Emergence of responsible sanctions without second order free riders, antisocial punishment or spite

While empirical evidence highlights the importance of punishment for cooperation in collective action, it remains disputed how responsible sanctions targeted predominantly at uncooperative subjects can evolve. Punishment is costly; in order to spread it typically requires local interactions, voluntary participation, or rewards. Moreover, theory and experiments indicate that some subjects abuse sanctioning opportunities by engaging in antisocial punishment (which harms cooperators), spiteful acts (harming everyone) or revenge (as a response to being punished). These arguments have led to the conclusion that punishment is maladaptive. Here, we use evolutionary game theory to show that this conclusion is premature: If interactions are non-anonymous, cooperation and punishment evolve even if initially rare, and sanctions are directed towards non-cooperators only. Thus, our willingness to punish free riders is ultimately a selfish decision rather than an altruistic act; punishment serves as a warning, showing that one is not willing to accept unfair treatments.