The WEIRD-Sample Critique

In brief

Joseph Henrich, Steven J. Heine, and Ara Norenzayan published "The Weirdest People in the World?" in Behavioral and Brain Sciences in 2010, coining an acronym that stuck: WEIRD, for Western, Educated, Industrialized, Rich, and Democratic. Their claim went beyond the old complaint that psychology's samples are narrow. They argued the narrow samples are also unusual. Across domains where cross-cultural data existed, from visual illusions to fairness norms to the structure of the self, WEIRD subjects routinely sat at or near the extreme end of the human distribution rather than near its center. A field that treated undergraduate psychology students as a stand-in for "people in general" was, on the evidence, generalizing from one of the least representative groups available.

The full treatment

The problem it answers

Psychology presents its findings as claims about the mind, not about a subpopulation. But its evidence base was, and largely remains, drawn from a thin slice of humanity. Henrich, Heine, and Norenzayan estimated that up to 96 percent of subjects in top psychology journals came from countries holding about 12 percent of the world's population, and that most were university undergraduates, recruited because they were cheap and available to researchers who were themselves professors at Western universities. Their most quoted estimate: a randomly selected American undergraduate was more than 4,000 times more likely to appear as a research subject than a randomly selected person living outside the West. Jeffrey Jensen Arnett had made a related point two years earlier in "The Neglected 95%" (2008), auditing six subfields' flagship journals and finding a large majority of authors and samples were American. The 2010 authors sharpened the objection: this was not simply a gap to regret, it was a validity problem, because WEIRD populations differ from the rest of the species on the dimensions psychology studies.

How the argument works, and what it claims

The paper is a review, not a single experiment. It surveys existing cross-cultural data across domains, perception, fairness, spatial cognition, categorization, moral reasoning, and the self, asking one question of each: where do WEIRD samples fall relative to everyone else tested? The recurring answer was that they were frequent outliers, not a random draw from similar humans. The claim is not that no finding generalizes, but that generalization cannot be assumed and must be checked domain by domain, since cross-cultural variation itself varies: basic color discrimination shows little of it, while social norms, self-concept, and reasoning style show a great deal. The practical demand was that researchers report where samples came from and stop describing undergraduate results as findings about "people" without qualification.

The key studies and demonstrations

Visual perception: Marshall Segall, Donald Campbell, and Melville Herskovits, in The Influence of Culture on Visual Perception (1966), tested nearly two thousand people across seventeen cultures on the Muller-Lyer illusion, in which two lines of equal length look unequal because of angled fins at their ends. American students in Evanston, Illinois, showed a strong illusion; several African societies, including the San of the Kalahari, showed a far weaker one or none. The proposed "carpentered world" hypothesis held that people raised amid rectangular architecture, unlike those raised in round huts, learn perceptual habits that manufacture the illusion, a supposedly hardwired feature of vision that turned out to be, in part, an artifact of environment.

Fairness: Henrich led a team running the ultimatum game, where one player proposes a split of a sum and a second can accept it or reject it and destroy it for both, in fifteen small-scale societies across five continents, reported in "In Search of Homo Economicus" (2001). Standard theory predicts near-zero offers, always accepted; American undergraduates typically offer close to half and reject low offers on principle. Neither held everywhere: average offers ranged from about 26 percent among the Machiguenga of the Peruvian Amazon, who rarely rejected low offers, to well over half among the whale-hunting Lamalera of Indonesia, who sometimes rejected offers above half as suspiciously generous. Fairness intuitions appeared in every society, but their content tracked local economic life, especially market integration, more than any single baseline predicted.

The self: Steven Heine and Darrin Lehman, with Hazel Rose Markus and Shinobu Kitayama, asked in a 1999 Psychological Review paper whether the "self-enhancement motive," the North American tendency to see oneself as above average and protect a positive self-image, was a human universal. Reviewing evidence from Japan, they concluded it was not: Japanese participants showed far less self-enhancement, sometimes mild self-criticism instead, a pattern consistent with a self defined through relationships rather than an independent, trait-based identity.

Related distinctions

Keep the critique apart from cultural relativism, the philosophical claim that moral standards hold only relative to a culture: this is an empirical claim, that psychological processes vary more than assumed and the variation should be measured, not a claim about whether cross-cultural judgment is legitimate. It is also distinct from the replication crisis that surfaced after roughly 2011: replication asks whether a finding holds up when repeated, while this critique asks whether a finding, even replicated perfectly in its original population, was ever entitled to the universal claim made for it.

Lineage

The paper's immediate precursor is Arnett's "Neglected 95%" (2008). Its deeper roots run through decades of cross-cultural psychology that made related points without dislodging the discipline's defaults: Segall, Campbell, and Herskovits on perception, Harry Triandis on individualism and collectivism, Richard Nisbett's The Geography of Thought (2003) on holistic versus analytic reasoning. The contribution of the 2010 paper was to gather this scattered evidence into one quantified argument, under a name compact enough to change how researchers describe their samples.

The strongest case for it

The 2010 paper does not rest on a single anecdote. It assembles converging evidence, across perception, economic behavior, and self-concept, that WEIRD populations are a specific, historically unusual configuration rather than a neutral default. It also changed practice: journals and reviewers now routinely ask where a sample was drawn from, and hedge claims that a finding demonstrated in Michigan describes "human" cognition rather than American undergraduate cognition. Henrich's follow-up book, The WEIRDest People in the World (2020), proposed a historical cause for WEIRD psychology itself: medieval Church policies against cousin marriage that dissolved intensive kin networks and favored an individualistic psychology suited to strangers and impersonal institutions, treating WEIRD as a hypothesis to be explained rather than an insult.

The strongest case against it

Will Bennis and Douglas Medin, in "Weirdness Is in the Eye of the Beholder" (2010), a peer commentary published alongside the original article, argued that calling WEIRD populations outliers is comparison-dependent: the evidence measures them against whatever non-WEIRD samples happened to already exist in the literature, itself an uncontrolled sliver of humanity. The "outlier" finding may reflect which groups researchers had already studied, not a stable fact about where the human distribution's center lies.

A second objection concerns the fix. Henrich, with Michael Muthukrishna, Adrian Bell, and colleagues, acknowledged in "Beyond WEIRD Psychology" (Psychological Science, 2020) that treating "non-WEIRD" as one contrast category understates the problem: a Tsimane forager in Bolivia and an urban professional in Seoul are both, technically, non-WEIRD, yet separated by enormous psychological distance. Adding a single non-Western site swaps one narrow sample for another rather than establishing generality; a real fix requires measuring cultural distance systematically, slower and costlier than a diversity statement offered as a formality.

A third, more skeptical line holds that variation in ultimatum offers or self-enhancement does not prove the underlying cognitive machinery differs. Norms of fairness or self-presentation could vary on the surface while resting on shared computational foundations, as languages differ in vocabulary while sharing deep grammatical structure. Some WEIRD findings, on this reading, document learned content layered on a uniform architecture, and the language of psychological peculiarity overstates how deep the differences run.

Where it stands now

The core sampling finding is not contested: psychology's evidence base remains disproportionately Western and undergraduate. A decade-later audit, Amber Gayle Thalmayer, Cecilia Toscanelli, and Arnett himself, "The Neglected 95% Revisited" (2021), found only modest improvement in the journals Arnett surveyed in 2008, with most of the gain going to other Western countries rather than the majority world. What changed is infrastructure. The Psychological Science Accelerator, founded by Christopher Chartier in 2018, coordinates hundreds of laboratories across dozens of countries to run the same study simultaneously in diverse populations, and large multi-site efforts such as the Many Labs projects, beginning in 2014, now routinely report results broken out by country. Tal Yarkoni's "The Generalizability Crisis" (2022) has since argued WEIRD sampling is one instance of a wider failure, that psychologists generalize beyond their sampled population and their sampled tasks alike. The debate has largely moved from whether the problem exists to how thorough a fix must be.

Test yourself

Think of a psychological finding you have taken as simply true of people: that we anchor on first numbers, that we fear loss more than we value equivalent gain, that we conform under social pressure. Ask where the original evidence came from. If the answer is undergraduates at a Western university, that does not make the finding false. It means you have not yet been told what it is a finding about.