What’s Wrong with Moral Foundations Theory, and How to get Moral Psychology Right

Once the exclusive preserve of philosophy and theology, the study of morality has now become a thriving interdisciplinary endeavor, encompassing research in evolutionary theory, genetics, biology, animal behavior, psychology, and anthropology. The emerging consensus is that there is nothing mysterious about morality; it is merely a collection of biological and cultural traits that promote cooperation.

Best known among these accounts is Jonathan Haidt’s Moral Foundations Theory (MFT). According to MFT: “Moral systems are interlocking sets of values, virtues, norms, practices, identities, institutions, technologies, and evolved psychological mechanisms that work together to suppress or regulate selfishness and make cooperative social life possible.” And MFT proceeds to argue that, because humans face multiple social problems, they have multiple moral values—they rely on multiple “foundations” when making moral decisions. These foundations include: Care, Fairness, Loyalty, Authority, and Purity.

  • Care: “The suffering of others, including virtues of caring and compassion.”
  • Fairness: “Unfair treatment, cheating, and more abstract notions of justice and rights.”
  • Loyalty: The “obligations of group membership” including “self-sacrifice, and vigilance against betrayal.”
  • Authority: “Social order and the obligations of hierarchical relationships, such as obedience, respect, and the fulfillment of role-based duties.”
  • Purity: “Physical and spiritual contagion, including virtues of chastity, wholesomeness, and control of desires.”

These moral foundations have been operationalized, and measured, by the Moral Foundations Questionnaire (MFQ; you can complete it here).

MFT and the questionnaire have had an enormous impact on moral psychology. The central papers have been cited hundreds of times. And there is now a huge literature applying MFT to bioethics, charity, environmentalism, psychopathy, religion, and especially politics. However, MFT has some serious problems, both theoretical and empirical.

Moral Foundations Theory has had an enormous impact on moral psychology. However, the theory has some serious problems, both theoretical and empirical.

The main theoretical problem is that MFT’s list of foundations is not based on any particular theory of cooperation, or on any explicit theory at all. Indeed, Haidt, has explicitly argued against taking what he calls an “a priori or principled” approach to moral psychology, and instead has advocated taking an “ad hoc” approach. The shortcomings of this ad hoc approach, however, are all too plain to see.

First, MFT’s list of foundations has critical omissions. Despite claiming to be an evolutionary-cooperative account of morality, MFT fails to include the four most well-established types of evolved cooperation: kin altruism, reciprocal altruism, competitive altruism, and respect for prior possession.

  • Kin altruism has no dedicated foundation in MFT. Although MFT argues that Care originally motivated investment in offspring, it is now applied to nonkin; and MFT treats “family” as just another type of “group.” The questionnaire (MFQ) does have two items pertaining to family, but they appear under Fairness and Loyalty, not Care.
  • Reciprocal altruism has no dedicated foundation in MFT. Instead, MFT conflates reciprocity—a solution to iterated prisoners’ dilemmas—with fairness—a solution to bargaining problems. And the MFQ has no items pertaining to reciprocity.
  • Competitive altruism—that is, costly signals of status, such as bravery or generosity—has no dedicated foundation in MFT, and no items in the MFQ.
  • Respect for prior possession—that is, property rights and the prohibition of theft—has no dedicated foundation in MFT. The MFQ’s only mention of property occurs in an item about inheritance, under Fairness.

Second, in addition to these omissions, MFT includes two foundations that are not distinct types of evolved cooperation: Care and Purity.

  • Care—like “altruism” or “benevolence”—is a generic category, not a specific type of cooperation. It doesn’t distinguish between the various distinct types of cooperation—kin altruism, mutualism, reciprocal altruism, competitive altruism and their corresponding psychological mechanisms—all of which involve caring for different people (including family, friends, strangers) for different reasons.
  • Purity is supposed to stem from the need to avoid “people w/ diseases, parasites [&] waste products.” But “avoiding pathogens” is not itself a cooperative problem, any more than, say, “avoiding predators.” And, indeed, MFT offers no connection between purity and cooperation. On the contrary, Purity is described as an “odd corner” of morality because it is not “concerned with how we treat other people.” Hence, categorizing Purity as a moral foundation is anomalous.

Thus, MFT’s theory-free approach results in egregious errors of omission, conflation, and commission. It misses some candidate moral domains, combines others, and includes noncooperative domains. Most egregiously, the lack of theory means that MFT cannot rectify these errors; it cannot make principled predictions about what (other) foundations there might be, thus it cannot make progress toward a cumulative science of morality.

MFT also has empirical problems. The main problem is that MFT’s five-factor model of morality has not been well supported by studies using the MFQ. Some of the original studies, as well as replications in Italy, New Zealand, Korea, Sweden, and Turkey, and also a 27 country study using the short-form MFQ, have found that MFT’s five-factor model falls short of the conventionally acceptable degree of model fit (CFIs < 0.90). These studies typically find that a two-factor model—“Care-Fairness” and “Loyalty-Authority-Purity”—is a better fit. And so despite MFT promising five moral domains, the MFQ typically delivers only two. The MFQ does not distinguish domains dedicated to Fairness, Loyalty, or Authority; nor does it establish that Care and Purity are distinct moral domains. Simply put, it does not establish that there are five moral foundations. Other research has taken issue with specific foundations, especially Purity and the link between disgust and morality; but that’s a story for another time.

To their credit, proponents of MFT acknowledge these problems. They accept that the original list of foundations was “arbitrary,” based on a limited review of only “five books and articles,” and never intended to be “exhaustive.” And they have positively encouraged research that could “demonstrate the existence of an additional foundation, or show that any of the current five foundations should be merged or eliminated.”

And so that is what my colleagues and I have done. But we have not done so by making yet more “ad hoc” suggestions. We have gone back to first principles, to the theory that can provide a rigorous, systematic foundation for a cooperative theory of morality—the mathematics of cooperation, the theory of non-zero-sum games. We call this approach Morality-as-Cooperation (MAC).

According to MAC, morality consists of a collection of biological and cultural solutions to the problems of cooperation recurrent in human social life. For 50 million years humans and their ancestors have lived in social groups. During this time, they faced a range of different problems of cooperation, and they evolved and invented a range of different solutions to them. Together, these biological and cultural mechanisms provide the motivation for cooperative behavior; and they provide the criteria by which we evaluate the behavior of others. And, according to MAC, it is precisely this collection of cooperative traits—these instincts, intuitions, and institutions—that constitute human morality.

For 50 million years humans and their ancestors have lived in social groups. During this time, they faced a range of different problems of cooperation, and they evolved and invented a range of different solutions to them.

Which problems of cooperation do humans face? And how are they solved? That’s where game theory comes in. Game theory makes a principled distinction between zero-sum and non-zero-sum games. Zero-sum games are competitive interactions that have a winner and a loser; one’s gain is another’s loss. Non-zero-sum games are cooperative interactions that can have two winners; they are win-win situations. Game theory also distinguishes between different types of non-zero-sum games and the strategies used to play them. Thus, it delineates mathematically distinct types of cooperation.

A review of this literature suggests that there are (at least) seven well established types of cooperation: (1) the allocation of resources to kin; (2) coordination to mutual advantage; (3) social exchange; and conflict resolution through contests featuring (4) hawkish displays of dominance and (5) dove-ish displays of submission; (6) division of disputed resources; and (7) recognition of prior possession.

In my research, I have shown how each of these types of cooperation can be used to identify and explain a distinct type of morality.

(1) Kin selection explains why we feel a special duty of care for our families, and why we abhor incest. (2) Mutualism explains why we form groups and coalitions (there is strength and safety in numbers), and hence why we value unity, solidarity, and loyalty. (3) Social exchange explains why we trust others, reciprocate favors, feel gratitude and guilt, make amends, and forgive. And conflict resolution explains why we (4) engage in costly displays of prowess such as bravery and generosity, why we (5) express humility and defer to our superiors, why we (6) divide disputed resources fairly and equitably, and why we (7) respect others’ property and refrain from stealing.

Our research has shown that examples of these seven types of cooperative behavior—help your family, help your group, return favors, be brave, defer to your superiors, be fair, and respect others’ property—are considered morally good all around the world and are probably cross-cultural moral universals.

Our research has shown that examples of these seven types of cooperative behavior—help your family, help your group, return favors, be brave, defer to your superiors, be fair, and respect others’ property—are considered morally good all around the world.

And we have used MAC’s framework to develop a new measure of moral values that promises, and delivers, seven moral domains: (1) Family, (2) Group, (3) Reciprocity, (4) Heroism, (5) Deference, (6) Fairness, and (7) Property. This new Morality-as-Cooperation Questionnaire (MAC-Q) introduces the four moral domains that were missing from MFT: Family, Reciprocity, Heroism, Property. And unlike the MFQ, it distinguishes Family from Group (Loyalty), Group (Loyalty) from Deference (Authority), and Reciprocity from Fairness.

So this principled approach to morality, grounded firmly in the underlying logic of cooperation, outperforms an unprincipled approach. MAC explains more types of morality than MFT. It can generate novel principled predictions about morality’s content and structure—predictions that have thus far been supported by psychological and anthropological research. And it leads to a more comprehensive and reliable measure of moral values.

Equipped with this new map of the moral landscape, we can now examine familiar ground in greater detail and survey previously unexplored territory. We can take a fresh look at the genetic basis, and the psychological architecture, of morality. We can reassess the relationship between morals and politics. And we can investigate how and why moral values vary around the world. Above all, by using a theory to generate new testable predictions, we can pave the way for a genuine science of morality.