Negative selection by intelligence level: details

Genes that contribute to getting a good education are eliminated by selection 

Alexander Markov, "Elements"

In modern human populations, the level of education received is highly heritable, that is, it strongly depends on genes. Dozens of alleles affecting this trait have been identified. At the same time, education, as a rule, negatively correlates with Darwinian fitness: educated people reproduce worse. This indicates a possible selection against the "genes of education". A new study based on data on 110,000 Icelanders born between 1910 and 1975 has shown that "education genes" are indeed subject to negative selection. These alleles, many of which are also correlated with increased intelligence, good health and a long life, reduce fitness regardless of whether a person has realized the propensity to receive a good education due to them. The study confirmed fears that the evolution of modern humanity is directed towards the deterioration of the genetic basis of traits associated with intelligence. Socio-cultural development is still more than compensating for genetic degradation, but over time its consequences may become significant.

1. Genetic degradation of mankind and its possible mechanisms

The idea that evolution in modern human populations is directed in a completely different direction than we would like was first described in detail and justified by the greatest evolutionary geneticist of the XX century Ronald Fisher in the final chapters of his famous book "The Genetic Theory of Natural Selection" (1930).

Research in this area has progressed slowly, including for reasons unrelated to science. Nevertheless, to date, a lot of data has accumulated, indirectly confirming that the genetic basis of traits related to physical and mental health is gradually eroding and degrading, especially among residents of developed countries. Geneticists discuss two main mechanisms of such degradation (see: M. A. Woodley of Menie, 2015. How fragile is our intellect? Estimating losses in general intelligence due to both selection and mutation accumulation).

The first mechanism is the weakening of purifying selection against mutations that disrupt the work of certain organs and systems of the body, including the immune system and the brain. Previously, such mutations reduced fitness (their carriers left fewer offspring on average) and therefore were cleaned out by selection. However, the development of medicine, social welfare and other benefits of civilization has made these mutations less harmful in an evolutionary sense. Or even completely harmless (neutral). Therefore, they began to accumulate freely in the gene pool. The development and functioning of the cerebral cortex depend on the coordinated work of a huge number of genes, which makes our cognitive abilities especially vulnerable to mutational degradation (G. R. Crabtree, 2013. Our fragile intellect).

The second mechanism is negative selection based on traits that help (or have helped in the past) to achieve success in life. Previously, such signs increased Darwinian fitness, and in modern society, perhaps, they began to reduce it – for example, due to the fact that talented, educated and economically successful people began to postpone the birth of children "for later". Thus, there is evidence according to which in the USA and other countries people with high intelligence leave fewer offspring (R. Lynn, 1999. New evidence for dysgenic fertility for intelligence in the United States). There is also reason to assume that in developed countries after the industrial Revolution, the orientation of selection on the grounds of social and economic success has changed to the opposite. For example, if earlier a skilled shoemaker and a shrewd merchant left more offspring than their crooked and stupid competitors, then since the XIX century everything has become the opposite (V. Skirbekk, 2008. Fertility trends by social status).

However, until now, scientists have mostly had only indirect evidence of the genetic degradation of mankind. There was a lack of direct evidence that the genes contributing to the development of the psychological and cognitive traits we value are indeed subject to negative selection, and that this selection is effective (that is, it leads to a decrease in the frequency of occurrence of the corresponding alleles). And thanks to the development of comparative genomics methods and the creation of large databases on medical genetics, such evidence has finally appeared, with which we can all be congratulated.

2. A comprehensive indicator of genetic predisposition to education

Icelandic geneticists have studied the relationship between genes that affect the level of education and Darwinian fitness using the example of a sample that includes a significant part (about half) of all Icelanders born from 1910 to 1975.

The level of education received, measured by the number of years spent on study (educational attachment, EA) – a phenotypic trait that is convenient for studying, which, apparently, is closely related to cognitive abilities and such traits of the psyche that we usually consider virtues (purposefulness, thirst for knowledge, the ability to plan your life, openness to new experiences). It has been reliably established that EA in modern human populations has high heritability. Differences between people in terms of education level are explained by genes by at least 30-40% (the remaining 60-70% are the result of different environmental conditions and ontogenetic noise, see: Developmental noise).

Recently, a sample of 20,000 Americans was able to show for the first time that the genetic basis of EA is subjected to negative selection (J. P. Beauchamp, 2016. Genetic evidence for natural selection in humans in the contemporary United States). The new work of Icelandic geneticists not only confirmed this conclusion on a more representative sample, but also showed a real decrease in the frequency of occurrence of alleles contributing to high EA during the XX century.

The work under discussion is largely based on a 2016 study that analyzed a sample of 293,724 people of European descent. The analysis of this sample revealed 74 specific regions of the genome (locus) that significantly affect the level of education (for more information, see in the news The level of education received partly depends on genes, "Elements", 16.05.2016). The data collected in the course of this work, Icelandic geneticists used to derive a formula for calculating a complex indicator of genetic predisposition to education – educational attachment polygenic score (POLYEDU). This indicator, with accuracy depending on the size of the sample used, reflects the joint contribution of many genes (ideally, all genes of the genome) to the predisposition of a given person to receive education. The POLYEDU indicator is composed of individual "weights" that were calculated for each of the 620,000 genetic markers (polymorphic loci), depending on how the allelic variants of each of these loci correlate with EA. If the presence of an allele is associated with a slightly increased mean EA, the allele is given a small positive weight, etc. The authors tried several ways to calculate POLYEDU and were convinced that methodological subtleties (for example, ways to circumvent statistical problems associated with nonequilibrium coupling) have little effect on the final results.

To avoid unnecessary difficulties in interpreting the results, when developing the formula for POLYEDU, all Icelanders who were included there (49,970 people) were excluded from the initial sample of 293,724 people, and 111,349 people from the UK Biobank database were added instead. The increase in the sample led to the fact that the individual loci that reliably correlated with EA were no longer 74, but 120.

The POLYEDU indicator calculated by the obtained formula allows us to explain 3.74% of the EA variability. In other words, this indicator characterizes about one tenth of the entire genetic basis of EA (since the entire genetic basis explains 30-40% of the variability). If the formula for POLYEDU is derived on the basis of a smaller sample – without taking into account data from the UK Biobank – then the resulting indicator will explain only 2.52% of the variability. Apparently, in order to calculate the weight of each allele even more accurately – so that the final indicator explains most of the genetically determined variability in EA – data is needed not for hundreds of thousands, but for millions of people. The POLYEDU indicator was calibrated so that its average value was zero and the standard deviation was one, and then it was measured in these "standard units".

3. "Education genes" reduce fitness

Using the obtained formula, the authors calculated POLYEDU for 109,120 genotyped (see: Genotyping) Icelanders born from 1910 to 1975. Individual POLYEDU values were compared with three reproductive indicators:

1) the total number of descendants left by a person over a lifetime (number of children, NC), not counting those who died in infancy (this indicator is close in meaning to Darwinian fitness);

2) the age of the parent at the time of the birth of the first child (age at first child, AGFC);

3) the average age of the parent at the birth of offspring (average age at child birth, AACB).

The results of the analysis are shown in Fig. 1.

Fig. 1. A table showing the correlation between POLYEDU and three reproductive indicators. The values in the "Effect" column mean that an increase in POLYEDU by one is accompanied by a change in this attribute by the specified amount. The number of children (NC) is measured in units, age (AGFC, AACB) – in years. Thus, the table shows that for women, an increase in POLYEDU per unit corresponds to a decrease in the number of children by 0.084, an increase in the age of birth of the first child by 0.59 years, and the average age of childbirth by 0.46 years. The corresponding values for men are -0.054, 0.44 and 0.37. Here and below are the figures from the article under discussion in PNAS.

It turned out that genes contributing to a high level of education significantly reduce reproductive success in both sexes. In addition, they contribute to the shift of the beginning of reproduction to a later age. All effects have a high level of statistical significance, but they are more pronounced in women than in men. At the same time, AGFC increases with POLYEDU growth more than AACB. This suggests (and additional calculations confirmed this assumption) that "education genes" reduce fertility mainly due to the fact that the birth of the first child is shifted to a later age (I use the incorrect term "education genes" so as not to write long formulations like "a complex of alleles, the presence of which is positively correlated with an increased level of education").

The results of a more detailed analysis of the effect of POLYEDU on fertility depending on the age of parents are shown in Fig. 2. "Education genes" sharply reduce early fertility, while they have a positive effect on childbearing at a later age. The early negative effect greatly outweighs the late positive, so that the total number of children is significantly lower in people with high POLYEDU. The graph in Figure 2 is arranged in such a way that it increases until people with low POLYEDU give birth to more children than people with high POLYEDU. Thus, the fracture occurs at the age of about 30 years, when the graph reaches a maximum. Up to this age, people with low POLYEDU values reproduce better After 30 years, the owners of high POLYEDU slightly reduce the gap in the number of children, but they can no longer catch up with their rivals. The overall assessment of POLYEDU's impact on fitness (the number of children over a lifetime) corresponds to the two left values on the graph (-0.084 for women and -0.054 for men). These are the same values that are shown in the table in Fig. 1.

Fig. 2. A graph showing the cumulative effect of the "genes of education" (effect of polygenic score) on the number of children born at the age shown on the horizontal axis, or later. The red icons correspond to mothers, the blue ones to fathers. The leftmost values corresponding to the age of 14 years show the influence of "education genes" on the total number of children born over a lifetime (14 years is the minimum age of childbearing in the studied sample). These values correspond to the figures given in the table (-0.084 for mothers and -0.054 for fathers) and represent the final assessment of the influence of "education genes" on Darwinian fitness. The rightmost values show the influence of "education genes" on the number of children born at the age of 40 or more.

In this paper, the authors focused on the cumulative effect of multiple genes, rather than on the role of individual alleles. The individual contributions of most genes to EA variability are very small, so the sample size should be larger to study them. But the authors still noted some alleles. For example, the minor (rarer) variant of the polymorphic locus rs192818565 negatively affects EA. At the same time, it apparently undergoes positive selection, that is, it increases Darwinian fitness (there are a number of effective methods for searching for traces of positive selection in genomes). It is also known that the presence of this allele correlates with neuroticism and reduced cranial volume. As it turned out, this is clearly a "bad" allele (who wants to be a neurotic with a small brain?) significantly correlates with a large number of children and the early onset of childbirth. As the authors rightly point out, "this is a striking example of a genetic variant that is associated with a phenotype that is usually considered unfavorable, and at the same time with increased fitness in the evolutionary sense."

4. Selection against the "genes of education" is not limited to selection against educated people

The next question that the authors tried to solve concerns the mechanism of POLYEDU's influence on fitness. Since it is known that POLYEDU affects education, and education negatively affects childbearing, it is logical to assume that the negative impact of POLYEDU on reproduction is entirely due to the positive impact of POLYEDU on education (EA). To check whether this is the case, the authors conducted a joint analysis of the influence of two factors (POLYEDU and EA) on reproductive performance. It turned out that EA affects women's reproductive success in about the same way as POLYEDU, reducing the total number of children (by 0.045 per year of study) and shifting the beginning of reproduction to a later age (by 4.2 months per year of study). However, the effects of these two factors cannot be reduced to each other: they act largely independently. If we calculate the effect of POLYEDU on fitness adjusted for EA, this effect is only slightly reduced, remaining highly reliable. For example, if we take only women who have received 10 years of education (this is a mandatory minimum in Iceland), then within this sample we will see that women with high POLYEDU gave birth to fewer children than women with low POLYEDU - despite the fact that everyone has the same level of education. Thus, genes that increase the propensity to receive education reduce fitness, regardless of whether a person has realized this propensity or not. The same thing happens if we calculate the effect of EA on fitness adjusted for POLYEDU: education reduces fitness regardless of how pronounced a person's genetic predisposition to education is.

In men, the negative effect of POLYEDU on fitness also cannot be explained by the influence of POLYEDU on education. Moreover, it turned out that education in itself does not reduce, but even slightly increases the reproductive success of men! Every extra year of education adds an average of 0.011 children to an Icelandic man. Thus, the genetic propensity for education reduces the fitness of men, despite the fact that education itself increases reproductive success.

It follows directly from this that "education genes", in addition to their influence on EA, have other phenotypic effects that negatively affect fitness. What are these effects?

The authors found that POLYEDU positively correlates with life expectancy, as well as with a number of parameters related to physical health. In particular, people with high POLYEDU have low triglyceride levels in the blood, are less likely to be obese and smoke less. In addition, POLYEDU positively correlates with intelligence indicators, such as the ability to solve problems from IQ tests. The positive correlation between intelligence, education and longevity of the "Elements" has been told before.

It is quite difficult to imagine the negative impact of physical health on reproduction. Rather, it's about intelligence. For example, we can assume that smart people plan their lives and try not to have children too early. This is the simplest and most obvious mechanism by which intelligence can reduce fitness and promote natural selection for stupidity. Since intelligence positively correlates with physical health, such selection can simultaneously make us weaker and more painful.

5. The genetic basis of education is really deteriorating

Thus, the study showed that POLYEDU negatively correlates with Darwinian fitness. This means that natural selection works against the "genes of education". Consequently, the frequency of occurrence of these genes in the gene pool of the population should decrease over time. The authors tested this by comparing the average POLYEDU values of Icelanders born in different years. It turned out that the frequency of occurrence of "education genes" really decreased during the entire period of time studied. This can be seen by the steady decline in the average POLYEDU value (Fig. 3). Thus, negative selection by "education genes" is quite effective: these genes are gradually culled from the gene pool. The observed rate of POLYEDU decline is approximately the same as expected based on the fertility data shown in Fig. 1.

Fig. 3. The average POLYEDU value among Icelanders, depending on the year of birth. The data is grouped by decades. A steady decline in POLYEDU has been seen for most of the 20th century.

The graph in Fig. 3 seems to show that the decline of POLYEDU is gradually slowing down (the blue line in the figure). However, additional calculations have shown that this slowdown is an artifact associated with the fact that people with high POLYEDU live longer on average. As the cohort ages, the average POLYEDU value in it increases due to the selective mortality of people with low POLYEDU. This should not have much effect on young cohorts, so the most adequate idea of the dynamics of the process seems to be given by the red trend line in Fig. 3.

6. On the way to "idiocracy"

Since POLYEDU positively correlates with IQ, negative selection by "education genes" should lead to population stupefaction. To assess the scale of stupidity, the authors made a plausible assumption: they suggested that those genes affecting education whose effects are not taken into account by the POLYEDU index affect fitness and IQ in the same way as genes whose effects are reflected in POLYEDU. In this case, it turns out that the deterioration of the genetic basis of IQ under the influence of negative selection by the genes of education should lead to a decrease in the average IQ of the population by 0.3 points per decade. If such selection continues for several centuries in a row, the consequences will be very tangible.

However, real indicators of people's intelligence are currently growing due to socio-cultural factors (see the Flynn Effect). The average rate of this growth from 1932 to 1978 was 3.0 points per decade, which more than covers the expected genetic degradation. But the Flynn effect has nothing to do with "intelligence genes" and biological evolution: the observed growth is too fast to try to link it with evolutionary genetics. It reflects social and cultural processes that contribute to the fuller disclosure of innate abilities for intellectual growth. Unfortunately, these abilities themselves degrade. For readers who will console themselves with the fact that all this concerns only Icelanders, I would advise them to read the articles referenced at the end of the news and think again. Unfortunately, the picture is quite complete and convincing.

We don't know how long the Flynn effect will work, but it's hardly worth counting on it in the long run. There is evidence indicating the weakening and even complete disappearance of the Flynn effect in some populations since the 1990s. However, we do not know how long the negative selection for education and intelligence will continue. After all, during anthropogenesis, the same genes were obviously subjected to positive selection, and no one knows how the direction of selection will change in the future.

Anyway, the study convincingly showed that Ronald Fisher's fears were not groundless. If the trends identified among Icelanders are also active in other countries (and the probability of this is very high) and if all this continues for several more centuries, then genetic degradation can become a serious problem for humanity. To resist it, it is necessary to develop science faster (while there is still someone). Of course (and fortunately), today we can no longer talk about the barbaric methods of artificial selection discussed by the founders of eugenics. But it makes sense to think about developing high-tech, effective and humane methods of correcting unfavorable evolutionary trends. Including methods related to genetic engineering, gene therapy and selection of gametes or early embryos.

A source:  Kong et al., Selection against variants in the genome associated with educational attainment // PNAS, 2017.

Several articles on the deterioration of genetic characteristics related to intelligence: 1) Jonathan P.
Beauchamp. Genetic evidence for natural selection in humans in the contemporary United States // PNAS. 2016. DOI: 10.1073/pnas.1600398113.
2) Michael A. Woodley of Menie. How fragile is our intellect? Estimating losses in general intelligence due to both selection and mutation accumulation // Personality and Individual Differences. 2015. V. 75. P. 80–84. DOI: 10.1016/j.paid.2014.10.047.
3) Gerald R. Crabtree. Our fragile intellect. Part I // Trends in Genetics. 2013.
4) Richard Lynn. New evidence for dysgenic fertility for intelligence in the United States // Social Biology. 1999. DOI: 10.1080/19485565.1999.9988992.
5) Vegard Skirbekk. Fertility trends by social status // Demographic Research. 2008. V. 18. P. 145–180. DOI: 10.4054/DemRes.2008.18.5.

Portal "Eternal youth" http://vechnayamolodost.ru  25.01.2017