Genes, intelligence and education: a heady brew of issues. Add class, race or gender, as has happened so many times over the past 100 years, and you have a simmering mixture ready to boil over at any moment.
The last time this occurred in Britain was in the 1970s, when psychologist Hans Eysenck, whose books were standard reading for young teachers, published Race, Intelligence and Education. But in recent months the debate over IQ, school performance and genetics has been revived in new form, mixing the classical human genetics of the past century with the modern molecular genomics that developed after the sequencing in 2003 of the 3 billion DNA bases that comprise the human genome.
The first public sign of the re-emergence of this debate came last October, when a 237-page letter to England's education secretary Michael Gove from his departing adviser, Dominic Cummings, was leaked to the press. In it, Cummings excoriates the British educational system for failing both the brightest and the least able students.
For Cummings, teachers are part of the problem, but much of it is also down to the failure to tune education to the genetic potential of individual students. Intelligence (IQ) and educational achievement are, he asserts, some 70 per cent heritable. For Britain to catch up with its Asian rivals, IQ screening should be used to identify the top-scoring 2 per cent of students, who should be fast-tracked into the sciences, while research to identify high IQ genes should be fostered.
A month later, London mayor Boris Johnson echoed Cummings in his Margaret Thatcher Lecture to the Centre for Policy Studies. We need, he said, to nurture the 2 per cent with IQs above 130, who are the successful innovators - "the cornflakes who come to the top of the packet if you shake it", as he put it. The implication is that the "16 per cent of our species" (Johnson's words) with IQs below 85 are a drain on society.
So where do these figures come from, how meaningful are they and - setting aside Johnson's characteristically provocative remarks, which were swiftly disowned by the Conservative Party leadership - what are their implications for education policy?
Cummings is generous in his references, particularly to Robert Plomin, professor of behavioural genetics at the Institute of Psychiatry, King's College London. Within the UK, Plomin is the current standard-bearer in the long quest by geneticists and psychologists to uncover the relative roles of genes and environment in determining - or at least shaping - intelligence.
Cracking the genetic code
Genetics is about both inheritance and difference. Some traits and diseases are inherited in a relatively straightforward and predictable manner, more or less independently of the environment in which the child grows up. Huntington's disease, a devastating neurological degeneration that attacks in mid-life, is one example of a disorder caused by a single gene. But most inherited physical and behavioural traits have more complex causes, involving the interaction of many genes with one another and with varying environments, beginning in the womb and continuing through infancy into adulthood.
This interaction means that the question often phrased as "how much do genes and environment each contribute to any individual's intelligence (or IQ)?" is meaningless - the only possible answer is that each contributes 100 per cent. Nature versus nurture is a false dichotomy: they cannot be disentangled in any person's life history.
So, for most of the past century, behavioural genetics has asked another question, which might in principle be answerable: how much of the difference between individuals is down to genetics and how much to environment?
The assumption that the geneticists started with was that a proportion is contributed by genes (G), a proportion by environment (E), and a small proportion by the interaction between genes and environment (GXE).
In classical human genetics, the size of these proportions has been estimated through the study of twins. The theory is that identical (monozygotic, MZ) twins have 100 per cent of their genes in common; non-identical (dizygotic, DZ) twins, like any other siblings in a family, share only 50 per cent of their genes. Any differences between MZ twins must therefore be environmental, whereas differences between DZ twins result from both genes and environment.
So, comparing the difference in a trait - for instance, IQ - between pairs of MZ twins and pairs of DZ twins enables the genetic contribution to be teased out. This is expressed as a percentage figure and is called heritability. At 100 per cent heritability, the difference is all genetic; at 0 per cent it is all environmental.
To this end, huge twin registers have been collected over many decades in the US, Scandinavia and the UK (this last currently directed by Plomin). By studying twins in this way, psychologists have calculated a broad heritability figure of 50-70 per cent for IQ. (Eysenck put it higher, at 80 per cent.)
But these figures are not what they seem. First, the heritability equation depends on the environment. If the environment is identical for all, then all the differences between individuals should be genetic and the heritability would be 100 per cent. If the environment is very varied, the genetic contribution would be much reduced. This is reflected in studies such as one by Turkheimer et al in 2003, which found that whereas the heritability of IQ is high in children from rich families, it is less than 10 per cent in those from poor and deprived backgrounds.
Second, the heritability equation assumes that there is a straightforward entity that we call "the environment". Of course, this is not realistic. "The environment" is a term that encompasses many disparate factors, from childhood diet to home and school conditions and on to a rapidly changing social, technological and cultural context.
Attempts have been made to distinguish between shared environments (twins live together, probably go to the same school and so on) and unshared environments (experience specific to the individual). But these, too, are simplifying assumptions. The same classroom can be experienced very differently: one child may be more favoured by the teacher, another may be sitting next to a bully. And, pulling in the other direction, MZ twins share more of their experiences than DZ twins: they often develop special bonds and private languages; their parents may dress them identically; strangers and sometimes even close acquaintances may confuse them. In the real world, the comfortable assumption by behavioural geneticists that they can parse out the environment seems somewhat naive.
One way round this has been to seek out the "natural experiment" of the rare cases of MZ twins adopted separately. However, the appetite for such studies has waned since a scandal involving the work of Cyril Burt, who, it is alleged, made fraudulent claims to have identified and studied such twins in the 1950s. In addition, the Abortion Act 1967, which provided a legal defence for abortion, led to a reduction in the number of children put up for adoption in the UK.
The third problem is that the heritability equations derive from the early days of genetics. They were originally intended not for human studies but for trials to improve crop yield in agriculture, where environments could be closely controlled. The early geneticists made a number of assumptions, primarily that genetic and environmental components could simply be added together to make nearly 100, with just a small GXE term for the interaction. But if genes and environment do interact to any substantial degree, the calculations don't work out. Even before the revolution in genetics in the aftermath of the sequencing of the human genome, people began to realise that things weren't that simple.
One environment (for instance, a diet rich in certain foods) might result in particular genes being expressed, while another might not. Or a person with a particular set of genes (an example often cited is those for "thrill-seeking") might be predisposed to seek out environments such as bungee jumping or car racing. G and E interact in varying ways at varying times during development.
A 'black hole'
Sequencing the human genome produced a result that shocked many geneticists. The human body contains some 40 trillion cells, and some 100,000 different proteins, but only about 22,000 genes: roughly the same number as a fruit fly. There is no way such a small number of genes could "code" for all those proteins, let alone such complex behaviours as intelligence. The explanation must lie in the multiple ways in which genes (segments of DNA) are interpreted and used by the cells they reside in as a child develops. The old assumptions went out of the window and a whole new science of epigenetics - the study of factors outside genes that can affect their expression - has grown up to examine these interactions.
Meanwhile, new technology has begun to replace the old-fashioned twin studies. Sequencing the first human genome took $3 billion (#163;1.8 billion), armies of researchers and more than a decade. It can now be done for around $1,000 in a matter of weeks. So, researchers have been forming huge DNA data banks and scanning them for genes that might be associated with diseases such as cancer or coronary heart disease - or complex behavioural measures such as IQ.
Initially, it was thought that it would be possible to pinpoint perhaps half a dozen or so "major" genes - the ones that did most of the work - for each disease. But the results of these "genome-wide association studies" (GWAS) came as yet another surprise. Not a few but many hundreds of genes seemed to be involved in disorders known to have a high heritability. At least 200 were found to be connected with schizophrenia, for instance. Furthermore, even when the effects of all these were added together, they accounted for only a small percentage of the heritability. Something had gone seriously wrong with the whole enterprise.
Senior geneticists began to talk of a "black hole" at the heart of the heritability estimates. Speaking of the relevance of genomics to medicine and human health, the editor of the journal Genetics in Medicine, James Evans, along with health and behavioural psychologist Theresa Marteau, said that it was "time to deflate the genomic bubble". Conclusions drawn from heritability estimates are part of pre-modern genetics and are long out of date, they believe. The clues lie not in genetics but in epigenetics.
So where does this leave the efforts to discover intelligence genes, and what relevance might such a discovery have to educational achievement? Well aware of the pitfalls of heritability, Plomin, among others, is using GWAS to try to identify high-IQ genes in collaboration with one of the world's major sequencing laboratories, the Beijing Genomics Institute. But the best guess is that, just as with the hunt for schizophrenia genes and the rest, what will turn up is many dozens or even hundreds of genes, each of which has a tiny influence on IQ scores.
So far, I've used the terms intelligence and IQ almost interchangeably. IQ theorists claim that the test taps into a general cognitive factor that underlies all intelligent behaviour - they call it "g". Others are sceptical. Educationalists argue that IQ measures ignore creativity, and artistic and emotional intelligence. Neuroscientists like me see many distinct brain processes (motivation, perception, memory and more) underlying how students respond to IQ tests.
IQ scores do correlate reasonably with school performance, but this is what they were designed to do. And both IQ and school performance correlate with the socio-economic status of a child's parents. How well a child performs at school is partly related to his or her IQ, but also to other important factors such as motivation and self-confidence. This is why high IQs, contrary to Boris Johnson's claim, do not automatically imply success in life; not all members of Mensa, a British society for people with high IQs, are successful high-flyers.
So, suppose we forget IQ and use the GWAS method to try to identify genes associated with school performance. A recent paper in the prestigious journal Science, signed by some 200 authors, describes a GWAS study of 126,559 individuals aimed at identifying genes associated with educational achievement. The result? Adding together all the genetic variants that they found accounted for only 2 per cent of the differences in educational achievement: nothing to write home about, and nothing to help influence educational policy.
Genetics and education
Cummings' hope, following from his reading of Plomin, is that IQ tests and genetic screening will make it possible to identify high performers, provide the specialist education their genetic profile requires and hence enable Britain to compete with the rising powers of China and India in terms of innovation.
Plomin's agenda, as expressed in his and Kathryn Asbury's recent book G is for Genes: the impact of genetics on education and achievement, is more liberal. His hope is that identifying genes that predispose a child towards, say, science or geography - or, at the other end of the spectrum, those that limit a child's cognitive ability - would make it possible to move from a "one size fits all" educational strategy to a personalised, gene-tailored curriculum for each individual. The results of GWAS suggest, however, that this is unlikely ever to be feasible. Faced with such findings, it seems unkind to point out that you don't need to do a gene scan to see whether a child is turned on by science. You could ask them, or, if you think they won't know their own mind, try an aptitude test.
None of this is to imply that genes are irrelevant. The unique set of genes that each of us (bar MZ twins) carry, and their epigenetic expression during development, are of endless fascination for biologists. And genetics could help to clarify the biological reasons why some children have specific learning difficulties, such as dyslexia.
But how far can this inform education policy? Perhaps the one significant finding from all this research into genetics comes from the study by Turkheimer et al comparing heritability in rich and poor families. This showed high heritability in richer families. That is, the environment was such that children were performing closer to their genetic, or epigenetic, potential. By contrast, there was low heritability of IQ in poorer families. This means that the more their environment - in all senses of that word - can be enriched, the closer children will be able to perform to their potential.
Policymakers and educators don't need genetics to help them make a better environment for all our children. What is lacking is the political will.
Neuroscientist Steven Rose is emeritus professor at the Open University in the UK. His most recent book, with sociologist Hilary Rose, is Genes, Cells and Brains: the Promethean promises of the new biology
Turkheimer, E, Haley, A, Waldron, M, D'Onofrio, B and Gottesman, I (2003) "SES modifies heritability of IQ in young children", Psychological Science, 14: 623-28
Evans, JP, Meslin, EM, Marteau, TM and Caulfield, T (2011) "Deflating the genomic bubble", Science, 331: 861-62
Rietveld, CA et al (2013) "GWAS of 126,559 individuals identifies genetic variants associated with educational attainment", Science, 340: 1467-71.