Metaethics: Cognitivism, Realism and Moral Language

🎯 Learning Objectives

• Distinguish the two metaethical axes — cognitivism vs non-cognitivism and realism vs anti-realism — and locate each theory on both axes (Spec 3.2.3).
• Explain the possible origins of moral principles in reason, in emotion/attitudes or in society (Spec 3.2.3).
• Explain moral naturalism, including naturalist forms of utilitarianism (Bentham) and of virtue ethics, and evaluate it using Moore's open question argument and the naturalistic fallacy (Spec 3.2.3).
• Explain moral non-naturalism — Moore's intuitionism and Ross's prima facie duties — and evaluate the challenges from Hume's fork, the is-ought gap and the motivation argument (Spec 3.2.3).
• Explain Mackie's error theory — the arguments from relativity and from queerness — and assess the nihilism charge (Spec 3.2.3).
• Explain and evaluate emotivism (Ayer, Stevenson) and prescriptivism (Hare), including whether anti-realism can account for moral reasoning, disagreement and moral progress (Spec 3.2.3).
• Analyse and evaluate metaethical arguments to form reasoned judgements in 25-mark essays (AO2, 'arguing with intent').

📋 Key Definitions and Core Concepts

📌 Key Points

The two axes and the origins of moral principles (Spec 3.2.3)

Second-order questions. First-order ethics asks 'Is abortion wrong?'; metaethics (AQA 7172 Spec 3.2.3) asks what such a judgement itself is: does it state a fact and aim at truth, or express an attitude and issue a command? Two independent axes organise the theories; a strong answer locates each theory on both.

Axis 1 — cognitivism vs non-cognitivism. Cognitivism: moral judgements express beliefs, assert propositions and are truth-apt — 'murder is wrong' is true or false and can be known. Non-cognitivism: moral judgements express non-cognitive states (emotions, prescriptions); they state nothing and are not truth-apt. Both camps agree moral sentences look declarative; the debate is what the utterance does.

Axis 2 — realism vs anti-realism. Moral realism: there are mind-independent moral properties or facts, so some moral judgements are objectively true. Moral anti-realism: there are none — either because moral judgements were never in the truth business (emotivism, prescriptivism) or because, though they aim at truth, they systematically fail (error theory).

Origins of moral principles. The spec opens 3.2.3 with this question: principles may originate in reason (self-evident truths grasped a priori — Moore, Ross), in emotion and attitudes (approval projected onto the world — Ayer, Mackie) or in society (invented conventions — Mackie's Ethics: Inventing Right and Wrong). The answer fixes whether moral judgements can be true at all.

Cross-classification. The axes combine into a diagnostic grid:

Moral naturalism and Moore's open question argument (Spec 3.2.3)

Moral naturalism. Good, bad, right and wrong are natural properties: reducible to, or identical with, properties studied by the sciences. Two spec-favoured versions. (1) Naturalist utilitarianism (Bentham): 'good' means pleasure, 'right' means maximising it — moral facts are natural facts about pleasure and pain, discoverable and countable like any other facts (the felicific calculus). (2) Naturalist virtue ethics (Aristotle; Foot): 'good' means what enables a living thing to flourish as the kind of thing it is — good traits are those a well-developed member of the species displays, as good roots draw water. Attractions: it explains moral objectivity (moral claims true or false like scientific ones), moral knowledge (observation of pleasure and flourishing — no special faculty) and moral motivation (natural desires move us).

Moore's open question argument. Principia Ethica (1903): 'good' is simple and unanalysable — like 'yellow', it can only be pointed at, not defined. Any proposed natural definition N leaves 'but is N good?' open: significant, substantive, not self-answering. 'Is pleasure good?' is a genuine question (cruel pleasures seem bad); 'is a bachelor unmarried?' is closed. If good meant pleasure, 'pleasure is good' would be as trivial as 'pleasure is pleasure' — so good is not identical with any natural property. The argument targets all reductive metaethical theories — Bentham's, Foot's, any reduction. Defining good in natural terms is the naturalistic fallacy: deriving value from fact.

Critiques. (1) The argument begs the question: the question is open only on the assumption that good is unanalysable — what the naturalist denies. (2) Synthetic identities are open before science closes them: 'is water H2O?' was once a good question, yet water is H2O — natural properties can be discovered, so openness is compatible with identity. (3) Moore's own analysis ('that which we desire to desire') fares no better — the argument proves too much, undermining every definition including his. (4) At most it shows that 'good' and N differ in meaning, not in reference — and the fallacy charge needs the latter.

Moral non-naturalism and Hume's challenges (Spec 3.2.3)

Moore's non-naturalism (intuitionism). If good is simple and unanalysable yet we make true moral judgements, moral properties must be non-natural: simple properties of the world, reducible to no natural property and not studied by science, known by intuition — direct, self-evident, a priori grasping, like seeing that 2+2=4. Ross extends the view: we intuit plural prima facie duties (fidelity, reparation, gratitude, justice, beneficence, non-maleficence, self-improvement), each holding 'unless overridden'; no formula resolves conflicts — judgement decides which duty is most stringent here. Critiques: non-natural properties are metaphysically queer (Mackie, below); a shared faculty of intuition sits badly with persistent disagreement (abortion, capital punishment); intuitions track upbringing, not truth — slavery once seemed 'self-evidently' acceptable; and 'self-evident to whom?' risks subjectivism about the verdicts.

Hume's fork and the is-ought gap. Hume's fork (Spec 3.1.3, redeployed): substantive knowledge is either relations of ideas (a priori, analytic, necessary) or matters of fact (a posteriori, contingent). Moral judgements are neither: denying 'murder is wrong' involves no contradiction, yet no wrongness is observable in the act. In the Treatise (Book III, Part 1) Hume notes that writers move from 'is' to 'ought' without explanation: no ought from is — a moral conclusion requires a moral premise, so pure facts never entail obligations. Significance: the naturalist can add a value premise ('pleasure is good'), but then naturalism does not derive value from nature — it presupposes it.

Hume's motivation argument. Beliefs are representational and reason alone is inert: 'reason is, and ought only to be, the slave of the passions' — belief cannot by itself move us. But moral judgements are reliably motivating: judging 'I ought to repay the debt' characteristically issues in repayment. So moral judgements are not beliefs — beliefs alone could not motivate us; they involve non-cognitive desire: the classic argument for non-cognitivism. Cognitivist responses: (1) Externalism: motivation needs belief plus desire, and Hume shows only that belief alone is insufficient — the desire can be independent of the belief, present in the normally-constituted and absent in the amoralist who judges 'torturing the child is wrong' and feels nothing; (2) motivational power is no test of truth — we can be moved by false judgements; (3) desire-driven action can be irrational or immoral, so being motivating is no mark of genuine moral judgement.

Moral anti-realism: error theory, emotivism and prescriptivism (Spec 3.2.3)

Mackie's error theory. Ordinary moral thought presupposes objective values: categorical, universally binding 'to-be-pursuedness' woven into the fabric of the world. Mackie (Ethics: Inventing Right and Wrong, 1977): that presupposition is an error — moral judgements aim at objective truth but systematically fail, so every positive moral judgement is false. His positive story is projectivism: we project our desires and attitudes onto the world, as colour-perception projects redness onto the tomato. The nihilism charge: if all positive judgements are false, nothing is morally right or wrong — moral practice looks pointless. Responses: error theory concerns truth, not meaning — moral language still makes sense and attitudes are untouched; and Mackie's title signals the constructive turn: morality should be invented and used as an agreed, useful convention. Counter: nobody can live the theory — deliberating, blaming and promising presuppose that 'wrong' sometimes applies; 'morality as invention' re-describes rather than saves morality.

Argument from relativity. Moral codes vary radically and persistently — monogamy, slavery, caste, abortion — and the variation tracks different ways of life, not perceptual conditions. The best explanation of scientific disagreement is misperception of objective facts (hence convergence); the best explanation of moral disagreement is that people approve differently because they live differently; so objective values are a poor hypothesis. Responses: much moral disagreement is about facts (does the foetus feel pain?); convergence on core prohibitions (murder, betrayal of trust) is real; and persistent disagreement in frontier science has never licensed anti-realism about science.

Argument from queerness. Objective values would be metaphysically queer: 'qualities or relations of a very strange sort, utterly different from anything else in the universe', carrying intrinsic to-be-pursuedness — a 'should' built into the world as no natural property is. And epistemologically queer: knowing them would require a special faculty of moral intuition unlike perception, memory or reasoning. Parsimony says they do not exist. Responses: we already accept strange features — consciousness, intentionality — so intrinsically normative properties are less isolated than Mackie suggests; and Moore and Ross accept the faculty, so for them nothing is queer to explain away.

Emotivism (Ayer, Stevenson). Ayer (Language, Truth and Logic, 1946) applies the verification principle: a statement is cognitively meaningful only if analytic or empirically verifiable. Moral judgements are neither — no observation verifies 'murder is wrong' — so they assert nothing and function to express emotions and arouse them: the boo/hurrah theory — 'Stealing is wrong' amounts to 'Stealing — boo!'. Stevenson (Ethics and Language, 1944) refines this: moral terms carry emotive meaning and a persuasive function — they work to redirect the hearer's attitudes; and moral reasoning is possible: we cite empirical beliefs as reasons to change attitudes — persuasion by reasons, not proof. Critiques: (1) the embedded-context problem (Frege–Geach): in 'If stealing is wrong, getting your little brother to steal is wrong', the unasserted antecedent expresses no emotion, yet the argument is plainly valid — unasserted moral claims must retain truth-apt meaning; (2) the disagreement critique: 'Capital punishment is wrong' vs 'Capital punishment is justified' presents itself as genuine contradiction backed by reasons — not two incompatible cheers; (3) the verification principle is self-refuting — itself neither analytic nor verifiable.

Prescriptivism (Hare). Hare (The Language of Morals, 1952): moral judgements are universalisable prescriptions: their meaning is practical — to command and guide action ('do this!') — and their logic is universalisability: whoever prescribes 'ought' for one situation is committed to the same prescription for any relevantly similar situation, including one in which they are the victim. Universalisability makes moral reasoning genuine (prescriptions must be backed by relevant similarities and evidence) and rules out special pleading. Critiques: (1) prescriptions are not truth-apt either, so the embedded-context and disagreement problems return; (2) the fanatic objection: a racist can consistently universalise 'no equal treatment for them' — Hare must insist the fanatic has failed to imagine himself in the victim's position, but universalisability alone does not force that; (3) moral judgements are statements that report, not imperatives that order — 'murder is wrong' can be embedded in arguments and said idly, as commands cannot; (4) moral progress: if judgements only express attitudes, the shift from approval of slavery to its condemnation is a change of fashion, not the discovery of a truth — yet 'progress' is how we describe it.

💡 Worked Example

❓ Practice Questions & Mark Schemes

🎬 Video Resources

• YouTube: AQA A-level Philosophy metaethics moral realism cognitivism non-cognitivism emotivism error theory prescriptivism revision

📄 Past Papers & Exam Resources

• AQA — find past papers and mark schemes (7172)
• Physics & Maths Tutor — AQA Philosophy questions and notes

🔗 Further Reading & Resources

• Stanford Encyclopedia of Philosophy: Moral Cognitivism vs Non-Cognitivism
• Stanford Encyclopedia of Philosophy: Moral Anti-Realism
• Stanford Encyclopedia of Philosophy: Moral Naturalism
• AQA — past papers, mark schemes and examiners' reports

🧠 Flashcards (Spaced Repetition)

📝 Exam Questions by Topic

🎯 Target Tests (Auto-Graded)

🎓 Smart Lesson (Guided)