Five lessons I’ve learned about genetics

I admit that I began reading to find evidence that supported my view. I was sure science would back me up. And then I, complete with a list of Harvard-format references, would win the next debate. The problem was that science didn’t agree with me. I could find pockets of supporting evidence, but the overwhelming consensus was that I was wrong. It made me phenomenally uncomfortable to see a consensus emerging that directly contradicted my beliefs about how the world should be. But ultimately, science is not about how the world should be. Science is about how the world is. And if I want to make it different, I need an accurate picture of how it is now.

This is a blog on reading about genetics – what I’ve learned, and what I want to do about it. The lessons are the things that struck me as a teacher most, and are largely based on reading about intelligence or other cognitive characteristics. I want to note up front that I am not an expert, and that writing this is part of my attempt to learn more. If and when I’ve made a mistake please let me know and I will rectify it as quickly as possible.

Lesson #1: Genetics explains much more of the variation between people than I was willing to accept
The most influential type of study for this blog is the twin study, which looks to explain the variation between people and attribute it to one of three categories of cause: the shared environment (factors that would be common to a pair of twins living in the same household and attending the same school); the non-shared environment (all other environmental factors); and genetics. Studies of cognitive ability tend to find that variation is 10-20% shared environment, 30-40% non-shared environment, and 50-60% genetics.

This was hard for me to accept. I wanted to believe that most variation is caused by things within our control. Instead the shared environment, of which school is only one part, explains under a fifth of the variation between people.

Before moving on to Lesson #2 it is important to stress that we are talking about variation here – not absolute levels. We can say that 50% of variation in intelligence is genetic in origin, we can say nothing about how much of your intelligence is caused by your genes.

Lesson #2: Environments change how genes are expressed
Your genetic code is fixed, but what you do with it isn’t. Geneticist Nessa Carey likens your genetic code to the script of a play, which is then interpreted extensively by the actors and director before becoming the performance that eventually appears on stage. Epigenetics is the study of these interpretations, which are as crucial to our biological functioning as the genetic code itself.

One thing we learn from epigenetics is that our environment shapes how we interpret our genetic code, and that these interpretations stick. Once we have scrawled annotations over our genetic script they will stay unless we actively rub them out – and when our children inherit our scripts they will find many of our annotations still in place.

Extreme or consistent environmental stimuli can create epigenetic modifications that change how your genes are expressed. For example, a stressful environment can lead to genes that control the production of cortisol (a stress hormone) becoming over-expressed, meaning that you become much more easily stressed in the future. Such a change will then persist, shutting off chunks of working memory and reducing executive function in years to come.

As many epigenetic modifications are heritable it is difficult for twin studies to separate their effect from the effect of genes themselves, and so it is possible that some of the causal impact of genetics is actually environmental in origin. As our ability to do more complex analysis with the genome itself increases we will no doubt find out whether this possibility means anything in practice.

Lesson #3: Environments correlate with genes
A tall child is more likely than a short child to be asked to try out for the basketball team. Where we may have a genetic propensity towards a certain area we tend to seek out (or be pushed towards) an environment that increases that propensity. This means that a small effect that begins life as just an inkling of interest or talent can easily grow into a specialised environment that exaggerates initially small effects. It is possible that correlations like this are responsible for a significant proportion of our genes’ impact. If we adapt environments, whether consciously or not, we will be magnifying genetic differences.

Lesson #4: Genetics does not determine outcomes
Twin studies observe the differences we see today and explain their origins. They do not have any say in how big these differences are or will be in the future. So the fact that 50% of variation in a characteristic is due to genetic causes today does not mean that it must be tomorrow. Nor does it mean that we must accept present levels of difference as necessary. The numbers we find in these studies are not natural constants.

Behavioural geneticist Robert Plomin says that studies tell you what is, not what could be. He likens our genetic understanding of intelligence to our understanding of weight. Whilst it is obviously the case that people can be genetically predisposed to put on more or less weight than each other, it is also true that with the right interventions almost anybody can achieve a healthy weight. The same is true for intelligence. There may be genetic predispositions, but we can make sure that everybody achieves an acceptable level by providing the right environment.

Lesson #5: Genetics assumes determinism
If a study assumes that all difference is caused by either genetics, the shared environment or the non-shared environment, then it is also making one other underlying assumption: that everything about a person has an external cause. It assumes that your intelligence or success is a function solely of your genes and your environment. But what if everything about us is not 100% deterministic?

I was hesitant to write this lesson down, for fear of seeming to criticise the entire body of work genetics has given us. Science has to operate by studying the relationship between cause and effect. I cannot challenge it for failing to account for independent free will. But I am nonetheless uncomfortable not accounting for it. I do not know nearly enough in this area to do anything more than speculate. But I do wonder whether the large influence of the non-shared environment, that bucket for everything we can’t put our finger on, may be substantially down to things like attitude and motivation that may not be fully caused by an external mechanism.

So what do I take from all this?
Firstly, that genetics plays an unquestionably big role in explaining who we are, and how our brains work. Even if some of the effect attributed to genes is in fact environmental in origin (due to epigenetics or gene-environment correlations) there is no doubt that genes have a huge influence.

But secondly, even though genes make us all different, they don’t determine our cognitive future. Long-term memory still has unlimited capacity; brain plasticity is still immense; and good teaching can still take advantage of this. Genes determine difference, but they’re no excuse for educational inequality.

4 thoughts on “Five lessons I’ve learned about genetics

  1. Ed Cadwallader

    Great post David, thank you. I’m someone, I think like you, who wanted to believe in a Lockean blank slate and has had to adapt their views to the evidence that this is not an accurate understanding of human development.

    I agree too that the heritability of intelligence does not justify educational inequality, but I think the equality we seek is unobtainable given the dominance of academic subjects in the curriculum. In a system where children compete for a place in an academic hierarchy those with educated parents have genetic and environmental advantages that are impossible to counteract, given that improving pedagogy benefits everyone, not just the disadvantaged.

    You’re right that small effects can be magnified by experience into larger differences. Children who arrive ‘school ready’, seeing they are superior in the domain school prizes above all others, respond with pleasure, work hard and are rewarded with more praise. Those who don’t enter an opposing cycle of withdrawal, punishment and defiance. Only a curriculum that expands the definition of success beyond the academic can deliver educational equality.

  2. andrewsabisky

    great post, some thoughts, comments, clarifications in response

    there has always been a great deal of debate over the assumptions of the twin model, though oddly many of the assumptions that we know get violated in practice actually lead to underestimates of heritability. A lot of heritability estimates don’t take assortative mating into account (since you have to measure the parents as well as the twins to do so), and this matters; quite a few published estimates of genetic influence are probably too low for this reason. Incidentally, measurement error in the phenotype leads to underestimates of heritability also and an excessively large non-shared environment term (since more random error increases the differences between twins). Part of the reason why some heritability estimates are higher than others is just because we can measure some phenotypes (height, IQ) more reliably than we can others (personality or sleep parameters). Of course, on the flip side, violations of the equal environments assumption does lead to overestimated heritability (the assumption states at MZ twins are not subject to more similar external environments than DZ twins; or, if they are, this extra environmental similarity does not affect their phenotypic similarity). Anyway, the new era of GCTA (genome-wide complex trait analysis) and other methods of estimating heritability direct from the genotypes of unrelated persons means that to some extent the whole discussion is moot. But I do think the assortative mating point is worth remembering. Some phenotypes turn out to be quite significantly more heritable than originally thought once assortative mating is taken into account (such as reading ability).

    Section 2 appears to confuse several concepts. Epigenetics is a confusing term that seems to have about three meanings. The first is simply a fancy word for development, as in Waddington’s “epigenetic landscape”. The second is differential gene expression depending on environmental stimulus, which is a perfectly respectable field of study (just a form of plasticity, really). I don’t really know how important this sort of thing is, but in theory it could certainly contribute towards stable MZ twin differences. The third is supposed inheritance of such differential gene expression (the idea that Holocaust survivors suffered from such trauma that somehow they managed to genetically transmit this trauma to their kids, to use one recent example). This third concept has a lot of people very hyped but the evidence for it is very, very bad. It is neo-Lamarckism under another name, and Lamarckism fell out of favour for a reason. This and other issues are discussed more fully by Kevin Mitchell here:

    I agree with your thoughts on rGE (gene-environment correlation). It is worth remembering that occasionally the signs can be negative (such as the kid genetically predisposed to hyperactivity who gets told to shut up and sit down all the time). But this is probably not the norm. In the active form, where individuals shape their own environments, rGE may be part of the reason why heritability of IQ and other traits tend to rise with age (though I would still ultimately classify this form of rGE as genetic variance, but opinions may differ). In the passive form, rGE messes up a lot of social science studies without genetic controls. For instance, number of the books in the home predicts child educational outcomes. Why? Well, maybe a direct causal effect, but number of books in the home is also to some degree a reflection of the parents’ genotypes, which of course the child shares….the moral is that even in 2015 a vast quantity of studies are severely flawed by a lack of control for passive rGE.
    I also mostly agree with point 4 but would qualify it in a number of ways. The intelligence-weight comparison is slightly misleading. We do know how to achieve real reductions in weight, at least in the short term. We also know how to raise intelligence test scores – teach to the test – but of course this is not the same as raising genuine mental performance. Intervention programs designed to increase mental functioning permanently have had largely disappointing results, particularly in the developed world (in much of the developing world there is still a lot of room to make everyone smarter via better prenatal nutrition). Around the borderline of IQ 70, independent functioning and full participation in civic life becomes increasingly difficult, especially in an ever more technologically complex world with fewer jobs available that place a premium on easily trainable repetitive work. Low IQ is also a substantial risk factor for many biomedical outcomes, such as all-cause mortality and schizophrenia. The lesson is that low IQ is a serious problem, both for the individual and for society, and that modern methods of education do not adequately fix it. Just to put this into perspective, many of the US black-white differences in important life outcomes – income, educational attainment – shrink substantially, go to zero, or even reverse sign after controls for IQ.
    I am not totally sure what your point 5 is saying, so I will ignore it 😉


Leave a Reply