by Ian J. Deary
The most recent and best-known dataset of this type is the Minnesota Study of Twins Reared Apart (MISTRA). In the Minnesota Centre for Twin and Adoption Research (MICTAR), Tom Bouchard and his colleagues have the privilege of bringing together from all over the world the MISTRA twins (identical and non-identical) – and some triplets – who were separated during childhood and typically for most of their lives until that point. For a week in the MICTAR they are taken through 50 hours of psychological and medical tests and questionnaires. Their physical state, abilities, personalities, work patterns, and personal lives are documented as fully as the time allows.
The human interest in the study alone is astonishing. Bringing twins or triplets together after they have spent most of their lives apart is something that appeals to our emotions and curiosity. Tom Bouchard communicates that side of things very well. He has pictures of two male identical twins who were both firemen and who looked identical despite not having spent their lives together (Figure 19). Here’s the report of Gerald Levy’s and Mark Newman’s meeting. “Both sport sideburns and moustaches of equal length and with similar curl, both wear metal-frame aviator-type eyeglasses. Their mannerisms are alike, their voices indistinguishable, their gaits identical.” Says Newman: “Every time we did something it seemed to be in unison. That’s when it really started to get scary.” But I don’t want to focus on the anecdotal side of the study, because the hard psychological facts themselves are just as astonishing.
19. One of the pairs of twins taking part in the Minnesota Study of Twins Reared Apart.
Among all these tests, one of the areas of function that is examined most scrupulously is mental ability. Each pair of twins gets a large set of cognitive ability, intelligence-style tests. Then the researchers correlate the test scores to discover whether one member of a twin pair tends to get the same score as the other member. Among the tests they receive is the full Wechsler Adult Intelligence Scales, an earlier version of the one that we met in Chapter 1. It takes about an hour and a half or more to administer, with a different examiner testing each member of the twin pair. How similar are pairs of identical twins who lived apart for most of their lives? Well, their total scores on the Wechsler battery of mental tests correlate at 0.69. This is a very high correlation and not that much different from pairs of identical twins who have spent their lives together and whose scores correlate at 0.88. For some other mental ability tests, the correlations from the Minnesota study were the same for the identical twins who were reared together and apart. For example, Raven’s Progressive Matrices is reckoned to be one of the best single tests of the general factor in human intelligence. The correlation of Raven’s test scores (with a vocabulary scale score added to it) for reared-apart identical twins was 0.78. For reared-together identical twins it was 0.76.
That’s the principal and surprising result. Identical twin pairs who spent their lives apart end up just about as similar in intelligence as those who spent their lives together.
Later we’ll look at whether the results of the Minnesota study could be due to things other than genetic similarity. For now, though, we must fully appreciate their near-incredible conclusions. The results indicate that, on intelligence, identical twins who have lived separate lives are almost as alike on intelligence test scores as identical twins who lived shared lives. Look back again at Figures 15 and 17, concerning identical twins reared together and apart. The things that tend to make members of a twin pair similar are the circles that have two arrows emitting from them. The ‘apart’ twins share only genes. The ‘together’ twins share genes and common (family) environment. What can we conclude, then, if we find that the ‘apart’ and ‘together’ twins are just about as alike on intelligence? The conclusion is that the C factor, common environment, has a negligible effect. Both types of twin pair share only the fact that they have identical genes, so the genes seem to be important. A counter-intuitive and rather unpalatable finding to assiduous parents: that family upbringing has very little effect on intelligence level. Most of us would begin with the opposite assumption.
Let’s just drive home how similar identical twins are even when they have lived apart for most of their lives. Look at Figure 20. (Don’t bother about the details in the graph at the moment. We shall come to that in a minute or two.) For now, just look at the numbers that run vertically up the side. They are IQ score differences: 10, 20, 30 points, and so on. Notice the top horizontal line. It occurs at about an IQ difference of 18. It’s the average difference between two people picked at random off the street.
20. A diagram to show the similarity of IQ scores in identical twins reared apart (MZA). It also shows that there was no correlation between time spent together and similarity in IQ scores.
Now let’s take the opposite extreme. I give a mental ability test to people and then test them all once again. I want to compare how the same person scores when tested twice. Now, each person won’t get exactly the same score. We’ll find that there is a bit of wobble. Some might be a bit sharper on one day than the other, might be more distracted, might just have had a cup of coffee, might be thinking about a fight they had at home that morning, and so on through the complexities of human conscious thought. We are not totally reliable machines and mental ability tests will not always register the same score on the same person. (If we took their temperature or their blood pressure we would not get exactly the same result when we tested them twice either.) The average amount of wobble for the same person is about 5 or so IQ points, and you can see that horizontal line drawn along Figure 20. It’s marked as ‘two testings of the same individual.’
We are now set to answer the key question. How alike are identical twins reared apart – people with the same genes but brought up in different environments? If the family environment is all-important and they have not shared it with their twin, then they might be as alike as our two hypothetical random strangers. If genes are more important they might be more like the same person tested twice. Have a look at Figure 20 again. The answer is the horizontal line marked ‘mean MZA difference’ (identical [‘monozygotic’] twins reared apart). It’s only marginally higher than the same person tested twice. Identical twins reared apart are very similar in intelligence.
All we know so far is that this remarkable study of twins reared apart has told us that identical twins who do not spent their lives in contact with each other turn out to have highly similar levels of intelligence – nearly as similar as the same person tested twice. Some of what we know points to that similarity being caused substantially by genetic similarity, but we can think of other possibilities too. In fact, three things come to mind straight away.
1 The twins spent time together in their mother’s womb.
2 Members of each twin pair might have been placed in very similar homes, even though they were separated in early life. Whoever arranged the adoptions might, with humane intent, try to arrange such a state of affairs. Therefore, the separated twins might have lived in very similar environments even though they did not spend their time in contact with each other.
3 Not all of these ‘separated’ twins were separated for all of their lives before they were united via the MISTRA to take their intelligence tests. Some had spent some childhood together and some had adult contact. So, again, there was some opportunity to share environmental influences that might have made their intelligence levels similar.
Bouchard and his team tried to look at the latter two of these possibilities. They looked at the similarities of the families and homes into which individuals from each separated twin pair were placed. They made estimates of the adopted parents’ social class, of the facilities in the family homes, and of the more psychological aspects of the family environment. Some of these family-related things did associate moderately strongly across twin pairs. Some of these aspects related weakly to intelligence level. But the conclusion was that the effect of being placed in similar environments made only a tiny contribution to the intellectual similarity o
f the identical twins reared apart. The main influence seemed still to be the genes.
Next, the team on the MISTRA measured the amounts of time that members of the various twin pairs had spent together during their lives. This is illustrated in Figure 20, along the bottom. Note the 40-odd little circles in the diagram. Each one represents a twin pair in the MISTRA study. Their placement in the diagram describes their time spent together during their lives and their similarity on IQ tests. The further along the horizontal axis the circle is (i.e. the further to the right of the diagram), the longer was the time they shared together. The further up the vertical axis of the diagram they are, the more different are the members of the twin pairs in their IQs. Now, if there was an association between similarity of IQ and amount of time spent together during life we’d expect to see the circles, broadly speaking, arrange themselves along a line from the top left of the Figure (no time spent together and more difference in IQ) to the bottom right (more time spent together and less difference in IQ). Instead, what we see is a random-looking scatter of circles. There seems to be no association between the amount of time spent together and the similarity in the IQs. Do notice the range, though: one pair of genetically identical twins have an almost 30 IQ point difference; four other pairs cluster around the 18-point difference line. In some cases, then, there have been massive effects of the environment, but overall there is little evidence for this.
Let’s get back to the details in Figures 15 to 18 and offer some numbers to indicate how intelligence gets passed on from generation to generation. Recall that the researchers in the field of behaviour genetics split the influences on our psychological make-up into three main sources: genes, environments shared with members of our family, and our individual or unique environmental experiences. The Minnesota studies that looked at separated twins reckoned that genes contributed about 70% to the influence on human intelligence differences and environment did the rest. Now, let’s be clear what that means. It does not say that my or your intelligence score is 70% genetic. It means that, when we look at the differences in mental abilities across a range of adult people (twins, actually) in Western, developed countries, the differences between them in their mental capabilities are affected by genes to that degree. The Minnesota project is just one as yet incomplete study, not even a very large one, and it has still to report its full results. Looking across all the available studies in behaviour genetic research, one sees estimates of the genetic influence on intelligence differences that go from as low as 30% to as high as 80%. Rather conveniently, they average out to about 50%, meaning that about half the differences between people in their intelligence levels might be attributable to genetic differences.
None of the psychologists that I regularly talk to about intelligence differences cares much whether intelligence differences are attributable 40% or 70% to genetic differences. What we do know now is that intelligence differences have some appreciable genetic origins. What is much more interesting is to try to answer the following, more detailed questions that result from knowing that only some of humans’ intelligence differences have their origins in the genes.
Does the size of the influence of the genes change across the human lifespan?
Strangely, it appears that the influence of genes on intelligence grows stronger as humans grow older. The proportion of intelligence differences attributable to genes might be as low as 20–40% from infancy to childhood, yet 60% or even quite a bit higher by the time we get into our 70s and 80s. To me, this was originally counter-intuitive. One’s guess would be that as we accumulate education and knowledge and insults to our brains from the environment over a long life then the genes might have less and less effect. Not so. The first study to show the very high genetic influence on intelligence test scores in old age was so surprising that it made it to the top scientific journal Science and was featured with pictures as a splash on their cover (Figure 21).
What do we know about the influence of the environment on intelligence test scores?
We can see from the above numbers that the environment does indeed have quite a large influence on human intelligence differences. If genes, on average, account for about 50% of the differences between people in intelligence, then the environment also accounts for about 50% of the differences. Recall that the influences of the environment may be partitioned into shared and unique effects – those we have in common with our siblings and those that we experience alone. My guess and yours, probably, would be that the lion’s share of the environment’s influence would arise from the effect of the family. It’s not so; by far the larger part of the environment’s influence can be traced to the non-shared, unique environment. Families have little effect (when divorced from the contribution of the genes). This is arguably the most shocking result in the genetic/environmental study of intelligence. It was the
21. The cover page of the research journal Science for 6 June 1997, which reported the high level of genetic influence on intelligence differences in twins in late life.
topic dealt with by Judith Harris in her book The Nurture Assumption (i.e. the incorrect assumption that we all tend to make that upbringing is a big influence on intelligence level).
I think that this issue is crucial, and I also think that the way it is extracted from twin studies can be a bit abstruse, so I now want to introduce you to another remarkable study that suggests how limited an effect family upbringing has on intelligence.
Key dataset 8
Adoptees
Go to Figure 22, which is about adoption. I shall now describe a scenario and then ask some questions for you to think about before we look at the evidence.
• a mother (call her the ‘birth mother’) gives up her new-born baby for adoption
• the baby is adopted by another family (call them the ‘adoptive mother and father’), who have their own child too
• the children grow up without the adopted child ever seeing the birth mother
22. A diagram to show the influences on intelligence in adopted and natural children.
• the birth and adoptive parents take intelligence tests and the children take intelligence tests at different ages as they grow up
• remember, the adopted child spends all of his/her life with the adoptive parents and his/her step-siblings and none with the birth mother.
Now ask yourself:
1 Will the adopted child grow to have an intelligence level more like the adoptive mother and father, whom he/she has lived with from birth, or the birth mother, whom he/she has never met?
2 Since the step-siblings have spent their lives together in the same families, will they come to resemble each other in their intelligence test scores?
Anyone who believes there is an influence of family upbringing and environment on mental ability level is likely to predict that the adopted child will come to resemble the adopted mother and father and step-siblings in intelligence. Before unveiling the relevant findings, let me explain why such a prediction follows these beliefs.
Back to Figure 22 – and a reminder of the conventions that we used when we looked at twins. Examine the influences on the intelligence of the adopted child. His/her genes (G) come from the birth mother and father. (Here we are addressing only the birth mother’s contributions.) The adopted person’s ‘common’ family environment (C) comes from the adoptive mother (and father). The unshared/unique environment (U) is by definition not shared with anyone else and so is not really of interest here. Look at the ‘own child’ of the adoptive mother (and father). Both the genes (G) and the family environment (C) come from the same mother (and father). Therefore, we have two children who are genetically unrelated yet spend a lifetime in the same family. If there is an effect of family environment and upbringing on intelligence, then we expect to see step-siblings in the same family have some resemblance in their intelligence level. Also, if there is an effect of family environment on intelligence we expect to see some resemblance between the adoptive mother and her adopte
d child, perhaps greater than any resemblance between the birth mother and this adopted child, whom she never sees.
The test of these ideas came with the Texas Adoption Project conducted by John Loehlin and his fellow researchers. This project examined information from a church-based scheme in Texas in which mothers who were not married had offered up their children for adoption. Most of the birth mothers and adoptive mothers and fathers were white and middle-class. The children were adopted very soon after birth and were adopted on a permanent basis. The birth mothers and the adoptive mothers were given intelligence tests, including the Wechsler tests that we described earlier. The children were given intelligence tests at different stages: at an average of about 8 years and then an average of about 18 years.
Loehlin and his fellow researchers studied the data on intelligence test scores in several different ways to examine the effects of genes and the environment. The correlations between the intelligence test scores of the group of adoptive parents and their adopted children were around 0.1; these parent–children pairs share only environment and not genes. This suggests a very small effect of common family environment. Sometimes, more puzzlingly, the correlations between the group of adoptive mothers and their adoptive children were negative. That is, there was a slight tendency for more intelligent mothers to have less intelligent (adopted) children. This result would suggest that spending time together was making the adoptive children less (!) like their adoptive mothers in intelligence. The correlations between the intelligence scores of the group of adoptive parents and their own natural children were often around 0.2 or a bit more; these parent–child pairs share both genes and family environment. Thus, the parents with higher intelligence test scores tended to have natural children with higher intelligence test scores. This suggests an effect of genes adding to that of the common family environment in intelligence test scores. Most surprisingly, the highest correlations of all – often around 0.3 – were found between the group of adopted children and the birth mothers they had not lived with or even met after the first few days of life, which again points to the effects of the genes. Adopted children appear to grow more similar in intelligence level to a birth mother they have never met (with whom they share 50% of their genes) than to an adoptive mother with whom they spend their lives.