Moral claims are based on what?

Before we can argue about what is right, we have to confront a deeper question: what kind of claim is “right” in the first place? Is morality something we discover, construct, feel, project, or invent? Are moral judgments attempts to describe reality, expressions of attitude, rational requirements, or useful fictions? Every argument about justice, rights, cruelty, duty, dignity, law, punishment, or human flourishing rests on hidden answers to those prior questions. Metaethics is the discipline that drags those hidden commitments into the open.

This is why metaethics sits at the foundations of moral philosophy. It does not ask whether lying is wrong or whether utility should be maximized. It asks what it would even mean for lying to be wrong, what sort of fact moral wrongness would be, how such facts could be known, why moral judgments seem to motivate action, and whether moral language tracks reality at all. At this level, the familiar surface of ethics breaks apart into its underlying architecture: metaphysics, semantics, epistemology, psychology, normativity, and practical reason. What looks from a distance like one debate about morality becomes, under analysis, a field of deep disputes about truth, meaning, motivation, and human rationality itself.

This compendium maps a major stretch of that terrain. It begins with the field’s founding questions and the challenge posed by the is-ought gap; moves through the semantic divide between cognitivism and non-cognitivism; examines the Frege-Geach problem and the pressure it places on expressivist views; and surveys the core realism/anti-realism landscape, from Platonic non-naturalism and Cornell naturalism to error theory, constructivism, and quasi-realism. The aim is not merely to list positions but to show their logical architecture: which problems generate which theories, what each view preserves, what each sacrifices, and why the central disputes of metaethics remain among the hardest in philosophy.

Metaethics matters because moral life cannot avoid it. Whenever we say that slavery was not merely disliked but wrong, that human rights are not merely preferred but real, that cruelty deserves condemnation, or that conscience binds even against law and custom, we are already taking a stand on metaethical questions. The field studies the conditions under which those claims are objective, knowable, action-guiding, and more than rhetoric. In that sense, metaethics is not philosophy’s detour away from morality. It is the inquiry into what must be true for morality to have the authority we ordinarily assume it has.

---

The field and its founding questions

Metaethics occupies a distinct level in the standard tripartite division of ethics. Applied ethics asks concrete questions (“Is physician-assisted suicide permissible?”). Normative ethics asks about general principles (“Should we maximize well-being, or follow universal duties?”). Metaethics steps back to a second-order vantage and asks what morality itself is: Do moral facts exist? What do moral terms mean? Can “ought” be derived from “is”? As the Stanford Encyclopedia of Philosophy defines it, metaethics is “the attempt to understand the metaphysical, epistemological, semantic, and psychological presuppositions and commitments of moral thought, talk, and practice.”

A crucial feature of this hierarchy: metaethical positions underdetermine normative theories. G.E. Moore, J.S. Mill, R.M. Hare, and Berkeley all endorsed utilitarianism despite holding radically different metaethical views. One can be a moral realist and a virtue ethicist, or an expressivist and a consequentialist. The levels are logically independent, though certain affinities exist — metaethical naturalists have tended overwhelmingly toward utilitarianism.

The field organizes around three foundational parent questions. The ontological question asks whether moral facts exist and, if so, what kind of entity a moral fact would be. The epistemological question asks how we could ever come to know moral facts — whether through intuition, empirical observation, rational reflection, or some other faculty. The semantic question asks what moral sentences mean: do they express truth-apt beliefs about the world, or something else entirely — attitudes, prescriptions, plans? Some formulations add a fourth dimension — the psychological question of how moral judgment connects to motivation — and a growing subfield addresses the practical question of how agents should orient given moral uncertainty (as developed in William MacAskill, Krister Bykvist, and Toby Ord’s Moral Uncertainty, 2020).

The is-ought gap, fact-value distinction, and the nature of moral claims

Hume’s guillotine

The most famous single passage in metaethics appears in David Hume’s A Treatise of Human Nature (1739–40), Book III (“Of Morals”), Part I (“Of Virtue and Vice in General”), Section I (“Moral Distinctions Not Deriv’d from Reason”), paragraph 27 in the standard Selby-Bigge/Nidditch numbering (T 3.1.1.27). Hume wrote:

“In every system of morality, which I have hitherto met with, I have always remark’d, that the author proceeds for some time in the ordinary way of reasoning, and establishes the being of a God, or makes observations concerning human affairs; when of a sudden I am surpriz’d to find, that instead of the usual copulations of propositions, is, and is not, I meet with no proposition that is not connected with an ought, or an ought not. This change is imperceptible; but is, however, of the last consequence.”

Hume’s observation has two components. First, moral philosophers habitually shift from descriptive premises (using “is”) to evaluative conclusions (using “ought”) without justifying the transition. Second, since “ought” expresses “some new relation or affirmation” entirely different from “is,” it seems “altogether inconceivable” how the former can be deduced from the latter. This is variously called Hume’s Law, Hume’s Guillotine, or the No-Ought-From-Is principle. Interpretive controversy persists: the dominant twentieth-century reading holds Hume is stating an absolute logical prohibition on deriving ought from is, while some scholars argue he is making a narrower point about the need for explanation. What is clear is that Hume’s own positive view was sentimentalist — “morality is more properly felt than judged of” (Book III, Part I, Section II) — making him a forerunner of expressivism.

The fact-value distinction

The fact-value distinction is closely related to but broader than the is-ought gap. Where the is-ought gap is specifically about the logic of inference — whether evaluative conclusions follow from purely descriptive premises — the fact-value distinction is an epistemological and ontological claim that facts and values belong to fundamentally different categories. Hume’s Fork (relations of ideas vs. matters of fact) prefigured it. The logical positivists took it to its extreme: A.J. Ayer argued in Language, Truth and Logic (1936) that since moral judgments are neither analytic truths nor empirically verifiable, they lack propositional content entirely. Max Weber defended fact-value separation in social science methodology. Critics like Hilary Putnam argued for fact/value entanglement; John Searle contended that institutional facts (“Jones promised to pay Smith five dollars”) can generate genuine oughts; Philippa Foot and Alasdair MacIntyre argued that functional concepts (“This is a bad watch”) demonstrate evaluative conclusions following from factual premises.

Moore’s open question argument and the naturalistic fallacy

G.E. Moore’s Principia Ethica (1903) mounted a different but complementary challenge to bridging the fact-value divide. Moore’s open question argument has this structure: for any proposed naturalistic definition of “good” (say, “good = pleasurable”), a competent speaker can always meaningfully ask, “I know it is pleasurable, but is it good?” If “good” simply meant “pleasurable,” this question would be as trivially closed as “I know he’s a bachelor, but is he an unmarried man?” The persistent “openness” of the question reveals that goodness is not analytically equivalent to any natural property. The naturalistic fallacy is Moore’s label for the alleged error of making such identifications — equating goodness with pleasure, evolutionary fitness, desire-satisfaction, or any other natural property. As W.K. Frankena noted in a landmark 1939 Mind article, the naturalistic fallacy is not strictly a logical fallacy but a charge of mistaken metaphysical reduction. The OQA provides the argumentative backing; the naturalistic fallacy is the diagnosis of what goes wrong. Moore himself was a moral realist who held goodness to be a real, simple, unanalyzable, non-natural property known by intuition — sharply distinguishing him from Hume’s sentimentalism.

Moral semantics: cognitivism, non-cognitivism, and the varieties of each

The semantic question — what moral sentences mean and what mental states they express — generates the field’s deepest divide.

The cognitivist thesis

Cognitivism holds that moral sentences express truth-apt beliefs. When someone says “Torture is wrong,” they are asserting a proposition that can be true or false, just as “Snow is white” can be. The state of mind behind a sincere moral utterance is a belief — a cognitive state that aims to represent how the world is. Cognitivism is shared by moral realists (who think some moral beliefs are true) and error theorists (who think all moral beliefs are false but still genuinely truth-apt).

Within cognitivist realism, two major varieties exist. Moral naturalism identifies moral properties with natural properties — the sort of properties that empirical science investigates. Moral non-naturalism holds that moral properties are real but irreducible to natural properties, known through intuition or rational reflection rather than empirical observation. Error theory occupies the cognitivist anti-realist quadrant: moral sentences express genuine propositions, but those propositions are systematically false because the moral facts they presuppose do not exist.

The non-cognitivist thesis

Non-cognitivism comprises two interlocking negative claims. First, semantic nonfactualism: moral sentences do not express propositions with substantive truth conditions. Moral predicates do not denote properties; moral sentences do not predicate properties of their subjects. Second, psychological non-cognitivism: the mental states conventionally expressed by moral utterances are not beliefs but non-cognitive attitudes — desires, approvals, disapprovals, plans, commitments. When someone says “Torture is wrong,” they are not reporting a fact but expressing disapproval, issuing a prescription, or endorsing a norm.

A critical distinction separates non-cognitivism from cognitivist subjectivism. The subjectivist who says “Stealing is wrong means I disapprove of stealing” does express disapproval, but by reporting a belief about her own psychological state — which is itself truth-apt. The non-cognitivist says the speaker expresses disapproval directly, without reporting anything. As the Stanford Encyclopedia puts it: “One can express dislike of something by saying that one dislikes it. But one can also express dislike of something by booing or hissing.”

The evolution of non-cognitivism through four stages

Emotivism was the first systematic non-cognitivist position. A.J. Ayer, in Language, Truth and Logic (1936), argued from logical positivist premises: since moral statements are neither analytic truths nor empirically verifiable, they lack propositional content. “Stealing money is wrong” is semantically equivalent to “Stealing money!!” uttered with a tone indicating disapproval — the so-called “Boo-Hurrah” theory. C.L. Stevenson developed a more sophisticated version in “The Emotive Meaning of Ethical Terms” (1937) and Ethics and Language (1944). His key innovation was the distinction between disagreement in belief and disagreement in attitude: moral disputes are fundamentally clashes of attitude, not factual disagreements. Stevenson introduced the concept of “emotive meaning” — “a tendency of a word, arising through the history of its usage, to produce affective responses in people” — and emphasized the persuasive function of moral language: “When you tell a man he oughtn’t to steal, your object isn’t merely to let him know that people disapprove of stealing. You are attempting, rather, to get him to disapprove of it.” Where Ayer was motivated by saving logical positivism from moral counterexamples, Stevenson was motivated by observations about actual moral practice and the “magnetism” of ethical claims.

Prescriptivism was R.M. Hare’s advance. In The Language of Morals (1952) and Freedom and Reason (1963), Hare argued that moral judgments are fundamentally universal prescriptions — not descriptions and not mere emotional ejaculations, but imperatives of a special kind. Moral sentences carry both descriptive meaning (implying certain non-moral properties) and prescriptive meaning (commending or prescribing action), with the prescriptive component being primary. The defining feature is universalizability: if you call an action wrong, you commit yourself to judging wrong any relevantly similar action by any agent, anywhere, at any time — including hypothetical cases where you occupy a different role. This constraint vindicates moral reasoning as a rational enterprise, where emotivism had struggled to account for it. Hare may have introduced the term “supervenience” into philosophical literature (1952, p. 145), arguing that moral properties supervene on natural properties because universal prescriptions demand consistency: you cannot morally distinguish two cases without finding some non-moral difference between them.

Norm-expressivism was Allan Gibbard’s contribution in Wise Choices, Apt Feelings (1990). Gibbard held that moral judgments express the acceptance of systems of norms — rules dividing actions into forbidden, permitted, and required. To call an action rational is to express acceptance of a system of norms that permits it. Gibbard’s formal apparatus captures the content of a normative judgment by specifying the set of world-norm pairs it rules out: a judgment that lying is wrong is incompatible with any pair whose factual component describes the action as a lie and whose normative component is a norm permitting lying. Moral judgments are further analyzed through rationality: “An action is wrong if and only if it fails to meet standards of action the intentional or negligent violation of which in a normal state of mind would be sufficient for finding the agent prima facie blameworthy” (Gibbard 1990, p. 45). In Thinking How to Live (2003), Gibbard revised the framework: normative judgments express planning attitudes rather than norm-acceptance, modeled by “fact-prac worlds” (pairs of a factual specification of the world and a complete contingency plan). This provided a more systematic compositional semantics capable of handling complex sentence embeddings.

Quasi-realism is Simon Blackburn’s program, developed in Spreading the Word (1984), Essays in Quasi-Realism (1993), and Ruling Passions (1998). Blackburn accepts the expressivist starting point — moral judgments are projections of attitudes, not descriptions of mind-independent facts — but pursues the ambitious project of “earning the right” to all the features of realist moral discourse: truth-talk, moral facts, moral knowledge, correctness, error, logical inference. The strategy deploys a deflationary theory of truth: since “is true” adds no robust metaphysical property, we can say “It is true that stealing is wrong” without ontological cost — we are simply reaffirming our attitude. Blackburn’s quasi-realism is best understood not as a position but as a philosophical program, proceeding step by step to show that each apparently realist feature of moral discourse can be reconstructed on an expressivist base.

The Frege-Geach problem and why it threatens expressivism

The most enduring objection to non-cognitivism is the Frege-Geach problem (also called the embedding problem), first raised by Peter Geach in “Imperative and Deontic Logic” (Analysis, 1958), developed in “Ascriptivism” (The Philosophical Review, 1960), and given its fullest formulation in “Assertion” (The Philosophical Review, 1965). Geach attributed the underlying insight to Gottlob Frege’s distinction between assertoric force (the act of asserting something) and propositional content (the sense of a sentence, which can remain the same whether asserted or unasserted). Geach called this “the Frege point”: “A thought may have just the same content whether you assent to its truth or not; a proposition may occur in discourse now asserted, now unasserted, and yet be recognizably the same proposition.”

The problem strikes at the heart of expressivist semantics through a deceptively simple modus ponens argument. Consider Geach’s own example:

• P1: If tormenting the cat is wrong, then getting your little brother to torment the cat is also wrong.
• P2: Tormenting the cat is wrong.
• C: Therefore, getting your little brother to torment the cat is wrong.

This is obviously valid. But “wrong” must mean the same thing in every occurrence, or the argument commits the fallacy of equivocation. Here lies the difficulty. According to the expressivist, when someone asserts P2, they express disapproval of cat-tormenting. But in the antecedent of P1, “tormenting the cat is wrong” appears unasserted — the speaker of the conditional is not expressing disapproval of cat-tormenting. They might not even think it wrong; they are merely stating a conditional relationship. If the meaning of “tormenting the cat is wrong” is constituted by the attitude it expresses, then in the unasserted context of P1’s antecedent, the phrase either has no meaning or a different meaning. Either way, the modus ponens argument would be invalid — but it plainly is not.

Expressivist responses and their limitations

Hare attempted to construct a logic of imperatives paralleling the logic of propositions, but the project was widely regarded as incomplete. Blackburn proposed that complex moral sentences express higher-order attitudes — attitudes about combinations of attitudes. The conditional in P1 would express disapproval of the combination: disapproving of cat-tormenting while not disapproving of getting your brother to torment the cat. Critics including Bob Hale (1993) and Mark van Roojen (1996) argued that reducing logical inconsistency to attitudinal conflict conflates distinct phenomena. Gibbard’s formal apparatus of world-norm pairs (later fact-prac worlds) provides the most rigorous response: the content of each sentence is captured by the set of world-norm combinations it rules out, and logical relations between sentences correspond to set-theoretic relations between these contents. Modus ponens validity is preserved because the possibilities ruled out by the premises include those ruled out by the conclusion.

Mark Schroeder argued in Being For (2008) that the Frege-Geach problem actually conceals at least eight distinct sub-problems, and that the negation problem is the most fundamental. “Murder is not wrong” should be logically inconsistent with “Murder is wrong,” but the expressivist must identify a specific attitude expressed by the negation that conflicts with disapproval of murder. It cannot be approval (too strong), mere absence of disapproval (absence of an attitude is not an attitude to be “expressed”), or disapproval of non-murder (changes the subject). The consensus view is that while Gibbard and Blackburn have made considerable progress on the original formulation, deeper versions of the embedding challenge remain genuinely open problems.

Moral ontology: the realism/anti-realism landscape

The core divide

Moral realism holds three commitments: moral sentences express truth-apt beliefs (cognitivism), some of these beliefs are true, and their truth is objective — holding independently of anyone’s attitudes or stances. As David Brink put it, “Moral Realism should be our metaethical starting point, and we should give it up only if it does involve unacceptable metaphysical and epistemological commitments” (Moral Realism and the Foundations of Ethics, 1989). Anti-realism denies at least one of these commitments. Non-cognitivists deny the first; error theorists accept the first but deny the second; constructivists accept both but deny the third (making moral truth dependent on attitudes or agreements).

Platonic non-natural realism

Non-naturalism holds that moral properties are real but not identical with or reducible to any natural property. Its historical antecedent is Plato’s Form of the Good — a non-physical, abstract entity grounding all value. Mackie explicitly invoked Plato’s Forms to illustrate what objective values would have to look like: “The Form of the Good is such that knowledge of it provides the knower with both a direction and an overriding motive; something’s being good both tells the person who knows this to pursue it and makes him pursue it” (Ethics: Inventing Right and Wrong, 1977, Ch. 1, §9).

Moore’s Principia Ethica (1903) is the paradigmatic non-naturalist text. As Thomas Baldwin observed, “twentieth century British ethical theory is unintelligible without reference to Principia Ethica; its history until 1960 or so being, in brief, that although Moore was taken to have refuted ethical naturalism, Moore’s own brand of ethical non-naturalism was thought to make unacceptable metaphysical and epistemological demands; so the only recourse was to abandon belief in an objective moral reality.” Moore held that “good” denotes a simple, unanalyzable, non-natural property — analogous to “yellow,” which cannot be defined in more basic terms but can only be directly apprehended.

Modern non-naturalists include Russ Shafer-Landau (Moral Realism: A Defence, 2003), David Enoch (Taking Morality Seriously, 2011), T.M. Scanlon (Being Realistic About Reasons, 2014), and Derek Parfit (On What Matters, 2011). Enoch’s indispensability argument contends we must posit objective moral truths because they are indispensable to our best deliberative practices — analogous to how mathematical entities are indispensable to our best scientific theories. The chief objections to non-naturalism are the epistemological challenge (if moral properties are causally inert, how can we detect them?) and Mackie’s argument from queerness.

Cornell naturalist realism

Moral naturalism identifies moral facts with natural facts — the kind science investigates. The Cornell Realists, so called because Richard Boyd, David Brink, and Nicholas Sturgeon were associated with Cornell University, developed this position in the 1980s. Key texts include Boyd’s “How to Be a Moral Realist” (in Essays on Moral Realism, 1988), Sturgeon’s “Moral Explanations” (1985), Railton’s “Moral Realism” (Philosophical Review, 1986), and Brink’s Moral Realism and the Foundations of Ethics (1989).

Boyd’s account draws a systematic analogy between moral properties and natural kinds. Just as “water” refers to H₂O — a natural kind identified by its causal/explanatory role — “goodness” refers to a cluster of natural properties that play a particular causal role. Boyd calls these “homeostatic property clusters”: properties that tend to co-occur because of underlying causal mechanisms. Moral terms pick out these properties through a causal-historical chain of reference (Kripke-Putnam semantics), not through analytic definition. This response to the open question argument is powerful: the OQA at most refutes analytic naturalism (where “good” is synonymous with some natural term), not synthetic naturalism (where “good” is identical with a natural property discovered a posteriori). Just as someone can coherently ask “I know it’s H₂O, but is it water?” without revealing that water isn’t H₂O, someone can ask “I know it’s pleasurable, but is it good?” without demonstrating that goodness isn’t a natural property.

The most influential objection is Horgan and Timmons’ Moral Twin Earth argument (1992): on a planet where a different cluster of natural properties causally regulates moral terms, those terms would refer to different properties — yet intuitively the two communities seem to be genuinely disagreeing, not talking past each other. Sturgeon’s response to Gilbert Harman’s explanatory impotence challenge (the claim that moral facts are never needed to explain observations) was also influential: Sturgeon argued that moral facts do enter into genuine causal explanations — “the wrongness of slavery explains the abolitionists’ indignation.”

Error theory: Mackie’s two arguments

J.L. Mackie’s Ethics: Inventing Right and Wrong (1977) opens with the declaration “There are no objective values.” Error theory accepts cognitivism — moral sentences are genuine assertions — but holds they are systematically false because the moral facts they presuppose do not exist. The standard analogy: an error theorist regards moral discourse the way an atheist regards theological discourse. The claims are sincere, truth-apt, and systematically mistaken.

Mackie’s argument from relativity (or disagreement) is abductive. The enormous variation in moral views across cultures — persistent, intractable, and deep — is best explained by the hypothesis that moral judgments “reflect adherence to and participation in different ways of life” rather than by positing objective moral facts to which some cultures have better access. Mackie presses the point concretely: is it really plausible that one culture enjoys superior access to the moral facts about marital arrangements while another is simply blind to them?

The argument from queerness is more powerful and has two strands. The metaphysical strand claims that objective moral properties would be “entities or qualities or relations of a very strange sort, utterly different from anything else in the universe” (1977, p. 38). They would have to be intrinsically action-guiding — possessing built-in “to-be-pursuedness” that motivates any rational agent who perceives them, independent of that agent’s desires. Nothing else in the natural world has this feature. The epistemological strand follows: “If we were aware of them, it would have to be by some special faculty of moral perception or intuition, utterly different from our ordinary ways of knowing everything else.” Crucially, the epistemological strand is dependent on the metaphysical — we would need weird epistemic faculties only if the properties themselves are weird. The metaphysical strand is load-bearing.

Later error theorists include Richard Joyce (The Myth of Morality, 2001), who adds an evolutionary debunking argument and advocates moral fictionalism — retaining moral discourse as a useful fiction for pragmatic benefits — and Jonas Olson (Moral Error Theory: History, Critique, Defence, 2014), who identifies four distinct queerness arguments concerning supervenience, knowledge, motivation, and irreducible normativity.

Expressivism’s main forms as ontological positions

Each expressivist variant denies moral facts while reinterpreting what moral discourse does, generating a progressive arc of increasing sophistication. Ayer’s emotivism dissolved moral claims into emotional ejaculations (“Boo!”/“Hurrah!”). Stevenson added the machinery of emotive meaning and the belief/attitude distinction. Hare’s prescriptivism reconceived moral claims as universal prescriptions subject to rational consistency constraints. Gibbard’s norm-expressivism provided formal apparatus (world-norm pairs) for handling complex logical embeddings. Blackburn’s quasi-realism pursued the most ambitious program: starting from an expressivist base, progressively earning the right to every feature of realist discourse — truth, facts, knowledge, objectivity — through deflationary semantics and a logic of attitudes.

The quasi-realist program faces a signature challenge: Jamie Dreier’s “creeping minimalism” objection (Philosophical Perspectives, 2004). If the quasi-realist successfully earns the right to say everything the realist says — moral truths exist, moral beliefs can be correct or mistaken, moral facts are mind-independent — then what distinguishes the two views? As Blackburn himself acknowledged: “I take it that its success would be a measure of the difficulty of defining a genuine debate between realism and its opponents” (Essays in Quasi-Realism, 1993, p. 15). Dreier suggests the distinction may lie in explanatory direction: the realist explains attitude by fact, the expressivist explains fact-talk by attitude.

Street’s Darwinian Dilemma

Sharon Street’s “A Darwinian Dilemma for Realist Theories of Value” (Philosophical Studies 127, 2006, pp. 109–166) poses a powerful evolutionary challenge to moral realism. The first premise is empirical: natural selection has profoundly shaped human evaluative tendencies — we value survival, kin-care, reciprocity, and punishing harm-doers because these tendencies promoted reproductive fitness. The dilemma concerns the relationship between these evolutionary influences and the independent moral truths realism posits. Horn 1 (“no relation”): if evolutionary forces bear no relation to moral truth, then our moral judgments have been shaped by forces indifferent to truth, and the chance they track it is vanishingly small — radical skepticism follows. Street uses the analogy of a ship at sea pushed by random currents: the chance of reaching any particular destination is astronomically small. Horn 2 (“tracking”): the realist claims evolution favored ancestors who grasped moral truth. But evolutionary biology already provides a complete, more parsimonious explanation — these judgments exist because they promoted survival, not because they tracked stance-independent truths. For non-natural moral facts especially, which are causally inert, there is no mechanism for evolutionary tracking. Realist responses include David Enoch’s third-factor explanation (survival is genuinely good, and evolution pushed us toward valuing it, so a third factor explains both the evolutionary pressure and the moral truth) and the companions in guilt strategy (Street’s argument, if sound, would also debunk epistemic norms shaped by evolution).

Parfit’s end-run around the is-ought gap

Derek Parfit’s On What Matters (Vols. 1–2, 2011; Vol. 3, 2017) attempts not to bridge the is-ought gap directly but to dissolve the need for a bridge. His position — which he called “non-realist cognitivism” or “non-metaphysical non-naturalist normative cognitivism” — holds four claims: (i) normative judgments are beliefs (cognitivism), (ii) some are true (against error theory), (iii) normative concepts are irreducible to natural properties (against naturalism), and (iv) normative truths have no positive ontological implications — they are not “made true” by natural or non-natural entities.

Parfit’s central strategy is the “partners in innocence” argument: normative truths are like mathematical and logical truths. “2+2=4” is necessarily true without being made true by physical mathematical objects; similarly, “causing unnecessary suffering is wrong” can be true without positing metaphysically queer moral entities. Against naturalism, Parfit pressed a triviality objection: if “rightness” just is “maximizing happiness,” then the utilitarian claim becomes the tautology “happiness-maximizing acts maximize happiness” — but moral debates are substantive, not tautological. Against Nietzsche and Schopenhauer, Parfit argued they were misled by the ambiguity of the German sollen (meaning both “should” and “shall/is commanded”), conflating normative truths with divine commands and thereby concluding that without God, morality is empty. Parfit’s Triple Theory argued at the first-order level that the best versions of rule consequentialism, Kantian contractualism, and Scanlonian contractualism converge on the same principles — “climbing the same mountain on different sides” — providing evidence for objective moral truth. Peter Singer called it “the most significant work in ethics since Sidgwick’s masterpiece”; critics including Mark Schroeder and Constantine Sandis argued the convergence presupposes consequentialist assumptions that Kantians would reject.

The logical architecture connecting these positions

The positions in metaethics form a tightly interlocking logical structure rather than a mere catalog. The is-ought gap and Moore’s open question argument create the foundational challenges that every subsequent position must address. Cognitivism branches into realism (naturalist and non-naturalist) and error theory; non-cognitivism branches from emotivism through prescriptivism to sophisticated expressivism. Each position arises as a response to the perceived failures of its predecessors: non-naturalism answers naturalism’s reduction but faces epistemological queerness; error theory preserves cognitivist semantics but bites the bullet of universal falsity; emotivism avoids ontological commitments but cannot explain moral reasoning or logical embedding; prescriptivism adds rational structure but still falls to Frege-Geach; sophisticated expressivism and quasi-realism address embedding through formal apparatus and deflationary truth, only to encounter creeping minimalism — the risk that a fully successful anti-realism becomes indistinguishable from the realism it sought to replace. This dialectical structure — where solving one problem generates the next — is not a defect of the field but its engine.

Position	Moral sentences express…	Truth-apt?	Moral facts exist?	Key figures and landmark texts
Non-naturalist realism	Beliefs about non-natural facts	Yes	Yes (sui generis)	Moore, Principia Ethica (1903); Shafer-Landau (2003); Enoch (2011); Parfit (2011)
Naturalist realism	Beliefs about natural facts	Yes	Yes (natural)	Boyd (1988); Sturgeon (1985); Railton (1986); Brink (1989)
Error theory	Beliefs (all false)	Yes	No	Mackie (1977); Joyce (2001); Olson (2014)
Emotivism	Emotions/attitudes	No	No	Ayer (1936); Stevenson (1944)
Prescriptivism	Universal prescriptions	No	No	Hare (1952, 1963)
Norm-expressivism	Norm-acceptance / plans	No (robust)	No	Gibbard (1990, 2003)
Quasi-realism	Attitudes (+ earned truth-talk)	Deflationary yes	No (projectivism)	Blackburn (1984, 1993, 1998)

From Construction to Supervenience

Moral philosophy’s middle territory — constructivism, relativism, epistemology, motivation, and supervenience — contains the discipline’s most consequential fault lines. These five domains determine whether morality can be objective without being discovered, whether disagreement undermines realism, how we know moral truths (if we do), what connects judgment to action, and why moral properties track natural ones. This chapter maps each domain at textbook depth, tracing argument structures and their interconnections across roughly a century of landmark contributions.

Constructivism: morality as neither found nor merely felt

Constructivism occupies a distinctive position in the metaethical landscape: moral truths are the outputs of a correct procedure or rational deliberation, not pre-existing facts we discover (realism) and not mere attitudes we express (expressivism). Christine Korsgaard captures this with her distinction between “substantive realism” — the view that correct moral procedures exist because moral truths exist independently — and “procedural realism” — the view that moral truths exist because correct procedures generate them. The analogy Sharon Street deploys is illuminating: asking whether a baseball player is “safe” is not asking about a stance-independent fact or expressing an emotion. The answer is constituted by the rules of the game applied to non-normative facts about the player’s position. Constructivism treats morality similarly.

John Rawls introduced the term “constructivism” into moral philosophy in his 1980 Dewey Lectures, “Kantian Constructivism in Moral Theory” (Journal of Philosophy 77: 515–572). His formulation was explicit: “Apart from the procedure of constructing the principles of justice, there are no moral facts.” The parties in his original position — rational agents representing free and equal citizens, characterized by two moral powers (a capacity for a conception of the good and a sense of justice) — choose principles behind a veil of ignorance that strips them of knowledge about their race, gender, class, talents, and conception of the good. They retain only general facts about human psychology and social organization. From this procedure emerge Rawls’s two principles: the Equal Liberty Principle (each person has equal right to the most extensive system of basic liberties) and the Difference Principle (inequalities are permissible only if they benefit the least-advantaged). The veil of ignorance is not a discovery mechanism for pre-existing truths — it is the constructive device through which fair principles are generated. The fairness of the procedure transfers to the principles chosen.

By Political Liberalism (1993), Rawls narrowed his ambitions significantly. He distinguished political constructivism — limited to principles for the basic structure of society — from moral constructivism that would address all moral truths comprehensively. Political constructivism is “freestanding”: it does not presuppose any particular metaphysical, religious, or comprehensive moral doctrine, and is designed to accommodate the “fact of reasonable pluralism” in democratic societies. This retreat left Rawls’s mature position compatible with multiple metaethical interpretations, including realism.

Korsgaard’s constitutivism and the transcendental argument from agency

Christine Korsgaard’s The Sources of Normativity (1996) globalizes constructivism beyond the political domain to all normativity. Her starting point is the reflective endorsement argument: human consciousness creates a gap between impulse and action that animals lack. We can step back from any desire and ask whether to act on it. This reflection “suspends the authority of given attitudes” — a desire cannot serve as a reason for action until we reflectively endorse it. Endorsement requires a principle, and principles derive from our practical identities: the descriptions under which we value ourselves and find life worth living (being a parent, an artist, a citizen). “It is the conceptions of ourselves that are most important to us that give rise to unconditional obligations,” Korsgaard writes, “for to violate them is to lose your integrity and so your identity.”

The critical move is a transcendental argument. If you value anything at all, you must value your capacity to confer value — your humanity (rational nature). “You must value your own humanity if you are to value anything at all.” But humanity is a shared property. If your own rational nature provides you with reasons, then rational nature as such provides reasons — including the rational nature of others. This yields Kant’s Formula of Humanity: treat humanity, in your own person or in any other, always as an end and never merely as a means.

This is constitutivism: the view that moral norms are constitutive of rational agency itself. Korsgaard develops this further in Self-Constitution (2009), arguing that an agent who violates moral principles is not merely immoral but is failing as an agent — analogous to a knife that cannot cut. The norms are not external constraints but constitutive standards of the activity of rational willing.

Street’s Humean constructivism diverges on convergence

Sharon Street’s “What is Constructivism in Ethics and Metaethics?” (Philosophy Compass, 2010) proposes a taxonomy distinguishing Kantian constructivism (Korsgaard) from Humean constructivism (Street’s own view). Both deny stance-independent normative facts. The crucial difference concerns convergence. Korsgaard’s Kantian version holds that formal constraints on practical reasoning guarantee that all rational agents converge on a single set of moral truths. Street’s Humean version denies this: formal constraints like coherence and consistency constrain but do not fully determine one’s reasons. “Similarities in human beings’ reasons ultimately depend for their existence on contingent similarities in people’s evaluative starting points and circumstances.” An ideally coherent Caligula might have no reason not to torture — a consequence Street accepts as a form of agent-relative normative truth.

For Street, normative truth consists in what is entailed from within the agent’s practical standpoint. A normative judgment is correct for agent A if and only if it withstands scrutiny from the standpoint of A’s other evaluative attitudes. This connects directly to her Darwinian Dilemma (Philosophical Studies 127, 2006): evolution shaped our evaluative attitudes for fitness, not truth, which is devastating for realism (which needs our attitudes to track stance-independent facts) but unproblematic for constructivism (which takes evaluative truth to be constituted by those attitudes).

T.M. Scanlon’s contractualism in What We Owe to Each Other (1998) offers a parallel but more restricted constructivism. His reasonable rejection test holds that an act is wrong if its performance would be disallowed by principles that no one could reasonably reject as a basis for informed, unforced general agreement. This is non-aggregative — it focuses on individual standpoints, not aggregate welfare — and constructivist in that wrongness is determined by the contractualist procedure. However, Scanlon favors non-naturalist realism about reasons at a deeper level (Being Realistic about Reasons, 2014), making his constructivism local rather than global.

The circularity objection and the scope problem

The most fundamental objection to constructivism mirrors the Euthyphro dilemma. Either the constructive procedure is constrained by prior moral values or it is not. If unconstrained, there is no reason to expect morally acceptable outputs; if constrained, those values are presupposed rather than constructed. Joseph Raz crystallized this: constructivism is “vacuous because its test yields results only by presupposing moral views which can only be established independently of it” (2003). Rawls’s original position, for instance, already embodies commitments to equality and fairness through the veil of ignorance. Korsgaard’s transcendental argument arguably presupposes the value of humanity as a realist premise. Constructivists respond that the circularity is “virtuous” — reason’s self-vindication is necessarily reflexive — or that constitutive norms need no external justification, just as the rules of chess need not appeal to anything outside chess.

The scope problem asks: who is bound by constructed norms? If morality is constitutive of rational agency, what about psychopaths who arguably lack the relevant rational capacities, or animals who cannot participate in construction? Humean constructivism faces this acutely: an ideally coherent agent with no concern for others literally has no moral reasons. The publicity problem (Hussain & Shah 2006; Lenman 2012) questions whether constructivism can maintain itself as a genuine “third way” or inevitably collapses into either realism (if it smuggles in independent moral constraints) or expressivism (if it offers no truth-conditions beyond practical commitments).

Relativism and subjectivism face the progress problem

The most common route to moral relativism begins with an observation and commits a fallacy. Descriptive relativism — the empirical observation that moral beliefs differ across cultures, documented by anthropologists from Westermarck (1906) through Benedict (1934) to contemporary cross-cultural psychology — is uncontroversial. Metaethical relativism — the philosophical thesis that moral truth itself is relative to a culture or framework — is a substantive and controversial claim. The move from the first to the second commits precisely the is-ought fallacy that Hume’s guillotine prohibits. As James Rachels formalized it: from the premise that different cultures have different moral codes, the conclusion that there is no objective moral truth simply does not follow. People in different societies once believed the Earth was flat; the disagreement did not show there was no truth about the Earth’s shape.

Gilbert Harman’s “Moral Relativism Defended” (Philosophical Review 84, 1975) offers the most sophisticated version. Harman proposes that the moral “ought” is a four-place predicate: Ought(Agent, Act, Considerations, Motivating Attitudes). The relativism lies in the final parameter. Morality arises when groups reach implicit agreements about their relations, and inner judgments — claims that someone ought or ought not to have acted in a certain way — presuppose that the agent shares the relevant motivating attitudes. We can comfortably say Stalin “should not have” pursued his course (because we can imagine him as sharing some moral concerns and acting against them), but saying the same about entirely alien beings “without the slightest concern for human life” seems oddly inappropriate. For agents outside any shared moral framework, only outer judgments (calling them “evil” or “savage”) are appropriate, and these are not relativized to agreements. This is a “soberly logical thesis about logical form,” not crude cultural relativism.

Simple subjectivism — “X is wrong” means “I disapprove of X” — fails on two counts identified by Rachels. First, it renders everyone morally infallible: whenever you sincerely report your attitude, you speak truly. But we recognize that our past moral views were sometimes wrong. Second, it eliminates genuine moral disagreement: if you approve and I disapprove, we are both correctly reporting our attitudes and not contradicting each other. These failures motivated the development of emotivism and, in a different direction, ideal observer theory.

Firth’s ideal observer rescues objectivity through idealization

Roderick Firth’s “Ethical Absolutism and the Ideal Observer” (Philosophy and Phenomenological Research 12, 1952) defines a hypothetical observer possessing six characteristics: omniscience regarding non-ethical facts, omnipercipience (vivid experiential understanding of all relevant circumstances), disinterestedness (no personal stakes), dispassionateness (freedom from distorting emotions), consistency (stable reactions), and being otherwise normal (recognizably human apart from these idealizations). “X is wrong” means “any ideal observer would disapprove of X.” Because the ideal observer is the same for everyone — anyone meeting those conditions would react identically — the theory yields universal moral truths despite grounding them in responses rather than mind-independent facts. This makes ideal observer theory a form of cognitivism, not relativism: moral claims are truth-apt propositions about what a specific (hypothetical) entity would feel.

The theory connects to Adam Smith’s impartial spectator in The Theory of Moral Sentiments (1759), though Smith’s figure is less idealized — more a workable conscience for everyday life than an omniscient entity. Rawls himself criticized the ideal observer framework for risking the fusion of distinct individuals into a single system of desire, violating the “separateness of persons.”

All relative and subjective views face a cluster of devastating objections. They cannot account for moral progress: the abolition of slavery was not merely a change in preferences but an improvement, and relativism cannot explain why. They cannot justify cross-cultural moral criticism: if morality is relative to cultures, condemning genocide or slavery in other societies is illegitimate. They generate a self-refutation problem: “moral truth is relative” is itself presented as a non-relative truth, and if it is only relatively true, outsiders have no reason to accept it. And they face the moral reformer problem: figures like Martin Luther King Jr. and the abolitionists opposed their cultures’ prevailing norms — on relativist grounds, they were morally wrong, which is absurd.

Moral epistemology: how we know what we ought to do

Intuitionism and the self-evidence of prima facie duties

Moral intuitionism holds that some moral truths are self-evident: knowable non-inferentially through adequate understanding, analogous to grasping axioms of mathematics. W.D. Ross’s The Right and the Good (1930) systematized this through seven prima facie duties: fidelity (keep promises), reparation (compensate for wrongs done), gratitude (reciprocate benefits received), justice (distribute goods by merit), beneficence (improve others’ condition), self-improvement (develop one’s own virtue and knowledge), and non-maleficence (do not injure others). These are not absolute obligations but pro tanto reasons — considerations that always carry genuine moral weight but can be outweighed. When duties conflict, “no general rules can, so far as I can see, be laid down” for adjudication; one must perceive the particular situation.

Ross explicitly drew the mathematical parallel: “The moral order… is just as much part of the fundamental nature of the universe as is the spatial or numerical structure expressed in the axioms of geometry or arithmetic.” Michael Huemer’s Ethical Intuitionism (2005) updates the case through Phenomenal Conservatism: “Other things being equal, it is reasonable to assume that things are the way they appear.” Moral intuitions are intellectual seemings that provide defeasible warrant, just as perceptual seemings justify empirical beliefs absent defeaters.

The objections are formidable. Cross-cultural disagreement about moral intuitions undermines the self-evidence claim. Evolutionary debunking arguments (Street 2006; Joyce 2007) hold that natural selection shaped our moral intuitions for reproductive fitness rather than truth-tracking, severing the reliability connection that self-evidence requires. And Ross’s admission that conflicting duties lack principled adjudication leaves the theory practically indeterminate at precisely the points where guidance is most needed.

Kantian rationalism and the empty formalism charge

Kant’s alternative grounds moral knowledge entirely in pure practical reason. The categorical imperative has three formulations expressing the same underlying principle: the Formula of Universal Law (“Act only in accordance with that maxim through which you can at the same time will that it become a universal law,” Groundwork 4:421 ), the Formula of Humanity (“Act so that you treat humanity, whether in your own person or in that of another, always as an end and never merely as a means,” 4:429 ), and the Formula of the Kingdom of Ends (“Act in accordance with the maxims of a member giving universal laws for a merely possible kingdom of ends,” 4:439 ). Autonomy in Kant’s sense is not arbitrary choice but the will’s capacity to legislate universal moral principles through reason alone — self-legislation. A free will and a will under moral laws are one and the same.

The persistent objection, articulated by Hegel (1802, 1821) and never fully laid to rest, is empty formalism: a purely formal consistency test cannot generate substantive moral content without smuggling in assumptions from existing social institutions. Kantians respond that the Formula of Humanity provides substantive content through the intrinsic value of rational nature, and that rational agency itself constrains outcomes more than the charge allows.

Sentimentalism, reflective equilibrium, and method in ethics

Sentimentalism grounds moral knowledge in emotion rather than reason. Hume’s famous declaration — “Reason is, and ought only to be the slave of the passions” (Treatise 2.3.3.4) — denies that reason can generate moral conclusions or motivate action. Smith’s Theory of Moral Sentiments (1759) refines this through the impartial spectator: we form moral judgments by asking whether a disinterested, well-informed observer would approve. Jesse Prinz’s The Emotional Construction of Morals (2007) updates the tradition as “constructive sentimentalism”: moral concepts are constituted by emotional responses that are culturally shaped, making morality a social construction like money — objective enough for genuine moral discourse but relative across cultures.

Rawls’s method of reflective equilibrium (A Theory of Justice, 1971) offers a different epistemological path, drawing on Nelson Goodman’s insight (Fact, Fiction, and Forecast, 1955) that we justify logical rules by testing them against accepted inferences and vice versa. Narrow reflective equilibrium seeks coherence between considered moral judgments and moral principles. Wide reflective equilibrium (elaborated by Norman Daniels, “Wide Reflective Equilibrium and Theory Acceptance in Ethics,” Journal of Philosophy 76, 1979) incorporates background theories — philosophical, psychological, scientific — as a third element, allowing much more extensive revision of initial convictions. “Considered judgments” are those formed under conditions minimizing distortion: calm reflection, full information, no personal stake.

The central objection is the coherence-vs-truth worry: equilibrium might systematize bias rather than track moral reality. If one begins with racist considered judgments, coherent systematization yields a coherent racist theory. Richard Brandt (A Theory of the Good and the Right, 1979) charged that reflective equilibrium merely systematizes existing prejudices. Defenders respond that wide equilibrium’s incorporation of background theories provides independent constraint, and that moral epistemology — like scientific epistemology — is unavoidably holistic.

Disagreement, debunking, and the calibration crisis

Moral disagreement creates epistemological pressure from multiple directions. The conciliationist literature in general epistemology — David Christensen (Philosophical Review 116, 2007), Richard Feldman (2006), Adam Elga (Noûs 41, 2007) — argues that when epistemic peers disagree, both should reduce confidence. Applied to ethics: if equally thoughtful moral philosophers persistently disagree about fundamental questions, this constitutes higher-order evidence that our moral faculties are unreliable. The argument from disagreement against moral realism holds that if objective moral facts existed and were epistemically accessible, we would expect convergence over time, as in science. The realist responds that disagreement is explained by self-interest, bias, and varying epistemic access — the same factors that explain scientific disagreement — and that persistent disagreement no more refutes moral realism than persistent cosmological disagreement refutes physical realism. Mackie deployed disagreement as part of his case for error theory (Ethics: Inventing Right and Wrong, 1977), arguing that moral diversity reflects “adherence to and participation in different ways of life” rather than differing apprehensions of objective moral facts.

The calibration problem intensifies these worries through empirical findings. Jonathan Haidt’s social intuitionist model (“The Emotional Dog and Its Rational Tail,” Psychological Review 108, 2001) presents evidence that moral reasoning is primarily post-hoc rationalization of rapid emotional responses. His moral dumbfounding studies demonstrate that people maintain strong moral judgments even when every stated reason is refuted — participants in the famous Julie-and-Mark incest vignette say “I know it’s wrong, but I just can’t come up with a reason why.”

Joshua Greene’s dual-process theory adds neuroscientific precision. His fMRI studies (Science 293, 2001; Neuron 44, 2004) show that personal moral dilemmas (the footbridge case — physically pushing a man to his death) activate brain areas associated with social-emotional processing, while impersonal dilemmas (the switch case — diverting a trolley) activate regions associated with cognitive control. Participants who gave utilitarian responses in high-conflict personal dilemmas showed longer response times and greater dorsolateral prefrontal cortex activity, consistent with effortful cognitive override of emotional responses. Greene’s provocative conclusion: deontological judgments are “preferentially supported by automatic-emotional processes,” while utilitarian judgments are “supported by conscious-controlled processes and deliberative reasoning.”

Peter Singer (“Ethics and Intuitions,” Journal of Ethics 9, 2005) weaponized Greene’s findings against deontological ethics: if deontological intuitions are biological residues of evolutionary history rather than rational insights, “it is not clear why we should regard them as having any normative force.” The debunking argument structure (formalized by Guy Kahane, Noûs 45, 2011) proceeds: if we can give a complete causal explanation of why we have certain moral intuitions — through evolution, emotional conditioning, cultural learning — without reference to their truth, this undermines their epistemic status. Selim Berker’s influential reply (“The Normative Insignificance of Neuroscience,” Philosophy and Public Affairs 37, 2009) argues that fMRI data alone cannot adjudicate between moral theories; the inference from “this judgment has emotional origins” to “this judgment is unreliable” commits a subtle is-ought fallacy. Moreover, debunking arguments risk over-generalization: the utilitarian intuition that suffering is bad is itself an “obvious candidate for evolutionary debunking” (Kahane), threatening to undermine all moral knowledge rather than selectively targeting deontology.

The internalism/externalism debate connects judgment to action

The precise distinction and its metaethical stakes

Motivational internalism holds that there is a necessary connection between sincere moral judgment and motivation: if S judges that action A is morally required, S is thereby motivated (at least to some degree) to do A. Motivational externalism holds that the connection is contingent: moral judgments are beliefs that can fail to motivate without conceptual incoherence. Michael Smith’s influential formulation in The Moral Problem (1994) articulates the dominant conditional internalism: “Necessarily, if an agent judges that it is right for her to φ in circumstances C, then either she is motivated to φ in C or she is practically irrational.” This permits depression, weakness of will, and acedia to sever the judgment-motivation link, but only at the cost of irrationality.

The stakes are enormous. If internalism is true, moral judgments cannot be pure beliefs — because on the Humean theory of motivation, beliefs alone cannot motivate (only desires can). This supports expressivism: moral judgments must be or include desire-like conative states. If externalism is true, moral judgments can be beliefs, supporting cognitivism and opening the door to moral realism. Smith identifies the resulting inconsistent triad: (1) moral judgments express beliefs (cognitivism), (2) moral judgments necessarily motivate in rational agents (internalism), and (3) beliefs alone cannot motivate (Humean theory). Any two can be held, but not all three. Rejecting (3) yields anti-Humean cognitivist internalism (Nagel, McDowell). Rejecting (2) yields externalist cognitivism (Brink, Shafer-Landau). Rejecting (1) yields expressivism (Blackburn, Gibbard).

The amoralist, the Humean model, and anti-Humean perception

The amoralist thought experiment tests internalism directly: can someone sincerely judge “torturing innocents is wrong” while remaining entirely unmoved? R.M. Hare (The Language of Morals, 1952) argued that such a person uses moral language in an “inverted commas” sense — merely reporting that an action violates standards others accept, without genuinely endorsing those standards. The amoralist’s usage is parasitic and degraded, not a genuine counterexample. Mackie (Ethics: Inventing Right and Wrong, 1977) took the amoralist seriously but drew a different conclusion: our moral concepts purport to refer to “objectively prescriptive” properties that are simultaneously factual and motivating, but no such properties exist — hence error theory. Smith’s fetishism argument challenges externalism from the opposite direction: if the connection between judgment and motivation were merely contingent, the morally good person would need a standing desire to “do whatever is right” — but this makes them a “moral fetishist” who cares about the abstract property of rightness rather than the substantive features (suffering, injustice) that make actions right.

The Humean theory of motivation — rooted in Hume’s declaration that “reason alone can never be a motive to any action of the will” (Treatise 2.3.3.1) — holds that beliefs have mind-to-world direction of fit (representing how things are) while desires have world-to-mind direction of fit (representing how the agent wants things to be). Only desires can generate motivation; beliefs constrain and channel desire-satisfaction but cannot initiate action.

Anti-Humean alternatives reject this division. Thomas Nagel (The Possibility of Altruism, 1970) argued that recognizing a reason is itself motivating — any desire present when one acts morally is a “logical shadow” of the motivating belief, produced by the recognition of the reason rather than prior to it. John McDowell (“Virtue and Reason,” The Monist 62, 1979) proposed that the virtuous agent possesses a perceptual sensitivity to moral features of situations that is simultaneously cognitive and motivating. The virtuous person’s practical reasoning is not belief-plus-desire but a unified perception-action: seeing what kindness requires just is being moved to act. Crucially, the virtuous person’s perception of a moral requirement silences competing selfish considerations — they do not even register as genuine reasons, distinguishing virtue from mere continent self-control.

Williams denies external reasons and threatens moral universalism

Bernard Williams’s “Internal and External Reasons” (1979; reprinted in Moral Luck, 1981) — one of the most cited philosophical texts of the late twentieth century — argues that an agent has a reason to act only if sound deliberation connects the action to something in the agent’s subjective motivational set: desires, values, commitments, dispositions of evaluation, patterns of emotional reaction. An external reason claim — “you have a reason to be moral even if it doesn’t connect to anything you care about” — is either false or trivially reducible to an internal reason. Williams’s argument: a reason must be capable of explaining action; nothing explains intentional action except something that motivates the agent; nothing motivates except what connects to the subjective motivational set; therefore, there are only internal reasons.

The implication is stark: an agent with no concern for others literally has no reason to act morally. Morality’s rational authority becomes hostage to individual psychology. Korsgaard responded (“Skepticism about Practical Reason,” 1986) that an agent’s imperviousness to a consideration might impugn the agent’s rationality rather than the consideration’s status as a reason — being a rational agent entails having moral motives. Parfit (On What Matters, 2011) rejected internalism entirely, defending objective normative reasons independent of subjective motivational sets. McDowell argued that the virtuous person’s moral sensitivity constitutes genuine reason-responsiveness irreducible to Williams’s framework.

What follows for the “why be moral?” question depends on which view prevails. If Williams is right, the question has no universal answer — morality rests on psychological contingency. If Kant and Korsgaard are right, rationality itself demands morality: a free will and a will under moral laws are the same. The resolution of this question determines whether moral philosophy can vindicate morality’s claim to universal rational authority or must accept that such authority extends only to those whose motivational sets happen to include moral concern.

Supervenience: why moral properties track natural ones

The supervenience of the moral on the natural is among the least controversial theses in metaethics — virtually all parties accept it — yet explaining why it holds generates some of the sharpest disagreements. The thesis states that no two possible cases can be alike in all natural respects but differ in moral respects. If action A is wrong in situation S, any exact natural duplicate of A in a duplicate of S is also wrong. R.M. Hare introduced the term into ethics in The Language of Morals (1952): “It is logically impossible to say that St. Francis was a good man and to maintain at the same time that there might have been another man placed exactly in the same circumstances, and who behaved in exactly the same way, but who was not a good man.” G.E. Moore had articulated the idea implicitly in “The Conception of Intrinsic Value” (1922): if anything possesses intrinsic value, “anything exactly like it must, under all circumstances, possess it in exactly the same degree.”

Strong versus weak supervenience and why the distinction matters

Jaegwon Kim (“Concepts of Supervenience,” Philosophy and Phenomenological Research 45, 1984) systematized three varieties. Weak supervenience is intra-world: within any given possible world, no two things differ morally without differing naturally. Strong supervenience holds across all possible worlds: no two things in any possible worlds differ morally without differing naturally. The crucial difference is a second necessity operator. Weak supervenience permits “utilitarian worlds” (where rightness tracks happiness maximization) coexisting with “Kantian worlds” (where it tracks the categorical imperative) — so long as within each world, natural duplicates share moral status. Strong supervenience rules this out, making it the philosophically more interesting claim. Kim himself noted that supervenience is a “superficial relation that points to some other ‘deeper’ relation” (1987) — it states a pattern of covariation without explaining it.

Blackburn’s challenge and the ban on mixed worlds

Simon Blackburn developed the supervenience argument against moral realism in “Moral Realism” (1971; reprinted in Essays in Quasi-Realism, 1993) and Spreading the Word (1984). He identified two theses that non-naturalist realists accept: (1) the moral is conceptually supervenient on the natural, and (2) no purely natural description logically entails any moral conclusion (the non-entailment thesis). Together, these mean that for any set of natural properties, there is a possible world where things with those properties are morally good and another where they are morally bad — but there is no “mixed world” where two natural duplicates differ morally. Blackburn found this ban on mixed worlds inexplicable under realism: “we are asked to make intelligible the notion of a state of affairs subject to the constraint that its existence does not follow from the naturalistic facts being as they are but its continued existence follows from the natural facts staying as they are.” For the quasi-realist projectivist, the explanation is straightforward: we project attitudes onto nature, and naturally similar things naturally elicit the same projection. The ban on mixed worlds reflects a constraint on attitude consistency, not a mysterious metaphysical harmony.

Jackson’s argument that supervenience entails naturalism

Frank Jackson’s From Metaphysics to Ethics (1998) pushes further. Given the truth-aptness of moral claims and the global supervenience of the ethical on the descriptive, every morally wrong act has some complete natural description, and any act with that description is also wrong. The disjunction of all such descriptions yields a (possibly infinitely disjunctive) natural property necessarily co-extensive with wrongness. Jackson’s crucial second step: necessarily co-extensive properties are identical. Therefore, moral properties just are natural properties at some level of description. This supports analytical descriptivism — moral terms can be analyzed through their role in “folk morality.” Unlike Cornell realists (Boyd, Brink, Sturgeon), who treated the moral-natural identity as synthetic and a posteriori (like water = H₂O), Jackson argues it is discoverable a priori through conceptual analysis.

Non-naturalists invoke necessary connections between distinct properties

Non-naturalists resist both Blackburn and Jackson by arguing that the supervenience relation is a necessary connection between genuinely distinct properties — and that such connections are not inherently mysterious. The standard analogy: triangularity and having interior angles summing to 180 degrees are necessarily connected yet arguably distinct properties (one concerns shape, the other angular measure). The connection is necessary without being analytic in a trivial sense. Russ Shafer-Landau (Moral Realism: A Defence, 2003) proposes that moral laws necessarily connect natural base properties to moral properties — brute necessities no more mysterious than the laws of mathematics. David Enoch (Taking Morality Seriously, 2011) similarly defends necessary moral principles that explain supervenience without reduction. Against Jackson’s second step, critics note that necessarily co-extensive properties need not be identical on richer metaphysical frameworks involving universals or tropes — there is “logical space for non-natural moral properties even if they are necessarily co-instantiated with natural properties.”

The broader significance of supervenience extends in every direction. It creates tension with the is-ought gap (tight moral-natural covariance seems to undermine Hume’s insistence on a logical gap between descriptive and normative). It connects to Moore’s open question argument (supervenience is consistent with non-naturalism but, if Jackson is right, ultimately collapses it). And it connects to moral explanation: if moral properties supervene on natural ones, natural properties do all the causal work, raising the question — first posed by Harman (1977) and contested by Sturgeon (1988) — of whether moral facts explain anything that natural facts do not already explain.

The five domains covered here form an interconnected web rather than isolated topics. Constructivism attempts to resolve the realism/anti-realism divide by making moral truths the outputs of rational procedures — but faces circularity objections that threaten to collapse it back into one of the camps it sought to transcend. Relativism and subjectivism fail to survive the moral progress objection, while the ideal observer theory reveals that making subjectivism sophisticated enough to handle these objections transforms it into a form of objectivism. Moral epistemology shows that every candidate source of moral knowledge — intuition, reason, sentiment, reflective equilibrium — faces serious challenges, compounded by empirical evidence from Haidt and Greene suggesting that much moral judgment is post-hoc rationalization of emotional reactions. The internalism/externalism debate connects moral psychology to moral ontology through Smith’s inconsistent triad, with Williams’s denial of external reasons threatening to sever morality from universal rational authority. And supervenience — the one thesis virtually everyone accepts — generates a puzzle that non-naturalists struggle to solve: why do genuinely distinct moral and natural properties necessarily co-vary?

The thread connecting all five domains is a single question: can morality be both objective and human? Constructivists say yes — through procedures. Relativists say no — objectivity is an illusion. Moral epistemologists map the terrain of how we might know. The internalism debate asks whether moral knowledge can move us. And supervenience asks whether the moral can be real without being natural.

Metaethics Arguments

Haidt’s moral foundations theory

Jonathan Haidt and Craig Joseph first proposed the theory in “Intuitive Ethics: How Innately Prepared Intuitions Generate Culturally Variable Virtues” (Daedalus, 2004), building on Richard Shweder’s cross-cultural work. The major elaboration appears in The Righteous Mind: Why Good People Are Divided by Politics and Religion (Pantheon, 2012). Key collaborators include Jesse Graham and Brian Nosek.

The six foundations (exact name pairs):

The original five (2004–2007):

1. Care/Harm — Tracks suffering, vulnerability, neediness. Evolutionary basis: the adaptive challenge of caring for vulnerable offspring; the attachment system and empathy for kin. Underlies virtues of kindness, nurturance, compassion; condemns cruelty and aggression.
2. Fairness/Cheating — Tracks reciprocity, proportionality, cheating in cooperative exchanges. Evolutionary basis: reciprocal altruism and cheater-detection mechanisms. Originally encompassed both equality and proportionality; in The Righteous Mind (2012), Haidt narrowed it primarily to proportionality (getting rewards in proportion to contributions), separating out equality concerns.
3. Loyalty/Betrayal — Tracks group cohesion, coalition membership, tribal identity. Evolutionary basis: long history of kin-based groups and intergroup competition. Underlies patriotism, self-sacrifice for the group; condemns betrayal and disloyalty.
4. Authority/Subversion — Tracks hierarchical social interactions, deference, dominance. Evolutionary basis: long primate history of hierarchical social structures. Underlies virtues of leadership, followership, deference to legitimate authority and tradition; condemns insubordination and subversion.
5. Sanctity/Degradation (also called Purity/Degradation) — Tracks contamination, disgust, purity. Evolutionary basis: disgust evolved to avoid pathogens and parasites. Underlies notions of the body as a temple, self-discipline, spirituality, naturalness; condemns degradation and defilement.

Sixth foundation (added in The Righteous Mind, Ch. 8, 2012):

6. Liberty/Oppression — Tracks domination, tyranny, autonomy, freedom from coercion. Evolutionary basis: egalitarian instincts in small bands that resisted alpha males. Added because the original five did not adequately capture libertarian and economic-conservative sensibilities. With this addition, Fairness became primarily about proportionality, while Liberty captured concerns about oppression and autonomy.

Political valence: Liberals rely heavily on Care/Harm and Fairness/Cheating (the “individualizing” foundations), with some support from Liberty/Oppression. Conservatives draw more equally on all six foundations, including the “binding” foundations: Loyalty/Betrayal, Authority/Subversion, and Sanctity/Degradation. Libertarians are most sensitive to Liberty/Oppression. Haidt’s central metaphor: the foundations are like “taste receptors” — “Moral matrices vary, but they all must please righteous minds equipped with the same six social receptors.”

2023 updates: Atari, Graham, and Haidt revised the theory, splitting Fairness into Equality and Proportionality as separate foundations. Ownership has been proposed as a potential additional foundation since 2013. An Honor foundation has been explored for Middle Eastern cultures. The theory explicitly remains open to modification.

Philip Kitcher’s The Ethical Project (2011)

Philip Kitcher, The Ethical Project (Harvard University Press, 2011). Kitcher was John Dewey Professor of Philosophy at Columbia.

Main thesis: Ethics is a “social technology” — not a set of divine revelations or discoveries of brilliant thinkers, but a set of practices that evolved over tens of thousands of years as humans worked out how to live together. Ethics is analogous to material technology: aimed at solving practical problems of group living.

The altruism-failures model (Part I): Kitcher begins with preethical ancestors who possessed a capacity for psychological altruism but one that was limited and fragile. Their altruistic dispositions made living together “tense and difficult.” Altruism failures are recurrent breakdowns in responsive altruism that created conflict and disrupted group life. The first ethicists overcame these problems by agreeing on rules for conduct — rules remedying recurrent altruism failures. This was the origin of ethics: deliberation around the campfire, negotiation of behavioral norms. With language and greater cognitive sophistication, ancestors could develop more complex behavioral rules. Ethical codes then evolved via cultural evolution, with more successful codes giving groups selective advantages.

Pragmatic naturalism (Part II): Kitcher resolves the tension between “mere change” and “genuine progress” in ethical evolution by modeling ethical progress on technological progress — discharging functions more efficiently or more fully. The initial function of ethical practice is the amelioration of altruism failures, but new functions emerged over time. Ethical truths are defined as “those acquired in progressive transitions and retained through an indefinite sequence of progressive transitions” — a pragmatic, not Platonic, conception of truth. Kitcher denies ethical expertise; philosophy’s role is “midwifery” — suggesting directions for renewed conversation.

Key concept — the “problem background”: Originally how to solve altruism failures; can change over time as group size increases and cooperation challenges evolve. Kitcher discusses John Woolman’s Quaker anti-slavery arguments as an example of progressive recognition that failing to respond to slaves’ desires constituted an altruism failure.

Normative framework (Part III): Proposes “dynamic consequentialism” — specifying a conception of the good that best discharges the functions of ethics, and norms/rules that promote this good. Proposes an egalitarian conception of the good and a method of deliberation under conditions of mutual engagement. The “unseen enforcer” (positing divine authority) may have been effective at encouraging conformity but led ethics down a dead-end path.

E.O. Wilson’s sociobiology and ethical implications

Key publications: Sociobiology: The New Synthesis (1975), On Human Nature (1978, Pulitzer Prize 1979), and the co-authored paper with Michael Ruse, “Moral Philosophy as Applied Science” (Philosophy 61, 1986).

The famous quote: From Sociobiology (1975, Ch. 27): “Scientists and humanists should consider together the possibility that the time has come for ethics to be removed temporarily from the hands of the philosophers and biologicized.”

Core claims: (a) Moral sentiments are products of natural selection: “Innate censors and motivators exist in the brain that deeply and unconsciously affect our ethical premises; from these roots, morality evolved as instinct” (On Human Nature). (b) “Human behavior is ultimately the means by which human genetic material is kept intact. Morality has no other demonstrable ultimate function.” (c) Genes hold culture on a “genetic leash” — culture evolves but its trajectories are powerfully guided by inborn biases.

The Ruse-Wilson paper (1986) — “Deceived by our genes”: Wilson and Ruse argued that the sense of objective moral obligation is an adaptive illusion produced by natural selection. We have been “deceived by our genes” into believing morality objectively binds us. This is a form of moral anti-realism/error theory: the felt bindingness of morality is a biological trick to promote cooperation.

The controversy: The Sociobiology Study Group (led by Richard Lewontin and Stephen Jay Gould) condemned sociobiology as providing genetic justification for social inequalities. At a 1978 AAAS meeting, protesters threw water on Wilson. Accusations of genetic determinism, racism, and sexism — charges Wilson denied and that were unfounded.

Descriptive vs. normative ambiguity: Wilson’s project oscillates between: (1) biology can improve ethics by helping us recognize Stone-Age moral biases (descriptive project); (2) biology reveals that objective moral claims are illusory (normative/metaethical project). Critics invoke Hume’s is-ought problem. The mathematics analogy cuts against Wilson: our mathematical capacities are also evolved, but this does not undermine the objectivity of mathematics — so why should the evolutionary origin of moral sentiments undermine ethics?

Paul Bloom’s Against Empathy (2016)

Paul Bloom, Against Empathy: The Case for Rational Compassion (Ecco/HarperCollins, 2016). Bloom was professor of psychology and cognitive science at Yale. Precursor article: “Against Empathy,” Boston Review (2014).

Narrow definition of empathy: Bloom targets emotional empathy (empathic concern) — “the act of coming to experience the world as you think someone else does,” literally feeling another’s pain. He explicitly distinguishes this from cognitive empathy (understanding what others think/feel without sharing the feeling) and compassion (caring about others’ well-being without vicariously experiencing their suffering).

Argument structure:

(a) Empathy is a poor moral guide: Bloom describes empathy as “a spotlight directing attention and aid to where it’s needed” — but spotlights illuminate narrow areas while leaving everything else in darkness.

(b) Empathy is biased: Biased toward the identifiable over the statistical (we respond to one named victim, not millions of anonymous ones); biased toward the similar (we empathize more with those who look like us); biased toward the proximate (nearby suffering over distant suffering).

(c) Empathy is innumerate (the “spotlight effect”): Empathy focuses on individuals, not statistics. It cannot scale. Key example: after the Sandy Hook shooting, the affluent town was overwhelmed with children’s gifts, often from poor donors — a massive empathy-driven misallocation of resources while equivalent suffering elsewhere went unaddressed.

(d) Empathy can motivate cruelty: Empathy for in-group victims can fuel desire for revenge and aggression against out-groups. Empathic distress leads to burnout and withdrawal.

(e) The alternative — “rational compassion”: Following Buddhist moral philosophy (citing Charles Goodman’s Consequences of Compassion), Bloom distinguishes empathy from compassion. Compassion is “simply caring for people, wanting them to thrive” — without the vicarious experience. Neuroscience supports the distinction: empathy and compassion activate different brain networks. Compassion is associated with positive affect and wellbeing; empathic distress leads to burnout. Bloom advocates a utilitarian-leaning approach: decisions based on cost-benefit analysis, not felt emotional resonance.

Nietzsche’s On the Genealogy of Morality (1887)

The three essays:

First Essay: “‘Good and Evil,’ ‘Good and Bad’” (GM I): Traces two moral systems. The aristocratic warrior class (“masters”) designated their own qualities — strength, power, health, nobility — as “good” (gut), and the qualities of the common as merely “bad” (schlecht), an afterthought denoting what is low or pitiable. Nietzsche supports this etymologically: in multiple languages, words for “good” derive from “noble” or “well-born,” while “bad” derives from “common” or “plebeian” (GM I: §§4–5).

The slave revolt in morality begins when the oppressed, unable to discharge hostility through direct action, develop ressentiment — a festering, creative hatred. “Ressentiment becomes creative and gives birth to values” (GM I: §10). The slaves invert the masters’ table: they relabel strength as “evil” (böse) and recast their own weakness, humility, and meekness as “good.” This is fundamentally reactive rather than active valuation — slave morality defines itself negatively by saying “No” to what is external.

Ressentiment differs from simple resentment: (1) it is a consuming, dominant passion, whereas the masters’ contempt is an afterthought; (2) it is a reactive affect produced in reaction to masters’ behavior, whereas masters’ values spring spontaneously; (3) ressentiment labels masters “evil,” while contempt labels slaves merely “bad.”

The priestly caste (especially the Jewish priestly caste) serves as the agent of transvaluation. Priests are physically weaker but intellectually subtler; their ressentiment is deeper and more spiritualized. “With the Jews there begins the slave revolt in morality” (GM I: §7).

The bird of prey and lamb parable (GM I: §13): Lambs naturally dislike birds of prey and conclude that being a bird of prey is evil. They infer that the opposite — being a lamb — is “good.” But this requires the fiction that the bird of prey could choose not to be one, that there is a neutral “subject” behind its actions. This grammatical-metaphysical fiction (the subject-predicate structure of language) creates the concept of the “soul” or “free will,” which makes the strong culpable for being strong. Slave morality “detaches subject from predicate, doer from deed,” enabling blame.

Second Essay: “‘Guilt,’ ‘Bad Conscience,’ and Related Matters’” (GM II): “Guilt” (Schuld) originally had no moral meaning — it derived from the material concept of “debt” (Schulden). A debtor was “guilty” and the creditor could exact punishment — not to produce remorse but to bring the creditor pleasure. Bad conscience emerges when humans are enclosed within society and can no longer discharge aggressive instincts outward. These instincts turn inward: “all instincts that do not discharge themselves outwardly turn inward” (GM II: §16). This internalization creates the “soul,” self-consciousness, and the capacity for self-torture. The concept of debt to ancestors evolves into religious guilt (debt to God), culminating in Christian original sin — an unpayable debt producing maximum self-torture.

Third Essay: “What Do Ascetic Ideals Mean?” (GM III): The ascetic priest takes the suffering of the weak and gives it meaning through the ascetic ideal: suffering is punishment for sin. “The meaninglessness of suffering, not suffering itself, was the curse that lay over mankind so far” (GM III: §28). The book’s final line: “man would rather will nothingness than not will.”

The genealogical method itself: Traces the historical emergence of moral concepts to reveal their contingency — showing that what appears natural or timeless is the product of specific historical power dynamics and psychological mechanisms. Genealogy does not merely commit the genetic fallacy; Nietzsche aims to evaluate values: “a critique of moral values, the value of these values themselves must first be called in question” (Preface §6). The method connects to Nietzsche’s broader perspectivism — there is no view from nowhere — and supports the “revaluation of all values” (Umwerthung aller Werthe).

Key scholarly interpretations: Brian Leiter’s naturalistic reading (Nietzsche on Morality, Routledge, 2002; 2nd ed. 2015) is the most influential anglophone interpretation. Leiter argues Nietzsche is primarily a methodological naturalist who holds a “doctrine of types” (each person has a fixed psycho-physical constitution) and targets “morality in the pejorative sense” (MPS) — morality characterized by the presupposition of free will, universality claims, and norms of altruism/equality that harm the flourishing of “higher types.” Contrasting readings: Alexander Nehamas emphasizes aestheticism/perspectivism; Maudemarie Clark stresses evolving views on truth; Deleuze and Foucault emphasize radical critique of subjectivity and power.

Foucault’s genealogical method applied to morality

Debt to and departure from Nietzsche: Foucault’s key programmatic text is “Nietzsche, Genealogy, History” (1971). Key borrowings: genealogy opposes the search for fixed origins (Ursprung); it traces descent (Herkunft) and emergence (Entstehung) — messy, contingent, power-laden processes. Key departures: Foucault avoids the will to power as a singular explanatory principle, analyzing more diffuse, relational, institutional forms of power. In Nietzsche’s genealogy, the weaker class imposes its morality; in Foucault’s, it is typically the more powerful class consolidating domination through moral categories. Foucault is less interested in origins and more in how power relations continuously constitute moral subjects in the present.

Discipline and Punish (1975): Traces the transformation from sovereign punishment (public torture as spectacle of sovereign power) to disciplinary power (the modern prison, normalization). The modern prison is not simply more humane; it is a new technology of power: “to punish less, perhaps; but certainly to punish better.” Disciplinary techniques produce “docile bodies” through surveillance, examination, normalization, hierarchical observation. The panopticon (Bentham’s design) serves as model: inducing self-regulation through internalized surveillance. Moral categories of normal/abnormal, deviant/healthy are produced through disciplinary mechanisms and human sciences (criminology, psychology, psychiatry) — not discovered by objective science.

The History of Sexuality, Vol. 1 (1976): The “repressive hypothesis” (that sexuality was repressed since the Victorian era) is wrong. Since the 17th century, there has been a proliferation of discourses about sex. Power/knowledge (pouvoir/savoir): “sexuality” is not a biological given but a historical construct produced through discourse. Medical, psychological, and juridical discourses didn’t describe pre-existing sexual identities — they constituted them. Power operates through incitement (making people confess desires and submit to expert interpretation), not merely prohibition.

Late ethics — The Use of Pleasure (1984) and The Care of the Self (1984): A dramatic turn toward ancient Greek and Roman ethics. Foucault analyzes through four dimensions: (1) ethical substance (ontologie) — what aspect of oneself is the material of moral conduct; (2) mode of subjection (déontologie) — how one relates to moral rules (for Greeks, voluntary aesthetic choice, an “aesthetics of existence”); (3) ethical work (ascétique) — self-forming activities (askēsis); (4) telos — the mode of being aimed at (a beautiful, moderate, free existence). Ancient ethics comprised “technologies of the self” — practices of self-transformation as part of a freely chosen art of living, irreducible to subjection by external power. This reveals a domain of practices of freedom distinct from modern obligation-centered morality.

Bernard Williams’s Ethics and the Limits of Philosophy (1985)

The Socratic question: Williams argues the proper starting point for ethics is Socrates’ question: “How should one live?” (Chapter 1). This is broader than modern moral philosophy’s focus on obligation. Moral philosophy since Kant has narrowed its focus to duty, losing this richer question.

Critique of “the morality system” (Chapter 10): Williams coins “the morality system” for the specific modern institution he opposes. Its features: (a) obligation-centeredness — moral obligation is the central concept; (b) the “peculiar institution” of moral obligation — treated as categorically binding, overriding all other considerations (“supremacy”); (c) blame — agents are blameworthy whenever they violate moral obligations, requiring a “voluntary” conception of action tied to blame and punishment; (d) reductiveness — all ethical considerations reduced to thin moral principles. Williams targets both Kantian and utilitarian theories as instances of the morality system.

Thick vs. thin ethical concepts (Chapter 8): Thick ethical concepts (e.g., cruel, brave, courageous, generous, treacherous, cowardly, chaste, lewd) are simultaneously “world-guided” and “action-guiding” (Williams 1985: 140). Their application depends on how the world is (descriptive content), and if a thick concept applies, it typically provides a reason for action. Thin ethical concepts (e.g., good, right, wrong, ought) are purely evaluative — action-guiding but not (or minimally) world-guided. Williams argues thick concepts “refute the simplest” fact/value distinction (1985: 140–5): to apply a thick concept correctly, one must simultaneously make descriptive and evaluative judgments; one cannot first describe facts then add evaluation.

Confidence vs. knowledge: Williams is skeptical about moral knowledge in a strong sense. He argues “reflection can destroy knowledge” in ethics (1985: 148) — when a society becomes reflectively aware that its ethical commitments are historically contingent, it cannot simply reaffirm them as objective knowledge. But ethical life remains possible. Williams distinguishes knowledge from confidence: a society can still live ethically with confidence — genuine but non-epistemically-guaranteed commitment to its values. Ethical life requires confidence, not certainty.

Williams’s “ground projects” concept

Appears primarily in “Persons, Character and Morality” (in Moral Luck, Cambridge University Press, 1981), connecting to arguments in “A Critique of Utilitarianism” (in Utilitarianism: For and Against, co-authored with J.J.C. Smart, 1973).

The concept: Ground projects are commitments, convictions, and pursuits that give a person’s life meaning and point — what make life worth living for that individual. They constitute one’s deepest identification and connect to “categorical desires”: desires that are not conditional on being alive but provide the reason for going on living. A person without ground projects would have no reason to continue living. Ground projects are not mere preferences; they are constitutive of identity and character: a creative vocation, devotion to a cause, deep relationships, commitment to pacifism.

The argument against utilitarianism: In “A Critique of Utilitarianism” (1973), Williams presents George (a chemist opposed to chemical weapons offered a job in a weapons lab) and Jim (who must choose whether to kill one person to save nineteen). Williams argues utilitarian answers violate integrity: (a) negative responsibility — utilitarianism holds agents responsible for what they fail to prevent, making the agent’s own actions have no special status; (b) integrity — demanding George take the job demands he abandon his ground project (pacifist commitment), destroying the very thing that gives his life meaning. Any moral theory claiming unconditional authority over agents — demanding abandonment of ground projects when calculation requires it — treats agents as “channels” for optimal outcomes rather than persons with their own character.

Connection to internal reasons: Williams’s internalism about reasons (“Internal and External Reasons,” in Moral Luck, 1981): an agent has a reason to act only if the action connects to the agent’s existing “subjective motivational set” (S) — desires, dispositions, commitments, emotional patterns. External reasons (applying regardless of motivational set) do not exist. Ground projects are central elements of S, providing genuinely one’s own reasons for action. This is why morality cannot simply override them.

The thick/thin ethical concepts debate (broader)

Williams’s formulation (1985): As above — thick concepts are both world-guided and action-guiding; thin concepts are purely evaluative. Thick concepts resist the fact/value distinction.

Philippa Foot’s role: In “Moral Arguments” (1958) and “Moral Beliefs” (1958–59), Foot argued moral terms have factual criteria for application. Natural Goodness (2001) develops a neo-Aristotelian account: moral goodness is a form of natural goodness, analogous to how a good specimen of a natural kind exemplifies species-appropriate functioning. Evaluation is inseparable from description of what a thing of that kind naturally does.

John McDowell’s contributions: In “Non-Cognitivism and Rule-Following” (1981), McDowell develops the anti-disentangling argument (the “shapelessness thesis”): Non-cognitivists (like Hare) claim thick concepts have separable descriptive and evaluative components. McDowell argues that if this were true, an “unsympathetic outsider” — someone not sharing the evaluative perspective — could master the extension of the concept by grasping only the descriptive component. But this is implausible: an outsider could not anticipate insiders’ use of “chaste” or “lewd” in new cases without grasping the evaluative perspective. Thick concepts are “shapeless” from a non-evaluative point of view: no purely descriptive pattern captures all and only things the concept applies to. Evaluation “drives the extension.” If thick concepts are shapeless, they cannot be disentangled, and non-cognitivism is undermined. McDowell draws on Wittgenstein’s rule-following considerations.

In “Values and Secondary Qualities” (1985), McDowell argues values are analogous to secondary qualities (colors): real properties perceptible only by creatures with appropriate sensibility.

The disentangling debate: Simon Blackburn (“Reply: Rule-Following and Moral Realism,” 1981; “Disentangling Disentangling,” 2013) argues evaluation “drives” but doesn’t “constitute” extension. Counter-example: “cute” used to appraise subservient behavior in women — if we cannot disentangle, we cannot properly criticize this use, since users could claim critics “failed to grasp the concept.” Hilary Putnam (The Collapse of the Fact/Value Dichotomy, 2002) defends inseparability.

Why this matters: If thick concepts resist disentangling: (1) the fact/value distinction is undermined; (2) non-cognitivism is challenged; (3) the is-ought gap is complicated; (4) moral realism gains support.

Williams’s Truth and Truthfulness (2002)

Central paradox: Contemporary culture simultaneously demands truthfulness (everyone fears deception) while expressing skepticism about truth (influenced by Nietzsche, Foucault, postmodernism, Rorty). Williams calls this self-undermining.

Genealogical method — the crucial innovation: Williams defines: “A genealogy is a narrative that tries to explain a cultural phenomenon by describing a way in which it came about, or could have come about, or might be imagined to have come about” (p. 20). Key distinctions:

• “Real” vs. “fictional” genealogies: Real genealogies trace actual historical processes (Nietzsche on slave morality, Foucault on prisons). Fictional genealogies construct imagined narratives (like social contract theories) to illuminate a practice’s functional value.
• “Vindicatory” vs. “debunking” genealogies: A debunking genealogy (Nietzsche’s Genealogy) traces a value to “shameful origins” (ressentiment, domination) to undermine its authority. A vindicatory genealogy shows that even after understanding contingent, functional origins, we have as much reason to respect the value as before. Williams’s genealogy of truthfulness is explicitly vindicatory.

The State of Nature narrative (Chapters 2–3): Williams imagines “a small society of human beings sharing a common language, with no elaborate technology and no form of writing” (p. 41). Two virtues naturally emerge: (1) Accuracy — the disposition to take care in forming beliefs, because some have “purely positional advantages” (they witnessed the event); (2) Sincerity — the disposition to assert what one actually believes (“trustworthiness in speech,” p. 94). Williams argues these virtues have genuine instrumental value that generates intrinsic value: trust requires that dispositions of accuracy and sincerity become stable character traits, intrinsically valued, not merely strategically adopted. The functional need for truth-telling generates dispositions that transcend their instrumental origins.

How Williams differs from Nietzsche and Foucault: Both Nietzsche’s and Foucault’s genealogies are primarily debunking/destabilizing. Williams’s is vindicatory: it uses the genealogical method not to debunk but to vindicate. The fact that values are historically contingent and functionally explicable does not mean they are not genuinely valuable.

Glaucon’s challenge in Plato’s Republic (Book II)

The tripartite classification of goods (357b–358a): Glaucon distinguishes: (a) pure intrinsic goods — chosen for their own sake, not consequences (joy, harmless pleasures); (b) mixed goods — valued both for their own sake and consequences (knowledge, sight, health); (c) pure instrumental goods — chosen only for consequences (physical training, medical treatment). Socrates declares justice belongs to (b) — the highest class. Glaucon reports “the many” classify justice as (c) — onerous, practiced reluctantly only for rewards.

The three-part devil’s advocacy (358e–362c):

First — social contract origin story (358e–359b): Justice originates as a compromise. By nature, doing injustice to others is good and suffering injustice is the worst. Since most people cannot consistently get away with injustice, they agree to neither do nor suffer it — a compact. Justice is a “mean” between the best (doing injustice without penalty) and the worst (suffering injustice without recourse). It is “strictly conventional in nature.”

Second — the Ring of Gyges (359c–360d): A shepherd discovers a ring granting invisibility. Using it, he seduces the queen, kills the king, takes the throne. The thought experiment: give both a just and unjust man such a ring. Both would act unjustly. No one is just willingly — people are just only because they lack power to be unjust with impunity.

Third — comparison of lives (360e–362c): The perfectly unjust person with a reputation for justice is happier than the perfectly just person saddled with a reputation for injustice. This strips away all extrinsic rewards to test whether justice is intrinsically valuable.

Adeimantus’s supplement (362d–367e): All existing defenses of justice — by fathers, poets, theologians — really praise the reputation for justice, not justice itself.

Socrates’ response: Proposes examining justice “writ large” in the city before transferring to the individual soul. This yields: (a) the tripartite soul (Book IV) — reason, spirit (thumos), appetite; (b) justice as psychic harmony — each part performing its proper function; (c) the tyrannical soul (Book IX) as maximally miserable — enslaved to appetites, internally chaotic.

The “Sachs problem”: David Sachs argued Socrates commits a fallacy: “Platonic justice” (psychic harmony) doesn’t map onto “vulgar justice” (not stealing, not lying). This remains debated.

Foundational status: Glaucon’s challenge establishes the template for all subsequent “Why be moral?” debates and recurs in Hobbes’s Foole, Hume’s sensible knave, and game-theoretic free-rider problems.

Peter Singer’s The Expanding Circle (1981; expanded edition 2011)

Main thesis: Moral progress consists in the progressive expansion of the circle of moral concern — from self to kin to tribe to nation to all humanity to all sentient beings. This is not merely change but genuine progress, driven by reason.

The evolutionary foundation: Singer accepts sociobiology’s account of altruism’s origins: kin selection (Hamilton) explains altruism toward genetic relatives; reciprocal altruism (Trivers) explains cooperation with non-kin. But biology alone cannot explain human ethics.

The “escalator of reason” (Chapter 4): Just as getting on an escalator to travel a short distance may carry you further than intended, once you begin reasoning morally — offering disinterested justifications — reason takes on “a logic of its own.” The demand for impartial justification pushes toward universality. If I justify my action by saying “it serves my interests,” I must explain why my interests count more than yours. This generates pressure toward the impartial standpoint. Singer draws on Sidgwick’s “point of view of the universe” and the universalizability principle. Key passage: “Once reasoning has got started it is hard to tell where it will stop.”

Reason transcends evolution: Singer addresses the debunking tension: reason transcends its evolutionary origins. Just as our mathematical capacity evolved for survival but discovers truths about abstract objects, our moral reasoning evolved for group living but can push us toward genuinely defensible moral positions. Evolution explains the origin of moral thinking but does not determine its content or limits.

Connection to utilitarianism: The only principled stopping point for moral concern is sentience. This connects to Animal Liberation (1975) and Singer’s broader utilitarian project.

2011 Afterword: Singer more firmly endorses objective normative truths, engages with Haidt’s social intuitionist model, and maintains that reason can and should override evolved moral intuitions.

Richard Joyce’s moral fictionalism

Key works: The Myth of Morality (Cambridge University Press, 2001), The Evolution of Morality (MIT Press, 2006), Essays in Moral Skepticism (Oxford University Press, 2016), Morality: From Error to Fiction (Oxford University Press, 2024).

Moral error theory: Following Mackie, Joyce holds all first-order moral claims are systematically false. Moral discourse is genuinely truth-apt (contra non-cognitivism) but the world lacks the properties needed. Specifically, moral claims presuppose objective prescriptivity — categorical imperatives applying regardless of desires — and no such properties exist. Analogy: just as atheism holds “God loves you” is truth-evaluable but false, moral error theory holds “Stealing is wrong” is truth-evaluable but false.

The “now-what” problem — three options: (1) Abolitionism — abandon moral discourse entirely; (2) Conservationism — continue believing moral claims even knowing they’re false; (3) Fictionalism — continue using moral discourse without believing it. Joyce advocates (3): (1) would be too costly (we lose practical benefits); (2) is epistemically irrational.

The critical hermeneutic/revolutionary distinction (terminology from John Burgess 1983, extended by Burgess and Gideon Rosen 1997):

• Hermeneutic fictionalism: Moral discourse already is fictional. People don’t really believe moral claims; they use them as make-believe. Joyce rejects this as descriptively implausible.
• Revolutionary fictionalism: Moral discourse currently involves genuine (but false) beliefs. We should reform our practice — moving from sincere belief to deliberate make-believe. Joyce endorses this. The “revolution” is in our attitude, not in what we say.

The pragmatic case: (a) Commitment devices — “The distinctive value of categorical imperatives is that they silence calculations.” Believing an act is wrong provides stronger motivation than purely prudential reasoning. (b) Bolstering self-control — moral beliefs function as bulwarks against weakness of will (akrasia). (c) Facilitating cooperation — moral discourse coordinates expectations and underwrites social cooperation.

Joyce’s account of fictive assertion: Joyce endorses a force-modifier account: when one fictively asserts “Stealing is wrong,” one pretends to assert the proposition, modifying the illocutionary force. It may not feel like pretense — it operates like deeply habituated make-believe.

Key objections: (a) The “bad faith” objection — Can a fiction motivate if you know it’s fictional? Joyce responds that fictive engagement can be psychologically powerful (novels, films genuinely affect emotions and behavior). (b) The instability objection — Won’t fictionalism collapse into genuine belief or abandonment? Joyce acknowledges this is the most serious challenge but argues we sustain other useful fictions without collapse. (c) Mark Kalderon’s hermeneutic fictionalism (Moral Fictionalism, Oxford University Press, 2005) — Kalderon defends the view that moral discourse already involves make-believe, not genuine belief, which Joyce finds descriptively implausible.

Historical precursor — Vaihinger: Hans Vaihinger, Die Philosophie des Als Ob (1911; English The Philosophy of ‘As If’, 1924). Vaihinger argued humans deploy consciously false assumptions (“fictions”) as pragmatic instruments. Scientific theories, mathematical idealizations, ethical concepts, even fundamental categories function as useful fictions. We must act “as if” ethical certainty were possible. This provides the most direct historical precedent for Joyce’s revolutionary moral fictionalism.

MacAskill, Bykvist, and Ord — Moral Uncertainty (2020)

William MacAskill, Krister Bykvist, and Toby Ord, Moral Uncertainty (Oxford University Press, 2020; open-access under CC BY-NC-ND 4.0).

Central problem: How should you act when uncertain not about empirical facts but about which moral theory is correct?

Maximize Expected Choiceworthiness (MEC): The book’s central proposal. The agent: (1) assigns credences (probability weights) to competing moral theories; (2) each theory assigns choiceworthiness values to available options; (3) the agent should choose the action that maximizes expected choiceworthiness — the credence-weighted sum across theories. Structurally analogous to expected utility theory but applied across moral theories.

Against “My Favourite Theory” (MFT): Just follow whichever theory you find most credible. Rejected because it ignores stakes. If credence is 51% Theory A / 49% Theory B, MFT says follow A even if A says stakes are trivially low and B says they are astronomically high. This seems reckless.

Technical challenges: (a) Intertheoretic comparisons — How compare “units” of choiceworthiness across theories? Authors propose a universal scale account and statistical normalization methods guided by the principle of equal say: for a given quantity of credence, every theory should exert the same degree of influence. (b) Merely ordinal theories — Some theories only rank options without specifying how much better. Chapter 3 uses tools from social choice theory (Borda Rule, Condorcet extensions). (c) Absolutist/deontological theories — The choiceworthiness of respecting a categorical imperative (conditional on Kantianism being true) may be very high, meaning even modest credence in Kantianism significantly influences MEC calculation.

Moral hedging: The practical upshot — even 10% confidence in a theory that claims very high stakes should receive significant weight. This produces natural “moral hedging.”

The fanaticism problem: If a theory claims infinitely high stakes, even tiny credence could dominate. Authors suggest the same solutions as for empirical fanaticism in standard decision theory.

Connection to effective altruism: The framework supports EA’s approach: given uncertainty, hedge across theories, which often recommends high-impact, broadly endorsed interventions. Prior work: Ted Lockhart’s Moral Uncertainty and Its Consequences (Oxford University Press, 2000) was the first book-length treatment.

The “Why be moral?” question — taxonomy of responses

Kantian response: Morality is constitutive of rationality. The categorical imperative is derivable from the structure of practical reason. To be irrational is to violate moral duty. An amoral person is irrational, not just immoral. Critique: rests on a controversial view of reason; purely instrumental conceptions create no such obligation.

Humean response: Morality serves our sentiments, desires, and interests. Being moral typically makes us happier and more socially successful. But for the “sensible knave” who “observes the general rule, and takes advantage of all the exceptions” (Enquiry Concerning the Principles of Morals IX.ii), there may be no rationally compelling answer. Hume admits: “if a man think that this reasoning much requires an answer, it would be a little difficult to find any which will to him appear satisfactory and convincing.” The answer is conditional: morality is rationally required if you have the right desires/sentiments.

Aristotelian/eudaimonist response: Morality is constitutive of human flourishing (eudaimonia). Virtues are not mere means to happiness but components of the good life. The just person is the happy person. Asking “Why be moral?” is like asking “Why flourish?” Critique: relies on a thick conception of human nature that some find metaphysically loaded.

The free rider problem: Even if moral cooperation is collectively beneficial, each individual has an incentive to defect — the modern version of the Ring of Gyges. Responses: (1) repeated-game arguments; (2) moral character arguments (cheating corrupts character); (3) Kantian universalizability; (4) evolutionary/psychological arguments (genuine moral emotions make free-riding psychologically costly).

Metaethical dimensions: Moral realists can say moral facts constitute genuine reasons for action. Anti-realists face harder challenges: error theorists appeal to pragmatic reasons; expressivists say “because you have pro-attitudes toward morality”; constructivists say “because you construct moral requirements through rational self-legislation.”

Moral disagreement and moral progress

The argument from disagreement (against moral realism): Traces to Sextus Empiricus and Mackie (1977). The best explanation of persistent, deep moral disagreement across cultures is that there are no objective moral facts. Requires an epistemic premise: if moral facts existed and were accessible, we would expect convergence; we don’t observe sufficient convergence; therefore moral facts likely don’t exist.

Realist responses: (a) Disagreement doesn’t entail no facts — disagreement exists in science too (David Enoch, “How Is Moral Disagreement a Problem for Realism?” 2009). (b) Significant moral convergence exists — slavery, religious toleration, women’s rights now far less controversial (Michael Smith 1994, Nicholas Sturgeon 1988). (c) Disagreement is explicable by distorting factors — cognitive biases, self-interest, limited information, cultural conditioning.

Anti-realist responses: Best explanation argument — the most parsimonious explanation is no moral facts. Expert disagreement is especially problematic (Brian Leiter 2014, 2021): if even expert moral philosophers disagree persistently after millennia, the inference to no objective facts is strong. Observed convergence may be explained by shared cultural influences, globalization, or evolutionary psychology rather than tracking moral truth (Hopster 2019, Fitzpatrick 2014).

Moral progress: Historical expansion of moral concern (abolition of slavery, suffrage, human rights, animal welfare). Realist view: genuine progress analogous to scientific progress — humanity gradually discovers moral truths. Anti-realist view: change by current standards, not truth-tracking progress. Singer’s expanding circle thesis is compatible with, and arguably supports, a form of moral realism.

Ronald Dworkin’s Justice for Hedgehogs (2011)

Harvard University Press, 2011. The title references Isaiah Berlin’s hedgehog/fox distinction; Dworkin claims to know one big thing: the unity of value.

The “one-system” / interpretive view: Moral judgments are neither descriptions of Platonic reality nor mere expressions of attitude. They are interpretive judgments — claims that aim to show moral concepts in their best light. Moral concepts are “interpretive concepts” rather than “criterial concepts”: their meaning requires value-laden interpretation.

Rejection of both external moral realism and anti-realism: Dworkin identifies and rejects: (a) external error skepticism (Mackie-style error theory); (b) external status skepticism (moral judgments only appear to be about truths but actually express attitudes); (c) external moral realism (Platonic “morons” or moral particles).

“Hume’s principle” and internal justification: Moral claims can only be justified by other moral claims — no metaphysical, scientific, or empirical fact can support or undermine a moral conclusion by itself. “Moral truths cannot be dislodged by empirical or metaphysical truths.”

Against “moral Archimedeans”: Dworkin attacks those who try to ground or debunk morality from an Archimedean standpoint outside moral practice. Neither the empiricist (“science shows morality is an illusion”) nor the metaphysician (“moral facts exist as Platonic forms”) makes a coherent move.

Unity of value (the “hedgehog” thesis): Genuine values cannot conflict and are necessarily mutually supportive. Ethics (how to live) and morality (how to treat others) cohere. Justice, liberty, equality, and democracy fit into a single integrated network. Living well means creating a life whose “performance value” matters alongside its “product value.”

Connection to legal philosophy: In Law’s Empire (1986), Dworkin proposed “law as integrity.” In Justice for Hedgehogs, he revises: law is simply “a branch, a subdivision, of political morality.” The concept of law is an interpretive concept.

Difference from constructivism and error theory: Unlike constructivism, Dworkin doesn’t claim moral truths are “constructed” by a procedure — they are discovered through interpretation. Unlike error theory, he vindicates “the ordinary view that moral convictions are straightforwardly true or false.” His position sustains human rights discourse because rights claims are genuinely truth-apt moral claims, justified within the moral framework.

P.F. Strawson’s “Freedom and Resentment” (1962)

Originally delivered as the Proceedings of the British Academy, 48 (1962): 1–25.

Main argument: Moral responsibility is grounded in our reactive attitudes — resentment, gratitude, indignation, love, hurt feelings, moral disapproval, guilt — rather than in metaphysical facts about free will. These attitudes constitute a “web” of interpersonal responses forming the natural foundation of moral responsibility practices.

Participant stance vs. objective stance: The participant stance: we engage with others as fellow agents, attributing good or ill will, holding them to moral expectations, responding with reactive attitudes. The objective stance: we treat a person as a natural phenomenon to be managed, predicted, or controlled — “handled” rather than engaged with as a moral equal.

Quality of will: Central to Strawson’s account — reactive attitudes track whether another person’s actions manifest goodwill, ill will, or indifference toward us. An accidental injury provokes less resentment than one caused by “contemptuous disregard” or “a malevolent wish to injure.”

Reactive attitudes as constitutive: Reactive attitudes are constitutive of moral responsibility, not merely evidence for it. Many interpreters read Strawson as proposing a “reversal”: rather than reactive attitudes being appropriate because someone is responsible, someone is responsible because they are apt targets of reactive attitudes.

The free will debate rendered practically irrelevant: Strawson argues: (a) determinism does not show that all human behavior manifests ill will — it does not generalize any excusing condition; (b) it is practically inconceivable that we could abandon reactive attitudes — too deeply rooted in human nature; (c) it would be irrational to abandon them since they constitute meaningful interpersonal relationships. “It cannot be a consequence of any thesis which is not itself self-contradictory that abnormality is the universal condition.”

Excuses vs. exemptions (terminology later formalized by Gary Watson): Excuses (Type 1): “He didn’t mean to,” “It was an accident.” These show the agent’s behavior didn’t actually manifest ill will. The agent remains within the moral community. Exemptions (Type 2): mental illness, extreme youth, severe cognitive impairment. These show the agent is not an apt target of demands for moral regard. We adopt the objective stance.

Influence: R. Jay Wallace, Responsibility and the Moral Sentiments (1994), extends Strawson by arguing accountability depends on fairness — whether it’s fair to hold people responsible given their rational competence. T.M. Scanlon’s contractualism draws on interpersonal accountability. Michael McKenna’s Conversation and Responsibility (2012) develops the Strawsonian framework by analogy with conversational exchange.

Companions-in-guilt strategy

A defensive strategy primarily used by moral realists against moral skeptics, error theorists, and evolutionary debunking arguments. Core idea: if arguments against moral facts are sound, they should equally undermine other normative domains (especially epistemic norms) — but since we cannot abandon epistemic norms without self-defeat, the arguments must fail.

Key proponents: Terence Cuneo, The Normative Web (Oxford University Press, 2007) — the most elaborate book-length treatment. Hallvard Lillehammer, Companions in Guilt (Palgrave Macmillan, 2007). Also: Renford Bambrough (1979), Russ Shafer-Landau (2003), Richard Rowland (2012, 2016), Ramon Das (2016, 2017).

Cuneo’s core argument (The Normative Web, p. 6): (1) If moral facts do not exist, then epistemic facts do not exist. (2) Epistemic facts exist. (3) So moral facts exist [from 1, 2 by modus tollens]. (4) If moral facts exist, then moral realism is true. (5) So moral realism is true.

Defense of premise (1) — the parity thesis: Features anti-realists find objectionable about moral facts equally characterize epistemic facts: (a) supervenience on descriptive facts; (b) intrinsic motivational force; (c) categorical/inescapable character; (d) metaphysical “queerness”; (e) failure to play explanatory roles; (f) pervasive disagreement.

Defense of premise (2): Epistemic nihilism (error theory about epistemic facts) is self-defeating — if there are no epistemic facts, there can be no reason for believing anything, including epistemic nihilism.

Against evolutionary debunking specifically: If evolutionary debunking arguments show evolutionary origins of moral beliefs undermine justification, the same applies to epistemic beliefs — since our epistemic capacities are also evolved. But abandoning epistemic norms is self-defeating: the debunking argument itself relies on epistemic norms. Therefore the debunking argument “proves too much.”

Two types (Lillehammer): (1) Arguments by analogy — moral claims share features with other claims whose credentials are not in question; (2) Arguments by entailment — denying moral facts logically entails denying epistemic facts.

Major critics: Christopher Cowie (2014, 2016); Jonas Olson (2014); Chris Heathwood (2009).

Kahane’s debunking argument structure

Guy Kahane, “Evolutionary Debunking Arguments,” Noûs 45(1): 103–125 (2011).

General structure of causal debunking arguments: Causal Premise: S’s belief that p is explained by [causal process] X. Epistemic Premise: X is an off-track process (does not track the truth of p-type claims). Conclusion: S’s belief that p is unjustified.

Key distinction: “on-track” processes = truth-tracking (e.g., visual perception under normal conditions); “off-track” processes = not truth-tracking (e.g., wishful thinking, evolutionary pressures unrelated to truth).

Targeted vs. global distinction: Targeted (local) EDAs — used within normative ethics to debunk specific moral intuitions (e.g., Greene’s argument that deontological intuitions in trolley cases are explained by evolved aversion to “up close and personal” violence). Global EDAs — used in metaethics to debunk all moral beliefs at once (Joyce, Street).

Key observation — presupposition of objectivism: EDAs presuppose the target is a metaethical objectivist/realist. A constructivist (like Street) who believes moral properties are constructed from whatever evaluative attitudes we have will be unpersuaded — for them, evolution tracking our attitudes is tracking moral truth.

The incompatibility problem: Targeted EDAs are incompatible with the global EDA. If the global EDA works, it debunks all evaluative beliefs, including replacements the targeted debunker relies on. If you resist the global EDA, the methods you use likely also protect the targeted beliefs.

Success conditions: For an EDA to succeed: (a) establish the causal premise with empirical rigor; (b) show the off-track process completely explains the belief, excluding on-track influences; (c) the target must be an objectivist/realist.

Metaethics of animal ethics

How metaethical positions affect animal ethics: Moral realism: if moral facts exist, whether animals are moral patients depends on those facts’ content. Sentientism (capacity for suffering as criterion for moral status) typically presupposes some form of moral realism or cognitivism. Non-cognitivists must recast the question as whether we should extend patterns of approval/disapproval to animal treatment.

Peter Singer — preference utilitarianism and the argument from marginal cases: Animal Liberation (1975) popularized “speciesism.” Core principle: the Principle of Equal Consideration of Interests — morally relevant interests must be weighed equally regardless of species. The Argument from Marginal Cases: traits cited to justify human moral superiority (rationality, language) are absent in some humans (infants, severely cognitively impaired) yet present in some animals. Consistency demands extending moral consideration. Builds on Bentham (1789): “The question is not, Can they reason? nor, Can they talk? but, Can they suffer?”

Tom Regan — rights-based approach: The Case for Animal Rights (1983). Any being that is a “subject-of-a-life” — has beliefs, desires, perception, memory, emotional life, a welfare that matters to them — possesses inherent value and must be treated as an end, never merely as means. Rights are “trumps” against overall good promotion. Metaethical commitments: broadly moral realist — inherent value is an objective property.

Christine Korsgaard — Fellow Creatures (2018): A Kantian constructivist argument for animal moral status — remarkable because Kant himself denied direct duties to animals. Korsgaard’s argument: as rational beings, we need reasons for our actions, requiring us to suppose some ends are absolutely good. Without metaphysical insight into intrinsic values, all we have is that things are good-for or bad-for us. In taking these to be good absolutely, we take ourselves as ends in ourselves. But animals are no different in having things that can be good or bad for them. So we are committed to regarding all animals as ends in themselves. She identifies two senses of “end-in-oneself”: (1) autonomous beings who make laws for ourselves; (2) beings who have a good. Animals qualify under (2) — “passive ends” whose good we must respect.

Metaethics of AI moral status

Schwitzgebel and Garza, “A Defense of the Rights of Artificial Intelligences” (Midwest Studies in Philosophy 39, 2015): The No-Relevant-Difference Argument: If entity A deserves moral consideration, and entity B does not differ in any morally relevant respect, then B also deserves consideration. There are possible AIs who do not differ in any morally relevant respect from humans. Therefore there are possible AIs deserving similar consideration. Bolstered by a gradual neuron-replacement thought experiment.

The functionalist approach: If moral status depends on mental states (suffering, preferences, consciousness), AI could have moral status if it has the right functional properties. Chalmers (The Conscious Mind, 1996) argues consciousness might arise in silicon as well as carbon. If consciousness is substrate-independent, functionally appropriate AI could be genuinely conscious.

The anti-realism challenge: If there are no objective moral facts, AI moral status becomes a matter of attitudes — whether we choose to extend moral consideration. If moral status is response-dependent (Strawsonian reactive attitudes), the question becomes whether we should extend reactive attitudes to AI.

Schwitzgebel’s “weird minds” destabilization worry: Our moral intuitions, evolved for interactions with biological beings, may break down when confronted with the range of possible AI minds — analogous to how folk physics breaks down at extreme scales.

Anti-realism and human rights discourse

The core challenge: Human rights language presupposes objective moral truths (“inherent,” “inalienable” rights). If anti-realism is correct, the philosophical foundation may collapse.

Richard Rorty’s pragmatist approach (“Human Rights, Rationality, and Sentimentality,” 1993): We cannot find foundations for human rights. “The emergence of the human rights culture seems to owe nothing to increased moral knowledge, and everything to hearing sad and sentimental stories.” Rorty advocates sentimental education — using narrative, literature, and stories of suffering to expand our sense of who counts as “one of us.” Two conditions for a functioning human rights culture: security and sympathy.

Charles Beitz, The Idea of Human Rights (2009): A political/practice-based account avoiding metaphysical commitments. Human rights are part of a “global discursive practice” understood within the international system of states — “matters of international concern whose violation by governments can justify international protective and restorative action.” Part of the “political turn” in human rights philosophy (following Rawls).

Allen Buchanan and Russell Powell, The Evolution of Moral Progress (2018): A “biocultural” theory: moral progress (gains in inclusivity) is possible because humans have “adaptively plastic” capacities for both inclusion and exclusion, triggered by environmental conditions. When conditions are secure, inclusivist dispositions are evoked. The modern human rights movement exemplifies the most important gains in moral inclusivity.

Moral responsibility and punishment theory

Retributivism presupposes moral desert — the wrongdoer deserves proportionate punishment. This arguably requires: (a) genuine moral responsibility (some form of free will); (b) objective facts about what’s deserved (moral realism or robust cognitivism).

Consequentialist theories (deterrence, incapacitation, rehabilitation) are more compatible with anti-realism — they don’t require backward-looking moral desert, only forward-looking assessments.

Joel Feinberg’s expressivist theory (“The Expressive Function of Punishment,” The Monist 49, 1965): Punishment has an expressive function that mere penalties lack: it expresses the community’s “strong disapproval” — “a symbolic way of getting back at the criminal.” Inaugurated the “expressivist turn” in punishment theory. Later expressivists include Jean Hampton (moral education), Antony Duff (Punishment, Communication, and Community, 2001).

Strawson’s framework bridges blame and punishment: If moral responsibility is constituted by the aptness of reactive attitudes like indignation and resentment, punishment can be understood as the institutionalized expression of these communal reactive attitudes.

Derk Pereboom’s hard incompatibilism (Living Without Free Will, 2001; Free Will, Agency, and Meaning in Life, 2014): Neither determinism nor indeterminism provides the control needed for basic desert moral responsibility. His Four-Case Manipulation Argument: In Case 1, neuroscientists directly manipulate Plum’s brain to cause him to kill White. Cases 2–3 involve progressively more indirect manipulation. Case 4: Plum is an ordinary person in a deterministic world. Pereboom argues there is no principled place to draw the line. If hard incompatibilism is correct: retributive punishment is ruled out. Pereboom and Gregg Caruso propose a public health-quarantine model: criminal behavior addressed through incapacitation, rehabilitation, and alteration of social conditions, justified by the right to self-protection, not by desert.

Moral Uncertainty and the Practical Turn

The framework of maximizing expected choiceworthiness, introduced through MacAskill, Bykvist, and Ord’s work, marks a deeper shift in metaethics than its technical apparatus might suggest. For most of the discipline’s history, the question of what to do when you do not know which moral theory is correct was treated as someone else’s problem — a practical inconvenience best resolved by first settling the theoretical questions. The practical turn reverses this priority. Agents cannot wait for metaethical convergence before acting. They face decisions now, under conditions of irreducible normative uncertainty, and the structure of rational response to that uncertainty is itself a metaethical question.

The intertheoretic comparison problem sits at the heart of this project. If an agent holds 60 percent credence in consequentialism and 40 percent in a Kantian framework, how does she weigh the choiceworthiness scores each theory assigns? Consequentialism might rate a given action at +50 and the Kantian framework at -200, but these numbers inhabit different scales with no natural conversion rate. MacAskill and collaborators explore several strategies: normalizing theories against each other using variance-based techniques that give each theory “equal say” per unit of credence, finding common ground where theories share ordinal rankings, and drawing analogies to social choice theory when only ordinal information is available. None fully resolves the problem. The fanaticism objection — that theories assigning extreme stakes (certain religious soteriologies, for instance, positing infinite payoffs for salvation) could dominate decision-making at vanishingly low credence — reveals a structural instability in any expected-value framework applied to normative uncertainty.

What makes this practically urgent rather than merely formally interesting is its connection to first-order ethical commitments. The effective altruism movement, which MacAskill co-founded, operationalizes moral uncertainty by treating the allocation of charitable resources as a decision problem under normative uncertainty across competing ethical frameworks. The choice between funding global health interventions, animal welfare reforms, and existential risk mitigation depends not only on empirical estimates of impact but on how one weighs different moral theories’ claims about what matters. MEC does not resolve these disputes; it reframes them as tractable decision problems rather than paralyzing impasses. The metaethical upshot is significant: the relationship between first-order ethics and metaethical commitments is not merely one of logical derivation but of practical navigation under uncertainty, and the tools for that navigation are themselves normative.

Psychopathy, Affect, and the Architecture of Moral Cognition

The psychopath, as characterized by Cleckley’s clinical portrait in The Mask of Sanity and operationalized through Robert Hare’s Psychopathy Checklist-Revised, provides the sharpest empirical test case for motivational judgment internalism. The internalist holds that genuine moral judgment necessarily carries motivational force. The psychopath appears to understand moral language, deploy it fluently, and yet remain wholly unmoved by it — Cleckley’s famous metaphor of “moral colorblindness” describes someone who can process moral sentences without grasping the evaluative dimension they express. If the psychopath genuinely judges that cruelty is wrong yet feels no pull whatever toward refraining from cruelty, internalism fails.

The debate hinges on whether the psychopath’s moral utterances constitute genuine moral judgments. Internalists like Michael Smith argue they do not — the psychopath uses moral vocabulary in “an inverted commas sense,” parroting words without the cognitive-affective state that constitutes real moral judgment. Externalists like David Brink and Shaun Nichols counter that this move is ad hoc, designed to protect the theory rather than describe the phenomenon. Nichols’s experimental work complicates the picture further. When subjects were presented with a psychopath who claims to know that hurting others is wrong, 85 percent attributed genuine moral understanding to the psychopath, suggesting that the folk concept of moral judgment does not build in the motivational component internalism requires. Yet later work by Björnsson and colleagues revealed that this result is sensitive to the factive structure of the verb used — “understands that X is wrong” pulls externalist intuitions, while “believes that X is wrong” pulls internalist ones — suggesting that our folk concept may be genuinely ambiguous rather than clearly favoring either side.

Jesse Prinz’s constructive sentimentalism offers a framework that absorbs the psychopathy data without collapsing into simple emotivism. Prinz argues that moral judgments are constituted by sentiments — stable dispositions to experience certain emotions — and that these sentiments represent response-dependent properties. When someone judges an action wrong, their judgment consists in a disposition toward disapprobation, and the property of wrongness just is the property of being the kind of thing that occasions such disapprobation. This is not mere expression of feeling in Ayer’s sense. Prinz’s sentiments carry representational content, allowing moral utterances to figure in conditionals and logical inferences, thereby addressing the Frege-Geach problem. The psychopath, on this account, lacks the constitutive emotional dispositions required for genuine moral judgment, and so Prinz endorses a qualified internalism: non-motivating moral talk is parasitic on the motivating kind, a derivative phenomenon rather than a counterexample.

Prinz’s more radical contribution lies in his constructivism about the emotions themselves. Against Haidt’s moral foundations theory, which posits innate moral modules, Prinz argues that moral emotions are culturally constructed from non-moral emotional raw materials. The empathy that underwrites harm norms, the disgust that underwrites purity norms, and the dominance emotions that underwrite authority norms all evolved for purposes unrelated to morality; cultures conscript them into moral service in different configurations. Cross-cultural variation in moral codes reflects not different applications of universal moral modules but genuinely different constructions. This yields content relativism: moral wrongness is indexed to the grounding norms of a community, and where communities differ in their grounding norms, there is no transcendent standpoint for adjudication. Prinz explicitly defends this relativism while denying it entails nihilism — moral rules, like money, are social constructions with genuine functional reality.

Fiery Cushman’s model-based/model-free framework represents the most sophisticated successor to Joshua Greene’s influential but coarse dual-process theory of moral cognition. Greene had framed the tension in trolley dilemmas as a contest between automatic emotional responses (generating deontological judgments) and controlled cognitive processes (generating consequentialist ones). Cushman, drawing on computational neuroscience, reframes the distinction entirely. The model-free system assigns cached values to action types based on accumulated reinforcement history — pushing, hitting, and lying carry intrinsic negative value regardless of their consequences in a given case — while the model-based system evaluates outcomes by simulating consequences and selecting actions based on expected value. Both systems involve affect, and both involve cognition; the distinction is structural, concerning what the value representation targets (actions versus outcomes), not whether emotion or reason is in play.

This reframing dissolves several problems with Greene’s account. It explains why people reject pushing the large man in the footbridge case (the model-free system has cached strong negative value for pushing) while accepting the switch case (flipping switches carries no such cached value), without requiring the implausible claim that consequentialist reasoning is emotionless. Cushman’s empirical work — showing that performing pretend harmful actions such as swinging a fake weapon at a mannequin elicits measurable aversive physiological responses even when subjects know no harm results — confirms that the model-free system assigns intrinsic value to action types independent of anticipated consequences. The philosophical significance is that deontological and consequentialist intuitions may not track competing moral truths but rather reflect the operation of two learning systems with different computational architectures, each adaptive in its own domain. Whether this undermines or vindicates deontological judgments depends on one’s prior metaethical commitments: a debunker sees evolutionary explanation as defeating justification, while a moral realist might argue that both systems track different morally relevant features of reality.

The “moral self” literature adds a further dimension. Augusto Blasi’s self-model, developed in the early 1980s, identifies moral identity — the degree to which being a moral person is central to one’s self-concept — as the critical mediating variable between moral judgment and moral action. Aquino and Reed’s empirical operationalization through the Self-Importance of Moral Identity scale confirms that moral identity internalization predicts prosocial behavior more reliably than moral reasoning alone. For moral exemplars, as Colby and Damon’s research shows, personal goals and moral goals become so deeply unified that the tension between self-interest and morality effectively dissolves. This finding has metaethical resonance: if moral motivation operates primarily through identity rather than through recognition of moral facts, the internalism/externalism debate may be asking the wrong question. The motivational force of moral judgment may be contingent not on the metaphysical nature of moral properties but on the psychological architecture of the agent’s self-concept.

Moral Luck, Desert, and the Limits of Responsibility

The 1976 joint symposium in which both Thomas Nagel and Bernard Williams published papers titled “Moral Luck” exposed a fracture in moral thought that remains unhealed. The control principle — the intuition that agents are morally assessable only for what depends on factors under their control — appears to be a bedrock commitment of ethical life. Kant enshrined it: the good will shines by its own light regardless of consequences. Yet our actual practices of moral assessment violate it systematically.

Nagel identifies four varieties of moral luck. Resultant luck is luck in how actions turn out: two equally negligent drivers, one of whom kills a child who darts into the road, face radically different moral consequences for identical negligence. Circumstantial luck is luck in the situations one faces: the potential Nazi collaborator who emigrates to Argentina before 1933 never confronts the moral test that would reveal — or constitute — his wickedness. Constitutive luck is luck in who one is: temperament, capacity for empathy, impulsiveness, and moral sensitivity are shaped by genes and environment, yet we blame people for cowardice and cruelty as if these traits were chosen. Causal luck generalizes the free will problem: if actions result from prior causes stretching back before the agent’s birth, nothing is ultimately under the agent’s control.

Williams’s contribution is more polemical and more unsettling. His Gauguin example introduces retrospective justification: whether Gauguin’s decision to abandon his family for art in Tahiti was rationally justified depends on whether he succeeds as an artist, a matter of luck that cannot be known at the time of decision. The lorry driver who, through no fault of his own, kills a child experiences what Williams calls “agent-regret” — a response whose constitutive thought is first-personal (“it would have been better had I not done that”) and that differs qualitatively from a bystander’s sorrow. These cases demonstrate that our moral psychology already treats outcomes beyond the agent’s control as relevant to moral assessment, regardless of what our theories say.

Constitutive luck is the most philosophically devastating variety because it undermines the natural retreat from other forms. When confronted with resultant or circumstantial luck, the Kantian retreats inward — what matters is the agent’s intentions, character, and quality of will. But constitutive luck reveals that these inner features are themselves products of factors beyond the agent’s control. As Nagel writes, the area of genuine agency “seems to shrink under this scrutiny to an extensionless point.” If we cannot locate any feature of a person that both grounds moral assessment and is genuinely under their control, the entire practice of holding people morally responsible appears to rest on a fiction.

Derk Pereboom’s hard incompatibilism draws the most radical conclusion from this line of thought. His position is that the kind of free will required for basic desert moral responsibility is incompatible with both determinism and indeterminism — that is, it is unavailable regardless of how the physics turns out. His four-case manipulation argument targets compatibilism with escalating force. In Case 1, neuroscientists manipulate Professor Plum’s brain moment-by-moment to cause him to murder Ms. White; clearly, Plum is not responsible. In Case 2, neuroscientists program Plum at birth to develop the character that will lead to the murder; still not responsible. In Case 3, Plum’s character is deterministically produced by ordinary social and environmental forces rather than neuroscientists; Pereboom argues no responsibility-relevant difference has been introduced. Case 4 is the actual deterministic world. Since there is no principled line between the cases, and since Plum satisfies every standard compatibilist condition — reasons-responsiveness, hierarchical mesh, capacity for moral reasoning — throughout, compatibilism fails.

The metaethical implications are far-reaching. If basic desert is illusory, retributive punishment loses its justificatory foundation: the claim that wrongdoers deserve to suffer simply because they knowingly did wrong presupposes exactly the kind of control that hard incompatibilism denies. Pereboom’s alternative is a quarantine model — society may restrain dangerous individuals for public protection, just as it quarantines carriers of infectious disease, but without claiming they deserve their confinement. Reactive attitudes require revision rather than abandonment: blame should be reconceptualized as moral protest aimed at future reform and reconciliation rather than as backward-looking retribution premised on desert. This connects directly to Strawson’s framework from “Freedom and Resentment,” but inverts its implications. Where Strawson argued that reactive attitudes are constitutive of interpersonal life and cannot coherently be abandoned, Pereboom insists that the desert-entailing forms of these attitudes — resentment understood as the belief that the wrongdoer deserves to suffer — can and should be replaced by forward-looking alternatives without destroying the relational fabric Strawson sought to protect.

The relationship between these responsibility debates and the realism/anti-realism divide is intricate. Moral realism does not strictly require robust moral responsibility — a realist might hold that there are objective moral facts about good and bad while accepting that basic desert is illusory. Conversely, expressivism may be particularly well-suited to accommodate responsibility practices, since Strawson’s reactive attitudes framework already treats blame and praise as fundamentally emotional responses rather than factual judgments about desert. The expressivist can say that blaming someone just is expressing moral protest, without any need for metaphysically loaded claims about what the wrongdoer deserves. Error theorists face a sharper challenge: if desert claims are systematically false, the question becomes whether fictionalism about responsibility — continuing to blame and praise as if desert were real because doing so coordinates social behavior — is psychologically stable and ethically acceptable.

Normativity Beyond Ethics: Reasons, Rationality, and Convergence

The companions-in-guilt strategy, introduced in earlier discussion as a defensive move for moral realism, extends into a broader inquiry about the structure of normativity itself. If the challenge to moral normativity — “what makes a moral reason a reason?” — applies equally to epistemic normativity and the normativity of rationality, then either all normative domains face the same foundational crisis or none does. This is not merely a tu quoque maneuver; it suggests that normativity may be a unified phenomenon requiring a unified account.

The normativity of rationality has become a distinct subfield, with the central debate running between three positions. John Broome draws a sharp distinction between requirements of rationality and normative reasons. Rational requirements, on his account, are wide-scope: the requirement to intend the means to your ends does not tell you to form a specific intention but rather to make a conditional true — either intend the means or give up the end. Wide-scope formulation avoids the bootstrapping problem, in which adopting any intention, however malicious, would generate reasons for further intentions. A person who intends to kill violently and believes a cleaver is the most violent method does not thereby acquire a reason to intend to use a cleaver. But Broome ultimately concludes that even wide-scope requirements resist assimilation to genuine normativity: rationality consists in satisfying these requirements, not in responding to reasons.

Christine Korsgaard takes the opposite view, grounding the normativity of rational requirements in the constitutive structure of agency itself. To be an agent at all is to be bound by rational requirements; they are not optional standards one might accept or reject but constitutive of the deliberative activity through which one acts. The regress problem — why be rational? — is blocked because the question cannot be coherently posed by anyone already engaged in reasoning. Critics object that constitutivism either proves too much (any constitutive standard of any activity would be normative, including the “standards” of torture) or too little (it establishes that agents must follow rational requirements but not that they have genuine reasons to do so).

T.M. Scanlon’s reasons fundamentalism offers a third path. Reasons are the most basic normative concept, and all other normative notions — value, obligation, rationality — are to be analyzed in terms of them. His buck-passing account of value holds that to be good is not to possess a property that itself provides reasons but to have other properties that provide reasons for pro-attitudes. Goodness does no normative work on its own; it “passes the buck” to the lower-order properties that constitute reasons. This elegantly unifies the normative domain but faces the wrong kind of reasons problem: an evil demon who will kill you unless you desire a cup of mud gives you a powerful reason for a pro-attitude toward the mud, yet the mud has no value. Distinguishing “right kind” reasons (those reflecting genuine value) from “wrong kind” reasons (pragmatic reasons for having the attitude) has proven remarkably difficult without circularity, since the distinction seems to presuppose the concept of value that the buck-passing account was meant to analyze.

Derek Parfit’s triple theory represents the most ambitious attempt at normative convergence. In On What Matters, Parfit argues that Kantianism, consequentialism, and contractualism — properly formulated — yield the same substantive moral principles. His strategy involves reformulating Kant’s Formula of Universal Law as Kantian Contractualism (everyone ought to follow the principles whose universal acceptance everyone could rationally will), then arguing that the optimific principles are uniquely universally willable, and finally connecting this to Scanlon’s contractualism through the claim that no one could reasonably reject the optimific principles. If three independently motivated traditions converge, this constitutes powerful evidence against the skeptic’s argument from disagreement and suggests that genuine moral progress — understood as movement toward principles all three traditions endorse — is possible.

The convergence claim has faced stiff resistance. Critics note that even if the three theories agree on which acts are right, they disagree profoundly about why — and that treating verdicts as all that matters already presupposes consequentialism. Susan Wolf argues that Parfit’s reconciliation departs so far from what Kant, Mill, and Scanlon actually held that it amounts to creating new theories rather than showing existing ones converge. The deeper metaethical question is whether convergence of normative theories, if achieved, would constitute evidence for moral realism. A realist sees convergence as discovery: just as independent scientific methods converging on the same result confirms the existence of what they measure, normative convergence confirms objective moral facts. An anti-realist can explain convergence through shared evolutionary pressures, cultural transmission, and the internal logic of consistency reasoning without positing moral facts to be discovered.

The Metaphysics of Moral Properties: Grounding, Causal Inertness, and Explanatory Gaps

The supervenience relation, discussed in its strong and weak forms in the prior treatment, has increasingly been recognized as too coarse-grained for the work moral metaphysics requires. Supervenience tells us that moral properties necessarily co-vary with natural properties — there cannot be a moral difference without a natural difference — but it says nothing about the direction of dependence or the explanatory relationship between the two levels. The grounding relation, as articulated by Gideon Rosen, Kit Fine, and Jonathan Schaffer, supplies the missing structure. When the fact that an action is wrong is grounded in the fact that it causes unnecessary suffering, this is an asymmetric, irreflexive, and explanatory relation: the natural fact makes it the case that the moral fact obtains, not vice versa.

Grounding captures what philosophers actually mean when they invoke supervenience in moral contexts. When a non-naturalist like Moore insists that moral facts depend on natural facts without being reducible to them, the grounding relation provides a precise formulation: moral facts are fully grounded in natural facts, but the concepts and properties involved are not identical. This allows non-naturalism an intermediate position — determination without reduction — that supervenience alone cannot articulate. Rosen connects grounding to real definition: if wrongness is grounded in suffering-production, then the real definition of this token of wrongness involves suffering-production, even if the property of wrongness is not analytically equivalent to any natural property. The hyperintensional character of grounding — its sensitivity to the manner in which facts are determined, not just their modal profiles — makes it suitable for distinguishing moral metaphysical positions that supervenience conflates.

Yet the deepest problem for non-naturalism is not about the right dependence relation but about epistemic access. If non-natural moral properties are causally inert — if they occupy no position in the causal order and produce no effects in the physical world — then there is no mechanism by which they could influence our cognitive faculties to produce reliable moral beliefs. This parallels Benacerraf’s dilemma for mathematical platonism and Hartry Field’s reliability challenge: what explains the striking correlation between our moral beliefs and the moral facts, if the facts are causally isolated from us? The challenge does not depend on accepting a causal theory of knowledge (now widely rejected) but takes the more general form of demanding an explanation for reliable belief-tracking in the absence of causal contact.

Non-naturalists have developed several responses. Enoch’s third-factor explanation proposes that natural selection shaped our moral faculties to promote survival, and survival happens to be genuinely good (a moral fact), so evolutionary pressures correlate our beliefs with moral truth not because moral facts cause our beliefs but because a third factor independently correlates with both. Critics find this ad hoc — it seems to assume a convenient alignment between evolutionary fitness and moral truth without explaining why such alignment should obtain. The companions-in-guilt strategy points out that identical reliability challenges arise for mathematics, logic, and modality; if we accept knowledge in those domains despite causal inertness, consistency requires accepting moral knowledge too. Clarke-Doane has argued extensively that the parallel with mathematics is genuine and that the moral case is no worse off. The supervenience-based epistemology response notes that since moral properties are grounded in natural properties, we are in causal contact with the grounding base — the natural facts that make moral facts obtain — and this indirect contact may suffice for knowledge without requiring direct causal interaction with moral properties themselves.

The explanatory gap between natural and moral facts remains, however, even if grounding provides the right structural relation. Explaining why certain natural configurations — particular distributions of suffering and flourishing, specific exercises of autonomy, certain patterns of fairness — ground specific moral facts requires bridging principles whose own status is contested. Are these bridging principles themselves moral facts, requiring their own grounds? The threat of regress looms. Or are they conceptual truths knowable a priori, as Cuneo and Shafer-Landau argue when they posit “moral fixed points”? The metaphysics of moral properties thus remains a domain where formal tools have advanced considerably while foundational questions persist.

Experimental Philosophy and the Status of Moral Intuitions

The rise of experimental philosophy, catalyzed by Weinberg, Nichols, and Stich’s 2001 paper showing cross-cultural variation in epistemic intuitions, brought empirical pressure to bear on a methodology that had operated largely unchallenged. If responses to Gettier cases vary significantly between Western and East Asian subjects, and if philosophical theories are built on the assumption that such intuitions are stable data points, the evidential foundations of analytic philosophy are shaky. Transposed to metaethics, the challenge is acute: moral intuitions — about trolley cases, about the permissibility of incest, about the scope of rights — vary not only across cultures but across socioeconomic strata, and sensitivity to framing effects and order of presentation persists even among trained philosophers.

The negative program in experimental philosophy uses these findings to undermine traditional armchair methodology. If intuitions are unreliable, the reflective equilibrium method — which treats considered moral judgments as data to be brought into coherence with principles — loses its foundation. The positive program uses empirical methods constructively: Joshua Knobe’s discovery that moral evaluations infiltrate supposedly non-moral judgments about intentionality (the side-effect effect, in which 82 percent of subjects say a CEO who foreseeably harms the environment does so intentionally, while only 23 percent say the same about foreseeable help) reveals that folk concepts are not neatly separable into descriptive and normative components. This finding, robust across cultures and age groups, challenges the assumption that the descriptive and evaluative can be cleanly factored apart — a challenge that resonates with the thick concepts debate already examined.

Nichols’s “Sentimental Rules” framework adds a further layer. Norms that “resonate with our emotions” enjoy higher cultural fitness and are more likely to persist across generations. The near-universality of harm prohibitions reflects not innate moral knowledge but the cultural selection of affect-backed norms. This has direct metaethical implications: if moral convergence across cultures is explained by shared emotional repertoires rather than by tracking objective moral facts, the realist’s argument from convergence is weakened. Important caveats apply, however. Subsequent replication studies have challenged some of the original cross-cultural findings — Seyedsayamdost’s 2015 failure to replicate the Gettier case differences, for instance, suggests the empirical picture is less tidy than early X-Phi work implied. The methodological question of what to do about disagreement among intuitions mirrors the metaethical question of what to do about moral disagreement, creating a recursive problem that no current approach fully resolves.

Convergence, Progress, and the Explanatory Challenge

Whether morality converges over time is among the most consequential empirical claims in metaethics. The realist argument, pressed most explicitly by Michael Huemer, points to the broad historical trajectory: abolition of slavery, expansion of suffrage, decline of torture as state policy, decolonization, recognition of rights for previously excluded groups. This convergence across disparate cultures is best explained, the realist argues, as progressive discovery of objective moral truths — analogous to scientific convergence on physical laws.

Peter Railton’s naturalist realism provides the most detailed mechanism for such progress. On his ideal advisor theory, an individual’s non-moral good is what a fully informed, vivid, and rational version of herself would want her actual self to want; moral rightness is what would be approved from a standpoint counting all affected individuals’ interests equally under conditions of full information. Moral progress, on this account, occurs when societies better approximate what would be endorsed from this idealized impartial standpoint. Increased empirical knowledge, expanded empathy, and improved institutional design allow closer approximation to social rationality. Slavery’s abolition reflects not mere shifting preference but better understanding of its victims’ interests and the irrationality of discounting them.

Anti-realists have underappreciated resources for explaining the same data. Jeroen Hopster identifies several mechanisms: convergent moral views spread through shared cultural influences such as colonialism and globalization rather than independent discovery; technological and economic convergence generates similar cooperation problems yielding similar normative solutions; consistency reasoning drives internal extension of initially limited principles (once “all men are created equal” is accepted, logical pressure extends it to women and other excluded groups); and self-interest under conditions of prosperity favors liberal norms. Philip Kitcher’s account of ethics as social technology explains moral progress as the overcoming of functional limitations in social cooperation, without positing objective moral facts to be discovered.

The question of whether moral progress is self-defeating for anti-realism depends on how “progress” is defined. If progress means movement toward mind-independent moral truth, anti-realists cannot claim it by definition. But if progress means increased coherence, expanded cooperative capacity, reduced suffering, or better functional performance of moral systems, anti-realists can account for it comfortably. The dispute ultimately reduces to whether the direction of moral change requires explanation by moral truth or can be fully explained by non-moral factors — evolutionary pressures, economic incentives, internal logic of existing commitments, and the dynamics of cultural transmission.

Metaethics and the Architecture of Political Life

The connection between metaethics and political philosophy is most visible in John Rawls’s attempt, in Political Liberalism, to construct a freestanding political conception of justice that does not depend on any comprehensive moral, philosophical, or religious doctrine. The central problem is how a just and stable society is possible among citizens divided by reasonable but incompatible comprehensive views. Rawls’s solution is a political conception supported by an overlapping consensus — citizens from different traditions endorsing the same political principles for their own internal reasons — justified through public reason that avoids appeal to comprehensive doctrines when deciding constitutional essentials and matters of basic justice.

This represents an explicit attempt to do politics without settling metaethics. Rawls’s political constructivism does not claim that principles of justice are true; it claims they are reasonable and publicly justifiable. The concept of truth is neither used nor denied but set aside. Joseph Raz termed this “epistemic abstinence” — refraining from asserting the truth of one’s comprehensive doctrine in political justification. The motivation is practical: given the fact of reasonable pluralism, insisting on the truth of one’s own doctrine would require the oppressive use of state power.

Whether this abstinence is stable has generated sustained criticism. The burdens of judgment — Rawls’s explanations for why reasonable people disagree, including complexity of evidence, vagueness of concepts, and different life experiences — appear to encode substantive epistemological commitments that some comprehensive doctrines would reject. A citizen who holds that God has revealed moral truth may find it incoherent to treat others’ rejection of that truth as “reasonable.” David Enoch argues that political liberalism cannot make sense of “reasonable citizens” without embracing significant epistemic commitments, which defeats the abstinence it claims.

Richard Rorty’s liberal ironism pushes the anti-foundationalist strategy further, dispensing not only with metaethical foundations for politics but with the aspiration to find them. Rorty’s ironist has “radical and continuing doubts” about her own final vocabulary and does not think it is closer to reality than alternatives. Combined with the liberal commitment that cruelty is the worst thing we do, this produces a stance that defends liberal politics through historical narrative and sentimental education rather than philosophical argument. Solidarity — an expanded sense of “we” built through literature, journalism, and imaginative identification with others’ suffering — replaces objectivity as the ground of political life.

The charge that anti-foundationalism about morality entails political nihilism is the sharpest objection both Rawls and Rorty must face. If moral claims lack objective standing, what prevents the powerful from simply imposing their preferences? Rorty’s response is that foundations have never actually prevented this — metaphysical claims about human nature have historically been used to justify cruelty at least as often as to oppose it — and that the pragmatic track record of liberal institutions provides a better defense than any philosophical proof. Yet critics note that some of the most effective resistance to oppression, including the American civil rights movement, drew explicitly on foundationalist moral claims. The relationship between metaethical position and political commitment may be less direct than either side assumes: moral realists and anti-realists alike have defended and opposed liberal democracy, and the practical implications of one’s metaethics appear to be underdetermined by the theoretical commitments alone.

The Deepest Unresolved Problems

Several problems persist at the foundations of metaethics, resistant to every framework surveyed across this primer. Their persistence is not a failure of the discipline but a mark of the genuine difficulty of the questions.

The regress of justification is inescapable. Every moral system bottoms out in something it cannot further justify — an axiom, a bedrock intuition, a constitutive commitment. The consequentialist cannot justify the claim that well-being is what matters without appealing to something beyond well-being; the Kantian cannot justify the categorical imperative’s authority without presupposing the value of rational agency it is meant to establish; the contractualist cannot explain why we should care about what principles no one could reasonably reject without invoking normative premises that themselves require justification. Korsgaard’s constitutivism attempts to block the regress by locating normativity in the structure of agency itself, but critics argue this either relocates the regress (why value agency?) or proves too much (any constitutive standard of any activity would be equally normative). Coherentism — justification through mutual support among beliefs — avoids the regress but raises the specter of elegant, internally consistent moral systems that are nevertheless wholly detached from anything worth caring about.

The motivation gap persists even if the justification problem were solved. Suppose moral facts exist and are knowable: it remains unclear how they reach agents with sufficient force to motivate action. The Humean theory of motivation holds that beliefs alone cannot motivate — a desire is always required — and if moral beliefs are cognitive states (as cognitivists maintain), they need a conative companion to produce action. Externalists accept this and deny that moral judgment is intrinsically motivating. Internalists insist that genuine moral judgment carries its own motivational force, but the psychopathy data and the moral identity literature suggest that the connection between judgment and motivation is mediated by emotional and identity-related factors that are contingent rather than necessary. The practical question of how moral knowledge translates into moral action remains wide open.

The normativity of normativity — the question of what makes a reason a reason — is perhaps the deepest problem of all. Scanlon’s reasons fundamentalism takes reasons as primitive: there is no further account of what constitutes a reason. This has the virtue of avoiding reduction but at the cost of explanatory silence at the foundational level. The companions-in-guilt strategy shows that the problem generalizes — epistemic reasons, prudential reasons, and logical reasons face the same foundational question — but generality does not constitute a solution. If normativity is primitive and irreducible, then at the base of all our normative practices lies something we can identify and use but never fully explain.

The is-ought gap, first articulated by Hume and sharpened by Moore, remains unclosed. Naturalist realists like Railton attempt to bridge it through a posteriori identities between moral and natural facts, but their success depends on contested claims about whether such identities preserve normativity. Non-naturalists maintain the gap’s integrity by positing a distinct normative domain, but then face the epistemic access problem. Constructivists dissolve the gap by making moral facts dependent on rational procedures, but face the circularity objection — the procedures themselves embody normative commitments.

What remains is what might be called the orientation problem: how agents actually navigate moral life in the absence of certainty about any of these foundational questions. Several responses are available, none fully satisfying, each with its own integrity. Pragmatic commitment holds that one can act on moral convictions without claiming certainty about their foundations, treating moral engagement as analogous to scientific inquiry — fallibilist, self-correcting, and provisional. Existential resolution, in the tradition of Kierkegaard and the early Sartre, holds that the absence of guaranteed foundations makes moral commitment more rather than less significant: one chooses in the face of uncertainty, and the choice itself is constitutive of who one is. Institutional embedding argues that individual agents need not resolve foundational questions because moral life is primarily sustained by institutions — legal systems, professional norms, cultural practices — that encode moral commitments and create contexts in which agents can act well without philosophical certainty. The “as if” strategy, anticipated by Kant’s postulates of practical reason and developed by moral fictionalists like Richard Joyce, holds that agents can organize their lives around moral commitments while remaining agnostic about their ultimate truth status, treating moral discourse as a useful framework rather than a body of fact. Finally, tragic endurance acknowledges that the foundational problems may be genuinely irresolvable, that moral life will always involve acting under conditions of uncertainty that cannot be eliminated, and that the appropriate response is neither despair nor false confidence but a sustained willingness to engage with moral questions honestly, knowing that final answers may never come.

---

Conclusion

This primer has traversed the full conceptual landscape of metaethics — from the semantic question of what moral sentences mean, through the ontological question of whether moral facts exist, to the epistemological question of how they could be known, the psychological question of how they motivate, the genealogical question of where they come from, and the practical question of what to do when certainty is unavailable. The terrain is structured by a set of interlocking distinctions that, once grasped, make the entire field navigable: cognitivism versus non-cognitivism divides theories of moral language; realism versus anti-realism divides theories of moral ontology; internalism versus externalism divides theories of moral motivation; and naturalism versus non-naturalism divides theories of moral properties. These distinctions are not merely taxonomic. Each marks a genuine philosophical fault line where arguments on both sides have real force, and where choosing one position generates determinate commitments and costs across other dimensions. The cognitivist who holds that moral sentences express beliefs faces the motivation problem; the non-cognitivist who holds they express attitudes faces the Frege-Geach problem; the realist who posits moral facts faces the epistemological access problem; the anti-realist who denies them faces the moral progress problem. These are not accidents of the current state of the debate but structural features of the problem space.

The analytical tools assembled across these chapters equip the reader to do several things that were not possible before. One can now identify the precise metaethical commitments embedded in any first-order moral argument and trace their implications: when someone claims that human rights are universal and objective, one can specify what ontological, semantic, and epistemological commitments that claim carries and what challenges it must meet. One can diagnose the mechanism by which a genealogical challenge — evolutionary, historical, or psychological — threatens a moral position, and assess whether the threat is genuinely debunking or merely genetic. One can evaluate whether a given moral disagreement is best explained by differing access to moral facts, differing non-moral beliefs, differing conceptual frameworks, or the absence of facts to discover. One can recognize when an apparently first-order ethical dispute is actually a metaethical dispute in disguise — when arguments about abortion or animal rights or distributive justice turn on unstated assumptions about the nature of moral properties, the reliability of moral intuitions, or the scope of moral reasons. And one can navigate the practical challenge of moral uncertainty with theoretical self-awareness, understanding why the choice between moral theories cannot be made on purely moral grounds and what decision-theoretic resources are available for acting under normative uncertainty.

What the primer cannot provide is resolution. The deepest questions — whether moral facts exist, whether they can motivate, whether they can be known, whether normativity itself can be explained — remain open not because the arguments are weak but because they are strong on multiple sides. The appropriate response to this situation is not skeptical paralysis or dogmatic commitment to whichever position one finds initially attractive, but the disciplined engagement that these tools make possible: tracking arguments precisely, identifying where they succeed and fail, maintaining awareness of the costs of one’s own position, and remaining open to revision. Metaethics does not tell us what to do. It tells us what we are doing when we tell each other what to do, and what the world must be like for that telling to make the kind of sense we need it to make. That inquiry, far from being idle, is among the most consequential philosophy can undertake — because every moral claim, every political institution, and every act of conscience rests, whether its agent knows it or not, on answers to exactly these questions.