Abstract
Male antisocial behavior is concentrated in the adolescent period of the life course, as documented by the curve of crime over age. This article reviews recent evidence regarding the hypothesis that the age-crime curve conceals two groups with different causes. Life-course persistent males show extreme, pervasive, persistent antisocial behavior from early childhood to adulthood. They are hypothesized to be rare, with pathological risk factors and poor life outcomes. In contrast, adolescence-limited males show similar levels of antisocial behavior but primarily during the adolescent stage of development. They are hypothesized to be common and normative, whereas abstainers from offending are rare. This article recaps the taxonomy’s 25-year history, concluding that it is standing the test of time in research, and making an impact on policy in early-years prevention and juvenile justice. Research is needed into how the taxonomy relates to neuroscience, health, genetics, and changes in modern crime, including digital crime.
Introduction
Male antisocial behavior harms victims, perpetrators, and society, and it is heavily concentrated in the adolescent stage of life. This article reviews new research stemming from a developmental taxonomy of antisocial behavior 1. Twenty-five years ago the taxonomy outlined two hypothetical prototypes: life-course persistent versus adolescence-limited (hereafter LCP and AL). According to this taxonomy, LCP individuals’ antisocial behavior has its origins in neuro-developmental processes and family adversity, beginning in childhood, building persistently thereafter, and continuing into midlife. In contrast, AL delinquents’ antisocial activities have their origins in age-graded social processes that begin with a maturity gap in adolescence and end when social adulthood is attained. According to the taxonomy, LCP antisocial behavior is rare, persistent, pervasive, and pathological, whereas AL antisocial behavior is common, relatively transient, situational, and near normative (Figure 1). This article emphasizes research since 2005, when the last comprehensive reviews appeared (e.g.2–5). The review is limited to research on male LCP and AL behavior; related topics such as girls’ delinquency or childhood-limited conduct disorder are omitted because findings have not reached consensus. Coverage is not exhaustive; instead this article selectively covers empirical research supported by rigorous designs, strong samples, and/or new methods. It identifies what’s new and flags what still needs to be learned.
The historical context of the developmental taxonomy of antisocial behavior
Before looking into what’s new, it is useful to revisit the 1980’s, to understand why at that time there was a need for a taxonomy of youth crime that juxtaposed two types of offenders, each framed in its own theory. Two influential reports had drawn scholarly attention to the importance of change in rates of antisocial behavior across age. Epidemiologist Lee Robins 6 reported an apparent paradox that puzzled psychiatry: Antisocial adults virtually always begin as children with antisocial misconduct, but most young people who engage in antisocial misconduct do not grow up to be antisocial adults. Criminologist Al Blumstein, leading the U.S. National Academy of Sciences panel on crime careers 7, enumerated the replicated facts about crime at that time. First, a small fraction of the population commits a large fraction of crime, a fact recently reconfirmed by meta-analysis 8. This fraction was termed “chronic offenders”. Chronic offenders are a subset of the 30%–40% of males convicted of non-traffic crimes in developed nations (Skardhammer, 2014). Second, chronic offenders tend to have a younger onset age, and younger onset predicts higher offending frequency, longer career duration, and thus a large share of offenses. Chronic offenders also commit a broad repertoire of crime types, including violence. Third, Blumstein’s panel confirmed that offending followed a curve over age, which became known as “the age-crime curve” (Figure 1a). Onset of illegal behavior was typically between ages 8–14 years (later in official data than in self-report data), and desistence from offending was typically between 20–29 years. The peak age of offending occurred in-between: among 15–19 year olds, when upwards of 90% of males break laws, according to self-report cohort studies1. These were the known facts at the time the LCP/AL taxonomy was developed. Two possible explanations were advanced to account for this adolescent crime peak: it could result from the few chronic offenders escalating their personal crime rate between ages 15–19, or it could result if the few chronic offenders were joined by large numbers of young people who offended around age 15–19 and then desisted within a few years. The developmental taxonomy was developed to explain the age-crime curve (Figure 1b).
Until the early 1990’s, biological and psychological theories did not account for the age-crime curve’s adolescent onset and peak, while simultaneously, sociological theories did not account for the age-crime curve’s steep post-adolescent drop of crime desistance. Psychobiological researchers typically conducted their studies in child clinics or adult prisons and thus trained their lenses on early childhood or adulthood, neglecting adolescence. They typically focused on the long-term stability of individual differences in traits such as impulsivity, neuroticism, autonomic nervous-system reactivity, or low intelligence, but psychobiological trait theories could not explain the adolescent peak of the age-crime curve without positing a sudden and dramatic population upshift in criminogenic traits followed by return to baseline a few years later. In contrast sociologically-oriented criminologists favored studying the adolescent age period. Historically, reliance on legal definitions of antisocial behavior had kept these delinquency researchers preoccupied with the question of why illegal offending onsets in adolescence. Adolescence was when crime peaks, and therefore offending could be most conveniently studied by administering surveys to adolescents in high-schools. But the resulting sociological research failed to address the long-term stability of antisocial behaviors that begin years before adolescence, even in the preschool years. In addition, most sociological crime theories at that time invoked causes such as low social class, school drop-out, cultural approval of violence, and deviant labels. However, such causal factors do not disappear in time to explain the marked downward shift in offending that ends the age-crime curve.
The developmental taxonomy was developed to integrate psychobiological and sociological theories. It proposed that psychological and biological theories applied best to LCP offenders, who behave in antisocial ways during childhood, adolescence, and adulthood (Figure 1c). It proposed that sociological theories applied best to AL offenders, whose antisocial acts elevate the age-crime peak (Figure 1d). The 1993 article also integrated into one developmental taxonomy the many disparate measures of behavior which violates the rights and safety of others: sociology’s self-reported and officially registered offending, and psychobiology’s childhood conduct problems and adult antisocial personality.
Research designs for testing the developmental taxonomy of antisocial behavior
The original 1993 article made testable predictions, and listed research design desiderata to test them. Samples should be population-representative to capture the population range of natural histories. The same individuals should be studied longitudinally to trace trajectories of antisocial behavior within individual lives, ideally starting in childhood and continuing into adulthood. Measures of antisocial behavior should allow for the emergence of new forms of antisocial behavior (e.g., automobile theft in adolescence, intimate-partner violence and workplace deviance in adulthood) and the forsaking of old forms (e.g., childhood tantrums, truancy). Reports of antisocial behavior should be gathered from multiple sources and settings to tap pervasiveness. In addition to official measures of offending, research should also analyze self-reports and informant reports to ascertain childhood-onset of antisocial conduct and to insure coverage of antisocial behaviors not captured in official crime data. Risk factors should be measured prospectively to avoid confusing consequences of an antisocial lifestyle with its causes. Among studies that met this challenge is the Dunedin Study 9, which followed a representative birth cohort of 1,037 children born in 1972–73 across the age-crime curve (Box figure 1).
Predictions from the 1993 taxonomy; how have they fared?
Prediction 1: The age-crime curve conceals two groups
The 1993 article stated that fewer than 10% of males should show extreme antisocial behavior that begins during early childhood and is thereafter sustained at a high level across time and circumstances. A much larger percentage of males should show similar levels of antisocial behavior during the adolescent age period but should lack a childhood history of stable, pervasive problem behavior.
What’s new? Group-based trajectory methods
About the time the developmental taxonomy was put forward, methods to test it in repeated-measures datasets were developed 10. Chief among them were modelling techniques designed to detect hypothesized groups with distinctive developmental trajectories of behavior within a population 11. Since the advent of group-based trajectory modelling methods, the existence of trajectory groups fitting the LCP and AL taxonomy has now been confirmed by reviews of more than 100 longitudinal studies 12–15, including the Dunedin Study 16,17, plus more recently published studies 18–21. Studies vary in sampling, geographic area, historical period, phases of the life-course observed, length of observation period, and data sources analyzed, and thus understandably on the number and shape of trajectories reported. However, reviews conclude that results are reasonably consistent with the taxonomy, detecting heterogeneity in the temporal course of offending, and pointing to a minimum of three groups: a low or non-offender group, a group whose offending peaks in adolescence but drops as adulthood begins, and a chronic/high-rate group that continues offending into adulthood.
What’s new? Digital crime
The internet and related communication technologies have altered the behavioral landscape, including the landscape of methods young people use to victimize others 22,23. The taxonomy was proposed in an era when the adolescent crime peak meant “street crimes,” such as shoplifting, house-breaking, and vehicle theft. But virtual internet-enabled crimes do not require physical strength or proximity; e.g., drug trafficking is accomplished via Telegrass. Research is needed on the taxonomy in digital natives.
Prediction 2: In adulthood AL and LCP groups go separate ways
The 1993 article stated that AL delinquents can profit from opportunities for desistence, because they retain the option of successfully resuming a conventional life-style. LCP delinquents may make transitions into marriage or work, but their injurious childhoods make it less likely that they can leave behind their longstanding antisocial style of solving life’s problems; they should express antisocial behavior at home and at work
What’s new? Cohorts followed through adulthood
When the taxonomy was published, few cohort studies of antisocial behavior had reached adulthood, rendering them unable to test whether LCP and AL groups have different outcomes 24. What’s new is that more cohorts have now collected data on duration of offending careers at least to midlife, when the age-crime curve of officially recorded crime drops to very low. The results of these long-term follow-ups are broadly consistent with the hypothesis that the age-crime curve comprises a small group of chronic offenders whose crime careers extend to at least midlife and a larger group of short-term offenders whose careers end younger. Prospective longitudinal studies from multiple countries that have tested the taxonomy while following cohorts since childhood report that LCP groups show poorer life outcomes into their late forties 25–27. Two studies extend the follow up period into the fifties. The Cambridge Study in Delinquent Development followed 411 London males from age 8 to age 56 using official conviction records and self-report interviews 20. Trajectory modelling revealed an adolescent high-peak group and a chronic high-rate group. Members of the chronic group had poor mid-life outcomes and were still being convicted at age 56. The most recent study that modelled crime trajectories up to age 51 found a small chronic high-rate offender group, men characterized by a childhood history of abuse and neglect 21. Two remarkable studies tested the taxonomy in offenders followed to age 70. A Netherlands study of over 5,000 offenders detected a small group still being regularly convicted in their seventies 28. In the Gluecks’ cohort of reform-school boys, trajectory models confirmed the existence of chronic offenders and early crime desistors, but the authors emphasized that all men’s offending declines precipitously with age; even chronic offenders rarely offend into late life 29.
What’s new? Many LCP offenders end up with ill-health or incarceration
Mid-life followups revealed an important finding unanticipated by the 1993 taxonomy: the LCP lifestyle often culminates in illness, hospitalization, sickness disability, and premature mortality 16,30–32. Antisocial lifestyles probably damage health, but alternatively, childhood neurodevelopmental risks may signal weak system integrity of the body, which emerges as poor adult health. Furthermore, the LCP antisocial lifestyle tends to land offenders behind bars, as shown in the Dunedin Study (Box), and other studies 21,28. Excess incarceration, sickness disability, and premature mortality preclude offending, selectively removing LCP offenders from criminal-record data. Studies testing the hypothesis of life-course persistence should measure illness and incarceration.
What’s new? No adult-onset antisocial behavior
Some studies of official crime-record data point to offenders first convicted as adults. The taxonomy asserted that LCP and AL offenders account for the bulk of crime under the age-crime curve; it did not address the possibility of an adult-onset group. However, this possibility has now been tested in several longitudinal studies that analyzed official conviction records alongside parent-, teacher-, or self-reported offending measures taken before the age when conviction is possible. They report that so-called adult-onset offenders, in fact, had clear histories of childhood conduct problems and juvenile offending, undetected by police 24,33.
What’s new? Surrogate methods to define LCP and AL
Research on taxonomy has revealed urgent need for a way to discriminate between LCP and AL persons at one time-point. Many researchers lack access to child-to-adult longitudinal data, and practitioners often lack prospective childhood information to decide if a client is on an AL versus LCP trajectory. Several expedients suggest themselves, including elevated scores on aggression 34, callous-unemotional traits 35, comorbid conduct disorder and hyperactive-impulsivity 36, consensus across reporters that antisocial behavior is pervasive across settings 37,38, and family history of substance-dependence 39. All have empirical merit, yet none has sufficient evidence to justify its use alone as a surrogate for child-to-adult data. Research to build a one-time-point assessment tool is needed.
Prediction 3: AL and LCP groups differ on childhood risk factors
The 1993 article stated that the strongest prospective predictors of persistent antisocial behavior are individual characteristics (e.g., difficult temperament, neuropsychological deficits, hyperactivity) and family characteristics (e.g., socioeconomic deprivation, poor parenting), but not age. Individual and family differences should play little or no role in short-term adolescent offending careers.
What’s new? More cohorts support the risk hypothesis
Accumulated research showing that LCP offenders have differential childhood risk factors has been described in reviews 2,4,5,40,41. A query has emerged: Whether AL offending shares the same risk profile as LCP offending, albeit at a lower level, which could indicate a quantitative rather than qualitative group difference in risk 41–43. This query provides an opportunity to clarify here that the taxonomy conceptualizes LCP offenders as abnormal on childhood risk factors, but AL offenders as ordinary, average, normative. During adolescence, large numbers of young men who come all varying backgrounds, from very low levels of risk to risk levels somewhat above the mean, engage in offending and join the AL group. The main risk factor for the AL group is the maturity gap itself, which is not a risk factor for LCP offending. LCP’s should show extreme risk on childhood risk factors, while AL’s should show population-normative levels (population-normative risk levels alas are not zero). Comparisons show that the LCP group scores significantly above their cohort mean on risk factors while the AL group scores near the mean (Box figure 2, panels a and b). It is the “low,” non-antisocial comparison group who scores significantly below the mean on risk factors 43,44. Earlier I reported an exception: Dunedin AL’s had above-normal sensation-seeking, a personality risk unanticipated in the 1993 article 4.
In recent years, some very large nationally representative surveys have weighed in on the taxonomy. The Collaborative Psychiatric Epidemiology Studies (CPES) of over 20,000 Americans reported that the LCP group had low childhood socioeconomic status, lack of maternal closeness, and a history of harsh discipline, relative to the AL group 45. The National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) in over 43,000 Americans identified 5% of respondents who had a greater variety and severity of antisocial behaviors characteristic of LCPs, and also had more extreme sociodemographic, psychiatric, and behavioral risk correlates, compared to other respondents (although this study did not define an AL group) 46. CPES and NESARC had the shortcoming that lifetime antisocial behaviors were retrospectively ascertained in a cross-sectional adulthood interview. The 1970 British Cohort Study of over 16,000 respondents measured child, parent and socioeconomic risk factors prospectively up to age 5 years, and found they predicted conviction at age 34 years 47. This study did not designate LCP or AL groups, but it is valuable for its very young risk measurements in a population-representative sample.Additional longitudinal studies with sample sizes from 400 to 1,000 that were well-designed to define LCP and AL groups have reported that they differ on child and family risks measured prospectively during childhood 16,20,21,39,43,48. As one example, the Pittsburgh Youth Study investigated parenting 49. For teens on an LCP path, parent-child communication was unhealthy from the earliest years. For teens on an AL path, parent-child communication was healthy in childhood, but after the teenager began offending it deteriorated.
What’s new in risk research? Genetics
The 1993 article noted that LCP children should be distinguished by traits that are heritable, such as low verbal ability and low self-control, but did not explicitly mention genes. Until recently, researchers who tested for genetic influence on behavior relied on biometric studies of twins or adoptees. A 2005 review of such studies 50, plus subsequent studies 34,38,51,52, provide evidence of elevated genetic influence on LCP-type antisocial development, as compared to AL.
Today, advances in the genomic sciences make it possible to investigate genetic influences on behavior at the molecular-genetic level, using genome-wide association studies (GWAS) 53. The effect size is miniscule for any single GWAS-detected genetic variant, but it is possible to aggregate the effects of millions of variants across the genome, thereby constructing a polygenic score which indexes each person’s genetic propensity along a continuum. A very large GWAS recently yielded a polygenic score for highest educational degree 54. We hypothesized that this educational-attainment polygenic score would be low among LCP individuals because genetically influenced traits that promote LCP development, such as low verbal ability and low self-control, also reduce educational attainment 55. As hypothesized, the Dunedin Study LCP trajectory group had a significantly lower mean polygenic score than the AL group (whose mean polygenic score was cohort-average). The educational-attainment polygenic scores’ prediction of offending careers was mediated by primary-school risk factors: lower cognitive abilities, lower self-control, academic difficulties, and truancy. Soon polygenic scores for conviction itself will be derived from GWAS 56, which can be tested for LCP-specific risk.
What’s new in risk research? Neuroimaging
Many well-designed studies have shown that abnormalities on neuropsychological tests are prospectively associated with LCP but not AL patterns of antisocial behavior 4,57–59. Thus, there is good evidence that LCP behavior is characterized by difficulties in the brain’s mental functions, particularly its verbal and executive functions. Advances in neuroscience now make it possible also to investigate brain-behavior associations using magnetic resonance imaging techniques (MRI). Children can now be scanned, so prospective neuroimaging data could feasibly be compared for LCP versus AL groups. However, at this point, what can be confidently said is that neuroimaging studies of children whose antisocial behavior onsets in childhood (variously defined) tend to show abnormal MRI findings of both brain structure and brain function during mental tasks 42,60–62.
Despite confirming that childhood antisocial behavior is accompanied by neural abnormalities, decisive tests of the hypothesis that neural abnormality should characterize LCP but not AL individuals are lacking. This lack is because neuroimaging studies are costly, precluding them from employing research designs suitable for risk-factor research 63,64. Designs in LCP/AL neuroimaging comparisons so far have used small unrepresentative middle-class samples, and been under-powered, cross-sectional, and retrospective 65. Some comparisons between a childhood-onset versus an adolescent-onset conduct disorder group yielded small group differences 66–68, but other comparisons in the same sample yielded nil group differences 42,66,68,69. However, it is not possible to evaluate reports of either hypothesis confirmation or disconfirmation on the basis of studies with insufficient designs. Ideally the design requirements of neuroimaging studies must be the same as for studies of any other risk factor: population-representative samples, adequate statistical power, repeated measures of antisocial behavior beginning in childhood, and prospective neuroimaging before adolescence to establish that MRI abnormalities are antecedent, not consequential to offenders’ lifestyles. In future, LCP/AL comparisons may be accomplished in large cohorts of young people followed longitudinally with repeated neuroimaging, such as IMAGEN. Unfortunately, many such cohorts recruit adolescent participants, omitting childhood MRI scans.
It is useful to mention that the field of developmental neuroscience reports MRI findings that are associated with risk-taking during adolescence 70–73. In a nutshell, the argument is that a lack of synchrony in brain maturation characterizes adolescence, thereby explaining why adolescence is a life-stage of excessive risk-taking and susceptibility to peer influence, as compared to childhood or adulthood. This approach points to a population-wide upshift in neural processes that promote delinquent offending, followed by a downshift a few years later, therefore fulfilling the requirement for a causal factor in AL offending. A normative brain maturity gap should be conceptualized among the causes of AL offending, an addition that could not have been foreseen in 1993. Research is needed to integrate new developmental neuroscience findings into the sociological theory of AL offending. Moreover, developmental neuroscience, which typically compares age groups, has not yet tackled brain-behavior differences between individuals who are on different trajectories of offending to test the LCP hypothesis.
Prediction 4: Adolescence-limited antisocial behavior in a maturity gap
When the existence of AL offenders was proposed to account for the peak in the age-crime curve, an explanation of their origin was required. The 1993 article stated that the strongest prospective predictors of this short-term offending should be attitudes toward adulthood and autonomy, knowledge of peers’ delinquency, cultural and historical context, and age. The hypothesized motivation that drives AL offending is a gap between biological and social maturation. Evidence that this maturity gap is associated with adolescent offending has been presented earlier 4,74,75.
What’s new about the maturity gap?
More studies have added support for the maturity-gap hypothesis. In 6,500 participants of the National Longitudinal Study of Adolescent Health, the maturity gap was measured as the discrepancy between pubertal stage and the amount of autonomy each teen reported 76. As hypothesized, the width of this measured gap predicted drug use and minor delinquent offenses, but not the violent offenses characteristic of LCP offending. Similarly, in the Social Network Analyses of Risk in Early Adolescence Study, the width of the gap between pubertal stage and decision-making autonomy predicted substance use and delinquency 77. In the Cambridge Study in Delinquent Development, a financial version of the maturity gap predicted offending among non-LCP participants 78.
One question relevant to the maturity gap hypothesis is whether AL offenders age out of crime and into conventional lifestyles as predicted. Studies following cohorts into young adulthood showed AL’s had continued continued difficulties, particularly related to substance abuse. More recent follow-ups into midlife of the Cambridge Study on Delinquent Development (Piquero, Farrington, Nagin, Moffitt, 2010) and the Dunedin cohort (Box figure 3, Box figure 4), indicate that more AL offenders have ordinary life outcomes. Nevertheless, this recovery is delayed beyond anticipations, a delay that recommends interventions to help AL offenders recover.
Something else is new that was unanticipated by the taxonomy: adolescence lasts longer. Since the 1980’s the transition from adolescence to adulthood has elongated, and traditional milestones of adulthood, such as completing education, marriage, first child, and a steady occupation, along with the psychological maturation they bring, now arrive years later than in previous generations 79,80. The taxonomy’s assumption that AL’s would exit the maturity gap and desist offending shortly after adolescence, and even the choice of the term “adolescence-limited,” was based on published age-crime curves calculated for a generation before adolescence began to elongate 7,81. It is possible that desistance is now delayed by late social maturation. There is evidence that the crime peak has shifted slightly older, but the reasons for this are under debate 82–86. Moreover, other historical shifts are relevant: the turn of the millennium saw an international drop in crime rates 87,88, particularly in teens 82. The 1993 taxonomy included ‘historical context’ among the causes of AL offending. Tests of this prediction are lacking and needed.
What’s new about snares?
Not every AL offender desists from offending at the same age, as revealed in the tapering off of the age-crime curve. The AL theory explained this variation by invoking the concept of ‘snares,’ experiences which can retard desistence from crime, such as addiction, a criminal record, imprisonment, victimization, or truncated education 89. The Mater University Study of over 3,000 Australians tested this hypothesis and reported that adolescent-onset offenders who continued offending further into adulthood had experienced a larger number of snares, particularly substance dependence, a criminal record, victimization, and early school-leaving 90.
What’s new about abstainers?
The 1993 article argued that AL offending is group social activity that is so highly prevalent as to be normative, as well as understandable from the perspective of contemporary teens. If this assertion is true, then the existence of abstainers, teens who commit no delinquency at all throughout their adolescent years, requires explanation. The article cited cohort studies showing that teens who self-report no delinquent acts are rare, and speculated that they must have one of three causes: barriers that prevent them from learning about delinquency, no maturity gap because of early access to adult roles, or personal characteristics unappealing to other teens that bar abstainers from risk-taking teen groups. The resulting research mainly focused on the hypothesis of unappealing characteristics. It suggests some abstainers do have characteristics that exclude them from delinquent groups, but that these characteristics pay off in life success. In the Dunedin cohort, abstainers described themselves on personality measures as over-controlled and lacking social confidence, and they were latecomers to sex for their cohort. However, their highly conscientious style became successful in adulthood. Since then, other studies have reported that abstainers are less involved with peers 91–94, and are less accepted in early adolescence but become more accepted later 95. Abstainers were reported to be latecomers to puberty 91, withdrawn, shy, socially anxious 96,97, methodical and conscientious 94, rational thinkers, and good at coping 92. Interestingly, the Cambridge Study in Delinquent Development yielded two abstainer groups: one characterized by low popularity and low school achievement who ended up poorly adjusted as adults, and another characterized by high honesty who ended up well-adjusted 98.
Impact of the developmental taxonomy
Impacts on clinical practice and policy-making can be difficult to track, because ideas that make their way from the ivory tower to policy circles often lose their provenance along the way, because policy-makers seldom reference academic journals. However, the taxonomy is cited in reports that guide policy in the UK (e.g., Home Office 201699; The Centre for Social Justice100 2008) and the USA (e.g., National Research Council 2013101; 2014102). Key themes in these reports are the notion that individual development is one driver of serious recidivistic crime, and the need for early-childhood prevention aimed at families and schools. Another key theme picked up in these reports is the need to appreciate heterogeneity within adolescent offenders, to distinguish the few who have adverse backgrounds and a poor prognosis from the many who have ordinary backgrounds and a better prognosis. The reports articulate the need to limit formal justice-system sanctions to fewer juveniles, and the need for non-punitive diversion approaches for more juveniles to avoid a damaging criminal record and give them room to reform. All of these themes chime with the taxonomy.
Policy-relevant research has analyzed the economic cost of LCP offending. Lifetime crime-career trajectories derived in a 1958 Philadelphia birth cohort were connected to the monetary cost calculated for each offense recorded for each trajectory member 103. LCP offenders, who committed crimes at high frequency while young and then turned to more serious crimes as adults, accounted for greater costs than did AL offenders, whose offending peaked during adolescence. Use of crime-costing methodology in the Cambridge Study in Delinquent Development similarly concluded that the cost of LCP offending is up to ten times greater than the cost of AL offending 104. These reports recommend that prevention and intervention resources should be invested in individuals whose offending fits an LCP pattern.In legal practice, the taxonomy is cited (alongside developmental neuroscience research 105) in Amicus Curiae briefs submitted for United States Supreme Court cases in 2009 and again in 2013 (Roper vs Simmons, overturning the death sentence for juveniles; Graham, Sullivan vs Florida, Miller vs Alabama, regarding life sentence without possibility of parole for juveniles). Both sides drew on the taxonomy, a reminder that scientists cannot always anticipate the uses of their work. Briefs arguing for sentences of death or life without parole for juveniles cited the taxonomy as evidence that juvenile offenders on a LCP trajectory have low probability of reform, and therefore should be permanently incapacitated. In contrast, briefs arguing against adult sentences cited the taxonomy as evidence that LCP offenders have cognitive deficits that render them incompetent to be prosecuted as adults. Briefs against adult sentences also cited the taxonomy as evidence that illegal behaviour is normative for adolescents, inferring that adult sentences for juveniles constitute “cruel and unusual” punishment. The court ruled that a juvenile cannot be sentenced to life without parole if their crime reflects “transient immaturity.” During sentencing, defense attorneys must show the offender has transient immaturity, whereas prosecutors must show the same offender lacks transient immaturity.
Conclusion
Evidence about LCP antisocial behavior provided impetus for the early-years prevention movement. However, less appreciated is that evidence about AL antisocial behavior provides impetus for movements to reform juvenile-justice and mental-health services in directions that are more supportive for young people. The taxonomy achieved its original goal, to account for the age-crime curve. Its dual theories of LCP and of AL development met their original goal of drawing from biological, psychological, and sociological theories to explain antisocial behavior. The theory must remain sufficiently flexible to stay as pertinent tomorrow as yesterday, while simultaneously keeping its defining tenets. Empirical research continues to sand and polish.