Here is the final rendition of the essay if you have the patience to read it! Thanks again for your comments.
Can cooperation emerge in a world of self-interested individuals without central authority?
The only thing that will redeem mankind is cooperation. – Bertrand Russell
In his Leviathan, Thomas Hobbes famously claims that without a central authority, the life of man would be ‘solitary, poor, nasty, brutish, and short’. In a state of nature in which man is competitive, diffident and vainglorious, cooperation would never be rational and as a result, mankind would be doomed to a perpetual war of all against all. Yet we know that cooperation does occur and that our civilisation is based upon it. Is this due to fear of retribution that the state will enforce on us if we do not cooperate? Or is it possible for cooperation between individuals, and consequently societies, to evolve without this so-called ‘Leviathan’. I shall argue in this essay that it is indeed possible for cooperation to evolve without the need for a central authority and also under what conditions it will emerge.
In order to analyse whether cooperation can occur amongst self-interested individuals we can use a tool from game-theory on which to model human interaction. It is known as the Prisoner’s Dilemma (PD) and it will be prudent to spend some time understanding it in its classic form. In the Prisoner’s Dilemma game, there are two players, each of which have two choices; cooperate or defect. Defection will always yield a higher payoff than cooperation, no matter what the other does. The dilemma arises from the fact that if both defect, they are both left worse off than if they had both cooperated. An illustration of the four possible outcomes can be seen in figure 1.
Fig. 1 the Prisoner’s Dilemma
Consider a real life example (AXELROD p.7). Two industrial nations have erected trade barriers to each other’s exports. This is equivalent to the punishment of mutual defection. The advantages of free trade would mean that both nations would be better off if these barriers were eliminated, i.e. the reward for mutual cooperation. However, if one nation raises its barriers while the other doesn’t it would find itself facing terms of trade that hurt its own economy. The result is, whatever one nation does, the other nation is better off retaining its own trade barriers. Since while the payoff of 1 is less than 3, it is better than the potential of the sucker’s payoff of 0. Therefore, each nation will retain its trade barriers; even though they would be better off had both nations cooperated with each other.
In his Leviathan, Hobbes argues that cooperation can never occur amongst individuals without an absolute sovereign and that in a state of nature, mankind is doomed to a ‘war of all against all’. The only way to emerge from this warlike state is for individuals to enter into a social contract in which they relinquish their power to this absolute sovereign who is able to enforce cooperation. For Hobbes, every interaction that mankind has with one another, is like a PD. There is always more incentive not to cooperate.
The Hobbesian State of Nature
The first reality of the state of nature, says Hobbes, is the relative equality of atomic individuals. He claims that the distribution of physical and mental endowments amongst individuals is such that even the weakest, slowest and dumbest can kill the strongest, fastest and smartest. As a result, in general, all human beings are vulnerable to mortal assault at the hands of all others. The second reality is that the goods that we can acquire (e.g. food, shelter) are relatively scarce. According to Hobbes, these realities will cause self-interested individuals to be competitive, diffident and vainglorious.
Individuals in the state of nature will have a desire for self-preservation which requires the acquisition of resources necessary for survival. So, says Hobbes, “if any two men desire the same thing, which neverthelesse they cannot both enjoy, they become enemies;†This results in competition between individuals, a direct consequence of which motivates them to also be diffident. The realisation that others are competing for the goods that one has seized or will want to seize breeds distrust. In this culture of distrust the most reasonable way, says Hobbes, for any man to make himself safe is to pre-emptively strike first. The final attribute of man which contributes to the war of all against all is his desire for ‘Glory’. This is the desire to raise the value of oneself in the eyes of others, often resulting in violence.
So what is the result of these three ‘principals of discord’? Hobbes says “The first maketh men invade for gain; the second, for safety; and the third, for reputation.[…] Hereby it is manifest that during the time men live without a common power to keep them all in awe, they are in that condition which is called war; and such a war as is of every man against every man.â€
Is Hobbes right in claiming that the state of nature will result in the failure of cooperation between individuals? Consider two persons; A and B. They can either choose to invade in order to seize the other’s resources (defect) or they can refrain from doing so (cooperate). If both refrain from invading, they will gain the reward for mutual cooperation. However, if A refrains from invading while B invades then A is left with the sucker’s payoff while B gains the temptation to defect, and vice versa… If both invade, their payoff is lower than mutual cooperation but higher than the potential sucker’s payoff. But this means, as with the trading nations, that ‘no matter what the other person does, it is rational for each person in this situation to be uncooperative’. This is because a person can always do better if they do not cooperate. If B cooperates, then A gains more if A does not cooperate. If B does not cooperate, once again A gains more by not cooperating.
According to this account of conflict, it is rational for persons in the state of nature to not cooperate with others. Hobbes even says that such behaviour is virtuous: “Where there is no common Power, there is no Law: where no Law, no Injustice. Force, and Fraud, are in warre the two Cardinall virtues.†(HOBBES)
Iterated Prisoner’s Dilemma
However, this is not a true reflection of reality. If people really were in a one-time Prisoner’s Dilemma game situation then we might concede that Hobbes’s account is correct. However, the reality is that interactions amongst human beings are much closer to the iterated prisoner’s dilemma game. (Axelrod) We can imagine a person A, living in the same area as person B, coming into contact with each other many times in the space of a lifetime. Axelrod says:
“What makes it possible for cooperation to emerge is the fact that the players might meet again. This possibility means that the choices made today not only determine the outcome of this move, but can also influence the later choices of the players. The future can therefore cast a shadow back upon the present and thereby affect the current strategic situation.†(AXELROD, p.12)
In a one-time PD game, it is always rational (i.e. in the interest of the individual) to defect. However, if the frequency of future interactions is great and therefore important enough that it would be better to cooperate than to defect, then there is a potential for self-interested individuals to cooperate. This seems a more realistic model of human interaction based on the potential for individuals living in a geographically close area to meet again and again.
So if we accept this ‘shadow of the future’, what are the requirements for cooperative strategy to evolve without a central authority? Firstly the strategy has to be able to get started in a world of unconditional defection. If this is not possible then Hobbes will be vindicated that the state of nature is a war of all against all. Secondly, the strategy has to be able to thrive in a world of different strategies and finally the strategy, once established, must be able to protect itself from less cooperative strategies. (AXELROD p.21) If these requirements can be met by a strategy that promotes cooperation then we will have successfully shown that cooperation can evolve amongst self-interested individuals, including those living in the state of nature. In his book, Evolution of Cooperation, Robert Axelrod provides significant evidence that such strategies exist and do occur.
Axelrod’s Tournaments
Axelrod invited academic colleagues all over the world to devise computer strategies to compete in an IPD tournament. The programs that were entered varied widely in algorithmic complexity; initial hostility; capacity for forgiveness; and so forth. The results of this first tournament were published and Axelrod invited more submissions for a second tournament. Each strategy played against every other strategy including a RANDOM strategy which unsurprisingly cooperated and defected at random.
What Axelrod found was remarkable. The results showed that it was possible for self-interested individuals to do well, even when employing a cooperative strategy. In fact, a cooperative strategy won both rounds of the tournament. In addition to this, in a hypothetical series of tournaments, in which successful strategies were proportionally better represented in the environment in subsequent generations; the evolutionary dominant strategies were again the ones which were cooperative.
The winner of both rounds was a program called TIT-FOR-TAT which incidentally was also the simplest. TIT-FOR-TAT always cooperates on the first move and then replicates whatever the other player does on its next move. So if the player cooperates, so will TIT-FOR-TAT, but if the other player defects, then TIT-FOR-TAT will ‘punish’ them by also defecting.
Having completed those tournaments, Axelrod then constructed a whole sequence of hypothetical future rounds of the tournament. In these, the more successful strategies became a larger part of the environment. This is analogous to the process of survival of the fittest via natural selection in which better strategies for survival will result in more offspring for that individual and therefore more replications of that strategy. If a strategy did badly in Axelrod’s hypothetical future rounds it was more likely to become a lesser part of the environment and eventually extinct.
Once again TIT-FOR-TAT came out on top proving it to be an extremely successful rule. This suggests that TIT-FOR-TAT would continue to thrive in a population, and that eventually it might be used by virtually everyone. So what is it that makes TIT-FOR-TAT so robust? Why did it perform so well in every tournament? Axelrod says:
“What accounts for TIT-FOR-TAT’s robust success is its combination of being nice, retaliatory, forgiving, and clear. Its niceness prevents it from getting into unnecessary trouble. Its retaliation discourages the other side from persisting whenever defection is tried. Its forgiveness helps restore mutual cooperation. And its clarity makes it intelligible to the other player, thereby eliciting long-term cooperation.†(AXELROD p.54)
It is clear that a society in which everyone employed a strategy of TIT-FOR-TAT would be one based on cooperation. As such, can TIT-FOR-TAT meet the requirements necessary to become a dominant strategy? The answer is yes. Firstly, TIT-FOR-TAT can invade a population of unconditional defectors, as long as some clusters of individuals, who employ the TIT-FOR-TAT strategy, have a sufficient proportion of their interactions with each other. It is conceivable that these cooperators could be able to distinguish each other via distinct chemical markers thus knowing whether to enter into an interaction or not. (SKYRMS p.56) These interactions will cause their average score to be higher than those who are unconditionally defecting. As a result, the proportion of TIT-FOR-TAT strategists will grow. Secondly, as the tournament has shown, TIT-FOR-TAT is extremely robust and is successful when interacting with almost every other strategy including unconditional defection. This satisfies the second requirement which is to thrive in a world of different strategies.
Finally we come to the requirement of being an evolutionary stable strategy (ESS) . An ESS is a strategy which if adopted by a population cannot be invaded by any competing alternative strategy. Is this true for TIT-FOR-TAT? Yes, TIT-FOR-TAT is an ESS provided that the future casts a large enough shadow and the invading strategy is to always defect. Since this is true for our state of nature, we can accept TIT-FOR-TAT as meeting all three requirements.
Does this mean that TIT-FOR-TAT is always the best strategy? The answer is no. What Axelrod has shown is that there is least one strategy (TIT-FOR-TAT) that can get started in a world of unconditional defection, can become dominant in that world and crucially, defend itself against invasion as an evolutionary stable strategy. Most importantly, it is a strategy that promotes, encourages and fosters cooperation amongst self-interested individuals. The previous paragraph on the robustness of TIT-FOR-TAT suggests that similar successful strategies would employ a combination of being nice, retaliatory, forgiving and clear, depending on the environment it was in.
A real life example of how a cooperative strategy can arise is the live-and-let-live system that emerged in the bitter trench warfare of World War I (AXELROD p.87). It was common amongst units who were engaged in trench warfare, to mutually avoid confrontation and conflict. This could be in the form of bombing areas which were known to not contain any troops, or to agree to a ceasefire during certain hours to allow the troops to eat. Each unit acts as a player in the PD game and mutual cooperation is the equivalent of this mutual restraint. This is a clear example of how cooperation can occur even amongst sworn enemies who clearly do not have each others interest at heart.
So, it seems that Hobbes’s war of all against all isn’t a necessity, but before I leave Hobbes behind, I would like to address an account put forward by Jean Hampton in which she attempts to support Hobbes’s war of all against all by denying that individuals in the state of nature are able to comprehend the benefits of long term cooperation. She calls this, the short-sightedness account of conflict.
The Shortsightedness Account of Conflict
Hampton rejects Hobbes’s account that it is never rational to cooperate. She accepts that, due to the reality of IPD games in human interaction, it is rational for self-interested individuals to cooperate in the long-term. Her argument then is that most individuals in the state of nature will not realise the long-term benefits of cooperation and so instead will defect to reap the short-term benefits. There may be some portion of the population with a high enough faculty of reasoning which will enable them to understand the rationality behind cooperation. However, they will soon realise that most of the population doesn’t have that rationality and so will expect them to defect. In turn, they will adopt the short-term strategy of defection for fear of being exploited. As a result, Hampton argues that we will once again be faced with the war of all against all.
I have two objections to her argument. The first objection is that Hampton doesn’t provide substantial support to the claim that the majority of the individuals in the state of nature will not understand the benefits of long-term cooperation. It doesn’t seem any more plausible than the opposite reality. One of the reasons that TIT-FOR-TAT did so well was because of its simplicity. Other strategies were able to recognise TIT-FOR-TAT and could act accordingly. This meant that other strategies didn’t need to understand the long-term benefits of cooperation, because the cooperation was being induced by the easily recognisable TIT-FOR-TAT.
The second is that foresight is not necessary for the evolution of cooperation (Axelrod ch.5). Evolutionary biology can provide us whole host of examples of self-interested organisms who have evolved to cooperate, all of which is decided on a genetic level via natural selection. (DAWKINS 1976) There was no need for these organisms to have a high faculty of reason, or to understand the long-term benefits of cooperation. Conversely, it is precisely the fact that cooperation does have long-term benefits which causes it to become a dominant strategy amongst many organisms with no sense of rationality.
So where has Axelrod taken us? It seems clear that cooperation, in some sense, can evolve amongst self-interested individuals without a central authority. We have even seen examples of the evolution of cooperation, some real and some hypothetical. But what is the scope of this cooperation? Axelrod’s tournaments have shown us how cooperation could evolve but it requires a fair number of constraints, including a large enough shadow of the future and a suitable ESS. This may show how us how cooperation could perhaps occur but because of its theoretical nature, it is perhaps not the best model of how cooperation amongst individuals on a grand scale, i.e. the cooperation we see in our world today, came about.
This concept has been developed by Brian Skyrms in his Evolution of the Social Contract. Skyrms differs to Axelrod in that his agenda is to explain how different aspects of the implicit social contract that exist between individuals and societies have actually came about; for example justice, commitment and importantly in our case, cooperation. Axelrod created an environment and showed us how cooperation could evolve, while Skyrms attempts to show us how cooperation has evolved. Although he provides a different tactic, Skyrms will provide us with another argument as to how cooperation can (and has) evolved amongst self-interested individuals.
The Evolutionary Account of Cooperation
The main thrust of his argument is that all these aspects of the social contract have come about due to the process of differential reproduction in evolutionary biology. To understand this we can take the example of justice. In an experiment subjects were asked to divide a dollar among themselves. Not surprisingly, all agreed to a fifty-fifty split. Skyrms shows us that although this is a Nash equilibrium based on rational choice theory, it is not the only one. In fact, there are an infinite number of strict Nash equilibria which it would be rational to adopt. Skyrms argues that it is the result of evolutionary dynamics which leads us to adopt this just choice. In the third chapter, Skyrms focuses on the evolution of what Petr Kropotkin called Mutual Aid, i.e. cooperation.
As with the original formulation of the PD, Skyrms acknowledges that the strategy of always defect is a Nash equilibrium. However, in the previous examples of the PD, we have assumed that the pairings of individuals is random. The argument put forward by Skyrms relies on the assumption that in reality the pairings are not random i.e. they are correlated in some way. According to him there is “rich biological literature showing that, in nature, pairing may not be random.†(SKRYMS p.53) He cites the tendency of individuals to interact with relatives, or neighbours, or one identifies as being of the right type etc. This is a crucial difference to Axelrod’s tournaments.
Skyrms then argues that it is this correlation between pairs that can produce cooperation between individuals, even in one-time PD games. Strategies become dominant in a society in proportion to their average fitness. With random pairings, a cooperative individual would have as much chance of playing a defector as a cooperator. It is the non-random pairing that Skyrms says is more realistic, which introduces conditional proportions that give the proportion of individuals using a given strategy who will interact with individuals using the various possible strategies. If cooperating individuals are pairing with other cooperating most of the time, then the average fitness of the cooperating strategy becomes greater and eventually becomes dominant.
Consider an example where correlation is established by sensory detection; cooperators and defectors give off different types of chemical markers. At first, everyone is paired randomly but if a cooperator is paired with a defector, it doesn’t interact. The ones who didn’t interact are then paired again with each other and interact. In a graph of expected fitness of cooperation and defection against the proportion of cooperators in the population, Skyrms shows that if there are enough cooperators, it can become a dominant strategy. (SKYRMS p.57) Note this is different to Axelrod because in this example, the individuals are employing a strategy of always cooperate or always defect. There is no foresight or learning involved. The dominance of a strategy is determined by its average fitness compared to other strategies. On an evolutionary scale, cooperators who play cooperators have a higher fitness than defectors playing against defectors; therefore cooperators will take over the population.
So what have we established? I have provided two possible explanations for how cooperation between self-interested individuals can evolve without a central authority. The first, from Axelrod, relies on the evidence that cooperation becomes a rational strategy if the importance of future interactions becomes sufficiently great. As a result, small clusters of ‘cooperators’ can invade a population of unconditional defectors, such as in the Hobbesian state of nature, providing that enough of their interactions with each other. Once established, cooperative strategies tend to dominate because they do well with each other and as a consequence they have the potential to be evolutionary stable strategies i.e. ones which no other strategy can invade. This account is not only hypothetical. I have provided a real-life example of evolution of cooperation in which foresight was a factor, that of trench warfare in World War I and also an account of evolutionary biology which does not need foresight for cooperation to occur. The second explanation, from Skyrms, provides us with an argument for how cooperation really has evolved and claims that it is evolutionary dynamics that provides the necessary conditions for cooperation. This enables the argument for evolution of cooperation to have a much wider scope because it is based on more realistic assumptions of how living beings interact.
So what are the consequences of these conclusions? The results of the iterated prisoner’s dilemma tournaments show how cooperation could possibly evolve but does not show that’s how cooperation in society has evolved. The evolutionary account is perhaps more helpful in that respect. If cooperation can occur without the need for a central authority, does it make that authority obsolete? Skyrms seems to suggest that the social contract is created by evolutionary dynamics and not by the state. Justice, commitment and cooperation can all be explained by evolutionary biology. Having said that it seems that the state created by evolution would be very different to the sorts of states we have now, ruled by government. Whether it would be preferable or not is another matter.
Bibliography
AXELROD, R. The Evolution of Cooperation, Basic Books, Inc., Publishers, New York 1984.
DAWKINS, R. The Selfish Gene, Oxford University Press; 2Rev Ed edition 1990.
HAMPTON, J. Hobbes and the Social Contract Tradition, Cambridge University Press, Cambridge 1986.
HOBBES, T. Leviathan 1660.
MAYNARD SMITH, J. Evolution and the Theory of Games, Cambridge University Press, Cambridge 1982.
SKYRMS, B. Evolution of the Social Contract, Cambridge University Press, Cambridge 1996.
