This blog post analyzes how well the selfish gene theory and game theory explain reciprocal altruism in animals.

Introduction

In his book The Selfish Gene, Richard Dawkins presents the theory of the selfish gene, which explains the principles of evolution based on the self-interest of genes. According to the theory of the selfish gene, the agents of evolution are genes, and genes aim to replicate themselves and spread widely over long periods. Genes use individuals—survival machines—as means to achieve their goals, and the observable behaviors of individuals can be explained through the selfishness of genes. Dawkins also explains reciprocal altruism—the behavior of animals helping each other despite belonging to the same species but lacking a blood relationship—through the theory of the selfish gene. However, in the same book, the author also explains reciprocal altruism among the same animals using game theory. His intention is to treat genes as analogous to humans in general game theory, explaining competition between animal individuals as a game between them.
Game theory is a framework that greatly aids in predicting and implementing interdependent decision-making. It explains how game participants, pursuing their own interests, end up engaging in reciprocal altruistic behavior where they mutually benefit each other. Thus, it appears very similar to the selfish gene theory. However, while the agents in games addressed by game theory play with specific strategies to maximize utility, animals, unlike humans, do not act rationally. Therefore, applying game theory to animal interactions is fundamentally unsuitable. This essay aims to demonstrate that applying game theory to explain animals’ reciprocal altruism is inappropriate, highlighting the greater differences between the theory of the selfish gene and game theory than their similarities.

Game Theory and Reciprocal Altruism

The issue this essay addresses is Dawkins’ use of game theory to explain reciprocal altruism. Reciprocal altruism is a concept frequently used to explain unusual behaviors between individuals of different species. According to Robert Trivers, who pioneered this theory, animal sociality evolved through acts of helping others not for immediate reciprocation but in anticipation of future repayment. This dimension of explanation can be applied to acts of kindness and reciprocity between individuals of the same species who are not related. Representative examples include birds that remove aphids from each other’s backs and vampire bats. These individuals of the same species perform specific acts for each other, expecting that while they help the other, the other will also reciprocate through some act.
Dawkins’ theory of the selfish gene posits that genes are the agents of evolution. Consequently, individual genes drive the behavior of organisms in ways that ensure the continued survival of genes of the same type. This appears to conflict with reciprocal altruism, where individuals of the same species but no blood relation engage in mutual aid. According to the selfish gene theory, this is because only identical genes, belonging to the same survival machinery, have a justification for helping each other.
Trivers’ reciprocal altruism is ‘altruism,’ but it is fundamentally behavior that expects a return for the act of kindness. The reason we can help even different species is because there is a benefit that returns to us in accordance with the help given. Suppose one individual performs an act of kindness toward another that can be explained by reciprocal altruism. The recipient of the kindness receives help in surviving thanks to the other individual. At this point, there are two possible courses of action: reciprocating or not reciprocating. If they reciprocate, there is no issue. However, if they do not reciprocate, the individual who initially extended kindness based on reciprocal altruism becomes evolutionarily disadvantaged. Yet, as Williams noted, reciprocal altruism originates from the ability to recognize other animals and perform kind acts. Consequently, it is possible to go further and recognize whether the recipient reciprocates the initial kindness. In this situation, if individuals exist that refrain from offering secondary acts of kindness only to those who do not reciprocate, they gain a relative advantage over individuals controlled by genes that do not know how to reciprocate. Ultimately, combinations where acts of kindness and reciprocation can be exchanged will prevail in competition. Dawkins refers to the gene that remembers non-reciprocal individuals and ceases to extend further kindness as the ‘resentment gene’. This strategy adopted by grudge-holders always places them in an advantageous position, regardless of whether other gene types emerge—whether genes that show kindness regardless of the other’s reciprocation or genes that never reciprocate regardless of the other’s kindness. Therefore, it can be recognized as an ESS. In such a system where the resentful gene prevails, reciprocal altruism naturally leads to symbiotic relationships, and indeed, numerous examples have been observed.
Having explained reciprocal altruism through the selfish gene theory, Dawkins then applies game theory to achieve the same goal. Reciprocal altruism resembles a game where two prisoners each hold two options: cooperation and defection. Dawkins argues that game theory—the study of what strategy prisoners should adopt to gain an advantage—can be applied to explain reciprocal altruism. From each prisoner’s perspective, choosing betrayal is the rational strategy. However, both prisoners know that if they both choose the rational strategy, they will face punishment or receive low rewards. Conversely, if both prisoners cooperate, they will receive high rewards. This situation is called the ‘Prisoner’s Dilemma’. In a simple Prisoner’s Dilemma game, there is no way to verify trust, and the game is ultimately destined to end in mutual betrayal, resulting in a bad outcome for both prisoners. However, if this Prisoner’s Dilemma is placed in a situation where it is repeated, the situation changes. According to Dawkins, two individuals in a reciprocal altruistic relationship can choose to perform or withhold acts of kindness and reciprocity, and since these choices can be made repeatedly over time, they are effectively playing a Repeated Prisoner’s Dilemma game. In a one-time game, the defection strategy yields the greatest payoff. However, in repeated games, multiple strategies are possible. To determine which strategy is most advantageous among the possible ones, Axelrod proposed strategies, translated them into a common programming language, and pitted them against each other. The winning strategy was ‘Tit for Tat’ (TFT). This strategy starts cooperatively in the first round and then simply mimics the opponent’s previous move in subsequent rounds. This proves to be the most advantageous strategy. Its context is quite similar to that of a ‘tit-for-tat’ player, who decides whether to cooperate in the next game based on whether the opponent cooperated in the previous game. Because a strategy of always being kind can naturally integrate into a TFT-based system, it is difficult to precisely view TFT as an ESS. However, considering aspects like its immunity to intrusion by strategies of betrayal and retaliation, a mixed strategy that fundamentally acts kindly while employing TFT at appropriate times can be seen as an ESS. Thus, since the ‘retaliator’ aspect of reciprocal altruism can be explained through game theory and TFT, it seems plausible at first glance to recognize game theory as a tool for understanding reciprocal altruism. In repeated games, teams where everyone always cooperates gain the greatest benefit, making them fit for survival and leading to evolution toward this reciprocal altruistic behavior. In other words, Dawkins explains that altruistic species arise because, based on game theory, engaging in secondary behavior maximizes the probability (or expected value) of passing on genes.

Comparing the Selfish Gene Theory and Game Theory

Thus, it appears both the selfish gene theory and game theory can be applied to explain reciprocal altruism. However, upon rigorously examining the conditions under which each theory holds, a significant divergence between the two becomes apparent. Since there is no disagreement that reciprocal altruism is ultimately a product of evolution, we will apply the theory of the selfish gene and game theory to the standard stages of evolution here and analyze their commonalities and differences.
Evolution is a process consisting of four stages: replication, mutation, competition, and selection. Genes replicate over an extremely large number of generations, and during this process, mutations occur by chance, giving rise to new genes. As diversity within the gene pool increases, genes compete with each other for dominance, and specific genes are selected through natural selection. Similarly, in reciprocal altruism, assuming all individuals were initially uncooperative, cooperative individuals emerged by chance through mutation. Since cooperative individuals were better suited for survival, they were selected by nature, leading all to become cooperative. After the initial genes existed in the early generations, mutations occur for various reasons as genes replicate through successive generations. According to the selfish gene theory, new genes capable of winning through reciprocal altruism, like the ‘one-time gene,’ emerge. Subsequently, existing genes and those arising from mutation compete within the gene pool. Survival machines controlled by the ‘selfish gene’ gain an advantageous position in reciprocal altruistic relationships with other individuals (the reason for this was mentioned earlier) and increasingly benefit in competition over time. This means the likelihood of gene dominance increases, and naturally, the genes that win the competition are selected.
While the theory of the selfish gene certainly discusses the ‘selfishness’ of genes, this is solely for the sake of explanatory convenience. It firmly establishes that there is no intervention by any rational being in the processes of mutation and selection. In other words, the process of variation occurring within the gene pool, and the process by which some genes win and are selected in competition with others, is based on chance, not on any rational thinking by someone. The term ‘selfish gene theory’ arises because the process of evolution, occurring at the gene level, appears to be driven by their selfishness. However, this name is fundamentally based on the common-sense assumption that genes themselves cannot think.
In contrast, applying game theory to animals’ reciprocal altruism is possible only under the premise that animals possess reason and consciously choose whether to cooperate or defect. Furthermore, it requires the premise that animals successfully devised and executed such strategies, thereby winning against competitors and being selected. Whether using the selfish gene theory or game theory to explain animal reciprocal altruism, the conclusion appears similar when viewed through the commonalities seen in ‘One-Han-Ja’ and TFT. However, the conditions under which each can be applied differ. In particular, while the theory of the selfish gene can well explain the mutation and selection stages during evolution, game theory involves logical leaps and thus cannot be considered an appropriate application.

Reasons Game Theory Cannot Be Applied to Reciprocal Altruism in Animals

When explaining reciprocal altruism in animals, the selfish gene theory is highly suitable for applying to the mechanically operating animal world. Conversely, according to game theory, it appears that the acting agent chooses because they predict the outcome of their own actions and their choice results in benefit to themselves, rather than due to ‘natural selection’ in evolution. The problem lies in the fact that the acting agents are irrational animals, incapable of predicting outcomes and devising strategies in the first place. While animals may superficially appear to think and act strategically, this is merely the result of genes that survived through competition and came to exist within those animals. Therefore, fundamentally, using game theory to explain animals’ reciprocal altruism is impossible, and caution is required to avoid such logical fallacies when discussing evolution going forward.
The explanation of evolution through the selfish gene theory was considered highly groundbreaking at the time of its presentation, and its significant impact led to attacks from various quarters. Viewed from another angle, the selfish gene theory is a concept with considerable influence and persuasive power. Beginning with The Selfish Gene, Dawkins has articulated his perspective on evolution in numerous publications. Whether people like his theory or not, he stands as one of the most prominent ethologists and evolutionary biologists of our time. Given his status as a popular and authoritative scientist, his theories and explanations demand greater scrutiny. This essay offers a critical perspective on applying game theory to animals’ reciprocal altruism.