Rebecca Padget, Tim Fawcett, Delphine De Moor, Safi Darden

Social animals often form differentiated social relationships with conspecifics. Developing closer partnerships with some than others can facilitate cooperative interactions in which individuals share resources or risk. When choosing a partner, individuals face a decision: a known partner might be sub-optimal if better options are available, but switching partners can be risky if others' behaviour is uncertain. This decision likely depends on how much information individuals have about the behaviour of others in the group compared to their current partner. To investigate the effect of information on cooperative partnership, we developed a Bayesian learning model to identify when individuals should switch partner given the cooperativeness of their partner and knowledge of the population. We then extended this to simulate a population of individuals interacting and switching partners, according to the Bayesian learning model. We found that when individuals had some information about the population, they switched more often than when they did not but that once individuals had a non-zero amount of information, there was little effect of increasing information further. This suggests that populations in which individuals have (or can store) no information about others might have more stable partnerships than populations in which individuals have knowledge of others' behaviour. Our model therefore demonstrates that individuals having more information can lead to social instability that might negatively effect the maintenance of cooperation through reciprocity.