Breaking the mould: genetics and education

Tonight I participated in BBC’s “The Moral Maze”, discussing the recent reactions to a report by Dominic Cummings, an advisor to the education secretary, that mentioned that genetic factors have a big impact on educational outcomes. This ties in with the recent book G is for Genes by Kathryn Asbury (also on the program) and Robert Plomin where they argue that children are not blank slates and that genetic information might enable personalized education. Ah, children, genetics, IQ, schools – the perfect mixture for debate!

Unfortunately for me the panel tore into my transhumanist views rather than ask me about the main topic for the evening, so I ended up debating something different. This is what I would have argued if there had been time:

Talent is unfair. While one can quibble about what it actually is, and acknowledge that it is something that emerges not just from the genes but also from their interaction with the environment, it is clear that different people are born with different aptitudes for different things. Some of these aptitudes help a life go well. So through no fault of their own, some people will have less chance of a good life.

If we were to make a choice behind a veil of ignorance between a world where there was more talent to go around and a world with less talent, it seems that the reasonable choice is to choose the world of talent. We would probably also want to choose a world where talent was more equally distributed than one where it was less equal. But even the less talented people in a talented but unequal world could benefit from the greater prosperity and creativity.

In practice talent needs plenty of help to develop: without support and good teachers innate potential is unlikely to matter. So the ability to help kids develop their potential (and help them overcome their less able sides) is important for actualizing that talent. Without it none of the above worlds would be preferable. But figuring out how to cultivate and stimulate kids is hard. Hence, any information that could help do this better would be welcome.

So my basic stance is that if genetic information could personalize education well, go for it!

But… I am less convinced than the geneticists that we can actually do it, at least in the near future. Genetics is hard. It is surprisingly tricky to establish how genes translate into actual outcomes since so much is interacting. Even when there are statistical differences between groups it might not tell us much. For example, I have GG at SNP rs363050, something which (according to one study) is associated with about 3 IQ points less in non-verbal IQ. Given that I am in the philosophy faculty at Oxford I can’t be that stupid – no doubt I have compensating genes. Or a really good upbringing. Or maybe the variation only matters in some people. Or with some environments. Knowing about my rs363050 would not have helped my teachers to teach me better. Giving some extra non-verbal tasks might have make sense on average to people in the rs363050 GG group, but it is not clear that it would have helped me. The teachers would have been better off looking at who I was and what tasks I did well or badly at. In cases like this looking at the phenotype, the actual behaviour and abilities, is much more revealing that any amount of genotype information.

This is of course one of the fears raised by the report and the book: what if our society starts to pre-judge children based on their genotypes? It certainly is a real risk, but as Asbury tried to point out, it would be judging that is not based on the science. In fact, it would be stupid – hiring people or channelling kids based on a weak marker for ability rather than actual demonstrations of ability will lead to massive mistakes. Maybe the science does lend itself too easily to simplistic caricatures, but the fault is not in the science itself or even pointing out that it might be useful, but in us as a society allowing oversimplifications rule decisions.

Labelling, even well-meaning labelling, can have detrimental effects even when it is based on real information. Being told you are a low performer will usually not motivate you. Teacher expectations can easily bias student performance, and vice versa. Genetic markers are ready-made labels – but only if we let them. Genetic determinism is a mistake, and we should not teach it – either through the curriculum, or through the structure of the school itself.

There is a second problem with personalised education. Getting something useful out of the genetic information requires not just good genetic data gathering, but also good educational data gathering. It doesn’t matter if we find associations between genes and grades if we have no idea how to influence things. This will require vast amounts of fairly detailed data and a close collaboration between the behavioural geneticists and educators – not a simple task, as neuroscience has realized when trying to help education. Just because we know how learning works in the brain doesn’t mean we can apply that cognitive knowledge well to education.

In the long run I am sure we will figure out a few useful things the genome does tell us about learning styles, talent or other things that matter for education that could not be detected by a skilled teacher. But that raises another problem: might the personalisation itself be unfair? I am not talking about the well-off getting better education (that is an issue regardless of genetics). Some kids will have genetic markers that enable useful personalisation that help them excel, and some kids will lack them – they will have to do with standard education. This is in a sense exactly the same unfairness as the random distribution of talent represents, but here it is a random distribution of personalizability. One can still argue that unequal distributions are OK as long as the worse off benefit (educational resources get allocated more efficiently), but it seems that we should strive for something better.

That something better is of course to enhance people’s ability from the start. It replaces the meta-problem of unequal personalizability with the first-order problem of fair enhancement – something that is going to be relevant anyway across a lot of domains if we can do it.

During the program I was prodded about the perceived perfectionism of transhumanism, that we want all people to develop themselves in the same direction, ending up in the same mould. This is of course a misunderstanding of what the aim is: it is getting away from a lot of bad states – lack of talent, illness, cruelty – and aiming at personally chosen goals. Given the diversity of human life projects I am not worried that we would become bland and homogeneous, just as education – when done well – does not shape us in the same mould but instead allow us to grow and develop our potential.