Social preferences are the part of your decision-making that cares about what happens to other people, not just yourself. In economics and behavioral science, the term captures a broad family of motivations: fairness, reciprocity, altruism, spite, and inequality aversion, among others. A comprehensive review of the research literature concludes that a large majority of people exhibit some form of social preference, while purely self-interested individuals are actually a minority.1American Economic Association / Journal of Economic Literature. Social Preferences: Fundamental Characteristics and Economic Consequences That finding upends the classical economic assumption that people are rational self-maximizers, and it has real consequences for how we design workplaces, public policy, and even artificial intelligence.
What Social Preferences Actually Are
When economists talk about social preferences, they mean any situation where your choices are influenced by the outcomes other people experience. If you split a windfall evenly with a stranger when you could keep it all, that is a social preference at work. If you reject a lopsided offer even though rejecting it leaves you with nothing, that is also a social preference, specifically inequality aversion. The concept is intentionally broad because human social motivation turns out to be messy and multidirectional.
Researchers typically break social preferences into a few major flavors:
- Distributional preferences: You care about how resources are divided. This includes a dislike of inequality, whether you are on the losing end or the winning end.
- Reciprocity: You reward kind behavior and punish unkind behavior, even when doing so costs you something.
- Altruism: You value other people’s well-being directly, sometimes for its own sake, sometimes because giving feels good (a motivation called “warm glow”).
- Spite or competitive preferences: You are willing to pay a cost to reduce someone else’s payoff, especially in competitive or intergroup settings.
One important nuance is that these preferences interact with each other. In bargaining situations, for instance, a second mover’s behavior can look like reciprocity on the surface but actually be driven by distributional preferences, where the person’s sense of what counts as “fair” creates a kink in how they evaluate payoffs.2Europe PMC. Distributional Preferences, Reciprocity-Like Behavior, and Efficiency in Bilateral Exchange Someone who rejects a stingy offer might not be punishing the offerer so much as refusing to accept an outcome that feels lopsided. The distinction matters because it changes predictions about how people will behave in different economic situations.
How Researchers Measure Social Preferences
You cannot just ask people whether they are fair. Self-reports are unreliable, because people either do not know their own tendencies or overstate their generosity. The standard approach uses experimental games where real money is on the line, so the choices have genuine consequences. These games have become a core toolkit in behavioral science.3Foundations of Human Sociality. Measuring Social Norms and Preferences Using Experimental Games: A Guide for Social Scientists
In the Dictator Game, one person gets a sum of money and decides how much, if anything, to give to an anonymous partner. The partner has no say. A purely self-interested person keeps everything; most people give away roughly a quarter to a third. In the Ultimatum Game, the partner can reject the offer, in which case both sides get nothing. This addition dramatically changes behavior: offers below about 20 to 30 percent of the total tend to get rejected, even though rejection means walking away empty-handed. That willingness to burn money to punish unfairness is one of the most robust findings in the field.
The Social Value Orientation (SVO) measure takes a different approach, asking people to make a series of allocation choices between themselves and an anonymous other. The pattern of choices classifies each person along a spectrum from competitive to individualistic to prosocial. Public goods games add a group dimension: participants decide how much of their endowment to contribute to a shared pool that benefits everyone. Free-riding is tempting but collectively destructive, and the tension between personal and group interest is exactly what makes these games useful for studying cooperation.
Social Preferences Start Surprisingly Early
If you have spent time around toddlers, you know they are not famous for sharing. But the capacity for fairness-based preferences emerges earlier than most people expect. In a study of two- and three-year-olds, children were asked to distribute resources either between two friends or between a friend and themselves. Three-year-olds were significantly more likely to distribute fairly than two-year-olds, and in both age groups children shared more equitably when they were not personally one of the recipients.4PubMed Central. Young Children’s Development of Fairness Preference The study also found an interesting asymmetry: children were more likely to object to unfairness when they were on the losing end than when they had the advantage. In other words, the sensitivity to inequality that drives adult behavior in Ultimatum Games already has roots in early childhood.
By age five, the picture becomes more sophisticated. When five-year-olds found themselves with fewer resources than a third party, they actively tried to involve that person in sharing. But when five-year-olds had more than the third party, they were less inclined to invite comparison and instead chose to share on their own terms.5PubMed. Preschool children involve a third party in a dyadic sharing situation based on fairness This is already a strategic use of fairness norms: children invoke the principle of equal division when it helps them and quietly set it aside when it does not. That strategic element is not a corruption of genuine social preference; it is part of how social preferences work in the real world, blending sincere concern for fairness with self-interest.
What the Brain Does During Fairness Judgments
Social preferences are not just cultural habits layered on top of a selfish brain. Neuroimaging research shows that experiencing inequality activates core motivational and emotional circuits. When people see that they have more than someone else (advantageous inequality) or less (disadvantageous inequality), both scenarios activate the putamen, orbitofrontal cortex, and insula, brain regions tied to motivation and reward processing.6PubMed Central. Overlapping and distinct representations of advantageous and disadvantageous inequality
But the two types of inequality are not neurologically identical. Having less than someone else engages emotion- and conflict-related areas, including the amygdala and the dorsal anterior cingulate cortex, which lights up when people experience distress or need to resolve competing impulses. Having more than someone else, on the other hand, engages brain regions associated with mentalizing and social cognition, areas you use when you think about what another person is feeling.7PubMed Central. Distinguishing neural correlates of context-dependent advantageous- and disadvantageous-inequity aversion This suggests that feeling guilty about having too much and feeling angry about having too little are psychologically distinct experiences, even though both push behavior in a similar direction (toward equality).
The hormone oxytocin also plays a modulatory role. Often simplified in popular media as the “love hormone” or “trust hormone,” oxytocin’s actual relationship to prosocial behavior is more complicated. A review of its neurobiology emphasizes that oxytocin is deeply embedded in the architecture of the social brain and modulates prosocial behavior, but it does so in context-dependent ways rather than simply making people nicer.8PubMed Central. Oxytocin and the Neurobiology of Prosocial Behavior In some situations oxytocin increases trust and generosity; in others it increases favoritism toward one’s own group. The neurobiological picture, in short, mirrors the behavioral picture: social preferences are real, biologically grounded, and far from simple.
Evolutionary Roots of Caring About Others
A persistent puzzle in evolutionary biology is why altruism exists at all. If natural selection rewards individuals who maximize their own survival and reproduction, helping others at a cost to yourself should be selected against. The classic answer involves kinship: you help relatives because they share your genes. But this is not the full story. A theoretical framework demonstrates that the most fundamental requirement for altruism to evolve is not genetic relatedness but assortment, the tendency for cooperative individuals to end up interacting with others who also cooperate.9PubMed Central. A simple and general explanation for the evolution of altruism Even extreme forms of self-sacrifice can theoretically evolve without genetic similarity, as long as helpful individuals cluster together often enough for the benefits to flow back to them or others like them.
This is relevant to understanding social preferences because it means the human tendency to care about fairness, to punish free-riders, and to cooperate with non-relatives is not some quirky cultural invention. It has deep evolutionary plausibility. The emotional responses that drive social preferences, the anger at unfairness, the guilt about unearned advantage, the satisfaction of cooperating, are likely the psychological mechanisms that evolution shaped to solve cooperation problems in our species’ long history of group living.
The Dark Side of Social Preferences
Social preferences are not always prosocial in the way we might hope. One of the most important and counterintuitive findings in this field is that the same psychological machinery that makes people generous to their own group often makes them hostile to outsiders. This is parochialism: a preference for favoring members of your own group, even at a cost. Experiments with indigenous groups in Papua New Guinea found that people who punished norm violators protected ingroup victims much more vigorously than outgroup victims, regardless of who had actually violated the norm.10Nature. Parochial altruism in humans
Research on intergroup conflict has attempted to disentangle the prosocial and parochial components. Using measures of social value orientation toward both ingroup and outgroup members, one study found that participants who were more prosocial toward their ingroup and simultaneously less generous toward outgroup members were the most inclined to engage in costly intergroup conflict.11PubMed Central. Untangling altruism and parochialism in human intergroup conflict Ingroup altruism and outgroup hostility appear to be separate psychological dimensions that combine to predict behavior. A person can be generous toward insiders without being hostile to outsiders, or hostile to outsiders without being especially generous to insiders. But when both tendencies are strong, the result is the willingness to sacrifice personal resources to harm the outgroup, the essence of intergroup conflict.
Testosterone modulates this dynamic, though its role is more nuanced than the popular image of “aggression hormone” suggests. In the presence of intergroup competition, testosterone is associated with increased parochial altruism, specifically a greater willingness to punish outgroup members on behalf of one’s ingroup.12PLoS ONE. Does Competition Really Bring Out the Worst? Testosterone, Social Distance and Inter-Male Competition Shape Parochial Altruism in Human Males
Antisocial Punishment and Why Cooperation Is Fragile
The ability to punish free-riders is widely considered one of the pillars of human cooperation. If shirkers face consequences, groups can sustain high levels of contribution to shared resources. But there is a complication that gets much less attention: antisocial punishment, where non-cooperators spend their own resources to punish cooperators. This sounds irrational, but it is observed repeatedly in experiments. Theoretical models show that when the full range of punishment strategies is allowed, including the ability to punish cooperators, the pro-cooperation effect of punishment largely disappears.13Nature Communications. The Evolution of Antisocial Punishment in Optional Public Goods Games In these models, punishment functions less as an altruistic enforcement tool and more as a self-interested weapon for protecting oneself against competitors.
Separately, modeling work confirms that when antisocial punishment is introduced into spatially structured populations, costly punishment no longer reliably promotes cooperation.14PubMed Central. Anti-social punishment can prevent the co-evolution of punishment and cooperation This is a sobering result. Many real-world institutions, from workplace norms to criminal justice systems, are built on the assumption that the ability to punish bad behavior is inherently prosocial. Antisocial punishment reminds us that punishment mechanisms can be co-opted: by bullies suppressing whistleblowers, by cartels punishing price-cutters, or by corrupt officials targeting reformers. Understanding that social preferences include a genuine capacity for spite, not just generosity, is essential for designing institutions that actually work.
Social Preferences at Work
One of the most practically important questions about social preferences is whether they affect economic productivity. Three field experiments tested how workers’ social preferences toward their employer influenced their effort. Workers exerted effort even when they had no private financial incentive to do so, consistent with some form of prosocial motivation. But their effort did not respond to how much the employer benefited from their work, which suggests the motivation was “warm glow,” the personal satisfaction of being helpful, rather than genuine altruism toward the employer.15American Economic Review. Estimating Social Preferences and Gift Exchange at Work
The experiments also tested gift exchange, the idea that an unexpected bonus or raise creates a sense of obligation that makes workers more productive. The gifts had no measurable effect on productivity itself, but workers did put in extra hours. The distinction matters: if you are an employer hoping that a surprise bonus will make people work faster, the evidence here says it probably will not. But it may make them willing to show up for additional shifts or stay late. Extra work and harder work are not the same thing, and social preferences operate differently on each.
Are Social Preferences Stable Over Time?
A reasonable worry is that what looks like a social preference in a laboratory game might just be a mood or a context effect. If your fairness behavior fluctuates wildly from week to week, it is hard to call it a preference in any meaningful sense. A study tracked the same individuals’ contributions across four separate public-goods-style events spanning six years. Despite large differences in overall contribution levels between events, the correlations between the same person’s behavior across events were substantial and, in most cases, statistically significant.16Elsevier. Social preferences are stable over long periods of time Someone who was relatively generous in 2005 was still relatively generous in 2011. The correlations were not perfect, around 0.2 to 0.4, which means context and circumstances still matter. But the person-to-person ranking holds up over years and across different types of contribution decisions. Social preferences look more like a personality trait than a situational response.
When Being Watched Backfires
A natural policy instinct is that making behavior more visible should encourage generosity. If people know their choices are observed, they should want to look good. The reality is more complicated. Research on observability and social image found that being watched alone had very little positive effect on giving behavior and sometimes backfired, leading some people to give less or even take from the other party.17Elsevier. Observability and social image: On the robustness and fragility of reciprocity The mechanism appears to involve social comparison: when people can see what others are doing, some are inspired to give more, but others react negatively to the comparison and become less generous. Inequality aversion, the same preference that motivates fairness in other settings, can drive the backlash. If you see that someone else gave less than you, you may feel exploited and reduce your own contribution.
The study found that making social norms more explicit, rather than just making behavior visible, could counteract the backlash. Transparency without norms is just surveillance, and surveillance without a clear behavioral benchmark can activate competitive or defensive instincts rather than cooperative ones. For anyone designing a charity campaign, a workplace recognition program, or a public health initiative, this finding is worth remembering: visibility alone is not enough, and it can make things worse if people lack a clear reference point for what “good” behavior looks like.
Fairness in Other Primates
One way to assess how deep social preferences run is to look at whether they exist in other species. Research on nonhuman primates has documented several parallels. Capuchin monkeys, for example, reject unequal rewards in experimental setups: a capuchin who has been happily accepting cucumber slices will refuse them if it sees a partner receiving a more desirable grape for the same task. Chimpanzees show more variable results, but prosocial behavior, helping others at some cost, has been documented in multiple primate species under various experimental conditions.18PubMed Central. Justice- and fairness-related behaviors in nonhuman primates These cross-species findings suggest that the roots of social preference predate human culture and language, though human social preferences are far more elaborate, flexible, and norm-governed than anything observed in other primates.
Psychopathy and Social Preferences
You might expect that people who score high on clinical measures of psychopathy would show markedly different social preferences, particularly lower generosity and lower sensitivity to fairness. A pilot study of male prisoners tested this directly, measuring both psychopathy scores and behavior across Dictator Games, Ultimatum Games, SVO, and other economic games. The result was striking: there were no significant correlations between psychopathy scores and any of the social preference measures, not in offers made, not in minimum offers accepted, and not in cooperation rates.19Nature. Social preferences and psychopathy in a sample of male prisoners—a pilot study
This is a small study and should be treated as preliminary. But it challenges the intuitive assumption that social preferences are simply a reflection of empathy or moral concern. The prisoners scoring high on psychopathy presumably have reduced empathy, yet their economic-game behavior was indistinguishable from other prisoners. One possible explanation is that social preferences as measured in these games are partly strategic or habitual rather than purely emotion-driven. A person can learn the “right” move in a sharing game without feeling the emotional pull that motivates fairness in others. Another possibility is that the lab setting captures only a thin slice of social behavior, and the real differences emerge in more complex, higher-stakes, or longer-term interactions where empathy plays a bigger role.
Social Preferences and Artificial Intelligence
An emerging frontier for social preference research is the alignment of AI systems. When machine learning models are trained to follow human preferences, they have to grapple with the fact that human preferences are not uniform. Different people value different outcomes, and a model trained on the majority’s preferences may systematically ignore or harm minority viewpoints. Work on reinforcement learning from human feedback has started incorporating social-preference-like frameworks to address this problem. One approach, designed to maximize the minimum satisfaction across diverse groups, achieved substantial improvements in accuracy for minority groups without reducing performance for majority groups.20arXiv. MaxMin-RLHF: Alignment with Diverse Human Preferences The logic is directly descended from the same distributional preferences that behavioral economists study in humans: how do you allocate a limited resource (in this case, model performance) fairly across people with different needs? AI alignment is, in a real sense, a social preference problem.
This connection runs deeper than just fairness metrics. The same tension between ingroup and outgroup preference that shows up in human experiments emerges in AI systems trained on data that reflects group-specific norms. And the observation that social preferences can include antisocial or spiteful elements is a useful warning for AI designers: human feedback can be motivated by the desire to punish or disadvantage, not just to improve. Building systems that faithfully reflect human social preferences means reckoning with the full spectrum, generosity and spite alike.
Why Societies Invest in Building Prosocial Preferences
If social preferences genuinely affect cooperation, public goods provision, and economic behavior, then societies have an incentive to cultivate them deliberately. This is not a new idea: every culture uses education, socialization, religious instruction, and civic rituals to encourage cooperative dispositions. Theoretical modeling of intergenerational preference transmission suggests that deliberate investment in prosocial preferences can improve outcomes for collective action problems, both as one-time interventions and as sustained institutional practices.21Springer International Publishing. Social creation of pro-social preferences for collective action The finding that social preferences are partially stable over time but also partially malleable creates a window: early socialization and institutional design can shift the distribution of social preferences in a population, not by turning selfish people into saints, but by nudging the large middle group toward more cooperative defaults.
The practical applications range from the design of tax systems and pension contribution defaults to how companies structure team incentives. Knowing that most people are not purely self-interested, but that their prosocial tendencies are sensitive to context, norms, and visibility, changes the design space. Policies that assume universal selfishness miss opportunities; policies that assume universal generosity get exploited. The evidence on social preferences suggests aiming for the realistic middle: most people will cooperate if the system makes cooperation easy, visible, and fair, and if the small minority of free-riders face clear consequences.