Reputation-based Strategies for the Evolution of Cooperative Behaviour
Cooperation between strangers can be difficult to explain. Several mechanisms have been shown to sustain cooperation among which one of the most general is Indirect Reciprocity. This describes how reputation-based social norms can distinguish between appropriate and inappropriate behaviours and sustain cooperation through the promise of future reciprocity from other members of a population. We present three experiments that investigate how a social norm’s ability to sustain cooperation is affected when: information flow is restricted to between neighbours, anyone can punish and anyone can be punished, and when people are capable of fine tuning their behaviour in response to their environment.
Using simulations and a series of agent-based models, we find that – in the two-person prisoner’s dilemma – restricting the flow of information and ensuring people learn from their neighbours, benefits the maintenance of good behaviour. In such scenarios, the best chances for cooperation occur when actions are judged harshly, ensuring that a good reputation once lost, is difficult to regain. For social norms to sustain cooperation in collective action problems, similar harshness is required through the ongoing threat of punishment. These situations can be highly cooperative if withdrawal from the social dilemma is possible and such behaviour is not judged to be morally worse than defection. However, if people are not able to punish badly behaving peers, then free-riding runs rampant unless the population considers defection to be worse than withdrawing from the social dilemma. We show that an improvement on this state of affairs, can be obtained when agents are able to fine-tune their behaviour when confronted with various reputational environments. Regardless of how actions are morally viewed, cooperation has a good chance if people can be sufficiently deliberate.
https://discovery.ucl.ac.uk/id/eprint/10168359/1/Reputation-based