Brief on Genetics and Crime
by Evan Balaban and Richard Lewontin

The idea that evil or criminal tendencies are “in the blood” is an old one. Belief in evil forces poised to take possession of human minds historically coexisted with the belief that some people are more prone to this type of corruption than others. These ideas were especially important to European social theorists in the Nineteenth Century, as reflected in the most popular literature of the time, where authors passionately explored their major facets. To what extent were criminal tendencies inherent in a person’s makeup, and, once there, to what extent could they be changed? How did criminal tendencies get into people in the first place: from chance events during pregnancy or early childhood that left an imprint, from a systematically lawless environment, or from like begetting like?

Charles Dickens’ Artful Dodger in Oliver Twist and the entire Macquart family described by Emile Zola in his sequence of twenty novels owed their criminal tendencies to hereditary defects. As Zola said in his preface to the first of the series, “heredity has its laws, just like gravity”, although we would regard Zola’s notion of those laws as rather quaint. In the late Nineteenth and early Twentieth centuries, the rise of psychological theories of personality development, social reform movements, and the emergence of behaviorism ushered in an age where the psycho-social environment became the chief causative agent in the development of anti-social and criminal behavior. The reaction to biological theories of the National Socialists in Germany seemed to make hereditary notions of social behavior widely unacceptable. Over the last half-century, the immense expansion of knowledge and technique in genetics and the extraordinary publicity given to genetic research accompanied by the repeated claims that genes “determine” the organism have refocused attention on heredity as a major determining factor causing individuals to commit crimes.

One might suppose that genetic determination of criminal acts would present a serious challenge to the legal system, for how can society hold a person responsible for an act that was the inevitable consequence of his or her biological nature? Many legal scholars, however, are not convinced that biological information purporting to predict criminal behavior would present any new problems for legal systems. This is because insanity defenses and the doctrine of diminished responsibility are well-established parts of modern legal systems and behavior- genetic or other biological information can be accommodated under these rubrics. The legal issue is whether a person is regarded as sane and therefore responsible, irrespective of the underlying causes of that insanity. The issue of causes is regarded as relevant only to the course of action to be taken in sentencing, but not for the guilt or innocence of the accused. So, if the possession of a given genotype only increased the probability of performing a criminal act but did not absolutely determine the action, there would be no serious challenge to legal doctrine.

The status of genetic arguments about “free-will” is therefore not a cause for special concern in the immediate future, but two elements that sit under the surface present problems. The first is the uncritical acceptance by many scholars who are not themselves scientists (and some who are,) of the notion that there has been or soon will be some kind of breakthrough in the understanding of genetic contributions to and predictions of human behavior.  Such a viewpoint may erroneously support events that make up a second concern: the danger to individual rights and to the rights of disadvantaged populations resulting from the “informal application” of tentative behavior-genetic findings.  Correlations of limited observations to information in DNA databases at early phases of criminal investigation could be used to select potential suspects in criminal cases — a common current practice that may now be extended.

Behavior genetics: The new hype

Human behavior geneticists have increasingly abandoned former research designs based on twin studies, and embraced newer methodologies based on direct correlations between DNA sequence variation and variation in performance on behavioral measurements4. The first stage of such an investigation is to survey a very large number of known human polymorphic genes to find statistical correlations between variant gene alleles and the incidence of some behavioral syndrome, say schizophrenia or abnormal aggressiveness. Among the genes for which statistically significant correlations are found, those known to code for gene products that would reasonably seem to be related to the central nervous system are identified as “candidate genes” for further investigation. Or, investigators simply pick combinations of genes with significant correlations and assume that later research will establish such relationships. There are two serious problems with any such approaches.

The first is a statistical fallacy. A correlation between two variables is considered “statistically significant” if a correlation as large or larger than the one observed would rarely occur by pure chance in a sample of observations, without any real causal connection. In conventional practice, “rarely” means 5 percent of the time or less. But if one surveys, say, 10,000 genetic polymorphisms for a correlation with a behavioral syndrome, then up to 500 of those polymorphisms could turn out to be significantly correlated to the behavior at the 5 percent level of significance even if none of them has any real causal relation to the syndrome. How is one to decide which, if any, of these genetic variations is causal? One possibility is to run a second, independent, study on a new sample and see how many of the original 500 significant correlations will show up again. But by pure chance 25 (5 percent) of the original correlations will appear in both samples. How many of these are real? And how many real connections have been missed because a relevant gene polymorphism only increases the probability of the behavior by a small amount too weak to be detected by the study design? This problem is not unique to genetic studies.

The next step is to look for candidate genes among those identified statistically, based on knowledge of which of these code for central nervous system proteins likely to be involved in behavior. The second problem arises at this stage. Behavior is more complicated than brains dictating behavior. For example, like other vertebrates, humans have a neuroendocrine system in which there is a constant feedback between nervous system states and the secretion of a variety of hormones that neurons are sensitive to, which, in turn, have their rates of production modulated by more general physiological clues such as blood sugar levels and products of muscle fatigue. One of the most serious errors in understanding human behavior is to suppose that the brain is an isolated and autonomous causal organ. Reciprocal causal interactions between the brain and the rest of the body make the number of candidate genes for any behavioral syndrome very large, encompassing a large part of any organism’s metabolic functions. This problem is exacerbated if the genes that have a real causal connection to the target syndrome do not completelydetermine the behavior but only make it somewhat more likely, say by a factor of 25 percent.

An example of the candidate gene approach that has become a poster child for the new human behavioral genetics as applied to criminal behavior, is a study which involved the comparison of individuals with sequence differences in a chromosomal region just in front of the gene that encodes an enzyme present in many parts of the brain called monoamine oxidase A, involved in the metabolism of several different brain neurotransmitters. A cohort of about 540 “Caucasian” boys growing up in New Zealand was monitored every few years from the ages of 3-21 years for a multidisciplinary health and development study.  Evidence of maltreatment during the first 11 years of life was gathered from these periodic observations, from parental reports, and from retrospective reports by study members and their families. Court records of violent convictions in adulthood were obtained (11 percent of the subjects had criminal records). A “disposition toward violence” was ascertained by using a personality questionnaire (containing items such as “When I get angry I am ready to hit someone,” and “I admit that I sometimes enjoy hurting someone physically”), to which subjects provided a number on a scale indicating how much they agreed or disagreed with the statement. Finally, subjects provided the investigators with the name of someone who knew them “well”, and the investigators mailed a questionnaire to these persons.

Molecular genetic variation was split into two categories: variants of this region of the monoamine oxidase A (MAOA) gene that previous studies had suggested to have “high enzyme activity”, and ones that previous studies had suggested to have “low enzyme activity” (based on studies in cancerous or non-neural cell lines that do not entirely agree with one other). The investigators classified their subjects according to the presumed activity of their alleles and the extent of their maltreatment as children (“no maltreatment,”, “probable maltreatment,”, “severe maltreatment”). Subjects’ convictions for violent offenses, proportion of conduct disorders, extent of disposition toward violence, and antisocial personality symptoms all correlated positively with maltreatment. But there were larger increases in these behaviors with maltreatment for possessors of “low activity” alleles than for possessors of “high activity” alleles. According to the authors, “These findings provide initial evidence that a functional polymorphism in the MAOA gene moderates the impact of early childhood maltreatment on the development of antisocial behavior in males” (Caspi et al (2002)8, p. 853).

Since this study also clearly shows that childhood maltreatment doesn’t do the holders of any allelic MAOA gene form any good, it is unclear what the proponents of this type of research would have us do based on allelic differences, especially since subsequent research disagrees on how big the effect of allelic variation is, and the extent of psychopathologies in possessors of the “high-activity” allele. While the original study suggested that a sizeable response difference to childhood maltreatment existed between the members of the two allelic variation categories, four more recent studies either showed no difference (2 studies) or only showed weak differences between these groups (2 studies)10.

Modern molecular behavior genetic work remains as tentative and inconclusive as its historical, non-molecular counterparts do, even though its proponents have become much more sophisticated about discussing the complexity of the link between genes, brains, and behavior. Yet a vocal sector of this community remains dedicated to advancing the proposition that individual behavioral variation can be meaningfully predicted from genetic sequence variation. They posit that we will not commit any egregious violations of individual rights by starting to apply such information. We believe this point of view reflects a fundamental conflict of interest, is scientifically flawed, ideologically motivated, and, if generally accepted, could potentially lead to significant setbacks in the rights of individuals in democratic societies, disproportionately affecting those individuals in socially and economically disadvantaged groups ,.

What is criminal behavior?

A third problem with the studies of the genetics of “criminal behavior” is in their extremely narrow view of what constitutes crime. In practice genetic studies have identified crime with crimes of violence against individuals, a bias that is inevitable when there is an attempt to connect genes and behavior. What would be the criteria for a candidate gene whose coded product led to an increased probability of, say, embezzlement, securities fraud, illegal computer hacking or automobile theft? While these are undoubtedly a consequence of mental states, what sort of biochemical or developmental variation can we identify that is conducive to a mental state that contravenes the Eighth Commandment?  Crimes of violence against individuals account for only about 10 percent of all reported criminal offences. About 10 percent of arrests are for violation of drug laws, but there is no common agreement about the mental states that lead to immersion in a culture of drugs, nor the genetic variations that could be the proximate causes of drug-dependency. Even for crimes of violence against individuals it would be naïve to suppose that these overwhelmingly arise from an “aggressive personality” as opposed to calculated or unforeseen acts of violence that arise in the commission of other offenses. The gap between mental states and the coding of gene products does not require simply more work on biochemical and molecular genetics and neurobiology. It requires the elucidation of fundamental causal connections between complex brain states and ongoing behavior that modern neuroscience is still unable to forge convincingly. It will be even more difficult to translate any such links into the abstract and arbitrary categories used to characterize criminal behavior.

The overwhelming danger of under-formalized investigative procedures

While the discussion so far has concentrated on crime and genetics, there is an analogous problem developing in the brain sciences and their applications to criminal behavior —– the “my criminal brain made me do it” argument. The increasing attempts to use brain imaging data to bolster biological arguments of culpability in courts, and to drive public policy decisions on how to treat criminals, is a problem which is receiving good critical coverage in the scientific press, but not in the popular press. Two groups of researchers have recently upped the ante by claiming to provide a scientific framework for criminal brain arguments, hypothesizing that there is an identifiable collection of connected brain regions that constitutes a discrete organ for “moral sense”, and that criminals have defects in just these regions of the brain. Nikolas Rose has previously provided a cogent review and analysis of these newer forms of “biologicization” of criminal behavior and the future challenges these will present in terms of discrimination and screening. He also argued that new ideas of biologically based criminal conduct raise the prospect of not-entirely-consensual “treatments’, increased use of preventive detention and other “pre-emptive” actions on the part of authorities.

Hacking echoed many of Rose’s sentiments, making the point that a shift in language from “crime” to “criminal behavior” signals a shift in the ownership of these human phenomena from the perpetrators of crimes, their victims and the criminal justice system to criminologists, psychologists, sociologists and biologists. He pointed out how this process of public reclassification could influence the behavior and attitudes of the classified and their classifiers, but did not think that biological classifications used by intellectual elites had much feedback on the behavior of the average person. Duster, on the other hand, reinforced Rose’s concerns about the effects of biological ideas primarily on their new owners —– the criminologists, psychologists and sociologists who teach and consult with law enforcement officials. Through them, tentative, theoretical ideas may easily shape the largely hidden procedures police forces and government agencies use to identify “likely” suspects of crimes via burgeoning information databases that may disproportionately scrutinize disadvantaged groups within larger societies. No matter if many of the fruits of these procedures are later thrown out — it is here that introduced biases lead to unequal treatment and rights violations that, if allowed to become accepted procedure, can corrode fundamental rights and freedoms for all members of society.

We share and amplify Duster’s concerns about the “dangerous intersection of allele frequencies in special populations and police profiling via phenotype”. DNA database information can be abused in a manner that can selectively disfavor those segments of the population most likely to run afoul of the criminal justice system, which in many countries is highly correlated with race and ethnicity. To make matters worse, much of the future research correlating genetic variation with behavioral attributes is likely to be corporate. Like the achievement tests that academic institutions and professional schools now rely on, which effectively control young people’s access to opportunities in the U.S., genetic tests employed by private testing companies are corporate products whose composition and scoring procedures are not open to scientific and public scrutiny. We worry that corporations or entrepreneurial academics who offer such services in crime-solving areas of law enforcement are not accountable to anyone. How much access would the public have to the contents of such tests, or to indiscriminate data mining from DNA databases that could be used by law officials or government agencies?

A related concern exists with the tactics employed by law enforcement officials. Take the example of the lie detector, which is based on flawed scientific premises and is inadmissible evidence in court, but which continues to be used in criminal investigations as a tool of intimidation against suspects. If the idea of brains predisposed to criminality or criminal genes is widely accepted, criminologists may consider trying to formalize the strategy of using meaningless biologically-based tests to intimidate suspects in order to leverage confessions.

Yet another cause for concern is the use of gene sequence information in “biological suitability” services. Suppose that private firms offer the government, military or other employers to screen people for their “natural potential,” for lack of criminal tendencies, or for the ability to withstand the stresses of particular lines of work — what “public proof” (such as double-blind studies conducted by a disinterested party) would a private corporation have to offer that any of their methods actually worked? What recourse would people denied opportunities have in the face of such determinations?

The greatest danger of behavior-genetic applications to crime is not in the courtroom, but in the back rooms during the initial phases of law-enforcement investigations where bizarre pseudoscientific theories, fads, and unproven technologies can mix with bias and prejudice free from explicit scrutiny. We believe this is the arena where the interaction between crime and behavior genetics is poised to have its most pernicious effects, and where counteracting these effects poses the greatest challenge.

Recommendations

We recommend that organizations with social justice concerns develop new initiatives to (1) combat the already well-entrenched idea in a variety of academic, public policy and media circles that molecular behavioral genetics will inevitably reveal new information about the control of human behavior that will of necessity change the way societies deal with crime, and (2) explore the variety of ways in which biological information can be abused in the initial phases of crime investigations, and develop effective strategies for regulating such abuses. Goal (1) can be accomplished in part by aiding the formation and public effectiveness of groups of scientific professionals such as the Neuroethics Society, a multidisciplinary group set up to address inappropriate use of brain scans and other applications of neuroscience. Aim (2) will require much more concerted effort to identify particularly pressing issues and place them on the agendas of civil liberty and public policy organizations.

Prepared and Accepted by the CRG Board on February 28, 2007

References

Emile Zola, “La Fortune des Rougons”, 1871

O. Onay (2006) The true ramifications of genetic criminality research for free will in the criminal justice system. Genetics, Society and Policy 2: 80-91.

See N. A. Farahany, W. Bernet (2006) Behavioral genetics in criminal cases: Past, present and future. Genetics, Society and Policy 2: 72-79; E. Parens (2004) Genetic Differences and Human Identities: On Why Talking about Behavioral Genetics Is Important and Difficult. The Hastings Center Report, January-February 2004, S1-S36; E. Parens E, A.R. Chapman, N. Press (eds) (2006) Wrestling With Behavior Genetics: Science, Ethics, and Public Conversation. Johns Hopkins University Press: Baltimore; D. Wasserman (2004) Is there value in individual genetic predispositions to violence? Journal of Law, Medicine & Ethics 32: 24-33; D. Wasserman, R. Wachbroit (eds) (2001) Genetics and Criminal Behavior. Cambridge University Press: NY.

D. Hamer (2002) Rethinking behavior genetics. Science 298: 71-72.

P. Appelbaum (2005) Behavioral genetics and the punishment of crime. Psychiatric Services 56: 25-27.

J.P.A. Ioannidis (2005) Why most published research findings are false. PLoS Medicine 2: 696-701; J.P.A.Ioannidis (2005) Molecular bias. European Journal of Epidemiology 20: 739–745.

This study has been approvingly discussed as an example of “new and improved” human behavior genetics by a number of scientist and non-scientist authors, including Hamer (2002, op. cit.), Parens (2004, op. cit.) Wasserman (2004, op. cit.), Appelbaum (2005, op. cit.).), and Farahany & Burnet (2006, op. cit.). 

A. Caspi et al. (2002) Role of the genotype in the cycle of violence in maltreated children. Science 297:851-854.

S.Z. Sabol, S. Hu, D. Hamer (1998) A functional polymorphism in the monoamine oxidase A gene. Human Genetics 103-273-279; J. Deckert et al (1999) Excess of high activity monoamine oxidase A gene promoter alleles in female patients with panic disorder. Human Molecular Genetics 8: 621-624; R.M. Denney, H. Koch, I.W. Craig (1999) Association between monoamine oxidase A activity in human male skin fibroblasts and genotype of the MAOA promoter-associated variable number tandem repeat. Human Genetics 105: 542-551.

J. Kim-Cohen et al. (2006) MAOA, maltreatment, and gene-environment interaction predicting children’s mental health: new evidence and a meta-analysis. Molecular Psychiatry 11:903-913.

R. Lewontin (2002) It Ain’t Necessarily So: The Dream of the Human Genome and Other Illusions, 2nd ed. New York review Books: New York.

E. Balaban (2006) Cognitive developmental biology: History, process and fortune’s wheel. Cognition 101:298-332.

R.E. Redding (2006) The brain-disordered defendant: Neuroscience and legal insanity in the twenty-first century. Villanova University School of Law Working Papers, Year 2006, Paper 61:51-127.

A. Abbot (2001) Into the mind of a killer. Nature 410: 296-298; A. Mandavilli (2006) Actions speak louder than images. Nature 444: 664-665.

A. Raine, Y. Yang (2006) Neural foundations to moral reasoning and antisocial behavior. Social, Cognitive and Affective Neuroscience 1: 203-213; M. Hauser (2006) The liver and the moral organ. Social, Cognitive and Affective Neuroscience 1: 214-220.

N. Rose (2000) The biology of culpability: Pathological identity and crime control in a biological culture. Theoretical Criminology 4: 5-34.

I. Hacking (2001) Degeneracy, criminal behavior, and looping. In: Wasserman D, Wachbroit R (eds) Genetics and Criminal Behavior. Cambridge University Press: NY, pp. 141-167.

T. Duster (2006) Behavioral genetics and explanations of the link between crime, violence and race. In: Parens E, Chapman AR, Press N (eds) Wrestling With Behavior Genetics: Science, Ethics, and Public Conversation. Johns Hopkins University Press: Baltimore, pp. 150-175.

ibid, pp. 166-171

The FBI and local law enforcement agencies are already using the services of DNAPrint Genomics of Sarasota, Florida, that sells what one of its founders once referred to as "racial inference" tests based on DNA sequence information (now referred to as "biogeographical ancestry makeup"); see S. Fink (2006)Reasonable Doubt: Questions about the forensic infallibility of DNA emerge even as police begin to use it to profile suspects by race. Discover Magazine 27 (7):54-59.