The Paradox of Intelligence: Heritability and Malleability Coexist in Hidden Gene-Environment Interplay (2024)

The publisher's final edited version of this article is available at Psychol Bull

2.1. What heritability is and what it is not

Genetic variation is a major determinant of most traits, and this maxim has been described as the “first law of behavior genetics” (Turkheimer, 2000). Ronald Fisher, one of the fathers of both population genetics and modern statistics, invented in the 1920s the now ubiquitous analysis of variance (ANOVA) as a way to break down the total variation of any trait into genetic and non-genetic components. Using this or other related statistical methods, we can describe the simple split of variance in the following way:

VP is the total variance (the squared deviation from the mean value) of the phenotype (or trait) in the population. Note that VP here is split into genetic factors and non-genetic factors: Vg is the variance associated with genetic differences in the population, and Ve is the variance of everything else, including shared environmental effects, unique environmental effects, epigenetic effects, and errors. Unfortunately, Ronald Fisher misleadingly described Ve as the “environmental variance,” while it is really the total variance left over after explaining a portion of the phenotypic variance with the genetic model (Templeton, 2006). (Later, we discuss the importance of this seemingly subtle conceptual difference.) From this, we can then represent heritability mathematically as Vg/VP. Or in plain English, heritability is the proportion of genetic variation in a trait in relation to the total variation. The more variation Vg can “capture” during the split, the higher the estimate of heritability.

While seemingly straightforward, in the natural environment there are many complications associated with the calculation of heritability. Often, we do not know which genes to focus on, and it is expensive (and interpretatively complex) to scan for the whole genome. In addition, we do not know the ways in which genes work together to produce variations in a trait. Due to these complexities, both human and nonhuman animal studies have historically used an indirect approach to estimate heritability: genetic relatedness as a proxy for genes per se. (Genetic relatedness is easier to measure because while we cannot specify the genes behind a trait, we typically can identify daughters, uncles, siblings, or other relatives.) The problem with this trick is that individuals who are related also tend to share environments. Do cousins have the same temperament because they are genetically similar or because they grew up with similar family traditions? For non-human animals raised under controlled conditions, researchers can split “genetic relatedness” from “shared environment” by distributing family members with known genetic variations into distinct environments (or the same environment for all different families, which accomplishes the same goal). For humans, we look for cases where these splits happened naturally, such as cases of adoption, siblings raised apart, and families with twins.

For the present purposes, we can limit our discussion now to the “twin method”, a strategy that is frequently used to estimate heritability in humans. The strategy exploited by twin studies is to compare monozygotic, “identical” twins (henceforth called “MZ twins”) to dizygotic, “fraternal” twins (henceforth called “DZ twins”). MZ twins share approximately 100% of genetic material, while DZ twins share on average 50%, just like any other siblings. The assumption of twin studies is that the shared family environment experienced by MZ and DZ twins will be relatively similar, while only the shared genetic material will differ. Therefore, the discrepancy between how “similar” MZ and DZ twins are can be used to estimate how heritable a trait is. For IQ scores, if the values among MZ twins correlate more than the values among DZ twins, this discrepancy in correlation is inferred to be the consequence of genetic influences.

Here is a simplified formula to calculate heritability based on differences between MZ and DZ twins:

• Heritability = 2 (rMZ − rDZ)

• rMZ = the average correlation of a measured trait among MZ twins

• rDZ = the average correlation of a measured trait among DZ twins

Since the difference in the resemblance of MZ and DZ twins is due to the difference in sharing either 100% or 50% of genes, the difference between rMZ and rDZ is multiplied by 2, yielding an index of the proportion of the trait that is heritable. (For more on the twin method and other models for estimating heritability in humans, see ).

There are a number of concerns one can raise about the estimates of heritability obtained from twin studies. MZ twins usually share a placenta, while DZ twins never do, and this may cause different patterns of gene expression (Gordon et al., 2012). And of course MZ twins may be treated differently than DZ twins, or might be encouraged to make more similar decisions in life (Kendler et al., 1993; ). But these often-discussed caveats (e.g., Eysenck, 1979) do not form the crux of our concerns here. The problems we have in mind are more substantive in nature, and they hinder people’s appreciation of the malleability of intelligence as well as its gene-environment interplay. There is first a problem of interpretation: Heritability is not the equivalent of genetic determination. Heritability is an estimate of the causes of differences in a trait among people (a population statistic), whereas genetic determination is a matter of what causes a trait to be expressed in an individual. Heritability is a measure of what is associated with variation in the trait (its causes of variation), and not a measure of what causes the trait. Second, there is a problem sometimes overlooked even by experts: estimates of heritability (owing to biases in the underlying calculations) grossly underestimate gene-environment interplay.

In some instances, genetically determined traits are associated with very low estimates of heritability. For example, the number of fingers on a human hand is genetically determined: the genes related to this trait code for instructions that lead to five fingers in almost everyone in any normal environment. However, the heritability of number of fingers in humans is very low. That is because the vast majority of variation in finger number is purely environmental, with traumatic amputations and prenatal complications being the leading causes (while genetic coding for other than five fingers is rare in humans). In contrast, traits that are not genetically determined can sometimes be highly heritable. For example, the tendency to vote in an election has a heritability of 0.53 (Fowler et al., 2008), and the affiliation to Democrat (liberal) vs. Republican (conservative) parties in the US has a heritability of 0.46 (Settle et al., 2009). If this strikes the reader as odd, remember: A heritable characteristic is not one that is passed down in the genes, and neither is political affiliation, at least not in any sense where this term has explanatory power. In fact, there might be dozens of correlations and interactions between genes and environmental factors such as wealth, education, and choices in life that contribute to one’s political affiliation. Unfortunately, typical measures of heritability grossly underestimate GE interplay, as we explore next.

2.2. First dibs to genetic effects: Why heritability can underestimate GE interplay

The construct of “general intelligence” (or the g factor) captures the remarkable phenomenon that performance across diverse cognitive tasks are positively correlated. In other words, people who do well in one task tend to do well in many other tasks. This principle is a well-supported by decades of evidence, and is a central feature of many (if not the majority of) modern theories of intelligence (Mackintosh, 2011). At first glance, the existence of general intelligence might suggest that one “thing” explains much of the variability in intelligence. Such a casual assumption would severely limit our appreciation of the basis for intelligence. It could bias us into a straightforward genetic view where there is no space (nor need) for a multitude of correlations and interactions with the environment. Although general intelligence is a single, abstract construct, it needn’t be a single, concrete property of the brain. Instead, the higher-order construct of g may emerge from the interplay between a multitude of systems, including genetic networks, influences from psychological traits like working memory capacity, attentional capacity, motivation and personality, and reciprocal relations between cognitive and environmental processes (as held, for example, by on the “mutualism” model of intelligence originally proposed in van der Maas et al., 2006). Intelligence is a hugely complex trait, and we should expect a high number of moving parts behind differences across individuals.

At this point, it is important to define “GE interplay”. GE interplay denotes both correlations as well as interactions between genes and environment. Gene-environment correlation means that individuals with particular genotypes for a trait are more likely to experience particular relevant environments. Meanwhile, a gene-environment interaction means that there is a difference in the effect of a given relevant environment on individuals depending on their genotypes for that trait. These effects will be explored in more detail below.

While historically ignored (or at least minimized), the study of GE interplay as a cause of variation in a trait is starting to gain momentum in our current scientific age of big data, complex networks, and systems biology (; ; Lazer et al., 2009; ; Rockman, 2008). Given their importance, we should recognize where GE interplays are captured in the estimation of heritability. The following formula is representative of a relatively complete model of heritability (and one that is used extensively in contemporary analyses):

As in the simpler formula described above, VP here is again the total variance of the phenotype (trait) in the population. Genetic variance (formerly Vg) is generally broken down into two parts: the variance associated with additive genetic variation (Ga), and the variance associated with gene-gene interactions (G×G). Additive genetic variation is also known as “independent” genetic variation because it is the portion transmissible from a single parent to the offspring, no matter what the other parent contributes. Ec is the variance associated with shared environments in the population. (In human studies, it is the environment shared by siblings in the same family.) Making up what we call “GE interplay”, rGE is the variance associated with gene-environment correlations, while the G×E is the variance associated with gene-environment interactions¹. Finally, Ee is everything else (now a smaller “everything else” compared to the simpler equation described above), which includes the variance associated with “unique” environments (in human studies, these are aspects of the environment that differ among siblings, such as being a firstborn versus a later-born child), personal decisions (like when one sibling takes the initiative and “decides” to follow a cognitively challenging line of study), epigenetic effects (the regulation of DNA transcription without alteration of the original sequence, through mechanisms like DNA methylation, histone modifications and non-coding RNAs), and error.

Independent environmental effects (Ec and Ee) as well as GE interplay (rGE and G×E) are the determinants of a trait’s “malleability”. The more the environment causes (directly or indirectly) the variation in a trait, the higher the trait’s malleability. Keep in mind, however, that even if a trait is not particularly malleable, interventions are still possible, albeit much harder to detect and/or implement. If there is no typical environment (within the normally encountered range of environments) that affects the trait, there may still be special environments (e.g., one created in response to our knowledge of the underlying biology) which can affect it. For example, the highly heritable intellectual disability that arises from phenylketonuria can be avoided if the carrier maintains a special diet that would not typically be encountered ().

As we discussed above, in the simple equation VP = Vg + Ve, Fisher defined Ve as the leftover variance (residual variance) that was not explained by Vg. Similarly, indirect effects such as genetic interactions, gene-environment correlations, and gene-environment interactions are also relegated to the residual genetic variance left over after subtracting off the additive genetic variance (Templeton, 2006). Because additive genetic effects (Ga) are accounted for before rGE and G×E, any contribution from genes to the variation in a trait will be overestimated as the direct effect from genes (Templeton, 2006). And, of course, any contribution from the environment to the variation in a trait will be underestimated because its fair share of the variance was already appropriated by Ga. In other words, Fisher designed the quantitative genetic parameters in his model in such a way that (as we like to say) first dibs went to genetic effects. This is a problem intrinsic to partitioning variance under any classic statistical tool (such as ANOVA or simple regression): Variance (outcomes) with multiple correlated sources (predictors) can be mispresented as belonging to the source first/better specified in the analysis. In the case of intelligence, while researchers can precisely specify genetic relatedness, they do not understand, posit, and/or search for all of the rGE and G×E sources, and so default to the assumption that the variance comes from G. As pointed by Tenesa and Haley (2013), lack of modeling frequently leads to an overestimation of genetic effects. And this problem is not limited to twin studies. Any model (e.g., the ACED model for the estimation of heritability) where GE interplay is not specified, there is a high risk of GE becoming incorporated into additive genetic effects, which of course bias the estimates of heritability ().

What exactly are additive, independent genetic effects (Ga)? Does it imply that a single isolated gene produces the trait? In any literal sense, this cannot be true. If you put a strand of DNA double helix in a tube and wait, no distinguishable trait will emerge from it. And what about the GE interplay of gene-environment correlations (rGE) and gene-environment interactions (G×E)? Do these parameters merely refer to the evident truth that an egg needs both genes and environment to create an adult chicken? The answer is “no”. As described above, the effects that these parameters capture are related to causes of variation, not deterministic causation. Additivity or independence is conceptually analogous to what a statistician would describe as a “main effect”: it is how much a predictor/input variable (a gene) relates to a predicted/output variable (a trait) when averaging across the other predictor variables (environment and other genes). An rGE effect is a statistical correlation between two predictor variables. In turn, a G×E effect is conceptually close to a statistical interaction: it is when the effect of a variable (gene) with another variable (an environmental factor) has multiplicative consequences, with the result (variation in a trait) being more than the sum of its parts².

Now that we understand what additive effects and GE interplay mean conceptually, it is useful to explore rGE and G×E more concretely.

2.3. The iceberg of hidden interactions on evolutionarily critical traits

As we have seen, the hidden nature of rGE in the variation of intelligence comes from our inability to specify factors that contribute to active and reactive correlations. The hidden nature of G×E reflects, like for rGE, a similar lack of knowledge on the modulation of intelligence. But in addition to that, as we see now, the search for G×E is made even harder due to a probable “hidden iceberg” in the past and current evolution of intelligence. Evidence from evolutionary genetics strongly suggests that traits related to survival and reproduction (like intelligence) have a large reservoir of hidden variance in the form of interactions (). This type of diversity comes from both gene-gene and gene-environment interactions, and is generally referred to as “hidden” variation because it has the potential to affect a trait, but is not expressed under typical/current conditions (). These interactions capture genetic variation and accumulate mutations that stay latent for long periods of time, since natural selection is “blind” to anything other than additive, independent effects (). And since additive genetic variation is what fuels evolution by natural selection, important traits like intelligence might only continue to evolve at a fast pace because of constant new G×G and G×E hidden effects. A useful analogy to represent this is the iceberg. The tip of the iceberg (the independent genetic variance) is the only type of variance that can be seen by natural selection. And it is also the only effect that is not “residual” in most methods in quantitative genetics, like the ones seen for heritability here. Just as we see only the tip of the iceberg, the effects of interactions cannot be seen by natural selection and are much harder to detect during empirical measurements.

A recent meta-analysis on interventions to increase intelligence in humans (Protzko, 2015) suggested that “under increased demands from the environment (due to an intervention) we can raise IQ. Once those demands are removed, the system adapts to the new, reduced demands. Protzko’s conclusions suggest that G×E interactions influence IQ in a way similar to muscular mass, where some people are more genetically prone to build muscle, but that tendency also continuously interacts with circ*mstances: if you play in a professional league, your environment will buff you further (until you get fired or retire), and if are an astronaut in the ISS, your environment will lead to muscular atrophy (until you return to Earth). So, one could ask: What are the advantages of being weak? And why aren’t all humans “blessed” with hypertrophied muscles cells like gorillas? No one knows for sure, but it is not hard to imagine how our ecology and history might have sometimes benefited weak humans. Or, better, humans that could match their strength according to their needs in a period of a few months (thus conserving resources). Similarly, one could ask: What are the advantages of having a low IQ? There might be many. And maybe intelligence’s plasticity is useful to match its environment as to save precious resources from our energy-hungry brain.³

In fact, evolution is not only a part of our argument here on the hidden iceberg. The t present article is also inspired by recent changes in the field of evolutionary biology. Some psychologists, and more so behavior geneticists, claim that there is nothing new about the recent evidence on GE interplay as it relates to the interpretation of what heritability means and to the interpretation of the broader questions regarding the mixing of nature and nurture in the development of complex traits. We disagree, and are immediately reminded of an analogy in the field of evolutionary biology. Studies of G×G interactions were only recently possible, as were the assessment of epigenetic effects, niche construction, and interacting phenotypes (such as in the case for social evolution, seen in ). These findings and ideas make up today what modern researchers call The Extended Evolutionary Synthesis, which provides a new framework to think about and understand evolutionary phenomena that differs from the gene-centric conception that dominated evolutionary thinking for almost a century (Pigliucci, 2007). The extended evolutionary synthesis revisits some neglected and/or unknown factors at play in evolution, including developmental processes, the role of epigenetic inheritance and plasticity, as well as networks of interactions between genes and environment (Danchin et al., 2011; Laland et al., 2015). In sum, the goal of the new synthesis is, in Danchin et al., (2011) words: “to go beyond DNA in order to build a broader conception of evolution”.

Something analogous, we imagine, could be in the making for the fields studying the genetics of psychological (and other complex) traits, and GE interplay might be critical to it. Intelligence is a prime candidate, since this trait presents a unique case of established moderate-to-high heritability while at the same time exhibiting remarkable malleability. Intelligence has potential for gigantic GE interplay, although the way this interplay manifests itself in the variation in IQ has contributed to its earlier dismissal (as we discuss in the next section).

We can conclude so far that there is not only a common problem with the interpretation of heritability’s meaning (genetic determination versus genetic cause of variation), but heritability estimates themselves overlook indirect environmental/genetic effects. Furthermore, intelligence is a trait with a huge potential for gene-environment correlations and interaction given its vast psychological and biological complexity. But how do we know that the environment or GE interplay exerts a central influence on the variation of intelligence? In fact, a multitude of empirical evidence indicates that intelligence has a substantial malleability coexisting with its high heritability.

Table 1

Principal evidence for the malleability of intelligence by gene-environment interplay (i.e., observations that cannot be accommodated by an “all gene” or “all environment” interpretation). Unless noted, IQ values represent means. “See pages” denotes page numbers in the present manuscript, and “n/a” indicates that the results were not discussed within this manuscript.

3.1. IQ gains from adoption and immigration

Adoption studies allow researchers to separate the effects of the environment and genes by comparing an adopted child with his/her biological siblings who were not adopted, or with other peers who were left behind. Being raised by different parents changes not only a child’s family environment, but also all factors contributed by the neighborhood, peer, and school environments. These often-drastic environmental changes make adoption studies a particularly powerful method to assess the malleability of intelligence. It should be reiterated that if the change in environment is small, evidence can be found to support the all-gene view of IQ. However, by their design, these studies cannot capture the potential for IQ’s malleability (though they are certainly important to show its limits and granularity). Although we will discuss such studies where appropriate, our emphasis here is on studies that illustrate the potential for gene-environment interplay.

In a French adoption study, Duyme et al. (1999) examined a group of 65 impoverished children adopted relatively late in life (between 4 and 6 years of age) who had an average IQ of 77 before adoption. When measured during adolescence, these adopted children showed significant gains in IQ, and the new values were significantly correlated with pre-adoption IQ (indicating that the measurements were reliable). The size of the gain was dependent on the socio-economic status (SES) of the adoptive families: an average gain of 7.7 IQ points in low SES adoptive families and 19.5 IQ points in high SES adoptive families. These IQ gains are far from trivial. To put them in perspective, a successful college graduate is, on average, 15 IQ points above the average of 100, and a child is considered “gifted” in the US if he/she is 30 IQ points above average.

Even in countries with very low inequality and relatively hom*ogeneous socio-cultural environment, changing from a relatively poor and uneducated family to a wealthier and educated family can substantially increase IQ. In a recent adoption study done in Sweden, Kendler et al. (2015) assessed the IQs of 436 pairs of separated siblings where at least one member was reared by biological parents and the other by adoptive parents. Adoption by parents with higher level of education was associated with a significant increase of 4.4 points in the child’s IQ in adulthood. Interestingly, the authors also found that in families with at least 2.5 steps higher education status than biological parents (i.e., the difference between no high school and some postsecondary education), the adopted-away siblings had 7.6 IQ points higher on average than their home-reared adopted siblings. On the other extreme, sibling sets in which the biological parental educational status was at least 2 steps higher than that of the adoptive parents, the adopted-away siblings had an IQ on average 3.8 points lower than their home-reared siblings (Kendler et al., 2015). This result suggests some role of GE interplay, as it seems that higher biological parental education (which is itself likely to correlate strongly with the biological parents’ IQ) was worse than higher adoptive parental education for stimulating intellectual development.

The studies above can be contrasted with a small study (n = 38) involving a full cross-fostering design by Capron and Duyme (1989). Like in the above, the authors here found that the mean IQ of children reared in upper SES homes was significantly superior to those reared in low SES homes. The IQ gains from high SES environments was 12 points; smaller than the 19.5 points from Duyme et al. (1999), but still remarkably high! However, children born to upper SES families but adopted early into lower SES families had mean IQ of 107.5 whereas the mean IQ of children born into low SES families but adopted early into high SES families had a mean IQ of only 103.6 (). One would expect the means to travel extensively in the opposite direction if the environment was as potent as it appears in aggregated data because these contrasting environments (top vs. bottom 13% of French society) are several standard deviations apart. So, while these data are compatible with GE interplay, it suggests that additive genetic effects might sometimes be strong enough to counteract improvement in SES.

In a meta-analysis, van Ijzendoorn et al. (2008) considered 75 studies (totaling more than 3800 children in 19 different countries) to compare the intellectual development of children living in orphanages to that of children living with adoptive families. On average, children growing up in orphanages had an IQ that was 16.5 points lower than their peers who were adopted. Not surprisingly, orphanages in countries with a higher Human Development Index (a combined measure of life expectancy, literacy, education, standards of living, and quality of life) had smaller detrimental effects on children’s intelligence (reduction of 11.9 IQ points) than countries with a lower Index (reduction of 21 IQ points). Also, children in orphanages with the most favorable caregiver-child ratio (maximally three children per caregiver) did not significantly differ from their adopted peers. These observations suggest that the typical orphanage has environmental conditions that are detrimental to the development of intelligence. In other words, environmental conditions related to orphanages are causes of variation in IQ.

Environmental effects on intelligence from international adoption are potentially much more powerful than those observed within a country or region. Economically prosperous countries can have up to 9X the GDP per capita of economically undeveloped nations. (By comparison, adoption within the US from a low SES family to a higher SES family is commonly associated with a 2–4X increase in family income.) The poorest 5% of the US population, for example, are richer than 60% of the world (Milanovic, 2013). In addition, there are many other environmental factors relevant to IQ that differ drastically across nations, such as educational opportunities, parental expectations and pressure, motivation, a culture of intellectualism versus anti-intellectualism, the availability of cognitively demanding jobs, etc.

Winick et al. (1975) examined 205 Korean orphans (all of the viable cases from a single adoption service from 1959–1967) who were adopted during early life by US parents, and divided cases into three categories (according to the conditions of the children before adoption): malnourished (n = 59), moderately nourished (n = 76), and a control of well-nourished children (n = 70). (Keep in mind that Korea as a whole was a poor and underdeveloped country at the time of the adoptions during the 1960s.) After at least six years with their American parents, the children were assessed for IQ. The mean IQ of the previously malnourished group was 102; the moderately nourished group, 106; and the well-nourished group, 112. Strikingly, the mean IQ of the children from the previously malnourished category was 10 to 40 points higher than the IQ of malnourished children living with their biological families in Korea or other poor populations (Galler, et al., 1983; Hertzig et al., 1972; Liu et al., 2003; S. A. Richardson, 1976). In a similar analysis, O’Connor et al. (2000) examined 111 children from Romania who were adopted (after the collapse of the Soviet Union) by families in the UK at the age of four. The authors found a considerable catch-up in children’s cognitive abilities from the time at the adoption to just two years later, at age six (although these adopted Romanian children were still slightly below the average IQ of adopted UK children.)

A meta-analysis of 62 studies from a multitude of countries (totaling 18,000 adopted children) found an average increase in IQ of 17.6 points within several years of adoption (van Ijzendoorn et al., 2005) – a remarkable cognitive gain over their biological, nonadopted siblings and their peers who stayed behind. The size of the IQ gains described in this meta-analysis is higher than typical for studies of adoption. That, we believe, is because the meta-analysis considered studies where children came from extremely low SES and were adopted by families in a developed country (so, as pointed out above, they experienced a more dramatic environmental change than would be typical for within-nation adoptions). Furthermore, no significant differences were observed between the ultimate IQs of the adopted children and their environmental siblings/peers. In other words, the new environment made the initially “dull” children climb up the IQ ladder as high as the typical child in their new environment.

While early adoption associated with immigration can have a dramatic positive impact on IQ, the effects of adult immigration (absent adoption) are less encouraging. Rindermann and Thompson (2016) compared immigrants to natives worldwide across a 10–15-year period. Rindermann and Thompson reported that generally, immigration was associated with a small drift in IQ toward that of the native inhabitants. Depending on the quality of the host country’s educational system and economic resources, increases in IQ of 1–4 points were typical, a level much lower than observed in typical adoption studies (Rindermann & Thompson, 2016). However, since most immigrants are not adopted (so would not obtain the benefits of adoption into higher SES families) and many are adults at the time of immigration, we might anticipate that the effects of immigration would be smaller than would accompany the subset of cases of immigration represented by adoption. Thus, Rindermann and Thompson’s analysis suggests that while immigration can itself impact IQ, age at the time of immigration and access to the resources of the host country are likely (as described above) to influence the degree of change in IQ.

Now we are left with a thorny question: How exactly can environmental factors explain the huge IQ gains associated with adoption? It is well established that the effect of shared environment component, Ec, on IQ is usually low. In an elegant study by Charles Murray, for example, he compared thousands of siblings who have grown up in the same home, with the same parents, but who have different IQs, and found little relevance of shared environment (Murray, 1998). More recent studies also estimate Ec to be low, fluctuating around 10–30%, (e.g., Johnson et al., 2010; Lyons et al., 2009). And therein lies the problem: in the current models of heritability, environmental factors would need a colossal change in order to create such massive variations in IQ. The gains seen in adoption studies tend to be around 15 IQ points, or one standard deviation. Therefore, we would need an environmental improvement of 5 standard deviations to support a corresponding increase of 15 IQ points. And if we take into account that some environmental factors do not change much across SES and countries, the required improvement in relevant environments might be over 6 standard deviations. That is the equivalent of a boy from the slums of the Bronx in NYC being adopted by the CEO of a Fortune 500 company (a 5-standard deviation change in US income distribution), or a girl from a lower-class family in Kenya immigrating to Norway (a 6-standard deviation change in global income distribution). Both examples are clearly too extreme to accommodate the typical case of adoption and immigration. To make matters worse, in most adoption studies that have measured it, shared environmental effects on IQ tend to fade by adulthood. Does that mean, then, that the environment has little room to be a relevant cause of variation in IQ? Yes, but only if we limit the consideration of environmental impact to Ec, or, in other words, only if we consider the independent effects of the (shared) environment. As we have seen, although it is recognized that gene and environment can be correlated, the assumption has been that it did not matter whether genes expressed themselves purely through biological mechanisms or through environmentally mediated paths. Genetics always got first dibs. However, as proposed by , G×E correlations “and the mechanism we believe causes it radically alter the implications of heritability estimates for the potential effects of environment on IQ”. To put it another way, the synergistic (whole-more-than-the-sum) effects of G×E interactions could easily explain the malleability of intelligence seen in adoption and immigration.

The evidence on intelligence’s malleability from adoption studies may appear to be in conflict with the common assertion that intelligence has highly heritability. As we saw, Plomin stated that “parents who don’t see their children after the first few hours of life are just as similar in terms of both weight and IQ to them after adolescence as are parents who reared their own kids. And adopted parents are zero similar!” (Wakefield, 2013). Similarly, psychologist Charles Locurto argued that standard adoption studies provide little evidence for the malleability of IQ (Locurto, 1990). He based his conclusion on the fact that the IQ of adopted individuals, even many years after the adoption, is still highly correlated with the biological parents (and, as we pointed above, the environmental influence of the adoptive family is drastically reduced). How can these patterns be reconciled with the massive IQ gains following adoption?

Indeed, substantial evidence shows that the correlations of adopted children with their biological parents are high, and the correlations with their adoptive family members are near zero by late adolescence (examples in Horn, 1983; Plomin et al., 1997). However, this pattern of correlations does not imply that the impact from the environment must be small. In fact, massive IQ changes consequent to adoption are entirely compatible with high IQ correlations with biological parents. Put more generally, the magnitude of a correlation is independent of changes in means. (While this is a mathematical truth, its implications can be easily overlooked when interpreting correlations.) A recent study explored this issue empirically by performing a longitudinal assessment of intelligence in thousands of twins, first at age 20, and later, at age 55 (Lyons et al., 2009). Using standard methods for separating the causes of variation between genetic components, shared environment, and unique environment, the authors were able to infer that the genetic component was responsible for most of the observed stability in IQ from age 20 to age 55 (71.3% of the longitudinal correlation was genetic). There was, however, a change of 10 IQ points in half the individuals, and a change of at least 20 points in one fifth of the individuals. The authors concluded that these massive changes were overwhelmingly (83.1% of the longitudinal correlation) due to aspects of the environment not shared by twins. Therefore, genetic factors were primarily responsible for stability, and environmental factors were primarily responsible for changes in the actual value of IQ. Does that mean, then, that genes are at least the main force for the stability of IQ? As discussed next, the answer is “it depends”. The very correlation between genetic effects and IQ (or, in other words, the heritability of IQ) also varies across lifespan and SES.

3.2. Heritability changes across lifespan and socio-economic status

The heritability of body weight is quite high at five years of age (heritability = 0.95), but it typically decreases across the lifespan, plateauing at approximately 0.60 by 50 years of age. This decline in heritability reflects the fact that while genes set some initial parameters for body weight, lifestyle choices have more dominant later influence. While genetics can determine the physiological predisposition for body fat, such influence is less pronounced as lifestyle choices accumulate with age. The changing heritability of body weight illustrates that this trait is not genetically determined, but rather, is quite malleable despite the initially strong genetic influence.

Like body weight, intelligence is also subject to a change in its heritability across the lifespan. However, the pattern of change for intelligence is quite different than that observed for body weight, and many other traits. Intelligence can be reliably quantified beginning at about age 4–5. In populations of this age, the heritability of IQ is estimated at approximately 0.22. By 16 years of age, the heritability of IQ is estimated to be 0.62. Even more striking, by age 50 (at which time the heritability of body weight has declined precipitously), the heritability of IQ is commonly estimated to be 0.80 (with estimates ranging as high as 0.90; Bouchard, 1997; Haworth et al., 2009). This increase in heritability is not simply an artifact of changes in our ability to measure IQ. While IQ can be difficult to measure accurately at age 4–5, its measurement at age 16 is as reliable as it is later in life. Instead, the increase in IQ’s heritability with age probably reflects an underlying role of GE interplay in creating IQ differences between individuals.

Gene-environment interplay may underlie the increase in IQ’s heritability in a way that may not be immediately intuitive. Importantly, the genome is largely established at birth, so a population is not gaining much genetic variation as it ages (except for age-dependent genes). This suggests that the increase in heritability across lifespan cannot be explained by pure changes in independent genetic effects. Alternatively, it is reasonable to expect that an individual’s intelligence influences that individual’s attraction toward a particular cognitive environment. Individuals with disparate intelligence are likely to find themselves pursuing very different cognitive challenges, while those with similar cognitive abilities are likely to gravitate to more similar cognitive challenges. Thus, those with similar IQs become more similar, while those with disparate IQs become less similar, with the net effect being an increase in the estimate of heritability. As this runaway process occurs (via the passive, active, and reactive rGE described above), genetic differences that underlie early differences in IQ can be amplified by the accumulation of cognitive challenges offered by different environments (; ). Children with slightly higher IQs end up mated with the environment (and choose an environment) that is appropriate for their cognitive abilities, which can in turn promote further gains in intelligence. Conversely, children with slightly lower IQs may gravitate toward less challenging environments, and may come to have relatively lower IQs as adults.

Part of the increase in heritability across lifespan could also be due to Ga×Ec interactions as well, and not only correlations. Interactions with the shared/familial environment are known to inflate estimates of Ga for a trait (). The Ec behind IQ variation is quite high during infancy (around 0.55) and early adolescence (around 0.30), but falls to single digits by ages around 17 to 23 (Flynn, 2016). So, during late teenage years and early adulthood, Ec could be channeled towards heritability by interacting with differences in genes to explain the sudden and quick raise in IQ’s heritability. And this process could be cumulative, as intelligence is expected to have a high number of G×E interactions later in life (as feedback loops and networks accumulate), and have fewer of them early in life (which as we have described is the opposite for what is expected for heritability in most traits). This theme is also discussed by Tucker-Drob and Briley (2014), and may have repercussions for the way we understand the genetic stability of intelligence. As Ga×Ec compound with age, the effects of shared environments become ever more tied to genotypic differences. The authors suggest that the accumulation of Ga×Ec “may also help to explain why the stability of the shared environment increases to such a high level. As recurrent objectively shared experiences increasingly differentiate individuals on the basis of their genes, it is possible that the only remaining shared environmental main effects are those that have resulted from particularly severe and lasting early environmental experiences that all humans respond similarly to” ().

It is important to note that Ga×Ec does not need to be completely cumulative. The reduction in Ec and the increase in heritability across age could, in part, be because as people age, both the environments as well as the genetic effects relevant to IQ get more diversified (due, respectively, to cognitively complexity of adult life, and to genes with late onset). That, in turn, can lead to more opportunities for Ga×Ec to happen. (Remember that Ec is not only about the factors related to the household, but also about factors related to school quality, friendship circles, socioeconomic status, etc.). Hence, if for a 6-year-old many of the Ec and Ga do not interact, by age 18 the number and complexity of Ec and Ga is such that there are many more opportunities for “collisions” in Ga×Ec. This Ga×Ec would be interpreted as Ga and would inflate estimates of heritability in adulthood. Caution is required here, though. Heritability reaches its peak much later in life, whereas EC has mostly faded by the early 20s. Hence, this non-cumulative Ga×Ec cannot in isolation explain all of IQ’s heritability increase across lifespan.

To summarize, because methods in quantitative genetics usually give first dibs to genetics, the rGE and G×E effects creating the IQ gap is counted as an independent genetic effect. Thus, the fact that IQ’s heritability increases with age is, counterintuitively, evidence of GE interplay, a process described by others to reflect the “transactional model” of heritability (Tucker-Drob et al., 2013).

Interestingly, the same pattern of changes in heritability that occurs across lifespan happens across socio-economic status. In a seminal study, Turkheimer et al. estimated genetic and environmental effects on IQ in 7-year-old twins in high and low SES families (). About 25% percent of the families sampled had incomes below the US poverty line (which comprises about 15% of the US population), and the median annual income of the sample was equivalent today to $37,000, which is moderately lower than the median income of the US population as whole ($52,000). Results from that study showed that among affluent families, most of IQ’s variation was associated with genetic variation, and almost none was associated with shared familial environment (heritability of 0.72, with the rest associated with unique environments). However, among the poorest families, the reverse was true: most of variation in IQ was associated with the shared familial environment, and little with genetic variation (heritability of 0.10). These results suggest that differences in family background matter more when that background is relatively impoverished. Subsequent twin studies have found a similar pattern. Tucker-Drob et al. measured cognitive abilities in infant twins, 25% of whom lived below the US poverty line (). At age 10 months, genes had almost zero effect on the variation in mental ability across all levels of SES, which is expected given what we have seen about changes in heritability across lifespan. At two years of age, children raised in high-SES homes (2 standard deviations above the sample’s mean SES) had genes accounting for nearly 50% of the variation in mental ability (with 35% of shared environment). On the other hand, 2-year-olds raised in low-SES homes (2 standard deviations below the mean SES) had genes accounting for less than 5% of the variation in mental ability (with 70% of shared environment). In a third study, Harden et al. assessed a sample of adolescent twins ()⁴. Among those from the lowest SES families (the lower 40^th percentile of US family income, equivalent to $40,000 today), genetic influences accounted for 39% of the variance in cognitive abilities (with 45% attributed to shared environment). Meanwhile, among the twins from the wealthiest families (94^th percentile of US income, equivalent to $200,000 today), genetic effects accounted for 55% of the variance in cognitive abilities, and 35% was attributable to shared environmental influences. The adolescents sampled were all takers of the National Merit Scholarship Qualifying Test, and so we expect very few to be living in extreme poverty. Because of that, Harden et al. (2007) concluded that “genotype-by-environment interactions in cognitive development are not limited to severely disadvantaged environments, as has been previously suggested.”

The results above suggest that environmental differences between low, middle, and high SES families influence the expression of genetic potential for intelligence. Differences in genes are more accentuated in favorable environments, while on the other extreme, differences in familial environment are strongest for IQ’s variation among poor families. In fact, Bronfenbrenner and Ceci (1994) predicted these results and the importance of GE interplay in their bioecological model a decade before any empirical support (see discussion above). This pattern was confirmed in a more recent meta-analysis (). Based on aggregated data from 11 studies that followed twin and adopted samples from birth to 18 years of age, Tucker-Drob et al., reported that in infancy, genes accounted for less than 25% of the variability in IQ, whereas the shared family environment accounted for approximately 60%. By late adolescence, this pattern had reversed, with genes accounting for 70% of the variability in IQ and the shared family environment accounting for near 0%. In the same paper, Tucker-Drob et al. traced a beautiful parallel between the meta-analysis results for heritability changes across life span with previous results (already detailed here) of heritability changes across socio economic status by Harden et al., 2007, and Tucker-Drob et al., 2011. In the parallel made by Tucker-Drob et al. 2013, the graph for heritability by lifespan and the graph for heritability by SES are remarkably similar! In both cases, the heritability of IQ is low precisely when GE is expected to be low (in young age due fewer relevant experiences, and in poor populations due to lack of opportunities), while heritability of IQ is high when GE is expected to be high (in old age due to accumulated experience, and in wealthy populations due to more opportunities). It’s hard to imagine how Ga alone could have affected the continuum of SES and lifespan in a similar manner to explain the same pattern of heritability changes. Independent genetic effects cannot be the only important way in which genes affect IQ’s heritability – G×E and/or rGE effects matter.

Common descriptions give the misimpression that IQ’s heritability is always high, and that the shared environment plays only a trivial role⁵. Given the data on IQ and SES discussed above, the opposite is more likely true. In a global scale, 80% of humans are below US’s and Europe’s poverty lines (for an in-depth and insightful analysis on global inequality, see Milanovic, 2013.), and so the shared environment is likely to be an extremely powerful cause of variation in intelligence in most populations. Moreover, it is plausible that these high values of the shared environment component reflect (the same way high values of the genetic component probably do) a high G×E (and/or rGE) contribution to the estimate of heritability.

3.3. The Flynn Effect

IQ scores are standardized, and so the average score of a population is necessarily 100. As tests are revised, the test maker designs each revision to maintain that average (based initially on large preliminary test samples). Consequently, if the population were to become “smarter”, the test maker would need to increase the test’s difficulty in order to maintain the average score of 100. This is exactly what occurred throughout the 20^th century, where in France for instance (where IQ scores were obtained for the entire population of 18-year-old males), IQ increased at least 20 points just between 1950 and 1980. Thus, although the average IQ score remained constant, the actual intelligence of test takers had increased. This phenomenon, referred to as the “Flynn Effect”, was first formally described in detail in 1984 by James Flynn (Flynn, 1984).

Although the Flynn effect is a worldwide phenomenon, it occurs predominantly in countries transitioning into what we consider today to be a “developed” society, both in social aspects like education and health access, as well as in economic aspects like per capita GDP and industrialization. In the United States, IQ increased by approximately 14 points between the years 1932 and 1978 (Flynn, 1984), and similar gains of three IQ points per decade were observed during the last century in France, Great Britain, the Netherlands, Australia, Canada, Germany, and Japan (). In recent decades, gains in IQ also began to emerge in developing countries such as Turkey, Sudan, and Dominica (Khaleefa et al., 2008; Meisenberg et al., 2005; Rindermann et al., 2013). Worldwide, a meta-analysis of 53 studies conducted in industrialized societies showed an increase of 17.6 IQ points occurred between 1951–2011, translating to an average increase of approximately 2.9 IQ points per decade (Trahan et al., 2014). These increases are far from trivial. According to the Wechsler IQ classification scheme, this increase translates to an equivalent shift (when comparing across the 20^th century) from “average” to “superior” intelligence.

Many explanations for the Flynn Effect have been considered, such as genetic heterosis (the “hybrid vigor” effect from miscegenation) (Mingroni, 2007), improvements in nutrition/health (Lynn, 2009), reduced pathogen stress (), reduced family size (), and test-takers’ familiarity with formal testing methods (Tuddenham, 1948). Although each of those factors may have some small explanatory value, they explain (both alone and in combination) only a small portion of the increase in intelligence throughout this era (Lynn, 2009; Woodley, 2011).

In contrast to the earlier attempts to explain this phenomenon, Dickens and Flynn () described a model with high explanatory value in which people’s IQs were shaped by both environment and genes, but in which environments were matched to individuals’ IQs. Dickens and Flynn argued that, as we have seen here, “heritability” includes both a direct effect of the genotype on IQ and also indirect effects such that the genotype interacts with and shapes the environment, and that these combined influences determine an individual’s ultimate cognitive capacity. That is, those with a greater IQ tend to seek stimulating environments that further increase IQ, and importantly, that more stimulating environments are available to those that reside in developed societies. These reciprocal effects result in GE interplay, or what Dickens & Flynn call “multiplier effects”. There are many changes that come with a country’s development that contribute to the Flynn Effect. James Flynn currently proposes that developed countries have more complex jobs, and have adopted a widespread, formal education focused in abstract and scientific thinking. To quote Flynn (Flynn, 2012):

“Our ancestors lived in a world that was concrete and utilitarian. In 1900, schoolchildren were asked, “What are the capitals of the 46 states?” Today they are asked, “If rural representatives dominated a state legislature, where would they put the capital?” (The answer is that, because they hate big cities, they would put the state capital [of New York] in Albany rather than New York City.) In other words, we take applying logic to hypothetical situations seriously.”

In a recent meta-analysis on the Flynn Effect, examined 271 independent samples spanning one century and four million participants from 31 countries. The authors used multiple meta-regressions on annual IQ test score changes, and included a wide range of predictor variables, including genetic miscegenation, blood lead levels, nutrition, pathogen stress, family size, test taking behavior, technology, education, and GE multipliers effects. After analyzing this extensive data set, the authors determined that IQ gains were most heavily associated with what they describe as hybrid factors (interacting biological and environmental factors), the most relevant of which were improvements of education, improvements in nutrition, and the GE multipliers proposed by Flynn (and discussed above). These results reinforce the conclusion that GE interplay is critical to the increases in intelligence that emerged across the 20^th Century.

3.4. Cognitive aging

In normative terms, IQ tends to be stable across the lifespan, e.g., an individual of average IQ at age 16 is likely to exhibit average IQ at age 90 (Deary, 2014). However, absolute performance on the IQ test declines precipitously as we age, such that by age 90, an individual will obtain a raw score (one that has not been age-normalized) that is comparable to a typical eight-year-old. This later effect (the impact of which can be observed across a multitude of cognitive tasks) is widely recognized and described as “cognitive aging” (Gerstorf et al., 2008; ). Numerous factors contribute to cognitive aging, some of which are related to the organic brain impairments that accumulate with age (Fischer et al., 2010; Kramer et al., 2004; ).

Though the general decline in cognitive capacity from aging seems to be ultimately unavoidable, it may be possible to slow the rate of decline. For example, it seems that improvements in health care and general fitness, as well as cognitive stimulation, play a greater role in cognitive function among the elderly than it does for younger individuals. In a rather unique study, Staff, et al, (2014) compared two groups from Scotland (the Aberdeen 1921 and 1936 Birth Cohort) who were born in different decades. When contrasting both groups at age 77, the later, more “modern” group outscored the earlier group by 16.5 IQ points, which was a significant increase relative to their IQ gap recorded in their teens. What might account for this dramatic difference between the two generations? It is not hard to imagine that the elderly in the more modern group had access to more modern medicine, had better dietary habits, and tended to exercise more often. Another plausible environmental cause of variation in the IQ of older people is, of course, improved technology and greater access to it.

In a recent study, Bordone et al. (2015) contrasted cognitive tests taken by Europeans at around 60 years of age during multiple time points between 2002 and 2012. They found that cognitive test scores of individuals in later years were higher compared to those from earlier years, a result that is predicted by the Flynn Effect. What is remarkable here was that the study also quantified the individual’s use of modern technology. The time frame that they considered (2002–2012) was one in which intense technological progress became widely available to the general population (note that the first “smart” phone was released in 2007). Indeed, test participants near the end of this time frame were significantly more likely to report use of modern technology such as PCs, the internet, and mobile phones. Even after controlling for sex, age, and education, the authors found that the adoption of modern technology explained as much as 54% of the observed increase in IQ among the more modern sample. The authors concluded that “Part of the Flynn effect observed for the older population in Europe is caused by an increasing level of cognitive stimulation brought forth by the permeation of everyday lives with cognitive challenges related to technological innovation and interactive media” (Bordone et al., 2015). In accordance with these results, a qualitative review suggested that computerized cognitive training programs (classic cognitive training tasks as well as neuropsychological software and video games) can delay cognitive decline of older adults (). Thus, although it is indisputable that aging is associated with a decline in general cognitive capacity, it is interesting to speculate that some of the age-related decline in cognitive capacity can be overcome by “cognitive training” and other environmental manipulations. While direct evidence with humans is still incomplete (Kramer et al., 2004), supporting experimental evidence has been obtained with laboratory animals (Markowska, 2002; ; Matzel et al., 2011; ; ).

3.5. Early Education and Cognitive Engagement

Early education and programs focused on cognitive stimulation in children can have dramatic effects on later academic success. Montessori pre-schools, for example, have a curriculum strongly based on developing executive functions, such as reducing disorder, impulsivity, and inattention, and promoting self-discipline, independent thinking, and orderliness. “Executive functions” (involving attention, impulse control, allocation of working memory, and orchestration of thought and action) are widely asserted to be core features of intelligence (). In one particular study, pre-school age children were randomly picked from a lottery to enter a Montessori public school and were compared over time to individuals who were also in the lottery but were not picked (). Upon completion of Kindergarten (age 5) and 6^th grade (age 12) Montessori students performed better on specific tests of executive functions than their peers who attended other schools during preschool. Furthermore, they received better grades in reading and math.

In a recent review, Diamond and Lee () looked at studies of children 4–12 years old who underwent intervention programs aimed at promoting executive functions, including progressively more challenging pre-school school curricula, computer- and non-computer games, and mindfulness training. The authors concluded that such training was overall associated with improved academic performance. Interestingly, participants of lower SES typically exhibited initially poorer executive functions, and these individuals benefited more from early intervention than those of initially higher SES or executive skills. This pattern is consistent with the idea we discussed before that the impact of G×E and rGE on intelligence depends on a cognitively challenging environment.

While the studies on early education have not explicitly assessed the impact of early cognitive engagement on later IQ, we have already seen here that IQ is highly predictive of functional cognitive-dependent outcomes such as years of education, academic grades, income, and career prestige (Gottfredson, 1998; Mackintosh, 2011). Thus, it is easy to imagine that one of the benefits of these early intervention programs is at least in part a reflection of their effects on IQ. As support for such a possibility, it is established that IQ is strongly related to measures of cognitive control, such as selective attention and the implementation of working memory (; ).

Based both on independent analyses and systematic reviews of available literature, the Nobel laureate in economics James Heckman has written extensively on the cognitive benefits of early education. In a summary of his analyses (https://heckmanequation.org/resource/invest-in-early-childhood-development-reduce-deficits-strengthen-the-economy/), Heckman concludes that

“the highest rate of return in childhood development comes from investing [in education] as early as possible, from birth to age five in disadvantaged families… Skills beget skills in a complementary and dynamic way. Those seeking to reduce deficits and strengthen the economy should make significant investments in early childhood education”.

Heckman’s conclusions, even though based as much on speculation as actual data, are consistent with what Flynn described earlier as “multiplier effects”, i.e., those with an initially greater IQ tend to seek stimulating environments that further increase IQ. The direct effect of the gene-environment interplay might initially be quite small, but the cycle of feedback could (over the course of a lifetime) ultimately lead to a large increase in the IQ of the population.

Government-sponsored early intervention programs such as Head Start have been the subject of ongoing criticisms, echoed in the popular media, that their benefits appear to dissipate within a year or two of the program’s termination (DeParle, 1993). In a recent meta-analysis of the results from 39 independent studies, Protzko (Protzko, 2015) concluded that while early (and relatively brief) environmental interventions (such as Head Start) do raise the IQ of otherwise impoverished individuals, these beneficial effects are not sustained after the intervention ends. While these results may not be immediately encouraging, keep in mind that after an intervention, the participant is likely to return to the cognitively impoverished environment from which he/she was selected. But, as Protzko writes “The interventions show a strong effect on IQ that does not immediately snap back but instead gradually fades over years”. Absent the opportunity to assimilate into an environment that is matched to their new cognitive capacity (a forced loss in gene-environment correlation), it would be difficult to maintain or amplify the initial benefits afforded by the early intervention. Thus, much like the inter-generational Flynn effect, increases in IQ might be amplified, or at least sustained, by greater access to opportunities that often are inequitably distributed. In simpler terms, the analysis of Protzko should not lead us to conclude that early intervention programs such as Head Start can have no long-term benefits. Rather, these results highlight the need to provide participants with continuing opportunities that would allow them to capitalize on what might otherwise be transient gains in cognitive abilities.

​

¹Here we are treating all types of rGE and G×E under the umbrella term of GE interplay. As noted in (), these two forms of GE are commonly intertwined in real life, and are difficult to assess separately (though some modern statistical methods can help). In addition, it is likely that from an evolutionary perspective, both rGE and G×E reflect adaptive effects of G on E as well as of E on G, so we should indeed expect both effects to contribute to individual differences (). And more to the main argument developed here, rGEs and G×Es are together a solution to explain why intelligence can possibly have high heritability as well as high malleability.

²Technically, the interaction in G×E is similar, but not the same as a statistical interaction. A statistical interaction will only occur when there is variation in both G and E (). Cases of potentially high G×E where an important environmental factor is the same to most people (e.g., exposure to pollen when studying the trait of hay fever) will result in no statistical interaction (). In addition, as we already noted, the interactions of G×E can be masked by main, independent effects.

³The reader might assume that there is convincing evidence that intelligence is highly stable (i.e., not plastic) across the life span (Deary, 2014). And indeed, that is the case. However, that stability does not mean that actual cognitive capacity does not change (an adult is clearly superior to a child). Rather, it means that an individual’s IQ relative to his/her age cohorts tend not to change with age. While this is an interesting fact, it says little about the malleability of intelligence or the presence of GE interplay. The relative stability of IQ allows for changes in environmental factors and of GE interplay in the same way that it needs to allow for changes in genetic factors (given that humans are not born already expressing all their genes). As we describe below, massive environmental manipulations can underlie similarly large changes in the IQ of some individuals, despite the fact the environments and IQs are relatively stable for most individuals.

⁴Note that quantitative conclusions based on the studies of Turkheimer et al. (2003) and Tucker-Drob et al. (2011) are somewhat complicated by a lack of detail regarding the range of incomes sampled, or the average incomes of the upper SES brackets. However, the reported results are qualitatively consistent with those reported by Hardin et al. (2007), and this latter study does not suffer from the same ambiguities.

⁵Some have argued that the relationship between environmental and phenotypical variance has been tested for, and those tests generally show little or no reason to suspect a G×E interaction. Were we to accept those earlier results at face value, they would be in clear conflict with the data reviewed above (as well as our broader conclusions). These arguments against G×E influences are based on what has been described as studies of “environmentality” (e.g., ; ; Thompson et al., 1993). To precisely perform such analyses, it would be necessary to quantify environmental variance. The difficulty arises in that while genetic variance (or its proxy, familial resemblance) can be precisely specified, estimates of environmental variance are at best qualitative. Environmental history is vague or unknown, it changes with time, the critical components of the environment are a matter of speculation, and any measurement of environmental variance would be “nominal” at best. Ronald Fisher was aware of this problem, and his revolutionary quantitative techniques were only able to split a trait’s variation by giving precedence to what could be known: the genetic effects (see Section 2.2 above). In this same tradition, rather than directly measure environmental variance, studies of environmentality compute the ratio of genetic and environmental covariance to the phenotypic correlation, yielding estimates of bivariate heritability and environmentality. In other words, this strategy assumes that the “left over” environmental variance is the inverse side of heritability. Thus, it is unavoidable that environmental effects will necessarily be low; all of the G×E interaction (which was not empirically determined) was already assigned to Vg!

• Bailey RC. Hereditarian scientific fallacies. Genetica. 1997;99(2–3):125–33. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/9463068. [PubMed] [Google Scholar]
• Bentley JP, Asselbergs FW, Coffey CS, Hebert PR, Moore JH, Hillege HL, van Gilst WH. Cardiovascular risk associated with interactions among polymorphisms in genes from the renin-angiotensin, bradykinin, and fibrinolytic systems. PLoS ONE. 2010;5(9):1–8. http://doi.org/10.1371/journal.pone.0012757. [PMC free article] [PubMed] [Google Scholar]
• Block N. How heritability misleads about race. Cognition. 1995;56(2):99–128. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/19226416. [PubMed] [Google Scholar]
• Bordone V, Scherbov S, Steiber N. Smarter every day: The deceleration of population ageing in terms of cognition. Intelligence. 2015;52:90–96. http://doi.org/10.1016/j.intell.2015.07.005. [Google Scholar]
• Bouchard TJ. IQ similarity in twins reared apart: findings and responses to critics. In: Sternberg RJ, Grigorenko EL, editors. Intelligence: Heredity and Environment. Cambridge University Press; 1997. [Google Scholar]
• Bouchard TJ. Genetic Influence on Human Psychological Traits. A Survey. Current Directions in Psychological Science. 2004;13(4):148–151. http://doi.org/10.1111/j.0963-7214.2004.00295.x. [Google Scholar]
• Bronfenbrenner U, Ceci SJ. Nature-nuture reconceptualized in developmental perspective: A bioecological model. Psychological Review. 1994;101(4):568–586. http://doi.org/10.1037/0033-295X.101.4.568. [PubMed] [Google Scholar]
• Butcher LM, Davis OSP, Craig IW, Plomin R. Genome-wide quantitative trait locus association scan of general cognitive ability using pooled DNA and 500K single nucleotide polymorphism microarrays. Genes, Brain and Behavior. 2008;7(4):435–446. http://doi.org/10.1111/j.1601-183X.2007.00368.x. [PMC free article] [PubMed] [Google Scholar]
• Capron C, Duyme M. Assessment of effects of socio-economic status on IQ in a full cross-fostering study. Nature. 1989;340(6234):552–554. http://doi.org/10.1038/340552a0. [Google Scholar]
• Chabris CF, Hebert BM, Benjamin DJ, Beauchamp J, Cesarini D, van der Loos M, Laibson D. Most Reported Genetic Associations With General Intelligence Are Probably False Positives. Psychological Science. 2012 http://doi.org/10.1177/0956797611435528. [PMC free article] [PubMed]
• Chandler CH, Chari S, Dworkin I. Does your gene need a background check? How genetic background impacts the analysis of mutations, genes, and evolution. Trends in Genetics. 2013;29(6):358–366. http://doi.org/10.1016/j.tig.2013.01.009. [PMC free article] [PubMed] [Google Scholar]
• Chaste P, Leboyer M. Autism risk factors: Genes, environment, and gene-environment interactions. Dialogues in Clinical Neuroscience. 2012;14(3):281–292. http://doi.org/10.2217/epi.11.19. [PMC free article] [PubMed] [Google Scholar]
• Cowan N, Fristoe NM, Elliott EM, Brunner RP, Saults JS. Scope of attention, control of attention, and intelligence in children and adults. Memory & Cognition. 2006;34(8):1754–1768. http://doi.org/10.3758/BF03195936. [PMC free article] [PubMed] [Google Scholar]
• Crossa J. From Genotype × Environment Interaction to Gene × Environment Interaction. Current Genomics. 2012;13(3):225–244. http://doi.org/10.2174/138920212800543066. [PMC free article] [PubMed] [Google Scholar]
• Danchin É, Charmantier A, Champagne FA, Mesoudi A, Pujol B, Blanchet S. Beyond DNA: integrating inclusive inheritance into an extended theory of evolution. Nature Reviews Genetics. 2011;12(7):475–486. http://doi.org/10.1038/nrg3028. [PubMed] [Google Scholar]
• Darabos C, White MJ, Graham BE, Leung DN, Williams SM, Moore JH. The multiscale backbone of the human phenotype network based on biological pathways. BioData Mining. 2014;7(1):1. http://doi.org/10.1186/1756-0381-7-1. [PMC free article] [PubMed] [Google Scholar]
• Deary IJ. The Stability of Intelligence From Childhood to Old Age. Current Directions in Psychological Science. 2014;23(4):239–245. http://doi.org/10.1177/0963721414536905. [Google Scholar]
• DeParle J. Sharp Criticism For Head Start, Even by Friends. The New York Times; New York: 1993. Retrieved from http://www.nytimes.com/1993/03/19/us/sharp-criticism-for-head-start-even-by-friends.html. [Google Scholar]
• DeYoung CG, Clark R. The gene in its natural habitat: The importance of gene–trait interactions. Development and Psychopathology. 2012;24(4):1307–1318. http://doi.org/10.1017/S0954579412000727. [PubMed] [Google Scholar]
• Diamond A, Lee K. Interventions shown to aid executive function development in children 4 to 12 years old. Science. 2011;333(6045):959–64. http://doi.org/10.1126/science.1204529. [PMC free article] [PubMed] [Google Scholar]
• Dick DM. Gene-Environment Interaction in Psychological Traits and Disorders. Annual Review of Clinical Psychology. 2011;7(1):383–409. http://doi.org/10.1146/annurev-clinpsy-032210-104518. [PMC free article] [PubMed] [Google Scholar]
• Dick DM, Agrawal A, Keller MC, Adkins A, Aliev F, Monroe S, Sher KJ. Candidate Gene-Environment Interaction Research: Reflections and Recommendations. Perspectives on Psychological Science. 2015;10(1):37–59. http://doi.org/10.1177/1745691614556682. [PMC free article] [PubMed] [Google Scholar]
• Dickens WT, Flynn JR. Heritability estimates versus large environmental effects: The IQ paradox resolved. Psychological Review. 2001;108(2):346–369. http://doi.org/10.1037/0033-295X.108.2.346. [PubMed] [Google Scholar]
• Docherty SJ, Kovas Y, Plomin R. Gene-environment interaction in the etiology of mathematical ability using SNP sets. Behavior Genetics. 2011;41(1):141–154. http://doi.org/10.1007/s10519-010-9405-6. [PMC free article] [PubMed] [Google Scholar]
• Duell EJ, Bracci PM, Moore JH, Burk RD, Kelsey KT, Holly EA. Detecting pathway-based gene-gene and gene-environment interactions in pancreatic cancer. Cancer Epidemiology Biomarkers and Prevention. 2008;17(6):1470–1479. http://doi.org/10.1158/1055-9965.EPI-07-2797. [PMC free article] [PubMed] [Google Scholar]
• Dutton E, Bakhiet SF, Essa YAS, Blahmar TA, Hakami SMA. A Negative Flynn Effect in Kuwait: The same effect as in Europe but with seemingly different causes. Personality and Individual Differences. 2017;114:69–72. http://doi.org/10.1016/j.paid.2017.03.060. [Google Scholar]
• Duyme M, Dumaret AC, Tomkiewicz S. How can we boost IQs of “dull children”?: A late adoption study. Proceedings of the National Academy of Sciences of the United States of America. 1999;96(15):8790–8794. http://doi.org/10.1073/pnas.96.15.8790. [PMC free article] [PubMed] [Google Scholar]
• Eichler EE, Flint J, Gibson G, Kong A, Leal SM, Moore JH, Nadeau JH. Missing heritability and strategies for finding the underlying causes of complex disease. Nature Reviews Genetics. 2010;11(6):446–450. http://doi.org/10.1038/nrg2809. [PMC free article] [PubMed] [Google Scholar]
• El-Soda M, Malosetti M, Zwaan BJ, Koornneef M, Aarts MGM. Genotype × environment interaction QTL mapping in plants: lessons from Arabidopsis. Trends in Plant Science. 2014;19(6):390–398. http://doi.org/10.1016/j.tplants.2014.01.001. [PubMed] [Google Scholar]
• Eppig C, Fincher CL, Thornhill R. Parasite prevalence and the worldwide distribution of cognitive ability. Proceedings Biological Sciences/The Royal Society. 2010;277(1701):3801–3808. http://doi.org/10.1098/rspb.2010.0973. [PMC free article] [PubMed] [Google Scholar]
• Fan R, Zhong M, Wang S, Zhang Y, Andrew A, Karagas M, Moore JH. Entropy-based information gain approaches to detect and to characterize gene-gene and gene-environment interactions/correlations of complex diseases. Genetic Epidemiology. 2011;35(7):706–721. http://doi.org/10.1002/gepi.20621. [PMC free article] [PubMed] [Google Scholar]
• Finkel D, Pedersen NL. Gene–Environment Interplay in Adulthood. Behavior Genetics. 2016;46(1):1–3. http://doi.org/10.1007/s10519-015-9775-x. [PubMed] [Google Scholar]
• Fischer H, Nyberg L, Karlsson S, Karlsson P, Brehmer Y, Rieckmann A, Bäckman L. Simulating neurocognitive aging: effects of a dopaminergic antagonist on brain activity during working memory. Biological Psychiatry. 2010;67(6):575–80. http://doi.org/10.1016/j.biopsych.2009.12.013. [PubMed] [Google Scholar]
• Flynn JR. The mean IQ of Americans: Massive gains 1932 to 1978. Psychological Bulletin. 1984;95(1):29–51. http://doi.org/10.1037/0033-2909.95.1.29. [Google Scholar]
• Flynn JR. Thinking in More Sophisticated Ways. The New York Times; 2012. Feb, Retrieved from http://www.nytimes.com/roomfordebate/2012/02/26/are-people-getting-dumber/thinking-in-more-sophisticated-ways. [Google Scholar]
• Flynn JR. Does Your Family Make You Smarter?: Nature, Nurture, and Human Autonomy. Cambridge University Press; 2016. [Google Scholar]
• Fowler JH, Baker LA, Dawes CT. Genetic Variation in Political Participation. The American Political Science Review. 2008;102(2):233–248. http://doi.org/10.2307/27644513. [Google Scholar]
• Galler JR, Ramsey F, Solimano G, Lowell WE, Mason E. The influence of early malnutrition on subsequent behavioral development. I. Degree of impairment in intellectual performance. Journal of the American Academy of Child Psychiatry. 1983;22(1):8–15. http://doi.org/10.1097/00004583-198301000-00002. [PubMed] [Google Scholar]
• Gerstorf D, Siedlecki KL, Tucker-Drob EM, Salthouse Ta. Executive Dysfunctions Across Adulthood: Measurement Properties and Correlates of the DEX Self-Report Questionnaire. Aging, Neuropsychology, and Cognition. 2008;15(4):424–445. http://doi.org/10.1080/13825580701640374. [PMC free article] [PubMed] [Google Scholar]
• Geschwind DH, Konopka G. Neuroscience in the era of functional genomics and systems biology. Nature. 2009;461(7266):908–15. http://doi.org/10.1038/nature08537. [PMC free article] [PubMed] [Google Scholar]
• Goes CP, Sauce B, Peripato AC. Milk ejection in mice LG/J × SM/J. Mammalian Genome. 2012;23(11–12):770–9. http://doi.org/10.1007/s00335-012-9431-1. [PubMed] [Google Scholar]
• Goodnight CJ. Epistasis and the Effect of Founder Events on the Additive Genetic Variance. Evolution. 1988;42(3):441. http://doi.org/10.2307/2409030. [PubMed] [Google Scholar]
• Gordon L, Joo JE, Powell JE, Ollikainen M, Novakovic B, Li X, Saffery R. Neonatal DNA methylation profile in human twins is specified by a complex interplay between intrauterine environmental and genetic factors, subject to tissue-specific influence. Genome Research. 2012;22(8):1395–1406. http://doi.org/10.1101/gr.136598.111. [PMC free article] [PubMed] [Google Scholar]
• Gottfredson LS. Mainstream science on intelligence: An editorial with 52 signatories, history, and bibliography. Intelligence. 1997;24(1):13–23. http://doi.org/10.1016/S0160-2896(97)90011-8. [Google Scholar]
• Gottfredson LS. Intelligence and the American ambivalence toward talent. Talent development IV: Proceedings. 1998:41–58. [Google Scholar]
• Hahn LW, Ritchie MD, Moore JH. Multifactor dimensionality reduction software for detecting gene-gene and gene-environment interactions. Bioinformatics. 2003;19(3):376–382. http://doi.org/10.1093/bioinformatics/btf869. [PubMed] [Google Scholar]
• Hallmayer J, Cleveland S, Torres A, Phillips J, Cohen B, Torigoe T, Risch N. Genetic Heritability and Shared Environmental Factors Among Twin Pairs With Autism. Archives of General Psychiatry. 2011;68(11):1095–1102. http://doi.org/10.1001/archgenpsychiatry.2011.76. [PMC free article] [PubMed] [Google Scholar]
• Hanley WB, Clarke JT, Schoonheyt W. Maternal phenylketonuria (PKU)–a review. Clinical Biochemistry. 1987;20(3):149–56. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/3308176. [PubMed] [Google Scholar]
• Harden KP, Turkheimer E, Loehlin JC. Genotype by environment interaction in adolescents’ cognitive aptitude. Behavior Genetics. 2007;37(2):273–83. http://doi.org/10.1007/s10519-006-9113-4. [PMC free article] [PubMed] [Google Scholar]
• Harlaar N, Butcher LM, Meaburn E, Sham P, Craig IW, Plomin R. A behavioural genomic analysis of DNA markers associated with general cognitive ability in 7-year-olds. Journal of Child Psychology and Psychiatry. 2005;46(10):1097–1107. http://doi.org/10.1111/j.1469-7610.2005.01515.x. [PubMed] [Google Scholar]
• Haworth CMa, Wright MJ, Luciano M, Martin NG, de Geus EJC, van Beijsterveldt CEM, Plomin R. The heritability of general cognitive ability increases linearly from childhood to young adulthood. Molecular Psychiatry. 2009:1–9. http://doi.org/10.1038/mp.2009.55. [PMC free article] [PubMed]
• Hayden EC. Ethics: Taboo genetics. Nature. 2013;502(7469):26–28. http://doi.org/10.1038/502026a. [PubMed] [Google Scholar]
• Hermisson J, Wagner GP. The population genetic theory of hidden variation and genetic robustness. Genetics. 2004;168(4):2271–2284. http://doi.org/10.1534/genetics.104.029173. [PMC free article] [PubMed] [Google Scholar]
• Hertzig ME, Birch HG, Richardson SA, Tizard J. Intellectual levels of school children severely malnourished during the first two years of life. Pediatrics. 1972;49(6):814–824. [PubMed] [Google Scholar]
• Horn JM. The Texas Adoption Project: adopted children and their intellectual resemblance to biological and adoptive parents. Child Development. 1983;54(2):268–275. [PubMed] [Google Scholar]
• Hyttinen V, Kaprio J, Kinnunen L, Koskenvuo M, Tuomilehto J. Genetic liability of type 1 diabetes and the onset age among 22, 650 young Finnish twin pairs: A nationwide follow-up study. Diabetes. 2003;52(4):1052–1055. http://doi.org/10.2337/diabetes.52.4.1052. [PubMed] [Google Scholar]
• Jablonka E. From replicators to heritably varying phenotypic traits: The extended phenotype revisited. Biology and Philosophy. 2004;19(3):353–375. http://doi.org/10.1023/B:BIPH.0000036112.02199.7b. [Google Scholar]
• Jensen AR. How much can we boost IQ and scholastic achievement? Harvard Educational Review. 1969;39(1):1–123. http://doi.org/10.1037/h0037827. [Google Scholar]
• Johnson W, Deary IJ, Silventoinen K, Tynelius P, Rasmussen F. Family background buys an education in Minnesota but not in Sweden. Psychological Science : A Journal of the American Psychological Society/APS. 2010;21(9):1266–1273. http://doi.org/10.1177/0956797610379233. [PMC free article] [PubMed] [Google Scholar]
• Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ. A test of the equal-environment assumption in twin studies of psychiatric illness. Behavior Genetics. 1993;23(1):21–27. http://doi.org/10.1007/BF01067551. [PubMed] [Google Scholar]
• Kendler KS, Turkheimer E, Ohlsson H, Sundquist J, Sundquist K. Family environment and the malleability of cognitive ability: A Swedish national home-reared and adopted-away cosibling control study. Proceedings of the National Academy of Sciences. 2015;112(15):4612–4617. http://doi.org/10.1073/pnas.1417106112. [PMC free article] [PubMed] [Google Scholar]
• Khaleefa O, Abdelwahid SB, Abdulradi F, Lynn R. The increase of intelligence in Sudan 1964–2006. Personality and Individual Differences. 2008;45:412–413. http://doi.org/10.1016/j.paid.2008.05.016. [Google Scholar]
• Kooke R, Johannes F, Wardenaar R, Becker F, Etcheverry M, Colot V, Keurentjes JJB. Epigenetic Basis of Morphological Variation and Phenotypic Plasticity in Arabidopsis thaliana. The Plant Cell Online. 2015;27(2):337–348. http://doi.org/10.1105/tpc.114.133025. [PMC free article] [PubMed] [Google Scholar]
• Kramer AF, Bherer L, Colcombe SJ, Dong W, Greenough WT. Environmental Influences on Cognitive and Brain Plasticity During Aging. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences. 2004;59(9):M940–M957. http://doi.org/10.1093/gerona/59.9.M940. [PubMed] [Google Scholar]
• Kueider AM, Parisi JM, Gross AL, Rebok GW. Computerized cognitive training with older adults: A systematic review. PLoS ONE. 2012;7(7) http://doi.org/10.1371/journal.pone.0040588. [PMC free article] [PubMed] [Google Scholar]
• Laland KN, Uller T, Feldman MW, Sterelny K, Müller GB, Moczek A, Odling-Smee J. The extended evolutionary synthesis: its structure, assumptions and predictions. Proceedings of the Royal Society B: Biological Sciences. 2015;282(1813):20151019. http://doi.org/10.1098/rspb.2015.1019. [PMC free article] [PubMed] [Google Scholar]
• Lathrope GM, Lalouel JM, Jacquard A. Path Analysis of Family Resemblance and Gene-Environment Interaction. Biometrics. 1984;40(3):611. http://doi.org/10.2307/2530905. [Google Scholar]
• Lazer D, Pentland A, Adamic L, Aral S, Barabasi AL, Brewer D, Van Alstyne M. Computational social science. Science. 2009;323(5915):721–723. http://doi.org/10.1126/science.1167742. [PMC free article] [PubMed] [Google Scholar]
• Le Rouzic A, Carlborg Ö. Evolutionary potential of hidden genetic variation. Trends in Ecology and Evolution. 2008;23(1):33–37. http://doi.org/10.1016/j.tree.2007.09.014. [PubMed] [Google Scholar]
• Lewontin RC. Further Remarks on Race and the Genetics of Intelligence. Bulletin of the Atomic Scientist. 1970;26(3):23–25. [Google Scholar]
• Lichtenstein P, Holm NV, Verkasalo PK, Iliadou A, Kaprio J, Koskenvuo M, Hemminki K. Environmental and Heritable Factors in the Causation of Cancer — Analyses of Cohorts of Twins from Sweden, Denmark, and Finland. New England Journal of Medicine. 2000;343(2):78–85. http://doi.org/10.1056/NEJM200007133430201. [PubMed] [Google Scholar]
• Lichtenstein P, Yip BH, Björk C, Pawitan Y, Cannon TD, Sullivan PF, Hultman CM. Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. The Lancet. 2009;373(9659):234–239. http://doi.org/10.1016/S0140-6736(09)60072-6. [PMC free article] [PubMed] [Google Scholar]
• Lillard A, Else-Quest N. The early years. Evaluating Montessori education. Science. 2006;313(5795):1893–4. http://doi.org/10.1126/science.1132362. [PubMed] [Google Scholar]
• Liu J, Raine A, Venables PH, Dalais C, Mednick SA. Malnutrition at age 3 years and lower cognitive ability at age 11 years: independence from psychosocial adversity. Archives of Pediatrics & Adolescent Medicine. 2003;157(6):593–600. http://doi.org/10.1001/archpedi.157.6.593.Malnutrition. [PMC free article] [PubMed] [Google Scholar]
• Locurto C. The malleability of IQ as judged from adoption studies. Intelligence. 1990;14(3):275–292. http://doi.org/10.1016/S0160-2896(10)80001-7. [Google Scholar]
• Loo SK, Shtir C, Doyle AE, Mick E, McGough JJ, McCracken J, Nelson SF. Genome-wide association study of intelligence: Additive effects of novel brain expressed genes. Journal of the American Academy of Child and Adolescent Psychiatry. 2012;51(4):432–440.e2. http://doi.org/10.1016/j.jaac.2012.01.006. [PubMed] [Google Scholar]
• Lykken DT, Bouchard TJ, McGue M, Tellegen A. Heritability of interests: A twin study. Journal of Applied Psychology. 1993;78(4):649–661. http://doi.org/10.1037//0021-9010.78.4.649. [PubMed] [Google Scholar]
• Lynn R. What has caused the Flynn effect? Secular increases in the Development Quotients of infants. Intelligence. 2009;37(1):16–24. http://doi.org/10.1016/j.intell.2008.07.008. [Google Scholar]
• Lynn R, Hampson S. The rise of national intelligence: Evidence from Britain, Japan and the U.S.A. Personality and Individual Differences. 1986;7(1):23–32. http://doi.org/10.1016/0191-8869(86)90104-2. [Google Scholar]
• Lynn R, Vanhanen T. National IQs: A review of their educational, cognitive, economic, political, demographic, sociological, epidemiological, geographic and climatic correlates. Intelligence. 2012;40(2):226–234. http://doi.org/10.1016/j.intell.2011.11.004. [Google Scholar]
• Lyons MJ, York TP, Franz CE, Grant MD, Eaves LJ, Jacobson KC, Kremen WS. Genes determine stability and the environment determines change in cognitive ability during 35 years of adulthood. Psychological Science. 2009;20(9):1146–1152. http://doi.org/10.1111/j.1467-9280.2009.02425.x. [PMC free article] [PubMed] [Google Scholar]
• Mackintosh N. IQ and Human Intelligence. Oxford; Oxford University Press; 2011. [Google Scholar]
• Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, Visscher PM. Finding the missing heritability of complex diseases. Nature. 2009;461(7265):747–753. http://doi.org/10.1038/nature08494. [PMC free article] [PubMed] [Google Scholar]
• Manuck SB, McCaffery JM. Gene-Environment Interaction. Annual Review of Psychology. 2014;65(1):41–70. http://doi.org/10.1146/annurev-psych-010213-115100. [PubMed] [Google Scholar]
• Markowska AL. Protective Effect of Practice on Cognition during Aging: Implications for Predictive Characteristics of Performance and Efficacy of Practice. Neurobiology of Learning and Memory. 2002;78(2):294–320. http://doi.org/10.1006/nlme.2002.4064. [PubMed] [Google Scholar]
• Markowska AL, Savonenko A. Retardation of cognitive aging by life-long diet restriction: implications for genetic variance. Neurobiology of Aging. 2002;23(1):75–86. http://doi.org/S0197458001002494 [pii] [PubMed] [Google Scholar]
• Matzel LD, Light K, Wass C, Colas-Zelin D, Denman-Brice A, Waddel AC, Kolata S. Learning & Memory. 5. Vol. 18. Cold Spring Harbor N.Y: 2011. Longitudinal attentional engagement rescues mice from age-related cognitive declines and cognitive inflexibility; pp. 345–56. http://doi.org/10.1101/lm.2034711. [PMC free article] [PubMed] [Google Scholar]
• Matzel LD, Wass C, Kolata S, Light K, Colas DC. Learning & Memory. 10. Vol. 16. Cold Spring Harbor N.Y: 2009. Age-related impairments of new memories reflect failures of learning, not retention; pp. 590–4. http://doi.org/10.1101/lm.1503209. [PMC free article] [PubMed] [Google Scholar]
• McGue M, Bacon S, Lykken DT. Personality stability and change in early adulthood: A behavioral genetic analysis. Developmental Psychology. 1993;29(1):96–109. http://doi.org/10.1037/0012-1649.29.1.96. [Google Scholar]
• Meaney MJ. Epigenetics and the biological definition of gene X environment interactions. Child Development. 2010;81(1):41–79. http://doi.org/10.1111/j.1467-8624.2009.01381.x. [PubMed] [Google Scholar]
• Meisenberg G, Lawless E, Lambert E, Newton A. The Flynn Effect in the Caribbean : Generational Change of Cognitive Test Performance in Dominica. The Mankind Quarterly. 2005;XLVI(1):29–69. [Google Scholar]
• Merilä J, Sheldon BC. Genetic architecture of fitness and nonfitness traits: empirical patterns and development of ideas. Heredity. 1999 Jun;83(Pt 2):103–9. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10469197. [PubMed] [Google Scholar]
• Milanovic B. Global Income Inequality in Numbers: in History and Now. Global Policy. 2013;4(2):198–208. http://doi.org/10.1111/1758-5899.12032. [Google Scholar]
• Miller EK, Cohen JD. An integrative theory of prefrontal cortex function. Annual Review of Neuroscience. 2001;24:167–202. http://doi.org/10.1146/annurev.neuro.24.1.167. [PubMed] [Google Scholar]
• Mingroni MA. Resolving the IQ paradox: Heterosis as a cause of the Flynn effect and other trends. Psychological Review. 2007;114(3):806–829. http://doi.org/10.1037/0033-295X.114.3.806. [PubMed] [Google Scholar]
• Moore AM, Brodie ED, III, Wolf JB. Interacting Phenotypes and the Evolutionary Process: I. Direct and Indirect Genetic Effects of Social Interactions. Evolution. 1997;51(5):1352–1362. [PubMed] [Google Scholar]
• Murray C. Income inequality and IQ. AEI Press; 1998. [Google Scholar]
• O’Connor TG, Rutter M, Beckett C, Keaveney L, Kreppner JM. The Effects of Global Severe Privation on Cognitive Competence: Extension and Longitudinal Follow-up. Child Development. 2000;71(2):376–390. http://doi.org/10.1111/1467-8624.00151. [PubMed] [Google Scholar]
• Ohta T. Slightly deleterious mutant substitutions in evolution. Nature. 1973;246(5428):96–8. http://doi.org/10.1038/246096a0. [PubMed] [Google Scholar]
• Philip Chen CL, Zhang CY. Data-intensive applications, challenges, techniques and technologies: A survey on Big Data. Information Sciences. 2014;275:314–347. http://doi.org/10.1016/j.ins.2014.01.015. [Google Scholar]
• Pietschnig J, Voracek M. One Century of Global IQ Gains: A Formal Meta-Analysis of the Flynn Effect (1909–2013) Perspectives on Psychological Science. 2015;10(3):282–306. http://doi.org/10.1177/1745691615577701. [PubMed] [Google Scholar]
• Pigliucci M. Do we need an extended evolutionary synthesis? Evolution; International Journal of Organic Evolution. 2007;61(12):2743–9. http://doi.org/10.1111/j.1558-5646.2007.00246.x. [PubMed] [Google Scholar]
• Plomin R, DeFries JC. Multivariate behavioral genetic analysis of twin data on scholastic abilities. Behavior Genetics. 1979;9(6):505–517. http://doi.org/10.1007/BF01067347. [PubMed] [Google Scholar]
• Plomin R, DeFries JC, Loehlin JC. Genotype-environment interaction and correlation in the analysis of human behavior. Psychological Bulletin. 1977;84(2):309–322. http://doi.org/10.1037/0033-2909.84.2.309. [PubMed] [Google Scholar]
• Plomin R, Fulker DW, Corley R, DeFries JC. Nature, Nurture, and Cognitive Development from 1 to 16 Years: A Parent-Offspring Adoption Study. Psychological Science. 1997;8(6):442–447. http://doi.org/10.1111/j.1467-9280.1997.tb00458.x. [Google Scholar]
• Protzko J. The environment in raising early intelligence: A meta-analysis of the fadeout effect. Intelligence. 2015;53:202–210. http://doi.org/10.1016/j.intell.2015.10.006. [Google Scholar]
• Purcell S. Variance Components Models for Gene–Environment Interaction in Twin Analysis. Twin Research. 2002;5(6):554–571. http://doi.org/10.1375/twin.5.6.554. [PubMed] [Google Scholar]
• Richardson K, Norgate S. The equal environments assumption of classical twin studies may not hold. The British Journal of Educational Psychology. 2005;75(Pt 3):339–350. http://doi.org/10.1348/000709904X24690. [PubMed] [Google Scholar]
• Richardson SA. The relation of severe malnutrition in infancy to the intelligence of school children with differing life histories. Pediatric Research. 1976;10(1):57–61. http://doi.org/10.1203/00006450-197601000-00011. [PubMed] [Google Scholar]
• Riemann R, Angleitner A, Strelau J. Genetic and environmental influences on personality: A study of twins teared together using the self- and peer report. Journal of Personality. 1997;65(3):449–475. http://doi.org/10.1111/j.1467-6494.1997.tb00324.x. [Google Scholar]
• Rietveld CA, Medland SE, Derringer J, Yang J, Esko T, Martin NW, Koellinger PD. GWAS of 126,559 Individuals Identifies Genetic Variants Associated with Educational Attainment. Science. 2013;340(6139):1467–1471. http://doi.org/10.1126/science.1235488. [PMC free article] [PubMed] [Google Scholar]
• Rindermann H, Schott T, Baumeister AEE. Flynn effect in Turkey: A comment on Kagitcibasi and Biricik (2011) Intelligence. 2013;41(3):178–180. http://doi.org/10.1016/j.intell.2013.02.003. [Google Scholar]
• Roberts RC. Behavior-genetic analysis. New York: McGraw-Hill; 1967. Some concepts and methods in quantitative genetics; pp. 214–257. [Google Scholar]
• Rockman MV. Reverse engineering the genotype-phenotype map with natural genetic variation. Nature. 2008;456(7223):738–44. http://doi.org/10.1038/nature07633. [PubMed] [Google Scholar]
• Rutter M. Developmental Catch-up, and Deficit, Following Adoption after Severe Global Early Privation. Journal of Child Psychology and Psychiatry. 1998;39(4) S0021963098002236. http://doi.org/10.1017/S0021963098002236. [PubMed] [Google Scholar]
• Rutter M, Silberg J. Gene-Environment Interplay in Relation to Emotional and Behavioral Disturbance. Annual Review of Psychology. 2002;53(1):463–490. http://doi.org/10.1146/annurev.psych.53.100901.135223. [PubMed] [Google Scholar]
• Salthouse TA, Ferrer-Caja E. What needs to be explained to account for age-related effects on multiple cognitive variables? Psychology and Aging. 2003;18(1):91–110. http://doi.org/10.1037/0882-7974.18.1.91. [PubMed] [Google Scholar]
• Sauce B, de Brito RA, Peripato AC. Genetic architecture of nest building in mice LG/J × SM/J. Frontiers in Genetics. 2012 May;3:90. http://doi.org/10.3389/fgene.2012.00090. [PMC free article] [PubMed] [Google Scholar]
• Schiff M, Duyme M, Dumaret A, Tomkiewicz S. How much could we boost scholastic achievement and IQ scores? A direct answer from a French adoption study. Cognition. 1982;12(2):165–196. http://doi.org/10.1016/0010-0277(82)90011-7. [PubMed] [Google Scholar]
• Settle JE, Dawes CT, Fowler JH. The Heritability of Partisan Attachment. Political Research Quarterly. 2009;62(3):601–613. http://doi.org/10.1177/1065912908327607. [Google Scholar]
• Shipstead Z, Engle RW. Interference within the focus of attention: Working memory tasks reflect more than temporary maintenance. Journal of Experimental Psychology: Learning, Memory, and Cognition. 2013;39(1):277–289. http://doi.org/10.1037/a0028467. [PubMed] [Google Scholar]
• Silventoinen K, Magnusson PKE, Tynelius P, Kaprio J, Rasmussen F. Heritability of body size and muscle strength in young adulthood: A study of one million Swedish men. Genetic Epidemiology. 2008;32(4):341–349. http://doi.org/10.1002/gepi.20308. [PubMed] [Google Scholar]
• Staff RT, Hogan MJ, Whalley LJ. Aging trajectories of fluid intelligence in late life: The influence of age, practice and childhood IQ on Raven’s Progressive Matrices. Intelligence. 2014;47:194–201. http://doi.org/10.1016/j.intell.2014.09.013. [Google Scholar]
• Sternberg RJ. Educating intelligence: Infusing the triarchic theory into school instruction. In: Sternberg RJ, Grigorenko E, editors. Intelligence, Heredity, and Environment. Cambridge University Press; 1997. [Google Scholar]
• Sullivan PF, Neale MC, Kendler KS. Genetic epidemiology of major depression: Review and meta-analysis. American Journal of Psychiatry. 2000;157(10):1552–1562. http://doi.org/10.1176/appi.ajp.157.10.1552. [PubMed] [Google Scholar]
• Sundet JM, Borren I, Tambs K. The Flynn effect is partly caused by changing fertility patterns. Intelligence. 2008;36(3):183–191. http://doi.org/10.1016/j.intell.2007.04.002. [Google Scholar]
• Tabery J. Beyond Versus: The Struggle to Understand the Interaction of Nature and Nurture. MIT Press; 2014. [Google Scholar]
• Templeton AR. Population Genetics and Microevolutionary Theory. Hoboken, New Jersey: Wiley-Liss; 2006. [Google Scholar]
• Tenesa A, Haley CS. The heritability of human disease: estimation, uses and abuses. Nature Reviews Genetics. 2013;14(2):139–149. http://doi.org/10.1038/nrg3377. [PubMed] [Google Scholar]
• Thompson LA, Detterman DK, Plomin R. Differences in heritability across groups differing in ability, revisited. Behavior Genetics. 1993;23(4):331–336. http://doi.org/10.1007/BF01067433. [PubMed] [Google Scholar]
• Trahan LH, Stuebing KK, Fletcher JM, Hisco*ck M. The Flynn effect: A meta-analysis. Psychological Bulletin. 2014;140(5):1332–1360. http://doi.org/10.1037/a0037173. [PMC free article] [PubMed] [Google Scholar]
• Tranter LJ, Koutstaal W. Age and Flexible Thinking: An Experimental Demonstration of the Beneficial Effects of Increased Cognitively Stimulating Activity on Fluid Intelligence in Healthy Older Adults. Aging, Neuropsychology, and Cognition. 2008;15(2):184–207. http://doi.org/10.1080/13825580701322163. [PubMed] [Google Scholar]
• Tucker-Drob EM, Briley Da. Continuity of genetic and environmental influences on cognition across the life span: a meta-analysis of longitudinal twin and adoption studies. Psychological Bulletin. 2014;140(4):949–79. http://doi.org/10.1037/a0035893. [PMC free article] [PubMed] [Google Scholar]
• Tucker-Drob EM, Briley Da, Harden KP. Genetic and Environmental Influences on Cognition Across Development and Context. Current Directions in Psychological Science. 2013;22(5):349–355. http://doi.org/10.1177/0963721413485087. [PMC free article] [PubMed] [Google Scholar]
• Tucker-Drob EM, Harden KP. Early childhood cognitive development and parental cognitive stimulation: Evidence for reciprocal gene-environment transactions. Developmental Science. 2012;15(2):250–259. http://doi.org/10.1111/j.1467-7687.2011.01121.x. [PMC free article] [PubMed] [Google Scholar]
• Tucker-Drob EM, Rhemtulla M, Harden KP, Turkheimer E, Fask D. Emergence of a Gene x Socioeconomic Status Interaction on Infant Mental Ability Between 10 Months and 2 Years. Psychological Science. 2011;22(1):125–133. http://doi.org/10.1177/0956797610392926. [PMC free article] [PubMed] [Google Scholar]
• Tuddenham RD. Soldier intelligence in World Wars I and II. American Psychologist. 1948;3(2):54–56. http://doi.org/10.1037/h0054962. [PubMed] [Google Scholar]
• Turkheimer E. Three Laws of Behavior Genetics and What They Mean. Current Directions in Psychological Science. 2000;9(5):160–164. http://doi.org/10.1111/1467-8721.00084. [Google Scholar]
• Turkheimer E, Haley A, Waldron M, D’Onofrio B, Gottesman II. Socioeconomic status modifies heritability of iq in young children. Psychological Science. 2003;14(6):623–628. http://doi.org/10.1046/j.0956-7976.2003.psci_1475.x. [PubMed] [Google Scholar]
• Tyler AL, Asselbergs FW, Williams SM, Moore JH. Shadows of complexity: What biological networks reveal about epistasis and pleiotropy. BioEssays. 2009;31(2):220–227. http://doi.org/10.1002/bies.200800022. [PMC free article] [PubMed] [Google Scholar]
• van der Maas HLJ, Dolan CV, Grasman RPPP, Wicherts JM, Huizenga HM, Raijmakers MEJ. A Dynamical Model of General Intelligence : The Positive Manifold of Intelligence by Mutualism. 2006;113(4):842–861. http://doi.org/10.1037/0033-295X.113.4.842. [PubMed] [Google Scholar]
• van Ijzendoorn MH, Juffer F, Poelhuis CWK. Adoption and cognitive development: a meta-analytic comparison of adopted and nonadopted children’s IQ and school performance. Psychological Bulletin. 2005;131(2):301–16. http://doi.org/10.1037/0033-2909.131.2.301. [PubMed] [Google Scholar]
• van Ijzendoorn MH, Luijk MPCM, Juffer F. IQ of Children Growing Up in Children’s Homes: A Meta-Analysis on IQ Delays in Orphanages. Merrill-Palmer Quarterly. 2008;54(3):341–366. http://doi.org/10.1353/mpq.0.0002. [Google Scholar]
• Wakefield M. Revealed: how exam results owe more to genes than teaching. The Spectator. 2013 Jul; Retrieved from http://www.spectator.co.uk/features/8970941/sorry-but-intelligence-really-is-in-the-genes/
• Wicherts JM, Dolan CV, van der Maas HLJ. A systematic literature review of the average IQ of sub-Saharan Africans. Intelligence. 2010;38(1):1–20. http://doi.org/10.1016/j.intell.2009.05.002. [Google Scholar]
• Winick M, Meyer K, Harris R. Malnutrition and environmental enrichment by early adoption. Science. 1975;190(4220):1173–1175. http://doi.org/10.1126/science.1198103. [PubMed] [Google Scholar]
• Woodley MA. Heterosis doesn’t cause the Flynn effect: A critical examination of Mingroni (2007) Psychological Review. 2011;118(4):689–693. http://doi.org/10.1037/a0024759. [PubMed] [Google Scholar]
• Zhao J, Li S, Wang L, Jiang L, Yang R, Cui Y. Genome-wide random regression analysis for parent-of-origin effects of body composition allometries in mouse. Scientific Reports. 2017 Feb;7:45191. http://doi.org/10.1038/srep45191. [PMC free article] [PubMed] [Google Scholar]
• Zuk O, Hechter E, Sunyaev SR, Lander ES. The mystery of missing heritability: Genetic interactions create phantom heritability. Proceedings of the National Academy of Sciences. 2012;109(4):1193–1198. http://doi.org/10.1073/pnas.1119675109. [PMC free article] [PubMed] [Google Scholar]