Genetics and Human Behaviour: The Ethical Context
Current findings: quantitative genetics
Antisocial behaviour
9.9 Estimates of heritability for antisocial behaviour from recent research in quantitative genetics cluster around 0.50. The most reliable estimates come from contemporary studies in the Netherlands, Britain, Norway, Sweden, Australia and the US, because these studies examine large, representative samples using sophisticated quantitative modelling techniques. A complementary meta-analysis of 51 twin and adoption studies yielded an estimate of heritability of 0.41 for the genetic influence on antisocial behaviour.(5) Estimates of heritability below 0.20 tend to emerge from studies with unusual design features; for example, observational measures, small sample sizes, very wide age ranges, small groups of girls, or adults being asked to report childhood symptoms retrospectively. Similarly, some, but not all, studies yielding estimates above 0.70 have non-optimal designs, such as small sample sizes or adults being asked to report their childhood symptoms retrospectively.
9.10 The sizes of estimates of heritability vary somewhat across different types of measures of antisocial involvement. Overall, this variation appears to be systematic. It reflects the fact that (all other aspects of methodology being equal) higher estimates of heritability are obtained from studies using measures that sample the most different types of antisocial behaviours. Measures that sample many behaviours yield more accurate estimates because their scores are less contaminated with error, and also because they are more sensitive to the full range of behaviour in the population. The lowest estimates of heritability emerge from observational measures, which sample only a narrow type of behaviour such as hitting a doll or arguing with a parent, over brief spans of time, usually minutes. The next lowest estimates emerge from measures of official offending in juveniles, which also sample a narrow range of behaviour, namely illegal behaviour, which is seldom detected. Fewer than half of juveniles who offend are arrested, and of those arrested 75% are arrested only once or twice. Medium estimates of heritability tend to emerge from measures of self-reported offending, symptoms of conduct disorder and official offending in adults, all of which tend to aggregate across a moderately wide sample of behaviours and moderately long periods of ascertainment.
9.11 The largest estimates of heritability tend to emerge from studies using measures able to array individuals along a continuum from non-antisocial to severely and persistently antisocial. These are studies using other-reported delinquent or aggressive behaviours (such as the Child Behaviour Check List (CBCL) externalising scale), and self-reported personality traits (such as the MPQ aggression scale). These studies tend to include a very large number of items inquiring about a variety of antisocial attitudes and behaviours. Some of these items, such as robbery, are exhibited rarely by people, but others, such as enjoying violent films, are exhibited commonly. As a result, the instruments are sensitive to population variation in the severity of antisocial behaviour.(6) Overall, the distribution of more than 100 estimates of heritability from recent papers approximates a bell-shaped normal curve. This distribution is to be expected from a sample of more than 100 imperfect estimates of a true effect that equals 50% in nature.
9.12 As well as the possibility that genes influence antisocial behaviour, it is also possible that antisocial experience can influence how genes are distributed in the population. This is an implication of the finding that men and women mate on the basis of similarity between the partners’ antisocial behaviour (this is called assortative mating), and that couples in which both people exhibit antisocial behaviour tend to have more children than the norm.(7) Assortative mating on a genetically-influenced phenotype, such as antisocial behaviour has consequences for genetic variation in the population. Because people form unions with other people like themselves, the result is that families differ more from each other on average than they would if people mated randomly. If successive generations mate assortatively, genes relevant to the phenotype will become concentrated within families. Consider height as an example. Whole families clearly differ from other families in terms of height, yet families are made up of persons who are similar in height. Part of the explanation for this phenomenon is likely to lie in the positive assortative mating that occurs for this trait.
9.13 Another important insight from research in behavioural genetics is that while genes influence tendencies towards antisocial personality and antisocial behaviour, they have relatively little influence on the probability of becoming officially designated as a delinquent at any particular court appearance. The designation of ‘delinquent’ is a characteristic influenced by the behaviour of co-offenders, police, parents, lawyers and judges, not merely by the behaviour of the young person.
Violence
9.14 Public debate about the implications of heritability for criminal responsibility often focuses on violent crime. Findings about genetic effects on violence are rare and inconsistent. Three studies report evidence of a value of zero for the heritability for violence,(8) whereas three studies report evidence that heritability for violence is about the same as for non-violent antisocial behaviour (0.50).(9)
9.15 Two main difficulties arise in the study of individual differences in official records of conviction for violent crime. First, even in the largest samples, official convictions for violent crime occur at very low rates. Inconsistent findings arise from such low indicators, because they cannot be reliably aggregated. Secondly, contrary to popular assumption, offenders designated ‘violent’ by virtue of court conviction are not necessarily the most serious, persistent criminals at the antisocial extreme. To illustrate, studies of murderers reveal that approximately half have lengthy histories of repeated assaults, rapes, robberies and other offence types, but the other half have committed a single extreme act after a lifetime free from crime. This indicates that the most serious of violent offences, homicide, as a legally constructed status, captures individuals likely to be quite heterogeneous in their genetic dispositions. Low base rates and heterogeneous participants may explain why studies using conviction data have found no heritability for violence.
9.16 The antidote to studying convictions is to use measures of violence that inquire about violent behaviours that have physical differences and cover a range of severity (fighting, hurting animals, robbery, hitting, aggravated assault using a weapon, gang-fighting, rape and domestic abuse). It is also helpful to use a reporting period long enough for research participants who are violently inclined to exhibit these relatively rare behaviours. This approach was used in two studies of self-reported violence in twins and siblings, which yielded significant estimates of heritability.(10)
9.17 Researchers in behavioural genetics usually do not single out violence for separate analysis. Researchers are dissuaded from doing so by the strong psychometric evidence that antisocial behaviour is a unified construct and therefore a separate research focus on violence is not warranted. Most studies of the structure of antisocial behaviour have suggested that items measuring physical aggression belong together with items assessing stealing, lying, fraud, vice, reckless irresponsibility and other forms of antisocial behaviour.(11)
9.19 Many studies in behavioural genetics have examined measures known to be strong, specific predictors of physical violence. For example, many have used the MPQ aggression scale and have reported strong estimates of heritability. The MPQ aggression scale measures attitudes, values, and beliefs that are consistent with approval of the use of physical violence.(12) Longitudinal research shows that the MPQ aggression scale empirically predicts future conviction for violent crime.(13)
9.20 Overall, the question of genetic influences for violent crime has not interested researchers in behavioural genetics as much as it has the general public. As a result, the evidence base is not sufficient to answer the question decisively. However, there is good evidence of heritability for antisocial traits and behaviours associated with risk for engaging in violent crime, and this suggests that heritable liability for violence is a reasonable hypothesis. An area overlooked by research in behavioural genetics is violence within relationships, a type of violence for which there are reliable aggregate measurement tools.
Sex differences
9.21 Differences between the sexes in heritability of antisocial behaviour may exist, but are small. It is unclear as yet whether these small differences should be interpreted as substantive, or as artefacts of sex differences in measurement. On balance, the results of model tests in large samples suggest that estimates of heritability may be slightly higher among males than females, but that sex-specific models of heritability cannot be justified.(14)
9.22 Estimates of heritability may be slightly smaller for females because measurements of antisocial behaviour among females represent less of the full range of antisocial severity, relative to measurements among males. The antisocial behaviour performed by females is less serious and less frequent than that of males, and females participate in antisocial activities for a much shorter period of the life course than males.(15) The relative rarity and brevity of females’ antisocial behaviour makes it difficult to obtain strong aggregate measures of it, and this may influence estimates of heritability downwards for females.
Footnotes5 Rhee, S. H. & Waldman, I. D. (2002). Genetic and environmental influences on antisocial behavior: a meta-analysis of twin and adoption studies. Psychol. Bull. 128, 490–529.
6 A study of 14,500 Danish adoptee families provides a good example of the importance of measures that sample different acts along a dimension to improve sensitivity to features of antisocial behaviour such as severity, frequency and persistence (Mednick, S. A., Gabrielli, W. F. & Hutchings, B. (1984). Genetic factors in criminal behaviour: evidence from an adoption cohort. Science 224, 891–3). When adoptee family members were classified simply as ‘not convicted’ or ‘convicted’ (a legal status applying to nearly one quarter of Danish males), the estimate of heritability was modest. However, when the number of convictions in an individual’s life-time from age 15 to 50 was considered, stronger estimates of heritability emerged for individuals who had repeatedly been convicted on many successive court dates, presumably reflecting what criminologists call a ‘crime career’.
7 See for example Farrington, D. P., Barnes, G. C. & Lambert, S. (1996). The concentration of offending in families. Leg. Criminol. Psychol. 1, 47–63; Rowe, D. C. & Farrington. D. P. (1997). The familial transmission of criminal convictions. Criminology 35, 177–201; Farrington, D. P., Jolliffe, D., Loeber, R., Stouthamer-Loeber, M. & Kalb, L. (2001). The concentration of offenders in families, and family criminality in the prediction of boys’ delinquency. J. Adolescence 24, 579–96; Kreuger, R. F., Moffitt, T. E., Caspi, A., Bleske, A. & Silva, P. A. (1998). Assortative mating for antisocial behaviour: development and methodological implications. Behav. Genet. 28, 173–86.
8 Bohman, M., Cloninger, R., Sigvardsson, S. & von Knoring, A. L. (1982). Predisposition to petty criminality in Swedish adoptees. I. Genetic and environmental heterogeneity. Arch. Gen. Psychiatr. 39, 1233–41; Mednick, S. A., Gabrielli, W. F. & Hutchings, B. (1984). Genetic factors in criminal behaviour: evidence from an adoption cohort. Science 224, 891–3; Sigvardsson, S., Cloninger, C. R., Bohman, M. & von Korring, A. (1982). Predisposition to petty criminality in Swedish adoptees. III. Sex differences and validation of the male typology. Arch. Gen. Psychiatr. 39, 1248–53.
9 Heritability is reported as 50% in Cloninger, C. R. & Gottesman, I. I. Genetic and environmental factors in antisocial behaviour disorders. In Mednick, S. A., Moffitt, T. E. & Stack, S. A., editors. (1987). The Causes of Crime: New Biological Approaches. Cambridge: Cambridge University Press. pp. 92–109; 32% in Rowe, D. C., Almeida, D. M. Jacobson, K. C. (1999). School context and genetic influences on aggression in adolescence. Psychol. Sci. 10, 277–80; 55% in Rushton, J. P. (1996). Self-report delinquency and violence in adult twins. Psychiatr. Genet. 6, 87–9.
10 Rowe, D. C., Almeida, D. M. & Jacobson, K. C. (1999). School context and genetic influences on aggression in adolescence. Psychol. Sci. 10, 277–80; Rushton, J. P. (1996). Self-report delinquency and violence in adult twins. Psychiatr. Genet. 6, 87–9.
11 Moffitt, T. E., Krueger, R. F., Caspi, A. & Fagan, R. W. (2000). Partner abuse and general crime: How are they the same? How are they different? Criminology 38, 201–35; Blumstein, A., Cohen, J., Das, S. & Moitra, S. (1988). Specialization and seriousness during adult criminal careers. J. Quan. Criminol. 4, 303–45; Farrington, D., Snyder, H. & Finnegan, T. (1988). Specialization in juvenile court careers. Criminology 26, 461–85.
12 Do not be misled by the names of measures; most scales labelled ‘aggression’, including the MPQ, do not measure physical aggression.
13 Moffitt, T. E., Krueger, R. F., Caspi, A. & Fagan, R. W. (2000). Partner abuse and general crime: How are they the same? How are they different? Criminology 38, 201–35.
14 Gjone, H. & Stevenson, J. (1997). The association between internalizing and externalizing behaviour in childhood and early adolescence: Genetic or environmental common influences? J. Abnorm. Child Psychol. 25, 277–86; Eaves, L. et al. (1997). Genetics and developmental psychopathology: 2. The main effects of genes and environment on behavioural problems in the Virginia study of adolescent behavioural development. J. Child Psychol. Psychiatr. 38, 965–80; Taylor, J., McGue, M., Iacono, W. G. & Lykken, D. T. (2000). A behavioural genetic analysis of the relationship between the socialization scale and self-reported delinquency. J. Pers. 68, 29–50; Finkle, D. & McGue, M. (1997). Sex differences and nonadditivity in the heritability on the Multidimensional Personality Questionnaire scales. J. Pers. Soc. Psychol. 72, 929–38.
15 Moffitt, T. E., Caspi, A., Rutter, M. & Silva, P. A. (2001). Sex Differences in Antisocial Behaviour: Conduct Disorder, Delinquency, and Violence in the Dunedin Longitudinal Study. Cambridge: Cambridge University Press.