Causation of Criminality: Genetic and Environmental Influences
Introduction
The causes of criminal behavior are diverse and complex; its development is influenced by both genetics and the environment. This paper examines some of the literature aiming to disentangle the effects of these factors, and also present a model for criminal behavior that supports moderate genetic and environmental influence. Many family, adoption and twin studies have been carried out since the 1920s. These studies provide empirical evidence for a larger genetic or environmental role by providing information on increased risk for families, the influence of genetic and environmental factors and the possible points of interference with the results. This evidence serves to put forth the knowledge that criminal behavior results from a combination of factors and development over many years before the event, leading to possible interventionist treatments (Gabrielli et al., 1983).
It must be pointed out that the definition of criminality promotes a vagueness on the subject matter being studied. In most studies, criminality is associated with the criminal act and the apprehension for the act, in order to have a definite population of subjects to study. This is a questionable basis for most studies, as the very definition of criminality on which the studies rely is liable to social change (Joseph, 2001). Another point of contention is the fact that the population of subjects only includes those who have been caught, without taking into account those who have not been apprehended (Gabrielli et al., 1983).
Genetics vs. Environment
In analyzing the role of genetics in crime and criminal behavior, one must keep in mind that genes do not influence criminal behavior singly; they are likely to contribute to personality characteristics, or other such dispositions that make an individual behave in a criminal manner (Lyons, 1996). The high increase in crime in most Western countries during the last 50 years points to a high probability that genetics does not solely contribute to criminality; the rate of mutation and adaptation in the genetic pool cannot change to such a high degree (Rutter & Smith, 1995). National differences certainly affect criminality, but not necessarily in a direct genetic way: for example, the murder rate for young people in the US is 10 times higher than it is in the UK (Rutter, 1996). The racial genetic explanation is also very dangerous in that it must take into account social attitudes and cannot be purely scientific. For example, Department of Justice data shows that in 1997, there were 725 black men in jail for every 100,000 black men in the population, with a parallel number of 138 for white men, even though they show no difference in the rate of antisocial personality disorders, which is a marker for subsequent criminal behavior (Dept. of Justice, 1997; Robins et al., 1991).
In discussing environmental influence, a prime fallacy is that everything is reducible to genetics, which it is not. Our response to the environment is governed by genetic mechanisms, but these responses are adaptations attributable as a crucial role of the environment (Gibbard, 2001). So far, there has been no research on what environmental factors are triggers for crime, or what genetic factors would allow an individual to seek out specific environments which predispose to crime (Raine, 1993). Currently, one can only determine an approximate amount of influence that the environment or genes play. The scientific community has yet to explain in detail the cause of emotional change from one situation to another, sometimes leading to violent and criminal behavior. Psychobiology can play an extremely important role in causation. Parental influences, peer pressures, social experiences and other environmental behaviorist factors can be just as responsible for behavior as an individual�s own judgments (Gabrielli et al., 1983). This problem leads to another loose hole in behavioral genetic analyses: it is still impossible to truly separate environment and genetic influence, due to the ambiguity of emotions and environmental causation.
However, three approaches to genetic investigation � twin studies, family studies and adoption studies � allow science to get a better picture of the etiology of antisocial behavior.
A. Twin Studies
Carey (1991) attempted to put forth a model explaining and breaking down genetic versus environmental influence for the published twin data of registered criminality in Denmark. His linear path diagram of fraternal twin influence for criminality showed genetics, common environment and unique environment affecting the phenotype of each sex twin, with reciprocal imitation parameters between the two phenotypes to account for the imitation of behaviors by the other twin. Carey also showed a correlation component for genetic and common environment influences that affect both twins. Included in this model was the idea of a threshold, in which phenotypic liabilities exceeding a certain threshold denote a liability for criminal behavior. The imitation parameters include a distinction between competition and cooperation; when the imitation parameter a > 0, prosocial behavior of one twin increases probability of prosocial behavior in the other twin, and when a < 0, the probability of antisocial behavior is increased. With this model, it is a simple task to measure MZ and DZ twin liabilities and concordances; if the twins are peers and influence each other, prevalence of criminality should be affected and influences calculated. For example, when heritability values (genetic variances) are higher than 0, phenotypic liability variances should be higher for MZ twins than they are for DZ twins. Because of this, it is possible to backtrack: from values of criminality for MZ and DZ twin pairs, one can calculate various variance effects, including concordance rates in twins.
Studies of twin concordance for criminality have shown higher concordance rates in monozygotic twins than in dizygotic twins. In a review of all twin studies conducted since 1929, Raine found that the average concordance rates were found to be 51.5% for MZ twins and 20.6% for DZ twins, suggesting a substantial genetic component (1993). This is only an average and therefore taken as a representation of the approximate genetic influence. The ranges for the concordance rates stretch very far: a study in Holland found a 100% concordance rate for MZ twins and a 20% concordance rate for DZ twins (Mednick et al., 1985). The first twin criminality study, conducted by the German psychiatrist Lange in 1929 found a 77% pairwise concordance for MZ twins and a 12% pairwise concordance for DZ twins (qtd. in Brennan et al., 1991). At the lower end of the spectrum, values were found to be 61% and 11% for MZ and DZ twins respectively in a Japanese study in 1961 (Mednick et al., 1985). A recent reanalysis of data from the Danish twin registry found data whose MZ and DZ twin differences were more similar to the averages calculated by Raine: 74% pairwise concordance for MZ twins and 47% for DZ twins (Carey, 1991).
In any case, the important fact gleaned from these studies is the fact that MZ twin concordance was larger than that of DZ twins. This means that the closer genetic relations of the MZ twins probably play a larger role in the development of similar criminal behaviors. These close genetic relations can also mean that MZ twins experience a more similar environment than DZ twins (Lyons, 1996). This is also a problem for twin studies; the existence of differential treatments of MZ and DZ twins can artificially raise concordances. It must be kept in mind that heritabilities calculated from MZ/DZ twin comparisons are not as yet applicable to the individual; heritability of traits are based on populations and variances, and not on individuals (Raine, 1993). Lyons (1996) does conclude in his study that environment provides a significant contribution to individual differences in criminal behavior, and that genetic factors also contribute a significant influence. However, Rutter (1996) does find some contradictions in concordance studies. For example, he mentions the fact that DZ twin correlations are higher than that for full siblings in most studies. Another pertinent point brought up was the fact that Danish adoptee heritabilities were found to be half of the heritabilities found in Danish twin data. Also, Swedish adoptees seem to have the same criminality as the general population, while in Danish adoptees, the rate was raised. This again brings into play the anonymous influence of rearing environment in the two countries.
Coid et al. (1993) studied a population of 490 twins who were seen at Maudsley Hospital from 1948 to 1988. Within this sample, 16.9% were found to have a criminal record, and of those, 68 were men, or 27.6% of the total number of men. The study found that there was a significant two-way interaction in this case: with 1 degree of freedom, the deviance or log-linear analyses turned out to be 21.015 with a significance of 0.001. Among the probands in this study, 21% had an early criminal record, with 83% of these being men. Under further probabilistic analysis, the chi square value came out to be 28.62, an extremely high and significant value. This study came to the conclusion that more men were criminal than women among the twin pairs.
Interestingly enough, data from the Danish Twin Registry also supported this conclusion from a different angle. According to the peer influence model of Carey (1991), male twins had a lower threshold of 1.28, compared to the female twins� threshold of 2.18. Prevalence values had a similar trend for MZ and DZ twins. For MZ twins, 13.42% of the males had a record as opposed to 2.59% of the females. This is of interest in separating genetic influences in that not only do the male twin pairs share common genes and Y chromosomes, but that there is also a strong environmental component that has not been fully analyzed by psychobiology or evolutionary psychology. By personal observation, current social attitudes and mores have a stronger impact in prompting criminal behavior in males than females in today�s society.
There is an alternate twin model, developed by Jones et al (1995). They use the theory of contagion, by which risk for a trait is increased when in the presence of individuals with the trait. In this model, contagion only operates within siblings if causes of the trait only affect one twin and not the other. Calculations of concordances are greater when contagion is present, as are prevalence rates. Furthermore, there are different pathways that contagion can follow. Those in which transmission is more likely is known as a preferred pathway. This model makes use of the fact that criminal prevalence is much greater in men than it is in women by providing each gender with its own set of parameters. However, this is only an environmental model and not yet a genetic model and is therefore not truly applicable in separating genetic influences from environmental ones. Brennan (1993) says that the current biometric models cannot account for all of the possible permutations of genetics and environmental similarities that can occur among twins. A few studies have been conducted of twins reared apart. These studies are not methodologically strong and suffer from biases in sampling, and so have not been used much. Brennan also points out a general fault with twin studies: they all report pairwise concordances instead of probandwise concordance rates, leading to errors such as sample bias.
B. Family Studies
Family studies have been conducted but with less success than twin studies. Generally, family studies seem to lack a good separation point for either genetic or environmental influences. Because family studies analyze the link between family members� criminalities, it also must involve the social-familial influences included in parenting a child. While it is true that parental criminality is one of the best single predictors of later crime, the cause of its success as a predictor cannot be reflected by either genes or environment (Raine, 1993; Brennan, 1991; Lee, 1982). One study, conducted by Gabrielli et al. (1984), did conclude that kids whose biological parents were criminal were also more likely to be criminal themselves. However, this conclusion was also connected with another quasi-conclusion, that urban environments played a larger part in criminality. However, the researchers admitted that this conclusion might be induced, because one would not know the level of correlation, if there were any at all.
C. Adoption Studies
Adoption studies examine criminality in adopted children and its relationship with criminal or non-criminal biological and adoptive parents. In most cases, adoption studies have provided strong support for conclusions from twin studies (Scarpa, 1997). However, adoption studies seem to be better at separating influences on behavior (Mednick et al., 1988). This is because the adoptive study design approaches a controlled experiment (Lee, 1982). Adoption studies usually use the forward or backward trace designs, using the parent and then the offspring as the variable respectively. Genetics is represented by the contribution of the biological parents to their offspring�s criminality. Environment can be varied; that is, the adoptive parents can supply the adoptees� total environment, or if the adoptee was separated from its biological parents some time after birth, a fraction of the total. However, adoption studies can be skewed because of selective placement or preferential placement, the efforts to either match home backgrounds or to place the adoptee in a �good� home. If any matching trait is genetically linked to the subjects of study, genetic influence estimates can be deflated.
A common conclusion in most adoption studies is that a child adopted near birth but having no contact with its biological parents has a higher likelihood of exhibiting criminal behavior if its biological parents is or are criminals (Van Dusen et al., 1983; Bohman, 1996). An analysis done by Mednick et al. (1988) found a 14.7% conviction rate for adoptees with noncriminal biological fathers and criminal adoptive fathers, a value very close to the 13.5% conviction rate for adoptees with wholly noncriminal parents. However, those adoptees whose biological father had been convicted were found to have a 20% conviction rate, a significant jump in rates observed. Those adoptees with wholly criminal parents were found to have a 24.5% conviction rate, the highest percentage. Brennan (1996) found that male chronically offending adoptees with chronically offending biological parents made up 1% of the male adoptees but were responsible for 30% of convictions within the cohort. One could conclude from this data that the adoptive father�s convictions have no association with the rate of adoptee convictions, and that the criminality of the biological parent had a significant influence on the child�s behavior. This data supports at least a partial genetic influence for criminality.
A separate Mednick et al. study (1984) found that as genetic relationship increases, the level of criminal concordance increases as well. For example, concordances for half-siblings were found to be 12.9%, but for full siblings, the concordance was increased to 20%. A study conducted by Moffitt (1987) found that more than one genetic pathway may be associated with criminal behavior, suggesting an inherently more complex interaction of similar purpose genetic factors. More evidence for only a partial genetic influence includes the fact that over 70% male adoptees in Mednick et al.�s study (1988) whose biological parents have three or more convictions were never convicted themselves. This means that a genetic correlation is not enough to produce convictions in adoptees; environmental or other factors must mediate in the process leading to development of the behavior.
The adoption study provides more concrete information about gender in criminal offending than does twin studies. A study done by Mednick et al. (1984) supported a stronger relationship between biological mother conviction and adoptee conviction. Though female criminal statistics were significantly lower than male statistics, there is also a significantly less amount of social pressure to commit crime. Therefore, criminal acts by a female can be more genetically induced and thus more closely related to adoptees� convictions. Crowe (1974) found a genetic-environmental interaction within adoptees that had a criminal mother and spent longer times in temporary placement. These adoptees were found to have the highest rates of conviction. As a sidenote, it should be pointed out that also among males, there was a statistically significant association between adoptee criminality and time spent in temporary placement, meaning environmental factors such as institutionalization and selection bias have an effect as well (Mednick et al., 1984).
Evidence for a moderate environmental influence can also be deduced from adoption studies. Two studies found that knowledge of biological parent crime by the adoptive parents did not affect adoptee convictions at all (Mednick et al., 1984; Brennan et al., 1993). Data from the Mednick et al. study (1984) shows that 63% of the biological parents were convicted after the adoption occurred, thus not transmitting that information to the adoptive parents. 16.1% of their children were later convicted. For the 37% of biological parents who were convicted before the adoption occurred, only 15.6% of their children were convicted. Due to the large jump in percentages of biological parent convictions before and after adoption and the similarity of the percentages of adoptee conviction, it can be concluded that there is no correlation there. Thus, the genetic correlation between biological parent and children did not necessarily account for a conviction, and the adoptive environment had somewhat of an effect.
A study of social class and crime conducted by Van Dusen et al. (1983) found evidence for some environmental correlation. The general trend was for criminal convictions to increase as the social class was lowered. When a child from an originally high class background was adopted by a low social class family, the number of convictions rose. In the opposite way, a low social class background child going into a higher class family lowered the number of criminal convictions. Also, children from a high social background adopted into another class were generally less criminal than children originally from the other classes. This would seem to show that the class environment does matter in criminality, but that a genetic predisposition would also affect values. Still, there is some doubt about these results as police bias in arrests could significantly change conviction records. Hesitation to arrest middle or upper class criminals could deflate criminality values. Van Dusen et al.�s analyses of the data using a joint biological-environmental model found that rearing environment has a measurable impact on adoptee criminality, that there is a biological factor associated with lower social class and criminality that may be genetically transmitted, that for males, the environment is more important, and that for females, the biological factor is more important.
Conclusions
Through a general compilation of conclusions from these three designs, it is clear that there are moderate effects of both genetics and environment. The closer genetic relation of MZ twins compared to DZ twins seemed to promote higher concordance rates for criminality and thus support genetic effects. Environmental conclusions from the twin studies include support for an increased socialization of crime and violence for males rather than females. Family studies support both genetic and environmental reasoning through the conclusion that parent criminality has an effect on child criminality. The adoption studies concluded that biological parents have a high effect on adoptee criminality, but also that other factors are needed to promote the actual development of criminal behavior. Other conclusions include the environmental influence of social classes on crime and support for stronger biological mother and adoptee conviction rates. There are flaws in each design that could contribute to inflations of values and misinterpretations, but science continues to move forward, correcting itself when it can. Much is not yet known about the development of criminal behavior and its root causes, but it has been shown that both genetics and the environment do matter at a moderate level, and that much research will be needed to separate the sociological and psychological aspects of crime from the predispositions and genetic heritage accounts of criminality.