What if we’ve been thinking about the genetics of intelligence from completely the wrong angle? Intelligence (as indexed by IQ or the general intelligence factor “g”) is clearly highly heritable in humans – people who are more genetically similar are also more similar in this factor. (Genetic variance has been estimated as explaining ~75% of variance in g, depending on age and other factors). There must therefore be genetic variants in the population that affect intelligence – so far, so good. But the search for such variants has, at its heart, an implicit assumption: that these variants affect intelligence in a fairly specific way – that they will occur in genes “for intelligence”.
An implication of that phrase is that mutations in those genes were positively selected for at some stage in humanity’s descent from our common ancestor with apes, on the basis of conferring increased intelligence. This seems a fairly reasonable leap to make – such genes must exist and, if variation in these genes in humanity’s evolution could affect intelligence, then maybe variation in those same genes can explain variation within the human species.
The problem with that logic is that we are talking about two very different types of variation. On the one hand, mutations that arose during human evolution that conferred increased intelligence (through whatever mechanism) will have been positively selected for and fixed in the population. How this happened is unknown of course, but one can imagine an iterative process, where initial small changes in, say, the timing of processes of brain development led to small increases in intelligence. Increased cognitive capabilities could have led in turn to the emergence of crude communication and culture, opening up what has been called the “cognitive niche” – creating an environment where further increases in intelligence became selectively more and more advantageous – a runaway process, where genetic changes bootstrap on cultural development in a way that reinforces their own adaptiveness.
That’s all nice, though admittedly speculative, but those mutations are the ones that we would expect to not vary in human populations – they would now be fixed. In particular, there is little reason to expect that there would exist new mutations in such genes, present in some but not all humans, which act to further increase intelligence. This is simply a matter of probabilities: the likelihood of a new mutation in some such gene changing its activity in a way that is advantageous is extremely low, compared to the likelihood of it either having no effect or being deleterious. There are simply many more ways of screwing something up than of improving it.
That is true for individual proteins and it is true at a higher level, for organismal traits that affect fitness (the genetic components of which have presumably been optimised by millions of years of evolution). Mutations are much more likely to cause a decrement in such traits than to improve them. So maybe we’re thinking about the genetics of g from the wrong perspective – maybe we should be looking for mutations that decrease intelligence from some Platonic ideal of a “wild-type” human. Thinking in this way – about “mutations that affect” a trait, rather than “genes for” the trait – changes our expectations about the type of variation that could be contributing to the heritability of the trait.
Mutations that lower intelligence could be quite non-specific, diverse and far more idiosyncratic. The idea of a finite, stable and discrete set of variants that specifically contribute to intelligence levels and that simply get shuffled around in human populations may be a fallacy. That view is supported by the fact that genome-wide association studies for common variants affecting intelligence have so far come up empty.
Various researchers have suggested that g may be simply an index of a general fitness factor – an indirect measure of the mutational load of an organism. The idea is that, while we all carry hundreds of deleterious mutations, some of us carry more than others, or ones with more severe effects. These effects in combination can degrade the biological systems of development and physiology in a general way, rendering them less robust and less able to generate our Platonic, ideal phenotype. In this model, it is not the idea that specific mutations have specific effects on specific traits that matters so much – it is that the overall load cumulatively reduces fitness through effects at the systems level. This means that the mutations affecting intelligence in one person may be totally different from those affecting it in another – there will be no genes “for intelligence”.
Direct evidence for this kind of effect of mutational load was found recently in a study by Ronald Yeo and colleagues, showing that the overall burden of rare copy number variants (deletions or duplications of segments of chromosomes) negatively predicts intelligence (r = -0.3).
If g really is an index of a general fitness factor, then it should be correlated with other indices of fitness. This indeed appears to be the case. G is weakly positively correlated with height, for example, and also strongly correlated with various measures of health and longevity. In a recent, outstanding review by Ian Deary, the following statistics are cited:
“One standard deviation advantage in intelligence was associated with 24% lower risk of death over a follow-up range of 17 to 69 years (Calvin et al. 2011 – [meta-analysis]). … The range of causes of death with which intelligence is significantly associated… include deaths from cardiovascular disease, suicide, homicide, and accidents, but not cancer.”
This correlation can be interpreted in two ways: one, less intelligent people have less healthy and/or riskier lifestyles (i.e., direct causation), or, two, both intelligence and rates of mortality at least partially reflect an underlying factor – general fitness.
Another good marker of general fitness is developmental stability. This refers to the robustness of the system and the ability of the genotype to reliably generate a phenotype within the species-specific normal range, despite genetic and environmental perturbations and intrinsic noise or randomness. It is a property that varies between people.
One can get a good measure of developmental stability by looking at how symmetric someone is. The two sides of the body develop independently from the same set of genomic instructions – if a particular genotype is very robust then it should generate a very similar outcome on each side of the body. If, however, the system is less robust, then the person may be more asymmetric in any number of features (arm lengths, finger widths, earlobe lengths, eye widths, etc.). This kind of asymmetry is called fluctuating asymmetry as the direction is random – one arm may be longer than the other, but it is equally likely to be the left or right (unlike the asymmetry of internal organs, for example, which is directional and a species-specific trait).
Fluctuating asymmetry should thus be a good indicator of general fitness and is fairly easy to measure (though it is important to look at multiple features to get an aggregate score in each individual). It is also a heritable trait – monozygotic twins are more similar to each other in degree of asymmetry than are dizygotic twins. There is no reason, however, to think this reflects variation in a set of genes whose function it is to make the organism more symmetric, or to make developmental systems more robust. Rather, mutations in any genes affecting development are likely to not just contribute to some specific phenotype, but also to generally decrease robustness of the system and increase variability.
You can probably guess what’s coming next – fluctuating asymmetry correlates negatively with various IQ measures. At least, most of the studies that have looked at it have found such a correlation – ranging from –0.2 to –0.4, which is fairly substantial. Not all studies have found this but a meta-analysis confirms a correlation with a value between –0.12 and –0.2. This correlation is weaker, but still significant, and means that there is at least some relationship between intelligence and symmetry. (It may also be an underestimate, as one study found that psychometric tests with heavier loadings on g showed greater correlations with fluctuating asymmetry). The most plausible interpretation is that this correlation reflects the effects on both parameters of a “latent variable” – general fitness.
This may, incidentally, also explain the recently demonstrated correlation between intelligence and physical attractiveness, which itself has been correlated with facial symmetry. (Gratuitously exemplified by the lovely, symmetrical Kate Beckinsale and her mirror-image right- and left-side dopplegangers).
Correlations of intelligence with measures of brain size, white matter integrity, network efficiency or other parameters may be similarly explained. They could be either independent correlates of general fitness or the structural measures could be the substrates of phenotypic differences in intelligence – the means by which general fitness affects g. According to that model, expecting to find defined sets of genes “for white matter integrity” would be as misguided as looking for genes “for intelligence”.
Instead, we may all carry many mutations that affect intelligence, negatively and mostly non-specifically, with the total burden determining how far away we each are from our archetypal Homo platonis.