Preference Evolution under Partner Choice
This paper demonstrates that when agents can choose their partners, evolutionary dynamics favor a specific class of preferences combining efficiency with parochialism, as these preferences enable agents to sustain efficient coordination while protecting themselves from exploitation.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
In the study of human behavior, economists and biologists have long asked a simple but profound question: why do we act the way we do? For decades, the standard answer was to treat our desires and values as fixed, unchangeable settings on a machine. We assumed that people simply have preferences—some want to be kind, others want to be rich—and then they act on those desires to get the best result for themselves. But a growing body of research suggests that these preferences are not fixed at all. Instead, they are the product of a slow, invisible evolutionary process. Just as physical traits like height or eye color change over generations because they help us survive, our internal motivations might also evolve. The logic is a loop: our preferences shape how we behave, our behavior determines how successful we are in life, and that success decides which preferences get passed on to the next generation. This raises a fascinating puzzle: if evolution favors success, why do we see such a mix of selfishness and kindness? Why don't we all become purely selfish, or purely selfless?
A new study by researchers Ziwei Wang and Jiabin Wu tackles this puzzle by looking at a crucial piece of the evolutionary puzzle that was often ignored: who we choose to be with. In many previous models, people were paired up at random, like drawing names from a hat. But in real life, we have a say. We choose our friends, our business partners, and our spouses. This paper explores what happens when evolution happens in a world where people can pick their partners. The researchers built a mathematical model to simulate a large population of people who can not only choose whom to interact with, but also coordinate their actions once they are matched. They wanted to see which types of personalities would survive and thrive when individuals have the power to sort themselves into groups.
The researchers tested two obvious candidates for evolutionary success: the purely selfish person and the purely efficient person. The selfish person cares only about their own gain. The efficient person cares about the total good of the pair, wanting to maximize the combined reward for both parties, regardless of who gets what. Intuitively, one might think the selfish person would win because they are tough and won't let others take advantage of them. Or perhaps the efficient person would win because they create the most wealth overall. The study found that neither of these simple strategies is the ultimate winner on its own. Pure selfishness fails because selfish people often get stuck in bad situations where they could have done better if they had cooperated, but they are too suspicious to try. Pure efficiency fails because efficient people are too nice; they will accept a bad deal just to keep the total reward high, making them easy targets for exploitation by others.
Instead, the study identified two more complex, hybrid types of personalities that dominate in this world of partner choice. The first is the "morally constrained selfish" type. Imagine a person who is fundamentally selfish but has a strict rule: they will only cooperate if the deal is fair and efficient for everyone involved. If they find themselves in a partnership that is wasteful or unfair, they will leave and find a new partner who agrees to a better arrangement. This type of person is smart enough to demand a good deal but disciplined enough to ensure the partnership works well. Because they can find each other and agree to split the rewards fairly, they protect themselves from being exploited while still enjoying the benefits of cooperation.
The second winning type is the "parochial efficient" person. This is someone who cares deeply about the group's success, but only if the other person is "one of their own." They have a strong preference to match with people who share their values. Once they find someone like themselves, they work together to create the maximum possible reward for the pair. This creates a powerful sorting effect: these people naturally flock together, forming tight-knit groups where they cooperate perfectly. Because they refuse to interact with outsiders who might try to take advantage of them, they avoid exploitation entirely. Their group-level selfishness acts as a shield, allowing them to reap the full benefits of their cooperation without fear.
The story changes, however, when the researchers added a layer of uncertainty. In the real world, we cannot always see inside someone's mind to know if they are selfish, efficient, or something else. We only see their outward behavior or labels. When the researchers simulated a world where people cannot directly observe each other's true nature, the "morally constrained selfish" type lost its advantage. Without being able to see who is who, these individuals cannot be sure that a potential new partner is actually like them. They might try to leave a bad deal, only to find that the new partner is actually a different type who will take advantage of them. The risk of being tricked becomes too high, so they stay in bad partnerships, and their evolutionary edge disappears.
In contrast, the "parochial efficient" type remains a winner even when information is incomplete. Because their primary drive is to find someone just like them, they are willing to leave almost any situation that doesn't feel right. If they are matched with someone who doesn't share their values, they derive no benefit from the interaction, so they have a strong incentive to keep looking. This drive to sort themselves out allows them to form perfect groups of their own kind, even without being able to read minds. Once they are together, they cooperate efficiently and thrive. The study shows that this specific combination of caring for the group's success and being picky about who you let into your circle is a robust strategy for survival.
The findings suggest that the evolution of human morality and social behavior is not a simple tug-of-war between greed and generosity. It is a sophisticated dance of selection and coordination. The most successful strategies are those that combine a desire for personal gain with a mechanism to ensure fairness, or a desire for group success with a strict filter on who gets to join the group. The researchers conclude that in a world where we can choose our partners, the winners are not the ones who are simply the nicest or the meanest, but those who are smart enough to find the right people and strong enough to stick with them. This helps explain why we see such complex social behaviors in nature and human society, where cooperation often thrives not because everyone is selfless, but because people have evolved ways to find and protect their own kind.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.