For most of the past century, analysis of the origins of crime has been dominated by sociological models. When Tony Blair declared in 1992 that his party would be "tough on the causes of crime," his audience presumed that he meant that Labour would try to eliminate crime-generating social ills such as poor housing, unemployment and inadequate schools. Discussion of the possible roots of offending and antisocial behaviour within individuals rarely formed part of elite public discourse. Punishment, the courts held, should be regulated by the severity of the crime, not the criminal's propensity to commit further offences.
One of the few challenges to this orthodoxy was made in the 1960s by Hans J Eysenck, for many years a professor at the Institute of Psychiatry. Eysenck believed that criminals' personalities could be rigidly categorised and that most of their behaviour was inherited. But his work on crime was attacked by mainstream sociological criminologists and had little influence on policy. Indeed, for most criminologists the concept of a personality more likely to commit crime was abhorrent.
The resistance to Eysenck was especially fierce because he was writing during the vogue for "radical criminology," when crime was seen as a social construct and the "labelling" of deviants an aspect of social control. Thirty years later, intellectual fashion has shifted beyond recognition, with, for example, a heavy new emphasis on the experiences of victims of crime. Nevertheless, investigation of the factors that put an individual at high risk of engaging in criminal and antisocial behaviour remains controversial, and most criminologists continue to steer well clear of it.
The act, and the new emphasis on risk assessment in general, entail a big shift from the principle that has governed sentencing in the past—that of punishment tailored to fit the crime, of proportionate "just deserts." Although it has been subjected to little public debate, this new approach requires penal decision-makers—other than those dealing with murder—to take a radical step: to assume some of the characteristics of the insurance actuary, and to base the length of incarceration on future probabilities. At the same time, the act contains an analysis of offending that departs significantly from sociological models. Under its terms, many of those judged to pose a high risk will have been assessed by forensic psychologists or psychiatrists, on the grounds that they exhibit a "dangerous severe personality disorder," or DSPD—a disorder that makes them likely to reoffend.
It is unfortunate that the term DSPD does not match any accepted clinical definition. Some of those who have already been so described are psychopaths—callous, emotionally affectless, careless of the damage their crimes cause to their victims. Others, however, have been diagnosed with conditions including borderline personality disorder and obsessive-compulsive disorder, as well as the much more common antisocial personality disorder. Nevertheless, the approach that the 2003 act represents poses important questions to which sociological theories of crime have no answers. Why do some people from deprived or abusive backgrounds become violent criminals, while others, whose upbringing appears to have been equally disadvantageous, go on to lead productive, law-abiding lives? Might there be ways to spot high-risk individuals before they commit serious offences, perhaps even in childhood? And are there interventions that might modify children's behaviour over the long term, diverting the course of those at high risk before they reach adulthood?
The focus on future risk requires a means to differentiate between individuals from similar environments. It places the offender, not the crime, at the centre of the penal decision-making universe, and asks those who make such sentencing decisions to base them on clinical assessments of the defendant's personality and its associated disorders. It hands great power over individuals' future to a group unused to wielding it—forensic psychologists and psychiatrists, and academic researchers in this field.
Even those most wedded to a sociological model of offending accept that a relatively small proportion of those convicted of criminal offences account for a very large proportion of total crime. David Farrington's study of every male born in Britain in 1956 found that as many as one third had been convicted of at least one non-traffic offence by the age of 30. But he and his colleagues also discovered that as few as 5 per cent were responsible for at least half of all known crime committed by the 1956 cohort. Other research suggests a group this small commits more than 70 per cent of all recorded offences, and more than 70 per cent of violent ones.
In a paper published in 1993, Terrie Moffitt, now at the Institute of Psychiatry, suggested that people convicted of crimes fall into two very different groups: those for whom offending is limited to adolescence, and those whose antisocial behaviour begins much earlier and then persists into middle age and beyond. The age of peak offending, Moffitt pointed out, is 17, and the majority of active criminals are teenagers. By their early twenties, however, the number of active offenders decreases by more than 50 per cent, and by 28, about 85 per cent of former offenders will have stopped committing crimes.
Moffitt cited several studies to propose that there is "a very small group of males who display high rates of antisocial behaviour across time." The same 5 per cent of boys first appear on the criminological radar screen in early childhood—being prone to aggression and violence, disobedience, recklessness, lying and theft. Even at primary school, they are likely to face exclusion and other sanctions. In adolescence, they are the youths who commit the more serious and violent crimes, and in adulthood they do not, like most of their peers, cease such behaviour. "The nomenclature may change," Moffitt wrote, "but the faces remain the same as they drift through successive systems aimed at curbing their deviance: schools, juvenile justice programmes, psychiatric treatment centres and prisons." An age of first arrest between seven and 11 is among the most reliable predictors of persistent adult offending.
The causes of "life-course persistent antisocial behaviour" are, wrote Moffitt, likely to lie "early in life, in factors that are present before or soon after birth." Behind the condition, she suggested, was an interactive process between some kind of neuropsychological condition and an individual's environment: "In a nurturing environment, toddlers' problems are often corrected. However, in disadvantaged homes, schools, and neighbourhoods, the responses are more likely to exacerbate than amend… thus over the years, an antisocial personality is slowly and insidiously constructed." Its emergence, she implied, was likely to be the consequence of complex interactions between nature and nurture, of genes and their environment. Among those whose antisocial behaviour was limited to adolescence, its origins were more restricted, arising from teenagers' vulnerability to peer pressure, their search for popularity and the frustrating quest for the resources and privileges available to adults which they were denied.
Later research in several countries has supported Moffitt's classification. Some of the most important work has come in a series of papers by Moffitt herself and her colleagues, drawn from the Dunedin multidisciplinary health and development study. This looked at a group of over 1,000 men and women born in New Zealand between April 1972 and March 1973, who have been assessed regularly ever since.
The Dunedin study reveals that boys and men who engage in life-course persistent antisocial behaviour are likely to have experienced many disadvantages. (Similar females have been omitted from this analysis because their numbers are so small—a fact which is likely to reflect both nature, in the shape of genetically determined propensity, and nurture, socially constructed gendered roles.) In childhood, between the ages of three and 13, the Dunedin boys with the worst conduct problems at home and school also displayed "neurological abnormalities, low intellectual ability, reading difficulties, hyperactivity, poor scores on neuropsychological tests, and slow heart rate." They were often being emotionally rejected by their parents as young as the age of five, as well as by their peers and teachers. In contrast, those whose offending and antisocial behaviour did not begin until adolescence tended not to suffer from these disadvantages. Although both groups behaved badly in adolescence, those on the life-course persistent path were also likely at that stage to experience weak bonds with other family members, continued educational difficulties and poor relationships. Their criminal convictions were more serious, and more often for crimes of violence.
In Dunedin, the 10 per cent of boys whose antisocial behaviour had started before adolescence were about three times as likely as the "adolescence- limited" group to be convicted of crimes after the age of 26, and they "tended to specialise in serious offences." Although some from the larger "adolescence-limited" group were still committing offences as adults, these tended to be relatively trivial. The difference between the groups was especially marked when it came to using violence: the 47 men deemed to be on the life-course persistent pathway accounted for five times their statistical share of all violent crime.
Not only were they committing much more crime, their lives were in much worse shape in other ways. At 26, they were much more likely than the "adolescence-limited" group to be abusing alcohol, and to have suffered symptoms of schizophrenia, paranoia and depression. They were more likely to have abused their partners, and while they had fathered more children, they were less likely to be helping to rear their offspring. More than half had no high-school qualifications, and only one of them had attended college.
A persistent antisocial personality is, to put it mildly, a daunting prospect. However, among those who fit this definition is a still more intractable minority: the group defined as psychopaths. There are no studies of their prevalence in the general population, but it is thought that they account for 20 per cent of all male prisoners. Their classification derives mainly from the work of Canadian psychologist Robert Hare. In his account, psychopaths are callous, emotionally "flat," driven solely by their own needs and welfare, manipulative, and careless of the suffering they inflict. Moreover, they appear indifferent to punishment, so that the threat of penal sanctions has little effect on them. According to Hare, psychopaths are between four and 12 times as likely as others to be reconvicted of a violent crime within two years of release from prison, while more than 80 per cent will have reoffended within four years.
There is now some evidence that suggests that like "regular" antisocial personality disorder, psychopathy may be discernible in children aged seven or younger. One marker is the emergence of "callous and unemotional" personality traits. Another is the presence of unprovoked aggression. Most aggressive children show "reactive" or impulsive aggression in response to a provocation or threat. In contrast, the markedly rarer proactive or "instrumental" type of aggression is unprovoked, and "typically involves planning and forethought."
The scientists working on the origins of persistent antisocial behaviour are not looking for "reductionist" or deterministic explanations of the kind sought by Eysenck. As the Institute of Psychiatry's Michael Rutter writes, much previous writing on mental and behavioural disorders has "made the implicit assumption that there is likely to be just one causal pathway" for each disorder, implying that the challenge for research is to find it. Instead, he states: "In the great majority of cases, both psychological traits and mental disorders are multi-factorial in origin."
The Dunedin study supports this conclusion. Moffitt and her husband, Avshalom Caspi, have examined some of the environmental childhood risk factors that appear to predict persistent antisocial behaviour, and found them to be diverse. They included harsh and inconsistent parenting that failed to reward good behaviour, conflict within the family, repeated changes in the main person providing childcare, and single parenthood. Also significant was a wide range of cognitive and neurological deficits, which meant children talked later and had difficulty in learning to read and write, as well as displaying hyperactivity, impulsive behaviour and anger—all factors that contributed to an infant personality that was difficult to control. In contrast, the backgrounds of those whose antisocial behaviour was confined to adolescence were found to be normal.
Studies of twins and adopted children had already established that antisocial behaviour is likely to be partly inherited. Identical twin pairs, who share the same genes, have been shown to be more likely to share antisocial traits than fraternal, non-identical ones. The children of criminals, even when adopted at an early age by non-abusive and non-criminal families, are more likely than average to become offenders themselves.
In a paper published in 2002, Moffitt, Caspi and some of their colleagues discuss why there are large differences between children in their response to maltreatment: "Although maltreatment increases the risk of later criminality by about 50 per cent, most maltreated children do not become delinquents or adult criminals." The authors turned to evidence that neurotransmitters—the chemicals such as serotonin, norepinephrine and dopamine that relay and amplify electrical signals between the brain's neurons—play a significant role in the generation of mood, behaviour and general mental health. Serotonin and norepinephrine are also known to be involved in multiple brain functions that regulate responses to stress.
The paper's hypothesis was that one of the factors that differentiates individuals' propensity for antisocial behaviour is a particular gene—the one responsible for generating the enzyme monoamine oxidase A (MAOA). This enzyme regulates neurotransmitter levels in the brain: one of its roles is to get rid of excess serotonin, dopamine and so on, in order to keep neurological circuits working smoothly.
In fact, there are five known variants—known as alleles or genotypes—of the MAOA gene, although three of them are rare. The authors of the 2002 paper examined the two main types. The low-activity allele, which programmes the body to produce low levels of the MAOA enzyme, is found in about one third of males. The more normal, high-activity allele is found in almost all of the rest. In order to test their hypothesis about the role of MAOA, the researchers went back to the Dunedin cohort. Its members' history had already been examined and described, so that it was already known that between the ages of eight and 11, 8 per cent of the cohort's children had suffered "severe" maltreatment, and 28 per cent had experienced "probable" maltreatment. As we have seen, the team already knew which members of the study had exhibited antisocial behaviour, and when. Now researchers also found which of the MAOA genotypes they had by examining their DNA.
As might have been expected, the Dunedin study found that maltreatment in childhood would, on its own, make someone more likely to commit crime and display antisocial behaviour. About 35 per cent of the maltreated men with the normal high-activity genotype had shown conduct disorder, and 20 per cent had a conviction for violence. But when the two risk factors were found together—the low-activity genotype and childhood maltreatment—the correlation with antisocial behaviour was far stronger. More than 80 per cent of the men in this category had exhibited conduct disorder, and more than 30 per cent had convictions for violence. As a group, they were all among the most violent third of men. No fewer than 85 per cent of the cohort's men with the low-activity genotype who had also been severely maltreated went on to develop antisocial behaviour.
In other fields of study, a correlation of this size would be considered significant. It is not a complete explanation of antisocial behaviour: after all, some of those in the life-course persistent group had the high-activity allele. The MAOA gene is likely to be one of many, each with its own pattern of environmental interaction, that plays some part in the development of antisocial behaviour. On the other hand, the risk among maltreated men with the low-activity MAOA genotype was about the same as the risk that a person with high levels of cholesterol has of developing heart disease and greater than the correlation between low bone density and fractures. To put it another way, it seemed that having the high-activity genotype conferred a measure of protection from the consequences of childhood maltreatment. It was possible, the paper's authors said, that this finding might inform future developments in pharmacology: a drug to regulate violent behaviour. To date, there has been no research on whether the effects of a low-activity MAOA genotype might respond to treatment, or how this might be achieved.
Studies of gene-environment interplay are notoriously hard to replicate, but in this instance, replication was not long in coming. By 2005, there had been four further papers that confirmed the original findings, the last by members of the original Dunedin team.
The huge social and financial cost of failing to prevent the onset of antisocial personality disorder and persistent offending may be taken as given. Prison gates are all too often a revolving door, through which offenders return within months of regaining their freedom. In recent years, the British government has invested heavily in jail-based cognitive-behavioural programmes, designed to improve offender's thinking skills and to curb their impulsiveness and tendency to use violence. So far, however, the evidence suggests that the quality of their delivery is patchy and their value in reducing rates of reconviction marginal, especially with prisoners at highest risk. Attempts to reduce recidivism among adults are not futile, and sometimes the simplest interventions—such as providing the basic skills of numeracy and literacy—have positive results. Many of those whom one might class as life-course persistent offenders in fact desist in middle age: even psychopaths often cease their most damaging activities in their forties.
Nevertheless, in Moffitt's words, there is a sense that when one tries to modify adult antisocial behaviour, "the horse is already out of the barn." It seems reasonable, she and her colleagues argue, to target attempts to restrict the development of antisocial behaviour on parents and on children themselves.
Research into parent training programmes is now relatively advanced in Britain and the US. In December 2005, its value was given the official imprimatur of the National Institute for Health and Clinical Excellence (Nice). The Institute of Psychiatry's Stephen Scott, one of the co-authors of the Nice evaluation and a doyen in the field, summarises the state of knowledge as to what constitutes effective parent training. Parents, he says, make a much bigger difference than teachers in the development of children's social skills, and by the same token—as the Dunedin and other studies indicate—poor parenting has a devastating impact. What seem to work, according to several studies by Scott and others, are rigorous behavioural approaches, administered in groups over a period of about three months. For some parents, the basic elements of this training—the setting of boundaries and consistent rules, the swift application of appropriate disciplinary sanctions and the giving of praise when things go well—are second nature. For others, it is clear, they are not.
Training, says Scott, is not effective in all cases: about a fifth of the children whose parents are trained will fail to improve significantly. But there is reason to believe that many families and children can be rescued from a vicious spiral, and placed in a virtuous one: "After the intervention, parents were giving their children far more praise to encourage desirable behaviour and more effective commands to obtain compliance," says Scott in one paper. Their relationships were significantly better.
It is possible that some of the children whose behaviour did not improve in this and other studies may also be those who exhibit youthful signs of psychopathy. If so, this would not be surprising—psychopaths do not fear punishment and so are unlikely to show concern over family sanctions, such as a "time out" for misconduct. Even here, however, the picture may not be hopeless. Not all children with conduct disorder and "callous and unemotional" (CU) traits will become adult criminal psychopaths, and while CU children do not respond to punishment, they do appear to modify their behaviour when good conduct earns rewards.
Once children reach adolescence, parent training becomes much less effective. However, researchers at the Institute of Psychiatry have recently begun to develop means of intervening with children of this age. Working with schools in London, and focusing on special institutions for those excluded from ordinary education, the institute has devised a short cognitive-behavioural course aimed at aggressive teenagers. The programmes are continuing, but the first, encouraging, results of a randomised controlled trial have just been submitted for publication.
The academy can be a compartmentalised place, with surprisingly little dialogue between disciplines, and mainstream sociological criminology is only beginning to become aware of the work described here. It may not evoke a favourable response. A recent issue of the journal Criminal Justice Matters, published by the Centre for Crime and Justice Studies at King's College London, contained a fierce attack on the work of Terrie Moffitt and others. The article accused researchers at the Institute of Psychiatry and elsewhere of "genetic fundamentalism—a belief in a mythic, not a real genetics," and suggested that twin studies that found a genetic component in antisocial behaviour were without value. Moffitt and her colleagues have, in fact, stressed that genetic predispositions must be "switched on" by childhood maltreatment, and that the important thing was to concentrate on eliminating this and other types of adverse environment.
Asked to give an after-dinner speech to Liberal Democrat lawyers, I caught a different glimpse of the hostility that behavioural genetic research into the causes of crime can evoke. After I had presented an account of some of the work described here, the response was viscerally critical. Speakers claimed that it was "deterministic," and would surely lead to a wanton attack on civil liberties. One distinguished legal practitioner went so far as to demand who had funded these investigations, claiming that they must have been cooked up according to some pre-ordered, authoritarian agenda.
None of this research is deterministic. Twin studies, work on gene-environment interactions and controlled trials of interventions all describe correlations, risks and probabilities, not inevitable consequences. It may be likely that a callous, unemotional child exhibiting severe conduct disorder will, left alone, end up as an adult criminal psychopath. It is not, however, certain, and any policy consequences of this work must above all else bear this in mind. At the same time, it seems intellectually bizarre to disregard the work, let alone to refuse to continue with lines of research that are in their relative infancy.
What, then, might the policy consequences be? One conclusion seems to emerge strongly: that early interventions with children and young people identified as members of groups at risk deserve substantial investment and appear to be very promising, offering the potential for cost-effective ways of reducing enormous individual and social harms. It seems difficult to argue that to leave antisocial personalities to develop without attempts at intervention, and thus to wait until they present themselves as violent adult criminals, is a sensible public strategy.
Such measures will not, however, be extended without controversy: some will find in them a version of the dystopic, authoritarian future depicted by Anthony Burgess in A Clockwork Orange. The many supporters of "just deserts" ideology will attack parent training programmes as punitive, and be likely to stigmatise those who undertake them. They will question the social, racial and psychological assumptions that underly diagnostic tools. The work described here is far removed from the eugenicist Francis Galton, or indeed Eysenck, but the memory of 20th-century history will not easily be shifted.
However, as we move from considering possible treatments for young people who are yet to commit serious crimes to adults who already have, the implications of the work described in this paper are likely to become even more contestable, and potentially, more troubling. On the one hand, the deepening understanding of psychopathy and persistent antisocial behaviour has enormous potential as a risk-assessment tool. It has obvious applications in the management and supervision of offenders within prison and after their release, and in the imposition of sentences. It is likely to enable probation and other offender management services to target their resources much more effectively, and to set parole licence conditions tailored more closely to the needs and risk profile of individuals.
But criminal justice policy is not derived from scientific evidence alone. It is conceived in a broader social and political setting. This includes several volatile elements, above all governments' desire to be seen as tough on crime and their need for quick results. This year has seen a string of high-profile cases in which offenders on probation have committed murders and sex crimes. One of the most horrifying, the rape and murder of the Reading teenager Mary-Ann Lenaghan, prompted one Sunday Times writer to ask: "If someone has not committed a crime, should we lock them up just in case? Tom Cruise in Minority Report did not like being hunted down for a murder he had not yet committed, but from the point of view of society wouldn't that be a good thing? Is it time our criminal justice system did more to investigate 'pre-crime?' If we did, Mary-Ann Lenaghan might still be alive."
We already have the 2003 Criminal Justice Act and its indefinite sentences, and in some US states, " sexually violent predator" laws consign some high-risk inmates to what amounts to lifelong preventative detention. One could envisage an ambitious home secretary, careless of the distinction between prediction and probability, who might decide that since the police already take DNA samples on arrest, it would be a simple thing to test those accused of violent or sexual crimes for their genotype.
For those on the receiving end, such "actuarial justice" fundamentally changes the premises of the law, and may in time create unmanageable control problems among prisoners left with few incentives to conform. Even psychopathic prisoners have a keenly developed sense of justice when it is applied to them, as I discovered during a discussion in 2005 with inmates at the Frankland prison DSPD unit near Durham. When they asked me to explain the provisions of the 2003 act, their reactions were anger and disbelief. "How can they keep you inside when you've already done your time? Surely it's against the law?" asked one, a serial violent offender.
Moreover, many persistent criminals eventually stop offending, although they may well retain other antisocial behavioural and personality traits. Existing tools do not give us much clue as to when an individual's offending may stop. Risk science may tell us that 80 per cent of criminal psychopaths will re-offend within four years of leaving prison. But even that dismal statistic leaves 20 per cent who will not.
In a paper to be published in a forthcoming volume on security and human rights, the criminologist Lucia Zedner sets out further possible pitfalls. Slowly, she contends, Minority Report's "pre-crime" is becoming a policy reality. Sentences or further detention based on risk assessment are "counterproductive to any effort to hold offenders to account, to convey blame, still less to engage their moral reasoning in potentially penitent… directions." If, in tabloid caricature, sociological criminology allowed offenders to shirk responsibility by blaming society, the new behavioural science might offer them the no less questionable excuse of blaming their crimes on the interaction between their parents and their genotype.
Evaluation of the implications of the work described in this essay has barely begun, and it needs to become the subject of a substantial interdisciplinary debate. Sociological criminologists must realise that to dismiss it as crude, positivist determinism, as their forebears once dismissed Galton and Eysenck, is inadequate. Equally, behavioural scientists require an awareness of the social, political and legal context in which their findings land. What is certain is that as knowledge of the human genome and its interaction with its environment expands, this is a field of study that will not go away.
