The Role of Gene-environment Interplay in the Development of Intelligence the Nature of Nurture

We present a review of empirical evidence that suggests that a substantial portion of phenotypic variance is due to non-linear (epigenetic) processes during ontogenesis. The role of such processes as a source of phenotypic variance in human behavior genetic studies is not fully appreciated. In addition to our review, we present simulation studies of nonlinear epigenetic variance using a computational model of neuronal network development. In each simulation study, time series for monozygotic (MZ) and dizygotic (DZ) twins were generated and analyzed using conventional behavior genetic modeling. In the results of these analyses, the nonlinear epigenetic variance was subsumed under the nonshared environmental component. As is commonly found in behavior genetic studies, observed heritabilities and unique environmentabilities increased with time, whereas common environmentabilities decreased. The fact that the phenotypic effects of nonlinear epigenetic processes appear as unsystematic variance in conventional twin analyses complicates the identification and quantification of the ultimate genetic and environmental causes of individual differences. We believe nonlinear dynamical system theories provide a challenging perspective on the development of individual differences, which may enrich behavior genetic studies. 2.1 Nonlinear Epigenetic Variance: Literature Review and Illustrative Simulation Study The nature-nurture question, as addressed in (behavior) genetic studies, concerns the decomposition in a linear statistical model of phenotypic (observed) variance into components attributable to latent factors. In humans, this decomposition is carried out in genetically informative designs, such as the twin design (e.g. Plomin, DeFries, McClearn, and McGuffin, 2008). Generally two general latent sources of variance are distinguished, namely the genetic and the environmental. Phenotypic variance attributable to the former may be decomposed into additive, dominance, and epistatic genetic variance, whereas phenotypic variance attributable to the latter can be further decomposed into shared environmental variance, and nonshared environmental variance, where shared influences are attributable to the environment cohabiting individuals (both human and infra-human) share. Regardless of the (genetically informative) design employed, it is assumed that genotypic and environmental factors reflect the underlying mechanisms causing phenotypic individual differences (e.g. Plomin et al., 2008).FF F An important characteristic of the behavior genetic methodology is that the partitioning of variance and the attendant causal interpretation pertains to phenotypic individual differences, and not to phenotypes themselves (Dolan & Molenaar, 1995; Lewontin, 1974; Oyama, 1985).F 2 F Indeed, there is an increasing complicating effects (Dick & Rose, 2002 F However, if we remain strictly within the domain of individual differences, a causal interpretation still requires caution. First, the validity of the interpretation may be undermined by the limitations of the statistical model employed to carry out the decomposition of phenotypic variance. These may concern the failure to take into account genotype-environment interaction, genotypic-environment covariance, assortative mating, etc. However, it should be noted that in principle, i.e., given a suitable design or given sufficient information, these effects can be modeled (e.g. Plomin et al.). effort to devise statistical models that can accommodate these ; Purcell, 2002; Eaves & Erkanli, 2003). At present, the issue 1 It is in this sense that we take the term “explained” seriously, i.e., genetic and environmental differences cause phenotypic differences. 2 What causes (say) cognitive functioning, and what causes individual differences in cognitive functioning are different questions (Oyama, 1985).