A common misconception is that people who disagree about a moral question must be following the same value to different conclusions, or else that one side has no values at all. They may instead be attending to different concerns: the harm a decision could cause, for example, or the obligations it creates within a group. Jonathan Haidt and his colleagues developed a multi-domain framework of ethical judgment to investigate such differences.
The framework proposes several recurring ethical concerns, while recognizing that cultures and individuals develop those concerns in different ways. It is a research framework rather than a test of who is good. It also differs from Lawrence Kohlberg’s account of moral development, which asks a different question about changes in reasoning.
Jonathan Haidt Moral Foundations: A Framework for Intuition
The Jonathan Haidt moral foundations model grew out of work by Haidt, Craig Joseph, and other researchers interested in the variety of concerns people treat as moral. Its central proposal is that moral judgment draws on several partially distinct sensitivities. These are shaped by social learning and cultural practices as people encounter particular situations.
An intuition, in this context, is a relatively quick evaluative response: a person might feel that a betrayal is wrong before producing an explanation. Deliberation can follow, and reasons can be exchanged or revised. The framework does not require every judgment to be instantaneous or deny that reflection matters; it suggests that conscious reasoning is not the only process worth studying.
Jonathan Haidt’s moral foundations approach supplies categories that researchers can turn into questions and test against responses. Calling a concern a moral foundation, however, is a theoretical proposal, not proof that it corresponds to one isolated brain circuit or an unchangeable instinct. Researchers continue to examine how best to define and measure these categories.
Five Concerns in the Early Model
The early account commonly described five domains. They are best understood as broad families of concerns, not boxes that can perfectly classify every decision.
Care and harm concern suffering, vulnerability, and protection. A bystander who helps an injured cyclist may be responding to the rider’s immediate need. The same concern can also inform disagreements over how to distribute limited medical resources.
Fairness and cheating concern how people are treated and whether they receive what is due. One person may emphasize equal treatment; another may focus on proportional rewards for contributions. Both can appeal to fairness while disagreeing about the rule that makes an outcome fair.
Loyalty and betrayal involve obligations to a group and reactions to those who undermine it. A teammate who stays to help after a difficult loss may express loyalty. Yet loyalty can also conflict with other concerns, as when reporting misconduct would protect outsiders but damage a colleague’s standing.
Authority and subversion involve respect for roles, legitimate leadership, and social order. A patient might follow a clinician’s instructions because of professional expertise; a citizen might respect a court’s authority while still questioning a particular ruling. The category describes a concern with authority, not a requirement to obey every command.
Sanctity and degradation involve ideas about purity, sacredness, or conduct regarded as defiling. Religious observances offer some examples, but the relevant meanings depend on the community and context. Disgust can accompany such judgments, although a feeling of disgust alone does not settle whether an act is wrong.
These categories often overlap. A whistleblower, for instance, may feel loyalty to coworkers, concern for people harmed by a policy, and respect for the institution’s stated rules. The model offers a vocabulary for separating those considerations; it cannot decide how they should be weighed in the case.
Why Liberty Was Added
Later work proposed liberty and oppression as another domain, focused on resistance to domination and coercion. A tenant challenging an abusive landlord and a resident opposing an intrusive rule may both appeal to freedom while disagreeing about which restrictions are justified. Liberty can overlap with fairness and care, especially when unequal power exposes someone to harm.
The addition illustrates a larger point about Jonathan Haidt’s foundational model of morality: its inventory has changed as researchers have refined their hypotheses and measures. A 2012 study of libertarian morality examined liberty-related concerns alongside other measures. A proposed category earns its place through evidence and debate, rather than by being fixed permanently in the original list.
How Researchers Study Differences
Researchers often use questionnaires that ask people to rate the relevance of particular considerations to moral decisions or their agreement with statements. A set of items is intended to capture each proposed concern. Scores then allow investigators to look for patterns across groups and to test whether a measure relates to other attitudes or behavior.
In a 2009 paper, Jesse Graham, Haidt, and Brian Nosek used several approaches, including judgments and analyses of religious sermons, to examine relationships between political identification and moral concerns. In their samples, self-described liberals and conservatives showed differences in emphasis. These were group-level patterns, not a way to infer a stranger’s priorities from a party label.
That distinction matters because distributions overlap and individuals do not always fit a group average. In a 2012 study, researchers found that participants’ impressions of typical liberals and conservatives exaggerated the size of differences in questionnaire responses. A useful account of variation should make stereotypes less tempting, not give them scientific decoration.
Questionnaires also have limits. Responses depend on how items are worded and on what a participant understands by terms such as respect, fairness, or purity. A pattern found in one sample may not appear in the same form in another language or political setting. Researchers must examine whether a measure functions comparably across groups before treating differences in scores as direct comparisons of underlying values.
Different From Kohlberg’s Stages
Kohlberg’s developmental theory concerned changes in the structure of moral reasoning. His stage framework examined how people justify judgments, including whether they appeal to punishment, social rules, or broader principles. It proposed an ordered developmental progression in forms of reasoning.
By contrast, Jonathan Haidt’s foundational model of morality, developed with collaborators, asks which kinds of concern figure in judgments and how their salience varies. An appeal to loyalty is not a developmental stage, and a questionnaire score on care does not identify a person’s position in Kohlberg’s sequence. The two approaches address different questions, even where both examine moral life.
Neither description resolves a moral dispute by itself. Understanding why someone values loyalty does not establish whether loyalty should outweigh the duty to prevent harm. That further question requires reasons and evidence about the circumstances, not just a classification of the participants’ responses.
What the Model Can and Cannot Settle
The categories help generate hypotheses, but their number, boundaries, origins, and measurement remain subjects of research. A questionnaire may group responses in a way that reflects the questions selected as well as real differences among respondents. Researchers can compare alternative models and test their predictions rather than assume the categories are nature’s final list.
The framework also distinguishes a description from a justification. Someone may sincerely experience a custom as sacred, while another person may offer strong reasons to change it. Explaining both reactions does not make the custom fair or unfair. Moral evaluation still involves examining consequences, rights, evidence, and the reasons people give one another.
Consider a worker deciding whether to report a dangerous practice. One colleague stresses loyalty to a team that could suffer reputational damage; another stresses care for people exposed to the danger. The two concerns can both be intelligible, even if the facts ultimately support reporting. The framework helps name the difference in emphasis; it is not a verdict on either person’s moral worth.
