Altruistic punishment (Fehr/Gächter)

For Ernst Fehr and Simon Gächter, altruistic punishment means that people punish others who shirk a joint contribution – at their own cost, although they themselves gain nothing from it. The benefit goes to others: whoever has been punished takes part more afterwards. In the experiment, anger was the driving force, and without the possibility of punishment, cooperation collapsed.

This entry deals with this one finding and what it says about punishment and cooperation, not with the two researchers’ work as a whole. For the concept itself, see Altruism; for mutuality, Reciprocity.

About the people. Ernst Fehr (born 1956) is an Austrian economist at the University of Zurich and one of the best-known representatives of experimental behavioural economics. Simon Gächter (born 1965), also Austrian, does his research at the University of Nottingham. Their joint paper Altruistic punishment in humans appeared in Nature in 2002.

What did Fehr and Gächter investigate?

240 students in Zurich played a public goods game on screens. In groups of four, each person received money units and could keep them or put them into a joint project. Every unit paid in returned 0.4 units to each group member. For the group, then, it paid off if everyone paid in everything – for the individual, however, it paid to give nothing and profit from the others. In economics, someone who does that is called a free rider.

The set-up was decisive: the groups were reshuffled after every round, nobody met the same person twice, everything remained anonymous. So whoever punished could gain no later advantage from it. In one condition, participants could punish others after each round: one punishment point cost the punisher one unit and the person punished three.

What came out?

Without the possibility of punishment, cooperation collapsed: in the last round 58.9 per cent no longer contributed anything. With the possibility of punishment, contributions rose from round to round. 84.3 per cent of participants punished at least once, 34.3 per cent more than five times – although every punishment cost them money themselves. Mostly those who contributed a lot punished those who contributed little, and all the more harshly the greater the deviation.

Fehr and Gächter see the driving force in negative emotions. In follow-up questions, most stated that they would be very annoyed if someone else contributed hardly anything. They call the punishing altruistic, therefore, not because of a noble motive, but because it costs the punisher and benefits others. They call human cooperation an evolutionary puzzle – their finding shows a mechanism by which it holds, not proof of where it comes from.

What has been learned since?

• Punishing brings satisfaction. In people who can punish unfair behaviour, the reward system in the brain becomes active (de Quervain et al. 2004).
• Moderate punishment supports cooperation – across many experiments, as does reward (Balliet, Mulder & Van Lange 2011). Over many rounds, punishment can also benefit the community as a whole (Gächter, Renner & Sefton 2008). It works particularly well where people already trust one another (Balliet & Van Lange 2013).
• Reward works just as well. Where contributions can be rewarded, groups cooperate similarly well, without the costs of punishment (Rand et al. 2009).
• An excess tips over. Sanctions experienced as self-serving or hostile undermine the willingness to help (Fehr & Rockenbach 2003). When people punish back, feuds arise and cooperation falls (Nikiforakis 2008). In some groups punishment even hits precisely the cooperative (Herrmann, Thöni & Gächter 2008). And those who punish most end up no better off: “Winners don’t punish” (Dreber et al. 2008).
• A punishment can become a price. When day-care centres introduced a fee for late pick-up, more parents came late – the punishment had turned the obligation into a service that could be paid for (Gneezy & Rustichini 2000).
• Those who punish are often considered trustworthy. People who sanction wrongdoing at their own cost are more readily trusted in experiments (Jordan et al. 2016).
• In child-rearing, the disadvantages of physical punishment prevail. Smacking goes along with more behavioural problems in children, not fewer (Gershoff & Grogan-Kaylor 2016). In the administration of justice, the certainty of a punishment deters more strongly than its severity (Nagin 2013).

What do experiments in other cultures show?

That people punish rule violations at their own cost is found in many societies – how strongly and against whom, however, differs markedly (Henrich et al. 2006). A Zurich study in Papua New Guinea showed that uninvolved third parties punish a rule violation more harshly when the victim belongs to their own group, and more mildly when the offender does (Bernhard, Fischbacher & Fehr 2006). Such differences stand side by side, not one above the other.

One limit applies to all of these studies. The anthropologist Bronisław Malinowski called for a community to be studied by participant observation over a long period, in order to understand its rules from within. A game with money at stake and rules from outside, by contrast, brings foreign norms of its own. It shows how people act in this game situation – not how they would live without it, and still less any original state of humanity.

Is altruistic punishment altruism?

This is where our critique begins. Whoever punishes out of anger is following a desire for revenge, even if others benefit from it. A portrait of Fehr says that revenge too is „ein solches uneigennütziges Verhalten“ (“such an unselfish behaviour”; our translation); Fehr himself calls it „häufig nichts anderes als die dunkle Seite des Gerechtigkeitssinns“ (“often nothing other than the dark side of the sense of justice”; our translation). When revenge is declared a virtue because it serves order, coercion becomes a good deed – a thought we trace back to Auguste Comte: he wanted to fit the individual into an order directed from above and, against dissenters, in the end also envisaged force (see Altruism in Comte).

D. Harald Alke puts it more briefly, from experience: “But revenge does not pay.” (Energy for Millions, 1993).

Our reading: too much punishment destroys any cooperation

Punishment can support cooperation – as a last resort, in moderation and experienced as fair. Where it comes from anger, turns out too harsh, hits the uninvolved or becomes a principle, it destroys what it is meant to protect. The question is therefore not “punishment, yes or no?”, but how much, when and from whom. For us the order is clear: first explain, acknowledge and reward, allow amends to be made, show forbearance – and only then, in moderation, sanction.

For us, what is altruistic is not revenge, but an order that needs it as rarely as possible.

What we read differently in 2017

In Selected Perspectives on Altruism (2017, in German) we reproduced the experiment inaccurately in a few places: it was not a computer simulation, but real decisions on screens; without punishment, cooperation did not decline slightly, it collapsed; and the study in Papua New Guinea was not a control experiment to Fehr and Gächter, but a separate, later investigation. Our critique of the idea of interpreting revenge as altruism remains – it now stands on the correct finding.

Related terms

• Altruism – the concept, its origin in Comte and its interpretations
• Altruism in Comte – what Comte meant by the word and where we disagree
• Reciprocity – mutuality and Nowak’s five rules of cooperation
• Collectivism and individualism – when group pressure tips over
• Shared intentionality (Tomasello) – norms, third-party punishment and cooperation
• Fairness expectations (Schmidt/Sommerville) – the expectation of fairness, already in toddlers
• Helping behaviour (Bierhoff) – when people help

More from Kyborg Institute in Google Search — Set as preferred source↗

By Tobias O. R. Alke