Research Review: Evaluating and reformulating the developmental taxonomic theory of antisocial ...

17 Pages·2013·0.49 MB·English

JournalofChildPsychologyandPsychiatry54:9(2013),pp924–940 doi:10.1111/jcpp.12102
Research Review: Evaluating and reformulating the
developmental taxonomic theory of antisocial
behaviour
Graeme Fairchild,1 Stephanie H.M. van Goozen,2 Andrew J. Calder,3 and Ian M. Goodyer4
1Academic Unitof Psychology, Universityof Southampton, Southampton, UK; 2Schoolof Psychology, Cardiff
University,Cardiff,UK;3MedicalResearchCouncilCognitionandBrainSciencesUnit,Cambridge,UK;4Department
of Psychiatry, Universityof Cambridge, Cambridge, UK
Background: Thedevelopmentaltaxonomictheoryproposesthattherearetwosubtypesofantisocialbehaviour.The
firstisaneurodevelopmentaldisorderwhichemergesinearlychildhoodandfollowsalife-coursepersistentcourse,
whereasthesecondemergesinadolescence,remitsinearlyadulthoodandreflectspeerprocessessuchasmimicryof
antisocial peers. The aim of this review was to evaluate the developmental taxonomic theory in the light of recent
empiricalresearch.Methods: Weconductedacomprehensiveliteraturereviewcomparingthesesubtypesofantisocial
behaviourbasedonsearchesonPubMedandotherscientificdatabasescoveringtheperiodfrom1993to2013. We
focusedonresearchencompassingpsychiatricepidemiology,personalityassessment,neuropsychology,neuroendo-
crinology, genetics, and structural and functional neuroimaging. Sixty one empirical studies were identified that
investigated one of these forms of antisocial behaviour separately or explicitly compared childhood-onset and
adolescence-onsetformsofantisocialbehaviour.Results: Empiricalresearchprovidessupportforthehypothesisthat
life-course persistent antisocial behaviour is a neurodevelopmental disorder which emerges in the transactions
betweenindividualvulnerabilitiesandenvironmentaladversity.Incontrasttothedevelopmentaltaxonomictheory,
however, empirical findings suggest that severe antisocial behaviour that emerges in adolescence frequently has a
negativeprognosisandisrarelylimitedtotheadolescentperiod.Inaddition,bothformsofantisocialbehaviourare
associatedwithemotionprocessingdeficits,changesinbrainstructureandfunction,alterationsincortisolsecretion,
and atypical personality traits (such as increased callous-unemotional traits). Conclusions:We conclude that the
developmental taxonomic theory is in need of revision, as differences between life-course persistent and adoles-
cence-onset forms of antisocialbehaviourappeartobe quantitative, rather thanqualitative, innature. Inaddition,
evidenceisaccumulatingthatadolescence-onsetantisocialbehaviourmayalsobeaneurodevelopmentaldisorder.To
accountfor thesimilaritiesbetweenthesegroups, despitethedifferences intheirage-of-onset, wepropose thatthe
quality of the child’s early environment moderates the relationship between individual vulnerabilities and the
age-of-onset of antisocial behaviour. Keywords: Antisocial behaviour, conduct disorder, developmental taxonomic
theory,epidemiology,neuropsychology,neuroimaging.
excluded a role for neuropsychological deficits or
The developmental taxonomic theory and its
neurobiological factors in the aetiology of AL antiso-
impact on developmental psychology and
cial behaviour: ‘Instead of a biological basis in the
psychopathology
nervous system, the origins of adolescence-limited
The developmental taxonomic theory was first
delinquency lie in youngsters’ best efforts to cope
described in a seminal and highly influential article
with the widening gap between biological and social
publishedbyTerrieMoffittin1993(Moffitt,1993).It
maturity.’(Moffitt,1993,p.692).Moffittinvokedthe
holds that within adolescent populations, there are
conceptofamaturitygaptoexplainthebehaviourof
two groups of offenders who differ systematically in
the AL group – essentially, they wish to be treated
thecourses,correlatesandcausesoftheirantisocial
like adults, but society treats them as children, so
behaviour: a life-course persistent (LCP) group who
they imitate their LCP antisocial peers in a mis-
show stable high levels of aggression and antisocial
guidedattempttoobtainstatusandtheprivilegesof
behaviour starting in childhood and continuing into
adulthood (e.g., access to alcohol). Consequently,
adulthood, and who are characterised by neuropsy-
these groups are considered to differ qualitatively,
chological impairments and exposure to childhood
ratherthanquantitatively–LCPantisocialbehaviour
adversity, and an adolescence-limited (AL) group
is a form of psychopathology, whereas AL antisocial
whose antisocial behaviours are primarily nonag-
behaviour is viewed as virtually normative. In line
gressive, and who do not show neuropsychological
with this view, Moffitt predicted that LCP antisocial
impairments (see Figure 1). In fact, Moffitt explicitly
behaviour would be relatively rare, whereas AL
antisocial behaviour would be common amongst
Conflictsofintereststatement:Noconflictsdeclared. adolescent populations.
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
PublishedbyJohnWiley&SonsLtd,9600GarsingtonRoad,OxfordOX42DQ,UKand350MainSt,Malden,MA02148,USA
ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsuse,distributionand
reproductioninanymedium,providedtheoriginalworkisproperlycited.
doi:10.1111/jcpp.12102 Evaluatingthedevelopmentaltaxonomictheoryofantisocialbehaviour 925
100% this must be taken into account by researchers
OR90% studying antisocial behaviour. In particular, the
HAVI80% theory implies that researchers must clearly distin-
OCIAL BE7600%% gwuhiesnhsbtuetdwyeinegngLroCuPpsaonfdadAoLlesacnetnistosc.iIaflthbeeyhdaovionuort
NTIS50% doso,theirstudiesmaydramaticallyunderestimate
A ADOLESCENCE- the contribution of neurobiological or genetic influ-
E OF 40% LIMITED ences on LCP antisocial behaviour. Consequently,
ENC30% the theory suggests that neurobiological research
VAL20% should focus on the LCP group, because including
E
R
P10% LIFE-COURSE individuals with AL antisocial behaviour may
PERSISTENT
obscure group differences and unnecessarily
AGE 5 10 1155 20 25 30 35 40 45
increaseheterogeneityoftheantisocialgroup.Third,
in terms of clinical practice, the view that LCP
antisocial behaviour is a form of psychopathology
Figure1 A schematic representation of the developmental tax- whereas AL antisocial behaviour is virtually norma-
onomictheoryofantisocialbehaviour,illustratingthehypothet-
tive implies that clinicians should devote their ener-
icalcoursesoflife-coursepersistent(LCP)andadolescence-limited
gies to treating or developing new interventions for
(AL) forms of antisocial behaviour. (Adapted with permission
fromMoffitt,1993,PsychologicalReview,100,p.677;Copyright the LCP group. Conversely, the theory can be inter-
AmericanPsychologicalAssociation,1993.) preted as implying that individuals with AL antiso-
cial behaviour are more likely to respond to
In the two decades since its publication, the psychological therapies, whereas LCP individuals
developmental taxonomic theory has had a major will be less responsive to treatment as the cumula-
influence on the fields of developmental psychology, tive consequences of their antisocial behaviour have
psychiatry,educationandcriminology.Atthetimeof been building up over many years.
writing this review, the 1993 article describing the In the remainder of this review, we will discuss
theory has been cited over 5,000 times according to epidemiological, psychometric, neuropsychological,
Google Scholar, making it one of the most highly neuroendocrinological, and neuroimaging research
cited papers in psychology. It has inspired several testing the developmental taxonomic theory, and
fruitfullinesofresearchandhelpedtobringtogether evaluate the extent to which the findings of these
the disciplines of developmental psychology, psy- diverse studies support the theory. It should be
chopathology and criminology. In addition, the dis- notedthattherewillbesomevariationregardingthe
tinction set out in the theory between LCP and AL terms used to describe patterns of antisocial behav-
antisocial behaviour has informed the categories of iour:‘LCP’and‘AL’areappropriatewhenconsidering
childhood-onset and adolescent-onset conduct dis- participants in longitudinal studies who have been
orderthatwerefirstincorporatedintotheDSM-IVin followed into adulthood, whereas cross-sectional
1994 (American Psychiatric Association, 1994), and studies cannot state with certainty that child-
have been retained in the recent DSM-5 (American hood-onset patterns of antisocial behaviour will be
Psychiatric Association, 2013). The purpose of this life-course persistent, nor can they tell whether an
review was to provide a timely overview and critique adolescence-onset pattern will be adolescence-lim-
ofthedevelopmentaltaxonomictheory,inthelightof ited, if the participants are tested during adoles-
recentempiricalstudieswhichindicatethatrevision cence. Consequently, wewill usethe terms LCP and
andreformulationofsomeaspectsofthetheorymay AL antisocial behaviour when considering findings
be required. from prospective longitudinal studies, but our
In addition to its influence on a variety of disci- default position, especially when describing
plines, the developmental taxonomic theory has cross-sectionalstudies(whichmakeupthemajority
wide-ranging implications for research on antisocial of the studies included in this review) will be to use
behaviour and clinical practice with children with the terms ‘childhood-onset’ and ‘adolescence-onset’
disruptive behaviour disorders. First of all, as the CD. Whilethesetermsarenotsynonymous,there is
theory has informed the distinction between child- considerable overlap between LCP and child-
hood-onset and adolescent-onset forms of conduct hood-onset antisocial behaviour, and also between
disorder set out in the DSM-IV and DSM-5, the AL and adolescence-onset antisocial behaviour.
validity of the theory is important for our diagnostic
andclassificationsystems.Specifically,isage-of-on-
set a meaningful way to parse the heterogeneity Methods
within antisocial behaviour, and is it useful for Relevant studies were identified using PubMed and
clinicians to make this distinction when assessing Web of Science to perform literature searches with
patients and formulating treatment strategies? Sec- the keywords ‘developmental taxonomic theory’,
ond, if there are qualitative differences in aetiology ‘developmental taxonomy’, ‘childhood-onset con-
between LCP and AL forms of antisocial behaviour, duct disorder’, ‘early-onset conduct disorder’ and
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
926 GraemeFairchildetal. JChildPsycholPsychiatry2013;54(9):924–40
‘adolescent-onset conduct disorder’. We restricted adversity and the development of antisocial behav-
our search to studies published in English, with a iour in adolescence (Fergusson et al., 2000; Rois-
searchperiodcovering1993–2013.Wealsosearched man, Monahan, Campbell, Steinberg, & Cauffman,
in the reference sections of key review articles and 2010). For example, a recent prospective longitudi-
papers from the Dunedin Multidisciplinary Health nalstudythatassessedantisocialbehaviourrepeat-
andDevelopmentStudy.Wherepossible,wefocused edly throughout childhood and adolescence found
on empirical studies which distinguished between little evidence for differences between child-
childhood-onset and adolescence-onset subtypes of hood-onset persistent (similar to LCP), adoles-
CD or between life-course persistent and adoles- cence-onset and childhood-limited antisocial
cence-limited forms of antisocial behaviour. This groups in exposure to childhood adversity or intra-
literature search yielded 61 empirical studies individual risk factors (Roisman et al., 2010). Inter-
directly relevant to the developmental taxonomic estingly, however, all three groups were elevated on
theory in the areas of epidemiology, personality theseriskvariablesrelativetocontrols.Thisresultis
assessment, genetics, neuropsychology, neuroendo- consistent with the findings of Fergusson et al.
crinology and neuroimaging (these studies are (2000), who showed that exposure to psychosocial
denoted by an asterisk in the reference list). adversity was higher for all groups on an antisocial
trajectory (including the adolescence-onset group)
compared with controls, although childhood-onset
Epidemiological evidence relevant to the offenders were exposed to the highest levels of
developmental taxonomic theory adversity and maladaptive family functioning (Fer-
The primary source of the findings that led to the gusson et al., 2000).
theory was a longitudinal birth cohort investigation, Further challenges to the developmental taxo-
the Dunedin Multidisciplinary Health and Develop- nomic theory have come from studies investigating
ment Study. Using age 18 follow-up data from this adult outcomes of adolescence-onset antisocial
study,Moffittandcolleagueswereabletoidentifythe behaviour which, according to the theory, should
two hypothesised groups of adolescent offenders typicallyfollowan‘adolescence-limited’course.Inan
described in the theory: one whose behavioural important study by Odgers et al. (2007), using data
difficulties started in early childhood, and another fromtheDunedinStudy,therewerenodifferencesin
whose antisocial behaviour emerged during adoles- conduct problems between the adolescence-onset
cence (Moffitt, Caspi, Dickson, Silva, & Stanton, andLCPgroupsatage26(thelastwaveofthestudy
1996). Childhood predictors of negative outcomes whenthelevelofconductproblemsweremeasured).
suchashyperactivity,lowverbalIQorlowsocioeco- Moreimportantly,the age32follow-updatashowed
nomic status were found to be far more common in that mental and physical health outcomes were
the LCP than the AL group, even though these highly negative for both the LCP and adoles-
groups showed similar rates of criminal offending cence-onset groups, relative to controls. The adoles-
at age 15 (Moffitt & Caspi, 2001; Moffitt, Caspi, cence-onset group was impaired relative to the
Rutter,& Silva,2001).However,Moffitt et al.(1996) control group on 9/15 mental health and 7/14
also identified a ‘recovery’ group of equal size to the physical health outcome measures (in comparison,
LCP group, and subsequent studies have confirmed theLCPgroupwasimpairedon13/15mentalhealth
the existence of a third trajectory of antisocial and 12/14 physical health measures). Adoles-
behaviour: the childhood-limited (CL) pathway, in cence-onset individuals were also more likely than
which individuals exhibit severe conduct problems controls to report engaging in violence or partner
in early childhood, but desist from engaging in abuse, or to have an adult conviction for violence.
antisocialbehaviourinadolescence(Aguilar,Sroufe, These findings contradict the notion that antisocial
Egeland, & Carlson, 2000; Fergusson, Horwood, & behaviourbeginninginadolescencewillremitasthe
Nagin,2000;Odgerset al.,2007,2008;Raineet al., individual enters adulthood, and suggest that the
2005).Interestingly,manypopulation-basedstudies ‘adolescence-limited’ term is a misnomer.
haveshownthatchildhood-limitedantisocialbehav- Recent research has investigated whether the
iour is actually the most common outcome of child- behaviours that make up the DSM-IV CD diagnosis
hood-onset conduct problems (Barker & Maughan, show similar or different developmental trajectories
2009;Nagin & Tremblay, 1999; Odgerset al., 2007, and relationships with adult criminality. A prospec-
2008). Thus, whilst the terms childhood-onset and tive longitudinal study which followed a cohort of
LCPantisocialbehaviourwereusedinterchangeably males from ages 12to31 foundthat aggression and
in Moffitt’s early writings, subsequent research has theft (i.e., a form of rule-breaking delinquency)
demonstrated that 50%–70% of individuals with showed distinct developmental trajectories (Barker
childhood-onset conduct problems outgrow their et al., 2007). Theft increased in frequency over
difficulties by adolescence. adolescence in the majority of the cohort (55%),
Inadditiontostudiesdocumentingtheexistenceof whereas only a small subgroup (13%) showed
a childhood-limited antisocial group, recent work increases in aggressive behaviour over the same
has reported an association between childhood period. Several other studies found that aggression
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
doi:10.1111/jcpp.12102 Evaluatingthedevelopmentaltaxonomictheoryofantisocialbehaviour 927
decreasesinfrequencywithageinmostindividuals, antisocial behaviour. This model has received some
whereas theft becomes more common during the empiricalsupportfromstudiesdocumentingsimilar
teenage years (Nagin & Tremblay, 1999), suggesting personality traits and impulse control difficulties in
thatthesesubtypesofantisocialbehaviourmayhave childhood-onset CD males and adolescence-onset
distinctaetiologiesandshouldbestudiedseparately CD females (Silverthorn, Frick, & Reynolds, 2001).
(Tremblay, 2010). To further investigate these Subsequent prospective longitudinal studies have
issues, a recent study by Burt, Donnellan, Iacono shownthatchildhood-onsetCDortheLCPtrajectory
andMcGue(2011)examinedwhetherage-of-onsetof doesexistinfemales,butitisrelativelyrare(withsex
CDorthesubtypesofantisocialbehaviourshownby ratios between 3:1 and 15:1 in favour of males;
the individual (aggression or rule-breaking) was a Kratzer & Hodgins, 1999; Moffitt et al., 2001; Lahey
better predictor of adult antisocial outcomes. Con- et al., 2006; Odgers et al., 2008; although for con-
sistent with the developmental taxonomic theory, flicting results from a high-risk sample see Keenan,
theyfoundthatindividualswithchildhood-onsetCD Wroblewski, Hipwell, Loeber, & Stouthamer-Loeber,
that persisted into adolescence were more likely to 2010).Thesestudieshavealsorevealedthatfemales
meetcriteriaforantisocialpersonalitydisorder(APD) mayshoweitheranadolescence-delayed-onsetoran
in adulthood than individuals who developed CD adolescence-limited trajectory (Fontaine, Carbon-
during adolescence. However, 15.5% of the latter neau, Vitaro, Barker, & Tremblay, 2009), and that
group still fulfilled criteria for APD at age 24, as adolescence-onset females are at increased risk for
compared with 54.2% of the former. Interestingly, negative adult outcomes relative to their peers even
when the authors controlled for the behavioural though they are likely to desist from showing anti-
subtypesmanifestedbytheindividual,CDage-of-onset social behaviour in adulthood (Odgers et al., 2008).
nolongerpredictedadultAPD.Theseresultssuggest Intriguingly, Odgerset al. (2008)foundthat females
that the forms of antisocial behaviour that the largely appeared to follow an adolescence-limited
person displays (aggression vs. rule-breaking) are trajectory,whereasadolescence-onsetconductprob-
more important in terms of predicting persistence lemswerefarmorelikelytopersistintoadulthoodin
into adulthood than age-of-onset of CD. In addition, males.ThissuggeststhattheLCPversusALdistinc-
rule-breaking in adolescence was the strongest tionmayactuallyapplybettertofemalesthanmales.
individual predictor of adult APD symptoms, rather Thefinalstudywewillconsiderinthissectionwas
than aggression. performedbyWalters(2011),whoappliedtaxometric
Another important study applied latent class analyses to data on the externalising symptoms
analysis to a very large epidemiological dataset to displayed by a large sample of individuals with
explore the heterogeneity of CD: it identified five antisocial behaviour, to examine whether the LCP
distinct classes of CD (Nock, Kazdin, Hiripi, & and AL groups could be distinguished empirically.
Kessler, 2006). Interestingly, only the ‘pure aggres- Therewasnoevidenceforataxonic(i.e.,categorical)
sive’ class of CD had a mean age of onset below age boundary between LCP and AL forms of antisocial
10; the most severe and impairing classes of CD behaviour, suggesting that the differences between
(termed ‘severe covert’ and ‘pervasive’) both had these subtypes are quantitative, rather than quali-
mean ages of onset above 11 years. This suggests tative, in nature. This challenges the developmental
that age-of-onset of CD is not related to severity of taxonomic theory which holds that these subtypes
antisocial behaviour in the manner predicted by the are aetiologically distinct, instead indicating that
developmental taxonomic theory or the DSM-IV they both fall on the same underlying dimension,
criteria. Thestudyalsoshowed thatthemostsevere withLCPindividualsmerelyhigherontheantisocial
classes of CD were associated with the greatest risk dimension than AL individuals (Walters, 2011).
for subsequent mental disorders. One limitation of In summary, recent epidemiological data have
thisstudywasitsrelianceonretrospectivereportsof challenged the developmental taxonomic theory in
CD symptoms and their age-of-onset. several ways: (a)there isa third trajectory of antiso-
In addition, epidemiological data have raised the cial behaviour, termed the ‘childhood-limited path-
question of whether the developmental taxonomic way’, which is marked by childhood-onset conduct
theory applies equally to males and females. Silver- problems, but desistance from serious antisocial
thorn and Frick (1999) argued that, as girls only behaviour by adolescence; (b) the term ‘adoles-
rarely show childhood-onset antisocial behaviour cence-limited’ antisocial behaviour appears to be a
but come close to catching up with their male peers misnomer, as many individuals with adoles-
by midadolescence (Cohen, Cohen, & Brook, 1993; cence-onset CD continue to show severe antisocial
Lahey et al., 2000; Moffitt et al. 2001), females may behaviour and experience poor mental and physical
follow a third developmental pathway to antisocial health outcomes as adults; (c) the developmental
behaviour: the delayed-onset pathway. According to antecedents of adolescence-onset/AL antisocial
this model, the personality and neuropsychological behaviourappeartodifferonlyquantitatively,rather
characteristics which distinguish male child- than qualitatively, from those of LCP antisocial
hood-onsetindividuals,suchaspsychopathictraits, behaviour, as both groups experience higher levels
are also observed in females with adolescence-onset of psychosocial adversity than controls; (d) aggres-
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
928 GraemeFairchildetal. JChildPsycholPsychiatry2013;54(9):924–40
siveandrule-breakingformsofantisocialbehaviour alsoshownintwoseparatestudiesthatfemaleswith
show different developmental trajectories, with adolescence-onset CD show elevated psychopathic
rule-breaking more strongly associated with adult and callous-unemotional traits relative to controls
APD than the age-of-onset of CD; (e) it is unclear (Fairchild, Stobbe, van Goozen, Calder, & Goodyer,
whetherthedevelopmentaltaxonomictheoryapplies 2010; Fairchild et al., 2013). In contrast to these
equally well to females and males, as the course of findings, however, Dandreaux and Frick (2009)
CD appears to differ by sex, with females rarely found that childhood-onset offenders had higher
showing LCP or childhood-onset CD, leading some levels of callous-unemotional traits than adoles-
researchers to propose the existence of a ‘delaye- cence-onset offenders, although unfortunately no
d-onset’ antisocial pathway in females; and (f) taxo- controlgroupwasincludedinthisstudy(Dandreaux
metric analyses suggest that differences between & Frick, 2009).
LCP and AL antisocial behaviour are quantitative, In summary, personality research has provided
rather than qualitative, in nature. little support for the contention that there are
qualitative differences in personality traits between
AL/adolescence-onset and LCP groups, or the pro-
Personality trait correlates of CD posal that only LCP individuals will exhibit atypical
Moffitt et al. (1996)measured personality profiles at personality profiles. Most studies in this area have
age 18 in the Dunedin cohort, using the Multidi- shown that AL or adolescence-onset CD individuals
mensional Personality Questionnaire (Tellegen, differfromcontrolsonmultiplepersonalityvariables
1982), finding that the LCP group scored lower on such as impulsivity and negative emotionality, con-
personalitytraitsrelatedtoaffiliativebehaviour,and tradicting the developmental taxonomic theory,
higherinaggression,impulsivity,andstressreaction although personality profiles are generally more
than controls. The AL group also differed from extreme in LCP or childhood-onset groups. In addi-
controls on several of these measures. The only tion, while there are exceptions to this pattern,
variablethatdifferedbetweentheLCPandALgroups several studies have reported elevated levels of
was social closeness, a personality trait reflecting psychopathicandcallous-unemotionaltraitsinboth
sociabilityandadesiretoaffiliatewithothers,which childhood-onsetandadolescence-onsetformsofCD.
was lower in LCP individuals. In a subsequent
follow-up at age 26, Moffitt, Caspi, Harrington, and
Milne(2002)foundthatALindividualsscoredhigher Behavioural and molecular genetics of CD
innegativeemotionality,andlowerinagreeableness, Very few behavioural genetic studies have been
constraint, and conscientiousness than controls. designedtotestthedevelopmentaltaxonomictheory
Although the LCP group scored higher than the AL by directly comparing the heritability of LCP and AL
group on negative emotionality, and lower on agree- forms of antisocial behaviour using prospective
ableness and social closeness, the AL group scored longitudinal data. However, several studies have
lower than the LCP group on constraint (reflecting investigated whether the relative contribution of
increased impulsivity). These findings suggest that genetic and environmental influences on antisocial
AL individuals show a distinct constellation of per- behaviour changes across the life span. The devel-
sonality traits that, while less pathological than the opmental taxonomic theory can be interpreted as
LCPgroup,neverthelessplacesthematelevatedrisk predicting that genetic influences should follow an
for psychopathology and interpersonal problems. U-shapedfunctionwithage:theseshouldbestrong-
Consistent with the latter results, several other est in childhood and adulthood, reflecting the fact
studies reported that adolescence-onset and child- that LCP is under genetic influence, and weakest
hood-onset antisocialbehaviour arebothassociated during adolescence, as the latter measure would be
with increased impulsivity relative to controls, or confounded by the presence of a high proportion of
failed to detect differences in impulsivity between AL individuals whose antisocial behaviour arises
these groups (Aguilar et al., 2000; Dandreaux & from environmental, rather than genetic, origins.
Frick, 2009; Taylor, Iacono, & McGue, 2000; White, RheeandWaldman(2002)performedameta-anal-
Bates, & Buyske, 2001). ysis of behavioural genetic studies of antisocial
A number of studies have investigated psycho- behaviour, but were unable to investigate whether
pathic or callous-unemotional traits in adolescents LCP antisocial behaviour was more heritable than
with CD. In three separate male samples, we found adolescence-onset/ALformsofantisocialbehaviour,
that both childhood-onset and adolescence-onset as insufficient longitudinal data were available to
forms of CD were associated with elevated psycho- enable them to test this hypothesis. However, their
pathic and callous-unemotional traits relative to meta-analysis found that the heritability of antiso-
healthy controls, whereas there were no significant cial behaviour was greater in childhood than in
differences between these CD subgroups on either adolescence or adulthood, contrary to the U-shaped
measure (Fairchild, van Goozen, Calder, Stollery, & function predicted by the developmental taxonomic
Goodyer, 2009; Fairchild, van Goozen, Stollery, theory. In contrast, an earlier meta-analysis found
Brown et al., 2008; Fairchild et al., 2011). We have that genetic influences on antisocial behaviour
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
doi:10.1111/jcpp.12102 Evaluatingthedevelopmentaltaxonomictheoryofantisocialbehaviour 929
increased and shared environmental influences results are in line with the developmental taxo-
decreased between childhood and adulthood (Miles nomic theory.
& Carey, 1997), again diverging from the U-shaped Jacobson, Prescott, and Kendler (2002) measured
patternpredictedbythetheory.Wewillnowconsider genetic and environmental influences on antisocial
specific behavioural genetic studies which are par- behaviourinalargesampleofmaleandfemaletwins
ticularly relevant to the developmental taxonomic (n = 6,806). They found that genetic influences on
theory. antisocial behaviour increased from childhood
In an important twin study, Taylor et al. (2000) (defined as below age 15) to adolescence and adult-
explicitly compared the heritability of child- hood. Notably, for male twins, genetic influences on
hood-onset and adolescence-onset forms of CD, childhood antisocial behaviour were weak (0.06),
findingevidenceforgeneticinfluencesontheformer, whereas genetic influences on antisocial behaviour
but not the latter, subtype of CD. The authors also in adolescence or adulthood were moderate in size
found higher rates of adult antisocial behaviour in (0.41 and 0.40, respectively). The study also
the first- and second-degree relatives of the child- observed unique genetic influences on adolescent
hood-onset CD participants than the relatives of antisocial behaviour that were not shared with
adolescence-onset CD or control participants. Inter- childhood antisocial behaviour, possibly reflecting
estingly, however, 95% of the cotwins who were genetic influences on biological processes activated
concordant for CD showed the same subtype of CD, duringpuberty.Theseresultsarenotconsistentwith
indicating that when genetically-identical individu- the developmental taxonomic theory.
als are concordant for CD they are likely to develop In another recent study, Van Hulle et al. (2009)
the same variant. Although limited by a relatively found that genetic influences on antisocial behav-
smallsamplesize(n = 70twinpairswithatleastone iourinadolescenceweredistinctfromthoseaffecting
cotwinmeetingDSM-III-RcriteriaforCD),thisstudy antisocial behaviour in childhood, although herita-
provides strong support for the developmental tax- ble influences were significant at both time points.
onomic theory. The authors interpreted their findings as evidence
In a subsequent study of 1,186 twin pairs, Eley, that youth showing persistent antisocial behaviour
Lichtenstein, and Moffitt (2003) found that aggres- from childhood into adolescence are influenced by
sive antisocial behaviour in childhood was highly one set of genetic factors, whereas a second set of
heritable (0.60), whereas aggressive antisocial genetic factors affect antisocial behaviour which
behaviour in adolescence was only moderately her- emerges around the pubertal transition. These
itable (0.46). Non-aggressive antisocial behaviour observations run counter to the developmental tax-
was moderately heritable when considering either onomic theory, as they suggest that adoles-
childhood or adolescence (0.49 and 0.44, respec- cence-onset CD may have a genetic aetiology (albeit
tively). Shared environmental influences on nonag- one that is distinct from childhood-onset CD).
gressive antisocial behaviour were moderate in size Taking a different perspective, Burt and Neiderh-
atbothtimepoints(0.35and0.42forchildhoodand iser (2009) examined whether genetic influences on
adolescence, respectively). Overall, this study sug- rule-breaking and aggressive antisocial behaviours
gestedthatchildhoodantisocialbehaviourwasmore declined or increased over the adolescent period.
heritable than adolescent antisocial behaviour; this Their findings demonstrated that heritability of
was particularly true for aggression. These results rule-breaking forms of delinquency increased in the
provide some support for the developmental taxo- transition from early- to late-adolescence, whereas
nomic theory, but they do not appear to show a genetic influences on aggression remained stable
qualitativedifferencebetweenchildhoodandadoles- overthisperiod.Interestingly,sharedenvironmental
cence in the heritability of aggression andantisocial influences were most pronounced in childhood and
behaviour. declineddramaticallyinmidadolescence.Inaddition
In a direct test of the developmental taxonomic to suggesting differential genetic influences on dif-
theory, Silberg, Rutter, Tracy, Maes, and Eaves ferent forms of antisocial behaviour, these results
(2007)investigatedgeneticandenvironmentalinflu- demonstrated distinct heritable influences on ado-
ences on antisocial behaviour at several points lescence-onset antisocial behaviours involving
during adolescence in a sample of 1,037 male rule-breaking.
twins. They found that heritability estimates were Burt (2009) subsequently performed a meta-anal-
highest (0.43) for antisocial behaviour in early ysis investigating genetic and environmental influ-
adolescence and lowest (0.05) in midadolescence. ences on aggressive and rule-breaking antisocial
There was also a transient genetic effect on antiso- behaviour.Shefoundthataggressivebehaviourwas
cial behaviour displayed around the pubertal tran- more heritable than rule-breaking behaviour (0.65
sition. In contrast, shared environmental influences vs.0.48,respectively). Rule-breakingbehaviourwas
were largest in midadolescence, demonstrating the influenced by the shared environment (0.16),
importance of environmental factors in the aetiology whereassharedenvironmentalinfluencesonaggres-
of adolescence-onset antisocial behaviour. These sion were weak. These results suggest aetiological
differences between aggressive and nonaggressive
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
930 GraemeFairchildetal. JChildPsycholPsychiatry2013;54(9):924–40
forms of antisocial behaviour, consistent with the adolescence-onset CD, which appeared to be
studies described above demonstrating that these explainedbyhigherlevelsofaggressionintheformer
behavioural subtypes show distinct developmental group. Unfortunately, this study did not control for
trajectories. The results also have implications for time of day of saliva collection, which makes it
the developmental taxonomic theory, as aggression difficult to interpret these results because cortisol
is more strongly associated with LCP than adoles- secretion shows a pronounced circadian rhythm. A
cence-onset/AL antisocial behaviour. recent study reported that life-course persistent,
In conclusion, behavioural genetic studies have childhood-limited, and adolescence-onset forms of
yielded mixed results with some studies showing conduct disorder were all associated with lower
that the heritability of antisocial behaviour wakingcortisollevels,whereasnoneofthesegroups
increases,whereassharedenvironmentalinfluences differed from each other (Haltigan, Roisman, Sus-
decrease, with age. On the other hand, several man,Barnett-Walker&Monahan,2011).Thisstudy
studies have reported that the heritability of antiso- is important because it employed a prospective
cial behaviour is highest in childhood and lowest in longitudinaldesignwithmeasurementsofantisocial
adulthood. Neither pattern is consistent with the behaviourobtainedfromage5onwards,andtimeof
developmental taxonomic theory, which predicts saliva sampling was standardised across partici-
that genetic influences on antisocial behaviour will pants. In our research investigating basal cortisol
bestrongestinchildhoodandadulthood,andweak- secretion, we found no differences between child-
estin adolescence(as onlyLCP antisocialbehaviour hood-onset or adolescence-onset CD participants
is considered heritable). While several twin studies and controls in morning cortisol levels, although
have provided support for an aetiological distinction both CD groups showed elevated evening cortisol
between childhood-onset and adolescence-onset levels relative to controls, suggesting a flatter circa-
antisocial behaviour, recent studies have revealed dian profile in CD (Fairchild, van Goozen, Stollery,
unique genetic influences on antisocial behaviours Brown et al., 2008).
that emerge during adolescence (particularly In relation to cortisol reactivity, early studies
rule-breaking delinquency) or puberty. Conse- showed that children with disruptive behaviour
quently, behavioural genetic studies have provided disorders showed blunted cortisol responses to
only equivocal support for the developmental taxo- stress (van Goozen, Matthys, Cohen-Kettenis, Bu-
nomic theory. itelaar, & van Engeland, 2000; van Goozen et al.,
Molecular genetic studies have documented a 1998), particularly if they were also low in anxiety.
robustandreplicableinteractionbetweenthemono- These findings suggest that childhood-onset CD is
amine oxidase-A (MAOA) gene and childhood mal- associated with reduced HPA axis responses to
treatment in the aetiology of antisocial behaviour stress. We recently extended these findings by dem-
(Caspi et al., 2002; Fergusson, Boden, Horwood, onstrating that both childhood-onset and adoles-
Miller, & Kennedy, 2012; Kim-Cohen et al., 2006). cence-onset forms of conduct disorder were
However, the majority of these studies appear to associated with reduced cortisol responses to a
have collapsed across childhood-onset and adoles- psychological stressor involving provocation and
cence-onset forms of CD or studied children only. frustration (Fairchild, van Goozen, Stollery, Brown
Therefore, to our knowledge, no studies have exam- et al.,2008).Therewerenosignificantdifferencesin
ined whether this interaction between maltreatment cortisolreactivitybetweenthesesubgroups.Overall,
and MAOA genotype is specific to childhood-onset these studies suggest that basal cortisol secretion
CD, as would be predicted by the developmental and cortisol reactivity to stress are altered in both
taxonomic theory. childhood-onsetandadolescence-onsetformsofCD,
which is inconsistent with the developmental taxo-
nomic theory.
Cortisol secretion and stress reactivity in CD
The relationship between antisocial behaviour and
hypothalamic-pituitary-adrenal(HPA)axisactivityis Neuropsychological and psychophysiological
complex and findings in the literature have been studies of CD
mixed (Alink et al., 2008; van Goozen, Fairchild, Moffitt (1993) originally focused on verbal and exec-
Snoek, & Harold, 2007). However, studies focusing utivefunctionswhenspecifyingtheneuropsycholog-
on school-aged children have largely observed a ical impairments involved in the aetiology of LCP.
negative relationship between basal or stress-in- However, she also noted that neuropsychological
duced cortisol and antisocial behaviour, which variation might underlie individual differences
appears to be strongest in clinic-referred samples. between children in temperament, emotional reac-
McBurnett, Lahey, Rathouz, and Loeber (2000) tivity,impulsecontrolandcognitiveabilities,leading
studied the relationship between CD, aggression, to a heightened risk for antisocial behaviour. In
andbasalcortisollevelsinasampleofclinic-referred subsequent years, a consensus has emerged that
boys.Theyfoundthatboyswithchildhood-onsetCD severe antisocial behaviour is associated with
had lower basal cortisol levels than those with impairments in emotion recognition and regulation.
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
doi:10.1111/jcpp.12102 Evaluatingthedevelopmentaltaxonomictheoryofantisocialbehaviour 931
In contrast, the evidence for deficits in executive in LCP individuals were secondary to exposure to
function in antisocial behaviour has been more childhood adversity (Aguilar et al., 2000).
mixed, especially when studies have controlled for Using cross-sectional designs, Pajer et al. (2008)
intelligencequotient(IQ)differencesbetweengroups. and Giancola, Mezzich, and Tarter (1998) both
investigated neuropsychological function in adoles-
cent girls with CD, finding impairments in multiple
Intelligence
aspects of executive function and visuospatial pro-
One of the most robust findings in criminology is cessing and lower IQ in the CD groups relative to
that delinquents or children with severe antisocial controls. However, neither study observed a rela-
behaviourhavelowerIQscoresthannondelinquents tionship between CD age-of-onset and severity of
(Rutter, Giller, & Hagell, 1998). Early work investi- neuropsychological deficits, contraryto thedevelop-
gating IQ in delinquents found that the 8-point mental taxonomic theory.
difference in full-scale IQ between delinquents and vanGoozen,Cohen-Ketteniset al.(2004)foundno
nondelinquents masked a striking difference evidence for neuropsychological deficits in executive
between subtypes: stable, childhood-onset delin- function or inhibitory control in children with oppo-
quents had a 17-point reduction in mean IQ, sitional defiant disorder (ODD) or CD; rather, the
whereas ‘temporary’ (adolescence-limited) delin- group differences they observed were specific to a
quents showed only a 1-point reduction (Moffitt, task with a motivational inhibitory component. Par-
1990). ticipants with childhood-onset ODD or CD were
UsingdatafromthePittsburghYouthStudy,Raine more likely to carry on playing despite receiving an
et al.(2005)foundthatoutof15different measures increasing ratio of losses to gains as the task
of neuropsychological function spanning verbal, progressed, suggesting that they were less sensitive
spatial and executive functions, the LCP group to punishment.
differed significantly from controls on 7/15 mea- In a study examining decision-making under risk
sures, whereas the AL group differed from controls in CD, we found that male adolescents with both
on 3/15 measures. However, when directly compar- childhood-onset and adolescence-onset forms of CD
ing these groups, the AL group differed significantly made more risky decisions than controls (Fairchild,
fromthe LCPgrouponjustasinglemeasure:verbal van Goozen, Stollery et al., 2009). Interestingly, the
IQ. While consistent with the proposal that neuro- participants with CD systematically adjusted their
psychological impairments should be most marked behaviourinlightoftherewardsandlossesavailable
inthosewithLCPantisocialbehaviour,thesedatado in each trial, but were more likely than controls to
notsupportthecontentionthatLCPandALformsof select the risky choice across a range of choices
antisocialbehaviourarequalitativelydifferentfroma varying in terms of the probability and the size of
neuropsychological perspective. potential gains and losses. These results imply that
Kratzer and Hodgins (1999) studied a Swedish individuals with both forms of CD are either less
birth cohort, relating different aspects of IQ, as sensitive to potential losses or more sensitive to
measured at age 13, to developmental trajectories gains, or both. In the same study, neither CD
of antisocial behaviour. Although LCP males scored subgroup showed a deficit in ‘cold’ executive func-
lower on all of the IQ subtests than AL offenders, tion(asmeasuredusingtheWisconsinCardSorting
adult starter offenders, and controls, the AL group Test) when factoring out group differences in IQ.
were also impaired on all IQ measures relative to Considered together, these studies provide little
controls. These prospective data again suggest that supportfordifferencesinexecutivefunctionbetween
IQ differences between LCP and AL offenders are childhood-onsetandadolescence-onsetformsofCD.
quantitative rather than qualitative, and that both Rather, these studies suggest that both CD sub-
forms of antisocial behaviour are associated with IQ groups show altered decision-making under ‘hot’ or
deficits. motivational conditions.
Executive functions and decision-making Emotion recognition, emotional reactivity and fear
conditioning
In an important prospective longitudinal study,
Aguilar et al. (2000) investigated the neurodevelop- As mentioned above, there is increasing agreement
mental and social origins of LCP and adoles- that deficits in emotion recognition and emotional
cence-onset antisocial behaviour. They found that reactivity/regulation are associated with antisocial
the LCP and adolescence-onset groups differed in behaviour(Davidson,Putnam&Larson,2000).Con-
rates of exposure to adversity, but not in tempera- sistent with this view, we found that males with
ment or neuropsychological functioning, in early childhood-onset CD and females with adoles-
childhood. Differences between these groups in cence-onsetCDshowedimpairedrecognitionofangry
neuropsychological functioning were only observed anddisgustedfacialexpressions(Fairchild,vanGoo-
in late childhood or adolescence, which was inter- zen,Calderet al.,2009;Fairchildet al.,2010).Inthe
preted as evidence that neuropsychological deficits studyfocusing on males,adolescence-onset CDwas
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
932 GraemeFairchildetal. JChildPsycholPsychiatry2013;54(9):924–40
associated with similar impairments to those work supports the view that adolescence-onset CD
observed in childhood-onset CD, but most compari- is quantitatively, rather than qualitatively, different
sonsbetweenadolescence-onsetCDindividualsand from childhood-onset CD.
controlsdidnotreachacorrectedlevelofsignificance.
Notably,therewerenosignificantdifferencesbetween
the childhood-onset and adolescence-onset CD Structural and functional neuroimaging
groupsinfacialemotionrecognition. studies of CD
Psychophysiological studies have reported Progress in understanding the neurobiological basis
reduced electrodermal responses to affective and of antisocial behaviour has accelerated in recent
neutral images (Herpertz et al., 2005) and reduced years, with a substantial number of studies using
startleresponsestoanacousticprobeinchildrenor structural or functional neuroimaging techniques
adolescentswithchildhood-onsetCD(Fairchild,van such as magnetic resonance imaging (MRI) to inves-
Goozen, Stollery, & Goodyer, 2008; van Goozen, tigate CD. The majority of these studies have been
Snoek, Matthys, van Rossum, & van Engeland, restricted to individuals with childhood-onset CD,
2004).Thesefindingsappeartoreflectageneralised probably due to the influence of the developmental
deficit in autonomic reactivity, rather than a taxonomic theory.
valence-specific effect (i.e., reduced responses to
negatively valenced stimuli). When investigating
Structural imaging studies
affective modulation of the startle reflex in adoles-
cence-onset CD, we found that this group also In one of the first studies in this area, Kruesi,
showed reduced startle responses when viewing all Casanova, Mannheim and Johnson-Bilder (2004)
slide types compared with controls (Fairchild, van investigated grey and white matter volume in male
Goozen,Stollery,&Goodyer,2008).Moreover,these adolescents with childhood-onset CD. The authors
reductions were numerically greater than those foundreducedgreymatterinthetemporallobeofthe
observed in the childhood-onset CD group. In a CD group relative to healthy controls. However, the
subsequent study, we found that girls with adoles- groups used were relatively small (n = 10) and
cence-onset CD also showed reduced startle differed in IQ. The analyses also relied on manual
responses relative to controls, again irrespective of tracing of brain structures and were unable to
slide valence (Fairchild et al., 2010). We also inves- localise the changes to specific regions of the tem-
tigated autonomic fear conditioning in child- poral lobe such as the amygdala. Sterzer, Stadler,
hood-onset and adolescence-onset forms of CD, Poustka and Kleinschmidt (2007) used voxel-based
finding statistically equivalent reductions in fear morphometry, an unbiased, automated method
conditioning in both groups in males (Fairchild, which provides more detailed information about
vanGoozen,Stollery,&Goodyer,2008).Inaddition, regional differences, to study grey matter volume in
femaleswithadolescence-onsetCDshowedimpaired male adolescents with childhood-onset CD and co-
autonomic conditioning relative to controls (Fair- morbid ADHD. They found that the childhood-onset
child et al., 2010). Lastly, heart rate responses to CD group showed reduced amygdala and anterior
psychosocial stress were attenuated in males with insula volume compared with controls. Huebner
both childhood-onset and adolescence-onset forms et al. (2008) found that male adolescents with
of CD (Fairchild, van Goozen, Stollery, Brown et al., childhood-onset CD (most of whom had comorbid
2008). Considered together, these findings suggest ADHD) showed volumetric reductions in left orbito-
that both childhood-onset and adolescence-onset frontal cortex and left amygdala extending to sur-
variants of CD are associated with impairments in rounding medial temporal lobe regions, such as the
emotion recognition, fear conditioning, and auto- hippocampus. A recent structural imaging study
nomic reactivity. observed reduced anterior insula and medial pre-
In conclusion, there is now considerable evidence frontal cortex volume in eight year-old children with
for neuropsychological impairments in LCP antiso- ODD or CD relative to controls (Fahim et al., 2011).
cial behaviour or childhood-onset CD. These find- Our recent structural MRI study replicated the
ings go beyond early predictions of deficits in IQ or abovefindingsofreducedamygdalavolumeinmales
executive function in LCP antisocial behaviour to with childhood-onset CD (Fairchild et al., 2011).
document impairments in facial emotion recogni- Crucially,however,wealsoobservedbilateralreduc-
tion, decision-making and emotional reactivity in tions in amygdala volume in males with adoles-
CD. However, in many cases, similar neurocognitive cence-onset CD (see Figure 2). These results
and psychophysiological impairments have been remained significant when controlling for ADHD
reported in adolescence-onset CD. In some studies, symptoms, suggesting that reductions in amygdala
the adolescence-onset CD group actually showed volume were not explained by ADHD comorbidity.
greater impairments than the childhood-onset CD This latter observation is consistent with the results
group, although in most studies they performed at of a voxel-based morphometry study which directly
an intermediate level between controls and child- compared adolescents with noncomorbid CD and
hood-onset CD participants. Consequently, this noncomorbid ADHD, finding widespread structural
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.
doi:10.1111/jcpp.12102 Evaluatingthedevelopmentaltaxonomictheoryofantisocialbehaviour 933
(A) T (B) T
5 5
L y = 3 L y = 4
4 4
3 3
2 2
1 1
HC > Combined CD HC > CO-CD
(C) T (D)
5
L y = 4 E.) 0.72
P.
4 e ( 0.70
m
u
3 er vol 0.68
att 0.66
m
2 y
gre 0.64
a
1 dal 0.62
g
y
m
A 0.60
HC CO-CD AO-CD
HC > AO-CD
Groups
Figure2 Groupdifferencesinamygdalagreymattervolume:(A)greymattervolumewasreducedinbilateralamygdalainacombined
groupofadolescentswithconductdisorder(n=63),relativetohealthycontrols(HCs;n=27);(B)significantbilateralamygdalavolume
reductioninparticipantswithchildhood-onsetconductdisorder(CO-CD)relativetoHCs;and(C)significantbilateralamygdalavolume
reductioninparticipantswithadolescence-onsetconductdisorder(AO-CD)relativetoHCs.Thecolourbars,rangingfromredtowhite,
representTstatistics.PanelDdepictsmeanvaluesforrightamygdalagreymattervolumeineachgroup.(FromFairchildetal.,2011,
AmericanJournalofPsychiatry,168(6),p.628;CopyrightAmericanPsychiatricAssociation,2011.)
abnormalities in the CD group relative to both the (and surrounding anterior temporal lobe) in adoles-
control and ADHD groups (Stevens & Haney-Caron, cents with CD. Surprisingly, these initial studies
2012). In a further study, we demonstrated that have found that the microstructural integrity of the
females with adolescence-onset CD also showed uncinate fasciculus is increased in adolescents with
reduced grey matter volume in the anterior insula CD (Passamonti et al., 2012; Sarkar et al., 2013;
compared with healthy controls (Fairchild et al., althoughseeFingeret al.,2012).Thesechangesmay
2013). Finally, a recent study observed reduced have functional consequences, as the uncinate fas-
cortical thickness in widespread temporal and pari- ciculus is implicated in emotion regulation and
etal regions, and reduced folding in insula and shows reduced microstructural integrity in individ-
prefrontal cortex in both childhood-onset and ado- ualswithaffectivedisorders(Trompdoet al.,2012).
lescence-onset forms of CD (Hyatt, Haney-Caron, & The study by Passamonti et al. (2012) focused spe-
Stevens,2012).Interestingly,whendirectlycompar- cificallyonmaleswithchildhood-onsetCD,whereas
ing these subtypes, adolescence-onset CD partici- Sarkaret al.(2013)includedbothsexesanddidnot
pants showed reduced cortical folding in the insula differentiate between childhood-onset and adoles-
and ventromedial prefrontal cortex relative to child- cence-onset subtypes of CD. It is interesting to note
hood-onset CD participants. Considered together, that these results are in the opposite direction to
these results indicate that both forms of CD are those obtained in adults with psychopathy or anti-
associated with reductions in grey matter volume social personality disorder, who show reduced
andcorticalthicknessinbrainregionsimplicatedin microstructural integrity in the uncinate fasciculus
emotion processing and regulation, such as the (Craig et al., 2009; Sundram et al., 2012). As the
amygdala, anterior insula, and orbitofrontal cortex. typical pattern of white-matter development in the
These structural changes may underlie the neuro- uncinate fasciculus is an inverted-U function with
psychological deficits observed in both subtypes of age (Lebel et al., 2012), one possible explanation of
CD, as described in the section above. these findings is that individuals with CD show
Recentstudieshaveuseddiffusiontensorimaging accelerated maturation of the uncinate fasciculus
methods to investigate anatomical connectivity in tract in childhood or adolescence, which is followed
the uncinate fasciculus white-matter pathway that by earlier or more pronounced reductions in micro-
connects the prefrontal cortex and the amygdala structural integrity in adulthood (Passamonti et al.,
©2013TheAuthors.JournalofChildPsychologyandPsychiatry©2013AssociationforChildandAdolescentMentalHealth.