Introduction

General cognitive ability, or general intelligence, may be the area in which we know the most from a behavior-genetic perspective. Bouchard and McGue (1981) noted over 140 studies of this domain, which yielded the largely consistent result that genetic differences account for approximately 50% of the observed variability in general cognitive ability. However, it was not until recently that we began to learn more about cognitive ability than this simple, albeit important, finding.

The conceptualization of general cognitive ability as variance shared by a number of tests of various specific abilities dates back to Spearman (1904). The development of general cognitive ability, conceptualized in this manner, from infancy through middle childhood, is the focus of this chapter. Longitudinal twin and sibling data can be used to address two aspects of the developmental process. The first aspect concerns the sources of observed variation in individual differences in general cognitive ability at each age of assessment. By comparison of correlations from identical (monozygotic) and fraternal (dizygotic) twins, as well as adoptive and biological siblings, behavior-genetic methodology allows us to partition the observed variation into variation due to differences between individuals in genetic makeup and environmental differences between individuals. These environmental differences can be further subdivided into those shared by members of the family and those nonshared environmental influences that are unique to individuals. We can then examine whether heritable and environmental contributions change across the developmental period in question.

In this chapter we report results from model-fitting analyses of data from the Colorado Adoption Project using the simple sibling and parent–off-spring models developed in the preceding chapter as well as extending the model to include age-to-age genetic correlations and to consider the multivariate case. As discussed in Chapter 7, model fitting has several advantages over less sophisticated approaches to the interpretation of behavioral genetic data: It yields appropriate parameter estimates given the assumptions of a model; it provides standard errors for these parameter estimates; and it provides goodness-of-fit tests to aid in the evaluation of alternative models.

Sibling model

A simple model can be used to represent the sibling adoption design because the essence of this design lies in the comparison between two correlations: the correlations for adoptive and nonadoptive siblings. The sibling model, illustrated in the path diagram of Figure 7.3, was applied to the adoptive- and nonadoptive-sibling correlations presented in Chapter 6. The model involves only three parameters: heritability and shared and nonshared environment. As specified in Equations (7.4) and (7.5), the model assumes that the observed nonadoptive-sibling correlation is a function of half the heritability of the trait and of shared environmental influence; the adoptive-sibling correlation arises only from shared environmental influence – in the absence of selective placement, heredity does not contribute to the resemblance of adoptive siblings.

In addition to the descriptive and predictive changes discussed in the preceding chapter, developmental change can be seen in terms of etiology – changes in genetic as well as environmental influences. Changes in environmental influences can be explored without behavioral genetics; for example, the effects of prematurity on individual differences in social and mental development tend to diminish during infancy (Kopp, 1983). Behavioral genetics, however, provides a particularly powerful and general approach to the study of developmental changes in etiologies of individual differences.

From a behavioral genetics perspective, three kinds of etiological change can be considered; these mirror the phenotypic changes described in Chapter 5. The most basic phenotypic change that can occur is a change in variance, although changes in the magnitude of phenotypic variance are difficult to interpret because measures are not comparable across ages. The analogous genetic concept – change in heritability – is easier to interpret because it refers to a proportion of phenotypic variance due to genetic differences among individuals rather than to the absolute magnitude of variance.

As mentioned in the preceding chapter, quantitative genetic theory recognizes that genetic and environmental influences change during development, and it proposes new concepts and methods for exploring developmental change as well as continuity. This is the core of a new subdiscipline, developmental behavioral genetics. When the field of behavioral genetics is surveyed from a developmental perspective, it is clear that the relative roles of genetic and environmental influences change during development (Plomin, 1986a). If this were not the case, there would be no need for the field of developmental behavioral genetics – the story in childhood would be just the same as that in adulthood.

The conclusion that the relative magnitudes of genetic and environmental influences change during development is founded primarily on cross-sectional comparisons across studies, for the obvious reason that most behavioral genetic studies are cross-sectional. Although the cross-sectional design can be illuminating, the lifeblood of developmental analysis of change and continuity is the longitudinal design (McCall, 1977; see also Chapter 5). The few longitudinal behavioral genetic studies, discussed below, add disproportionately to the weight of these conclusions because the same subjects are studied at different ages and, at each age, subjects are usually studied within a relatively narrow age band.

It is reasonable to expect that descriptive and explanatory relationships in development involve complex interactions rather than simple main effects. For example, an easy temperament might buffer a child against a difficult environment; conversely, stress may have a disproportionate effect on vulnerable children (Garmezy & Rutter, 1983). Organismic specificity in reaction to environments is one of the major hypotheses that emerges from a thorough review of early experience and human development:

Both from basic and applied data it has become increasingly clear that the relationship of early experience to development will be mediated by the nature of the organism on which the experience impinges. Unfortunately, virtually nothing is known about the specific organismic characteristics which mediate differential reactivity to the early environment. (Wachs & Gruen, 1982, p. 247)

In this chapter, we explore interactions using the CAP data in early childhood. The word “interaction” has many connotations, and it is important to be clear about its use. We limit our search for interactions to statistical interactions, the type of interaction typically derived in analysis of variance that involves the sum of squares remaining after main effects and within-cell variation is removed: “The phenomenon is well named. Interaction variations are those attributable not to either of two influences acting alone but to joint effects of the two acting together” (Guilford & Fruchter, 1973, p. 249).

Behavioral genetic theory, methods, and research provide a unique perspective on nature and nurture during infancy and early childhood, that is, on the genetic and environmental origins of individual differences in behavioral development. The words “nature” and “nurture” each have warm associations until they are brought together. One of our goals is to emphasize the conjunction “and” rather than the projective test provided by the dash in “nature–nurture” or the explicit hostility in the phrase “nature versus nurture.” We believe that the perspective of behavioral genetics is as useful for understanding environmental influences in development as it is for exploring the role of heredity, and we hope that this book will convince developmentalists of the importance of both genetic and experiential factors in the origins of behavioral differences during infancy and early childhood. At the simplest level, the components-of-variance approach – which we explore in terms of simple correlations as well as by means of model-fitting analyses – often indicates that genetic variance is significant and invariably shows that nongenetic factors are important.

The decomposition of phenotypic variance into genetic and environmental components of variance is the standard fare of behavioral genetic research. Somewhat newer is an emphasis on the decomposition of the environmental component of variance into two components, one shared by family members, which increases their phenotypic resemblance, and the other not shared; correlations for genetically unrelated children reared together in the same adoptive homes are especially powerful for detecting the “bottom line” influence of growing up in the same family.

Interest in as well as understanding and acceptance of behavioral genetics are often hindered by a single issue: confusion between individual differences and group differences. That is, behavioral genetic theory and research address individual differences (variance), whereas most psychological research involves group comparisons (means). In this chapter, we contrast these two approaches and then discuss the advantages and disadvantages of an individual-differences perspective. In part, the relative neglect of an individual-differences approach is due to its apparent atheoretical orientation. For this reason, the next chapter considers quantitative genetics as the basis for a general theory of the origins of individual differences.

The group-differences approach focuses on average differences, such as gender, age, cultural, or species differences, among groups of individuals within a population. In contrast, the etiology of differences among individuals in a population is the focus of individual-differences research. The point of this chapter is not that the individual-differences approach is better than the group-differences approach. The two approaches are perspectives, and perspectives are neither right nor wrong, only more or less useful for a particular purpose. However, we do argue that the two approaches differ in important ways that affect theories and research. In the following section, the basic distinction between the two approaches is examined more closely.

In this chapter we step back from the numerous analyses described in previous chapters in order to gain some perspective on what we know and what we need to learn about the origins of individual differences during infancy and early childhood. As presaged in Chapter 1, the ratio of what is known to what is not known is small; however, the ratio is more impressive when we consider how few studies have addressed the issue of the origins of individual differences in infancy and childhood.

Principles

Rather than summarizing the preceding chapters, we have attempted to abstract some principles that outline what is known about nature and nurture in infancy and early childhood. In our book on infancy, several principles were drawn from the infancy results of the Colorado Adoption Project (Plomin & DeFries, 1985a). These principles involve some general points, such as the following. The etiology of individual differences in infancy includes heredity, variations in family environment are related to individual differences in infancy, and the relative extent of genetic and environmental influence varies for different characters. We have no doubt that these general principles hold for early childhood as well as for infancy.

One other general principle should be added to our earlier list: Individual differences among children are substantial and reliable.

One reason for the relative disregard of individual differences in psychology is that research on this subject appears atheoretical and usually addresses correlation rather than causation. In this chapter, we suggest that quantitative genetics provides the basis for a general theory of the etiology of individual differences of scope and power rarely seen in the behavioral sciences. After a brief overview of quantitative genetics, we describe a general theory of individual differences in terms of 10 propositions and then consider the theory in the context of current trends in the philosophy of science.

We will not concern ourselves with the philosophical intricacies of the word “theory.” The term obviously means different things to different psychologists, as illustrated by formal differences among the best-known theories in psychology, such as learning theories, personality theories, and Piagetian theory. Nonetheless, from the pragmatic view of a behavioral researcher, theories should clarify our thinking by describing, predicting, and explaining behavior. At the very least, theories should be descriptive, organizing and condensing existing facts in a reasonable, internally consistent manner. However, they should also make predictions concerning phenomena not yet investigated and allow clear tests of these predictions to be made. At their best, theories explain phenomena as well as describe and predict them.

Genotype–environment interaction, the topic of the preceding chapter, denotes an interaction in the statistical, analysis-of-variance sense of a conditional relationship: The effect of environmental factors depends on genotype. In contrast, genotype–environment correlation literally refers to a correlation between genetic deviations and environmental deviations as they affect a particular trait. In terms of a 2 × 2 table depicting low versus high genotypes reared in low versus high environments, evidence for genotype–environment interaction is obtained from a comparison of cell means (e.g., cells 1 and 4 vs. cells 2 and 3). In contrast, genotype–environment correlation is indicated by the frequency of individuals in the cells (e.g., more children of “high genotype” are likely to experience the “high environment”). In other words, genotype–environment correlation describes the extent to which children are exposed to environments on the basis of their genetic propensities. For example, if shyness is heritable, children genetically predisposed toward shyness will have shy parents on the average who are likely to provide a “shy” environment for their children – that is, modeling shy behavior and providing relatively few opportunities for interactions with strangers. Such proclivities can be reinforced in interactions with nonfamily members: Reactions of unfamiliar children and adults to a shy child are unlikely to be rewarding or successful for the child, thus enhancing the child's tendencies toward shyness.

The full adoption design allows us to investigate the etiology of individual differences in behavior in a direct and straightforward manner. The design is both simple and powerful, and the summary statistics it yields provide a broad description of this etiology, as we have seen in the preceding chapter. The correlation between the behavior of the adopted child and that of its adoptive parents provides direct evidence of the importance of shared environment independent of inherited, or genetic, influences. The correlation between the behavior of the adopted child and that of the biological parents from whom the child has been separated since birth provides direct evidence of the importance of genetic influences independent of the home environment. In nonadoptive families these two influences are always confounded and there is no direct way to evaluate their relative importance.

Simple correlations estimated from adoption data provide a very broad description that, for many purposes, may be quite sufficient for an understanding of the etiology of individual differences. Alternatively, a model of transmission may be assumed that facilitates estimates of genetic and environmental transmission parameters. The adoptive-parent/adopted-child correlation provides an estimate of the proportion of variance due to shared environmental influences (c2), whereas the biological-parent/adopted-child correlation estimates one-half of heritability (h2).