2.4 Why Cooperate?
2.4.1 Reciprocity
The effects of reciprocity have been studied most notably by Fehr and G¨achter in [57] where they propose that reciprocity has powerful implications for economic domains and is an important determinant in the enforcement of social contracts and norms. They define two types of reciprocity in this work: positive reciprocity (discussed in section 2.4.1.2), which is concerned with cooperative reciprocal behaviour and negative reci- procity (discussed in section 2.4.1.1), which is concerned with retaliatory behaviour. The paper goes on to discuss how under certain conditions, specifically in competitive market environments where incomplete contracts may exist, reciprocity dominates as a strategy instead of self-interested, rational behaviour. A number of laboratory experi- ments are considered, the first of which is a public goods experiment where positive and negative reciprocity can exist. To maintain contributions to the public good, players must demonstrate reciprocal behaviour (positive reciprocity). It is more than likely, however, that some participants will free-ride as they will be motivated by self-interest. On observation that another player is free-riding, a player may then free-ride them- selves causing a reduction in the free-rider’s pay-off; this is the game’s form of negative reciprocity.
An extension to this experiment is then proposed whereby participants are informed of how other participants are playing. Participants are then given the opportunity to punish others by placing a fine upon free-riders at a cost to the punisher (important since if it costs nothing to punish then self-interested free-riders would have no qualms in punishing cooperative individuals). Provision of punishment allows reciprocators to induce reciprocal behaviour in free-riders and its effect is dramatic with Fehr and Schmidt in [59] presenting evidence that a minority of reciprocal participants are capable of converting a majority of defectors into cooperators.
2.4.1.1 Negative Reciprocity
The work presented in [57] does not provide an answer as to what motivates reciprocal subjects to reciprocate. An answer regarding negative reciprocity is posited by Fehr and G¨achter in two separate papers, the first of which, [56], presents evidence to support the hypothesis that emotions are guarantors of credible threats. In this work, Fehr and G¨achter set up public goods games with four types of treatments with varying degrees of punishment: the more severe, the more costly it is to invoke:
Partner treatment with punishment: 6 groups of 4 participants play 10 repeated rounds of the game and punishment of free-riders is possible.
Partner treatmentwithout punishment: 6 groups of 4 participants play 10 repeated rounds of the game and punishment of free-riders is not possible.
Stranger treatmentwith punishment: 6 groups of 4 participants play 10 repeated rounds of the game and punishment of free-riders is possible. Participants are anonymous and, after each round, participants are randomly allocated to other groups and play resumes.
Stranger treatment without punishment: 6 groups of 4 participants play 10 re- peated rounds of the game and punishment of free-riders is not possible. Partic- ipants are anonymous and, after each round participants are randomly allocated to other groups and play resumes.
The results obtained from the investigation indicate that punishment has a positive effect upon rates of cooperation and a link between severity of punishment and degree of free-riding observed appears to exist. Fehr and G¨achter attempt to determine what the proximate source of punishment is and propose that emotions play a key role in motivating punishment as other suggestions appear to fall short of explaining the re- sults obtained. They cite Hirshleifer [90] and Frank [62] as being proponents of the idea that free-riding elicits strong negative emotions amongst cooperators and these emo- tions instigate a desire to punish free-riders. To ascertain the emotions experienced by participants when encountered with free-riders, Fehr and G¨achter asked participants to indicate the intensity of negative feelings towards a free-rider or towards themselves if they free-ride. The results show that:
More intense negative emotions are elicited when cooperators contribute more and free-riding occurs.
Less intense negative emotions are elicited when cooperators contribute little and free-riding occurs.
Participants expect more intense negative emotions directed towards themselves if they free-ride whilst others contribute large amounts.
Participants expect less intense negative emotions directed towards themselves if they free-ride whilst others contribute small amounts.
Fehr and G¨achter marry their hypothesis regarding emotion with the results obtained from the public goods experiments in three ways:
“...if negative emotions trigger punishment one would expect that the majority of punishment activities is executed by those who contribute more against those who contribute less. This is the case both in the Stranger- and the Partner-treatment. Between 60 and 70 percent of all punishment activities follow this pattern.”
“...remember that non-strategic punishment increases with the size of the negative deviation from the average. This is exactly what one would expect if negative emotions are the cause of the punishment because negative emotions are the more intense the more the free rider deviates from the others’ average contribution.”
“if negative emotions cause punishment, the fact that most people are well aware that they trigger strong negative emotions [...] in case of free riding renders the punishment threat immediately credible. Therefore, we should detect an immediate impact of the punishment opportunity on contributions at the switch points between the punishment and the no-punishment condition. Remember that this is exactly what we observe. The introduction (elimination) of the punishment opportunity leads to an immediate rise (fall) in contributions.”
Fehr and G¨achter conduct a very similar set of experiments in [58] and the results obtained with regards to both parts of the experiment correlate with the results ob- tained in [57]. However, in [58], the authors specifically mention anger as the emotion that motivates punishment from cooperators. There appears to be a notion of fairness that human beings are emotionally sensitive towards since in both sets of experiments participants punished free-riders more when their contribution fell further away from the average group contribution to the public good. In accordance with this, participants reported that they felt greater anger at free-riders when the average group contribution to the public good was high and less anger when the average was lower. In addition to this, participants in both [58] and [57] expected other group members to be more angry with them if they engaged in free-riding whilst average group contribution to the public good was higher.
2.4.1.2 Positive Reciprocity
Whilst anger may indeed provoke punishment in one-shot and multiple interaction ver- sions of the Prisoner’s Dilemma game, an explanation for why positive reciprocity occurs is required. This issue is tackled by Bartlett and DeSteno in [9] who propose that the emotion of gratitude acts functionally to encourage individuals to repay favours even if such behaviour proves costly in the short-term. In [9], the authors provide evidence
from external sources such as [126] to support the hypothesis that people who experience gratitude frequently are more likely to engage in pro-social behaviours such as provid- ing favours and volunteering time to help others. In their experiments, Bartlett and DeSteno set-up three studies: the first aims to demonstrate that gratitude has a direct effect on helping behaviour that is costly to the individual. In this study, experimental confederates provide the participant with a favour and then ask for help afterwards. The results obtained show that there is indeed a positive correlation between the intensity of gratitude felt and time spent helping others.
The second study aims to provide evidence to counter the argument that those who participated in the first study may have been more grateful and more helpful towards the experimental confederate due to an awareness of the social norms of gratitude (par- ticipants received a favour so as to elicit gratitude from them). To show that gratitude is the driver of pro-social behaviour rather than awareness of pro-social norms, participants from the first study were asked for help by both the experimental confederate (who still performs a favour for the participant) and a complete stranger. The results show that all participants in this study help both the confederate and the complete stranger when asked to. As explained by Bartlett and DeSteno, the reciprocity norm cannot explain the help offered to complete strangers.
The third study aims to provide experimental evidence that the help provided to strangers in the second study was not due to a spill-over effect from the gratitude felt towards the confederate. Therefore, in this study, the experimental confederate still provides a favour to the participant but only strangers request help from the participants. In one version of this study, participants were asked a question designed to draw attention to the confederate’s favour, this version is referred to as the gratitude-source version. The standard version of the study does not draw attention to the confederate’s favour. The results revealed that gratitude was most intense amongst participants in the gratitude- source version of the study but participants in the standard version of the study helped the stranger significantly more. Given these results, Bartlett and DeSteno argue that the hypothesis pertaining to gratitude promoting pro-social behaviour rather than pro- social norms is fortified. It is also argued that, by drawing the attention of participants to the pro-social behaviour of the confederate, then pro-social norms such as “pay it forward” are made salient. Consequently, there should be an increase in pro-social behaviour generally, as proposed by Reno et al. in [156]. The results however show that this is not the case and participants from the gratitude-source version of the study spend significantly less time helping a stranger than those participants from the standard version of the study.
DeSteno et al. extend their argument for gratitude being the champion of pro-social behaviour in [43]. In this investigation, economic games are used instead of requests for assistance since economic games pit potential immediate gains against immediate losses whereas requests for assistance only impact upon the time of individuals; there are usually no immediate, tangible gains resulting from such behaviour. Further to
this, the authors cite Vohs et al. [208] as support for the argument that the presence and consideration of money in such games decreases feelings of interdependence and pro-social helping. Ergo, testing for pro-social behaviour as a result of gratitude in such games can potentially provide strong evidence that this emotion is responsible for cooperation between individuals in human society. DeSteno et al. propose that gratitude increases the likelihood of cooperation even when real money is at stake. The experimental set-up for this paper is practically identical as that previously described in Bartlett and DeSteno’s previous work [9]. Instead of an experiment confederate asking for help however, half the participants play a public goods game against the confederate and half against a stranger. The decisions and actions of each player were made in private and the games were of a one-shot nature. Therefore, the impact of strategic considerations upon potential long-term gains/losses were non-existent in the games i.e. any gratitude displayed was not motivated by impure altruism (helping in anticipation of some long-term pay-off). The results from [43] show that participants who reported feeling gratitude towards the confederate also contributed more on average to both the stranger and the confederate. What is especially important about these results is that they show that acts of cooperation are not brought about by consideration/awareness of pro-social norms since participants cooperated with strangers as much as they did with the known confederate.
2.4.1.3 Upstream Reciprocity
Nowak and Roch in [136] outline how positive upstream reciprocity is motivated by grat- itude. Upstream reciprocity is defined by the authors as behaviour where an individual,
x, helps another individual, y, because x received help from another individual, z. In this work Nowak and Roch are essentially bolstering the findings made in [43] (or lay- ing down their foundations since, chronologically, Nowak and Roch’s paper was written before DeSteno et al.’s) by using random-walks to calculate whether cooperation will survive in a population. In these games, a simulated player has two parameters: p and q; p is the probability that a player will pass on cooperation when it is received from another and q is the probability that a player will spontaneously cooperate. Therefore, each player, S, can be defined as S(p, q) and four strategies/player-types are defined, these are:
S(0,0) - Defectors. Players of this type never spontaneously cooperate, neither do they pass on cooperation received from others.
S(0,1) - Classical cooperators. Players of this type spontaneously cooperate but do not pass on cooperation from others.
S(1,0) - Players of this type never spontaneously cooperate but always pass on cooperation from others.
S(1,1) - Players of this type spontaneously cooperate and always pass on cooper- ation from others.
In order to promote cooperation, players incur a cost,c, to themselves which is less than the benefit, b provided to the recipient of the cooperation therefore b > c. For games that only allow upstream reciprocity Nowak and Roch demonstrate that natural selection always reduces cooperation rates. This is due to the fact that strategies with less cooperation always out-compete highly cooperative strategies and pure defectors always dominate. They therefore conclude that upstream reciprocity alone does not enable the evolution of cooperation within a population.
However, Nowak and Roch note that the outcome changes dramatically if players are allowed to reciprocate directly i.e. if they are allowed to cooperate with a player that has just cooperated with them or if players are allowed to randomly compare their current pay-off to the pay-off of a neighbour and change their strategy if it is found that this neighbour has a higher pay-off values (spatial reciprocity). If direct reciprocity allows cooperation then upstream reciprocity evolves too and, in models that include spatial reciprocity, cooperation evolves much more easily with upstream reciprocity than without. In both types of model, Nowak and Roth report that classical cooperators are always out-competed by strategies whose p > 0 and upstream reciprocity greatly enhances the level of altruism in a population.
Overall, their study proposes that gratitude and other positive emotions may be the key to understanding phenomena such as upstream reciprocity and such emotions may evolve through the process of natural selection.