooperation Without Genes, Games Or Cognition

In this paper I describe how the spatial dynamics of autocatalytic interactions among entities in a virtual world led to a type of co-operation that typically would be studied from a game theoretical perspective. The entities were very simple and completely identical at the start of the simulation. They just aggregated and performed aggressive interactions in which winning was self-reinforcing. Patterns of reciprocation emerged at the level of the group, particularly in loose assemblages. These patterns appeared not to be due to global spatial structures as I have suggested before (Hemelrijk, 1996ab), but arose from local series of Titfor-Tat like interactions. These involved pairs of individuals that ‚collaborated‘ by taking turns in chasing away a third entity. Runs with varying parameter values showed that particularly those entities that were designed to be more aggressive, were more prone to co-operate. The processes responsible for the Tit-for-Tat like patterns are outlined. It is suggested that similar epiphenomena may contribute to cooperation found in real animals (such as reciprocation of support in primates, communal predator inspection by fish and co-operative hunting by lions). The explanatory value of game theoretical approaches to cooperation in real animals, such as the Prisoner‘s Dilemma is questioned. Firstly, because in that paradigm the ability to co-operate is attributed to a quality of the individual (be it in terms of genes or cognition) and arises from pay-off rationalisation, secondly because co-operation (or its absence) is studied in isolation, separate from other behavioural acts.