October 16, 2018   6 mins

Here’s something that we’re all going to have to come to terms with pretty sharpish. We could, if we wanted, select embryos for intelligence right now. It would be a deeply imperfect procedure, sure. But, if it were legal, we could select embryos that, on average, were slightly more intelligent than average. At an individual level it probably would’t have all that much immediate effect – but, on a national level, it may be a very different story.

And we’re going to get better at it. That means that, as a society, we’re going to have to make big decisions on this, soon. As it stands, we know many bits of DNA that are involved in intelligence; we have the technology to select embryos for them, or to (crudely) edit embryos to include some; and while Britain has stringent legal oversight of genetic research and its clinical applications, that is far from true everywhere.

Here’s how it works. First, stop talking about “the gene for” intelligence. Genes are ‘coding’ regions of the chromosomes: the bits that tell your cells to make specific proteins. There are about 20,000 of them.

Genes, though, only make up about 2% of your actual genome. The other 98% doesn’t code for proteins, but can have profound effects. Nowadays geneticists are much less likely to talk about “genes”, and much more likely to talk about changes to the individual building blocks of DNA anywhere on the genome. The different forms of those building blocks are called “single nucleotide polymorphisms” or SNPs, pronounced “snips”.

Also, for almost all complex traits, it’s not “a” gene or SNP, it’s absolutely loads of them. For instance, one big study looked at 1.1 million people’s genomes, and found 1,271 SNPs relating to intelligence. It estimated that those SNPs explain about 11% of the variance in educational attainment and about the same in cognitive ability. (The “variance” is a measure of how spread-out the scores are around the average.) A review in Nature Genetics this year estimated that our current knowledge allows us to explain about 20% of the overall variance in intelligence in terms of specific SNPs.

We know, from twin studies, that intelligence is at least 50% heritable, so there’s a lot more to find. But still, that’s a big deal. For comparison, the impact of (non-negligent, non-abusive) parenting on IQ is probably essentially zero; the most generous estimates put it at 10%.

So, we know lots of SNPs that are involved in IQ. Using statistical techniques, we could predict quite effectively which of two people was more intelligent simply by looking at their genomes.

And, right now, we can look at the genomes of embryos before they’re implanted during IVF. It’s already done for various diseases caused by single mutations – it’s called preimplantation genetic diagnosis. At the fertility clinic, a couple might have 10 or so viable embryos; the clinicians would test them all, and if one of them were to have the gene for cystic fibrosis or sickle-cell anaemia or Huntington’s disease, you throw that one away and implant one of the ones that doesn’t.

But now we can do it for the complex, ‘polygenic’ conditions. Things like diabetes, many cancers, and depression are heavily genetic. There’s no one SNP that causes them, but you can estimate an embryo’s risk by looking at the whole genome.

That’s easy, now. Less than 20 years ago, it took the Human Genome Project more than a decade and over a billion dollars to sequence one single genome. Now you can do the same thing for about £1,000 in less than 24 hours. There’s at least one company offering to do exactly this with “type 1 diabetes, type 2 diabetes, coronary artery disease” etc and so on. This isn’t, necessarily, selling snake-oil to idiots: a new study thinks you can identify people at a greater than 50% risk of diabetes from their genetics alone.

You could do exactly the same thing with intelligence. You could sequence the genomes of a couple’s embryos prior to implantation and select the one with the selection of SNPs most correlated to intelligence. And this is before we get onto the possibilities of actually editing the genome, with tools such as CRISPR

There’s an objection some people make at this point. We can’t select for intelligence, they say, because we don’t know what intelligence is. IQ tests, they say, only measure how good you are at IQ tests.

That’s not really true. How well you do at IQ tests is highly correlated with how well you do at school and work, and how much you earn, to your physical and mental health, to your reaction time, even to your per-year risk of death. It also correlates with all the many different ‘kinds’ of intelligence, such as emotional intelligence. (All this holds even after you control for race and social class: this isn’t, or isn’t simply, about posh white people making a test that posh white people do well at, in a society that makes it easy for posh white people in general.) There have been awful misuses of IQ research, but it is clearly measuring something real. And it is clearly highly genetically influenced.

So there are incentives for parents to test their embryos for the SNPs correlated to IQ. One statistically minded blogger I follow estimates that, with current knowledge and technology, selecting the highest-scoring embryo from 10 would correlate to about a 3-point bump in IQ. Nick Bostrom of Oxford University says that research finds that each 1-point gain in IQ roughly correlates to a 1% increase in income. It’s not huge, it’s wildly variable, and it may be that the causal direction isn’t as straightforward as that, but it’s certainly not negligible.

If you look at a national level, though, things get interesting, if controversial. While individual IQ scores are only fairly predictive – someone with a high IQ is more likely to have a high-paying job, but there’s lots of variation and a bad score on an IQ test definitely does not mean you’re going to have a terrible life (lots of professors have below-average IQs) – national IQ effects are much stronger. One, admittedly controversial, study suggests that a 10-point jump in average IQ correlates to a doubling of the size of the national economy.

But how accurate is that? Many people have argued against it, others in support of it. It’s also likely that the causation is complex: a bigger economy leads to higher IQs, as well as the other way around. But even if the figure is off by an order of magnitude, and a 10-point jump in IQ merely leads to a 10% increase in GDP, that is still big. You’d need to do a lot of genetic engineering for it to have an effect – it would probably have to be state-supported – but over a few generations it could be dramatic.

In the West, we’re wary about human genetic engineering. A recent survey found that 80% of Americans opposed editing an embryo’s genes to improve its intelligence. There are concerns over its safety, which presumably will reduce as the technology matures.

But there is also a fear of ‘playing god’ and interfering with nature. More seriously, to my mind, people worry, not unfairly, that it could lead to the creation of a genetic underclass; with rich people able to afford cleverer children, poorer people not.

Where you stand on all this will be largely a matter of temperament: different people will have different feelings about the same evidence. But arguably, everything humans have done since the invention of the flint axe is ‘playing god’, and everything that humans do is easier for rich people to do. In fact, rich people have traditionally been able to boost their intelligence through better nutrition and education. Genetic editing at least has the advantage that the benefits will be in the gene pool for everyone, and as it gets cheaper it’ll be more accessible for lower earners, just as food and education has become.

And the advantages are obvious. Smarter countries are happier, richer, and longer-lived. Being smarter is correlated with being happier and longer-lived. My own feeling is that – with due attention being paid to the caveats, the risks and the downsides – genetic selection for intelligence (and other traits) will probably be, on balance, a good thing, Brave New World notwithstanding.

In any case, it’s going to happen whether we like it or not. China is heartily embracing the possibilities. The facility at BGI Shenzhen is the largest genetics research centre in the world and does huge amounts of research into human intelligence. The Chinese population is less wary of genetic engineering than Western populations are. And while “germline” editing – editing the DNA in our eggs and sperm that would be passed on to our children and their children – is banned in many Western countries, including Britain, it’s not in China (or several other countries). China was the first country to edit human embryos, back in 2015.

Essentially, we can’t stop this sort of genetic engineering. The incentives, at national and individual level, make it inevitable. China is simply going now where we’re all going to follow later. It would be better to acknowledge the reality and try to shape it – rather than try to hold back the flood.


Tom Chivers is a science writer. His second book, How to Read Numbers, is out now.

TomChivers