- Email: [email protected]
Cooperation and Sociality
Cooperation and Sociality T. N. Sherratt, Carleton University, Ottawa, ON, Canada D. M. Wilkinson, Liverpool John Moores University, Liverpool, UK ã 2010 Elsevier Ltd. All rights reserved.
Introduction
Kin Selection
One of the fathers of modern sociobiology, E.O. Wilson, defined a society as ‘a group of individuals belonging to the same species and organized in a cooperative manner.’ By this definition, all societies consist of aggregations of individuals, but not all aggregations are societies. Male mosquitoes, for instance, may form swarms but they lack the requisite level of cooperation to be considered social. Bird flocks and wolf packs, on the other hand, are generally considered societies because their members not only form groups, but also cooperate – for example, through alerting one another to the presence of predators, or through hunting in packs. Sometimes, the cooperative acts within societies come at a significant cost to the cooperators themselves, and such behaviors are termed ‘altruistic.’ Worker honeybees, for example, forego their own reproduction to help their queen reproduce and will even die in her defense. Potential benefits of social living include a reduction in the rate of predation, improved foraging efficiency, improved defense, and improved care of offspring. For example, due to increased vigilance, the success rate of goshawk attacks on pigeons tends to decrease with increasing numbers of pigeons in a flock. Likewise by huddling together, emperor penguins help save energy and maintain a constant body temperature, thereby ensuring the successful incubation of their eggs. While it is often easy to see the benefits of group living, it is harder to understand why individuals do not free-ride on those benefits while giving nothing in return. Therefore, to understand how societies function and persist, we must understand the stability of the cooperative relationships that help define them. In his last presidential address to the Royal Society of London in November 2005, Robert May argued: ‘The most important unanswered question in evolutionary biology, and more generally in the social sciences, is how cooperative behavior evolved and can be maintained.’ Here, we review some of the principal solutions to understanding the evolution of cooperation – and hence societies – that have emerged over the past 50 years. Many of the solutions we discuss have also been applied to understanding the origin and stability of other forms of cooperation – including cooperation between cells in multicellular organisms and examples of cooperation between members of different species.
While a form of a gene can spread because it enhances the carriers’ own survival and reproduction, it can also spread in a population because it assists relatives who tend to share alleles identical by descent, even if this occasionally comes at a cost to the bearer’s own reproductive success. Thus, cooperative behavior can frequently spread and be maintained because it favors the survival and reproduction of close relatives. While J.B.S. Haldane, R.A. Fisher, and others toyed with the idea in the 1930s, it was Bill Hamilton (in 1964) who recognized its full importance and who developed a formal quantitative theory, in which he introduced the logic of ‘inclusive fitness.’ This general body of theory is now known under the heading ‘kin selection,’ although strictly speaking this refers to the narrower subset of conditions in which individuals assist other individuals that share the same copy of a gene through their close genealogical relatedness. Even if natural selection favors nonaltruists over altruists within groups of individuals likely to share the altruism trait, the proportion of altruists may nevertheless increase if they do better overall than alternative groups without altruists. Perhaps the easiest way to understand the logic of kin selection is through Hamilton’s rule, which states that a form of a gene that causes an individual to perform an altruistic act will tend to spread so long as rb – c > 0, where b (broadly speaking) is the fitness benefit to the recipient from the altruistic act, c is the fitness cost to the altruist, and r is a measure of relatedness. The rule is shorthand for a full population genetics model, and, strictly speaking, it is only correct if we define the terms in very particular ways. For example, r formally measures how genetically similar two individuals are when compared to two random ones in the population with which the altruist will compete for entry into the next generation (indeed, r can be negative). Through its simplicity, Hamilton’s rule serves to highlight the composite minimum conditions for kin-based cooperation, which are both ecological (mediated through b and c) and genetical (mediated through r). Acts of kin-based cooperation are not confined to complex animals and can, for example, be seen in the social amoeba (‘slime mold’) Dictyostelium discoideum, in which solitary cells start to aggregate under harsh conditions to produce a ‘slug.’ Some cells eventually produce
396
Cooperation and Sociality
the spores that can colonize new areas, but only at the expense of a minority of other cells that collect together to hold the spore-producing cells aloft. The cells involved in producing the colony are frequently (but not always) genetically identical, so this extreme altruism can potentially be understood in terms of kin selection; rather like individual cells in a multicellular organism, here helping others reproduce is tantamount to helping yourself. The colonies maintained by wasps, ants, and bees represent the classical examples of highly cooperative social groups. Here, sterile masses of individuals work by gathering food, cleaning the nest, and repelling predators, all in the service of a small reproductive minority. In a series of seminal papers, Hamilton proposed that their unusual genetics (specifically, their haplodiploid sex determination, in which fertilized eggs develop into females while unfertilized eggs become males) might help to explain the prevalence of cooperation in these groups. One implication of the genetics is that females are more related to their full sisters (relatedness 0.75) than they would be to their own offspring (0.5). Given this asymmetry, it is easy to see how sisters might be selected to forego their own reproduction if such behavior can help queens produce more sisters. Unfortunately, while haplodiploidy may well help to explain the evolution and maintenance of cooperation in the highly social (‘eusocial’) insects, it cannot provide the complete solution. Once one factors in the 0.25 relatedness of sisters to brothers, the overall relatedness among siblings in haplodiploids is not especially high. Multiple queens and multiple matings with different males further act to reduce the average relatedness of offspring within a colony. While the manipulation of sex ratio and the ability of females to preferentially favor their sisters may each play a role in enhancing relatedness between the donor and the recipient of cooperation, other factors that also help tip the balance include punishment of workers that attempt to lay their own eggs. Thus, even if high relatedness does not provide the whole explanation, then it may help level the playing field considerably, making cooperative behaviors more likely. Another good example of kin selection is seen in the formation of particular types of microbial mat. Laboratory populations of the bacterium Pseudomonas fluorescens rapidly diversify when maintained in unshaken broths, and a particular form – known as the ‘wrinkly spreader’ (produced by single mutations with large effects) – tends to build up at the liquid–air interface, creating a surface scum. The ability to live at the boundary layer allows the wrinkly spreader bacteria to avoid the oxygen-deprived conditions deeper in the water column, but it comes at some cost. To form and maintain the mat, the wrinkly spreaders have to make a cellulose polymer (glue), a metabolic cost that is not borne by other forms of the bacterium. Despite this cost, the mat can initially develop by kin selection (individuals helping to bind to the surface
397
share the same trait for making glue), but it undergoes periodic collapses as nonglue-making ‘cheats’ invade. The stability of some forms of cooperation may depend on both kin selection and other more direct forms of return. Providing ‘parental’ support to young that are not your own is relatively common in vertebrates. For example, in populations of birds such as Florida scrub jays and long-tailed tits, and mammals such as meerkats (suricates) and brown hyenas, there are nonbreeders that help raise young produced by dominant breeders. Why do not nonbreeders go it alone, rather than help look after another’s offspring? Kin selection may play an important role – indeed, groups in cooperatively breeding species are typically made up of extended families, so that subordinates often help their relatives. Moreover, studies have shown that helpers sometimes provide their closer kin with preferential care. However, direct fitness benefits may also be important in maintaining cooperation in this type of system and may be even more important than kinship itself. In some cases, helpers may be forced into helping behavior to avoid punishment, but by helping they may also sometimes increase their chance of inheriting the territory of the breeding pair (they are ‘paying the rent’). Likewise, the increased survival chances from grouping together may sometimes outweigh the costs of helping. In meerkats, for example, the foraging success and survival of all group members increases with the size of the group (an example of the ‘Allee effect’). It is these and other nonkin routes to cooperation that we now consider.
Reciprocal Altruism Kin selection may help to explain many cases in which individuals incur costs that benefit others, but as we have already seen, some of these examples involve additional phenomena that further maintain cooperation. At an extreme, how do we explain examples of cooperation among nonrelatives? Included in these examples are food sharing among unrelated crows and impala that groom unrelated individuals within the herd. Perhaps by helping others, the donor might subsequently be helped by the receiver when its own need arises? Evolutionary biologist Robert Trivers presented just such an explanation in the early 1970s, referring to the phenomenon as ‘reciprocal altruism.’ To understand how cooperation might be maintained under these circumstances, mathematical modelers have spent a great deal of time and effort elucidating the types of strategy that would do well in a simple game, known as the two-person iterated Prisoner’s Dilemma. In each round (‘iteration’), two players decide simultaneously whether to ‘cooperate’ (C) or ‘defect’ (D), just as two prisoners accused of a joint crime may decide to cooperate with each other by staying quiet under interrogation, or defect on their criminal partnership by talking to the police in exchange for a lighter sentence. In the Prisoner’s
398
Cooperation and Sociality
Dilemma (see Table 1), mutual cooperation (‘CC’) pays more to both players than mutual defection (‘DD’), but defecting while your partner cooperates (‘DC’) pays the defector most of all (reflecting a ‘temptation’ to defect) and a sole cooperator least of all (‘CD,’ the ‘suckers payoff ’). Many researchers consider the Prisoner’s Dilemma the key metaphor for understanding cooperation, primarily because it captures the temptation to defect, but also because it can reflect some of the damaging effects of pure self-interest. However, in a one-off game, the most rewarding strategy is always to defect because whether your partner cooperates or defects, your best option is to defect. So, how can cooperation ever evolve? One solution to the problem arises if the players have a chance of meeting again. In an iterated two-player game (played repeatedly, where the number of rounds is not known in advance by the players), the set of potential strategies is enormous – especially if long memories of previous interactions are allowed. One such strategy called ‘tit-for-tat’ (TFT – cooperate on first move, thereafter follow the partner’s previous move) has been widely recognized as a successful strategy in these iterated games. TFT is thought to be particularly effective because it is ‘nice’ (in that its starting move is to cooperate), retaliatory (in that it follows a defection from the partner with defection), and forgiving (in that it subsequently matches any cooperative act with cooperation). However, TFT is not without its weaknesses: for example, if mistakes are occasionally made, two tit-for-tatters can get stuck into indefinite rounds of defection. As an alternative, the win–stay, lose–shift strategy (WSLS – keep to your strategy if your previous exchange was high paying (DC or CC) but otherwise change) can correct occasional mistakes, although it is still open to drift in a population of cooperators. While such analyses help explain why certain types of cooperative behavior are more successful than others, they should be viewed as providing aids to thinking – not specific quantitative predictions about behavior in the real world. Can direct reciprocation explain examples of cooperation that we see around us? There do appear to be some good examples of reciprocation maintaining altruism, but Table 1 An example of the payoffs involved in a two-player Prisoner’s Dilemma Player A decision
Player B decision
Payoff to player A
Cooperate
Defect
Cooperate Defect
3 (R) 0 (S)
5 (T) 1 (P)
Participants must simultaneously decide whether to cooperate or defect. The defining inequalities of the dilemma require that T > R > P > S (and, for technical reasons, R > [S þ T ]/2), such that the temptation ( T ) to defect exceeds the reward (R) from cooperation, which in turn exceeds mutual punishment (P). The sucker’s payoff (S) the least that can be expected.
not many. The classical example is reciprocal blood sharing in vampire bats, in which females regurgitate blood meals to roost mates who have failed to obtain food in their recent past. While nest mates are often related, there appears to be more structure to the interaction. In particular, experimentally starved bats who received blood, subsequently gave blood to the former donors more often than one would expect by chance. Likewise, in laboratory experiments, cotton-top tamarin monkeys gave more food to a trained conspecific who regularly offered them food in the past, compared to an individual who never gave them food. Studies of grooming have also produced some clear cases of reciprocal altruism. For example, on the African savannah, impala frequently approach one another and begin grooming. Like the vampire bat example, the benefit, in this case removing parasites, may be high, but the costs of grooming in terms of time, fluids, and energy may be relatively low. Here, individuals deliver grooming in bouts (‘parcels’ of 6–12 licks), and the number of bouts received and delivered is remarkably well matched: in this case, defection involves simply walking away or doing nothing. While the relationship is based on reciprocation, it seems very likely that parceling up the cooperative acts in this way helps reduce the temptation to defect. Business deals often show a similar structure to avoid exploitation – half paid in advance and the other half paid when the job is complete. Male red-winged blackbirds in North America also appear to cooperate, sometimes coming to the aid of neighboring males in defending their nests and territories from potential predators such as American crows. One possibility is that the helpers are in fact the true fathers of some offspring on the neighboring territory and are selected to help out of sheer self-interest, that is, simple parental care. Alternatively, or in addition, the helper may benefit directly by removing any potential predator from the neighborhood (‘not in my back yard’), and any benefit to the neighbor is incidental (a by-product ‘mutualism’). R. Olendorf and colleagues recently put these and other explanations for cooperation to the test and ruled out any kin-based explanations on the basis of genetic analyses. However, they also looked for evidence of reciprocity by examining patterns of nest defense against a stuffed crow and simulating cheating by making it appear that a neighbor was not helping with the defense (a ‘defection’). As anticipated, male blackbirds tended to decrease their defense against a potential nest predator after their neighbor appeared to defect in the earlier trial, suggesting that reciprocation was having an important role in maintaining cooperation – ‘‘I’ll help mob your predators, if you help mob mine.’’
Indirect Reciprocity By its very nature, direct reciprocity requires repeated dealings among the same sets of individuals, so it cannot
Cooperation and Sociality
apply to cases of helping strangers we might never see again. However, what if others were looking on? Perhaps by helping others one might gain sufficient reputation as a ‘nice’ individual that strangers would be willing to help you when your own need arose. So, instead of ‘You scratch my back, and I’ll scratch yours,’ one could consider another, seemingly even more vulnerable, guiding principle ‘You scratch my back and I’ll scratch someone else’s.’ This is called the ‘indirect reciprocity’ route to cooperation. Although it may at first seem strange, bear in mind that what matters is that acts of cooperation are returned, not who returns them. Examples of the importance of maintaining an untarnished reputation are widespread in human societies. For example, eBay in part relies on reputation to maintain honest transactions when it provides scores of partner satisfaction. Being a good person or good company to deal with, does not in itself explain cooperation, but it begins to suggest a role for reputation in partner choice. Building on earlier arguments by Richard Alexander, mathematical modelers have demonstrated the theoretical plausibility of cooperation via indirect reciprocity by showing that behavioral rules can evolve in which individuals are more prepared to help strangers if these strangers have a reputation for cooperating. Of course, since reputable individuals tend to provide assistance to similar reputable individuals, kin selection may also play a key role here. Can indirect reciprocity explain cooperation in humans? After all, humans frequently help others who may never have an opportunity to reciprocate, and it is possible that such acts of kindness are recognized and rewarded by others. Staged laboratory games support this view. For example, in a recent experiment, human subjects were repeatedly given the opportunity to give money to others, having been informed that they would never knowingly meet the same person with reversed roles (all donations were anonymous). Despite the anonymity, the personal histories of giving and not-giving were displayed for participants to see at each interaction. As one might expect, the authors found that donations were significantly more frequent to those receivers who had been generous to others in earlier interactions. There are far fewer examples of indirect reciprocity in non-humans and they mainly include examples of cooperation between species rather than within species. One recent example comes from work on cleaner fish mutualisms. The cleaner fish Labroides dimidiatus remove skin parasites from their fish clients, but there is an apparent temptation for them to take a little more at the expense of the client by feeding on their mucus. Clients are faced with the challenge of getting cleaners to feed against their preferences if they are to come away unscathed. Field observations indicate that client fish almost always invite a potential cleaner to draw closer
399
and inspect them if they have had the opportunity to see that the cleaner’s previous interaction ended without conflict. By contrast, clients invite particular cleaners far less frequently if they observed that the last interaction of the cleaner ended with conflict, such as being chased away. So, a good reputation is good for the cleaner’s business, and it may be an important way in which clients avoid ‘defections.’
Strong Reciprocity Everyday observation suggests that there is a sense of ‘fair play’ in human societies. This has been backed up by more formal, if abstract, experiments. For example, in the wellknown ‘ultimatum’ game in economics, two players A and B have to agree on how a monetary reward has to be shared. Player A (the proposer) has one chance to suggest how the money is to be shared (e.g., 60% to player A, 40% to player B), but player B (the responder) can accept or reject the proposed division. If the bid is rejected, then both receive nothing, but if the bid is accepted then the proposal is implemented. The logical optimum is to offer the responder an almost negligible amount (1 cent, say, because 1 cent is better than nothing). However, this logical response may be grossly inadequate, because a common result in this type of game is that responders tend to reject proposals if the offer is anywhere less than about 25% (even when the sum is quite considerable). Other primates may also exhibit what we might think of as a sense of justice. In an intriguing study entitled ‘Monkeys reject unequal pay,’ Sarah Brosnan and Frans de Waal investigated what happens when capuchin monkeys previously trained to exchange a pebble for a piece of cucumber started to see others being rewarded with a more favored food (a grape) – the monkeys tended to go on strike, refusing to exchange a pebble for cucumber even though the alternative was no reward at all. Perhaps punishment can play a role in maintaining fair play and hence cooperation? Some may not see it as cooperation at all, bordering more on enforced slavery than on acts of ‘kindness.’ Nevertheless, when we see apparent examples of altruism we need to ask whether the threat of punishment is helping to maintain it. In a series of staged repeated games among human volunteers, Ernst Fehr and Simon Ga¨chter showed that students are prepared to take on costs in order to punish those who had earlier shirked their opportunity to contribute to a public good. According to Fehr and Ga¨chter, this altruistic punishment is simply a consequence of a ‘negative’ emotional reaction to the sight of somebody free-riding (although this proximate negative reaction may ultimately have evolved for other reasons). As one might expect, those that were punished for not contributing learned their lesson and cooperated more in subsequent rounds.
400
Cooperation and Sociality
Moreover, those games that prevented altruistic punishment altogether saw a marked reduction in the mean amount of cooperation over time. The behavioral tendency to cooperate for the public good yet punish noncooperators has been called ‘strong reciprocity.’ Strong reciprocity may explain many examples of human-based cooperation, but it is difficult to understand how altruistic punishment might evolve as a consequence of natural selection. After all, if altruistic punishment is costly, then an individual that free-rides and lets others do the policing would tend to leave more offspring. The temptation to sit back and let others punish defectors has been termed a ‘second-order defection’ or ‘twofold tragedy.’ So, if strong reciprocity can explain cooperation, perhaps it has only replaced the problem with another one further down the line – why should you be the one to punish? Kin selection may provide one potential solution to this question, but note that kinship can reduce the underlying incentives to defect in the first place. For example, in many eusocial Hymenoptera, worker-laid eggs are killed by the queen and other workers. In a comparative analysis, Tom Wenseleers and Francis Ratnieks found that fewer workers reproduced when the effectiveness of policing worker-laid eggs was higher, indicating that these sanctions were an effective deterrent. However, higher relatedness among colony workers led to less policing, not more, a result which is consistent with the view that less policing is needed when workers are highly related. So, self-restraint based on kin selection can achieve for free what expensive policing could bring about.
Escaping from Prison All adaptive explanations of altruism involve taking the ‘altruism out of altruism,’ either by showing how the actions can benefit other individuals carrying the same traits, or by showing how the nature of the interaction is such that it is in the ultimate interests of the altruist to cooperate. However, this commonality should not be taken to mean that all cooperation can be related back to the two-player Prisoner’s Dilemma (or n-player version of it which can give rise to the ‘tragedy of the commons’). One example of cooperation which is almost certainly not represented by a Prisoner’s Dilemma comes from recent work on a species of fiddler crab on the northern coastlines of Australia, where males aggressively defend their burrows from other wandering males (intruders). Patricia Backwell and Michael Jennions found that male fiddler crabs may sometimes leave their own territories to help neighbors defend their territories against these floating intruders. Why be a good Samaritan? It turns out that reciprocity cannot explain it because the ally that came to the neighbor’s assistance was always bigger than the neighbor itself. Here, it may directly benefit a resident to help its neighbor to defend a territory, so that
it can avoid having to renegotiate the boundaries with a new and potentially stronger individual. In this way, there is no temptation to cheat – large allies are helping themselves, and it is only incidental that helping the neighbor keep its territory is part and parcel of maintaining status quo. Our interpretation of cooperation gets tested further when we observe that some individuals may actually pay a cost to acquire or enhance the by-product benefits produced by another (a phenomenon known as ‘pseudoreciprocity’). Many lycaenid caterpillars, for example, produce sugary secretions that are consumed by ants and in turn the ants protect these individuals from predation. The sugar may be viewed as an investment, yet the protection arises as a consequence of general territorial ant defense. Likewise, many flowers attract pollinators using nectar. One might wonder why flowers do not save themselves the trouble and produce less nectar. Flowers are essentially in a ‘biological market,’ however, governed by simple laws of supply and demand, such that any flower that offers less than conspecifics may experience reduced pollination. So, while kin selection may be at the heart of much intraspecific cooperation, sometimes cooperation can be maintained by a complex interplay of several types of interaction including direct self-interest, reciprocity, reputation, partner choice, and the threat of punishment.
Acknowledgment This study is based in part on a longer review in Sherratt and Wilkinson (2009) which provides full references to many of the studies described here. We thank our editor Joan Herbers for her constructive comments and helpful advice. See also: Kin Selection and Relatedness.
Further Reading Axelrod R (1984) The Evolution of Cooperation. London: Basic Books. Axelrod R and Hamilton WD (1981) The evolution of cooperation. Science 211: 1390–1391. Connor RC (1995) Altruism among non-relatives-alternatives to the Prisoner’s Dilemma. Trends in Ecology and Evolution 10: 84–86. Doebeli M and Hauert C (2005) Models of cooperation based on the Prisoner’s Dilemma and the Snowdrift game. Ecology Letters 8: 748–766. Dugatkin LA (1997) Cooperation Among Animals: An Evolutionary Perspective. Oxford: Oxford University Press. Fehr E and Ga¨chter S (2002) Altruistic punishment in humans. Nature 415: 137–140. Hamilton WD (1964) Genetical evolution of social behaviour I. Journal of Theoretical Biology 7: 1–16. Krause J and Ruxton GD (2002) Living in Groups. Oxford: Oxford University Press. Noe¨ R and Hammerstein P (1995) Biological markets. Trends in Ecology and Evolution 10: 336–339. Nowak MA (2006) Five rules for the evolution of cooperation. Science 314: 1560–1563.
Cooperation and Sociality Nowak MA and Sigmund K (2005) Evolution of indirect reciprocity. Nature 437: 1291–1298. Sherratt TN and Wilkinson DM (2009) Big Questions in Ecology and Evolution. Oxford: Oxford University Press. Trivers RL (1971) Evolution of reciprocal altruism. Quarterly Review of Biology 46: 35–57.
401
West SA, Griffin AS, and Gardner A (2007) Social semantics: Altruism, cooperation, mutualism, strong reciprocity and group selection. Journal of Evolutionary Biology 20: 415–432. West SA, Griffin AS, Gardner A, and Diggle SP (2006) Social evolution theory for micro-organisms. Nature Reviews Microbiology 4: 597–607.
Introduction
Kin Selection
One of the fathers of modern sociobiology, E.O. Wilson, defined a society as ‘a group of individuals belonging to the same species and organized in a cooperative manner.’ By this definition, all societies consist of aggregations of individuals, but not all aggregations are societies. Male mosquitoes, for instance, may form swarms but they lack the requisite level of cooperation to be considered social. Bird flocks and wolf packs, on the other hand, are generally considered societies because their members not only form groups, but also cooperate – for example, through alerting one another to the presence of predators, or through hunting in packs. Sometimes, the cooperative acts within societies come at a significant cost to the cooperators themselves, and such behaviors are termed ‘altruistic.’ Worker honeybees, for example, forego their own reproduction to help their queen reproduce and will even die in her defense. Potential benefits of social living include a reduction in the rate of predation, improved foraging efficiency, improved defense, and improved care of offspring. For example, due to increased vigilance, the success rate of goshawk attacks on pigeons tends to decrease with increasing numbers of pigeons in a flock. Likewise by huddling together, emperor penguins help save energy and maintain a constant body temperature, thereby ensuring the successful incubation of their eggs. While it is often easy to see the benefits of group living, it is harder to understand why individuals do not free-ride on those benefits while giving nothing in return. Therefore, to understand how societies function and persist, we must understand the stability of the cooperative relationships that help define them. In his last presidential address to the Royal Society of London in November 2005, Robert May argued: ‘The most important unanswered question in evolutionary biology, and more generally in the social sciences, is how cooperative behavior evolved and can be maintained.’ Here, we review some of the principal solutions to understanding the evolution of cooperation – and hence societies – that have emerged over the past 50 years. Many of the solutions we discuss have also been applied to understanding the origin and stability of other forms of cooperation – including cooperation between cells in multicellular organisms and examples of cooperation between members of different species.
While a form of a gene can spread because it enhances the carriers’ own survival and reproduction, it can also spread in a population because it assists relatives who tend to share alleles identical by descent, even if this occasionally comes at a cost to the bearer’s own reproductive success. Thus, cooperative behavior can frequently spread and be maintained because it favors the survival and reproduction of close relatives. While J.B.S. Haldane, R.A. Fisher, and others toyed with the idea in the 1930s, it was Bill Hamilton (in 1964) who recognized its full importance and who developed a formal quantitative theory, in which he introduced the logic of ‘inclusive fitness.’ This general body of theory is now known under the heading ‘kin selection,’ although strictly speaking this refers to the narrower subset of conditions in which individuals assist other individuals that share the same copy of a gene through their close genealogical relatedness. Even if natural selection favors nonaltruists over altruists within groups of individuals likely to share the altruism trait, the proportion of altruists may nevertheless increase if they do better overall than alternative groups without altruists. Perhaps the easiest way to understand the logic of kin selection is through Hamilton’s rule, which states that a form of a gene that causes an individual to perform an altruistic act will tend to spread so long as rb – c > 0, where b (broadly speaking) is the fitness benefit to the recipient from the altruistic act, c is the fitness cost to the altruist, and r is a measure of relatedness. The rule is shorthand for a full population genetics model, and, strictly speaking, it is only correct if we define the terms in very particular ways. For example, r formally measures how genetically similar two individuals are when compared to two random ones in the population with which the altruist will compete for entry into the next generation (indeed, r can be negative). Through its simplicity, Hamilton’s rule serves to highlight the composite minimum conditions for kin-based cooperation, which are both ecological (mediated through b and c) and genetical (mediated through r). Acts of kin-based cooperation are not confined to complex animals and can, for example, be seen in the social amoeba (‘slime mold’) Dictyostelium discoideum, in which solitary cells start to aggregate under harsh conditions to produce a ‘slug.’ Some cells eventually produce
396
Cooperation and Sociality
the spores that can colonize new areas, but only at the expense of a minority of other cells that collect together to hold the spore-producing cells aloft. The cells involved in producing the colony are frequently (but not always) genetically identical, so this extreme altruism can potentially be understood in terms of kin selection; rather like individual cells in a multicellular organism, here helping others reproduce is tantamount to helping yourself. The colonies maintained by wasps, ants, and bees represent the classical examples of highly cooperative social groups. Here, sterile masses of individuals work by gathering food, cleaning the nest, and repelling predators, all in the service of a small reproductive minority. In a series of seminal papers, Hamilton proposed that their unusual genetics (specifically, their haplodiploid sex determination, in which fertilized eggs develop into females while unfertilized eggs become males) might help to explain the prevalence of cooperation in these groups. One implication of the genetics is that females are more related to their full sisters (relatedness 0.75) than they would be to their own offspring (0.5). Given this asymmetry, it is easy to see how sisters might be selected to forego their own reproduction if such behavior can help queens produce more sisters. Unfortunately, while haplodiploidy may well help to explain the evolution and maintenance of cooperation in the highly social (‘eusocial’) insects, it cannot provide the complete solution. Once one factors in the 0.25 relatedness of sisters to brothers, the overall relatedness among siblings in haplodiploids is not especially high. Multiple queens and multiple matings with different males further act to reduce the average relatedness of offspring within a colony. While the manipulation of sex ratio and the ability of females to preferentially favor their sisters may each play a role in enhancing relatedness between the donor and the recipient of cooperation, other factors that also help tip the balance include punishment of workers that attempt to lay their own eggs. Thus, even if high relatedness does not provide the whole explanation, then it may help level the playing field considerably, making cooperative behaviors more likely. Another good example of kin selection is seen in the formation of particular types of microbial mat. Laboratory populations of the bacterium Pseudomonas fluorescens rapidly diversify when maintained in unshaken broths, and a particular form – known as the ‘wrinkly spreader’ (produced by single mutations with large effects) – tends to build up at the liquid–air interface, creating a surface scum. The ability to live at the boundary layer allows the wrinkly spreader bacteria to avoid the oxygen-deprived conditions deeper in the water column, but it comes at some cost. To form and maintain the mat, the wrinkly spreaders have to make a cellulose polymer (glue), a metabolic cost that is not borne by other forms of the bacterium. Despite this cost, the mat can initially develop by kin selection (individuals helping to bind to the surface
397
share the same trait for making glue), but it undergoes periodic collapses as nonglue-making ‘cheats’ invade. The stability of some forms of cooperation may depend on both kin selection and other more direct forms of return. Providing ‘parental’ support to young that are not your own is relatively common in vertebrates. For example, in populations of birds such as Florida scrub jays and long-tailed tits, and mammals such as meerkats (suricates) and brown hyenas, there are nonbreeders that help raise young produced by dominant breeders. Why do not nonbreeders go it alone, rather than help look after another’s offspring? Kin selection may play an important role – indeed, groups in cooperatively breeding species are typically made up of extended families, so that subordinates often help their relatives. Moreover, studies have shown that helpers sometimes provide their closer kin with preferential care. However, direct fitness benefits may also be important in maintaining cooperation in this type of system and may be even more important than kinship itself. In some cases, helpers may be forced into helping behavior to avoid punishment, but by helping they may also sometimes increase their chance of inheriting the territory of the breeding pair (they are ‘paying the rent’). Likewise, the increased survival chances from grouping together may sometimes outweigh the costs of helping. In meerkats, for example, the foraging success and survival of all group members increases with the size of the group (an example of the ‘Allee effect’). It is these and other nonkin routes to cooperation that we now consider.
Reciprocal Altruism Kin selection may help to explain many cases in which individuals incur costs that benefit others, but as we have already seen, some of these examples involve additional phenomena that further maintain cooperation. At an extreme, how do we explain examples of cooperation among nonrelatives? Included in these examples are food sharing among unrelated crows and impala that groom unrelated individuals within the herd. Perhaps by helping others, the donor might subsequently be helped by the receiver when its own need arises? Evolutionary biologist Robert Trivers presented just such an explanation in the early 1970s, referring to the phenomenon as ‘reciprocal altruism.’ To understand how cooperation might be maintained under these circumstances, mathematical modelers have spent a great deal of time and effort elucidating the types of strategy that would do well in a simple game, known as the two-person iterated Prisoner’s Dilemma. In each round (‘iteration’), two players decide simultaneously whether to ‘cooperate’ (C) or ‘defect’ (D), just as two prisoners accused of a joint crime may decide to cooperate with each other by staying quiet under interrogation, or defect on their criminal partnership by talking to the police in exchange for a lighter sentence. In the Prisoner’s
398
Cooperation and Sociality
Dilemma (see Table 1), mutual cooperation (‘CC’) pays more to both players than mutual defection (‘DD’), but defecting while your partner cooperates (‘DC’) pays the defector most of all (reflecting a ‘temptation’ to defect) and a sole cooperator least of all (‘CD,’ the ‘suckers payoff ’). Many researchers consider the Prisoner’s Dilemma the key metaphor for understanding cooperation, primarily because it captures the temptation to defect, but also because it can reflect some of the damaging effects of pure self-interest. However, in a one-off game, the most rewarding strategy is always to defect because whether your partner cooperates or defects, your best option is to defect. So, how can cooperation ever evolve? One solution to the problem arises if the players have a chance of meeting again. In an iterated two-player game (played repeatedly, where the number of rounds is not known in advance by the players), the set of potential strategies is enormous – especially if long memories of previous interactions are allowed. One such strategy called ‘tit-for-tat’ (TFT – cooperate on first move, thereafter follow the partner’s previous move) has been widely recognized as a successful strategy in these iterated games. TFT is thought to be particularly effective because it is ‘nice’ (in that its starting move is to cooperate), retaliatory (in that it follows a defection from the partner with defection), and forgiving (in that it subsequently matches any cooperative act with cooperation). However, TFT is not without its weaknesses: for example, if mistakes are occasionally made, two tit-for-tatters can get stuck into indefinite rounds of defection. As an alternative, the win–stay, lose–shift strategy (WSLS – keep to your strategy if your previous exchange was high paying (DC or CC) but otherwise change) can correct occasional mistakes, although it is still open to drift in a population of cooperators. While such analyses help explain why certain types of cooperative behavior are more successful than others, they should be viewed as providing aids to thinking – not specific quantitative predictions about behavior in the real world. Can direct reciprocation explain examples of cooperation that we see around us? There do appear to be some good examples of reciprocation maintaining altruism, but Table 1 An example of the payoffs involved in a two-player Prisoner’s Dilemma Player A decision
Player B decision
Payoff to player A
Cooperate
Defect
Cooperate Defect
3 (R) 0 (S)
5 (T) 1 (P)
Participants must simultaneously decide whether to cooperate or defect. The defining inequalities of the dilemma require that T > R > P > S (and, for technical reasons, R > [S þ T ]/2), such that the temptation ( T ) to defect exceeds the reward (R) from cooperation, which in turn exceeds mutual punishment (P). The sucker’s payoff (S) the least that can be expected.
not many. The classical example is reciprocal blood sharing in vampire bats, in which females regurgitate blood meals to roost mates who have failed to obtain food in their recent past. While nest mates are often related, there appears to be more structure to the interaction. In particular, experimentally starved bats who received blood, subsequently gave blood to the former donors more often than one would expect by chance. Likewise, in laboratory experiments, cotton-top tamarin monkeys gave more food to a trained conspecific who regularly offered them food in the past, compared to an individual who never gave them food. Studies of grooming have also produced some clear cases of reciprocal altruism. For example, on the African savannah, impala frequently approach one another and begin grooming. Like the vampire bat example, the benefit, in this case removing parasites, may be high, but the costs of grooming in terms of time, fluids, and energy may be relatively low. Here, individuals deliver grooming in bouts (‘parcels’ of 6–12 licks), and the number of bouts received and delivered is remarkably well matched: in this case, defection involves simply walking away or doing nothing. While the relationship is based on reciprocation, it seems very likely that parceling up the cooperative acts in this way helps reduce the temptation to defect. Business deals often show a similar structure to avoid exploitation – half paid in advance and the other half paid when the job is complete. Male red-winged blackbirds in North America also appear to cooperate, sometimes coming to the aid of neighboring males in defending their nests and territories from potential predators such as American crows. One possibility is that the helpers are in fact the true fathers of some offspring on the neighboring territory and are selected to help out of sheer self-interest, that is, simple parental care. Alternatively, or in addition, the helper may benefit directly by removing any potential predator from the neighborhood (‘not in my back yard’), and any benefit to the neighbor is incidental (a by-product ‘mutualism’). R. Olendorf and colleagues recently put these and other explanations for cooperation to the test and ruled out any kin-based explanations on the basis of genetic analyses. However, they also looked for evidence of reciprocity by examining patterns of nest defense against a stuffed crow and simulating cheating by making it appear that a neighbor was not helping with the defense (a ‘defection’). As anticipated, male blackbirds tended to decrease their defense against a potential nest predator after their neighbor appeared to defect in the earlier trial, suggesting that reciprocation was having an important role in maintaining cooperation – ‘‘I’ll help mob your predators, if you help mob mine.’’
Indirect Reciprocity By its very nature, direct reciprocity requires repeated dealings among the same sets of individuals, so it cannot
Cooperation and Sociality
apply to cases of helping strangers we might never see again. However, what if others were looking on? Perhaps by helping others one might gain sufficient reputation as a ‘nice’ individual that strangers would be willing to help you when your own need arose. So, instead of ‘You scratch my back, and I’ll scratch yours,’ one could consider another, seemingly even more vulnerable, guiding principle ‘You scratch my back and I’ll scratch someone else’s.’ This is called the ‘indirect reciprocity’ route to cooperation. Although it may at first seem strange, bear in mind that what matters is that acts of cooperation are returned, not who returns them. Examples of the importance of maintaining an untarnished reputation are widespread in human societies. For example, eBay in part relies on reputation to maintain honest transactions when it provides scores of partner satisfaction. Being a good person or good company to deal with, does not in itself explain cooperation, but it begins to suggest a role for reputation in partner choice. Building on earlier arguments by Richard Alexander, mathematical modelers have demonstrated the theoretical plausibility of cooperation via indirect reciprocity by showing that behavioral rules can evolve in which individuals are more prepared to help strangers if these strangers have a reputation for cooperating. Of course, since reputable individuals tend to provide assistance to similar reputable individuals, kin selection may also play a key role here. Can indirect reciprocity explain cooperation in humans? After all, humans frequently help others who may never have an opportunity to reciprocate, and it is possible that such acts of kindness are recognized and rewarded by others. Staged laboratory games support this view. For example, in a recent experiment, human subjects were repeatedly given the opportunity to give money to others, having been informed that they would never knowingly meet the same person with reversed roles (all donations were anonymous). Despite the anonymity, the personal histories of giving and not-giving were displayed for participants to see at each interaction. As one might expect, the authors found that donations were significantly more frequent to those receivers who had been generous to others in earlier interactions. There are far fewer examples of indirect reciprocity in non-humans and they mainly include examples of cooperation between species rather than within species. One recent example comes from work on cleaner fish mutualisms. The cleaner fish Labroides dimidiatus remove skin parasites from their fish clients, but there is an apparent temptation for them to take a little more at the expense of the client by feeding on their mucus. Clients are faced with the challenge of getting cleaners to feed against their preferences if they are to come away unscathed. Field observations indicate that client fish almost always invite a potential cleaner to draw closer
399
and inspect them if they have had the opportunity to see that the cleaner’s previous interaction ended without conflict. By contrast, clients invite particular cleaners far less frequently if they observed that the last interaction of the cleaner ended with conflict, such as being chased away. So, a good reputation is good for the cleaner’s business, and it may be an important way in which clients avoid ‘defections.’
Strong Reciprocity Everyday observation suggests that there is a sense of ‘fair play’ in human societies. This has been backed up by more formal, if abstract, experiments. For example, in the wellknown ‘ultimatum’ game in economics, two players A and B have to agree on how a monetary reward has to be shared. Player A (the proposer) has one chance to suggest how the money is to be shared (e.g., 60% to player A, 40% to player B), but player B (the responder) can accept or reject the proposed division. If the bid is rejected, then both receive nothing, but if the bid is accepted then the proposal is implemented. The logical optimum is to offer the responder an almost negligible amount (1 cent, say, because 1 cent is better than nothing). However, this logical response may be grossly inadequate, because a common result in this type of game is that responders tend to reject proposals if the offer is anywhere less than about 25% (even when the sum is quite considerable). Other primates may also exhibit what we might think of as a sense of justice. In an intriguing study entitled ‘Monkeys reject unequal pay,’ Sarah Brosnan and Frans de Waal investigated what happens when capuchin monkeys previously trained to exchange a pebble for a piece of cucumber started to see others being rewarded with a more favored food (a grape) – the monkeys tended to go on strike, refusing to exchange a pebble for cucumber even though the alternative was no reward at all. Perhaps punishment can play a role in maintaining fair play and hence cooperation? Some may not see it as cooperation at all, bordering more on enforced slavery than on acts of ‘kindness.’ Nevertheless, when we see apparent examples of altruism we need to ask whether the threat of punishment is helping to maintain it. In a series of staged repeated games among human volunteers, Ernst Fehr and Simon Ga¨chter showed that students are prepared to take on costs in order to punish those who had earlier shirked their opportunity to contribute to a public good. According to Fehr and Ga¨chter, this altruistic punishment is simply a consequence of a ‘negative’ emotional reaction to the sight of somebody free-riding (although this proximate negative reaction may ultimately have evolved for other reasons). As one might expect, those that were punished for not contributing learned their lesson and cooperated more in subsequent rounds.
400
Cooperation and Sociality
Moreover, those games that prevented altruistic punishment altogether saw a marked reduction in the mean amount of cooperation over time. The behavioral tendency to cooperate for the public good yet punish noncooperators has been called ‘strong reciprocity.’ Strong reciprocity may explain many examples of human-based cooperation, but it is difficult to understand how altruistic punishment might evolve as a consequence of natural selection. After all, if altruistic punishment is costly, then an individual that free-rides and lets others do the policing would tend to leave more offspring. The temptation to sit back and let others punish defectors has been termed a ‘second-order defection’ or ‘twofold tragedy.’ So, if strong reciprocity can explain cooperation, perhaps it has only replaced the problem with another one further down the line – why should you be the one to punish? Kin selection may provide one potential solution to this question, but note that kinship can reduce the underlying incentives to defect in the first place. For example, in many eusocial Hymenoptera, worker-laid eggs are killed by the queen and other workers. In a comparative analysis, Tom Wenseleers and Francis Ratnieks found that fewer workers reproduced when the effectiveness of policing worker-laid eggs was higher, indicating that these sanctions were an effective deterrent. However, higher relatedness among colony workers led to less policing, not more, a result which is consistent with the view that less policing is needed when workers are highly related. So, self-restraint based on kin selection can achieve for free what expensive policing could bring about.
Escaping from Prison All adaptive explanations of altruism involve taking the ‘altruism out of altruism,’ either by showing how the actions can benefit other individuals carrying the same traits, or by showing how the nature of the interaction is such that it is in the ultimate interests of the altruist to cooperate. However, this commonality should not be taken to mean that all cooperation can be related back to the two-player Prisoner’s Dilemma (or n-player version of it which can give rise to the ‘tragedy of the commons’). One example of cooperation which is almost certainly not represented by a Prisoner’s Dilemma comes from recent work on a species of fiddler crab on the northern coastlines of Australia, where males aggressively defend their burrows from other wandering males (intruders). Patricia Backwell and Michael Jennions found that male fiddler crabs may sometimes leave their own territories to help neighbors defend their territories against these floating intruders. Why be a good Samaritan? It turns out that reciprocity cannot explain it because the ally that came to the neighbor’s assistance was always bigger than the neighbor itself. Here, it may directly benefit a resident to help its neighbor to defend a territory, so that
it can avoid having to renegotiate the boundaries with a new and potentially stronger individual. In this way, there is no temptation to cheat – large allies are helping themselves, and it is only incidental that helping the neighbor keep its territory is part and parcel of maintaining status quo. Our interpretation of cooperation gets tested further when we observe that some individuals may actually pay a cost to acquire or enhance the by-product benefits produced by another (a phenomenon known as ‘pseudoreciprocity’). Many lycaenid caterpillars, for example, produce sugary secretions that are consumed by ants and in turn the ants protect these individuals from predation. The sugar may be viewed as an investment, yet the protection arises as a consequence of general territorial ant defense. Likewise, many flowers attract pollinators using nectar. One might wonder why flowers do not save themselves the trouble and produce less nectar. Flowers are essentially in a ‘biological market,’ however, governed by simple laws of supply and demand, such that any flower that offers less than conspecifics may experience reduced pollination. So, while kin selection may be at the heart of much intraspecific cooperation, sometimes cooperation can be maintained by a complex interplay of several types of interaction including direct self-interest, reciprocity, reputation, partner choice, and the threat of punishment.
Acknowledgment This study is based in part on a longer review in Sherratt and Wilkinson (2009) which provides full references to many of the studies described here. We thank our editor Joan Herbers for her constructive comments and helpful advice. See also: Kin Selection and Relatedness.
Further Reading Axelrod R (1984) The Evolution of Cooperation. London: Basic Books. Axelrod R and Hamilton WD (1981) The evolution of cooperation. Science 211: 1390–1391. Connor RC (1995) Altruism among non-relatives-alternatives to the Prisoner’s Dilemma. Trends in Ecology and Evolution 10: 84–86. Doebeli M and Hauert C (2005) Models of cooperation based on the Prisoner’s Dilemma and the Snowdrift game. Ecology Letters 8: 748–766. Dugatkin LA (1997) Cooperation Among Animals: An Evolutionary Perspective. Oxford: Oxford University Press. Fehr E and Ga¨chter S (2002) Altruistic punishment in humans. Nature 415: 137–140. Hamilton WD (1964) Genetical evolution of social behaviour I. Journal of Theoretical Biology 7: 1–16. Krause J and Ruxton GD (2002) Living in Groups. Oxford: Oxford University Press. Noe¨ R and Hammerstein P (1995) Biological markets. Trends in Ecology and Evolution 10: 336–339. Nowak MA (2006) Five rules for the evolution of cooperation. Science 314: 1560–1563.
Cooperation and Sociality Nowak MA and Sigmund K (2005) Evolution of indirect reciprocity. Nature 437: 1291–1298. Sherratt TN and Wilkinson DM (2009) Big Questions in Ecology and Evolution. Oxford: Oxford University Press. Trivers RL (1971) Evolution of reciprocal altruism. Quarterly Review of Biology 46: 35–57.
401
West SA, Griffin AS, and Gardner A (2007) Social semantics: Altruism, cooperation, mutualism, strong reciprocity and group selection. Journal of Evolutionary Biology 20: 415–432. West SA, Griffin AS, Gardner A, and Diggle SP (2006) Social evolution theory for micro-organisms. Nature Reviews Microbiology 4: 597–607.