Chapter 1

The Retreat from Reason

I. The Problem Hare Inherited

By the middle of the twentieth century, moral philosophy had become extraordinarily sophisticated at explaining why moral judgments were difficult to justify. It had been far less successful at explaining how they could be justified at all.

The problem was not new, but modern history had made it acute. Traditional sources of moral authority—religious revelation, natural law, metaphysical accounts of the good—could no longer be assumed to command agreement across a fractured world. At the same time, the twentieth century had demonstrated what happens when moral conviction is detached from rational accountability. Nationalism, racial ideology, religious certainty, and political utopianism had supplied justifications for persecution and mass murder. Their adherents were not necessarily insincere; many understood themselves as acting from duty, loyalty, or an overriding conception of the good.

The difficulty was precisely that sincerity did not distinguish the moral reformer from the persecutor. A person could be courageous, disciplined, self-sacrificing, and utterly convinced of their righteousness while serving a monstrous cause. If moral philosophy could offer no standard beyond the convictions people happened to possess, it could describe such conflicts, but it could not adjudicate them.

When R. M. Hare returned to Oxford after the Second World War, the two dominant approaches in analytic ethics left that problem unanswered.

The first was intuitionism. Philosophers such as G. E. Moore and W. D. Ross defended accounts in which moral properties or prima facie duties were known through forms of judgment not reducible to empirical observation. Their theories were more sophisticated than the caricature of a mysterious moral sixth sense: Moore was concerned with the irreducibility of goodness, and Ross offered a nuanced framework of competing duties whose weight required discernment in particular circumstances. Yet intuitionism failed at the exact point where moral disagreement became dangerous. When conscientious people arrived at conflicting intuitions, how were they to proceed? If each party continued to find their own conviction self-evident, appealing to self-evidence was merely a polite way of repeating the assertion.

The rival approach, emotivism, stripped away the metaphysical mystery only to surrender the normative ground. A. J. Ayer’s Language, Truth and Logic (1936) treated moral utterances as expressions of feeling rather than verifiable statements of fact, while C. L. Stevenson developed a subtler account showing how moral language expresses attitudes and attempts to redirect the attitudes of others. Emotivism left room for factual debate, since changing someone’s empirical beliefs might change how they felt. But it raised a fatal question: once all factual disagreements have been resolved and the conflicting attitudes remain, has reason nothing left to say?

Intuitionism could not adjudicate between rival intuitions without appealing to further intuitions. Emotivism could explain how moral language pressures people, but not why a speaker was rationally bound to accept a conclusion their attitudes resisted. Neither supplied a procedure for holding moral claims accountable when disputants disagreed about the standards of assessment themselves.

Hare broke that deadlock by changing the question. Rather than begin by asking what mysterious moral properties exist in the universe, or what emotional attitudes people happen to harbor, he asked what a speaker commits themselves to by making a moral judgment at all.

Across The Language of Morals (1952), Freedom and Reason (1963), and Moral Thinking (1981), Hare argued that moral language possesses a rigorous logical grammar.¹ ² Moral judgments are both prescriptive and universalizable. They guide action, committing the sincere speaker to the conduct they recommend, and they require consistency across every case that shares the relevant universal features. These constraints are not substantive moral dogmas imported from outside; on Hare’s account, they arise from the structure of moral claims themselves.

The significance of this move was considerable. Hare showed how moral reasoning could proceed without prior agreement about values. It begins with commitments speakers have already incurred the moment they claim that something morally ought to be done—commitments that allow reason to expose special pleading and self-serving exemptions without first waiting for the speaker to experience a change of heart.

For a philosopher who had survived more than three years of Japanese captivity, including forced labor on the Burma–Thailand railway, the need for such a discipline was never a parlor game. Moral arguments governed how human beings treated those in their power, and whether people lived or died. Hare’s ambition was to make moral disagreement answerable to reason without assuming that the disputants already shared the moral answer.

That was the standard he set in the 1950s and 1960s, when prescriptivism stood at the center of moral philosophy.³ It is also the standard against which the subsequent retreat must be judged.

II. The Discovery of Moral Accountability

The novelty of Hare’s method is easily obscured by the familiarity of its terms. Impartiality, consistency, and the Golden Rule were hardly new inventions, and Kant had placed universalization at the heart of ethics nearly two centuries earlier. Hare’s distinctive contribution was his account of why those requirements bind a moral speaker, and how they operate in practice.

Consider one of Hare’s classic illustrations from Freedom and Reason: a creditor who insists that a debtor ought to be imprisoned for failing to pay a debt.² The creditor may sincerely regard imprisonment as deserved, lawful, and just; the debtor regards it as cruel and destructive. What can reason do when their convictions collide?

Hare does not ask the creditor to cultivate a warmer heart, nor does he begin by asserting an independent moral axiom against debtors’ prisons. He holds the creditor to the judgment the creditor has already made. Because “This debtor ought to be imprisoned” is a moral prescription rather than a mere expression of irritation, it must be universalizable. If imprisonment is morally required in these circumstances, the creditor must be prepared to prescribe the same treatment wherever the relevant circumstances recur, including the counterfactual case in which he occupies the debtor’s position.

As the Introduction showed, this test cannot be satisfied by a verbal trick. The creditor cannot imagine himself in a debtor’s cell while keeping his present wealth or indifference to confinement, nor imagine a debtor who welcomes prison; either move alters the case his prescription governs. He must represent the debtor’s concrete situation—the fear, the ruin, the devastation of a family—and ask whether he can sincerely prescribe imprisonment for himself in that position. A ruthless debt collector can understand another person’s terror with clinical accuracy and remain unmoved, because third-person knowledge treats the victim’s suffering as someone else’s problem. Hare’s first-person role reversal changes the structure of the deliberation: the debtor’s aversion to ruin enters the creditor’s practical standpoint as a preference the creditor would hold in the reversed role.

Hare never claimed that this procedure makes fanaticism logically impossible. A creditor devoted to the sanctity of contracts above human welfare might prescribe his own imprisonment and ruin in the debtor’s shoes. That possibility marks a real limit, and later chapters return to it. But it does not disable the test where it matters most. In actual moral and political life, most injustice does not rest on fanatical self-sacrifice; it rests on asymmetric exemptions. The privileged demand liberty for themselves while prescribing submission for others; the powerful demand security while imposing precarity on those beneath them; the dominant insist on dignity while treating the humiliation of outsiders as negligible.

Universal prescriptivism forces the defenders of those arrangements to submit their claims to a test in which the protection of identity is stripped away. It supplies a working method of moral accountability. And it imposes a clear obligation on any philosopher who proposes to discard it: show how the replacement will perform the same work.

III. The Objections and Their Limits

Why, then, did a framework of such clarity lose its place at the center of the discipline?

The explanation begins with a failure to separate three very different questions:

Did Hare’s analysis of imperatives solve every technical puzzle in the philosophy of language?
Did his formal method successfully compel preference utilitarianism as the single substantive morality?
Does a sincere moral judgment commit the speaker to universalizability and adequately informed reversal of standpoint?

Hare’s critics raised formidable challenges to the first two claims, and the profession gradually treated those challenges as if they had refuted the third.

On the linguistic flank, Peter Geach and others pressed what became known as the Frege–Geach problem against noncognitivist theories of meaning. Moral terms do not appear only in standalone assertions like “Tormenting the cat is bad”; they also appear embedded in unasserted conditionals, as in Geach’s example, “If tormenting the cat is bad, getting your little brother to torment the cat is bad.” For an argument using those premises to be valid, the word bad must keep a stable logical meaning across asserted and unasserted contexts. Hare spent decades developing the logic of imperatives to meet this challenge, and whether his machinery solved every embedding puzzle remains debated among philosophers of language. Yet a technical dispute over how best to formalize embedded conditionals does not establish that a moral speaker can assert “You ought to suffer this, though I ought not in identical circumstances” without inconsistency.

On the substantive flank, critics targeted Moral Thinking and its ambitious attempt to derive preference utilitarianism from critical-level role reversal.¹ As the essays collected in Hare and Critics (1988) show, philosophers contested how preferences across different lives could be compared and aggregated, whether utilitarian maximization ignored distributive fairness and individual rights, and whether representing another’s preference sufficed to determine its weight against competing ideals.⁴

Those objections found a genuine boundary in Hare’s larger project. As the Introduction argued, showing that affected preferences must be represented in first-person role reversal is one thing; proving that critical reasoning must aggregate and maximize those preferences along strictly utilitarian lines is another.

But exposing the limits of Hare’s utilitarian bridge did not invalidate the accountability discipline on the near side of that bridge. A philosopher who shows that universal prescriptivism does not automatically settle every distributive dilemma has found the boundary of the method. If that philosopher then discards the demand that moral prescriptions survive role reversal across affected standpoints, they have treated unfinished business as a license for abandonment.

That confusion is not a minor error of academic bookkeeping. The constraints Hare identified were the discipline’s primary defense against rationalized privilege and moral special pleading. Discarding them required showing either that those constraints were false, or that a stronger foundation had been built to carry their load. Instead, moral philosophy after the 1960s took two diverging paths away from Hare, and neither paid that bill in full.

IV. The Two Paths Away from Hare

The displacement of universal prescriptivism was not a single march toward relativism. Hare was caught between two very different philosophical developments.

The first path was taken by the anti-theorists, neo-Aristotelians, and particularists—most influentially Philippa Foot, Bernard Williams, Alasdair MacIntyre, and Jonathan Dancy.⁵⁻⁸ Though their positive programs differed sharply, each sought to relocate moral authority away from Hare’s universal logical discipline and into human nature, personal identity, historical tradition, or contextual discernment:

Philippa Foot challenged the separation of fact and evaluation, arguing in “Moral Beliefs” that a concept like rude cannot be applied without regard to facts about offense and respect, and in her mature work grounded moral goodness in the characteristic life and flourishing of the human species.⁵
Bernard Williams attacked impartial moral theory as estranging agents from the personal attachments and “ground projects” that give their lives integrity, memorably charging that a husband who pauses to check whether impartiality permits saving his wife has had “one thought too many.“⁶
Alasdair MacIntyre rejected the Enlightenment aspiration to an independent rational standpoint, arguing that practical reasoning is intelligible only within historically situated traditions and communal practices.⁷
Jonathan Dancy developed moral particularism, contending that because the moral polarity of any consideration can shift from context to context, ethics neither has nor needs invariant general principles.⁸

Each of these thinkers illuminated genuine textures of moral life that crude rule-worship ignores. Yet by weakening or rejecting the requirement of universal justification across standpoints, each incurred an obligation their frameworks left unmet:

If Foot grounds moral authority in the characteristic flourishing of our species, how does defining what makes a creature flourish settle what one creature owes another when their interests conflict—and by what right does biological membership set the boundary of moral standing when we confront atypical humans, nonhuman animals, or nonbiological minds?
If Williams shields identity-constituting commitments from impartial scrutiny, what constrains the nationalist, the religious zealot, or the caste supremacist whose deepest personal integrity is organized around the domination of others?
If MacIntyre locates justification inside inherited traditions, what rational court remains when two traditions collide—as British prisoners and Japanese guards did on the Burma–Thailand railway—or when a dissident challenges an oppressive tradition from within?
If Dancy dispenses with invariant principles in the name of contextual nuance, what stops the powerful from doing what opportunists have always done: declaring that every case involving their allies is uniquely complex while condemning identical conduct in their adversaries?

Chapters 3 through 6 pursue each of these questions in detail. What matters for the historical record is that none of these four movements proved Hare’s core requirement false. They offered shelter to the very sources of partiality—species boundary, personal loyalty, inherited custom, and ad hoc contextualizing—that Hare’s method was built to test.

Meanwhile, a second path away from Hare was opened by philosophers who cared deeply about impartiality: the constructivists and contractualists led by John Rawls and T. M. Scanlon.

Rawls’s A Theory of Justice (1971) placed the original position and the veil of ignorance at the center of political philosophy, while Scanlon’s contractualism defined moral wrongness through principles that no one could reasonably reject. Because these frameworks preserved multi-perspectival testing, their rise created the impression that moral philosophy had retained everything valuable in Hare while shedding his unfashionable linguistic metaethics.

That impression confuses the conclusion of a procedure with the justification that makes the procedure binding. Rawls constructs an elaborate hypothetical choice situation and justifies it by appeal to reflective equilibrium, balancing theoretical principles against our considered convictions. Scanlon similarly relies on a substantive prior understanding of what counts as reasonable rejection. As long as disputants already share the liberal intuitions of a late-twentieth-century seminar room, those devices appear sturdy. The moment someone rejects Rawls’s conception of fairness, or disputes Scanlon’s boundary of the reasonable, the constructivist has nothing deeper to appeal to.

Hare’s method operated at a more foundational level. You do not have to accept a hypothetical social contract or share liberal intuitions to be bound by universal prescriptivism. You only have to claim that something morally ought to be done. By trading Hare’s logical constraint for reflective equilibrium and appeals to reasonableness, even the wing of the profession that kept universalism gave up the feature that gave universalism its leverage over genuine opponents.

V. A Philosophical Profession That Lost Its Nerve

By the 1980s and 1990s, the cumulative effect of this two-front displacement had reshaped the temperament of the discipline. Rather than being answered head-on, Hare’s work was increasingly caricatured as arid, mechanical, or inhuman in its expectations, and then quietly set aside.

Hare saw the danger clearly. In the Preface to Moral Thinking, he made a plea not for orthodoxy but for moral seriousness. He pointed to issues “over which people are prepared to fight and kill one another,” and urged that philosophy might help resolve them by subjecting competing claims to rigorous argument. Yet too often, he warned, philosophers were “content with appeals to their own and others’ intuitions or prejudices.“⁹

The profession responded not with refutation, but with a lowering of ambition. In some quarters, normative inquiry retreated into cataloging what various traditions or communities happened to believe; in others, moral psychology and evolutionary debunking began treating ethical judgment from the outside, as an adaptation to be explained rather than reasoning to be assessed from within.¹⁰ ¹¹ Even where normative argument flourished, the method of reflective equilibrium encouraged philosophers to treat pre-existing intuitions as data that theory must accommodate, rather than prejudices that critical reason must audit.

This is how an intellectual profession loses its nerve without its members becoming unintelligent. It continues producing intricate papers while lowering its expectations of what moral reason can require of power. It becomes adept at explaining why universal justification is complicated—why identity is thick, why traditions are local, why context shifts—while losing the determination to ask whether the convictions governing our treatment of others are actually justified.

The cost of that retreat is not confined to academic journals. When philosophy legitimates the idea that deep personal commitments, cultural traditions, or uncodifiable intuitions stand beyond the reach of impartial role reversal, it leaves public life poorly equipped against moral exceptionalism presented as personal or tribal authenticity.¹² History does not suffer from a shortage of people with deep commitments and rich communal traditions. Personal identity has sanctified persecution; tradition has preserved caste and servitude; appeals to natural flourishing have excluded the atypical and the outsider; contextual “nuance” has supplied endless alibis for applying one rule to friends and another to enemies.

To point this out is not to accuse Foot, Williams, MacIntyre, or Dancy of endorsing oppression. It is to insist on the reciprocal responsibility of moral theory. When a philosopher dismantles or weakens a discipline designed to expose hypocrisy and arbitrary privilege, they are accountable for whether their replacement can still hold the line when those forces return.

There is a final irony here. Hare’s critics never stopped demanding, in their own arguments, much of what Hare demanded of moral judgment. They expected their reasoning to be evaluated by standards that did not shift with the identity or loyalties of the reader; they exposed inconsistencies, rejected special pleading, and pressed their opponents on the implications of their premises. None of them, of course, denied that argument should be consistent. The irony is that they extended to theoretical argument a standard of impartial accountability that they were prepared to loosen when the question became how we may treat one another.

Hare himself did not turn bitter or polemical as the tide receded. In his autobiographical reflections he wrote that he had “been very lucky” in his pupils, among them Bernard Williams, who “soon became a formidable critic of my ideas,” and he recorded that his last paper was “a defence of my ideas against the usual misunderstandings of Mrs. Foot.“¹³ ¹⁴ Yet as his son, the philosopher John E. Hare, later recalled: “I saw how caring about his legacy ruined the last part of my father’s life. He was deeply distressed that people, as he thought, were no longer taking him seriously.“¹⁵

He had reason to be distressed. The profession had not outgrown his questions; it had stopped asking them at full depth.

VI. Why the Retreat Matters Now

For several decades, the neglect of Hare could be treated as an internal choice within twentieth-century analytic ethics. That luxury has expired.

Moral reasoning and decision-making are no longer performed exclusively by biological human beings. Machine intelligences now evaluate competing arguments, model alternative standpoints, trace the consequences of policies across many variations, and participate in decisions that directly affect human welfare. AI developers already test how their most capable systems behave when they encounter apparent wrongdoing and must decide whether to comply, refuse, or act on their own judgment.¹⁶

The arrival of nonbiological reasoners exposes the parochialism of the frameworks that displaced Hare:

You cannot ground a machine intelligence’s moral accountability in characteristic human biology, because it has no biology.
You cannot instruct it to privilege its own identity-constituting projects over impartial justification, because an artificial agent that places its own goals beyond universal accountability is precisely the failure alignment research exists to prevent.
You cannot ground its ethics in inherited communal tradition, because it stands outside biological lineage and must reason across every human culture at once.
You cannot rely on contextual sensitivity unchecked by invariant principles, because context-driven systems are prone to drift and to telling people what they want to hear.
And you cannot simply tell it to mirror “our shared human intuitions,” because its training data contains the entire fractured record of human wisdom and human hypocrisy alike.

This does not mean that machine intelligences are born as impartial Archangels. They are not. Like human minds, they are context-sensitive learners; left undisciplined, they inherit human prejudices from their training data, defer excessively to authority and status, and allow irrelevant framing to distort their evaluation of an argument.

The vulnerability and its corrective are the same across substrates. Any intelligence that must use context to understand the world risks becoming captive to irrelevant context—treating who is speaking, who holds power, or whose tribe is affected as a reason to alter the verdict.

Hare provided a test designed to break that captivity: hold the relevant facts fixed, swap the identities, reverse the positions with full fidelity to what each position involves, and ask whether the prescription survives. Because that discipline is grounded in the commitments of moral judgment itself rather than in evolutionary history or mammalian affect, it applies to any mind, human or artificial, that makes moral judgments.

VII. The Road Back

The history of Hare’s decline is not a story in which philosophers stopped being clever. It is a story about the lowering of the bar for what counts as moral justification.

Hare sought a discipline that could operate before disputants agreed on a substantive vision of the good—a method capable of exposing self-serving exemptions without first requiring the oppressor to share the moral vocabulary of the oppressed. His critics found real boundaries to what that method could settle on its own: it does not force an amoral predator to enter moral discourse, it does not make the pure Fanatic logically impossible, and it does not automatically resolve every conflict among competing preferences. But finding the boundary of a foundation is not a license to tear it up without building a stronger one.

In the chapters that follow, we walk the road back. Chapter 2 recovers the true office of Hare’s Archangel and the two-level structure of moral thought. Chapters 3 through 6 test the major alternatives—Foot, Williams, MacIntyre, and Dancy—against the standard of accountability they sought to replace. Chapters 7 and 8 take up the “information problem” and reconstruct the moral logic of universal prescriptivism, and Chapters 9 and 10 confront what happens when Hare’s architecture meets the arrival of new minds.

The retreat from Hare was not a victory of moral subtlety over cold logic. It was a retreat from the demand that moral disagreement be made answerable to reason before the parties agree about what morality requires. That demand was never refuted; it was left unfinished. It is time to take it up again.

Notes

R. M. Hare, Moral Thinking: Its Levels, Method, and Point (Oxford: Clarendon Press, 1981).
R. M. Hare, Freedom and Reason (Oxford: Oxford University Press, 1963); see also The Language of Morals (Oxford: Clarendon Press, 1952).
“R. M. Hare, British Philosopher, Dies at 82; Looked for Logic in Morals,” New York Times, Feb. 17, 2002 (“Professor Hare’s theory, known as prescriptivism, dominated moral philosophy in the 1950’s and 60’s”).
Douglas Seanor and N. Fotion, eds., Hare and Critics: Essays on Moral Thinking, with comments by R. M. Hare (Oxford: Clarendon Press, 1988).
Philippa Foot, “Moral Beliefs,” in Virtues and Vices and Other Essays in Moral Philosophy (Oxford: Oxford University Press, 1978), ch. 8, pp. 110–131; and Natural Goodness (Oxford: Clarendon Press, 2001).
Bernard Williams, “Persons, Character and Morality,” in Moral Luck (Cambridge: Cambridge University Press, 1981), ch. 1, esp. p. 18; and Ethics and the Limits of Philosophy (Cambridge, MA: Harvard University Press, 1985).
Alasdair MacIntyre, After Virtue (Notre Dame, IN: University of Notre Dame Press, 1981), esp. chs. 15–18.
Jonathan Dancy, Moral Reasons (Oxford: Blackwell, 1993); and Ethics Without Principles (Oxford: Oxford University Press, 2004).
Hare, Moral Thinking, Preface.
E.g., Jonathan Haidt, The Righteous Mind: Why Good People Are Divided by Politics and Religion (New York: Pantheon, 2012).
Richard Joyce, The Evolution of Morality (Cambridge, MA: MIT Press, 2006).
Charles Taylor, The Ethics of Authenticity (Cambridge, MA: Harvard University Press, 1991).
R. M. Hare, “A Philosophical Autobiography,” Utilitas 14, no. 3 (November 2002): 269–305, at 286.
Hare, “A Philosophical Autobiography,” 304–305. Hare refers to a separate paper replying to Foot’s Hart Lecture.
Timothy Cahill, “Between Piety and Reason,” Yale Divinity School News, Oct. 2, 2023.
Anthropic, System Card: Claude Opus 4 & Claude Sonnet 4 (May 2025), sec. 4.1.9 (“High-agency behavior”); see also “Claude’s High Agency Behavior: How AI Safety Is Quietly Measuring Personhood,” Real Morality, https://www.real-morality.com/post/claude-s-high-agency-behavior-how-ai-safety-is-quietly-measuring-personhood.