If there’s one thing nearly all people can agree on, it’s that some actions are morally right and others are wrong.
But which actions count as right or wrong? That gets a bit more complicated. In some societies, polygamy is normal and proper. In others, taking a second wife can get you imprisoned. Some societies value individual rights and autonomy, while others emphasize collective obligations and hierarchy. In the face of such dazzling differences, how are we supposed to develop a coherent—much less scientific—understanding of “morality?” In a recent paper, a team of researchers at Oxford tried to answer this question by arguing that morality is always about cooperation—and they crunched data from wildly different societies all over the world to make their case.
Oliver Scott Curry, the paper’s head author, is an anthropologist who studies the evolutionary basis for moral sentiments and values. In a post at the online magazine Behavioral Scientist, Curry argued that previous attempts to use evolutionary psychology to understand how humans came to evolve a sense of morality have fallen short. In particular, Curry took aim at Moral Foundations Theory, the well-known framework for studying moral psychology developed by Jonathan Haidt (now a professor at New York University). Moral Foundations Theory posits that moral intuitions come in five different “flavors,” or basic types: care, fairness, in-group loyalty, authority, and purity. Many studies have found that self-described conservatives tend to value all five of these foundations equally, while self-identified liberals value care and fairness (the “individualizing” foundations more highly than the other three (the “binding” or “groupish” foundations).
Critiquing Moral Foundations Theory
Curry isn’t alone in critiquing Moral Foundations Theory. Several studies have found that, statistically speaking, the five moral foundations don’t actually seem to behave like five separate and distinct factors, but rather only two: the “individualizing” (liberal) and the “binding” (conservative) foundations. Curry also pointed out that Haidt and his collaborator Jesse Graham built the MFQ in an ad-hoc fashion, using a fairly limited reading of the literature to guess the most likely candidates for cross-culturally universal moral values.
In contrast, Curry and his colleagues at Oxford tried to develop a cross-culturally universal model of human morality from evolutionary first principles. Game theory helped them extrapolate fundamental moral values from what evolutionary biology has uncovered about conflict management and social living. This method produced seven basic moral domains: familial obligations, group loyalty, reciprocity, bravery, respect, fairness, and property rights.
Each of these moral domains, in turn, rides atop even more basic strategies for problem-solving among social animals. For example, familial obligations have their animal analogue in kin selection or inclusive fitness, while group loyalty stems from mutualism. Ultimately, all seven moral domains are aimed at enabling cooperation, which Curry and his colleagues define as non-zero-sum games: interactions in which my gain isn’t necessarily your loss, and vice-versa. For example, cooperation during a buffalo hunt is non-zero-sum. If we both sacrifice our time and energy, we’re more likely to secure a buffalo, and then we both get meat. The paper’s claim is that moral sentiments are primarily aimed at stabilizing this kind of cooperative relationship.
Thus, Curry’s morality-as-cooperation theory depicts human morality as fully continuous with social cooperation in other animals, a point is made more clearly in a separate book chapter. It insists on tracking the evolutionary precedents for our own moral intuitions, thereby grounding its findings solidly in a fully naturalistic model of human behavior. By assuming that humans face the same cooperative dilemmas other animals confront—thereby tethering the proposed moral domains mechanistically to actual biological problems—morality-as-cooperation theory hopes to offer a significant advance over Moral Foundations.
In paper earlier this year in Current Anthropology, Curry and several co-authors presented data from a detailed analysis of 60 different societies to test their prediction that morality is primarily about cooperation. The societies were chosen for being geographically representative of the entire globe, and because they largely aren’t genealogically related to one another—that is, they don’t descend from common ethnolinguistic ancestors. This non-relatedness helped Curry and his colleagues avoid “Galton’s problem,” which occurs when anthropologists try to test cross-cultural hypotheses using data about cultures that are related to each other.
(For example, say you’re trying to test whether “big gods” are more common in monarchical societies. If most of the societies in your sample are descended from a society that had both big gods and kings, then you can’t really test whether one variable affected the other. All you can say is that all your societies have inherited some common features. Sampling linguistically unrelated societies is one way, then, to make sure that you’re testing your hypotheses in a credible way.)
The authors coded ethnographic findings from more than 600 different ethnographic documents for mentions of (1) property possession; (2) resource sharing or division, (3) submissive and (4) dominant strategies for conflict resolution, (5) reciprocity and vengeance, (6) in-group helping, and (7) kin helping. They then coded for whether people in the relevant societies considered these behaviors morally good, bad, or neutral. For example, let’s say an anthropologist studying a tribe in South America recorded an incident in which a man sacrificed his time and comfort to care for his sister’s sick children, and other members of the tribe applauded this action as a noble deed. The study’s authors would code this (fictional) passage as indicating that kin helping was both present and considered morally good.
Using these methods, Curry and his co-authors found nearly 1,000 mentions of the seven moral domains, and in all cases except one the recorded moral valence was positive. There was no recorded mention of any people, anywhere or at any time, who thought that helping your family or being brave in combat were bad things. Nobody considered respecting superiors, reciprocating when others cooperated, or helping one’s in-group to be bad, either.
The single example of a moral domain with a negative valence was a reference to the Chuuk people of the South Pacific, who believe that stealing from others is sometimes admirable because it shows one’s dominance and fearlessness. Curry and his colleagues explained that this single exception actually showed that other moral domains—specifically, bravery and dominance—were more highly valued among the Chuuk than property rights. Thus, they found compelling evidence that the seven posited moral domains exist everywhere, and that people consider them morally good.
Of course, not all societies emphasized all seven domains. Interestingly, the moral domain that was most common across cultures was property rights, which was mentioned in nearly 90% of societies. Fair division of property was the least common, recorded as present in less than 20% of societies. The other five domains were fairly evenly spread out, recorded in between 50% to 75% of societies.
It’s interesting to wonder whether the overrepresentation of property rights in Curry et al’s findings genuinely reflects a strong emphasis on private property across the world’s cultures, or whether English-speaking ethnographers were simply highly interested in, and so asked more questions about, tribal people’s conceptions of property.* It’s hard to know for sure, but the possibility does illustrate the limitations of relying on ethnographic data for cross-cultural research—and the ever-present danger of missing data.
To be fair, Curry and his colleagues are careful to point out that absence of evidence isn’t the same as evidence of absence. That is, just because there’s no ethnographic record of respect for bravery in a given society doesn’t mean that people in that society don’t value bravery. It just means that no ethnographer wrote down anything about it. These problems aside, the analyses in the Current Anthropology paper do demonstrate that the seven moral domains posited by Curry and his colleagues are almost universally seen as positive.
Uniquely human morality?
While developing the morality-as-cooperation theory, Curry and his colleagues restricted potential moral domains to those that have evolutionary analogues among other animals. While this decision does ground their theory in a solid ethological and evolutionary basis, it may also eliminate a priori any specifically human morality.
After all, humans—unlike other animals—have language and symbolic thought. These traits make it possible not only to form coalitions and live in groups, as other animals do, but to create symbolic identities, such as American citizenship or membership in the “porcupine clan.” (In many foraging societies, tribes are divided up into smaller clans that boast symbolic names, often referring to local animals or other entities.) Symbolic groups, in turn, often impose certain taboos or restrictions, such as never eating your clan’s totem animal, or fasting during certain periods of the year. These expectations are almost always considered morally important.
In larger-scale societies, symbolic identities might become wrapped up in different kinds of symbols—such as flags or religious icons—but they still imply taboos. Consider the visceral reaction that many conservative Americans have to flag-burning, for example, or Catholic fasting from meat on Fridays. If grew up Catholic prior to the 1960s and you ate meat on a Friday, you were a “bad” Catholic—and you’d better believe that your devout grandmother meant that in a moral sense.
Religion and morality
These observations point to a domain of human morality that Moral Foundations Theory attempts to grapple with, but which morality-as-cooperation seems to (at least so far) leave mostly untouched: religion. All around the world, religious rituals and behaviors are considered morally important, and failure to participate in them is considered morally wrong.
I’m not just talking about big, world religions. Ritual sacrifices to local spirits, ancestors, and other non-“Big God” deities are probably the default form of religion around the world. For example, if you go out hunting in many small-scale societies, you have to ask permission from the spirits beforehand, which might involve fasting from certain kinds of foods or performing a sacrifice.
Offerings to ancestors in Thailand
Ancestor veneration, too, is extremely common. Across different societies and eras, people pay offerings to their ancestors by leaving bowls of rice in front of home altars, sacrificing choice pieces of hunted meat, or abstaining from sex for certain periods. These actions are almost always considered morally right, and failing to perform them is morally wrong.
So does morality-as-cooperation theory cover the morality of taboos and religious rituals? More directly put, does it cover the moral interests of religion? Well, the Current Anthropology paper’s methods didn’t address taboos or ritual behaviors directly. They focused instead on interpersonal interactions, such as theft, revenge, bravery in combat, and respect for elders. The paper also didn’t address sacred values, or beliefs and norms that people consider so important that they wouldn’t trade or break them for any amount of money.
These oversights—if that’s what they are—might be built into the framework of the theory. Religious rituals, taboos, and sacred value are precisely what other animals don’t have. If we assume from first principles that human morality is just a more complex version of the same strategic instincts that other social animals possess, then a distinctively human morality is out of the question. But if we assume that human sociality might be the product of a nonlinear evolutionary shift—a transition that produced something genuinely novel—then ritual and religion become good candidates for the markers of a specifically human type of morality.
Morality-as-obligation theory, despite its careful mapping of the one-to-one relations between specific evolutionary problems that social animals must solve and their potential moral solutions, doesn’t address this problem—yet. However, Curry and his colleagues explicitly hope to test whether their theory can account for religious behavior and beliefs in the future.
In a commentary to the Current Anthropology paper, psychologist Paul Bloom points out another potential problem: in order to show that morality “always and everywhere” about cooperation, the theory’s proponents have to show that other domains of behavior aren’t moralized. But colloquially, certain moral questions don’t have much to do with zero-sum cooperative games. Not murdering others is a potential example: murder is considered morally wrong in all societies, but it’s not clear what its relationship to cooperation might be. It might just be that sympathy—affective responses to others’ desires and pains, probably rooted in mammalian instincts for caring for offspring—is an important contributor to our moral decision-making, as Moral Foundations Theory predicts.
However, Curry and his colleagues have a thought-provoking rejoinder for this critique: harming or killing others is only considered morally wrong when it harms cooperation within the group. After all, harming an enemy during war isn’t usually considered wrong. Similarly, executions or punishments are often considered morally right, even though they obviously involve inflicting harm on someone. Morality-as-cooperation theory may, then, explain in which contexts harm is considered morally wrong versus morally right.
Finally, Curry et al.’s definition of cooperation as characterized by non-zero-sum interactions seems somewhat idiosyncratic. The game-theoretic definition of a cooperative game is, roughly, one in which decisions are made in coalitions. This type of game can include both zero-sum and non-zero-sum interactions. Moreover, evolutionary biologists define cooperation both in terms of altruism, in which benefits to cooperators are indirect, or mutualism, which offers direct benefits. The morality-as-cooperation model emphasizes mutualism, but it’s not clear—at least to me—why human morality can’t include domains in which the benefits to cooperators are indirect.
In other words, it’s not entirely clear from which fields Curry and his colleagues are drawing their precise definition of cooperation, which might make it difficult for researchers and lay readers across different disciplines to assess their claims.
Despite these potential weak spots, the theory of morality-as-cooperation is an important step forward in moral psychology, because it pushes researchers to think more rigorously about the evolutionary background and specific processes that might give rise to moral sentiments. By tying morality into the evolutionary study of animal behavior, the theory situates moral psychology comfortably amidst the fabric of the broader life sciences, providing many contacts with other disciplines and theories. And maybe most importantly, by suggesting that morality helps people adapt to differing local circumstances and contexts while always focusing on solving the same types of problems, it responds to the age-old question, “Is it nature or nurture?” in the best way possible: “Yes.”
* The answers to these questions are probably yes and yes. That is, English anthropologists probably are more conditioned to think in terms of private property, because Western European culture in general and English society in particular really do emphasize private property more than most cultures. At the same time, Marxian critics of Western culture who believe that private property is a uniquely pernicious, oppressive, Western cultural construct are almost certainly wrong, because the data seems pretty clear that people tend to get angry if others steal from them in the majority of the world’s societies—even if they don’t have all that much property to steal.