In this blog post, we will examine whether using game theory to explain reciprocal altruistic behavior in animals is indeed appropriate by comparing it to the theory of the selfish gene.
Introduction
In ‘The Selfish Gene’, Richard Dawkins presents two theories—the selfish gene theory and game theory—to explain reciprocal altruistic behavior in animals, which appears to contradict his central claim, “individuals are selfish,” made early in the book. In Chapter 10, Dawkins uses the selfish gene theory to explain why reciprocal altruistic behavior was bound to evolve. However, in the following chapter, he explains the same behavior by applying game theory, specifically the concept of ESS (Evolutionarily Stable Strategy). At first glance, it may seem that both the theory of the selfish gene and game theory describe the same phenomenon, and one might feel that applying either theory leads to the same conclusion. However, there is a fundamental difference in the process by which they reach that conclusion.
Game theory is a theory developed to explain human behavior, in which individuals predict the consequences of their actions and, based on those consequences, choose strategies that are advantageous to themselves. However, Dawkins applies this theory to explain animal behavior as well. Of course, it is true that game theory appears similar to the theory of the selfish gene in that it explains how reciprocal altruism can emerge as individuals pursue their own interests. However, there are more significant differences than similarities between the theory of the selfish gene and game theory, and considering these differences, it can be argued that explaining reciprocal altruism in animals using game theory is inappropriate.
In this article, I will examine why it is inappropriate to explain animal behavior—which, unlike that of rational humans, is described as operating according to a behavioral system designed by genes—using game theory, focusing on the differences between the selfish gene theory and game theory.
Dawkins’ Explanation of Reciprocal Altruistic Behavior Using the ESS Concept
In his book, Dawkins presents the behavior of birds removing parasites—ticks—from each other’s bodies as a prime example of reciprocal altruism. While a bird can easily remove a tick attached to its torso with its beak, it is difficult for a bird to remove a tick attached to its head on its own. Consequently, the behavior of removing ticks from each other’s heads arises naturally, and this behavior is precisely a prime example of reciprocal altruism.
The most widely known example, also introduced in the book, is the vampire bat. Vampire bats, which inhabit the tropical regions of Central and South America, have a very high metabolic rate; if they go without food for about 2–3 days, they are highly likely to starve to death. Consequently, it has been observed that individuals who have successfully hunted regurgitate blood to share with those who have failed to do so. Subsequent research has confirmed that this behavior occurs not only among relatives but also among individuals who have maintained long-term social relationships. Thanks to this mutual cooperation, the survival prospects of vampire bats are greatly enhanced.
Dawkins also explains this reciprocal altruism—in which individuals of the same species but with no blood relation help one another—through his theory of the selfish gene. He argues that this behavior evolved through natural selection because individuals who cooperate with one another have an advantage in survival and reproduction over those who do not. Even intuitively, it is easy to understand that reciprocal altruism is an evolutionarily advantageous behavior.
Dawkins does not rely solely on this intuition but provides a scientific explanation. Initially, only individuals that did not help one another existed, but through random mutations, individuals “designed” to cooperate emerged; as these individuals exhibited higher survival rates, they were selected by natural selection and gradually became dominant within the population. Through this process, Dawkins convincingly explains reciprocal altruism in animals using the theory of the selfish gene.
However, Dawkins does not stop there; he further applies game theory. In the latter part of ‘The Selfish Gene’, he uses the Iterated Prisoner’s Dilemma to explain reciprocal altruism in animals. If we map the behavior of birds removing ticks from one another onto the Iterated Prisoner’s Dilemma, removing ticks for a partner represents a cost in terms of time and energy, while having a partner remove one’s own ticks constitutes a reward. While it may seem advantageous in the short term to help no one and pursue only one’s own interests, in situations where relationships continue over multiple interactions, a strategy of sustained cooperation yields the greatest long-term benefit.
To explain this, Dawkins uses the concept of the ESS. An ESS is a strategy that cannot be displaced by another strategy when the majority of a population adopts it.
He presents the “Hawk” and “Dove” model as a prime example. Doves engage only in threatening behavior rather than actual combat, while Hawks actively engage in combat. When a Hawk and a Dove encounter each other, the Dove retreats, so the Hawk wins. When two Hawks encounter each other, they fight, risking serious injury, while two Doves face off without actually fighting until one backs down.
Dawkins quantifies these situations to run simulations. He assigns 50 points for a victory, 0 points for a defeat, -100 points for a serious injury, and -10 points for wasting time without an actual fight. Under these conditions, if all individuals are doves, the average score is high; however, once a few hawks appear, an equilibrium forms at a certain ratio over time. In other words, while the average efficiency of the group as a whole may decline somewhat, a stable state emerges in which no single strategy completely replaces the others—and this is the ESS.
To apply this ESS concept to reciprocal altruism, Dawkins posits three types of individuals. First is the “sucker,” who always helps others; second is the “cheater,” who only takes help and never returns it; and third is the “grudge-holder,” who helps when helped but refuses to cooperate again if betrayed.
Dawkins ran computer simulations in a scenario where “suckers” made up the majority, while “grudge-holders” and “cheaters” were in the minority. Initially, the number of suckers declined sharply as cheaters continuously exploited them. Later, once the suckers disappeared, the cheaters could no longer reap one-sided benefits, and the grudge-holders gradually increased as they excluded the cheaters. Eventually, the cheaters disappeared, leaving only the grudge-holders.
Since “retaliators” maintain a strategy of cooperating only when others cooperate, the group as a whole ultimately becomes a society that sustains reciprocal altruism. Based on these results, Dawkins explains that reciprocal altruism in animals today has established itself as an evolutionarily stable strategy. However, there is a significant problem with this explanation.
Similarities and Differences Between the Selfish Gene Theory and Game Theory
Evolution fundamentally occurs through four processes: replication, mutation, competition, and selection. Applying the selfish gene theory to explain the evolution of reciprocal altruistic behavior, as we saw earlier, a cooperating individual emerges through a random mutation in a group consisting solely of non-cooperating individuals, and this individual competes with the existing individuals. Subsequently, as the cooperative individuals gain an advantage in survival and reproduction, they undergo natural selection and gradually become dominant within the population. The key concepts here are “chance” and “mutation.” Mutations occur in genes independently of an individual’s will; if the result is advantageous for survival, it spreads through natural selection, and if not, it disappears. In this process, natural selection determines which mutations survive.
Game theory, on the other hand, is not a theory that explains the process of natural selection. Game theory was developed from the outset to explain human behavior; it is the choices of rational agents—not natural selection—that produce outcomes. Players analyze the results of their previous actions and predict future outcomes to adjust their strategies in a way that maximizes their own interests. This process of strategic choice is at the heart of game theory.
However, the animals to which Dawkins seeks to apply game theory do not behave in this manner. Dawkins himself describes animals in his book as beings that act according to a program designed by their genes. In other words, animals do not analyze the results of their actions or predict the future to adjust their strategies; they simply carry out the behavioral patterns dictated by their genes. Of course, whether explained by game theory or the theory of the selfish gene, the ultimate conclusion—that cooperative individuals survive—remains the same.
However, the two theories show a decisive difference in the process. In the theory of the selfish gene, an individual’s survival and prosperity are the result of natural selection, and the individual does not choose the direction of its own evolution. In contrast, in game theory, the agent creates its own survival and prosperity by choosing and modifying its own behavior. This is precisely the most important difference between the two theories.
Why Game Theory Cannot Be Applied to Reciprocal Altruism in Animals
As discussed earlier, reciprocal altruistic behavior in animals is explained not as the result of thought or reasoning, but as behavior that arises from a behavioral system designed by genes. According to the theory of the selfish gene, individuals merely carry out behaviors programmed by their genes, and genes cannot predict the consequences of their actions. Furthermore, whether an individual survives is determined solely by natural selection, not by the individual’s own choice.
In contrast, game theory presupposes the existence of beings capable of predicting the future based on the outcomes of past actions and selecting more advantageous strategies based on those predictions. In other words, game theory was developed to explain human behavior characterized by foresight and strategic thinking, and historically, it has evolved through the analysis of decision-making in human society.
In game theory, the reason a particular strategy survives is not because nature selected it, but because the actor chooses that strategy after considering their own interests. However, Dawkins applies game theory to animal behavior without fully considering this premise, using it to develop his own logic.
Conclusion
From the very beginning of ‘The Selfish Gene’, Dawkins describes all living organisms, including humans, as “survival machines” that exist for the survival and reproduction of genes. Genes design the behavior of survival machines to maximize their own replication, and individuals act according to that program. From this perspective, Dawkins’ explanation of reciprocal altruism in animals through the theory of the selfish gene can be considered a sufficiently persuasive argument.
However, the problem lies in his attempt to explain the same phenomenon using game theory. In simulations using the repeated Prisoner’s Dilemma, a society composed of “conformists,” “cheaters,” and “retaliators” ultimately results in only the “retaliators” surviving. Since “retaliators” maintain a strategy of cooperating when their partner cooperates but never cooperating again when betrayed, the entire society eventually reaches a state of mutual cooperation. This result is presented as evidence supporting Dawkins’ claim that reciprocal altruism is an evolutionarily stable strategy.
Looking only at the conclusion, it might seem that there is no major problem since the same result is obtained even when game theory is applied. However, game theory is, by its very nature, a theory designed to analyze the decision-making of rational human beings. Game participants analyze the results of previous choices, predict the future, and then select the strategy most advantageous to themselves. In contrast, according to Dawkins’ explanation, the genes are the agents of actual behavior, and individuals merely act according to the genes’ instructions. The assumption that genes analyze past outcomes, predict the future, and select new strategies does not align with the basic premise of the selfish gene theory.
A “survival machine” designed by genes—which possess neither reason nor the ability to predict the future—and game theory, which presupposes reason and strategic thinking, start from entirely different points. Applying game theory directly to reciprocal altruism in animals is akin to interpreting a fly’s behavior of rubbing its front legs as a plea for food, just as a human might do. Humans act with intent, anticipating the other’s response, but flies do not act based on such decision-making. Similarly, I believe it is logically problematic to apply game theory directly when explaining reciprocal altruism in animals.
Therefore, while Dawkins’ attempt to apply game theory to explain reciprocal altruism may be an interesting interpretation, it has limitations as a theory that explains the actual mechanisms of animal behavior, and it is difficult to view it as sufficiently persuasive from a logical standpoint.
