2 Genotype-Environment Correlation.

Paradoxically, the factors that make humans difficult to study genetically are precisely those that make humans so interesting. The experimental geneticist can control matings and randomize the uncontrolled environment. In many human societies, for better or for worse, consciously or unconsciously, people likely decide for themselves on the genotype of the partner to whom they are prepared to commit the future of their genes. Furthermore, humans are more or less free living organisms who spend a lot of time with their relatives. If the problem of mate selection gives rise to fascination with the complexities of assortative mating, it is the fact that individuals create their own environment and spend so much time with their relatives that generates the intriguing puzzle of genotype-environment correlation. As the term suggests, genotype-environment correlation (CorGE) refers to the fact that the environments that individuals experience may not be a random sample of the whole range of environments but may be caused by, or correlated with, their genes. Christopher Jencks (1972) spoke of the ``double advantage'' phenomenon in the context of ability and education. Individuals who begin life with the advantage of genes which increase their ability relative to the average may also be born into homes that provide them with more enriched environments, having more money to spend on books and education and being more committed to learning and teaching. This is an example of positive CorGE. Cattell (1963) raised the possibility of negative CorGE by formulating a principle of ``cultural coercion to the biosocial norm.'' According to this principle, which has much in common with the notion of stabilizing selection in population genetics, individuals whose genotype predisposes them to extreme behavior in either direction will tend to evoke a social response which will ``coerce'' them back towards the mean. For example, educational programs that are designed specifically for the average student may increase achievement in below average students while attenuating it in talented pupils. Many taxonomies have been proposed for CorGE. We prefer one that classifies CorGE according to specific detectable consequences for the pattern of variation in a population (see Eaves et al., 1977). The first type of CorGE, genotype-environment autocorrelation arises because the individual creates or evokes environments which are functions of his or her genotype. This is the ``smorgasbord'' model which views a given culture as having a wide variety of environments from which the individual makes a selection on the basis of genetically determined preferences. Thus, an intellectually gifted individual would invest more time in mentally stimulating activities. An example of possible CorGE from a different context is provided by an ethological study of 32 month-old male twins published a number of years ago (Lytton, 1977). The study demonstrated that parent-initiated interactions with their twin children are more similar when the twins are MZ rather than DZ. Of course, like every other increased correlation in the environment of MZ twins, it may not be clear whether it is truly a result of a treatment being elicited by genotype rather than simply a matter of identical individuals being treated more similarly. That is, the direction of causation is not clear. Insofar as the genotypes of individuals create or elicit environments, cross-sectional twin studies will not be able to distinguish the ensuing CorGE from any other effects of the genes. That is, positive CorGE will increase estimates of all the genetic components of variance and negative CorGE will decrease them. However, we will have no direct way of knowing which genetic effects act directly on the phenotype and which result from the action of environmental variation caused initially by genetic differences. In this latter case, the environment may be considered as part of the ``extended phenotype'' (see Dawkins, 1982). If the process we describe were to accumulate during behavioral development, positive CorGE would lead to an increase in the relative contribution of genetic factors with age, but a constant genetic correlation across ages (see Chapter ). However, finding this pattern of developmental change would not necessarily imply that the actual mechanism of the change is specifically genotype-environment autocorrelation. The second major type of CorGE is that which arises because the environment in which individuals develop is provided by their biological relatives. Thus, one individual's environment is provided by the phenotype of someone who is genetically related. Typically, we think of the correlated genetic and environmental effects of parents on their children. For example, a child who inherits the genes that predispose to depression may also experience the pathogenic environment of rejection because the tendency of parents to reject their children may be caused by the same genes that increase risk to depression. As far as the offspring are concerned, therefore, a high genetic predisposition to depression is correlated with exposure to an adverse environment because both genes and environment derive originally from the parents. We should note (i) that this type of CorGE can occur only if parent-offspring transmission comprises both genetic factors and vertical cultural inheritance, and (ii) that the CorGE is broken in randomly adopted individuals since the biological parents no longer provide the salient environment. Adoption data thus provide one important test for the presence of this type of genotype-environment correlation. Although most empirical studies have focused on the parental environment as that which is correlated with genotype, parents are not the only relatives who may be influential in the developmental process. Children are very often raised in the presence of one or more siblings. Obviously, this is always the case for twin pairs. In a world in which people did not interact socially, we would expect the presence or absence of a sibling, and the unique characteristics of that sibling, to have no impact on the outcome of development. However, if there is any kind of social interaction, the idiosyncrasies of siblings become salient features of one another's environment. Insofar as the effect of one sibling or twin on another depends on aspects of the phenotype that are under genetic control, we expect there to be a special kind of genetic environment which can be classified under the general category of sibling effects. When the trait being measured is partly genetic, and also responsible for creating the sibling effects, we have the possibility for a specific kind of CorGE. This CorGE arises because the genotype of one sibling, or twin, is genetically correlated with the phenotype of the other sibling which is providing part of the environment. When above average trait values in one twin tend to increase trait expression in the other, we speak of cooperation effects (Eaves, 1976b) or imitation effects (Carey, 1986b). An example of imitation effects would be any tendency of deceptive behavior in one twin to reinforce deception in the other. The alternative social interaction, in which a high trait value in one sibling tends to act on the opposite direction in the other, produces competition or contrast effects We might expect such effects to be especially marked in environments in which there is competition for limited resources. It has sometimes been argued that contrast effects are an important source of individual differences in extraversion (see Eaves et al., 1989) with the more extraverted twin tending to engender introversion in his or her cotwin and vice-versa. Sibling effects typically have two kinds of detectable consequence. First, they produce differences in trait mean and variance as a function of sibship size and density. One of the first indications of sibling effects may be differences in variance between twins and singletons. Second, the genotype-environment correlation created by sibling effects depends on the biological relationship between the socially interacting individuals. So, for example, the CorGE is greater in pairs of MZ twins because each twin is reared with an cotwin of identical genotype. If there are cooperation (imitation) effects we expect the CorGE to make the total variance of MZ twins significantly greater than that of DZ's, which in turn would exceed that of singletons (Eaves, 1976b). Competition (contrast) effects will tend to make the MZ variance less than that of DZ's. Other effects ensue for the covariances between relatives, as discussed in Chapter 8. Sibling effects may conceivably be reciprocal, if siblings influence each other, or non-reciprocal, if an elder sibling, for example, is a source of social influence on a younger sibling.