Evolution of coordinated punishment to enforce cooperation from an unbiased strategy space

Abstract

The emergence and maintenance of punishment to protect the commons remains an open puzzle in social and biological sciences. Even in societies where pro-social punishing is common, some individuals seek to cheat the system if they see a chance to do so—and public goods are often maintained in spite of cheaters who do not contribute. We present a model accounting for all possible strategies in a public goods game with punishment. While most models of punishment restrict the set of possible behaviours, excluding seemingly paradoxical anti-social strategies from the start, we show that these strategies can play an important role in explaining large-scale cooperation as observed in human societies. We find that coordinated punishment can emerge from individual interactions, but the stability of the associated institutions is limited owing to anti-social and opportunistic behaviour. In particular, coordinated anti-social punishment can undermine cooperation if individuals cannot condition their behaviour on the existence of institutions that punish. Only when we allow for observability and conditional behaviours do anti-social strategies no longer threaten cooperation. This is due to a stable coexistence of a minority supporting pro-social institutions and those who only cooperate if such institutions are in place. This minority of supporters is enough to guarantee substantial cooperation under a wide range of conditions. Our findings resonate with the empirical observation that public goods are resilient to opportunistic cheaters in large groups of unrelated individuals. They also highlight the importance of letting evolution, and not modellers, decide which strategies matter.

Why do inequality and deprivation produce high crime and low trust?

“…Our findings also have implications for the literature on the evolution of cooperation. This has shown that punishment can be an effective mechanism for stabilising cooperation 28 , 29 , but have not considered that the deterrent effects of punishment may be different for different individuals, due to variation in their states. Our findings could be relevant to understanding why some level of exploitation persists in practice even when punishment is deterrent overall.…”

Mediating artificial intelligence developments through negative and positive incentives

“…These results show that, similarly to peer incentives in the context of one-shot social dilemmas (such as the Prisoner’s Dilemma and the Public Goods Game) [ 31 – 40 ], strategies that target development speed in DSAIR can influence the evolutionary dynamics, but interestingly, they produce some very different effects from those of incentives in social dilemmas [ 41 ]. For example, we have shown that strong punishment, even when highly inefficient, can lead to improvement of safety outcome; while punishment in social dilemmas can promote cooperation only when highly cost-efficient.…”

“…Incentives such as punishment and rewards have been shown to provide important mechanisms to promote the emergence of positive behaviour (such as cooperation and fairness) in the context of social dilemmas [ 31 – 40 , 42 , 43 ]. Incentives have also been successfully used for improving real world behaviours such as vaccination [ 44 , 45 ].…”

Cost efficiency of institutional incentives in finite populations

“…Institutional incentives such as punishment and reward provide an effective tool for promoting the evolution of cooperation in social dilemmas. Both theoretical and experimental analysis has been provided [Gürerk et al, 2006, Sasaki et al, 2012, García and Traulsen, 2019, Baldassarri and Grossman, 2011, Dong et al, 2019, Bardhan, 2005, Wu et al, 2014]. However, past research usually ignores the question of how institutions’ overall spending, i.e.…”

“…They include kin and group selection [Hamilton, 1964, Traulsen and Nowak, 2006], direct and indirect reciprocities [Ohtsuki and Iwasa, 2006, Krellner and Han, 2020, Nowak and Sigmund, 2005, Han et al, 2012, Okada, 2020], spatial networks [Santos et al, 2006, Perc et al, 2013, Antonioni and Cardillo, 2017, Peña et al, 2016], reward and punishment [Fehr and Gachter, 2000, Boyd et al, 2003, Sigmund et al, 2001, Herrmann et al, 2008, Hauert et al, 2007a, Boyd et al, 2010], and pre-commitments [Nesse, 2001, Han et al, 2013, Martinez-Vaquero et al, 2017, Han et al, 2016, Sasaki et al, 2015]. Institutional incentives, namely, rewards for cooperation and punishment of wrongdoing, are among the most important ones [Wang et al, 2019, Sigmund et al, 2001, Han and Tran-Thanh, 2018, Sigmund et al, 2010, Vasconcelos et al, 2013, Chen et al, 2015, Wu et al, 2014, García and Traulsen, 2019, Góis et al, 2019, Powers et al, 2018]. Differently from other mechanisms, in order to carry out institutional incentives, it is assumed that there exists an external decision maker (e.g.…”