Nature and Nurture
about half an hour, not including Q&A
Good morning all. My name is Andrew Sabisky. Fresh from completing my masters in educational psychology, I’m here today to talk to you one of my major research interests: genes, and how they combine with parenting, teaching, and peer groups to produce academic achievement and behaviour. All matter to some degree. For centuries, scientists, philosophers, and the general public have argued about the importance of each of these factors. I will give you some answers based on over a century’s worth of research, and a brief discussion of the importance of these findings for education.
Scientific enquiry into the respective contributions of nature and nurture to behaviour has a long history, but for our present purposes the story should begin with Charles Darwin’s cousin. Sir Francis Galton, the Victorian polymath. Galton’s interest in eugenics has probably blackened his name in history, for he deserves to be remembered as one of the greatest geniuses of the era. He made vital contributions to meteorology, biology, anthropology, psychometrics, and perhaps most of all to statistics, single-handedly inventing the concepts and mathematics of correlation and regression. One of Galton’s major works was a book called Hereditary Genius (Galton 1869), in which he claimed that eminence - or what we now call “socioeconomic status” - tended to run in families. This led Galton to claim that the primary contribution to eminence was genetic.
This may seem somewhat premature. Families, after all, share an environment as well genes, and eminent families are well placed to educate and stimulate their children to a far higher degree than less eminent families. It’s only natural, therefore, to expect eminence to run in families - but this doesn’t have to be for genetic reasons at all.
slide with some basic Galton facts
In fact, though it is rarely mentioned in the literature, Galton was well aware of this problem, and attempted to resolve it by studying the biographies of the adopted sons of eminent men, who share an environment with their fathers, but none of their genes. He then compared the later eminence of these adopted children against the achievements of blood children, and found that, on average, adopted children did not do as well. Galton had pioneered the adoption study, which remains today one of our best ways to disentangle the respective contributions of nature and nurture to development. It has been supplemented by the twin study over the course of the 20th century. The findings from these studies comprise the field of behaviour genetics, a vast literature that I am going to attempt to summarize for you today, and discuss its implications for education.
The adoption study
The principles of the adoption design are also fairly straightforward. The traits of children adopted at very young ages are compared to those of their adoptive siblings, adoptive parents, and also their biological parents. If the adopted children most closely resemble their biological parents over time, then we can conclude that the post-natal family environment has had little influence on their development, and that genetic influences remain the most powerful.
A fascinating subset of the adoption study is the virtual twin design. Parents will occasionally adopt a child at birth who is almost exactly the same age as a biological child that they already have, thus creating virtual twins, who share no genes but have near-identical post-natal environments. This situation is quite rare, but when it does arise it allows scientists to design very powerful studies free of many of the usual confounds.
The twin study
The logic of the twin study is beautiful in its simplicity. Identical twins raised together share a family environment and 100% of their genes: they are almost perfect clones, in genetic terms. Dizygotic (non-identical) twins, however, are no more genetically similar than full siblings: they share an environment to the same degree as identical twins, but only 50% of their genes. Therefore, if monozygotic twins are more similar for a trait than dizygotic twins, we can conclude that this is due to their greater genetic similarity, and are able to quantify the precise degree of genetic and environmental contributions to that trait across our sample of the population. Even better, we are able to calculate what percentage of the environmental variance is due to the environment that the twins share (the family environment), and what percentage is due to the environment that is unique to each twin.
I’m going to give you a stripped-down, simplified version of the mathematics here , since I think it’s important to show that my claims do make sense and aren’t just glorified witchcraft. To calculate the percentage of genetic variance for a trait - let’s imagine IQ - we take the difference between the identical twin correlations and the non-identical twin correlations, and multiply that difference by 2 (because identical twins are twice as genetically similar as non-identical twins). This produces the heritability coefficient, h2. To calculate the total environmental contribution to a trait, we subtract h2 from 1. But how do we go about partitioning the environmental variance into “family” (shared) and unique (non-shared) components?
The logic is as follows. Identical twins, as already stated, share a family environment and all their genes. Therefore, all their similarities are the result of these two factors, and any differences between them must be the result of the environment unique to each twin. To calculate the importance of this unique environment to the variance of the trait in question, therefore, we simply subtract the identical twin correlation for the trait from 1. To calculate the importance of the family environment, we subtract the heritability coefficient from the identical twin correlation, based on the principle that since identical twin correlation is due to their shared genes and their shared environment, whatever quantity of the correlation that is not caused by genes must be down to the shared environment.
(while talking this through, have a slide here displaying the basic equations for h2, total E, and non-shared + shared E, as well as a table with some sample figures and the estimates for G+E)
There are a few other variants on this basic twin study design. An experimental - rather than statistical - method for calculating the importance of family environments is to examine monozygotic twins who have been separated at birth and raised in different families, and compare their similarities to those of monozygotic twins raised together. The Minnesota Study of Twins Reared Apart is particularly famous in this field, and gave rise to some amazing stories. The “Jim twins”, MZ twins separated at birth and raised in different Ohio working-class families, both married women named Linda, then divorced them and remarried women named Betty. Both had childhood pets named Toy, and called their sons James Allen and James Alan. They did this all without any knowledge of the life of their genetic counterpart. Amongst the many other identical twin pairs separated at birth, one were very prone to the giggles, despite both having dour adoptive parents; one pair both handled dogs; one pair were both captains of volunteer fire departments; one pair of women always wore seven rings; and one pair of men left love notes around the house for their wives (Lykken et al. 1992)
While these personal idiosyncrasies of identical twin pairs are unlikely to run in families, they do speak to extraordinarily strong genetic influences on behaviour.
Conclusions of twin and adoption studies
Twin and adoption studies have allowed behaviour geneticists to assign estimates of the contributions of genes and environments to a very wide range of traits and behavior. Some are very obvious: general intelligence (IQ) and the major personality traits have been exhaustively studied. Students of psychopathology have also learned much from behaviour genetics; all kinds of conditions have been subject to twin and adoption designs, from autism to alcoholism. Some rather more surprising behaviours have also been studied for evidence of genetic contributions. Table 1 presents a wide range of results, with sources listed. I have presented a range of scores where possible, or where no one study has a methodology clearly superior to the others in the field.
Trait/Outcome | Heritability | Source |
IQ (general intelligence) | 50-80% | The literature is vast: Plomin et al 1994 is a good start |
GCSE scores (core subjects) | 58% | Shakeshaft et al. (2013) |
Conscientiousness | 75% | Riemann et al (1997) |
Bullying (at age 10) | 61% | Ball et al (2008) |
Altruism | 56% | Rushton et al (1986) |
Alcoholism | 20%-70% (wide range of values appear in the literature, may differ significantly between sexes) | Prescott et al 2005, Walters 2002 |
Autistic traits (at age 8) | 64-92 % (depending on diagnostic cut-off) | Ronald et al 2006 |
ADHD | 70-80% (consensus estimate) | Martin et al (2002) |
Obesity | 40-75% (consensus estimate) | O’Rahilly & Farooqi (2008) |
Time spent watching TV | 30-45% | Plomin et al (1990) |
Smoking initiation | 44% | Vink et al (2005) |
Nicotine dependence | 75% | Vink et al (2005) |
Political conservatism | about 40-60% | Martin et al (1986), Bouchard et al (2003) |
Conduct disorder (age 11-18) | 53% | Gelhorn et al (2005) |
I’ll give you a few moments to digest this table. What you’ll see is that the heritabilities are typically around 50%, sometimes much higher. This means that across the population, some 50% of the differences between individuals for the traits studied are due to differences in genotypes[KA9] .
Another important and highly counter-intuitive finding of this field is that the shared environment, while it often accounts for a decent percentage of the variance in traits when measured in childhood, has a negligible - often zero - effect when these traits are measured in adulthood. Whilst families have some importance when we are young, their effects do not last into adulthood. This effect has been found over and over again across almost every single trait we have studied using behaviour genetic methods. In the long run, parents, children, and siblings resemble each other because of their shared genes, not because of their shared experiences.
Table 2 presents the percentage of variance accounted for by the shared environment. Unless noted otherwise, these figures are given for traits measured in adulthood. As before, it is worth remembering that when these traits are measured in childhood, the family environment is significantly more important.
Trait/Outcome | Shared environment % | Source |
IQ (general intelligence) | generally about 0 in adulthood but about 20-30% in early childhood | again, a vast literature: Plomin et al 1994 is fairly representative |
GCSE scores | 36% | Shakeshaft et al (2013) |
Conscientiousness | 0 | Riemann et al (1997 |
Bullying (at age 10) | 0 | Ball et al (2008) |
Altruism | 0 | Rushton et al (1986) |
Alcoholism | generally low in males (around 10%) but may be much higher in females (50-60%) | Prescott et al 2005, Walters 2002 |
Autistic traits (at age 8) | 0 | Ronald et al 2006 |
ADHD | 0 | Burt (2009, 2010) |
Obesity (age 8-11) | 10% | Wardle et al (2008) |
Time spent watching TV (age 3-5) | 18-34% | Plomin et al (1990) |
Smoking initiation | 51% | Vink et al (2005) |
Nicotine dependence | 0 | Vink et al (2005) |
Political conservatism | <10% (but much higher in adolescence) | Hatemi et al (2009) |
Conduct disorder (age 11-18) | 10- 25% (my own rough estimate from the literature but may be higher in some disadvantaged samples) | Gelhorn et al (2005) Burt (2009) |
It is important to note that family environments are more important when we are younger, but their effects “wash out” by the time we reach early adulthood. This is probably due to an effect that Sandra Scarr called “niche building”. Imagine a violent, aggressive child born to a conscientious, peaceful family. In the child’s early life, his family will be able to control his behaviour: send him to a good school, keep him away from bad company, cushion the worst effects of his conduct, and so on. Over time, however, the child starts to become less controllable. He will most likely, by hook or by crook, choose his own peer group. By adulthood he is free from parental influence and able to create his own environment in accordance with the demands of his genotype. This process also explains why intelligence becomes more heritable over developmental time.
After genes and the environment shared between twins, the final component of the variance in these outcomes and traits is the non-shared environment: the environment unique to each twin. This includes the influence of those the twins do not experience together: teachers and peer groups will often be included in this category, as well as random developmental noise and measurement error. Whatever the non-shared environment actually is - the picture is somewhat muddy - it has a powerful effect, often account for up to 40% or more of the variation in adult traits and outcomes.
These figures above almost all come from twin studies, but it is a strength of the field that adoption studies paint a similar picture. The correlation between the IQs of adopted children and their adoptive parents is insignificant, and, for the reasons discussed earlier, becomes even weaker over time.
This is called convergent validity - when two very different lines of evidence, each subject to criticism but of very different types, both provide similar data, we can be more confident in our conclusions .
Implications for education
At this point, some of you may be feeling rather depressed . If a large percentage of the differences between individuals occurs because of their genes, then aren’t we all wasting our time with this whole teaching business? Well, no, and here I’ll explain why. Far from establishing that teaching is a waste of time, behaviour genetics is a strong confirmation of its utility. Perhaps it also points to some directions for the future, and I will briefly explore those also.
It is important to remember that heritability is a population statistic. It measures the respective contributions of genes and environment to the differences between individuals, not to the development of individual phenotypes. No one can say what percentage of your personal intelligence is due to nature or nurture - in fact, the question is meaningless. What we can estimate is what percentage of the difference in intelligence between yourself and the person sitting next to you is due to nature or nurture - a subtly important distinction.
This realization that heritability is a population statistic should make us realize that when environments are equalized, heritability is higher. This is a key point: let me illustrate with an example.
Imagine we lived in a world where half the population, through no merit of their own, lived in the lap of idle luxury, and the other half, through no fault of their own, in the most grinding, awful poverty you can imagine: children grew up with no formal education, primitive technology, and malnutrition and iodine deficiency were rampant, along with crime, infectious diseases, and so on. In such world, if you measure the whole population, differences in outcomes and traits would be largely due to the environment, because your environment, in this scenario,
a) determines your outcomes
b) will significantly shape your core traits (via such mechanisms as malnutrition and disease impairing intelligence).
Genetic factors, in our anti-meritocracy, will not significantly influence individual differences. Now what happens if we equalize environments? Let’s imagine our anti-meritocracy becomes a Communist Utopia where all experience very similar upbringings. In the absence of significant environmental variation between individuals, most differences between individuals must be down to genes.
Our 21st century Western world is clearly neither a feudal nightmare nor Utopia, yet significant heritability for IQ and academic outcomes is the norm, as seen above. This can be taken to mean that teachers are doing a fairly good job in some respects. Thanks to universal education, environments are relatively equal, and consequently heritability is pretty high. Everyone is getting a reasonably fair shot at success, and consequently genes significantly influence outcomes. To quote the great behavioural geneticist Robert Plomin, “heritability can be taken as an index of societal meritocracy ”. That does not mean, of course, that a more meritocratic world could not be achieved.
Most importantly, heritability estimates capture a snapshot of the world as it is: they give limited information about the world as it might be. Classroom misbehaviour is clearly under significant genetic influence. The naughty kids are that way, to be blunt, in significant part because their genes incline them so (Viding et al. 2008) Does this mean we can do nothing about their behaviour? Of course not! In another world, although the rank ordering of persons is unlikely to change, overall behaviour could easily significantly improve. With better behaviour management, although they will still most likely behave a bit worse than their classmates, the genetically naughty kids could progress from “incredibly disruptive” to “a minor, somewhat charming pain in the butt”. Nothing in heritability estimates precludes this kind of change. Remember also that outcomes, in principle, are more modifiable than the fundamental traits that underlie them.
This does, however, lead me on to my final point, which is a word of warning. A great deal of educational time and resource is spent discussing various “achievement gaps” and how to
remedy them - particularly in America, but also in the UK. This may very well be time and money wasted. Politicians and education reformers alike labor under the misapprehension that it is more favourable environments that lead to the successful doing well, and that if we could replicate those environments, achievement gaps would disappear. As we have seen, individual differences in intelligence and academic outcomes are largely a result of genes, not environments. The pursuit of educational equality of outcome may turn out to be a hunt for the Questing Beast.
But just because today’s education is meritocratic does not mean that it is efficient. High heritability says nothing about whether or not students are learning the maximum that they are capable of. Frankly, no one believes that they are. It is one of the great failings of educational psychology that no one has really put much time and effort into figuring out the maximum that students of different IQs and personalities are capable of learning, given the best possible teaching. I strongly suggest that academia must embrace this research program over the course of the next decade. Only then can we truly produce personalized curricula that will enable each student to achieve their genetic potential. We know that we cannot make men more than what they are. The fault lies not in our stars, but in ourselves. Yet where the limits of that potential are, we cannot yet say. It is for us all to find out together.
Thank you.
Thanks
I am deeply grateful to Kathryn Asbury, whose comments greatly improved an earlier draft of this presentation.
Further reading
Asbury K. & Plomin R. G is for Genes: The Impact of Genetics on Education and Achievement. Wiley 2013
Sesardic N. Making Sense of Heritability. Cambridge University Press 2005
Mitchell, K.J. (2007) The genetics of brain wiring; from molecule to mind. PLoS Biology Apr 17;5(4):e113. (Open Access).
http://www.plosbiology.org/article/info:doi/10.1371/journal.pbio.0050113
References
Ball H.A., Arsenault L., Taylor A., Maughan B., Caspi A., & Moffitt T.E. (2008). Genetic and environmental influences on victims, bullies, and bully-victims in childhood. Journal of Child Psychology and Psychiatry and allied disciplines 49 (1) 104-112.
Galton, F. Hereditary Genius. Macmillan 1869
Hatemi P.K., Funk C.L., Medland S.E,. Maes H.M., Silberg J.L., Martin N.G., & Eaves L.J. (2009). Genetic and environmental transmission of political attitudes over a lifetime. Journal of Politics 71 (3) 1141-1156
Lykken D.T., McGue M.,Tellegen A ., & Bouchard T.J.Jr. (1992). Emergenesis: genetic traits that may not run in families. American Psychologist 47 1565-77.
Martin N.G., Eaves L.J., Heath A.C., Jardine R., Feingoldt L.M., & Eysenck H.J. (1986). Transmission of social attitudes. Proceedings of the National Academy of Sciences 83 4364-4368
O’Rahilly S. & Farooqi I.S. (2008). Human Obesity: A Heritable Neurobehavioral Disorder That Is Highly Sensitive to Environmental Conditions. Diabetes 57 (11) 2905-2910
Plomin R., Pedersen N.L., Lichtenstein P., & McClearn G.E (1994). Variety and stability in cognitive abilities are largely genetic in origin. Behavior Genetics 24 (3) 207-215
Plomin R., Corley R., Defries J.C., & Fulker D.W. (1990). Individual differences in television viewing in early childhood: nature as well as nurture. Psychological Science 1 (6) 371-377.
Prescott C.A., Caldwell B.C., Carey G., Vogler G.P., Trumbetta S.L., & Gottesman I.I. (2005). The Washington University Twin Study of Alcoholism. American Journal of Medical Genetics Part B (Neuropsychiatric Genetics) 134B:48–55
Ronald A., Happe F., Bolton P., Butcher L.M., Price T.S., Wheelwright S., Baron-Cohen S., & Plomin R. (2006). Genetic Heterogeneity Between the Three Components of the Autism Spectrum: A Twin Study. Journal of the American Academy of Child & Adolescent Psychiatry 45 (6) 691-699`
Rushton J.P., Fulker D.W., Neale M.C., Nias D.B.K., & Eysenck H.J. (1986). Altruism and Aggression: the heritability of individual differences. Journal of Personality and Social Psychology 50 1192-1198
Shakeshaft N.G., Trzaskowski M., McMillan A., Rimfeld K., Krapohl E., et al. (2013). Strong Genetic Influence on a UK Nationwide Test of Educational Achievement at the End of Compulsory Education at Age 16. PLoS ONE 8(12): e80341. doi:10.1371/journal.pone.0080341
Viding E., Jones A.P., Frick P.J., Moffitt T.E., & Plomin R. (2008). Heritability of antisocial behaviour at 9: do callous-unemotional traits matter? Developmental Science 11 (1) 17-22
Vink J.M., Willemsen G., & Boomsma D.I. (2005). Heritability of smoking initiation and nicotine dependence. Behavior Genetics 35 (4) 397-406
Walters G.D. (2002). The heritability of alcohol abuse and dependence: a meta-analysis of behaviour genetic research. The American journal of drug and alcohol abuse 28 (3) 557-584