Yale College football fans. Tim Clayton/ Getty


Spot the WEIRDo

Too much research is done on Western, Educated, Industrialised, Rich and Democratic students. Can science widen its base?

by Robert Colvile + BIO

Yale College football fans. Tim Clayton/ Getty

If a boy is to become a man among the Etoro people of New Guinea, he must ingest the semen of an elder member of the tribe, via ritualised fellatio. The same belief is held by the nearby Kaluli. But in their case, the ceremonies require the semen to be delivered via the initiate’s anus, not his mouth. The Etoro despise the Kaluli’s practice, finding it disgusting and unnatural.

If you were asked to sum up the nature of modern adolescence, the Kaluli and the Etoro would not be the first examples that spring to mind. But a few years ago, three researchers in Vancouver became convinced that much of what we think we know about humanity as a whole is in fact the result of studying a tribe that is just as divorced from the mainstream of human experience: the American undergraduate.

For many years, scientists assumed that evolution had wired most of humanity in pretty much the same way. This meant that studying one person’s brain was as good as another’s. So why not use those brains that were closest to hand? But in the mid-2000s, the psychologists Steven Heine, Joseph Henrich and Ara Norenzayan at the University of British Columbia became convinced that this wasn’t actually true. The trio had all done work on how behaviour differs between cultures: Heine on our sense of self; Norenzayan on religion and belief; and Henrich on fairness and reciprocity. What they’d found was that ideas that were taken as universal turned out to be surprisingly culturally specific.

A key discovery came when Henrich asked the Machiguenga people of Peru to play the Ultimatum Game. This is a staple behavioural test in which one player is given some cash, and asked to divide it between himself and a stranger: if the stranger rejects the offer, neither of them gets anything. Logically, we should each accept even the smallest offer – it’s free money! – but, actually, both sides seem to have an innate sense of fairness, making larger offers than is economically rational and blocking smaller ones. Except, it turned out, for the Machiguenga. They made – and accepted – very low offers. ‘They just didn’t understand why anyone would sacrifice money to punish someone who had the good luck of getting to play the other role in the game,’ said Henrich. This was something that had never been seen in the US. Nor had the contrasting phenomenon in which people in many countries reject so-called ‘hyper-fair’ offers that give the lion’s share to the second player.

In a 2010 paper for Behavioral and Brain Sciences titled ‘The Weirdest People in the World?’, Henrich, Heine and Norenzayan argued that what we think of as science is all too often ‘WEIRD’ science: Western, Educated, Industrialised, Rich and Democratic. Between 2003 and 2007, 96 per cent of experimental volunteers in the leading psychology journals were WEIRD; 68 per cent of papers relied exclusively on US subjects; and in the prestigious Journal of Personality and Social Psychology, 67 per cent of total subjects were US psychology students. ‘Many fields have a model organism that they study,’ says Heine now. ‘A lot of medicine is done with mice, a lot of genetics is done with fruit flies. And in psychology, the model organism is the American undergraduate.’

According to Henrich, Heine and Norenzayan, the WEIRD sample was not just biased, but hideously unrepresentative. Given the peculiarities of Western college students, the trio wrote, they might ‘represent the worst population on which to base our understanding of Homo sapiens’. These peculiarities were, they said, apparent in fields as diverse as ‘visual perception, fairness, cooperation, spatial reasoning, categorisation and inferential induction, moral reasoning, reasoning styles, self-concepts and related motivations, and the heritability of IQ’.

Take the Müller-Lyer illusion: the fact that a line with arrows on the end pointing inwards looks shorter than one with arrows pointing out

Except that it doesn’t if you aren’t raised in a Western environment full of straight lines and sharp angles. Likewise, Westerners are the only ones whose ethical reasoning is governed by abstract, ‘post-moral’ considerations such as notions of justice and rights. We don’t even have normal feet: ours are flat and narrow with underdeveloped big toes, ‘the product of a lifetime of having one’s feet bound in cushioned shoes’.

it turns out that WEIRDoes are at the extreme end of a spectrum along which the rest of humanity is scattered

Some of the differences between WEIRD subjects and others are either/or. For example, English is an egocentric language whereas most others are allocentric: English-speakers describe objects’ location in relation to themselves or other people, and not to other objects (for example, ‘the bike is five metres to my left’ rather than ‘the bike is next to the fire hydrant’). But mostly, it turns out that WEIRDoes are at the extreme end of a spectrum along which the rest of humanity is scattered.

Perhaps the best example of this is individualism. Americans are the most individualistic of all the planet’s people (as well as being the most patriotic, litigious, philanthropic, populist, optimistic…) and college students are the most individualistic of Americans. While this makes them narcissistic, it does not seem to make them selfish. WEIRD subjects are among the most cooperative when it comes to playing those redistribution games. Being exposed to the market actually teaches you to play fair, and to expect others to do the same. Individualism means you have internalised your moral code as opposed to depending on your peers for guidance, making you less likely to cheat when unobserved. This is heartening, and fascinating, but it also means that if you erect a universal theory about the nature and extent of human altruism using these particular building blocks, it will be wrong.

The WEIRD bias doesn’t just affect the findings of studies. It affects the kinds of studies that are carried out. Our cultural bias means that not only do we ignore concepts that might be important in other countries – such as face, caste or honour – but that you often end up testing for an English-language concept (‘shame’, for example) which might have no direct equivalent in another society, or have different connotations. In the field of psycholinguistics, for example, the fundamental concepts that are generally accepted also happen to be easy to express in English: proof either that God is an Englishman, or that English-language brains studying English-language brains have come up with English-language theories.

It is not just students who are the cause of cultural bias, either. Most studies of children and development are from families ‘with the time, resources and motivation to bring their infant to participate in a development study at a university laboratory’, as Anne Fernald of Stanford University wrote in her response to the WEIRD paper. These are, she pointed out, ‘even less diverse than the college students who predominate in studies with adults’. At the 2010 International Conference on Infant Studies, less than one per cent of the 1,000 presentations included participants from disadvantaged families, even though they make up 20-40 per cent of children in the US. Other scientists argued that cultural bias even affects studies that compare humans with apes. The humans might be WEIRD but the apes are ‘BIZARRE’: the products not of the jungle but of Barren, Institutional, Zoo and other Rare Rearing Environments.

Then there is the fact that students are, well, students. Not only are they, by definition, well-educated, but they are generally immature. If you play the Ultimatum Game with people of various ages, their responses change over the time. Your level of social trust does not stabilise until your mid-20s or even, for some games, your 30s. ‘All the baselines for proper cognitive functioning are based on 20-year-old university students,’ says Bradley Love, a researcher at University College London (UCL). But the brain changes throughout our lifespan, undergoing changes when we learn new things.

This not only means that we get a distorted picture of what a functioning brain looks like, but that we might have a mistaken idea of when senility sets in. ‘There’s some work coming out suggesting that a lot of the effects of ageing that we’ve attributed to cognitive decline might be down to a different knowledge base,’ says Love. ‘Older people have more concepts, more words, more information to sort through. Even if the mental hardware doesn’t change, it’s like searching through a hard drive with three files vs 3 million files. The information just won’t show up as quickly.’

And even if you can get the right people, of the right age, from a diverse range of cultures into the lab, their behaviour there might not be at all representative of how they would act in real life. For example, there is a theory about our appetite for novelty that holds that our willingness to try something new – a different restaurant or product – is determined by the length of time since our last switch: the longer the time that passes, the more likely we are to want novelty. In the lab, this theory holds true. But by studying people’s real-life choices – in the form of customer data from one of Britain’s big supermarket chains – Love found that it’s actually the opposite way round: ‘Ten per cent of people do follow the laboratory pattern but, with most people, the longer they’ve bought something for, the less likely they are to explore another option.’ When Love’s team sent coupons to customers offering discounts on new products, using their new formula, redemption rates doubled.

‘What economics or psychology research tends to do is put people in the most unrealistic scenario you can ever imagine,’ says Nichola Raihani, another UCL researcher. ‘You’re faced with a computer, you’re completely anonymous, you’re never going to meet the person you’re interacting with, no one’s ever going to know about your decisions. Oh, and you’re forced to play with a particular person. It’s like taking all the things we know are important for social interactions, removing them, and then saying: “When you’ve removed all these things, what do people do?”’

The fact that so much of what we think we know about human nature might apply only to US undergraduates – and, beyond that, to US undergraduates under the artificial conditions imposed by the laboratory – ought to set alarm bells ringing. Heine and his colleagues expected a furious backlash against their findings. Instead, he says, the response to their work has been near-universal praise, but not much action: ‘There was a conference a few months ago that Joe Henrich was speaking at. He was introduced by Jonathan Haidt as one of the authors of this paper that we all read, we all think is great, and none of us is doing anything about.’

You get publicity, and tenure, for fascinating truths about the human condition, not about the US undergraduate. The temptation is to generalise and universalise

Studying the literature, it is possible to find fascinating cross-cultural work. For example, a recent paper by Simon Gächter and Jonathan F Schulz at the University of Nottingham involved asking 2,568 young people across 23 different countries to roll a die and report the outcome, with higher numbers translating into higher earnings. Their personal honesty levels turned out to correlate with the level of rule-breaking in the country as a whole, suggesting that those living in corrupt countries become more dishonest themselves.

Other studies have used cross-country comparisons to test whether particular traits are in-built due to evolution or arise from particular cultural factors. One, the largest ever study on relationship dissolution, asked 5,705 women across 96 countries who had been cheated on by their male partner to fill in an online questionnaire. It turned out that there was, universally, a period of grief and betrayal – but that, after they had recovered, most women ‘won’ by becoming better able to detect the signs of a bad partner. True, there was an online bias to the study (some, indeed, have suggested turning WEIRD into WEIRDO), but it was still better than drawing participants solely from the student body at Georgia Tech.

Still, for every such study, there are a dozen more that stick to the same old WEIRD model. Much of the reason for this comes down to convenience. If you’re working at a small university in a small city, the easiest volunteers to recruit are those on campus: you certainly won’t have the money to repeat your study around the world, even if those countries had the research infrastructure to support your work. Plus, the incentives of the academic world are geared towards rewarding those who publish frequently, with novel findings. You get publicity, and tenure, for revealing fascinating truths about the human condition, not fascinating truths about the US undergraduate. The temptation is always to generalise and universalise.

‘Doing studies in the lab is way easier,’ says Raihani. ‘If you’re in a competitive environment where you’re trying to publish, and keep up with the competition, that produces incentives to take the quickest path, to go for the safe option.’ There are also, she points out, many advantages to working in a lab: not least the degree of control you have over an experiment.

Getting out into the world can also be an expensive and time-consuming business. Raihani gives the example of the honesty study: ‘That’s an amazing undertaking. But it took the guy who did the data-collection three years.’ Especially since all the separate studies had to be done by one person, or different people, to prevent the data being skewed because half the subjects were interviewed by a beautiful young woman, and half a grouchy middle-aged man.

Yet even if you could afford to test your results among every culture on the planet, the findings might still not be granular enough. Shakti Lamba of the University of Exeter has carried out a series of studies among the Pahari Korwa, a forager-horticulturalist tribe in rural India, based on fairness games, or asking them to share valuable resources. It turned out that the variations across the tribe’s separate communities were just as great as those between different cultures. In other words, to talk about ‘Pahari Korwa norms’ is as misleading – and patronising – as it would be to talk about ‘American norms’ when the qualities under examination encompass individuals as diverse as Donald Trump and Oprah Winfrey.

This is not to say that cross-cultural studies, or lab studies, don’t have value. ‘I don’t think it’s necessarily a problem to study the American undergrad, to the extent that you limit your conclusions to the American undergrad,’ says Heine. ‘The problem is when we don’t limit our conclusions, and start saying: “This behaviour is part of human nature, and evolved on the African savannah millions of years ago.” And that’s where we’re making a really big leap.’

One shortcut he suggests, if you’re trying to prove universality, is ‘to try to find a cultural group that seems to be the opposite of WEIRD on as many dimensions as possible – say a subsistence population in New Guinea or Burkina Faso. And if you get a similar pattern of results, that’s a good step towards helping you make that argument.’ The worry for Heine is that the WEIRD problem has been overshadowed by another, bigger one: the replicability crisis. Recently, scientists – especially in fields such as psychology and behavioural economics – have become uncomfortably aware that many flagship results do not hold up: not just in other cultures, but full stop.

‘The replicability crisis is a crisis of internal validity – can we trust the findings of our studies? Do they mean what we think they mean?’ says Heine. ‘And one solution people all seem to be in agreement on is that we need larger sample sizes and more effort to include replication studies. And both of these require more test subjects – which increases the incentive to get convenient samples.’

So how to avoid the danger that, in making science more accurate, we also make it more WEIRD? One solution that some scientists are adopting in order to widen their sample – or just find subjects more easily – is Amazon’s Mechanical Turk (known within the field as ‘MTurk’). This offers researchers access to a pool of people who are willing to carry out tasks, such as answering a researcher’s questions, for a small amount of money.

This is not, however, a magic wand. MTurk is, as Love points out, a weird sample as well as a WEIRD one. ‘You’re selecting people who are odd, because they’re interested in doing these things online for not that much money – and they’re not just doing one or two studies, like in a lab, they’re doing 50. They’re almost like professional subjects.’ The pool of volunteers is also rather more shallow than it appears: a recent paper found that most scientists are reaching only approximately 7,300 people, rather than the 500,000 claimed by Amazon. There are also indications that a small contingent of ‘Western’ MTurk volunteers are actually from India and other parts of the developing world, spoofing their IP addresses in order to make some easy income.

Recruiting volunteers on Facebook rather than MTurk might solve this problem. While it is still a biased sample, since it excludes those without computers or smartphones, it is a vastly more representative sample of humanity, as well as one in which people are more likely to use their real names and locations. And there are also other databases for scientists to explore. Just as Love used supermarket sales data, Raihani and her colleagues recently trawled through data on online charity donations to explore the idea of ‘costly helping’ – that instead of brandishing our tail feathers to attract a mate, we donate excessive amounts to their chosen charity. They found that people did indeed compete to try to top the most recent donation, but only when they were male and the fundraiser was an attractive female.

Part of the solution to the WEIRD problem, therefore, is to abandon the idea that there is one particular gold standard of experiment. (Or, indeed, one particular gold standard of human behaviour.) Instead, theories and results can be tested in the lab, in different cultures, against the findings of big data, and via volunteers from online tools, with each individual result being treated as representing a probability rather than a certainty.

This is not just about getting better data, but about saving lives. In his book The Geography of Madness (2016), Frank Bures shows how all manner of illnesses are culturally determined. In Nigeria, China and Singapore, people under stress might worry that their genitals are being sucked into their body. But there are also Western conditions that do not crop up elsewhere, or do so in very different forms – including pre-menstrual stress. Even the idea that babies should crawl before they walk is based on cultural convention, not biological inevitability.

Western ideas about mental illness largely came from WEIRD research, ideas that were then exported across the world, in the belief that they were universally applicable. In his book Crazy Like Us (2010), Ethan Watters depicts Western psychiatrists almost as missionaries, fanning out across the globe brandishing the DSM (the Diagnostic and Statistical Manual of Mental Disorders) instead of the Bible. In many places, these new arrivals ended up creating the very illnesses they claimed to be curing. In Hong Kong, for example, the Chinese equivalent of anorexia – whose sufferers did not tend to fear fatness, or wrongly perceive themselves as being overweight – was displaced by the US version. In Rwanda post-genocide or Sri Lanka post-tsunami, sufferers of what was labelled post-traumatic stress disorder were pushed into having one-on-one sessions with psychiatrists, as opposed to subsuming their grief in the group. Our coping strategy was imposed on theirs.

There might be no unifying theory of everything, not every disease can be traced to a single gene, not every human comes with the same in-built operating system

Perhaps the most interesting example comes from Japan. This has long been one of the most fruitful markets for Western pharmaceutical firms. But there was one exception. The ‘worried well’ generally did not seek psychiatric help: it was for those who were severely mentally ill. This meant that sales of SSRIs, the anti-depression drugs such as Prozac that are popped in their millions in the West, were virtually non-existent. As Watters recounts, pharma firms sought to change this by essentially redefining the meaning of depression. Adverts appeared describing it as a ‘cold for the soul’ – and sure enough, sales shot up. Yet it is unclear whether the ad campaigns persuaded genuinely depressed Japanese to go to the doctor, or whether they created unnecessary diagnoses. Given the fact that SSRIs raise the risk of suicide in some patient groups (while reducing it in others), it might be the case, although it is impossible to know, that the medicalisation of Japanese depression has cost more lives than it saved.

It’s easy to look at such developments, and at the WEIRD bias more generally, and think that they’re the result of ignorance, or arrogance: a case of the West needing to check its scientific privilege. Actually, it emerges as much from optimism, from the Enlightenment idea that we are, or can be, one great big brotherhood of man. Abandoning that model of humanity will be difficult, because it feeds into a wider narrative about human progress: that science has long been on a journey from complexity to simplicity. The goal has been to burrow past the mess and froth of our daily lives to discover the hidden rules that lie beneath, the basic mechanisms that produce the extraordinary effects we see around us.

In fields as diverse as physics, biology and psychology, we’re finding that it isn’t so easy after all: that there might be no unifying one-line theory of everything, that not every disease can be traced to a single gene, that not every human comes with the same in-built operating system. As a result, we are, in many areas, making a journey back from simplicity to complexity.

Psychologists might have to abandon the idea that our differences are no more than skin-deep – and to work out, via cross-cultural and other studies, which traits and behaviours are part of our in-built human hardware and which are the result of the cultural and environmental operating systems that we layer on top. That will be a fascinating process, but one which is much more complicated, and messy, than the one that went before it.

As part of this process, those of us who are not scientists might also have to abandon some of our cherished illusions. In our own minds, we’re all the Etoro, convinced that the society we see around us is the natural order of things, the natural endpoint to which other cultures should aspire. As a result, we’ve not only spread our mental monoculture around the globe, but have been ignorant of how strange – how WEIRD – we really are.