Age changes in heritability are incompatible with Charney's theory

There is substantial evidence that heritability of cognitive ability increases with age (Deary et al. Reference Deary, Spinath and Bates2006; Haworth et al. Reference Haworth, Wright, Luciano, Martin, de Geus, van Beijsterveldt, Bartels, Posthuma, Boomsma, Davis, Kovas, Corley, DeFries, Hewitt, Olson, Rhea, Wadsworth, Iacono, McGue, Thompson, Hart, Petrill, Lubinski and Plomin2010), but Charney emphasizes that monozygotic (MZ) twins become increasingly unlike each other in terms of epigenetic profile as they get older, either because of stochastic errors in cell division or because of encountering different environments. Thus older twins are more discordant for epigenetic factors, but they are more concordant for IQ and a variety of other traits than younger twins. Indeed, IQ has relatively low heritability in early childhood, with linear increases into the young adult period (Haworth et al. Reference Haworth, Wright, Luciano, Martin, de Geus, van Beijsterveldt, Bartels, Posthuma, Boomsma, Davis, Kovas, Corley, DeFries, Hewitt, Olson, Rhea, Wadsworth, Iacono, McGue, Thompson, Hart, Petrill, Lubinski and Plomin2010). Thus the relatively greater genetic discordance of older MZ twins for epigenetic factors does not make them less alike in IQ; rather they become more alike.

Charney argues that perinatal stress makes MZ twins more similar. But being concordant for epigenetic factors due to prenatal stress does not increase MZ concordance for IQ early in life when concordance for IQ is relatively low; and relative discordance for epigenetic factors in adulthood does not decrease concordance for IQ in adulthood. These results are clearly independent of any greater similarity MZ twins may have because they are exposed to more stress in utero – Charney's suggested explanation for MZ twin similarity.

Phylogenetic adaptations and contingent alternate strategies

As Charney notes, a great many epigenetic events are random or are linked to pathology. As indicated by the IQ example above, Charney does not present a case that there are important epigenetic effects on adaptive traits like cognitive ability or personality.

Many of the processes highlighted by Charney are stochastic. But phylogenetic adaptations reliably arise across the range of environments normally encountered by a given species. Like genetic point mutations, most of these stochastic events are likely to be maladaptive or neutral. The development of adaptations requires the smooth meshing of genes. Thus it is not surprising that many epigenetic events are linked with pathology.

However, some adaptations function as contingent strategies in which genes are turned on or off depending on environmental triggers (e.g., as a result of maternal influence). Behavior-genetic studies of contingent adaptive strategies should result in evidence for substantial shared environmental influence and low heritability if mothers treat offspring the same, as indicated by the mouse maternal licking studies. Charney challenges “the principle of minimal shared maternal effects.” However, recent studies of attachment – a central developmental construct (e.g., Sroufe et al. Reference Sroufe, Egeland, Carlson and Collins2005) – show strong effects of shared maternal environment (Bakermans-Kranenburg et al. Reference Bakermans-Kranenburg, van Uzendoorn, Bokhorst and Schuengel2004; Bokhorst et al. Reference Bokhorst, Bakermans-Kranenburg, Pasco Fearon, van Ijzendoorn, Fonagy and Schuengel2003; O'Connor & Croft Reference O'Connor and Croft2001; Pasco Fearon et al. Reference Pasco Fearon, van Ijzendoorn, Fonagy, Bakermans-Kranenburg, Schuengel and Bokhorst2006; Roisman & Fraley Reference Roisman and Fraley2008). For example, Roisman & Fraley found that shared environment explained 53% of the variance in attachment security, unshared environment explained 30%, and the remaining 17% were due to additive genetic variance.

Behavior-genetic models can thus be quite informative on variation resulting from environmental programming of adaptive systems. Moreover, lack of maternal effects would be surprising given evolutionary and life history perspectives on the importance of maternal care, particularly in mammals. Research indicates large intercorrelations between markers of high-quality environments (including secure attachment, delayed maturation, low fertility) and adaptive outcomes in children, with parenting variables accounting for 20%–50% of the variance in child outcomes (Maccoby Reference Maccoby2000).

Finally, minimal maternal effects are logically inconsistent with behavioral data on parent–child interaction showing bidirectional influence between child and parent. If the child's behavior shapes the parent's behavior, how is it possible that the parent's behavior has no effect on the child?

This type of adaptive phenotypic plasticity makes great sense to us as evolutionists. Epigenetic processes grant the genome greater flexibility than a rigid DNA code. Its great adaptive advantage stems from its sensitivity to fluctuating environmental conditions such as the availability of food. Nature and nurture in concert shape developmental pathways and outcomes, resulting in a “blurring of boundaries” between genes and environment.

G × E Interactions are unlikely design features of complex adaptations

Charney continues in the tradition of earlier critics of behavior genetics who emphasize the possibility of extensive G × E interactions (Gottlieb Reference Gottlieb1997; Meaney Reference Meaney2010; Wahlsten Reference Wahlsten1990). Currently all G × E interactions that have been identified involve single genes that have multiple variants and are linked with pathology. These findings provide clear cases in which one allele is less functional than the more common allele, predisposing people carrying the allele towards pathological outcomes in normal environments (e.g., the PKU gene or the alanine allele associated with diabetes). However, caution should be exercised in extrapolating these findings to complex polygenic traits such as IQ, where no such G × E interactions have ever been identified, despite repeated attempts to do so. Whereas genotype-environment correlation (Cov[G,E]) results in maximizing the fit between organisms to environments, G×E interactions actually imply a genetic load, as there is selection against some variants in some normal environments (MacDonald & Hershberger Reference MacDonald, Hershberger, Burgess and MacDonald2005).

In general, findings support the importance of additive genes. For example, Hill et al. (Reference Hill, Goddard and Visscher2008) summarized data from animal and human genetics indicating that for fitness-related traits, typically around 50% of the phenotypic variation is due to additive genetic variation and that about 80% of genetic variation is additive. Additive genes have their effects on a wide range of normal genetic backgrounds and across a wide range of normal environments, thus fitting easily into the architecture of complex adaptations. The presence of complex, unpredictable, or idiosyncratic interactions would make it very difficult for natural selection to construct complex adaptations.

Despite this, it remains true that some genes may produce G×E interactions important for psychiatry and medicine because they result in pathology in some environments. The point here is that such genes are not likely to be part of the story of normal development of complex adaptations in the Environment of Evolutionary Adaptedness or even in the vast majority of contemporary environments.