Abstract: Humans willingly pay a cost to punish defecting partners in experimental games. However, the psychological motives underpinning punishment are unclear. Punishment could stem from the desire to reciprocally harm a cheat (i.e. revenge) which is arguably indicative of a deterrent function. Alternatively, punishment could be motivated by the desire to redress the balance between punisher and cheat. Such a desire for equality might be more indicative of a fitness-leveling function. We used a two player experimental game to disentangle these two possibilities. In this game, one player could choose to steal $0.20 from their partner. Depending on the treatment, players interacting with a stealing partner experienced either advantageous inequality, equal outcomes or disadvantageous inequality. Players could punish stealing partners, but some players had access to effective punishment (1:3 fee to fine) whereas others could only use ineffective punishment (1:1). Players who had access to effective punishment could reduce disadvantageous inequality by tailoring their investment in punishment whereas ineffective punishment did not change the relative payoffs of the individuals in the game but could be used to exact revenge. Players punished regardless of whether stealing created outcome inequality or whether punishment was ineffective at removing payoff differentials, suggesting that punishment was at least partly motivated by the desire to inflict reciprocal harm. However, in the effective punishment condition, players' tendency to punish increased if stealing resulted in disadvantageous inequality and, when possible, punishers tailored their investment in punishment to create equal outcomes. Together these findings suggest that punishment is motivated by both a desire for revenge and a desire for equality. The implications of these findings are discussed.
Alex’s Notes: Have you ever punished another? Why? Punishment is costly to administer, both in terms of executing the punishment itself and in terms of the possibility of provoking retaliation from the target. In fact, the evolutionary rationale for punishment has yet to be determined. Nonetheless, two main hypotheses have been proposed.
- Under the revenge hypothesis, the punisher harms another in response to an intentional harm that was previously done to them by the punished, even if this punishment cannot undue the former insult. The punishment thus serves to deter the punished (or bystanders) from harming the punisher in the future.
- Under the fitness-leveling hypothesis, punishers are motivated primarily by the desire to equalize the playing field, and any deterring function is purely a byproduct.
To investigate the basis for punishment, the study at hand utilized nearly 5,000 American men and women from an online labor market to play a game. In this anonymous game, the participants were evenly split into a role as player 1 (P1) or P2, both of which were given one of five initial endowments. In stage one of the game, P2 could chose to steal $0.20 from P1 or do nothing. Then, in stage two, P1 would be informed of P2’s decision and chose if they wished to punish P2, up to a maximum of four times.
Simple, right? But the manipulations are ingenious. You see, as mentioned earlier, punishments are costly to inflict. Thus, half the P1s could only inflict ineffective punishment whereby the punishment both costs P1 and inflicts on P2 a loss of $0.05 (fee to fine ratio of 1:1). Conversely, the other P1s could inflict an effective punishment that cost $0.05 but reduced P2s endowments by $0.15 (fee to fine ratio of 1:3). Moreover, the starting endowments for P2 were grouped into five categories ranging in amount. The overall set-up is summarized in the table below.
|Treatment||Stage 1 payoff (P1:P2)||P2 stole (yes/no)||Stage 2 payoff (P1:P2)||
(from P1 point of view)
What would you do?
Not surprisingly, P1 was significantly more likely to punish a stealing P2 than a non-stealing P2 in both effective and ineffective treatments. This tendency increased further when the stealing resulted in a disadvantageous inequality for P1, but only when P1 had an effective punishment. Moreover, when punishment was effective and stealing created a disadvantageous inequality, P1 only punished enough to equalize outcomes. Yet, when the stealing resulted in an advantageous inequality, the effective punishers chose to use the harshest punishment available.
Revenge… or equality… or…
Both! As usual, the middle-road makes the most sense and is what this study supports. Depending on the circumstances, people use punishment both as a form of revenge and to equalize the playing field. The former appears to occur more often when the punisher maintains advantage over another who has done him harm, while the latter appears to occur more often when the punisher loses advantage upon another’s insult.
Now while it is currently not clear what the most realistic fee to fine ratio is to use for punishment in laboratory settings in order to mimic the cost to impact ratio of punishment under real-world settings (since the fee-to-fine ratio of punishment is likely to vary with individuals), this study supports the notion that context is key. So next you time you feel like punishing, play a game and find out why.