He then further extends the analogy, asking us to presume that the lack of nutrients in the suboptimal environment is due to a careless laboratory worker who failed to add the proper amounts of nitrogen and zinc. A chemist is called in to diagnose the cause of the difference between the two groups and determines that one solution lacks sufficient nitrogen. The nitrogen is added, the experiment is repeated, and the difference is narrowed but not eliminated because the chemist did not test for differences in the trace quantities of zinc required for optimal growth. Lewontin's point here is that identifying some environmental causes and correcting them may not completely erase group differences until all environmental causes have been identified and corrected. Thus, failure to erase a gap through environmental intervention does not necessarily offer evidence of a genetic cause for that gap.
Lewontin's analogy with corn plants is a good visual representation of heritability, and it also represents the concept's origin. Heritability and methods for measuring it arose from plant and animal breeding, where its utility is very powerful. When heritability is high, plant and animal breeders can rapidly improve inherited characteristics through artificial selection because they are able to select individuals that are genetically most predisposed to inherit genes conferring the traits they wish to improve. When heritability is low, however, progress from artificial selection is slow and unproductive because environmental differences mask genetic differences. Therefore, much of modern plant and animal breeding is aimed at implementing measures to maximize heritability, usually through reducing environmental variation, such as planting experimental plants in environments that are as uniform as possible or providing experimental animals with adequate food in exactly the same rations for each individual. Plant- and animal-breeding experiments designed to measure heritability are usually highly controlled, with elaborate statistical and experimental designs to ensure accurate results.
By contrast, measuring heritability in humans is notoriously challenging. Researchers, for obvious ethical reasons, must not subject humans to highly controlled environments as they do for experimental plants and animals. And even if they were to do so, the results would have little meaning because heritability in the real world is dependent on actual environments, not experimental conditions. As Nisbett and his colleagues put it in their 2012 review:
The concept of heritability has its origins in animal [and plant] breeding, where variation in the genotype and environment is under the control of the experimenter, and under these conditions the concept has some real-world applications. In free-ranging humans, however, variability is uncontrolled, there is no “true” degree of variation to estimate, and heritability can take practically any value for any trait depending on the relative variability of genetic endowment and environment in the population being studied.36
Some of the most common estimates of heritability in humans are based on studies of identical twins who were separated at birth and raised apart. The idea here is that identical twins are genetically identical. Therefore, any differences between them are assumed to be entirely environmental (in other words, heritability is zero for identical-twin pairs). Measurement of how much two twins differ for any characteristic, therefore, offers an estimate of purely environmental variation between them. By combining these differences among multiple pairs of identical twins reared apart, researchers can derive an average estimate of environmental variation for the environments of these twin pairs without the confounding effects of genetic variation. If they then apply these estimates of environmental variations to people who do differ genetically but are raised in similar environments to the twins, they can derive estimates of heritability for the people who are genetically different. The assumption—which is key—is that any variation greater than that observed between twins must be genetic. In some cases, identical twin pairs are compared to fraternal twin pairs, who differ genetically as much as full siblings but have the comparative advantage of having shared the same womb and been born on the same day, like identical twins. Alternatively, researchers can compare variation among relatives with different degrees of genetic similarity—such as full siblings, half siblings, and cousins—who are raised in similar environments as a means of statistically estimating heritability.
Regarding such estimates of heritability, Herrnstein and Murray conclude that
the genetic component of IQ is unlikely to be smaller than 40 percent or higher than 80 percent. The most unambiguous direct estimates, based on identical twins raised apart, produce some of the highest estimates of heritability. For purposes of this discussion [differences between racial groups], we will adopt a middling estimate of 60 percent heritability, which, by extension, means that IQ is about 40 percent a matter of environment.37
Critics of The Bell Curve rightly point to this generalization of “a middling estimate of 60 percent heritability” as a gross misapplication of the concept. According to Nisbett and colleagues’ 2012 review of current research, “the heritability of intelligence test scores is apparently not constant across different races or socioeconomic classes,”38 reinforcing the long-standing caveat that heritability should not be generalized from one population to another.
Moreover, reliance on estimates of heritability based on environmental variation measured in identical twins has also been criticized because, even when twins are reared apart, environmental similarities may be confounded with the genetic identity of twins. For example, adopted children are typically raised in more affluent homes in environments that tend to be more uniform and less representative of overall environmental variation, thus resulting in potential overestimates of heritability on the basis of twin studies.
Also, the shared uterine environment of twins (identical or fraternal), which is unrepresentative of people who are not twins, can have a confounding effect on heritability estimates, even when identical and fraternal twins are compared. Identical twins typically share the same placenta, whereas fraternal twins have separate placentas, resulting in different uterine environments. The uterine environment is especially pertinent to measurements of IQ later in life, particularly if prenatal care is poor or if a birth mother has abused alcohol, tobacco, or illicit drugs during pregnancy because these factors can permanently affect brain development, and, in the case of twins, such factors may or may not affect the twins similarly, leading to misinterpretation about the degree of genetic influence on the brain.
Similar obstacles confront researchers who conduct nontwin studies for measuring heritability in humans. One of the most serious obstacles has been the inability to directly estimate genetic variation. Historically, researchers have estimated genetic variation by the degree of relatedness of subjects, such as comparing identical twins, full siblings (including fraternal twins), half siblings, and cousins to people who are unrelated. Such indirect measures of genetic variation are highly problematic, however. For instance, the degree of genetic variation between full siblings is not generalizable because it depends on how genetically different their lines of ancestry are. For instance, it is often said, as Herrnstein and Murray put it, that “full siblings share about 50 percent of genes.”39 Such a statement represents a seriously naïve understanding of human genetics. In reality, full siblings share about 50 percent of the variants that are heterozygous in each parent and 100 percent of the variants that are homozygous in each parent. The more diverse the lines of ancestry are for an individual's parents, the greater the degree of heterozygosity in the parents and the greater the degree of genetic variation in their children. Therefore, genetic variation among full siblings varies considerably from one family to another; full siblings whose parents have similar ancestry are genetically more similar to one another than full siblings whose parents have genetically diverse ancestry. The same holds true for half siblings, cousins, and other people who are genetically related. Differing degrees of genetic diversity from one family to another result in different heritabilities among biological full siblings or other relatives, further
dispelling the notion that heritabilities in humans can be generalized.
To their credit, many scientists who conduct research on heritability in humans recognize these and other limitations and clarify them when reporting research. Scientists often use complex and elaborate scientific and statistical methods to derive estimates of heritability that are as reliable as possible despite the large margins of error that are typical of heritability measures in humans. Most of these scientists are well-trained researchers who appropriately apply scientific methods to their work. Unfortunately, not everyone is as attentive; overly simplistic generalizations of heritability for IQ in humans often lead those who advocate hereditarian models to unsubstantiated conclusions that may be based more on political ideology than on scientific evidence.
Published studies on IQ and the interaction between genetic and environmental variation often report contrary conclusions. For instance, Herrnstein and Murray determined that, according to the studies they reviewed, socioeconomic status and shared environment (the fact that children raised in the same family share many of the same environmental influences) had little effect on heritability for IQ. Rushton and Jensen arrived at similar conclusions. By contrast, other researchers offer evidence that socioeconomic status and shared environment have a considerable effect on heritability for IQ. These latter studies often, though not always, portray a positive correlation between heritability and socioeconomic status, especially when shared environment is taken into account. Heritabilities tend to be highest in families in which parents are well educated or families of high socioeconomic status, whereas heritabilities are lowest for impoverished families and when parents have attained little education.40 However, yet other studies have led researchers to arrive at the opposite conclusion: heritabilities are highest in families of low socioeconomic status when compared to the more affluent.41
In many of these contradictory studies, there is nothing wrong with the science or the interpretations. In fact, it is no surprise that studies on heritability of IQ may show widely differing results. They often are conducted on different populations in different places with people whose degree of genetic relatedness differs, and with subjects of different age groups and educational backgrounds. They are excellent examples of the malleability of heritability estimates in humans. By focusing on a selected subset of studies, advocates of a particular point of view can assemble what seems to be ample evidence to support their preferred models, when, in fact, the real situation is highly complex, varied, and inconclusive.
The most reliable compilations of data are those that identify major trends among exceptionally large numbers of people over decades of time. One such trend is known as the Flynn effect, named after psychologist James R. Flynn of the University of Otago in New Zealand. His own words best describe it: “‘The Flynn Effect’ is the name that has become attached to an exciting development, namely, that the twentieth century saw massive IQ gains from one generation to another.”42
These gains are evident in essentially every country where IQ tests have been administered long enough to detect a trend—thirty countries as of 2012.43 In countries that had been modernized by the beginning of the twentieth century, the gains have averaged approximately three IQ points per decade.44 Less developed countries have also almost universally experienced gains.45 These gains, however, appear to be peaking in the most highly developed countries, such as Sweden. According to the 2012 review article by Nisbett and colleagues (one of whom is Flynn), “If Sweden represents the asymptote that we are likely to see for modern nations’ gains, the IQ gap between developing and developed nations could close by the end of the 21st century and falsify the hypothesis that some nations lack the intelligence to fully industrialize.”46
As psychologists and geneticists have pointed out, genetic changes over so short a period are insufficient to explain the Flynn effect, and, therefore, changing environments must be its principal cause. A number of environmental causes have been proposed, such as improvements in nutrition and education. Notably, the gains are not equivalent across the board but are, in most cases, greatest among groups of people who historically have scored lowest on IQ tests, and scores dramatically improve as economic and educational status improves. According to Nisbett and colleagues, “It seems likely that the ultimate cause of IQ gains is the Industrial Revolution, which produced a need for increased intellectual skills that modern societies somehow rose to meet.”47 Flynn, as a coauthor, agreed, but he also proposes a more specific explanation consistent with the Industrial Revolution and changes in education. He notes that gains are not evenly distributed across the subject areas that IQ tests measure but, instead, are greatest for analytical reasoning and pattern recognition and lowest for basic arithmetic. Education during the latter part of the twentieth century has placed greater emphasis on logic and abstract reasoning, mirroring the areas of IQ for which nations experienced the greatest gains.48
Perhaps the most serious deficiency for arguments supporting hereditarianism, especially racial hereditarianism, is the fact that nearly all measurements of genetic variation for IQ have been indirect. Genetic variation has been assigned essentially by default, under the presumption that any variation unexplained by environment must be genetic. One of the greatest advances in human genetics in recent years is the ability to measure genetic variation directly by tracking variants in DNA. In previous chapters, we've already seen how direct detection of variants in DNA can explain genetic variation for skin, hair, and eye pigmentation; inherited conditions; and susceptibility to a wide range of diseases, both infectious and noninfectious.
The ultimate confirmation of the underlying genetic influence on variation for any human trait is identification of causal variants in DNA. In the late 1990s, a number of psychologists expressed hope that genes and variants governing variation for intelligence would soon be identified.49 However, in spite of much research, identification of such genes and variants has proven elusive. Robert Plomin of King's College London has contributed some of the most significant research on variation for IQ, particularly through large-scale and long-term twin studies in the United Kingdom. In a 2013 article, Plomin laments that predictions made in the late 1990s regarding gene identification were “overly optimistic.”50 He also refers to the problem of missing heritability, defined as failure to confirm heritability estimates with DNA analysis:
Genetic designs such as the twin method would no longer be needed if it were possible to identify all of the genes responsible for heritability. However, it has proven more difficult than expected to identify genes for complex traits, including g, which has led to the refrain of “missing heritability.”51
During the first decade of the twenty-first century, a few genes and variants were identified as being associated with variation for intelligence. However, in 2011, a group of sixteen scientists representing institutions in the United States and Europe published a large-scale study that attempted to confirm these associations. They phrased the title of their article as a sentence, which aptly summarizes their main conclusion: “Most reported genetic associations with general intelligence are probably false positives.”52
Thus far, no single variant in any gene has been definitively associated with variation for IQ or g. However, when multiple variants are examined in the aggregate, a statistically detectable association with IQ and g has been identified. However, each variant appears to have no more than a very minor effect on variation for IQ or g. Only when many variants are examined in large numbers of individuals in the aggregate can a genetic influence on variation for IQ and g be detected.
Although large-scale studies on DNA variants and their relationship to IQ and g are relatively recent and few, a consensus on two major points appears to be emerging. First, variation for IQ and g is heritable to some degree, with heritability estimates differing among the populations studied.53 In this sense, direct estimates of genetic variation based on DNA studies are consistent with earlier studies relying on indirec
t estimates, which likewise produce varying heritability estimates. Second, no single gene or variant appears to have a major effect on variation for IQ or g. Instead, a large number of DNA variants, each with a very minor effect, combine in complex ways to influence variation for IQ and g. Adding to this complexity is the significant and often-changing influence of environmental variation.
Everyone Is African Page 13