biology, G factor (psychometrics), Genetics, Genome-wide association study, GWAS, heritability, Human, intelligence, Intelligence quotient, People, Popular science, science, science and society
A study from the University of Edinburgh claims to have found the basis of our intelligence in thousands of genes scattered throughout our genome. Although the discovery was made possible by a new statistical method and modern sequencing technology, how the results are interpreted hinges upon a century-old debate about what intelligence is and how we measure it. Will we ever be able to measure something so indefinable or discover the genes behind it?
In 1904 the English psychologist Charles Spearman noticed that people who do well in one subject also tend to do well in other, unrelated subjects; Spearman used this observation as the basis for his theory of intelligence. He developed a statistical method to calculate the “general factor” behind this performance; the g-factor tries to capture a person’s overall intelligence, the core ability that makes them perform similarly at different mental tasks. In the century since Spearman, researchers have continued to elaborate this theory. The modern incarnation treats intelligence as a hierarchy, with g subdivided into broad classes such as “crystallized intelligence”, which is based on long-term memory and acquired knowledge, and “fluid intelligence”, which requires on-the-spot thinking to deal with unfamiliar material and novel situations. Despite disagreements between experts about what g actually measures, it remains the most widely accepted metric of intelligence and correlates well with things like academic achievement, job performance and income.
There’s been continuous debate about the genetic contribution to intelligence. Although many studies have shown that g is highly heritable, it’s been difficult to eliminate possible effects from a shared environment, particularly in the womb. The advent of whole-genome sequencing offers a solution to this dilemma by making it possible to directly compare the genomes of different people and look for correlations with intelligence measurements. In a genome-wide association study (GWAS), researchers take advantage of the hundreds of thousands of little differences in our genome that make every person unique. People who share a characteristic (like a high g score) will have genomes that are similar in the area responsible for that trait; when all of the differences across the genome are compared, this island of similarity will stand out, pointing scientists towards the genes responsible for that trait. Ideally, there will be one region of high similarity, but that often isn’t the case, especially for complex traits which are influenced by many genetic and environmental factors. Unfortunately, intelligence is just such a trait. Although several GWAS in the past few years have tried to discover the genes behind intelligence, none have succeeded.
To overcome this hurdle, the researchers used a statistical technique they had recently developed to analyze height, another complex trait. Led by Dr. Ian Deary and Dr. Peter Visscher, the team measured the crystallized and fluid intelligence of over 3500 volunteers and conducted a GWAS to see how these scores correlated with differences in their genomes at nearly 550,000 locations. Although none of the locations stood out individually, the new method enabled the team to also consider the impact of similarity at all the locations together. The summed effect across the entire genome explained 40% of the variation in crystallized intelligence and 51% of the variation in fluid intelligence. In other words, the study revealed that intelligence isn’t driven by a handful of genes with strong effects but by thousands of genes each of which has a very small influence. These findings were later confirmed by an independent study at Harvard, which showed that 47% of the variation in g was accounted for by 630,000 locations throughout the genome.
These GWAS showed that g has a genetic basis, but since it still isn’t clear what g actually measures this might not translate into finding genes responsible for intelligence . In addition to doing well academically and professionally, people with a high g score also tend to be taller and healthier, have more symmetric bodies and live longer. These correlations have led some researchers to suggest that the “general factor” measured by g may not be overall intelligence but something more like biological fitness. Under this interpretation, these GWAS have shown that thousands of genes affect general fitness rather than general intelligence; their effect on intelligence is indirect, since people who are more fit will tend to perform better on intelligence tests as well as living longer, healthier lives.
By investigating the whole genome at once, the researchers at Edinburgh were able to directly demonstrate a genetic contribution to intelligence for the first time. They also showed that this results from the combined action of many genes, each making a small contribution, rather than a few “intelligence genes”. Although their research conclusively demonstrates a genetic basis for intelligence, the precise nature of this genetic basis and how it translates into an effect on intelligence remains unclear.
Davies, G., Tenesa, A., Payton, A., Yang, J., Harris, S., Liewald, D., Ke, X., Le Hellard, S., Christoforou, A., Luciano, M., McGhee, K., Lopez, L., Gow, A., Corley, J., Redmond, P., Fox, H., Haggarty, P., Whalley, L., McNeill, G., Goddard, M., Espeseth, T., Lundervold, A., Reinvang, I., Pickles, A., Steen, V., Ollier, W., Porteous, D., Horan, M., Starr, J., Pendleton, N., Visscher, P., & Deary, I. (2011). Genome-wide association studies establish that human intelligence is highly heritable and polygenic Molecular Psychiatry, 16 (10), 996-1005 DOI: 10.1038/mp.2011.85
Chabris, C., Hebert, B., Benjamin, D., Beauchamp, J., Cesarini, D., van der Loos, M., Johannesson, M., Magnusson, P., Lichtenstein, P., Atwood, C., Freese, J., Hauser, T., Hauser, R., Christakis, N., & Laibson, D. (2012). Most Reported Genetic Associations With General Intelligence Are Probably False Positives Psychological Science, 23 (11), 1314-1323 DOI: 10.1177/0956797611435528
Sounds like we need someone really smart to figure it out. 😉
Artem Kaznatcheev said:
Usually when I see fitness mentioned in a post about genetics or evolution, I think of of the biological definition of something like direct fitness: number of offspring that survive to adulthood (or maybe number of grandchildren if we are looking at life histories). However, it doesn’t look like this is what you mean by your use of the term “general fitness”. It seems like you mean something like “physical healthiness” or well-being. Am I misunderstanding something? Can you clarify what you mean by general fitness?
I don’t really mean “physical healthiness”, though I could see why you would think so. I’ve got biological fitness in mind, but in an indirect way which I’m happy to clarify. I’ll try two different approaches and see if either helps:
(1) If you’re familiar with evolutionary biology, what I’m actually talking about isn’t really ‘fitness’ but ‘mutational load‘, which I thought would be too complicated an idea to introduce in this post. The idea is that the thousands of correlations found across the genome by the GWAS aren’t many genes with small effects, but simply a measure of the genome-wide genotype, which might correlate with something like mutational load, which might affect intelligence.
(2) If that’s too much evolutionary biology for you, another way to think about it is to ask: Does higher intelligence make humans more fit (yes, biological fitness) or are humans with a higher fitness (thanks to their genes, etc) also likely to be more intelligent? To make up an example: there might be a mutation that makes someone extremely efficient at maintaining their body temperature. This person would spend less energy staying warm and the extra energy might (somehow) enable them to have more offspring (ie, be more fit) — maybe they’ve got enough extra energy to find and impress a very high quality mate or to mate with more partners. Some of the extra energy could also be used during the development of their brain or directly in mental tasks, making them more intelligent. The effect on intelligence could be entirely secondary and might not change their fitness at all. In that case, the body-warmth mutation might lead to higher fitness which would be associated with higher intelligence (or being taller, etc) even though the gene is actually a metabolic gene that has nothing to do with intelligence (or height). I’m not suggesting anything quite so crude is actually happening, but rather that thousands of similar effects (which is what I called “general fitness”) might be what the GWAS is picking up.
That was excellent question; thanks for asking it! I hope my answer was helpful. If not, somebody please ask for more clarification. 🙂
Artem Kaznatcheev said:
saying mutation load is more or less the same as saying direct fitness (since you can pretty safely assume that most mutations are either neutral or deleterious on direct fitness). However, this (and parts of approach 2) still puts you into an awkward stance of having to argue that intelligence correlates (in the indirect case you discuss; or causes in the direct case) higher fitness (in the biological sense). This is a non-trivial argument to make, see for instance the ideas of Satoshi Kanazawa. I usually find it safest to simply avoid the word “fitness” when talking about “intelligence”: the first term is already hard enough to define, the second is nearly impossible (as you pointed out).
I see what you’re saying, but I think we may be talking at cross purposes. I’m not trying to argue that intelligence is or isn’t linked to fitness — I have no idea if that’s the case and I would surprised if anyone really does. I am trying to raise questions about what the “g factor” actually measures. It’s generally considered a measure of intelligence, but it also correlates with things like height and even body symmetry. Given that the GWAS found that g> is associated with thousands of sites scattered throughout the genome, I’m asking (and I’m not the first) whether it might actually be measuring something like mutational load rather than intelligence. Yes, that would mean that mutational load (and height, symmetry, etc) would have to correlate with mutational load, but I don’t think that’s an unreasonable possibility.
For what it’s worth, I’m partial to the idea that sexual selection has played an important role in the evolution of intelligence (or at least brain size).
Reblogged this on The Science Blog.
Pingback: The Week In Science (Feb 23 – Mar 3) | Science Communication Blog Network
Pingback: Games, culture, and the Turing test (Part I) | Theory, Evolution, and Games Group