📖 Résumé de lecture · non citable

The Righteous Mind - Haidt (2012) (EN)

Résumé de lecture, chapitre par chapitre — rapporte ce que le livre dit.

Reading summary, part by part

🔬 Built from the reading dossier, not from the book. Source of this summary:
_dossiers/The Righteous Mind - Haidt (2012) (EN) — DOSSIER.md (150 candidates of 920) and
… — SONDES.md (targeted probes). ⚠️ This is a method claim with a consequence: what the
dossier did not surface is not in this summary, and 764 candidates were cut by the cap. The
summary is not a claim of coverage.

⚠️ Chiffres révisés le 2026-08-25. Le dossier a été régénéré avec dossier_lecture.py v1.4, qui corrige un défaut : la v1.3 ne scannait pas le texte situé avant le premier en-tête détecté. Pour ce livre la tête représentait 1 087 mots sur 105 559, soit 1 % — l'écart est marginal et aucune citation de ce résumé n'est affectée (toutes re-vérifiées). ⛔ Sur un autre livre du corpus (Pocket Guide, Porges 2017) la même tête valait 96 % du volume.

---

Identity

⛔ What could NOT be read, and is therefore not asserted here. This .txt carries no extractable chapter headers and no table of contents — the numbered headings it contains are in-chapter lists ("1. BRAINS EVALUATE INSTANTLY AND CONSTANTLY"), not chapter titles. Chapter numbers and titles are therefore absent from this summary. What is anchored: the three Parts, each with its stated central metaphor, and the author's own end-of-chapter recaps, located by probe. ⚠️ A per-chapter summary of this book would require a different extraction, not a better reading.

---

The three parts, as the book states them

The book is built on three principles, and each Part opens by naming its own central metaphor — the author's device, not an interpretation added here.

⚠️ The word-counts above are the detected sections, and the detection is imperfect: three of the seven detected sections are the Parts' announcement paragraphs in the introduction, not the Parts themselves. Read them as orders of magnitude — Part III is the longest by a wide margin — not as an inventory.

---

Part I — Intuitions come first

The claim is not that reasoning is absent but that it arrives second, and in a role: the rider serves the elephant. The book supports it with six lines of experimental work, which its own recap lists — brains evaluating instantly, social and political judgments running on quick intuitive flashes, bodily states influencing moral judgment, psychopaths who reason but don't feel, babies who feel but don't reason, and the timing of affective reactions in the brain.

The second half of Part I turns the same claim toward function: moral reasoning is strategic, "more like a politician searching for votes than a scientist searching for truth". Its sharpest empirical consequence, and the one this corpus should keep:

Verified quotations

"Schwitzgebel even scrounged up the missing-book lists from dozens of libraries and found that
academic books on ethics, which are presumably borrowed mostly by ethicists, are more likely to be
stolen or just never returned than books in other areas of philosophy. [...] In other words,
expertise in moral reasoning does not seem to improve moral behavior, and it might even make it
worse"

(off 189184 · probed, 1 occurrence) — ⚠️ the […] marks an elided footnote numeral ("49") that sits between the two sentences in the file. ⛔ It is marked because an unmarked splice of two sentences is a fabricated quotation, even when both halves are verbatim.

⚠️ What this does not license. The finding is about ethicists' book-borrowing. It is used in the book as an illustration of a general claim; reprised as evidence for that general claim, it would be over-extended. The book's own hedge — "it might even make it worse" — is part of the sentence and must travel with it.

---

Part II — More than harm and fairness

Two moves, and they are of different kinds.

The descriptive move. Cultures differ in what they moralize; WEIRD societies are statistical outliers, with a moral domain unusually narrow and largely limited to the ethic of autonomy. The book states the status of this claim itself, and the wording matters:

Verified quotations

"Moral pluralism is true descriptively ." (probed, 1 occurrence)

⛔ RT-2, in the author's own hand. Descriptively is his word, not a caution added here. Any reuse that turns this into a normative pluralism (therefore all moral matrices are equally valid — ⛔ my formulation, not the book's) contradicts the source. See Part III below, where the book itself goes normative — and says so.

The theoretical move. Moral Foundations Theory. Its status in the book is explicitly provisional, twice over:

Verified quotations

"Five good candidates for being taste receptors of the righteous mind are care, fairness, loyalty,
authority, and sanctity." (probed, 1 occurrence)

and innateness is defined as "organized in advance of experience" — "like the first draft of a book that gets revised as individuals grow up within diverse cultures". By the end of Part II the list is six, not five: the Liberty/oppression foundation is added, and the earlier two-foundation description of liberals is revised by the author himself:

Verified quotations

"Liberals have a three-foundation morality, whereas conservatives use all six."
(probed, 1 occurrence)

⛔ Consequence for the corpus, and it is the main one. Within a single book, the number of foundations is 5, then 6; the liberal profile is 2, then 3; the theory's own figure is labelled a "first draft". A theory that revised itself twice between its own covers cannot be cited as a settled count. This is not a defect of the book — it says all of it openly — it is a reason not to report "the moral foundations" as an established fact.

---

Part III — Morality binds and blinds

The longest part, and the one where the book takes its largest risk: multilevel selection, the "groupish overlay", the hive switch, and religion as a group-level adaptation rather than a parasite.

The book's epistemic position here is stated, and it is the most useful sentence in the volume for this corpus:

Verified quotations

"I can’t say for sure that human nature was shaped by group selection—there are scientists whose
views I respect on both sides of the debate." (probed, 1 occurrence)

and, on the same argument:

Verified quotations

"It is not far-fetched, and it should not be a heresy to suggest that this is how we got the
groupish overlay" (probed, 1 occurrence)

Verified quotations

"This is not a just-so story." (probed, 1 occurrence)

⚠️ Read those three together. "not far-fetched", "should not be a heresy", "not a just-so story" — these are defences against an objection, not evidence for a claim. The author is asking for the hypothesis to be admissible, not asserting it is established. ⛔ Group selection in this book is argued, contested, and declared contested by its own author. No E-level above the contested band can be drawn from it here, and the framework is not applied in this summary — that is canon 42's work, not this document's.

On religion, the account is layered and the first layer is conceded as accidental:

Verified quotations

"Our ability to believe in supernatural agents may well have begun as an accidental by-product of a
hypersensitive agency detection device" (probed, 1 occurrence)

⛔ And here the book crosses, explicitly, from description to prescription:

Verified quotations

"when we talk about making laws and implementing public policies in Western democracies that
contain some degree of ethnic and moral diversity, then I think there is no compelling alternative
to utilitarianism" (probed, 1 occurrence)

⚠️ **This is the point at which The Righteous Mind stops being a descriptive psychology.** The author declares moral pluralism true descriptively in Part II and then argues for a normative utilitarianism (a "Durkheimian" one) for public policy in Part III. Both statements are his; they are not in contradiction, because he separates the registers. A reuse that keeps the pluralism and drops the register distinction misrepresents the book — and a reuse that presents the normative conclusion as a finding of moral psychology commits exactly the error the book's own framing avoids.

---

What this book is good for in this corpus

• The rider/elephant model and its evidence base — mechanism, well anchored, stated with hedges.
• The WEIRD critique — descriptive, explicitly labelled as such by the author.
• The self-revision of Moral Foundations Theory — a documented case of a theory changing its own

count mid-volume. Useful as an epistemic object, not only as content.

• The descriptive/normative seam — a rare case of an author drawing the line himself, in writing.

What it must not be used for

• ⛔ A settled count of moral foundations (5 or 6, depending on the chapter).
• ⛔ Group selection as an established mechanism — the author declines to claim it.
• ⛔ Expertise in ethics does not improve behaviour (⛔ my formulation, not the book's) as a

general finding — it rests on one library study, with the author's own "might" attached.

• ⛔ Any efficacy, intervention, or applied claim. The book proposes none.

---

Provenance

⚠️ Two probes returned more than one occurrence and are reported as such rather than as unique anchors: "organized in advance of experience" (4) and "Intuitions come first, strategic reasoning second" (11 — it is the Part I principle, repeated by design). Neither is quoted above as a located one-off.

⛔ Tool documentation: _method/OUTIL_dossier_lecture.md.