Jim Springer and Jim Lewis were identical twins, born to a fifteen-year-old unmarried girl. They stayed together for a month before being adopted by families of similar backgrounds who lived about forty miles apart. Lewis’s adoptive mother visited the local courthouse to settle the adoption papers and was told by an official: ‘They named the other little boy “Jim” too.’ Before then, she hadn’t known her son was a twin. Eventually, she told her son, as did Springer’s adoptive mother. In February 1979, when the twins were thirty-nine, they had an emotional meeting. After the case was reported, a Minneapolis-based psychologist, Thomas Bouchard, who had studied twins for a decade, saw his chance. Two weeks later he met them in his office, and a series of detailed interviews began.
It turned out that both had worked at McDonald’s, and at petrol stations, albeit at different times. One had worked as a security guard, the other as a deputy sheriff. They had both taken holidays in the same part of Florida, drove Chevrolets, had dogs called Toy and had married women called Linda and Betty. One had a son called James Allan, the other James Alan. Both preferred maths to spelling, enjoyed carpentry, chewed their nails, smoked Salem and drank the same beer. Both had haemorrhoids, gained 10lb at the same age, started experiencing migraines at eighteen and had similar heart problems and sleep patterns. The story spread around the globe, finding its way into school textbooks and all over the World Wide Web. Some saw this as a case of telepathy between identical twins, others one of the extraordinary pull of genetics.1
Reality was rather less sensational. Over the half-century before 2018, 1,894 separated twins have been tested but none of the others have shown anything like the level of overlap of the Jim Twins.2 There are clear genetic links to heart problems, migraines and tendency to gain weight, and no doubt to sleep patterns, preference for maths and nail-biting too. But Salem cigarettes, dogs called Toy and wives called Linda and Betty? Pure chance, just as there are serendipitous coincidences in most people’s lives that might feel meaningful when considered in isolation. Further investigation suggested they were hardly carbon copies. They lived with similar families, of similar backgrounds, in the same neck of the woods, so we would expect parallels in the choices they had made. But they had different stories to tell. They had changed to different kinds of jobs (Lewis a furnace worker; Springer a meter reader). They had very different hairstyles and facial hair. One turned out to be better at writing, the other at speaking. Springer was ‘more easy-going’ and Lewis ‘more uptight’.3 Springer stayed with Betty; Lewis divorced and married for the third time, to a woman called Sandy.4 Also, curiously, we were never told whether they had the same IQs.
In other cases highlighted by Bouchard, the differences were more marked. But inevitably, the facts chosen by the news media are the headline-grabbers. The Jim Twins always take pride of place but if back-up is needed, their story tends to be followed by that of the former Nazi twin and his Jewish-raised brother. In Trinidad, Oskar Stohr and Jack Yufe were separated at six months old. Oskar was brought up as a Catholic in Germany and joined the Hitler Youth. Jack remained behind, brought up as a Jew. They were reunited in their forties; they had similar speech patterns, gaits and food tastes, and both flushed the toilet before using it. Then there were the ‘Giggle Sisters’, Daphne Goodship and Barbara Herbert. Again, one was brought up Jewish and the other Catholic and they were reunited in their forties. Unlike their adoptive families, they were both incessant gigglers and had a fear of heights.
Details pointing in the opposite direction were played down. In one of Bouchard’s early cases, one twin was adopted by a fisherman, brought up in a home with few books and was not educated beyond early high school; the other was adopted by a cosmopolitan family, lived all over the world and became an electronics expert for the CIA. His IQ was twenty points above his brother’s. In another pair, the IQ gap was twenty-nine points, without any hint of extreme deprivation.5 It seemed that significant differences in educational levels, social class and exposure to knowledge could prompt significant differences in adult IQ.
Most cases showing gaps between separated identical twins are ignored by the media. One much-covered exception involved two pairs of Colombian identical twins, who were mixed up in a hospital error and brought up separately as pairs of fraternal twins, one pair in a poor rural area, the other in a lower-middle-class urban area. When they met, after discovering the error as young adults, the initial reports focused on similarities. Then Nancy Segal, a professor from California State University, formerly Bouchard’s lead researcher, subjected them to interviews, questionnaires and IQ tests. She expected similarity among the identical twins but although their IQ results were not released, she acknowledged they were significantly less alike than anticipated. ‘The Columbian pair really made me think hard about the environment,’ she said.6 Elsewhere she added: ‘I came away with a real respect for the effect of an extremely different environment.’7
Twin studies begin with the eugenicist Francis Galton. In the 1870s he wrote to thirty-five pairs of apparently identical twins and twenty-three pairs of apparently fraternal twins (‘apparently’ because Galton, like his cousin Charles Darwin, was unaware of genetics, so judged on looks alone). He used their anecdotes to draw conclusions: the twins who looked alike had similar personalities, ailments and life interests throughout life, whereas those who looked different became even more different as they aged. Galton claimed that with both sets the ‘external influences have been identical; they have never been separated’ and therefore the results proved ‘nature prevails enormously over nurture.’8
Galton’s sole method of determining whether the twins were identical was appearance. But, as we now know, some fraternal twins look very similar and some identical twins do not look identical, and his conclusions were drawn entirely from what his group said about themselves in response to his letters. Still, today’s hereditarians are inclined to give him a thumbs-up. The science writer Matt Ridley, for one, endorses Galton’s results:
He would have to wait more than a century to see that the study of twins did in the end prove much of what he had suspected. To the extent that they can be tested apart, nature prevails over one kind of (shared) nurture when it comes to defining differences in personality, intelligence and health between people within the same society. Note the caveats.9
Galton used his conclusions on twins to promote eugenic beliefs about purifying the population, which, a decade after his death, became the essence of the Nazi credo. It was under the Nazis that the next twin studies took place, through Josef Mengele’s notorious research in Auschwitz that involved 1,500 twin pairs. The physician, known as the ‘White Angel’, the white-coated man who determined who went to the gas chambers, would order any twins among incoming prisoners to step out so that he could experiment on them. He ordered two Jewish doctors to assist him. Twins were treated better than the rest – sometimes given sugar and little gifts – but invariably as a prelude to horrendous experiments. These included injecting their eyes with chemicals to try to change their colour, amputating limbs, removing organs and sterilising women without using any anaesthetic. Most died, or were killed later. In one case, one of his assistants injected chloroform into the hearts of fourteen pairs of Roma twins, after which Mengele dissected their bodies. He sewed a pair of Roma twins together to create conjoined twins; they died of gangrene. He connected one girl’s urinary tract to her colon. Sometimes he simply shot them and then dissected them.
These revelations gave twin studies a bad name but they continued, mainly because those wanting to find whether a particular trait had a genetic origin had little alternative. Twin studies have since been used in several countries to test anything from whether vitamin C can prevent colds (it can’t) to whether homosexuality has a genetic origin (minor influence for gay men and even smaller for lesbian women).10
All these studies, like Galton’s, involved comparing the behavioural traits of monozygotic (single egg, identical) twins and dizygotic (two egg, fraternal) twins. This method was also used by a team headed by Robert Plomin, the British psychologist we met in Chapter 9, who acknowledged failure when it came to finding a single gene for intelligence. His subsequent research prompted a wave of media publicity, including a lead story in the Guardian entitled ‘Genetics accounts for more than half of variation in exam results’,11 which concluded that ‘genetics accounts for almost twice as much of the variance of GCSE scores (53 per cent) as does shared environment (30 per cent).’12 Their method was to use a national sample of 11,116 sixteen-year-old twins, and compare the marks of identical and fraternal twins, on the assumption that both shared the same environment but only the identical twins were genetic carbon copies. Towards the end of their published paper, the team noted potential problems, including ‘the equal environments assumption – that environmentally-caused similarity is equal for MZ and DZ twins – and the assumption that results for twins generalize to non-twin populations’.13
To illustrate why this method is more than just a minor concern, meriting no more than a sentence at the end of a paper, I’ll use two examples of twins in my classes at school. One pair – I’ll call them the Thompson twins – were so identical that I never managed to tell them apart. They always seemed to get the same marks, be in the same sports teams and have the same friends. Then there were the Wellingtons, Amy and Mary, whom I knew in Texas. They didn’t look anything like each other, and everyone related to them differently. It never seemed a big deal to Amy, but she was regarded as unusually pretty and everyone wanted to be her friend. Without having to try hard she was the life of the party, and was effortlessly sporty. People gravitated towards her and everything came easily, so she didn’t put effort into her schoolwork and preferred hanging out, smoking cannabis and having fun. Mary was regarded as average-looking and very ‘straight’. She had a small group of friends, steered clear of cannabis, focused on academic work and excelled. I have no clue what their IQs were, but I’d expect a gap between the Wellingtons, whereas it would surprise me if the Thompsons were more than a point apart.
At first blush this seems to confirm Plomin’s assumption that the gap between identical twins is narrower because of their shared genes. But it could also be influenced by real environmental differences. Everyone treated the Thompsons alike but the Wellingtons less so; although they were born just a few minutes apart, their life experiences were distinct. Partly because they looked so different, they were treated differently. Amy was doted on by teachers, coaches, schoolmates and family members. Mary did not get the same attention and grew up without any expectation of adoration. They carved out different paths early in life; Mary became more studious, more focused, less casual. She read more, studied harder and excelled in areas her sister didn’t bother to contest.
I’ve known fifteen sets of twins and triplets since childhood. Eight said they were identical and seven fraternal, which is unusual.14 The identical twins looked alike, were treated alike by their teachers, had the same friends, went to the same parties. They saw themselves as part of a set. It was different for the fraternal twins. There were significant distinctions in how they were treated by classmates, teachers and parents. They found separate niches and were regarded as individual people, leading to different life experiences. Genetic determinists might say: ‘Aha! Your fraternals were treated differently because they were genetically different.’ And they’d be right. But this raises a problem for the hereditarian assumption that identical and fraternal twins share each other’s environments in the same way. It doesn’t matter why they were treated differently. The mere fact of it – whether through looks, abilities or inborn cognitive differences – means we cannot assume the differences in the abilities are primarily genetic, which is precisely what is assumed by people such as Plomin, who insist on devising percentages for the nature-nurture components of IQ and academic performance.
To test this, we might ask if any of the differences in the fraternal twins’ experience affect IQ scores. Amy drank more alcohol, smoked cannabis and spent less time on school work than her sister, but had more affirmation and was more self-confident as a result. As we shall see, all these factors could affect the results of IQ or academic tests. For fraternal twins there is also a significantly higher chance of different experiences that can affect outcomes – an inspiring mentor, a life-changing overseas trip, a bump on the head – and these different experiences are likely to be more marked as they grow older.
There’s an additional environmental factor that twin studies ignore: the experiential gap starts in the womb because fraternal twins each have their own placenta, and one twin might be bigger than the other, whereas three quarters of identical twins share a placenta, and some share the same amnion (the membrane that protects the embryo). They are also more likely to share viruses and to be exposed to the same accidents. One meta-analytical study that focused on the impact of the foetal environment on IQ concluded it accounted for 20 per cent of the explanation for IQ differences or, in the case of twins reared apart, it explained 20 per cent of the reason for their similar IQs.15 Although this is a biologically related explanation, it is not a genetic one, yet the design of twin-based research studies ignores the likelihood that the life experiences of fraternal twins, both in the womb and subsequently, are less similar than those of identical twins. These studies start with the false assumption that the twins’ shared environment is the same and their conclusions suffer as a result.
The other method of studying twins to determine heritability – finding separated identical twins – started in the UK in the 1950s under Charles Spearman’s errant disciple, Cyril Burt. When Sir Cyril died of cancer in 1971 his reputation in his profession was unmatched, but it soon collapsed. In retrospect, the clues might have been picked up earlier, when he made the false claim that he, not Spearman, was the father of factor analysis in IQ testing. Burt first tried this claim before Spearman’s death in 1945 but apologised after being put down by the older man. Once Spearman died, however, Burt campaigned to undermine his mentor’s influence, and stopped citing his publications. Also, he was in the habit of writing letters, under various aliases, to his own journal, and shortly before his death he ordered that all his notes and records be burnt. But this was small potatoes compared to what was to come.
Burt made his claims about the heritability of IQ in a series of academic papers reporting studies of separated identical twins that he claimed to have conducted. Unlike in China, where the one-child policy led to many cases of separated identical twins, in the UK these were and are notoriously hard to find, for the simple reason that parents and adoption agencies keep twins together. Yet somehow, Burt found fifty-three sets. Initial queries were raised by the psychologist Leon Kamin, of Princeton, whose team investigated Burt’s work, publishing their study in 1974.16 Kamin noted that Burt had increased his claimed twins’ sample from twenty-one in 1955 to fifty-three in 1966, yet the average correlation in their IQs remained unchanged, down to the third decimal place. In the 1950s Burt claimed a correlation in the IQs of separated identical twins of 0.771; in the 1960s a paper published under the name of his supposed assistant, Margaret Howard, also cited a correlation of 0.771. Later, Burt and another supposed assistant, J. Conway, published a paper with the same correlation, 0.771. Kamin’s Princeton team deemed this coincidence so unlikely as to fit into the category of impossible.
Some of the details of the separated twins Burt claimed to have found were more suggestive of a fertile imagination than real life. They included a pair of illegitimate identical twins from a well-to-do birth mother; one twin grew up on the Scottish country estate of a prosperous family, the other was brought up by a shepherd, a case that sounds as if it emerged from a reading of Perdita’s story in Shakespeare’s A Winter’s Tale.17 What is even more curious is the close correlation in their reported IQs (118 and 121).18 As we have seen, when separated identical twins are brought up in different class and educational backgrounds their IQs tend to vary considerably, by twenty-nine IQ points in one case.
Two years after Kamin’s exposé, the medical correspondent of the Sunday Times, Oliver Gillie, made the direct accusation that Burt had faked his data. He also wrote that Burt’s two collaborators, Margaret Howard and J. Conway, either didn’t exist or were not in contact with him at the time, even though Burt had included their names as authors or co-authors of some of his papers and they had reviewed his books positively. There was no record of either woman having worked at UCL, where Burt was based, of either having published anything independently, and no one remembered them working with Burt. Towards the end of his life, when asked, Burt said they had both ‘emigrated’ but was unsure where, which is odd; one would think he would ask where they were going when they said they were leaving the UK. Subsequent research suggested it is likely that both women did exist, although it is doubtful either wrote the papers or reviews attributed to them. And there is no evidence of the existence of a third woman, Miss M. G. O’Connor, whom Burt also named as a collaborator.19
Burt’s backers leapt to his defence, particularly those who’d used his twin research. They included his former student Hans Eysenck, who wrote to Burt’s sister that it was all a plot by ‘left-wing environmentalists’.20 Arthur Jensen and J. Philippe Rushton (who relied on Burt’s twin data to support his claims about innate intelligence) claimed that the correlations were unsurprising, while the eugenicist sociobiologist W. D. Hamilton said Burt’s opponents were either wrong or had made gross exaggerations. Two other Burt supporters, Ronald Fletcher and Robert Joynson, wrote books backing him.21 Fletcher provided evidence of Howard and Conway’s probable existence, while conceding the absence of evidence for O’Connor.22
But by then the killer blow had landed. It came from a surprising source: Burt’s close friend, Leslie Hearnshaw, whom Burt’s sister commissioned to write an official biography. Hearnshaw started as a Burt enthusiast, but his 1979 book, Cyril Burt: Psychologist,23 shattered Burt’s reputation. Hearnshaw concluded all the allegations against Burt were true, that the fraud had begun nearly thirty years before Burt’s death, and that all his studies on separated twins were invented. He also accused Burt of fabricating other claims, including those about declining intelligence levels in Britain. Today, it is widely accepted that none of Burt’s postwar research can be trusted. The entry in the Encyclopaedia Britannica puts it gently: ‘[A]ll sides agreed that his later research was at least highly flawed and many accepted that he fabricated some data.’24 Others, less gentle, regard Burt and his career as a salutary example of scientific fraud and establishment gullibility.25
Burt’s descent into pumping out fake news may have been influenced by the effects of suffering from Meniere’s disease, a hearing and balance disorder that can cause heightened ‘anxiety and neuroticism’,26 as well as by personal setbacks and his fear that the hereditarian case was being undermined by evidence of the brain-altering impact of early intervention education schemes. These interventions were based on Mark Rosenzweig’s research in the early 1960s, which showed that an enriched environment increased the volume of the cerebral cortex of rats.27 Stephen Jay Gould, however, softened his earlier condemnatory approach, arguing that ‘the very enormity and bizarreness of Burt’s fakery forces us to view it not as the “rational” programme of a devious person trying to salvage his hereditarian dogma when he knew the game was up (my original suspicion, I confess) but as the actions of a sick and tortured man.’28
The next big wave came in 1979, with the Jim Twins case. After this fortuitous success, Thomas Bouchard received generous grants from the overtly racist Pioneer Fund, set up in 1937 to promote eugenics. The fund was inspired by a visit to Nazi Germany made by its anti-Semitic billionaire founder, Wickliffe Preston Draper, who later said his mission in life was to ‘prove simply that Negroes were inferior’.29 Draper funded segregationist causes and opposition to civil rights and backed the compulsory sterilisation programme in North Carolina. He was succeeded by fellow Nazi-backer Harry Laughlin, who campaigned for eugenics and segregation in the American South. The Pioneer Fund’s current stated aim is to research or aid research ‘into the problems of heredity and eugenics … and research and study into the problems of human race betterment’.30 In recent years it has funded people promoting racist psychology with the aim of emphasising IQ variation among races.
Bouchard approached the Pioneer Fund himself. And you don’t pick an overtly racist funder naively. Nor was it a one-off. Bouchard joined the usual suspects of racist psychology in signing Linda Gottfredson’s race-based ‘Mainstream Science on Intelligence’.31 Presumably he did so because he agreed with its agenda. He also offered an enthusiastic back cover endorsement for Rushton’s overtly racist book, Race, Evolution and Behavior. Some of Bouchard’s cheerleaders, such as Ridley, have played down the Pioneer Fund issue:
Their motive in supporting Bouchard’s research is presumably that they want to believe genes influence behaviour, so they give money to a researcher who seems to be getting results which support such a conclusion. Does that mean that Bouchard and all his many colleagues … have faked their data to please funders? Seems pretty far-fetched.32
Ridley is right, but we have only to look at how the racist agendas of the nineteenth-century skull-fillers and twentieth-century IQ pioneers distorted their use of data to make us wary of research that carries an unacknowledged political load. As we shall see, Bouchard consistently overestimated the genetic contribution and understated the environment.
Bouchard’s Minnesota Center for Twin and Adoption Research was building its twins larder when a new book by clinical psychologist Susan Farber was published.33 She went through several major studies of identical twins brought up separately (ninety-five sets in all) and found that in 90 per cent of the cases the twins were selected because they were notably similar. The reason for this bias was that the easiest way of finding separated twins was to advertise; those who responded had already contacted their siblings and believed they were identical because of their similarities. Many had been reunited years before and had substantial contact since, reinforcing their points of similarity. Farber also questioned what ‘reared apart’ meant, finding that most spent years together, often being adopted by relatives, which meant that they grew up in the same kinds of families. When she examined their reported IQs, she found that the more separately they were brought up, the greater the IQ gap. The largest reported study (other than Burt’s fraudulent one) involved forty-four pairs of separated identical twins. The data show thirteen pairs lived together for at least their first two years, and five of those for four years or more. Of the remaining thirty-one who were separated before their second birthdays, nine were reunited by the time they were twelve. Most were brought up in similar environments.34
The late Leon Kamin, the Princeton psychology professor who first exposed Burt’s twin studies fraud, was dismissive of the publicity-seeking Bouchard and the entire twins industry, saying it was ‘not serious scientific work’. He went on to argue:
If you lived in a science fiction world, you could scatter these twins randomly and then go back and find them but they don’t get put into homes by storks. They are put into homes by adoption agencies and sometimes the parents. … [T]here is a grotesquely underestimated degree of previous contact and … an enormous amount of pressure on the twins to come up with cute stories. The whole thing is a media show-business hype that simply strains credibility.35
Bouchard absorbed the criticisms and tried to avoid the potential pitfalls. Because of the publicity he’d courted, separated twins had started contacting his Pioneer-funded team of eighteen researchers. The twins were given honoraria, subjected to a battery of fifty hours of tests, including of IQ, and asked thousands of questions about family, childhood, personal interests, jobs, values and aesthetic judgements. By 1990, Bouchard’s ‘Minnesota Study of Twins Reared Apart’ had analysed data from fifty-six separated identical pairs. He took into account the time his twins had spent together (an average of 5.1 months before separation and 20.3 months following the first reunion) but, contrary to Farber, concluded that there was ‘little evidence’ that the degree of contact before or after separation ‘counted for much’.36
Adult IQ was 70 per cent heritable, Bouchard claimed, explaining that his figure was higher than previous estimates because his studies involved adults and the genetic component of IQ increases with age. This is a common perception among twin researchers; Plomin tells us that the heritability of intelligence increases from 40 per cent in childhood to 65 per cent in young adulthood and to 80 per cent at the age of sixty-five.37 But the explanation offered for this increasing ‘heritability’ (a term relating entirely to genetics) is environmental. Plomin, for example, explains that genetic effects ‘could be amplified as we increasingly select, modify and create environments correlated with our genetic propensities’.38
The IQ theorists Jim Flynn and William Dickens saw it differently, using an example of bright separated identical twins who go to the library, get into top-stream classes and attend university. Flynn asks: ‘What will account for their similar adult IQs? Not identical genes alone – the ability of those identical genes to co-opt environments of similar quality will be the missing piece of the puzzle.’39 Unlike Bouchard and Plomin, he places this ‘multiplier effect’ in the environmental column, saying that if we use a figure of 45 per cent for the heritable contribution to childhood IQ, it will have to be reduced because the children would have already done some matching of their genes to their environments. ‘Let us say that one-fifth of the value is due to matching. Then the direct effect of genes on IQ accounts for only 36 per cent of IQ variance.’40 And this becomes more marked as the child grows up. The change is therefore due not to increasing heritability but rather to environmental matching; to the way the person finds peers, classes and experiences that match their genetic inheritance.
This ‘positive feedback loop’ works better in a middle-class milieu than a deprived one in which opportunities for intellectual stimulus are limited. Evidence comes from a 2003 study by Eric Turkheimer, professor of psychology at the University of Virginia, who examined data on twins from a study of children, many from working-class families, who had taken IQ tests at the age of seven. For the poorer children, the IQs of identical twins differed just as much as in fraternal twins. Turkheimer concluded that the impact of growing up poor trumps genes: ‘If you have a chaotic environment, kids’ genetic potential doesn’t have a chance to be expressed. Well-off families can provide the mental stimulation needed for genes to build the brain circuitry for intelligence.’ He added that the wealthier the family, the more likely it is that the child’s genetic potential will be ‘maxed out’.41 But again, there’s another way of seeing it. Instead of speaking of genetic potential being realised in wealthy homes, you could talk of the well-to-do environment enhancing an inherited edge.
These different readings of the same data illustrate the difficulty of attaching percentages to the heritability of behaviour. Some geneticists warn against following this route. One of Britain’s leading neuroscientists, Steven Rose, put it bluntly: ‘Heritability estimates become a way of applying a useless quantity to a socially constructed phenotype and thus, apparently, scientising it – a clear-cut case of Garbage In, Garbage Out.’42
Full DNA analysis has given us more tools with which to measure heritability but it does not necessarily get us much closer to finding precise nature-nurture ratios for behaviour. Even under experimental conditions, in which we can draw on mathematical data about the comparative contributions of genes and environment, it is tricky, because of their interaction. The Stanford biologist Paul Ehrlich notes that this ‘cannot be decomposed into nature or nurture because the effect of each depends on the contribution of the other’.43
This view is reinforced by a study, published in Nature Genetics in 2018, in which eleven scientists looked at data from 54,888 Icelanders. It found that heritability estimates for traits such as height, which, they say, is really 55.4 per cent, and educational attainment (17%), have been seriously overestimated (Plomin, for instance, puts the heritability of school achievement at 60%44) because environmental effects have been included in the genetic category through the misreading of twin studies, which placed nurture in the nature column by the underestimation of the contribution of family and other environments. The Nature Genetics study used a different method, relatedness disequilibrium regression, of estimating heritability, which, the authors say, avoids the problem of conflating environmental and genetic contributions under the genetic heading, prompting far lower heritability estimates.45
Another problem with Bouchard’s figures is that they are based on ‘the assumption of no environmental similarity’ in the circumstances of separated twins.46 This is impossible, so he concedes similarity could account for 3 per cent of the result, even though almost all were brought up in middle-class homes. Only a few were ‘reared in real poverty and none were “retarded” [sic]’, which means ‘this heritability estimate should not be extrapolated to the extremes of environmental disadvantage still encountered in society.’47 He adds that non-middle-class environments might increase the environmental percentage, recognising that his data merely suggests that for adults ‘in the current environments of the broad middle class, in industrial societies, two-thirds of the observed variance of IQ can be traced to genetic variation.’48 He concedes that ‘in individual cases, environmental factors have been highly significant,’ referring to the case in which the twins’ IQs were twenty-nine points apart. Richard Nisbett, a psychology professor and IQ theorist at the University of Michigan, believes twin researchers are wedded to the idea that adopting families are different, when the opposite is true; they’re mostly middle class and well motivated to give their children a good start in life. ‘Adoptive families, like Tolstoy’s happy families, are all alike,’ he says.49
In Bouchard’s later studies, several aspects of common human behaviour (sense of humour, religious affiliation, food preferences, social and political attitudes) were found not to be heritable. By the late 1990s, he was toning down his claims: ‘There probably are genetic influences on almost all facets of human behaviour,’ he said, ‘but the emphasis on the idiosyncratic characteristics is misleading. On average, identical twins raised separately are about 50 per cent similar – and that defeats the widespread belief that identical twins are carbon copies. Obviously, they are not. Each is a unique individual in his or her own right.’50
Bouchard’s earlier figure of 70 per cent heritability for adult IQ is at the high end of the scale. What do other IQ hereditarians say? Plomin has plumped for 50 per cent as a lifetime average51; others go for 40 or 45 per cent. Flynn suggests 36 per cent for children, while Kamin provocatively suggested it could be zero.52 Surveying the twin-related literature in this area, Ridley said that studies have ‘converged’ on the same conclusion: that IQ is ‘approximately 50 per cent additively genetic; 25 per cent influenced by shared environment; and 25 per cent influenced by environmental factors unique to the individual’.53
With the possible exception of Kamin – who was being flippant – no one seriously disputes that intelligence, however measured, is heritable. Despite all the problems with twin studies, we can assume that identical twins will usually have more similar IQs than fraternal twins or siblings, and that one of the reasons is that identical twins have identical DNA. But it would seem that the heritability figure of 50 per cent, in a middle-class population, is based on a failure to consider the limitations of twin studies and particularly the inclination to put environmental influences in the genetic column as the twins grow older. It therefore overstates the genetic contribution to individual IQ, which is not the same as denying the role of nature.
Where does this leave us? Ridley talks of the ‘sweeping successes’54 of twin studies. But, at least with respect to IQ, the opposite conclusion seems more apposite. None of the studies managed to eliminate, or sufficiently reduce, the inbuilt problems of using twins to devise a convincing heritability figure. If we lived in a dystopian world in which all identical and fraternal twins were separated and randomly scattered around the globe, some among the American middle classes, rather more among the Chinese working class, many in the slums of Africa, Asia and the Americas, in rural areas, rain forests and deserts, and compared their IQs, what would the heritability figure be: 20 per cent; 10 per cent? Or would the idea of deducing percentages be abandoned altogether?
Heritability percentages relate to the question of how much genetic factors play a role in individual differences in a particular trait in a particular population. This raises the question of how to define populations when assessing the validity of statements such as: ‘IQ is highly heritable.’ If the population is white middle-class American, that might appear to be the case. Widen it to include all white Americans, and the genetic contribution shrinks. Widen it further to include all Americans, and it shrinks some more. Make the population the whole world and the genetic contribution to differences in IQ would be tiny. Look dispassionately at the record of twin studies and it is hard to avoid the conclusion that they arrive at heritability percentages by little more than a thumb-suck. All too often, twin studies tell us less about IQ than they claim and more about those who choose to follow this branch of research.
Another way to assess the genetic impact on IQ is to study adopted children who are not twins. Two French psychologists, Christiane Capron and Michel Duyme, perused thousands of records from public and private adoption agencies. Their results showed how genes are expressed through environment. Children from well-off families placed in well-off families had average IQs of 119.6, but children from well-off families placed in poor families had IQs averaging 107.5. Children starting out with poor parents before being adopted by poor families had an average IQ of 92.4 but those from poor families placed in well-to-do homes averaged 103.6.55 Incidentally, there are several possible environmental reasons why children with poor biological parents had lower IQs; poorer mothers are more likely to smoke, drink, take drugs and experience stress and violence when pregnant, and poorer children are more likely to be neglected in the months or years before adoption.
One hereditarian shibboleth is that only extreme poverty affects IQ, and other environments make little difference. As Ridley put it: ‘[L]iving on a few thousand dollars a year can severely affect your intelligence for the worse. But living on US$40,000 a year or US$400,000 a year makes little difference.’56 This was put to the test in a study of severely underprivileged children who were adopted between the age of four and six. Most had been abused and neglected, going from institution to institution or foster home to foster home. Their IQs averaged 77. Nine years on they were tested again; all had improved but some had improved more than others. Those adopted by farmers or labourers scored 85.5, by middle-class families, 92 and well-off families, 98.57 Contrary to the idea that IQ is fixed by nature, this illustrates that it can be hugely influenced by social circumstance. And, contrary to Ridley’s idea that the level of wealth made little difference, there was a six-point gap between the middle class and the rich, and a 12.5-point gap between the stable working class and the rich. This fits with the best-known adoption study, carried out by the American psychologists Skodak and Skeels, in 1949. They tested ninety-eight poorly educated working-class biological mothers whose IQs averaged 85.7. Their children, whose adoptive parents were better educated, had an average IQ of 116.8 at two years old and 108 by the age of 13.58
These studies discredit claims that family environment, schools and income make little difference to educational and IQ outcome. Perhaps the most extreme claim is Robert Plomin’s, who has been on a perpetual quest to find ballast for his nature over nurture convictions.59 In Blueprint, published in 2018, he argues that the most important environmental factors, such as families and schools, ‘account for less than 5 per cent’ of the explanation for academic performance. Genetics – ‘the blueprint that makes us who we are’ – accounts, he says, for 50 per cent. The other 45 per cent he puts down to ‘random experiences’ whose effects don’t last: ‘after these environmental bumps we bounce back to our genetic trajectory.’60
When we consider the results of studies showing long-term IQ changes as a result of adoption into wealthier, more-educated families, this claim seems ridiculous. Assessing the evidence on adoption studies, including those involving twins, Nisbett said that IQ theorists such as him had consistently underestimated the impact of nurture and of family in particular: ‘Unfortunately, for many years I bought the claims of the hereditarians that family environments don’t matter much.’ Illustrating this point by referring to the impact of cross-class adoptions, he added: ‘Raising someone in an upper-middle-class environment versus a lower-class environment is worth 12 to 18 points of IQ – a truly massive effect.’61
Overall, the impact of environment on IQ has been understated and the genetic impact overstated, which is not to say it doesn’t exist. Even Stephen Jay Gould, a fervent critic of hereditarianism, said it would be hard to find any aspect of life that had no genetic component. ‘The hereditarian fallacy is not the simple claim that IQ is to some degree “heritable”. I have no doubt that it is, though the degree has clearly been exaggerated …’62 Instead, he said, a major fallacy was to conflate heritable with inevitable. ‘The claim that IQ is so-many-percent “heritable” does not conflict with the belief that enriched education can increase … “intelligence”. A partially inherited low IQ might be subject to extensive improvement through proper education. And it might not. The mere fact of its heritability permits no conclusion.’63
There is a second, even more serious, fallacy concerning race: the conflation of IQ inheritance within a population group with IQ inheritance between population groups.64 Studies of the heritability of IQ have all been ‘within-group’, examining people brought up in the same population group and social class. But to assume that IQ might be, say, 50 per cent heritable among white middle-class Americans does not mean that differences between the IQs of white Americans and black Americans are also 50 per cent heritable, because there is no reason to assume that variation within a group is the same thing as variation between groups. Variation within the group may involve a mix of genetic and environmental factors; the precise mix will be influenced by the size of the group, economic well-being, age and socio-economic range. Differences between groups may be entirely environmental.
I’ll paraphrase an illustration suggested by the Harvard biologist Richard Lewontin: let’s say we have a bag of mixed seeds taken from genetically diverse varieties of wheat. We take two cupfuls and grow them separately, in the same conditions, except for one detail: one batch receives special nutrients. After a month the plants are measured. Within each batch there is some variation in size, a result of the different genes; within each population, heritability is high. There are height differences between the batches; the batch that got the nutrients is taller, on average. The differences between the batches are entirely a result of the nutrients – in other words, an environmental cause.65 The relevance to the variation in IQs between black and white Americans does not need spelling out.
Let’s consider something that is more heritable than IQ: height. If we take the tallest nation in South-East Asia, whose men average 1.74 metres in height, and compare them to the shortest, whose men are thought to average 1.66 metres, we can’t assume the difference is genetic simply because of the hereditability of height. The fact that the ‘within-group’ heritability is high tells us nothing about the possibility that receiving better nutrition and suffering fewer diseases might increase the average height of the shorter nation. (The two nations are South Korea and North Korea, and the average height of the 8cm-taller South Koreans continues to rise because of better nutrition.) Another example would be Japan, where the average height of eleven-year-old boys has increased by 12cm since the mid-twentieth century. ‘Heritability’ refers to the degree of variation in a trait directly caused by genes in a population but it doesn’t relate to the proportion of that trait attributed to genes. If the environmental factors relevant to height change to affect all members of a population (such as improved nutrition in South Korea), the trait’s mean value might be altered (taller South Koreans) but its heritability will remain the same. However, degrees of heritability can change if some members of a population experience environmental effects. For example, if some North Koreans are malnourished and others well fed, then variation in that population would be greater.
If we apply this to IQ, we can see why group comparisons are facile. The environmental factors relating to IQ might differ substantially between groups. As with the height of the North and South Koreans, the assumption that the differences were the result of genetics would be wrong. What is the equivalent of better nutrition that could prompt a gap in average IQ scores between populations? IQ differences are not the same as height differences in wheat or people; one factor X is not enough. We need a more comprehensive answer.