Source
Haidt (2001), LIT-511 (Psychological Review 108(4)), read in NOTE-403.
What was actually shown
The model (pp. 7–8). Six links: intuitive judgment, post-hoc reasoning, reasoned persuasion of others, social persuasion, reasoned judgment and private reflection. Moral judgment consists “primarily of links 1 through 4”. Reasoning overriding intuition (link 5) is rare. Private reflection works mostly by role-taking, which triggers a competing intuition (link 6).
Four problems for rationalism. Affective evaluation is fast and pervasive. Reasoning is motivated: people are “intuitive lawyers”. People confabulate reasons for their behaviour (Nisbett & Wilson). Moral reasoning predicts moral action weakly once IQ is partialled out. Most of this evidence comes from outside morality, as the paper says.
Harmless wrongs (p. 5). In Haidt, Koller & Dias (1993), Americans and Brazilians condemned harmless offensive acts, and affect predicted the judgments better than claims of harm. Haidt & Hersh found the same for sexual morality, with “moral dumbfounding”.
Manipulated intuition (p. 17). Wheatley & Haidt: a hypnotically induced pang of disgust attached to a word raised condemnation of stories containing it, in a small sample of highly hypnotizable participants.
What could have come out otherwise. People could have withdrawn condemnation once they agreed that no harm was done. They did not.
What this does not say
- That this is established. The author calls the model “a plausible alternative … not … an established fact” (pp. 2–3).
- That disgust drives moral condemnation. The hypnosis study is small, and later work finds disgust does not amplify condemnation (Landy & Goodwin’s meta-analysis and replications, reviewed in LIT-101, NOTE-132).
- That reasoning never changes judgments. Links 5 and 6 allow it. Because they do, any such case fits the model, which only says they are rare.
- That moral action is driven by emotion rather than reasoning. The argument is from correlations and clinical cases (psychopathy, VMPFC damage), not experiments.
- Anything about why morality has the content it has. It is a proximate account and is compatible with a cooperative ultimate function, which it invokes (pp. 18–19).
Connections
- Moral Foundations Theory (THEORY-539) takes this as its premise: the foundations work through fast intuitions.
- It supplies one answer to Bloom’s objection to morality-as-cooperation (THEORY-468) that the theory says nothing about proximate mechanism.
- THEORY-056 has the same shape for emotion regulation: one motive state checking another, with no regulating system above them.
Note of 2026-10-29: what holds it back
Read in full (NOTE-403), but its promote_when is unmet. The record holds no replicated experiment from outside Haidt’s lab that manipulates affect and supplied reasons separately and measures moral judgment. The one direct manipulation the paper reports, Wheatley and Haidt’s hypnotic disgust, is the result later work disputes (LIT-101, NOTE-132). Narrowing to what the paper proposes would free nothing: its one dependent, THEORY-539, presupposes it and is held by its own unmet promote_when.