What have studies of twins indicated about the causes of personality development?

A multitude of prior twin and adoption studies have examined genetic and environmental influences on criminality (Coccaro & McNamee, 1998, Eysenck & Eysenck, 1978, Goldsmith & Gottesman, 1996, Hutchings & Mednick, 1975, Tehrani & Mednick, 2001), juvenile delinquency (Jacobson et al., 2000, Jacobson et al., 2000, Rowe, 1983, Rowe, 1986, Taylor et al., 2000), and adult antisocial behavior (Cadoret et al., 1990, Cadoret et al., 1987, Crowe, 1974, Dilalla & Gottesman, 1989, Krueger et al., 2001, Lyons et al., 1995, McGuffin & Thapar, 1998). These studies have for the most part utilized behavioral measures related to criminality as well as criteria and symptom counts for the diagnosis of Antisocial Personality Disorder (ASPD) from the DSM-III (APA, 1980), DSM-III-R (APA, 1987), or DSM-IV (APA, 1994).

Research of this kind has consistently revealed substantial genetic contributions to the antisocial phenotype (c.f. Dilalla & Gottesman, 1989). A review of the literature also suggests that while the shared or family environment may promote adolescence-limited delinquency (Moffitt, 1993), genetic influence is more prominent among those individuals who start young and continue to engage in a pattern of life-course persistent antisocial behavior (see also Dilalla & Gottesman, 1989, Lyons et al., 1995, Moffitt, 1993, Taylor et al., 2000; for exceptions see Rowe, 1983, Rowe, 1986). For example, Taylor, Iacono, and McGue (2000), reported probandwise concordance rates for twins selected as either early starter (i.e. life-course persistent) or late starter (i.e. adolescence-limited) antisocial phenotypes. With respect to the early starter phenotype, co-twins of monozygotic (MZ) pairs were at a substantially greater risk than co-twins in dizygotic (DZ) pairs (55 and 29% probandwise concordances, respectively). In contrast, for the late starter phenotype, there was relatively little differential risk to co-twins based on zygosity (MZ and DZ probandwise concordance, 43 and 39%, respectively).

Furthermore, in behavior genetic studies of adult criminality, MZ twins typically have higher concordance rates than DZ twins (Dilalla & Gottesman, 1989, Eysenck & Eysenck, 1978, McGuffin & Thapar, 1998, Tehrani & Mednick, 2001). In a review of genetic studies of adult criminality, Goldsmith and Gottesman (1996) calculated MZ and DZ pairwise concordance rates of 52% and 23%, respectively, indicating that MZ twins are more than twice as likely to be concordant for criminal conduct than DZ twins. In another review, Dilalla and Gottesman (1989) noted an average MZ pairwise concordance of 51% and DZ pairwise concordance rate of 22% across several studies.

In addition, adoption studies of criminality have provided evidence for genetic contributions to the etiology of antisocial behavior. Several investigations and reviews have shown that adoptees with biological parents who were criminal had higher rates of criminal convictions than control samples of adoptees without such pedigrees (Crowe, 1974, Hutchings & Mednick, 1975, Tehrani & Mednick, 2001). Data from two sets of adult adoptees also support the importance of genetic influences on antisocial behavior as diagnosed by DSM- III criteria (Cadoret et al., 1987, Cadoret et al., 1990).

Although these findings regarding heritable influences on antisocial behavior have appeared consistently across several studies and designs, their relevance to the construct of psychopathy is unclear. Essentially, these studies have all focused on the etiology of behaviors associated with ASPD, delinquency, and criminality. These investigations have not examined psychopathy in terms of a constellation of maladaptive personality traits as delineated by Cleckley (1941/1988).

A system that exists for assessing this alternative conceptualization of psychopathy is Hare's (1991) Psychopathy Checklist-Revised (PCL-R). Probably the best validated and most widely used measure of psychopathy, the PCL-R has been consistently utilized in forensic settings and has shown excellent psychometric properties. The PCL-R is a diagnostic inventory consisting of 20 items that is scored on the basis of a structured interview and review of collateral file information.

Factor analysis of the PCL-R has yielded a latent two-factor model of psychopathy (Harpur, Hare, & Hakstian, 1989). Factor 1, encompasses emotional detachment and affective-interpersonal features of psychopathy, including many of the core personality traits suggested by Cleckley (1941/1988) such as lack of guilt, shallow affect, and narcissism. Factor 2, in contrast, relates to an impulsive, antisocial lifestyle, and comprises many chronic behaviors typical of ASPD such as a parasitic dependence, juvenile delinquency, and irresponsibility. These two facets of psychopathy show differing patterns of relations with various external criteria, including personality trait measures (Harpur et al., 1989, Patrick, 1994, Verona et al., 2001) and indices of emotional reactivity (Patrick, 1994, Patrick et al., 1993).

Based on the operational criteria used in the previously mentioned behavior genetic designs, these studies appear to have tapped the antisocial behaviors related to Factor 2 of the PCL-R, but not the core personality traits of Factor 1. Although the two factors are moderately correlated, it is important to determine whether there is differential heritability for the personality dimension of psychopathy. However, the PCL-R was constructed specifically for use within prison populations where researchers have access to file data. An alternative approach is to use a well-validated self-report personality inventory to investigate the etiology of psychopathy in non-criminal populations. This issue is vital given the fact that many psychopathic individuals may function successfully outside of prisons at a sub-clinical level (Ishikawa et al., 2001, Levensone et al., 1995, Widom, 1977).

While analysis of normal range personality traits relevant to psychopathy and other antisocial phenotypes has also revealed genetic contributions (Taylor et al., 2000, Tellegen et al., 1988, Zuckerman et al., 1980), well-known self-report psychopathy scales such as the Minnesota Multiphasic Personality Inventory (MMPI) Psychopathic Deviate (Pd) scale (McKinley & Hathaway, 1944) and the California Psychological Inventory (CPI) Socialization (So) scale (Gough, 1960), appear not to assess the core personality traits outlined by Cleckley. Empirically, these instruments correlate primarily with the antisocial deviance facet (Factor 2) of the PCL-R, rather than the affective-interpersonal facet (Factor 1; Hare & Cox, 1978, Lilienfeld & Andrews, 1996). Studies examining the Constraint superfactor of the Multidimensional Personality Questionnaire (MPQ; Tellegen, 1978/1982), which is closely linked to externalizing and antisocial behavior (Krueger, 1999, Krueger et al., 1996, Krueger et al., 2001, Krueger et al., 1994), are insufficient as well. Evidence from an analysis by Verona et al. (2001), shows that normal range traits related to Constraint also correlate primarily with Factor 2 rather than Factor 1 of the PCL-R. Consequently, twin studies using these self-report indices of personality, although informative, may fail to measure important aspects of psychopathic personality per se.

To facilitate research aimed at comparing the validity of the personality and behavior-based approaches, Lilienfeld and Andrews (1996) introduced the Psychopathic Personality Inventory (PPI). The PPI was constructed specifically to assess the core personality traits of psychopathy rather than the antisocial behavior facet. The criterion validity of the PPI has been demonstrated through correlations with the PCL-R. Analyses from a study in which the PPI and PCL-R were administered to prison inmates (Poythress, Edens, & Lilienfeld, 1998) revealed that most of the unique variance in PPI scores was associated with the affective-interpersonal features (Factor 1) rather than the antisocial deviance features (Factor 2) of the PCL-R. This attests to the validity of the PPI as a measure of the core personality facet of psychopathy.

This synopsis of the behavioral genetic research on psychopathy leads us to the primary objectives of the present study. Previous behavior genetic research has addressed issues related to etiological influences on antisocial behavior as characterized by ASPD and chronic criminality. The current investigation, in contrast, seeks to determine the relative genetic and environmental contributions to variance in the core personality facet of psychopathy. We used the PPI to assess this dimension. As such, this study was the first behavior genetic study to examine the etiology of psychopathy using a personality-based approach.

We expected the results to reveal a substantial genetic influence. While it is difficult to predict the degree and type of genetic loading, twin studies of personality in some instances have suggested effects of genetic non-additivity for complex personality traits (Finkel & McGue, 1997, Tellegen et al., 1988). Non-additive genetic effects consist of dominance, epistasis and emergensis. The first two effects involve the interaction among genes within and across loci, respectively. Emergenesis, however, entails several traits, genetically determined independently from one another, combining in a configural manner.

In summary, this investigation sought to assess the relative genetic and environmental contributions to variance in the personality trait dimension of psychopathy. Previous investigations have repeatedly demonstrated the heritability of relevant maladaptive personality traits, as well as antisocial behavior operationalized by criminality and criteria from the DSM. However, it is unclear whether these findings are applicable to the personality domain of psychopathy.