Encyclopedia of Personality and Individual Differences

Living Edition
| Editors: Virgil Zeigler-Hill, Todd K. Shackelford

Behavioral Genetics

  • John C. LoehlinEmail author
Living reference work entry

Latest version View entry history

DOI: https://doi.org/10.1007/978-3-319-28099-8_734-2

Keywords

Twin Study Genetic Contribution General Cognitive Ability Behavior Genetic Adopted Child 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Synonyms

Definition

The study of how genes and environments contribute to behavioral variation.

Introduction

Behavioral (or behavior) genetics has a long past, dating back to the breeding of animals for behavioral characteristics. For example, some 50 different breeds of dogs are mentioned in surviving Greek and Roman documents, in such behaviorally distinct categories as hunting dogs, shepherd dogs, guard dogs, war dogs, and pets (Brewer et al. 2001).

The present account of behavior genetics will, however, be primarily focused on differences among individuals, not breeds. It will chiefly address individual differences among humans. Work with other species – in laboratories and in the wild – will be mentioned, but not reviewed in detail. For a more comprehensive account of both human and animal behavior genetics and the biological background, the reader may wish to consult a standard textbook on behavior genetics, such as Plomin et al. (2013).

The behavior genetic research on human individual differences will here be addressed in terms of two broad questions raised by Anastasi (1958): “How much?” and “How?” Anastasi actually asked three questions. The first, “Which one?,” refers to whether variation on a behavioral trait is a result of genetic or environmental factors. A straightforward answer to this question is an animal breeding study. Such a study selects males and females high on some trait to mate with one another and males and females low on the trait also to mate. All the animals are otherwise subject to identical conditions. If genes are involved, over several generations the high and low lines are expected to diverge. For example, in one study two separate lines of mice were bred for showing high activity in a laboratory apparatus known as an “open field,” two separate lines of mice were bred for low levels of activity, and two lines were bred at random to act as control groups (DeFries et al. 1978). After 30 generations of selection, there was no overlap in activity levels between the high lines and the low lines – that is, the least active mouse in a high line was more active than the most active mouse in a low line. As expected, the control lines remained intermediate. Thus genes can be shown to influence open-field behavior in mice. Environment can also be shown to influence mouse open-field behavior. A study by Henderson (1967) involved mice that had previously been exposed to a relatively strong stimulus (electric shock) and an intermediate stimulus (sound of a buzzer) or left undisturbed. The mice that had received the intermediate stimulus were, on the whole, the most active in the open-field apparatus, showing that environment had an effect. However, not all mouse strains responded equally, suggesting that the effects of an environmental stimulus could depend on the genes and thus that an either-or question (Which one?) was inadequate.

On the other hand, the question “How much?” addressed to human traits asks about the extent to which differences in some characteristic in a human population reflect three causal sources: genes, environmental features shared by persons growing up in the same family, and environmental features unique to individuals within families. Anyone with even mild curiosity, observing the considerable variety in the people around her/him on almost any characteristic, may wonder what is responsible for it. Is it the difference in the individuals’ genes? Is it the difference in the families in which they were raised? Is it the effect of events unique to each individual? If all three, how much does each account for? Studies involving twins and adoptees, to be described shortly, address such “How much?” questions.

Finally, the question “How?” asks for the details of how genetic differences or environmental differences get transformed into behavioral differences. This is a harder question, but one increasingly being addressed in behavior genetic and related research.

Note that despite the word “genetics” in “behavior genetics,” the goals of the discipline, as addressed both in the “How much?” and “How?” questions, are as much environmental as genetic. Indeed, it has sometimes been suggested that the immediate impact of behavior genetic research may be more on the environmental than the genetic side. Knowing the extent to which differences in a behavioral outcome, such as academic achievement, are a function of environmental variables shared by family members (such as socioeconomic status) or of environmental variables unique to individuals (such as an inspiring teacher) can point the way to fruitful directions both for research and for social policy.

Answering “How Much?” Questions

Our interest may be aroused by the behavior of a particular individual, but it should be emphasized that the “How much?” questions are not for explaining any one person’s behavior, but are addressed to the proportion of individual differences in some trait in some population which results from genetic variation in that population, from environment shared in families, or from environment not shared. The primary methods that behavior geneticists have used for answering such “How much?” questions for humans are the twin study and the adoption study, plus elaborations of these.

Twin Studies

The basic twin study involves comparing two kinds of twin, identical twins (monozygotic, MZ) and fraternal twins (dizygotic, DZ). MZ twins come from a single combination of egg and sperm that has split in two very early in prenatal development and resulted in two individuals with the same set of genes. DZ twins come from a pair of simultaneous fertilizations that result in two individuals who are genetically ordinary brothers and sisters, that is, who share on average half their segregating genes. If both kinds of twin are reared together in families, as is typical, they will both share common family influences; moreover, they will both tend to be exposed to particular features of the family environment at the same age.

In the simplest case, if P stands for observed resemblance (“phenotypic” in genetic parlance), G stands for resemblance due to genes, and C stands for resemblance due to shared family environment, then P MZ = G + C, because MZ twins share both genes and family environment, and P DZ = 1/2G + C, because DZ twins share environment, but only half their genes.

If we take the observed resemblance P to be measured by the correlation between the twins on some trait in some population, from the above equations, we can deduce G to be equal to 2(P MZ – P DZ ) and C to be equal to 2(P DZP MZ). Thus if MZ twins are correlated 0.65 on some trait and DZ twins 0.40, G comes out to be twice the difference between these, or 50%, and C comes out to be the excess of twice P DZ over P MZ, or 15%. What we have left after both G and C are accounted for (100% − 50% − 15% = 35%) represents factors that make even MZ twins different, which includes environmental events not shared by twins, plus such things as random errors in gene expression or trait measurement.

Now of course it can get more complicated than this. If parents are genetically similar for the trait, due to marriages among related individuals or to a tendency for similar people to mate (“assortative mating”), one needs to use a slightly higher figure than 0.5 to describe the genetic resemblance between siblings (including DZ twins) – the resemblance between MZ twins will not be affected. Or if some of the differences between twins are due to errors in measuring the trait involved, one may wish to deduct measurement error (i.e., 1 – reliability) from the unshared environment component. Or if the effects of genes on a trait do not simply add up, but depend on the other genes present, MZ differences will not be affected, because MZ twins, being genetically identical, will share gene configurations as well as individual genes, whereas gene configurations will tend to be broken up in the transmission of genes from parents to offspring, lowering the genetic resemblance of DZs below 0.5. And there may be correlations and interactions between genes and environments to be taken into account: persons with certain genetic makeup may preferentially seek out certain environments, or some specific combinations of genes and environments may have effects unpredictable from either. Investigators actually carrying out research in a given area must pay attention to such factors, but broad surveys typically assume that genetic and environmental contributions to the variation of a given trait in a particular population mostly just add together and that the average degree of genetic resemblance between ordinary siblings on most traits does not depart radically from 0.5. A meta-analysis of thousands of twin studies of a variety of human traits over 50 years found that most of the findings were consistent with the simple formulas given above (Polderman et al. 2015). We consider some exceptions later.

Twin studies of higher-level cognitive functions. Many twin studies have been done for higher-level cognitive traits, as measured by IQ tests or tests of general cognitive ability. In the Polderman et al. review just mentioned, MZ twins were correlated on average 0.710 on tests of higher-level cognitive functions, DZ twins on average 0.441. Thus, the application of the simple formulas above suggests that individual differences for these traits in these populations were due about 53% to genetic differences, 17% to environmental differences shared by families, and 29% to a residual category including events unique to individuals, errors of measurement, and the like. But these figures vary by age. Table 1 shows a breakdown from Polderman et al.’s meta-analysis.
Table 1

Estimates from twins of genetic and environmental contributions to higher-level cognitive functions at different ages, from the meta-analysis of twin studies by Polderman et al. (2015)

Age range

Genes

Shared environment

Unshared and other

Pairings

0–11

0.451

0.316

0.233

215,731

12–17

0.606

0.080

0.314

75,766

18–64

0.795

−0.113

0.318

9296

65+

0.658

−0.055

0.397

10,030

Note that after childhood, genetic differences account for the bulk of observed individual differences in higher cognitive skills, the proportion due to having shared a family environment drops sharply, and the proportion due to other factors remains a fairly steady minority, increasing slightly with age. The column headed Pairings refers to the number of paired scores involved in the estimates: it is the largest for the childhood category, still quite large in adolescence, and smaller, but still substantial, in adulthood and old age. It is larger than the actual numbers of twins involved, for two reasons. In longitudinal twin studies, of which there are a number in the data, a particular pair of twins will typically be measured at several ages, contributing several pairings to the total. Or twins may be given a number of different measures on a given occasion, again increasing the number of paired scores. This, for example, is quite typical in the case of temperament and personality measures.

Twin studies of temperament or personality. Many twin studies during the past 50 years have been centered on, or have included, measures of personality or temperament. Table 2 summarizes the results of such studies in the same fashion as Table 1 for higher cognitive functions.
Table 2

Estimates from twins of genetic and environmental contributions to temperament and personality at different ages, from the Polderman et al. (2015) meta-analysis

Age range

Genes

Shared environment

Unshared and other

Pairings

0–11

0.543

0.068

0.399

213,182

12–17

0.587

−0.081

0.594

72,119

18–64

0.423

0.001

0.576

338,257

65+

0.386

−0.030

0.644

6746

Several resemblances and some differences from the results for higher cognitive functions may be noted. First, the genes contribute substantially to personality resemblance, as to cognitive resemblance, at all ages. However, the genetic contribution tends to go down with age for personality, rather than up. For personality, the contribution of shared family environment is minor at any age; it is individual experiences (or other factors contributing to twin differences) that are playing an increasing role.

Twin studies of psychopathology. Many of the studies included in Polderman et al.’s meta-analysis were focused on physiological variables or psychopathological conditions. A good many of these involved much smaller samples due to the rarity of the condition or the difficulty of the measurement. However, an example of a psychopathological condition for which a reasonable amount of data was available is conduct disorder, as shown in Table 3.
Table 3

Estimates from twins of genetic and environmental contributions to diagnoses of conduct disorder at different ages, from the Polderman et al. (2015) meta-analysis

Age range

Genes

Shared environment

Unshared and other

Pairings

0–11

0.503

0.181

0.316

240,984

12–17

0.504

0.126

0.370

91,616

18–64

0.652

0.014

0.334

8092

65+

Insufficient data to analyze

Whether or not someone is diagnosed with a conduct disorder again appears to be more a matter of genes than of environment; there is a modest contribution of shared family environment, decreasing with age, and most of the environmental influences are unique to the individual. Also, one might conjecture that conduct disorder tends not to be much of an issue for people over 65.

Adoption Studies

Another way of addressing the question “How much?” is via families in which one or more children are adopted into the family at birth. These children share a family environment with each other and with any existing biological children, but (unless adopted from relatives) not genes. The correlation between adopted children (first equation below) provides a direct estimate of the effect of shared environment, an estimate arrived at indirectly in twin studies.

\( \begin{array}{l}{P}_{Unrelated\; Adoptees}= C,\\ {}{P}_{Ordinary\; Siblings}=1/2 G+ C.\end{array} \)

Twice the difference between the correlations provides an estimate of the genetic contribution G. For example, if for some trait in some population, ordinary siblings are correlated 0.40 and adoptive siblings 0.15, we can estimate from the adoptees that 15% of resemblance on the trait is due to shared family environment and, from twice the difference between the correlations, that 50% is due to genes.

In adoption studies, we can also compare correlations of parents with their biological and adopted children.

\( \begin{array}{l}{P}_{P arent, Adopted Child}={C}_{P C},\\ {}{P}_{P arent, Biological Child}=1/2\ G+{C}_{P C}.\end{array} \)

Again, twice the difference estimates the genetic contribution, G. In this case C PC represents the environmental contribution to the child’s similarity to the parent, which may well include some of the factors involved in the C of the previous equations (socioeconomic status and the like), but is not necessarily identical to it. Thus, if parent-child correlations on some trait are 0.40 for biological children and 0.10 for adopted children, our estimate for G is twice the difference, or 60%, and for C PC 10%. Adoption studies are considerably rarer than twin studies, because recruiting twins is easier than recruiting the necessary combinations of individuals from families who have adopted one or more children. Accounts of two recent US adoption studies may be found in Petrill et al. (2002) and Horn and Loehlin (2010).

Twins Reared Apart

An interesting design, but one requiring great effort to collect enough data for stable analyses, combines the twin and adoption studies. It is the study of identical twins who have been reared apart in separate families.

\( \begin{array}{l}{P}_{MZ\; apart}= G,\\ {}{P}_{MZ\; together}= G+ C.\end{array} \)

The MZ apart correlation directly estimates G, the genetic component of trait variation, and the difference between the two correlations estimates the shared environment component C. Although the study of MZ twins reared apart is more dramatic, the study of DZ twins reared together and apart is also informative:

\( \begin{array}{l}{P}_{DZ\; apart}=1/2\ G,\\ {}{P}_{DZ\; together}=1/2\ G+ C.\end{array} \)

Here, twice the DZ apart correlation estimates G, the genetic contribution, and the difference between the two correlations again estimates C, the effect of shared family environment. A large study of twins reared apart, which for the most part supported the results of ordinary twin and adoption studies, was carried out at the University of Minnesota. An account may be found in Segal (2012).

More Complex Designs

It is quite possible to extend the above designs to encompass a multiplicity of groups. Thus, a twin family design may include the children of MZ and DZ twin pairs. Since MZ pairs are genetically identical, an MZ twin parent is genetically as similar to his twin’s children as he is to his own, but a DZ twin is not. Or one can bring spouses or parents of twins into the design. Or one can study blended families resulting from divorce and remarriage, in which the children may be genetically full siblings, half siblings, or unrelated (e.g., Reiss et al. 2000). In principle, one could write and solve a series of equations of the sort just illustrated, but in practice, overall models of the design are created and are fit by computer to the data. This process can result in overall estimates of genetic and environmental contributions to trait variation, breaking down shared environment into that shared by parent and child and that shared by siblings.

A meta-analysis of personality traits (Vukasovic’ and Bratko 2015) included both twin studies and adoption and other family studies (Table 4).
Table 4

Estimates of genetic contributions to personality traits in twin and family designs, after Vukasovic’ and Bratko (2015)

Design

Effect of genes

95% Confidence interval

Number of studies

Twin studies

0.47

(0.45–0.49)

38

Adoption and family studies

0.22

(0.17–0.28)

16

Clearly, twin studies are yielding higher estimates of the contribution of the genes to individual differences in personality than are adoption and other family studies. One reason this could happen is that the genetic contribution to personality traits may include a substantial nonadditive component. That is, the effects of individual genes may not just add together, but a substantial role may be played by genetic dominance (the effect of a gene depends on the other gene it is paired with at the same chromosomal locus) or epistasis (the effect of a gene depends on genes present at other chromosomal loci). Because gene configurations tend to get broken up in the transmission of genes from parent to child, ordinary family correlations, including DZ correlations, mostly reflect additive effects of genes. Not so for MZ correlations, identical twins share gene configurations as well as individual genes. For traits involving substantial nonadditive genetic components, designs that depend on comparisons of MZ twins with other groups, such as the classical twin design, will give higher estimates of the genetic contribution than will designs that do not. Here we are beginning to get into the question “How?,” but before we pursue this further, let us look briefly at studies of individual differences in other species and at multivariate and longitudinal extensions of behavior genetic designs.

Studies of Individual Differences in Nonhuman Species

Successful domestication of animals is in itself evidence that genetic variation for behavioral traits exists in wild populations. Studies in a number of such populations have made estimates of the genetic contribution to individual differences in species as diverse as garter snakes, bighorn sheep, and squid. A review of such studies (Réale et al. 2007) reported estimates of the genetic contribution to individual differences on behavioral traits. These estimates ranged from 0.00 to 0.89 across traits and species, with a median of 0.21.

Laboratory Studies of Animals

Many laboratory studies with lower animals have been carried out by behavior geneticists. Such studies allow the investigator to control the environments to a high degree in order to isolate effects due to genes. They also allow the control of mating patterns in a manner not feasible with humans and thus permit genetic manipulations of various kinds.

Selection studies. These have already been mentioned, with an example involving high and low levels of activity of mice in an open-field apparatus. Other selection studies have involved, among other traits, rats learning to perform in a particular maze (Tryon 1940) and fruit flies choosing to move toward or away from the light at choice points (Hirsch and Boudreau 1958).

Inbred strains. Another useful possibility in lower animal studies is inbred strains. Brother-sister mating over many generations can result in a strain of genetically identical individuals. One might think of an inbred strain as analogous to identical twins multiplied to 20 or 100, or however many one’s study requires, with the reservation that identical twins have not been subject to an inbreeding process which may eliminate certain categories of genes. Behavioral differences between inbred strains reared under the same conditions reflect genetic differences between the strains. Behavioral differences occurring when members of an inbred strain are subjected to different environmental conditions demonstrate effects of environment. The Henderson study mentioned earlier provides an example. It involved four inbred strains and crosses among them. The effect of the environmental stimulus largely depended on two of the strains and their crosses and was more sensitive to the strain of the female than of the male parent, suggesting that the relation between genetic and environmental effects on behavior may not always be a simple one.

Multivariate Behavior Genetic Designs

So far, we have considered behavior genetic designs as breaking down differences in causal factors for single traits. Such methods can also do this for the correlations or covariances between two traits, and this can be extended to the pattern of relationships among a set of traits.

Multivariate designs have been much less widely employed than univariate ones. They require larger samples, for one thing. In assessing resemblances or differences between two or more traits, the error in measuring each of them enters in. Moreover, the decision whether to analyze correlations or covariances may be important. Two traits may each be only slightly influenced by the genes, but have 100% of their genetic variance in common, and hence loom large in a study of genetic correlations, but not contribute much to genetic covariance.

An example of a multivariate behavior genetic design is a study by Loehlin and Martin (2013), which analyzed genetic and environmental correlations from two samples totaling 3159 same-sex twin pairs. Similar personality factors emerged in both genetic and within-family environmental sources of covariation.

Longitudinal Behavior Genetic Designs

Most of the designs considered in this chapter can be made longitudinal by measuring the individuals concerned more than once. Studies of this kind can support the results of cross-sectional studies carried out with different individuals at different ages. For example, the increasing genetic contribution to general cognitive ability with age was confirmed in a meta-analysis of longitudinal behavior genetic studies spanning infancy to adolescence (Briley and Tucker-Drob 2014). Longitudinal studies can also address the nature of the increasing genetic contribution. The same meta-analysis found that in the early years of life, this mostly reflected new genes coming on line, whereas later on it mostly reflected the amplification of the effects of existing genes.

Longitudinal studies can also address continuity and change. They suggest that the genes contribute to the continuity of cognitive ability throughout the life span, but shared environment does only in childhood (Plomin and Spinath 2004).

Gene-Environment Correlation and Interaction

As noted earlier, an additional source of complexity in analyzing genetic and environmental contributions to individual differences is that genes and environments may be correlated or may interact (Plomin et al. 1977).

Gene-environment correlation may take various forms: it may be passive, in that parents may supply environments to their children that are correlated with the genes they provide them – intellectual parents, for example, provide their children with genes favoring cognitive ability and are likely to have many books around the house. Or the correlation may be active, in that individuals with certain genotypes may seek out environments that enhance their predispositions – i.e., the amplification referred to in the preceding section. Because of gene-environment correlation, many environment measures, such as parents’ treatment of their children or the family’s socioeconomic status, turn out to have an appreciable genetic component (Plomin and Bergeman 1991).

With respect to passive correlations, adoption studies are helpful, since different parents supply the genes and the environments. Longitudinal studies, as noted, may be useful in assessing active gene-environment correlation.

Gene-environment interaction, in this context, does not refer to the obvious fact that interactions between genes and environments are necessary for human development. It refers, rather, to interaction in the statistical sense – the fact that some outcomes may depend on particular gene-environment combinations and be unpredictable from either genes or environment separately. A particular instance, a greater genetic contribution to cognitive performance in higher socioeconomic groups, has been reported in two US samples at different age and socioeconomic levels (Turkheimer et al. 2003; Harden et al. 2007). However, a broad investigation of the topic of G × E interaction in cognitive ability in 14 studies in 4 countries and 4 age groups concluded that “we did not find a consistent pattern of results between age groups, data sets, and countries” (Molenaar et al. 2013).

Attempting to Answer “How?” Questions

Answering “How much?” questions has, on the whole, been highly successful. Behavior geneticists have shown that individual differences on a great variety of behavioral traits reflect genetic differences to a substantial degree. No longer do psychologists attribute almost everything to environment, as many did before the rise of behavior genetics in the 1950s and 1960s.

Behavior geneticists, however, have also found that individual differences reflect environmental differences. And, strikingly, their studies have shown that – at least past childhood – the environmentally caused differences mostly do not reflect the environmental features that family members share, but rather environmental features that affect different family members differently.

A Different Strategy: Assessing Genetic Effects at Population, Not Family, Levels

With the development of relatively fast and inexpensive ways of genotyping individuals, it has become feasible to compare genetic and phenotypic similarity in individuals while excluding close relatives, for whom environmental similarity may exist. In a way, this is just a new approach to “How much?” questions, but it can go beyond them in providing, for example, evidence about the effects of genes of different frequencies and about the relative plausibility of different evolutionary mechanisms for maintaining genetic variability in a population.

A study of this kind is Verweij et al. (2012), which is also an example of the large, multiauthored studies becoming increasingly common, studies that pool the data from several samples, often from different countries, in order to address genetic issues requiring a lot of power in the analyses. The study did not attempt to locate particular genes associated with a trait, but rather asked about the extent to which genetic similarity (measured over hundreds of thousands of genetic loci that differ among humans) corresponds to phenotypic similarity (measured by four subscales of Cloninger’s Tridimensional Personality Questionnaire). The analysis also obtained an index of inbreeding for each individual, based on the length of runs of homozygosity in his or her genotype, reflecting the extent to which an individual’s ancestors were drawn from a limited genetic pool. Large existing samples, one from Australia and three from Finland, were used for the analyses. If there were pairs of individuals in the samples who were more closely related than second cousins, one from each pair was eliminated. Thus the sample consisted of individuals, with no close relatives involved. The methodology is referred to as genome-wide complex trait analysis (GCTA).

The results were used to assess the likelihood of different mechanisms for maintaining genetic variability in a population. They were judged to fit best with a model of mutation-selection balance, a model that assumed that new mutations continuously arose and were continuously but gradually selected against, so that at any given time many, particularly those with small effects, would still be present. The estimates of the effects of genes on the personality traits were lower than those arrived at in the usual twin and family studies – about one-fifth as large. The authors speculate that their methods are largely picking up the additive effects of relatively common genetic variants and that the discrepancy largely reflects the effects of numerous relatively rare genetic variants and/or a substantial nonadditive contribution of genes affecting personality. Evidence for the latter had earlier been provided by large twin and family studies in the USA and Australia (Lake et al. 2000). The Australian samples overlapped, but the twin and family relationships from which the conclusions were drawn in the Lake et al. study were eliminated in Verweij et al.

The Role of Particular Genes in Accounting for Individual Differences

As the possibility arose of measuring genes at many loci at which humans show variation, an obvious step toward answering the “How?” question about human individual differences was simply to correlate genetic differences at specific loci with trait differences. If several loci were found that together could account for a large proportion of the genetic contribution to differences on a trait – i.e., the “How much?” that had been estimated from twin and adoption studies – a major step would have been taken along the road to figuring out “How?”

The If has proved to be a big one. Many genetic loci are uniform across the human species, but a large number are not. Such varying loci are called “single-nucleotide polymorphisms” (SNPs). Modern methods permit assessing hundreds of thousands of them – even millions. This leads to an immediate statistical problem: if one were to assess relationships between a hundred thousand SNPs and a psychological trait in a dozen people, one would expect to find a large number of relationships merely by chance. In the early days, many associations between genes and behavioral traits were reported, but in further studies, the associations failed to replicate. More recently, larger and larger samples have been used in such “genome-wide association studies” (GWAS), with built-in replication schemes included. And as the authors of one such study of general cognitive ability, carried out with 7100 subjects and ~2.5 million SNPs, concluded: “Taken together, our results concur with other recent studies: they support a substantial heritability of [general cognitive ability], arising from a very large number of causal SNPs, each of very small effect” (Kirkpatrick et al. 2015, p. 1).

Similar results have been found for personality traits. A meta-analysis of genome-wide association studies (van den Berg et al. 2016) located 29 samples that included some measure of the trait extraversion – samples from Australia, the USA, the UK, and at least a half-dozen European countries. Genes were assessed in these samples for 1.1–6.5 million SNPs in a total of 63,030 subjects, with an additional 9783 individuals in a cross replication sample. After allowing for the number of comparisons, no association of a SNP with extraversion was deemed to be statistically reliable. None of the SNPs with the strongest observed associations had been previously reported to be related to personality, psychopathology, or brain functioning, and all failed the cross replication test. As in the earlier-mentioned study by Verweij et al., a GWAS analysis accounted for some, but much less genetic variation than that estimated from twin and adoption studies, suggesting that either the genetic variation mostly came from rare variants not sampled by the measured SNPs or resulted from nonadditive genetic effects of gene combinations rather than individual genes.

Within-Family Environmental Variation

Twin and family studies had shown that for most traits, the bulk of the substantial environmental contribution to individual differences was within families rather than shared by family members. So behavior geneticists attempted to isolate the causes of this within-family variation. They have had, on the whole, results similar to those of the gene finders. One review of studies on this topic (Turkheimer and Waldron 2000) concluded that the demonstrable effect of environmental differences upon behavioral differences between siblings was small and got smaller in better controlled studies, e.g., when genetic differences were incorporated into the design. They suggest that the relationships between phenotypes, genotypes, and environments are so complex that “some aspects of the development of complex human behavior may remain outside the domain of systematic scientific investigation for a very long time” (p. 93).

One study (Loehlin and Martin 2011) looked at personality structure assessed solely from within-family environmental differences via an analysis of MZ twin differences, which are by definition not genetically caused, and compared it with personality structure resulting from both genetic and within-family environmental differences, via the differences within DZ twin pairs. The structures were very similar. In a later paper (Loehlin and Martin 2013), the authors concluded that personality structure – how behaviors hang together – is a feature of the phenotype. Thus, it can be found in any phenotypic data, but won’t be separately localizable in the structure either of the genes or the environment – although, presumably, complex causal chains lie between both of them and the phenotype.

Turkheimer and his colleagues (Turkheimer et al. 2014) urge a “phenotypic null hypothesis for the genetics of personality,” which holds that genetic variance is not an independent mechanism of individual differences in personality – these, and their consequences, occur at the phenotypic level. Rather, the genetic contribution to personality is complex and unfocused. More closely related individuals are more similar and share more genes, but that in itself does not tell us much about personality. Where behavior genetic designs are most useful is in rejecting particular causal hypotheses. To take one example, an earlier age at first sexual intercourse is related in the population to adult delinquency, suggesting that the latter may be a consequence of the former. However, it turns out that within MZ twin pairs, an earlier age at first sexual intercourse tends to go with less adult delinquency (Harden et al. 2008), suggesting that the relationship in the population has some other explanation – for example, that the same genes or common environmental factors affect both.

Conclusion

Behavior geneticists have learned a great deal about the “How much?” question from twin, adoption, and family studies – knowledge that renders the “Which one?” question obsolete. They have learned less about the “How?” question, except that, for normal human behavioral traits, substantial relationships between particular genes and particular dimensions of variation do not appear to exist. The ability to split environmental differences into those shared by family members and those not shared and to look at environmental effects controlling for genetic differences (e.g., within MZ twin pairs) should, however, continue to be useful. And much scope remains for extending these methods into multivariate and longitudinal designs.

Cross-References

References

  1. Anastasi, A. (1958). Heredity, environment, and the question “how?”. Psychological Review, 65, 197–208.CrossRefPubMedGoogle Scholar
  2. Brewer, D. J., Clark, T., & Phillips, A. (2001). Dogs in antiquity. Warminster: Aris & Phillips.Google Scholar
  3. Briley, D. A., & Tucker-Drob, E. M. (2014). Genetic and environmental continuity in personality development: A meta-analysis. Psychological Bulletin, 140, 1303–1331.CrossRefPubMedPubMedCentralGoogle Scholar
  4. DeFries, J. C., Gervais, M. C., & Thomas, E. A. (1978). Response to 30 generations of selection for open-field activity in laboratory mice. Behavior Genetics, 8, 3–13.CrossRefPubMedGoogle Scholar
  5. Harden, K. P., Turkheimer, E., & Loehlin, J. C. (2007). Genotype-by-environment interaction in adolescents’ cognitive aptitude. Behavior Genetics, 37, 273–283.CrossRefPubMedGoogle Scholar
  6. Harden, K. P., Mendle, J., Hill, J. E., Turkheimer, E., & Emery, R. E. (2008). Rethinking timing of first sex and delinquency. Journal of Youth and Adolescence, 37, 373–385.CrossRefPubMedPubMedCentralGoogle Scholar
  7. Henderson, N. D. (1967). Prior treatment effects on open-field behavior of mice – A genetic analysis. Animal Behaviour, 15, 365–376.CrossRefGoogle Scholar
  8. Hirsch, J., & Boudreau, J. C. (1958). The heritability of phototaxis in a population of Drosophila melanogaster. Journal of Comparative and Physiological Psychology, 51, 647–651.CrossRefPubMedGoogle Scholar
  9. Horn, J. M., & Loehlin, J. C. (2010). Heredity and environment in 300 adoptive families. New Brunswick: Transaction.Google Scholar
  10. Kirkpatrick, R. M., McGue, M., Iacono, W. G., Miller, M. B., & Basu, S. (2015). Results of a “GWAS plus”: General cognitive ability is substantially heritable and massively polygenic. PloS One, 10, e112390.CrossRefGoogle Scholar
  11. Lake, R. I. E., Eaves, L. J., Maes, H. H. M., Heath, A. C., & Martin, N. G. (2000). Further evidence against the environmental transmission of individual differences in neuroticism from a collaborative study of 45,850 twins and relatives on two continents. Behavior Genetics, 30, 223–233.CrossRefPubMedGoogle Scholar
  12. Loehlin, J. C., & Martin, N. G. (2011). What does a general factor of personality look like in unshared environmental variance? Personality and Individual Differences, 51, 862–865.CrossRefGoogle Scholar
  13. Loehlin, J. C., & Martin, N. G. (2013). General and supplementary factors of personality in genetic and environmental correlation matrices. Personality and Individual Differences, 54, 761–766.CrossRefGoogle Scholar
  14. Molenaar, D., van der Sluis, S., Boomsma, D. I., Haworth, C. M. A., Hewitt, J. K., Martin, N. G., et al. (2013). Genotype by environment interactions in cognitive ability: A survey of 14 studies from four countries covering four age groups. Behavior Genetics, 43, 208–219.CrossRefPubMedPubMedCentralGoogle Scholar
  15. Petrill, S., Plomin, R., DeFries, J. C., & Hewitt, J. K. (Eds.). (2002). Nature, nurture, and the transition to adolescence. New York: Oxford University Press.Google Scholar
  16. Plomin, R., & Bergeman, C. S. (1991). The nature of nurture: Genetic influence on “environmental” measures. Behavioral and Brain Sciences, 14, 373–414.CrossRefGoogle Scholar
  17. Plomin, R., & Spinath, F. M. (2004). Intelligence: Genetics, genes, and genomics. Journal of Personality and Social Psychology, 86, 112–129.CrossRefPubMedGoogle Scholar
  18. Plomin, R., DeFries, J. C., & Loehlin, J. C. (1977). Genotype-environment interaction and correlation in the analysis of human behavior. Psychological Bulletin, 84, 309–322.CrossRefPubMedGoogle Scholar
  19. Plomin, R., DeFries, J. C., Knopik, V. S., & Neiderheiser, J. M. (2013). Behavioral genetics (6th ed.). New York: Worth.Google Scholar
  20. Polderman, T. J. C., Benyamin, B., de Leeuw, C. A., Sullivan, P. F., van Bochoven, A., Visscher, P. M., et al. (2015). Meta-analysis of the heritability of human traits based on fifty years of twin studies. Nature Genetics, 47, 702–709.CrossRefPubMedGoogle Scholar
  21. Réale, W., Reader, S. M., Sol, D., McDougall, P. T., & Dingemanse, N. J. (2007). Integrating animal temperament within ecology and evolution. Biological Reviews, 82, 291–318.CrossRefPubMedGoogle Scholar
  22. Reiss, D., Neiderheiser, J. M., Hetherington, E. M., & Plomin, R. (2000). The relationship code. Cambridge, MA: Harvard University Press.Google Scholar
  23. Segal, N. L. (2012). Born together – Reared apart. Cambridge, MA: Harvard University Press.CrossRefGoogle Scholar
  24. Tryon, R. C. (1940). Genetic differences in maze-learning ability in rats. In 39th yearbook of the National Society of Studies in Education (Part 1, pp. 111–119). Bloomington: Public School Publishing Co.Google Scholar
  25. Turkheimer, E., & Waldron, M. (2000). Nonshared environment: A theoretical, methodological, and quantitative review. Psychological Bulletin, 126, 78–108.CrossRefPubMedGoogle Scholar
  26. Turkheimer, E., Haley, A., Waldron, M., D’Onofrio, B., & Gottesman, I. I. (2003). Socioeconomic status modifies heritability of IQ in young children. Psychological Science, 14, 623–628.CrossRefPubMedGoogle Scholar
  27. Turkheimer, E., Pettersson, E., & Horn, E. E. (2014). A phenotypic null hypothesis for the genetics of personality. Annual Review of Psychology, 65, 515–540.CrossRefPubMedGoogle Scholar
  28. van den Berg, S. M., de Moor, M. H. M., Verweij, K. J. H., Krueger, R. F., Luciano, M., Vasquez, A. A., et al. (2016). Meta-analysis of genome-wide association studies for extraversion: Findings from the genetics of personality consortium. Behavior Genetics, 46, 170–182.CrossRefPubMedGoogle Scholar
  29. Verweij, K. J. H., Yang, J., Lahti, J., Veijola, J., Hintsanen, M., Pulkki-Raback, L., et al. (2012). Maintenance of genetic variation in human personality: Testing evolutionary models by estimating heritability due to common causal variants and investigating the effect of distant inbreeding. Evolution, 66, 3238–3251.CrossRefPubMedPubMedCentralGoogle Scholar
  30. Vukasovic’, T., & Bratko, D. (2015). Heritability of personality: A meta-analysis of behavior genetic studies. Psychological Bulletin, 141, 769–785.CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.Department of PsychologyThe University of Texas at AustinAustinUSA

Section editors and affiliations

  • Julie Schermer
    • 1
  1. 1.The University of Western OntarioLondonCanada