What is the difference between cooperation and reciprocity

Table 1. Points received by the participant and the computer with each card color combination and the number of trials in each condition. The blue-blue color combination resulted in points for the participant and the computer. Any other combination resulted in 20 points for both. The computer was programmed to choose the blue card 12 times and to choose the green card four times at random. In this condition, the participant and the computer always received the same number of points.

The blue-blue combination resulted in points for the participant and 20 points for the computer. The remaining possible combinations resulted in 20 points for the participant and 20 points for the computer. In this condition, the computer was programmed to choose only the blue card. With that, participants were able to receive by choosing the blue card, in which case the computer received 20 points. This computer program was considered cooperative as choices favored participants earns.

As with the condition described above, the participant received points with the blue-blue combination. However, the computer was programmed to choose the green card on all trials. Thus, regardless of the participant's choice, equity in the distribution of points was the only possible outcome. The computer choice did not allow participants to earn points on any trial, thus defined as uncooperative.

The blue-blue combination resulted in 20 points for the participant and points for the computer. The other three possible combinations resulted in 20 points for the participant and 20 points for the computer. The computer was programmed to choose the blue card on every trial.

This condition followed both CI and UI. Participants choice of green card did not allow the computer to receive points, and thus defined aversion to inequity. All participants were exposed to the four conditions described above; however, they were assigned to one of two sets of experimental instructions. The instructions for half of the participants did not mention that points earned during the experimental task would be later exchangeable for money.

The instructions for the other half of the participants had an additional sentence stating that points earned during the experiment would be later exchanged for money. Participants were exposed to a mixed design Johnston and Pennypacker, in which one of the independent variables — cooperative or uncooperative interaction — was manipulated within-subjects, and the other independent variable — payoff nature i. In total, 40 participants met this criterion. The rationale for this design was to observe participants' choices in DI-Test conditions after interactions in which the computer was programmed to be cooperative or uncooperative with the participant.

Figure 3 represents a flowchart of experimental design and conditions order of exposure. Figure 3. Flowchart of experimental design and conditions order of exposure. As we show in Figure 3 , after receiving or not information that points would be exchanged for money left hexagon and rectangle , participants responded to the equity condition center circles.

Thereafter, participants responded to one of the sequences of the conditions rectangle on the right. The experimental design illustrated in Figure 3 was planned to investigate whether 1 a history of cooperative or uncooperative interaction would affect aversion to inequity differently, whether 2 this influence depends on the order in which UI and CI conditions were presented, and whether 3 aversion to disadvantageous inequity is modulated by the information that points would be later exchanged for money.

Our hypotheses were:. Hypothesis 1. Several studies e. On the other hand, the literature on cooperation shows that it is strongly affected by reciprocal interactions Silverstein et al. The main hypothesis of the present study is that participants' choices during a DI-Test will be affected by reciprocal cooperation. Hypothesis 2. A diversity of rewards tangible and non-tangibles has been used in studies investigating the connection between DI and cooperation in non-humans, adults, and children e.

We tested whether or not receiving information that earned points would be exchanged for money affected levels of aversion to inequity in a DI-Test in a between-subjects design see Figure 3. In this hypothesis, we predicted that participants would show higher levels of aversion to inequity when receiving information about the points' monetary value.

During DI-Test conditions, Participants choices of blue cards allowed the computer to earn more points, and choices of green card prevented the computer from earning more points. Thus, the proportion of blue-card choices was used as a measure of DI aversion. A within-subjects design was employed to evaluate participants' proportion of blue-card choices after interacting with a cooperative and uncooperative computer hypothesis 1.

The effects of receiving information about the nature of outcomes money or points was evaluated in a between-subjects design hypothesis 2. The dependent variable is the dichotomous nominal variable of the participants' choice of a blue or green card, with green as the reference category. We used the binomial distribution and logistics link function resulting in a repeated measure binary logistic regression model. The participant identification variable was modeled as a random effect.

The fixed main effects are the card choices by the computer programming during cooperative and uncooperative conditions and the participant's choice of cards in DI-Test conditions.

The fixed interaction effect is composed of the programming variables of the computer card choice and the conditions of the experiment order of interactions and information about monetary outcome.

The present study evaluated participants' choices in the DI-Test conditions after interactions with a cooperative and uncooperative computer. A GLMM omnibus test was used to assess hypothesis 1. This result confirms hypothesis 1 — participants showed higher proportions of reciprocity after cooperative interactions.

Figure 4A depicts the proportion of blue-card choices in the DI-Tests after cooperative and uncooperative interactions separately for the participants who received information about money and those who did not. This figure shows that choices of blue cards were substantially higher in the DI-Test following the CI interaction than the UI interaction. Figure 4. Estimated marginal means of participant's blue cards choices under disadvantageous inequity test conditions as a function of previews interactions with cooperative and uncooperative computer.

A Depicts participants aggregated data for proportion of blue choices, showing that computer choices prior to DI-Test directly affected participants' choices. B Depicts participants proportion of choices of blue card in each DI-Test, showing that prior interactions directly affected participants' choices in DI-Test conditions regardless of the order of exposition. The within-subjects design employed to evaluate hypothesis 1 allowed testing to determine whether the order of presentation of cooperative and uncooperative interactions differentially affected DI aversion.

The post-hoc analysis revealed no significant difference between the proportion of blue-card choices when accounting for order of presentation of cooperative and uncooperative interactions. This post-hoc pairwise analysis revealed reciprocal cooperation after interacting with a cooperative computer and aversion to DI after interacting with an uncooperative computer was not affected by presentation order.

A second post-hoc analysis was carried out to evaluate hypothesis 2 — whether receiving information on the monetary outcomes would affect participants' reciprocity under DI-Test conditions. Thus, the conclusion is that the monetary outcomes within each DI-Test condition had no significant main effect on reciprocity. Figure 5 shows the estimated marginal mean proportion of blue-card choices after cooperative and uncooperative interactions for participants who did and did not receive information about monetary outcome.

Figure 5. Estimated marginal means of participant's blue cards choices under disadvantageous inequity test conditions as a function of the information about monetary outcomes. The results of the present study support the main hypotheses hypothesis 1 : Despite disadvantageous inequity, the participants cooperated under DI-Test conditions in a similar proportion that the computer was programmed to cooperate under prior conditions.

That is, choices in the DI-Test followed a reciprocal cooperation strategy: After interacting with a cooperative computer CI , the participants were also cooperative, and after interacting with an uncooperative computer UI , the participants were also not cooperative.

This reciprocal cooperation strategy was observed regardless of conditions CI and UI order of presentation. As far as we know, this is the first study that clearly shows how reciprocity can alter aversion to disadvantageous inequity in a within-subject design. Our data suggest that the same participant may be or may be not averse to disadvantageous inequity due to recent past cooperative history. These data support the hypothesis that by interacting multiple times, participants have the possibility to learn about their partner's choices and modulate their behavior accordingly in future interactions, as suggested by Baker and Rachlin Thus, a well-controlled experimental history of social interaction that involves different magnitudes of gains i.

Results replicated findings in children e. These data may be consistent with the general proposition of Tomasello et al. The probability of future interactions and their role in cooperation can be seen as a byproduct of experience in which learning about another's intention to be cooperative or uncooperative. We also asked whether information about the value of the outcomes points alone or points exchangeable for money would affect levels of inequity aversion during DI-Test conditions hypothesis 2.

Our data show no significant difference between aversion to inequity and information about experiment outcome. These results contradict studies that have investigated whether the nature of earnings points, money, candy, toys affected participants' choices Harbaugh et al. One possible explanation for the inconsistency across studies is that the influence of the history of successful or unsuccessful cooperation in CI and UI exceeded the possible influence of points vs.

One alternative way to investigate differences between points vs. Results are also consistent with assertions by Brosnan about the relationship between cooperation and inequity aversion. For Brosnan, such strategies can play an important role in long-term cooperative actions, particularly when unrelated individuals benefit from reciprocity and mutualism.

Importantly, our results show that interactions can modulate shifts in aversion to disadvantageous inequity and reciprocal cooperation within the length of an experimental session. Several experimental studies of inequity aversion showed that non-human primates Brosnan and de Waal, , babies Schmidt and Sommerville, , very young children LoBue et al. In the study by Corbit et al. In contrast to the present results, however, they also refused disadvantageous distributions of gains.

Such differences may be attributable to the type of task and specific experimental design. In the present study, we could specify the reciprocal relationship that was established between the participants. In Corbit et al. In the present study, with the same experimental task across conditions, it was possible to evaluate whether the participants' choices were reciprocal to the confederate's choices or not.

Thus, our results highlight the importance of considering reciprocal cooperation and aversion to inequity as complementary phenomena. Results can also be analyzed through a theory of conformity. Although individuals seek to maximize earnings in social interactions, choosing the most used strategy i. We create cliques of four nodes, and exactly one tie to a non-clique member in a regular way Fig. In the regular lattice network, when the UR strategy is present in a large number at the outset, it is able to gain dominance in the population.

The reason for this is that in a regular lattice the UR strategy also channels largely embeddeded and hence reliable information. Contrary to intuitive arguments, increasing the population size does not turn down the success of direct reciprocity in favor of indirect reciprocity Fig. TFT, when present, produces higher levels of cooperation and becomes dominant more often than indirect reciprocity strategies.

In the case of the copy-the-best strategy update, a saturation point for cooperation emerges when increasing population size. Moreover, large populations under this evolutionary rule do not produce cooperation at all if TFT is absent, as UD becomes dominant. The effect of population size on cooperation is largely different when we consider the copy-the-better strategy update. This evolutionary update rule provides more favorable conditions for cooperation than the copy-the-best rule in general 63 , because it keeps conditional cooperation alive, while it does not help the proliferation of overly successful defection strategies who benefit from cheating with multiple partners.

Simulation runs in which TFT is present reach full cooperation with population size over It is interesting to note that cooperation rates are lower for smaller than for larger population sizes.

Hence, the slowlier adoption rule works more efficiently in a larger network where it has more time to spread Figs 3 and S8. The dominance of direct reciprocity depends largely on the assumption that TFT recalls perfectly the previous actions of its partners, so that discriminatory practices can be applied to all partners.

To test the robustness of direct reciprocity success against indirect reciprocity, we relax the assumption of its perfect memory. We assume that agents playing TFT remember the last action of the partner with a given probability.

If they have perfect recall maximum efficiency , then they reciprocate the previous action of all partners. Otherwise, they may forget the previous action of some partners and revert to their basic behaviour.

The domination of direct on indirect reciprocity strongly depends on perfect recall by the TFT rule. As soon as the perfectness is relaxed, the TFT strategy loses dominance in the population and it is substituted in his role by Connected Reciprocity CR. Concerning cooperation, two equilibria emerge: one where cooperation ends up being played half of the times, and one where the whole population defects.

Given the imperfectness of TFT, the direct consequences of Connected Reciprocity for all members of the triad makes this strategy viable. In contrast, Unconnected Reciprocity proves to be inefficient as its good intentions without the enforcement of local social control are easily exploited with defection. Effect of TFT efficiency perfectness of recall on which strategies gain absolute dominance and on the proportion of cooperation. Simulation runs with imperfect recall could clearly be separated into two scenarios.

The Right Panel reports the proportion of simulations that end up in each scenario i. Results are provided for an initialization where the population is equally divided among the types of agents. The rule of strategy update is the copy-the-best strategy. Network density has a non-monotonous effect on the success of Connected Reciprocity and also on the proportion of cooperation in the case of the copy-the-best update rule.

High densities allow the acquisition of more complete information. In these conditions, the best strategies are those that exploit the most partners with defection This is different when the copy-the-better update rule is used, where a larger density implies more cooperation Fig.

The number of closed triangles is larger in a denser network that creates place for a better application and control of the Connected Reciprocity strategy further details in SI Fig. Cooperation increases when forgiveness of this kind is introduced, particularly when TFT is missing from the initial population Fig.

Just like for TFT in the dyad 9 , 10 , 65 , 66 , some forgiveness helps to break the vicious retaliation circles of defection in a triad of CR strategies. High levels of forgiveness, though, are problematic for cooperation, as they imply the neglect of relevant information and constrain effective punishment of a defecting partner.

The level of cooperation is higher in the copy-the-better rule for all levels of forgiveness and it is plateaud at a very high level of cooperation for a larger range of forgiveness values.

In summary, we analyze the relative effectiveness of variants of indirect reciprocity in the presence and in the absence of TFT. Motivated by the sociological debate on the nature of social capital 38 , 41 , 42 , 43 , 50 , 67 , we focus on the relative efficiency of Connected and Unconnected Reciprocity strategies. Connected Reciprocity benefits from social closure and relies on information from those individuals who are also tied to the partner.

Unconnected Reciprocity gains information from anyone and reciprocates the action of the partner towards the source of information.

In this way, our main question origins in the dilemma whether indirect reciprocity is able to operate efficiently due to cohesive aspects of social capital in closed circles or because it utilizes any available information about the partner, also from those who are not direct interaction partners.

This is also the case in larger populations where - according to some arguments - direct reciprocity is supposed to be replaced by indirect reciprocity. The impact of population size itself depends on the evolutionary update rule considered. On one hand, if individuals copy only their best performing neighbors, then large populations sustain a lower proportion of cooperation than small populations. On the other hand, if individuals are satisfied with updating to a neighbor strategy that simply performs better than their own, then cooperation is favored even in larger populations.

The well-known cooperation boosting effect of the copy-the-better update rule is crystallized in larger populations that do not let cooperative strategies disappear suddenly.

In the presence of the latter, indirect reciprocity strategies dominate and establish cooperation. Results show that indirect reciprocity strategies are able to maintain cooperation, but Connected Reciprocity is a better performer. This is a robust result which characterizes the situation when direct reciprocity suffers from imperfect recall, and also the case when direct reciprocity is excluded from the initial set of strategies.

The evolution of cooperation is one of the fundamental problems of human social organization 7 , Simple reciprocal and trigger strategies have shown to be prevalent in this process 3 , 69 , In line with the findings of Roberts 4 , direct reciprocity outperforms indirect reciprocity because it can immediately identify and punish defections of previous partners. The perfect tailoring of reciprocation, however, is constrained by individual memory capacities. Our results show that direct reciprocity in fact loses dominance as imperfect recall is introduced for TFT.

It is important to emphasize that not population size per se, but individual memory constraints are responsible for the decline of direct reciprocity. This potentially implies that cognitive constraints that had to be complemented with communication could have helped hominid groups to achieve impersonal cooperation. When direct reciprocity fails to gain dominance in our model, the relative strength of indirect reciprocity strategies is considered.

The basic idea behind indirect reciprocity strategies is that individuals are able to observe interactions in which they are not directly involved In human societies with the facilities of advanced human communication, and gossip in particular, direct observation is not necessary and the feasibility of indirect reciprocity strategies is further improved 71 , 72 , 73 , There is a quite large extent of ambiguity in the literature about what indirect reciprocity means exactly.

While indirect reciprocity has been defined in various ways, we focus on a fundamental difference in what source of information the strategy accounts for. According to Unconnected Reciprocity, any third-party information could be useful for conditioning behavior against an interaction partner. Due to information flow in open triads, Unconnected Reciprocity might produce global dissemination of behavior more easily.

This is the key aspect of social capital characterized by the presence of structural holes 50 , and weak, far-reaching ties In contrast, Connected Reciprocity only conditions cooperation on the information from mutual partners and has immediate positive externalities. This strategy benefits from social capital that is conceptualized differently: it builds on the reliability and accountability of closed and cohesive microstructures.

We show that Unconnected Reciprocity that benefits from information from indirectly related individuals loses the competition with Connected Reciprocity that builds on the strength of social circles, substantiating the relevance of social capital in closed triads.

In game theoretic terms, local play and information in cohesive micro networks create a correlation device that allows for the clustering of cooperators that establish the success of Connected Reciprocity Closed triads secure the chances of indirect reciprocity by allowing retaliation that has factual consequences for each member of the triad in one or two steps.

In contrast, Unconnected Reciprocity relies on information from individuals who were not and will not be interaction partners. As play never happens between the source and the recipient of information, defection is a better response even to the cooperative act of the partner to the third party. Observation of behavior of the partner with an unrelated third party therefore is insufficient to enforce a circle of cooperation, as the third party could always exploit the benefits of the structural hole Despite the superiority of Connected Reciprocity, cooperation has been achieved also with its Unconnected counterpart, due to the fact that at least the partner in the brokerage position is constrained by indirect punishment from its other contacts playing the UR strategy.

The relative success of Connected Reciprocity benefiting from closed triads can be linked to the empirical observation that social networks tend to be small world structures 76 that provide even more favorable conditions for Connected Reciprocity. Note that the viability of indirect reciprocity strategies in social dilemmas relies on strong assumptions that communication is frequent and factual 14 , 26 , 33 , 37 , 77 , 78 , For indirect reciprocity to work, the reliability of information has to be known with a certain accuracy 12 , 33 , When honesty is not hard-wired, indirect reciprocity cannot prevail: the possibility of deception might nullify all model results The error has been fixed in the paper.

Axelrod, R. The complexity of cooperation: Agent-based models of competition and collaboration Princeton University Press, Riolo, R. Evolution of cooperation without reciprocity.

Nature , — Reciprocity, culture and human cooperation: previous insights and a new cross-cultural experiment. PubMed Google Scholar. Roberts, G. Evolution of direct and indirect reciprocity. Google Scholar. Trivers, R. The evolution of reciprocal altruism. Q Rev Biol 46 , 35—57 The evolution of cooperation. Science 21 , The evolution of cooperation Basic books, Righi, S.

Emotional strategies as catalysts for cooperation in signed networks. Adv Complex Syst 17 , Nowak, M. The research is published in Psychological Science , a journal of the Association for Psychological Science. Previous research had produced evidence in support of both reciprocity and conformity, but Romano and co-author Daniel Balliet of Vrije Universiteit Amsterdam observed that no studies had tested which process would win out if the two were pitted directly against each other.

In the first activity, participants imagined that their spaceship had crashed and had to decide which 15 pieces of equipment to bring with them as they escaped.

They were told that their score would be combined with those of their group members, who were supposedly completing the task at the same time. The purpose of this activity was to foster a sense of group cohesion and belonging among participants.

Then, in a second activity, the participants played a game with their group members and another partner also programmed by the researchers. In each round, a group member and the partner each received tickets and had to decide how many to give to each other. Each ticket given away doubled in value — for the participant, the best outcome would occur if she kept all of her tickets and her partner gave away all of his tickets.

In this case, the participant would have a total of tickets. If both the participant and her partner gave away their tickets, they would each end up with total. How does the neighbourhood hold together? There are seven principles. Because of their wide application, six of them are described in separate entries: Boundaries , Character , Culture , Play Think of it like this: there are forms of interaction with people which have a direct and serious instrumental purpose: I am here to teach you Greek; I would like to order a goulash; please dig that ditch; what have you done with my socks?

And there are all other forms of interaction: the reason we are talking about Barcelona is because we happen to be having sandwich lunch together and we are not strangers: if we were baboons we might do a spot of mutual grooming.

But those will not always be enough to prevent conflict, so the seventh principle is about coping with it. Conflict prevention and resolution has a large and deeply researched literature.

The first is that conflict is a symptom that reveals the strengths or weaknesses of the community—and these in turn are reflections of practically everything it does.

So the whole of Lean Logic , from that point of view, is about conflict resolution, or prevention. The second point is that, in the neighbourhood, as in the primary group, conflict is by no means always a sign that things are falling apart. When conflict does break out, repairs do not always follow quickly, but they can do so. The elaborate reconciliation, grooming and bonding that occurs after conflict in primate groups e.

Indeed, Play Think of it like this: there are forms of interaction with people which have a direct and serious instrumental purpose: I am here to teach you Greek; I would like to order a goulash; please dig that ditch; what have you done with my socks?

There is an echo of this in the account of a retired fish-porter, Mr. Don Ruth, who had spent his working life at the old Billingsgate fish market in London. He spoke of the friendly joke-laden atmosphere that used to exist among porters and, as an illustration, he described the fights that took place—under strict rules, and always in the same place on the first floor, and attended by seconds and others to ensure fair play.

Shortly after a fight was over, the contestants could be seen comparing their bruises and congratulating each other on their tactics—which is as close to baboon-like mutual grooming as human males usually get. Afterwards, the two were likely to remain firm friends, bonded by their ordeal and the reconciliation.

This is the reciprocity in the context of a wider locality or parish which has developed its potential as an effective Lean Economy. It is the scale for substantial initiatives that take the community forward. The reciprocity is not the close collaboration and cooperation of the household and neighbourhood—here there must be a balance between what you give and what you take—but transactions are not impersonal. There is an incentive to maintain the flow of services rendered and received, not least because the health of the parish as a whole depends on its own responsibility and common purpose , and reciprocities are integral to its structure.

There are precedents. The obligations of membership of medieval communities were embedded in personal relationships, kinship and loyalties. Family relationships extended through the locality in a web of interconnectedness. The community assisted its poor with food, goods, stock animals and hospitality. R12 It built and maintained churches and bridges and, within its central ethic of caritas , there was day-to-day care, duty, affection and reciprocal service.

All this was celebrated in festival and carnival—in particular, in the celebrations of the church year. R14 It was a time of not only conspicuous consumption but conspicuous giving, substantial and regular enough for it to take the place of tax. That was demonstrated by the famous case of the Orangi district in Karachi—a place characterised by official inertia and paralysis in planning and building regulations. There was a gross failure in the supply of food, water, shelter and sanitation—typhoid, malaria, diarrhoea, dysentery and scabies were endemic—and no systems of social order prevailed other than those imposed by warring local mafias and drug gangs.

Then, in the s, the community took action. The Orangi example is, justifiably, seen as a model by people who recognise that the authorities cannot cope and that, if they want action, they will have to do it themselves.

And now there are Transition Towns, natural lean thinkers , recognising that the most creative initiatives are local. The parish is at the creative edge between vision and local detail. Here we are moving from bonding to bridging. The scale on which this reciprocity is most comfortable is that neglected and supposedly obsolete social unit—the county.

On this scale, there is no particular obligation or expectation for informal reciprocity.