Philosophers usually sort the study of morality into three levels: meta-ethics, which asks what moral claims are and whether they can be true; normative ethics, which seeks general standards of right conduct; and applied ethics, which brings those standards to bear on particular issues.
Ethics is a single field, but its questions sit at different distances from the choices we actually make. When someone asks whether a particular act of lying is wrong, they are working at one level. When they ask which general theory best explains why lying tends to be wrong, they have stepped back. When they ask whether the word wrong even reports a fact, they have stepped back further still. These three vantage points correspond to applied ethics, normative ethics, and meta-ethics, and keeping them apart prevents a good deal of confusion in moral argument.[1]
The division is now standard in textbooks and reference works, though the boundaries are porous and many real debates draw on all three at once. The ordering below moves from the most abstract level to the most concrete, since that is how the levels tend to support one another. NoteThe term meta-ethics dates to the early twentieth century, but recognizably meta-ethical questions appear in Plato and in Hume's treatment of moral judgment.
Meta-ethics
Meta-ethics studies the nature, meaning, and status of moral thought and talk. It does not try to settle whether any particular act is good or bad. Instead it asks what we are doing when we call an act good: are we describing a feature of the world, expressing an attitude, or issuing a kind of command. Its questions are semantic, metaphysical, epistemological, and psychological rather than directly practical.[2]
Several long-running debates organize the field. The metaphysical question divides moral realists, who hold that there are moral facts independent of what anyone thinks, from anti-realists, who deny this. A related semantic question divides cognitivism, which treats moral statements as beliefs that can be true or false, from non-cognitivism, which treats them as expressions of feeling or attitude. Expressivism and error theory are prominent anti-realist positions, the first holding that moral talk voices stances rather than reports facts, the second holding that moral claims do report facts but are uniformly false.
Meta-ethics also inherits Hume's observation that one cannot validly derive a conclusion about what ought to be from premises only about what is. This is-ought gap, and the related charge of a naturalistic fallacy pressed by G. E. Moore against defining good in natural terms, set much of the agenda for twentieth-century work.[3] NoteMoore argued that for any proposed natural definition of good, it remains an open question whether things with that property are in fact good.
Normative ethics
Normative ethics takes up the practical task that meta-ethics sets aside: it tries to formulate and defend general standards of right and wrong conduct. Where meta-ethics asks what goodness is, normative ethics asks what is good and what we ought to do. Its output is theories, principles, and criteria that a person could in principle use to evaluate actions and characters.
Three families of theory dominate the literature. Consequentialism holds that the rightness of an act depends on its outcomes; utilitarianism, which directs us to maximize aggregate well-being, is its best-known form. Deontology holds that some acts are required or forbidden regardless of outcome, a view associated with Kant's categorical imperative and its demand that we act only on principles we could will as universal law. Virtue ethics, descending from Aristotle, shifts the question from which acts are right to what a good person is like, treating settled traits of character as the primary subject of evaluation.[4]
Normative ethics seeks to arrive at moral standards that regulate right and wrong conduct, whether by naming the duties we should follow, the consequences we should weigh, or the virtues we should cultivate. a common textbook framing
Applied ethics
Applied ethics uses the resources of the other two levels to address specific, often contested, practical questions. Its subject matter is concrete: the permissibility of abortion or euthanasia, the treatment of animals, the obligations of business, the duties we owe to future generations or to the environment. Sub-disciplines such as bioethics, business ethics, and environmental ethics fall under this heading.
Work in applied ethics rarely proceeds by deducing an answer from a single theory. More often it weighs considerations drawn from competing normative theories, attends closely to the facts of the case, and looks for principles that both sides of a dispute can accept. Because it must yield guidance under disagreement and uncertainty, applied ethics is where moral philosophy most visibly meets law, medicine, and public policy.
The three compared
The cleanest way to hold the levels apart is by the question each asks. Meta-ethics asks about the status of morality; normative ethics asks about the content of morality; applied ethics asks about the verdict in a case.
| Level | Central question | Sample issue |
|---|---|---|
| Meta-ethics | What are moral claims, and can they be true? | Is moral realism correct? |
| Normative ethics | What general standards make actions right? | Do consequences alone determine rightness? |
| Applied ethics | What should be done in this case? | Is a given medical practice permissible? |
The levels are commonly described as hierarchical, with meta-ethics providing the foundation, normative ethics the general principles, and applied ethics the case-by-case verdicts. The picture is useful but should not be read too strictly. A person can reason well in applied ethics while holding no fixed meta-ethical view, and progress at the applied level sometimes forces revision higher up. The dependence runs in both directions.
A current case: AI ethics
The ethics of artificial intelligence shows the three levels at work together. At the applied level it asks concrete questions: whether an automated system may decide a loan, who answers for a self-driving car's harm, what counts as fair use of training data. These questions draw on normative theory, since a consequentialist and a deontologist may weigh the same automated decision differently. And they can press meta-ethical questions too, for instance whether a machine could ever have moral status, or whether moral judgments are the kind of thing that could be encoded as a rule a system follows.[5]
The case illustrates the general lesson. The three levels are distinct vantage points on one subject, and a careful argument keeps track of which one it occupies. Mistaking a meta-ethical disagreement for a practical one, or expecting a normative theory to settle a question it was never meant to reach, is among the more common sources of confusion in moral debate.