Crime Causation: Biological Theories

Epidemiological evidence that genetic factors contribute to criminal behavior come from three sources: family, twin, and adoption studies. The limitation of family studies is the inability to separate the genetic and environmental sources of variation. Therefore, given the limited utility of family studies to separate issues of nature versus nurture, this section will focus on two other epidemiological research designs that are better equipped to test for genetic effects.

Twin studies. Twin studies support the contention that a heritable trait may increase risk for criminal behavior. Twin studies compare the rate of criminal behavior of twins who are genetically identical or monozygotic twins (MZ) with twins who are not, or dizygotic twins (DZ) in order to assess the role of genetic and environmental influences. To the extent that the similarity observed in MZ twins is greater than that in DZ twins, genetic influences may be implicated.

The twin design, however, does present some problems to this interpretation. The use of twin studies to test questions of heritablilty are limited in that it is a rare occurrence for the twins to be reared in separate environments. Moreover, Dalgaard and Kringlen suggest that the greater similarity of MZ twins may be attributed to their shared environmental experiences. In line with this hypothesis, Carey (1992) suggests that MZ twins may imitate one another more than DZ twins, and that this phenomenon could lead to an overestimation of heritability. Consequently, any review of twin studies must keep these limitations in mind.

Earlier twin studies reported considerable variations in the pairwise concordance rates (among monozygotic twins from 100 percent to 25 percent and in dizygotic twins from 81 percent to 0 percent). Several methodological flaws in earlier twin studies made it difficult to draw conclusions regarding genetic liability to criminal behavior. First, the operational definition of "criminal behavior" varied from mild incidental offenses to long-term incarceration. A potentially more serious methodological concern is that, with the exception of Dalgaard and Kringlen's study and the twin study that follows, all other twin samples suffered from biased samples.

Using an unselected sample of 3,586 twin pairs in Denmark, Christiansen reported 52 percent of the monozygotic twins were (proband-wise) concordant for criminal behavior whereas only 22 percent of the dizygotic twins were (probandwise) concordant for criminal behavior. A marked increase of probandwise concordance for criminal behavior among monozygotic twins suggests that the MZ twins inherit some biological characteristic(s) that increases their joint risk for criminal involvement.

Results from more recent twin studies are largely in agreement with results obtained from earlier twin studies. Variability in criteria for criminal behavior and sample composition does not appear to change the genetic effect, an outcome which suggests that criminal behavior and correlates of antisocial behavior (i.e., antisocial symptom counts, conduct disorder) may be genetically mediated. The twin design, as discussed earlier, is limited in that the assumption of equal environments is often violated. Studies comparing the concordance rates in MZ twins reared apart can avoid this problem, but it is difficult to obtain such subjects. Christiansen has noted that several of the earlier twin studies had cases in which a set of monozygotic twins were raised in separate environments; these preliminary data suggest that studying MZ twins reared apart may be an important behavioral genetics tool to investigate the etiology of criminal behavior. To the present authors' knowledge, only one modern twin study has employed this type of research design to test whether criminal behavior may be genetically mediated.

Twins reared apart. Grove and others investigated the concordance of antisocial problems, as measured by the Diagnostic Interview Schedule (DIS), among a sample of thirty-two sets of monozygotic twins reared apart (MZA) who were adopted by nonrelatives shortly after birth. Because this was a nonclinical sample, very few subjects met Diagnostic and Statistical Manual-III criteria for antisocial personality. To remedy this limitation, symptoms that contribute to the overall DSM-III diagnoses were counted to assess for subclinical manifestations of antisocial problems. Grove found substantial overlap between the genetic influences for both childhood conduct disorders (correlation of .41) and adult antisocial behaviors (correlation of .28). Although these findings are based on a small number of subjects, the Grove findings are congruent with the findings from other twin studies and extend the twin literature by evaluating MZ twins raised in separate environments.

Adoption studies. Another epidemiological design that may more cleanly parcel out most environmental effects is the adoption design. Adoption studies provide a natural experiment to test the existence and strength of inherited predispositions. Adoptees are separated at birth from their biological parents. Thus, similarities between the adoptee and biological parents can be regarded as estimates of genetic influences, while similarities between the adoptee and the adoptive parents may be thought of as estimates of environmental influences. Moreover, the adoption design allows for the assessment of interaction effects between environmental and genetic influences. Adoption studies have been carried out in three different countries: the United States, Sweden, and Denmark.

Iowa. The first adoption study to explore the genetic transmission of criminal behavior was carried out in Iowa by Crowe. The sample consisted of fifty-two adoptees (including twenty-seven males) born between 1925 and 1956 to a group of forty-one incarcerated female offenders. A group of control adoptees were matched for age, sex, race, and approximate age at the time of adoption. Seven of the fifty-two adoptees sustained a criminal conviction as adults whereas only one of the control adoptees had a conviction. Since these adoptees were separated from their incarcerated mothers at birth, this tends to implicate a heritable component to antisocial behavior.

A separate series of adoption studies carried out in Iowa by Cadoret and colleagues (1980, 1983, 1985, 1987, 1995) have supported Crowe's original findings. These independent replications lend support to the notion that criminal behavior may have important genetic influences.

Several characteristics of the Iowa adoption studies carried out by Cadoret and colleagues should be noted. First, the genetic factors of interest, namely the antisocial status of the biological parents, were ascertained from "poorly maintained adoption agency records" or incomplete prison and hospital records. Second, a high refusal rate of adoptee interviews introduces the possibility that adoptees who consented to be interviewed may be qualitatively different from those who declined. Third, in two of the Cadoret studies, antisocial status of the adoptees was determined from telephone interviews (1987, 1995). In short, what is needed is the use of criminal national registries that would provide a better opportunity to assess lifetime, cumulative records for all subjects (both biological and adoptive parents and adoptees). This condition is difficult if not impossible to meet in the United States. Such requirements, however, have been met by adoption studies from two Scandinavian countries, Denmark and Sweden.

Sweden. Bohman examined the criminality and alcoholism rates among 2,324 Swedish adoptees and their biological parents, as determined by a check with national criminal and alcohol registries. Preliminary findings led Bohman to conclude prematurely that biological fathers who were criminal only (without alcohol abuse) were not more likely to have criminal, adopted-away children than biological fathers with no criminal record (12.5 percent vs. 12 percent). He did not differentiate between criminality alone in the biological fathers and criminality accompanied by alcohol abuse in the biological fathers. Further statistical analysis reveals that when these two groups are separated, there are significantly more criminal-only sons (without alcohol abuse) of criminal-only biological fathers than there are criminal-only sons of other fathers (8.9 percent vs. 4.9 percent, p (significance level) < 0.05).

One of the chief findings to emerge from the Swedish Adoption Study is evidence for a distinct, highly heritable form of alcoholism and criminality that may be transmitted from father to son (Cloninger et al., 1981). Cross-fostering analyses revealed the emergence of two distinct subtypes of alcoholism that could be differentiated based upon genetic and environmental influences. The first subtype proposed by Cloninger, Type I alcoholism, appears to be affected by environmental factors, such as the socioeconomic status of the adoptive parents. Type I alcoholics were found to have a late onset of alcohol abuse (i.e., after age twenty-five) and did not engage in criminal behavior.

Type II alcoholism, in contrast, appears to have a strong genetic component. Type II alcoholics are typically males with alcohol and criminal registrations. The biological fathers of these Type II alcoholics had an early onset (i.e., before age twenty-five) of recurrent alcoholism and criminality (sample size, n = 36). Environmental factors, such as low socioeconomic status and alcoholism in the adoptive parents, were not found to influence the frequency of Type II alcoholism. Moreover, the male adoptees' risk of Type II alcoholism was not increased by an interaction between genetic and environmental factors. These findings were later replicated in independent adoption studies carried out in Sweden by Sigvardsson and others (1996) and in a reanalysis of the Danish Adoption Project (Tehrani and Mednick, forthcoming). Although the utility of the Type I, Type II paradigm in clinical samples has received mixed support, these data suggest the existence of a highly heritable form of criminality and alcoholism that is genetically transmitted from father to son.

Denmark. Mednick, Gabrielli, and Hutchins carried out a study of the genetic influence on criminal behavior using an extensive data set consisting of 14,427 Danish adoptees (ranging in age from twenty-nine to fifty-two years) and both sets of biological and adoptive parents. They found that adopted-away sons had an elevated risk of having a court conviction if their biological parent, rather than their adoptive parent, had one or more court convictions. If neither the biological nor adoptive parents were convicted, 13.5 percent of the sons were convicted. If the adoptive parents were convicted and the biological parents were not, this figure only increased to 14.7 percent. When examining sons whose biological parents were convicted and adoptive parents remained law-abiding, however, 20 percent of the adoptees had one or more criminal convictions. Moreover, as the number of biological parental convictions increased, the rate of adoptees with court convictions increased.

There were cases where a biological father, mother, or both contributed more than one child to this population. Some of these children, either full or half-siblings, were placed in different adoptive homes. There were 126 male-male half-sibling pairs placed in separate adoptive homes. Of the 126 male-male half-sibling pairs in the study 31 pairs had at least one member of the sibship convicted. Of these 31 pairs, 4 pairs were concordant for convictions (concordance rate = 12.9 percent for half-siblings). The study yielded 40 male-male full-sibling pairs who were adopted into separate homes. Fifteen pairs had at least one member of the sibship sustain a criminal conviction; of these 15 pairs, 3 pairs were concordant for convictions (concordance rate = 20 percent for full siblings). Although the numbers are small, these findings suggest that as the level of genetic relationship increases, the level of concordance increases.

These data, obtained from three different countries and in different laboratories, lend support to the notion that criminal behavior appears to have a strong genetic component. In addition, the combination of genetic and environmental factors, or gene-environment interactions, has also been the subject of investigation. Accordingly, several adoption studies have noted significant interactive effects when environmental variables are also taken into account.