PsychologicalBulletin Copyright2007bytheAmericanPsychologicalAssociation 2007,Vol.133,No.1,149–182 0033-2909/07/$12.00 DOI:10.1037/0033-2909.133.1.149 The Evidence for a Neurobiological Model of Childhood Antisocial Behavior Stephanie H. M. van Goozen Graeme Fairchild CardiffUniversity UniversityofCambridge Heddeke Snoek Gordon T. Harold UniversityMedicalCentreUtrecht CardiffUniversity Childrenwithpersistentantisocialandaggressivebehaviorarediagnosedashavingdisruptivebehavior disorder.Theauthorsreviewevidencethatantisocialchildren,andespeciallythosewhopersistwiththis behaviorastheygrowolder,havearangeofneurobiologicalcharacteristics.Itisarguedthatserotonergic functioning and stress-regulating mechanisms are important in explaining individual differences in antisocialbehavior.Moreover,lowfearofpunishmentandphysiologicalunderactivitymaypredispose antisocialindividualstoseekoutstimulationortakerisksandmayhelptoexplainpoorconditioningand socialization.Theauthorsproposeatheoreticalmodelhighlightingtheinterplaybetweenneurobiological deficitsandcognitiveandemotionalfunctioningasmediatorsofthelinkbetweenearlyadversityand antisocialbehaviorproblemsinchildhood.Implicationsforinterventionprogramsarediscussed. Keywords:conductdisorder,aggression,serotonin,autonomicnervoussystem,cortisol This review is concerned with the role of neurobiological sys- temperament and a nonoptimal environment in which ineffective tems in the development and maintenance of antisocial behavior. socialization plays a key role: A difficult child is more likely to Inconsideringneurobiologicalinfluencesonbehavior,therecog- elicitharsh,inconsistent,andnegativesocializationbehaviors,and nitionthatanybehavioristheoutcomeofacomplexinterplayof as a result, a difficult temperament ultimately develops into anti- individual, developmental, and social factors is important. In the socialbehavior(Patterson,Reid,&Dishion,1992).Althoughthere caseoftheetiologyofchildhoodaggressionandantisocialbehav- are factors that contribute to antisocial behavior in childhood ior,contributingfactorsmayplayamoreorlessimportantroleat becoming chronic, not all antisocial children become antisocial differentpointsinchildhoodandadolescence.Forexample,harsh adolescents, and not all antisocial adolescents become antisocial parentaldisciplinewithcruelpunishmentappearstoplayacausal adults (Robins, 1978). A study of the neurobiological factors role in the development of antisocial behavior in childhood, involvedinantisocialbehaviorcanhelpexplainwhythesebehav- whereas the absence of parental supervision appears to be an ioralpatternspersistordesistovertime. important factor in antisocial behavior in late childhood or ado- Inthisreview,wearguethatserotonergicfunctioningandstress- lescence (Lahey, McBurnett, Loeber, & Hart, 1995). Another regulating mechanisms, including the hypothalamic-pituitary- example is that the early (perinatal) hormonal environment is adrenal(HPA)axisandtheautonomicnervoussystem(ANS),are assumed to play a role in shaping temperament and its develop- importantinexplainingindividualdifferencesinantisocialbehav- mentbeforetheonsetofimportantsocialinfluencessuchaspeer ior.Moreover,lowfearofpunishmentandphysiologicalunderac- interactions. tivitymaypredisposeantisocialindividualstoseekoutstimulation With regard to young children, it is likely that the origin of ortakerisksandmayhelptoexplainpoorsocialconditioningand antisocial behavior is to be found in a combination of difficult socialization.Inlightofthefindingsfromthestudiesthatwereto be reviewed, it was our firm belief that an investigation of the neurobiological factors involved in children’s antisocial behavior wouldultimatelyindicatewhichindividualsaremostatrisk,which Stephanie H. M. van Goozen and Gordon T. Harold, School of Psy- typesofinterventionsarelikelytobemostbeneficial,andwhich chology,CardiffUniversity,Cardiff,UnitedKingdom;GraemeFairchild, individualsaremostlikelytobenefitfromintervention. DevelopmentalPsychiatrySection,UniversityofCambridge,Cambridge, UnitedKingdom;HeddekeSnoek,DepartmentofChildPsychiatry,Uni- Terminology and Scope versityMedicalCentreUtrecht,Utrecth,theNetherlands. Preparation of this article was supported in part by a grant from the Antisocialbehaviorinchildrencanbeoperationalizedindiffer- Wellcome Foundation. We thank Tony Manstead, Menno Kruk, and entways,andthishasbeenthesourceofmuchdebate.Antisocial KatherineSheltonfortheirhelpfulcommentsonaversionofthisarticle. behaviorcanbedefinedintermsofpsychiatricdiagnoses(oppo- Correspondence concerning this article should be addressed to StephanieH.M.vanGoozen,SchoolofPsychology,CardiffUniversity, sitionaldefiantdisorder[ODD],conductdisorder[CD],disruptive TowerBuilding,ParkPlace,CardiffCF103AT,UnitedKingdom.E-mail: behavior disorder [DBD]; Diagnostic and Statistical Manual of [email protected] MentalDisorders,4thed.[DSM–IV];AmericanPsychiatricAsso- 149 150 VANGOOZEN,FAIRCHILD,SNOEK,ANDHAROLD ciation, 1994), in terms of the violation of social or legal norms aggressively when provoked. Thus, although the distinction be- (delinquency, criminality), or as aggressive behavior (Plomin, tweenhostileandinstrumentalaggressionhasbeenappealingfrom Nitz,&Rowe,1990).Theseoperationalizationsarerelatedbutnot aphenomenologicalordescriptivepointofview,weshouldkeep synonymous.Giventhatfewstudieshaveexaminedneurobiolog- inmindthatthedifferentsubtypes,iftheyexistatall(seeBush- icalfactorsinchildhood-onsetantisocialbehavior,wedecidedto man & Anderson, 2001), are not mutually exclusive, and that it adopt relatively broad inclusion criteria, such that studies exam- maywellbeunreasonabletoexpectadichotomyinasituationin ininganyofthefouroperationalizationsareincludedinthepresent which there is a complex interplay between different brain struc- review (i.e., psychiatric, antisocial, violent, and aggressive; see turesandmultiplehormonalandarousalsystems. Lynam, 1996; Moffitt, 1993, 2005). Following Rhee and Wald- man’s (2002) suggestion, we use the term antisocial behavior to Antisocial Behavior in Child Psychiatry refertothisomnibusoperationalization. When aggressive and antisocial behavior becomes a pervasive patternthataffectsdiversedomainsofchildren’sfunctioning,one Antisocial Behavior and Aggression Subtypes referseithertoanODDoraCD(DSM–IV;seeTable1).Theterm Aggression,definedasbehaviordeliberatelyaimedatinflicting disruptivebehaviordisorderencompassesbothODDandCD.The physical and/or psychological damage on persons or property, prevalenceofthesedisordersisrelativelyhigh:2.0%forCDand representsaproblemofsignificantclinicalandsocialconcern.In 3.2%forODD(Lahey,Waldman,&McBurnett,1999).Theprob- psychiatry, aggression does not constitute a separate diagnostic lembehaviorofchildrenwiththesedisordersisoftenquitestable entity itself, but it appears in several psychopathological condi- and persistent (Offord et al., 1992). Conduct problems in child- tions, the most important being CD, substance use disorder, neu- hoodareassociatedwithahostofnegativeoutcomesinadulthood; rological disorders involving the frontal and temporal lobes, and they predict not only future antisocial behavior (Fombonne, personalitydisorderssuchasborderlineandantisocialpersonality Wostear,Cooper,Harrington,&Rutter,2001;Zoccolillo,Pickles, disorder(APD;asdescribedintheDSM–IV;AmericanPsychiatric Quinton, & Rutter, 1992; Zoccolillo & Rogers, 1991) but also Association,1994). substance abuse and dependence in adulthood (Kazdin, 1995; Despite its common occurrence and clinical importance, there Offord&Bennett,1994),earlypregnancyinantisocialgirls(Bar- havebeenfewattemptstoclassifyaggressivebehaviorsystemat- done et al., 1998), persistent health problems (Bardone et al., ically for clinical purposes (Vitiello & Stoff, 1997). The psycho- 1998),andotherformsofpsychiatricillnesssuchasdepression. socialcomplexityofthisbehavioranditsetiologicalheterogeneity Althoughtheshort-termeffectivenessofinterventionstrategies aremajorobstaclestoanyclinicallyvalidsubtyping.Thissituation (e.g., parent management training, cognitive behavioral therapy) contrasts with the literature on nonhuman species: Subtypes of hasbeendemonstrated(Kazdin,2001),thelong-termeffectiveness aggression have been identified using behavioral, neuro- of treatment appears to be limited (Offord & Bennett, 1994). In anatomical, and biochemical approaches (Miczek, Fish, De particular,thehighpersistence,poorprognosis,andlimitedeffec- Bold, & de Almeida, 2002; Moyer, 1976). In humans, two tiveness of treatments of childhood antisocial behavior lend im- major forms of aggression have been identified: an impulsive- portance to the investigation of biological correlates of antisocial reactive-hostile-affective(impulsive)subtypeandacontrolled- behavior in childhood. An understanding of these factors in anti- proactive-instrumental-predatory (controlled) subtype, which are social children should generate hypotheses concerning both the qualitatively different from each other with respect to their phe- underlying neurobiological mechanisms and the etiology of anti- nomenologyandneurobiologicalfeatures(Vitiello&Stoff,1997). social behavior. Furthermore, biological studies of antisocial be- Impulsive aggression is typically explosive, uncontrolled, ac- haviorcouldleadtonewapproachestothetreatmentofpsychiatric companied by anger or fear, and characterized by high levels of conditions that are associated with aggression. Such approaches arousal. It may also be self-destructive. There is an increasing might involve pharmacological interventions to manipulate indi- bodyofresearchthathaslinkedimpulsiveaggressionandneuro- rectlyorperhapsevendirectlythebiologicalsubstratesofaggres- transmitter abnormalities, specifically a reduction of serotonergic sion or lead to hypotheses that influence the content of activity(Stoff&Vitiello,1996;seebelow).Incontrast,individuals psychotherapy. whomanifestnonimpulsive,controlled(predatory)aggressionare We critically evaluate the evidence that antisocial behavior, as lesslikelytobeaffectivelyunstable;theiraggressionisinstrumen- observedinchildpsychiatricpatientswithDBD,stemspartlyfrom tal,inthesensethatitisnormallyusedtoachieveagoalbeyond changes in the developing neurobiological systems of these chil- harmingavictim.Thelevelofarousalofindividualsengagingin dren.Inparticular,wefocusontherolesoftheHPAaxisandthe thisformofaggressionisthoughttobelow,asevidencedbytheir 5-hydroxytryptamine (5-HT; serotonin) neurotransmitter system, low baseline heart rate (HR) and skin conductance (SC) levels, the functions of which may be altered in children with antisocial althoughitisnotknownwhethertheirarousallevelsarealsolow behavior,particularlythatwhichhasitsonsetearlyinlife. duringthecommissionofviolentorantisocialactsthemselves. Atpresentitisunclearwhetherthedistinctionbetweenimpul- Clinical Subtyping and Overlap sive and controlled aggression applies to childhood aggression, although it has been suggested that child psychiatric cases with Conduct problems and attention-deficit/hyperactivity disorder aggressive behavioral problems more often show impulsive ag- (ADHD) are highly overlapping behavioral disorders, and have gression than controlled aggression (Vitiello & Stoff, 1997). It is been found to co-occur in 30%–50% of cases in epidemiological likely, however, that those individuals who employ instrumental andclinicalsamples(Lynam,1996).Althoughtheissueofwhether forms of aggression under some circumstances will also behave DBD and ADHD are distinct categories is still controversial, NEUROBIOLOGYINANTISOCIALCHILDREN 151 Table1 DiagnosticandStatisticalManualofMentalDisorders(4thed.)DiagnosticCriteriafortheDisruptiveBehaviorDisorders: OppositionalDefiantDisorderandConductDisorder Criteria Subcriteria Oppositionaldefiantdisorder Apatternofnegativistic,hostileanddefiantbehavior 1.Oftenlosestemper lastingatleast6months,duringwhichfour(or 2.Oftenargueswithadults more)ofthefollowingsubcriteriaarepresent: 3.Oftenactivelydefiesorrefusestocomplywithadults’requestsorrules 4.Oftendeliberatelyannoyspeople 5.Oftenblamesothersforhisorhermistakesormisbehavior 6.Isoftentouchyoreasilyannoyedbyothers 7.Isoftenangryandresentful 8.Isoftenspitefulorvindictive Thedisturbanceinbehaviorcausessignificant impairmentinsocial,academic,oroccupational functioning. Thebehaviorsdonotoccurexclusivelyduringthecourse ofapsychoticormooddisorder. Criteriaarenotmetforconductdisorderand,if individualisage18yearsorolder,criteriaarenot metforantisocialpersonalitydisorder. Conductdisorder Arepetitiveandpersistentpatternofbehaviorinwhich eitherthebasicrightsofothersormajorage- appropriatesocietalnormsorrulesareviolated,as manifestedbythepresenceofthree(ormore)ofthe followingsubcriteriainthepast12months,withat leastonecriterionpresentinthepast6months: Aggressiontopeopleandanimals 1.Oftenbullies,threatens,orintimidatesothers 2.Ofteninitiatesphysicalfights 3.Hasusedaweaponthatcancauseseriousphysicalharmtoothers(e.g.,abat,brick, brokenbottle,knife,gun) 4.Hasbeenphysicallycrueltopeople 5.Hasbeenphysicallycrueltoanimals 6.Hasstolenwhileconfrontingavictim(e.g.,mugging,pursesnatching,extortion, armedrobbery) 7.Hasforcedsomeoneintosexualactivity Destructionofproperty 8.Hasdeliberatelyengagedinfiresettingwiththeintentionofcausingseriousdamage 9.Hasdeliberatelydestroyedothers’property(otherthanbyfiresetting) Deceitfulnessortheft 10.Hasbrokenintosomeoneelse’shouse,building,orcar 11.Oftenliestoobtaingoodsorfavorsortoavoidobligations(i.e.,“cons”others) 12.Hasstolenitemsofnontrivialvaluewithoutconfrontingavictim(e.g.,shoplifting, butwithoutbreakingandentering;forgery) Seriousviolationsofrules 13.Oftenstaysoutatnightdespiteparentalprohibitions,beginningbefore13years 14.Hasrunawayfromhomeovernightatleasttwicewhilelivinginparentalorparental surrogatehome(oroncewithoutreturningforalengthyperiod) 15.Oftentruantfromschool,beginningbeforeage13years Thedisturbanceinbehaviorcausessignificant impairmentinsocial,academic,oroccupational functioning. Ifindividualisage18yearsorolder,criteriaarenotmet forantisocialpersonalitydisorder. researchers accept the notion that the two domains are at least Indeed, an important objective of our review of the neurobio- partially independent (Hinshaw, 1987). Of note is that ADHD logical bases of childhood antisocial behavior is to highlight and DBD have different background factors and correlates, evidence that ADHD and DBD are separate diagnostic catego- including socioeconomic status, family hostility, parental anti- ries with distinct neurobiological impairments. social behavior, and cognitive and achievement variables (Ly- nam, 1996). The focus of the present review is on antisocial Antisocial Children: A Developmental Problem behavior; thus, studies examining the role of neurobiological variables in clinical samples of ADHD (alone) will not be Robins’s(1966)conclusionthattomorrow’santisocialadultcan included.However,ifasamplecomprisedchildrenwithcomor- be found among today’s antisocial children still poses one of the bid DBD and ADHD, the study in question was included. mostchallengingquestionsindevelopmentalresearch.Childhood 152 VANGOOZEN,FAIRCHILD,SNOEK,ANDHAROLD conduct problems are a major risk factor for adult disorders that Etiological Factors: Genes and Family Environment arecharacterizedbyantisocialbehavior,andaggressioningeneral notonlyispredictiveofantisocialoutcomesinadulthoodbutalso Research on the etiology of antisocial behavior has focused is stable across generations (Huesmann, Eron, Lefkowitz, & almostexclusivelyontheroleofdysfunctionalfamilyinfluences, Walder, 1984; Lynam, 1996; Robins, 1966). Lynam (1996) ob- such as economic problems, parental psychopathology, coercive served that “although all chronic offenders have a history of parenting, physical abuse, and family conflict (Moffitt & Caspi, childhoodantisocialbehavior,childhoodantisocialbehaviorisso 2001). Often, these variables are considered to be environmental common that it predicts chronic offending only weakly, if at all. influences, and the possibility that they may also reflect genetic This asymmetry hobbles researchers’ attempts to identify the de- influences is not considered (see Moffitt’s, 2005, discussion of velopmentalcorrelatesofthechronicoffender...Thegoldengrail how bad parenting should be treated as a phenotype in future ofhigh-riskresearchhasbecometheidentificationoftheminority behavioral-geneticresearchonantisocialbehavior).Thisisunfor- of children who are most likely to persist in their antisocial tunatebecausedisentanglingtheinfluencesofnatureandnurtureis behavior from among the multitude of children who engage in afirststeptowardthegoalofexplainingtheetiologyofantisocial someantisocialacts”(p.211). behavior(Rhee&Waldman,2002,p.490). Lynam(1996)identifiedchildrenwithacombinationofADHD It is well known that antisocial behavior is concentrated in a andCDcharacteristicsasbeingatgreatestriskforchronicoffend- relatively small percentage of families (Farrington, Jolliffe, Loe- ing and explained their persistence by pointing out that their ber,Stouthamer-Loeber,&Kalb,2001),thatchildrenwithdisrup- behavioral,neuropsychological,andphysiologicaldeficitsaresim- tivebehavioralproblemsaremorelikelytohaveparentswhoshow ilartothoseofadultpsychopaths. antisocialbehavior(e.g.,Biederman,Munir,&Knee,1987;Lahey AdecadehaspassedsincethepublicationofLynam’sseminal et al., 1987; Stewart, deBlois, & Cummings, 1980), and that the paperandwearenowinapositiontoprovideamorecomprehen- intergenerational transmission of antisocial behavior has been sive understanding of the nature of the (neurobiological) factors well-documented(Huesmannetal.,1984). andprocessesthatcauseandmaintainthesedeficits.Indoingso, Meta-analyticstudies(Rhee&Waldman,2002)haveconcluded weattempttoexplainwhychildrenwithantisocialproblemshave that genes influence 40%–50% of population variation in antiso- neuropsychological problems and difficulties in response regula- cialbehavior,buttheproportionisgreater(60%–65%)foraggres- tion(Newman&Wallace,1993;VanGoozen,Cohen-Kettenis,et sive antisociality (Tackett, Krueger, Iacono, & McGue, 2005). al., 2004). We also explain why antisocial children have distinct Biological characteristics related to aggressive behavior may be social information-processing deficits that cause their behavioral the result of genetic variation. Within the domain of adolescent problems to become persistent (Milich & Dodge, 1984), and we antisocial behavior, studies that have distinguished between ag- propose an integrative theoretical model linking genetic factors, gressionanddelinquencyorrule-breakinghavereportedagreater early adversity, cognitive and neurobiological regulatory mecha- genetic component (65%) for persistent aggressive behavior than nisms,andchildhoodantisocialbehavior. for delinquent behavior (35%; Eley, Lichtenstein, & Moffitt, 2003). In terms of adult antisocial behavior, monozygotic twins have been shown to have a higher (35%–52%) concordance rate The Present Review forantisocialbehaviorandforcriminalconvictionthandizygotic The primary goal of our review is to discuss the evidence that twins(13%–23%).Adoptionstudiesarealsoconsistentinshowing specific neurobiological systems are involved in childhood-onset ageneticeffectonadultcriminality.Forexample,adopteesfrom antisocialbehavior.Thesefactorsareresponsiblefortheseverity abiologicalparentwithacriminalconvictionshowahigherrateof of the behavioral problems observed in antisocial children, but criminality compared with adoptees whose biological parents do they also play a role in their persistence, because they influence not exhibit antisocial behavior, even when the former group are children’sinteractionswiththeirenvironment. separatedfromtheirbiologicalparentssoonafterbirth(Bohman, We first briefly introduce two main factors in the theoretical Cloninger, Sigvardsson, & von Knorring, 1982; Cadoret, Yates, modelthatarepresentedlaterinthearticle(seeFigure1),namely Troughton, Woodworth, & Stewart, 1995). There also appears to genetic factors and early childhood adversity. In this section, we be a genetic influence on CD in adolescence, although it is con- seek to explain how the children’s genotype, through evocative siderablysmallerthanthatforadultAPD(Lyonsetal.,1995).This gene-environment processes, can create environments that rein- latter finding may reflect the fact that all the participants who forcetheirantisocialbehavior,therebyincreasingtheriskofper- engaged in antisocial behavior during adolescence were included sistenceinbehavioralproblems. in the analysis. According to Moffitt (1993), this strategy would Next,wepresenttheevidenceinsupportoftheinvolvementof runtheriskofincludingalargeproportionofadolescence-limited differenthormonalparametersinadultantisocialbehavior,before antisocials for whom the evidence for genetic influences is con- weproceedtodescribeanddiscussthefindingsfromresearchon siderablyweakerthanitisforindividualswithlife-coursepersis- children.Wepointouttheimplicationsofthesefindingsfortheory tentantisocialbehavior. and clinical practice. In a similar fashion, data are reviewed In considering the possible role of genetic mechanisms, it is suggestingthattheANSandserotonergicsystemaredysfunctional importanttorecognizethatinteractionsbetweengeneticpredispo- inindividualswithpervasiveantisocialbehavior.Wethendiscuss sitionsandtheenvironmentinwhichtheyareexpressedappearto the possible causes of deviant biological correlates in childhood becrucialintheetiologyofantisocialbehaviorproblems(Rutter, antisocialbehavior,andtowardtheendofthereview,wepresent Silberg, O’Connor, & Simonoff, 1999). A genetic predisposition an integrative theoretical model of the neurobiological bases of toward aggressive or antisocial behavior may be expressed in early-onsetantisocialbehavior. adverserearingenvironmentsinwhichthechildreceivesharshor NEUROBIOLOGYINANTISOCIALCHILDREN 153 Familial Influences Biological, Cognitive, Emotional Behavioral Outputs Mechanisms Genetic A Factors Neurobiological Deficits B E C D I G H Early Childhood F Disinhibited Cognitive J Antisocial Adversity and Emotional Problems Behavior Problems K Early Behavior Problems Time Figure1. Theproposedtheoreticalmodelrepresentsthedirectandindirectpathwaysthroughwhichfamily factors(earlychildhoodadversity,geneticfactors)arehypothesizedtoinfluenceearly-onsetantisocialbehavior problems.Boldpathsinthemodelconstitutetheoreticallyinformedandsupportedpathwaysonthebasisofour reviewoftheevidence.Dashedpathsrepresentsubstantiveinfluencesthatmayvaryonthebasisofourreview ofpriorevidence.Factorsrepresentedbycircles(e.g.,earlychildhoodadversity)constitutelatentindexesof primary theoretical constructs (i.e, they are respectively composed of correlated but conceptually distinct elements).Geneticfactorsconstituteamanifestindexandarethereforerepresentedbyabox(seeBollen,1989, forconventionsrelatingtopathmodelrepresentation).Pathsfromgeneticfactorsandearlyproblembehaviors arepresentedaslightgrayanddashed,nottosuggestthattheyrepresentnonsignificantinfluences,butratherthat they are secondary (genetic) or optional (estimation of initial symptom levels) in estimating this proposed theoreticalmodel.Notealsothatthismodelhighlightstheimportantroleofneurobiologicalmechanismsinthe etiologyofchildhoodantisocialbehavior.Itdoesnot,however,representalife-coursemodelofthepersistence anddevelopmentofantisocialbehavioracrosschildhood,adolescence,andadulthood.Rather,byhighlighting the role of neurobiological mechanisms underlying childhood antisocial behavior, this model identifies an importantsiteofintervention.Effectiveintervention,ofcourse,islinkedtopersistenceandfuturedevelopment ofantisocialbehavior. inconsistentdiscipline,orisexposedtohighlevelsofinterparental developmentofaggressiveorantisocialbehavior(Cadoret,Yates, conflict or marital breakdown (El-Sheikh & Harger, 2001; El- Troughton,Woodworth,&Stewart,1995;Caspietal.,2002;Foley Sheikh,Harger,&Whitson,2001).Conversely,theeffectsofsuch etal.,2004;seeRaine,2002,forareview). a predisposition may be minimized if the child is raised in a Thechild’sgenotypecanalsoevokenegativebehaviorfromthe positiveenvironmentinwhichtheparentsexpresswarmthtoward environment(e.g.,badparenting,negativepeercontacts),because their offspring and/or adopt a consistent, authoritative parenting the individual’s genetically influenced behavior (i.e., tempera- style. Unfortunately, antisocial individuals are more likely than ment)leadshimorhertocreate,seekout,orotherwiseendupin other people to provide an adverse rearing environment for their environments that match the genotype (Rutter & Silberg, 2002). offspring(Rutter,Giller,&Hagell,1998).Onestudythatprovides Antisocialbehaviorcanbringabouteachoftheseprocessesatany some insight into the importance of Gene (cid:1) Environment inter- point in the life course, and these active evocative gene- actions in the etiology of antisocial behavior is that of Bohman environmentprocessesareofenormousimportanceinunderstand- (1996).Inthisadopteestudyofpettycriminality,itwasshownthat ing the continuity of antisocial behavior (Caspi & Moffitt, 1995; individualslackingageneticpredispositionorenvironmentalrisk Moffitt,2005). factorshadonlya3%chanceofbecomingrecidivistcriminalsin adulthood. Individuals exposed to environmental risk had a 6% Temperament, Evocative Interactions, and Early rateofcriminality,whereasthosewithageneticpredispositionhad Adverse Effects a 12% incidence of criminality. In contrast, the incidence of criminalityinindividualswithbothgeneticandenvironmentalrisk Early brain development is especially vulnerable to the effects factorspresentwasashighas40%.Otherstudiesalsoillustratethe of environmental stressors, and this applies to both prenatal and importanceofconsideringGene(cid:1)Environmentinteractionsinthe postnatal development (Dawson, Ashman, & Carver, 2000; 154 VANGOOZEN,FAIRCHILD,SNOEK,ANDHAROLD Huizink,Mulder,&Buitelaar,2004).Evidenceforsuchenviron- cents(17–18years)havesometimesbeenreportedtobehigherin mentaleffectsonbraindevelopmentcomesfromstudiesexamin- aggressiveindividuals(Dabbs,Jurkovic,&Frady,1991;Scerbo& ing pregnancies with elevated risk because of maternal smoking, Kolko, 1994), several studies of prepubertal children have not poornutrition,exposuretoalcoholanddrugs,ormaternalpsycho- found a relationship between aggression and testosterone (Con- pathology(e.g.,Lundyetal.,1999),andatypicalcaretaker–child stantinoetal.,1993;VanGoozen,Matthys,Cohen-Kettenis,Thi- interactions such as those involving depressed mothers, irregular jssen, & Van Engeland, 1998). Moreover, lower levels of testos- and unpredictable parenting, or exposure to abuse and neglect terone have been found in children at risk for substance abuse (e.g., Ashman, Dawson, Panagiotides, Yamada, & Wilkinson, (Dawes et al., 1999). Furthermore, recent evidence indicates that 2002;Carlson&Earls,1997).Itisnowknownthatsuchenviron- testosterone levels are related to social dominance, rather than mentaleffectscanattenuatetheinfant’sstresssystemsthroughthe aggressionperse(Mazur&Booth,1998;Rowe,Maughan,Worth- mediating effects of early learning and the amygdala (Susman, man, Costello, & Angold, 2004). Thus, high testosterone levels 2006). The dampening or downregulation of the stress system to maybeassociatedwithpeergroupleadership,bothwithingroups chronic stressors over the first year of life is clearly an adaptive of healthy children and within antisocial groups, such as gangs mechanism(Gunnar&Donzella,2002),avoidingchronicarousal (Roweetal.,2004). and excessive energy expenditure that would ultimately result in In prepubertal children, an important part of the androgenic seriouspathophysiologicalconsequencesorevendeath. activityisofadrenalratherthangonadalorigin.Fromaroundthe Some children are born with a more easy-going temperament age of 6, children exhibit a gradual increase in androgens of than others. It has been established that early temperament and adrenalorigin,aperiodcalledtheadrenarche(C.R.Parker,1999), later personality are related (Caspi, Henry, McGee, Moffitt, & anditisnotuntilpubertythatgonadalandrogens,suchastestos- Silva,1995)andthatpersonalityisingeneralhighlystable(Caspi terone, become important. Therefore, it is quite likely that the &Bem,1990).Inthecaseofdifficultorhard-to-managechildren, testosterone–aggression relationship does not emerge until after their behavior evokes distinctive responses from others (parents, puberty,andresearchinprepubertalchildrenshouldthereforealso peers, and teachers) that are more likely to exacerbate than to focusonadrenalandrogens,suchasDHEA,itssulfate(DHEA-S) amelioratethechild’sexistingtendencies(Lynam,1996;Moffitt, and androstenedione. Higher levels of DHEA-S have been found 1993). Moreover, environmental factors that are independent of inbothchildandadolescentpatientswithCD(Dmitrieva,Oades, thechild’sgeneticmakeuportemperamentcanactascontributory Hauffa, & Eggers, 2001; Van Goozen, Matthys, Cohen-Kettenis, causesofpersistence.Suchfactorsincludemaritaldiscord(Harold Thijssen, & Van Engeland, 1998; Van Goozen, Van den Ban, et &Conger,1997),disruptedparenting(Erel&Burman,1995),or al., 2000). However, there are also studies that have not found parentalpsychopathology(Downey&Coyne,1990). differences in DHEA-S between aggressive and nonaggressive children(Constantinoetal.,1993). ThepositiverelationshipobservedbetweenDBDsymptomsand Neuroendocrine Features of Antisocial Behavior plasma levels of DHEA-S is of interest because DHEA-S has In line with findings from aggressive adults with APD GABA antagonistic(i.e.,neuroexcitatory)actionsandcouldcon- A (Virkkunen,1985;Woodman,Hinton,&O’Neill,1978),whichis tributetoincreasedaggression(Majewska,Demirgoren,Spivak,& almostalwaysprecededbyCDinchildhood(Robins,1978),there London,1990).ItisalsopossiblethathigherDHEA-Slevelsadd is now convincing evidence for changes in the functioning of toalargerpoolofendogenoustestosterone.Theresultsofthefew severalinterrelatedneuroendocrinesystemsinchildrenwithanti- studiesconductedsofardemonstratethatadrenalandrogenfunc- social behavior. In this section, we focus on the androgens, tes- tioningisanimportanttopicforfutureresearchontheoriginsof tosterone,anddehydroepiandrosterone(DHEA),beforediscussing aggressioninchildren. in detail the role of HPA axis hormones such as corticotropin- releasinghormone(CRH),adrenocorticotropichormone(ACTH), HPA Axis Hormones andcortisol. ThestresssystemhasitsperipherallimbsintheHPAaxisand the sympathetic adrenal medullary systems. Psychological stress Androgens associated with various life events has been widely studied with The rationale for considering androgens to be involved in ag- respect to the concentration and secretion of the adrenal stress gressivebehavioristhatmaleshavebothhigherconcentrationsof hormone cortisol. Events generally found to be stressful, as de- androgensandhigherlevelsofaggressivebehaviorthanfemales. finedbyanincreaseincortisol,includeanticipationofthedeathof Although an association between androgens and aggression has afamilymember,hospitaladmission,surgery,mentalperformance beenclearlyestablishedinanimals(e.g.,Higleyetal.,1996),the tests,publicspeaking,andanticipationofstrenuousexercise(L.N. evidence in humans is less clear cut (Archer, 1991; Archer, Parker,1989).However,considerableindividualvariationincor- Graham-Kevan,&Davies,2005).Inadults,elevatedtestosterone tisolreactivityisevidentinmoststudies.Theintensityanddura- levelsincerebrospinalfluid(CSF),plasma,andsalivahavebeen tionofastressresponseanditslong-termconsequencesappearto linked to antisocial behavior and violent crime (e.g., Banks & depend on the degree of stressor controllability, as perceived by Dabbs, 1996; Dabbs, Frady, Carr, & Besch, 1987; Dabbs, Jurk- theindividual(Huether,1996). ovic,&Frady,1991;Dabbs&Morris,1990;Ehrenkranz,Bliss,& Adrenalcortisolsecretionrepresentsthefinalstepinaneuroen- Sheard, 1974; Virkkunen et al., 1994). Studies of children and docrinecascadebeginningintheparaventricularnucleus(PVN)of adolescentshaveyieldedmixedresults(e.g.,Grangeretal.,2003; thehypothalamus.Inresponsetoactivationbylimbic,cortical,and Maras et al., 2003). Although testosterone levels in older adoles- other afferent inputs, CRH is released into the portal venous NEUROBIOLOGYINANTISOCIALCHILDREN 155 systembythePVN.CRHstimulatesthereleaseofACTHfromthe beenequivocalfindingsacrossstudies(seeTable2).Somestudies anteriorpituitary.ThereleaseofACTH,inturn,leadstoactivation havefoundassociationsbetweenreducedbasalcortisolconcentra- of the adrenal glands, resulting in the synthesis and release of tions and aggressive behavior (Kariyawasam, Zaw, & Handley, cortisol. Once secreted into general circulation, cortisol acts on a 2002; McBurnett, Lahey, Rathouz, & Loeber, 2000; Oosterlaan, variety of target cells to mobilize the physiological response to Geurts, Knol, & Sergeant, 2005; Pajer, Gardner, Rubin, Perel, & stress.Cortisolalsocrossestheblood–brainbarriertoactatsites Neal,2001;Shoal,Giancola,&Kirillova,2003;VandeWiel,Van inthecentralnervoussystem,whereitactivatesnegativefeedback Goozen,Matthys,Snoek,&VanEngeland,2004;Vanyukovetal., mechanisms that inhibit the release of CRH and ACTH, thereby 1993);otherstudiesfoundnosuchrelationship(Azaretal.,2004; reducingHPAaxisactivity(e.g.,Chrousos&Gold,1992). Kruesi et al., 1989; Scerbo & Kolko, 1994; Schulz, Halperin, The HPA system has a diurnal rhythm. Cortisol is secreted in Newcorn, Sharma, & Gabriel, 1997; Stoff et al., 1992; Van pulses,andthefrequencyofthesepulsesvarieswithtimeofday, Goozen, Matthys, Cohen-Kettenis, Buitelaar, & Van Engeland, thehighestfrequencybeingintheearlymorning,resultinginhigh 2000) or even a positive relationship (Van Bokhoven, Van early-morning levels (Deuschle et al., 1997). As the day Goozen, et al., 2005). There have also been studies that found progresses, the pulses become less frequent and cortisol levels associations between reduced basal cortisol concentrations and thereforedecrease.TheHPAsysteminchildrenreachesfunctional aggressiontowardpeers(Tennes,Kreye,Avitable,&Wells,1986) maturitybythe4thyearoflife(Genazzanietal.,1983;Herman, andhostilitytowardsteachers(Tennes&Kreye,1985)inhealthy Arthur-Smith, Hammock, & Josephs, 1988), although recent ex- school children. Interestingly, McBurnett et al. (1991) found that perimentalstudieshaveobservedanormalcircadianrhythmin12- anxious children with CD had higher cortisol levels than did to18-month-oldinfants(Goldbergetal.,2003). childrenwithCDalone. In the majority of studies referred to above, the findings have HPA Axis and Aggression been correlational in nature, with the result that it has not been possible to draw causal inferences between low cortisol concen- Studies of antisocial adults have observed a negative relation- trationsandantisocialbehavior.Therehavebeenveryfewstudies ship between cortisol levels and the magnitude of behavioral withadesigncapableofshowingthatlowcortisollevelsprecede deviation (e.g., Bergman & Brismar, 1994; Virkkunen, 1985; theonsetofantisocialoraggressivebehaviorandcanthereforebe Woodman, Hinton, & O’Neill, 1978). Lower levels of cortisol used to predict which individuals will exhibit a pervasive pattern could mean that these individuals are physiologically under- of antisocial behavior throughout adolescence and into young aroused(seeArousalTheoriesofAggressionsection,below),that adulthood. The study that comes closest to demonstrating a lon- the negative feedback mechanisms acting on their HPA axes are gitudinal, albeit noncausal, link between low cortisol levels and hypersensitive,orthattheyhaveanincreasedthresholdforstress aggressivebehavioristhatofShoaletal.(2003).Thiswasa5-year (Kruesi, Schmidt, Donnelly, Hibbs, & Hamburger, 1989). Most longitudinal study investigating the relationship between cortisol studies have focused on cortisol levels as an index of HPA axis concentrations at ages 10–12 years and aggressive behavior, as activity.Althoughthemeasurementofcortisolprovidesimportant measured using the Youth Self-Report Scale (Achenbach, 1991), information in this respect, cortisol is secreted by the adrenal at ages 15–17 years. The results showed that low cortisol was cortex and therefore reflects a relatively late response. Only four moderately predictive of aggressive behavior 5 years later and, studieshavemeasuredACTHorCRHinaggressiveorantisocial further, that this relationship appeared to be mediated by the individuals. Virkkunen and Linnoila (1993) and Virkkunen et al. effectsofcortisolonapersonalityvariabletheauthorstermedlow (1994) found lower levels of ACTH in the CSF of alcoholic, selfcontrol(Shoaletal.,2003).Afurtherstudywithalongitudinal impulsive offenders with an APD compared with healthy control component was performed by McBurnett et al. (2000). They participants, and they found no differences in CSF CRH levels. demonstrated that clinic-referred boys with consistently low cor- Susman et al. (1999) reported that CRH levels were lower in tisol levels in samples obtained 2 years apart showed the highest pregnant teenage mothers with DBDs compared with teenage levelsofaggressiveCDsymptomsovertime.Alimitationofthis mothers without symptoms of DBD. In contrast, Dmitrieva et al. studywasthatitfailedtocontrolforthetimeofsalivacollection. (2001)foundhigherplasmaACTHlevelsinadolescentswithCD. Anotherpointofinterestisthatbasalglucocorticoidconcentra- MeasuringACTHandCRHforscientificpurposesischalleng- tions have been reported to be moderately heritable in humans ingbecausethesamplinghastobeperformedinplasmaorCSF. (Inglis et al., 1999; Meikle, Stringham, Woodward, & Bishop, Given the low tolerability of the measurement procedures in- 1988),soitseemsunlikelythatbasalcortisollevelscouldfallafter volved, particularly lumbar puncture for ascertainment of CSF the onset of antisocial behavior. Finally, it should be noted that samples,andtheaccompanyingethicalissues,relativelyfewstud- parental antisocial personality symptom counts have been shown ies have examined the role of these parameters in childhood to be inversely related to cortisol concentrations in their children aggression. However, given the importance of the HPA axis sys- (Vanyukov et al., 1993). This suggests that cortisol may be in- tem in psychopathology in general, and antisocial behavior in volvedintheintergenerationaltransmissionofantisocialbehavior particular, it will in future be necessary to develop methods andprovidesfurtherindirectevidencethatthissteroidplaysarole of studying the influence of these higher level peptides and intheetiologyofantisocialbehavior. hormones. In addition to the difficulties arising from the correlational natureofthemajorityofstudies,anotherfeatureofresearchinthis Empirical Findings in Children areaisthatresultshavebeenrelativelymixed.Thismaybedueto Few studies have been conducted on cortisol levels or the the use of different methods for collecting cortisol (salivary vs. cortisol response to stress in aggressive children, and there have plasma vs. urinary free cortisol, and under basal conditions vs. 156 VANGOOZEN,FAIRCHILD,SNOEK,ANDHAROLD Table2 StudiesontheRelationshipBetweenHypothalamic-Pituitary-Adrenal(HPA)AxisIndicesinAggressiveChildrenandAdolescents Participants Meanage Agerange Study HPAaxisindex Sample (years) (years) Outcome Kruesietal.(1989) Urinarycortisol 19ADHD/CD/ODDboys 10.5 Nodifferences 19HCboys 10.5 McBurnettetal.(1991) Salivacortisol 67clinic-referredCDboys 8–13 CD/ANX(cid:2)(cid:3)CD/ANX(cid:4) Stoffetal.(1992) Plasmacortisol 8maleadolescentDBDboys 14.7 Nodifferences 8maleadolescentHCboys 15.3 Vanyukovetal.(1993) Salivacortisol 78boyswithPSUDfather; 10–12 NegativecorrelationwithCDsymptom 72boyswithHCfather count Scerbo&Kolko(1994) Salivacortisol 40clinic-referredDBDboys 7–14 Negativecorrelationwithinattention/ overactivity;positivecorrelation withinternalizingbehavior Mossetal.(1995) Salivacortisol 81boyswithPSUDfather; 10–12 BoyswithPSUDfather(higher Stress 103boyswithHCfather aggressiveanddelinquency)(cid:5)boys withHCfather Schulzetal.(1997) Plasmacortisol 50clinic-referredADHDboys: Nodifferences 23aggressive 9.0 27nonaggressive 9.0 VanGoozen,Matthys, Salivacortisol 21ODDboys 10.2 ODD(cid:5)NC;cortisolincreaseduring Cohen-Kettenis, Stress 31HCboys 9.6 stressstrongestinHE/HA;cortisol Gispen-deWied,etal. decreasestrongestinHE/LA (1998) Mossetal.(1999) Salivacortisol 120boyswithPSUDfather 10.8 BoyswithPSUDfather(cid:5)boyswith Stress 178boyswithHCfather 11.0 HCfather:cortisolresponsepriorto evokedrelatedpotentialtesting McBurnettetal.(2000) Salivacortisol2- 38clinic-referredDBDboys 7–12(at Correlationbetweenlowcortisoland yearfollow-up entry) persistence/earlyonsetaggression VanGoozen,Matthys,et Salivacortisol 26DBDchildren 10.1 Baseline:DBD(cid:6)HC al.(2000) Stress 26HCchildren 10.0 Stress:DBD(cid:5)HC Dmitrievaetal.(2001) PlasmaACTH 28CDboys 13.2 CD(cid:3)NC 13HCboys 13.9 Pajeretal.(2001) Plasmacortisol 47CDgirls 16.6 CD(cid:5)NC 37HCgirls 16.0 Kariyawasametal. Salivacortisol 32ODD/ADHD 10.6 ODD/ADHD(withoutstimulant (2002) 25HC 11.3 medication)(cid:5)HC Shoaletal.(2003) Salivacortisol 314HCboys 10–12 Lowcortisolrelatedtolowself-control 5-yearfollow-up andmoreaggressivebehavior5 yearslater Snoeketal.(2004) Salivacortisol 26HC Baseline:nodifferences Stress 15ODD Stress:HC(cid:6)ADHD(cid:3)ODD(cid:6) 31ODD/ADHD ODD/ADHD 23ADHD VandeWieletal.(2004) Salivacortisol 22DBDchildren Lowbaselinecortisolrelatedtomore Stressandtreatment severeDBD;lowcortisolstress follow-up responserelatedtolessimprovement afterintervention Oosterlaanetal.(2005) Salivacortisol 25childrenvaryinginlevelsof 6–12 High(aggressive)CDsymptom antisocialbehavior severityassociatedwithlowcortisol VanBokhoven,Van Salivacortisol 194maleyouths 13 HighercortisolinCDboys(n(cid:6)19); Goozen,etal.(2005) highercortisolinphysically aggressiveboys(n(cid:6)75);higher cortisolinhighreactiveaggressive boys(n(cid:6)39) Note. ADHD(cid:6)attention-deficit/hyperactivitydisorder;CD(cid:6)conductdisorder;ODD(cid:6)oppositionaldefiantdisorder;HC(cid:6)healthycontrolsubjects; ANX(cid:6)anxiety;DBD(cid:6)disruptivebehaviordisorder;PSUD(cid:6)psychoactivesubstanceusedisorder;HE(cid:6)highexternalizing;HA(cid:6)highanxiety;LA(cid:6) lowanxiety;ACTH(cid:6)adrenocorticotropichormone. during stress), different definitions of antisocial behavior or ag- cortisol concentration, which is particularly problematic if they gression, and varying methods of assessing antisocial behavior failedtocontrolforthetimeofdayofsalivaorplasmacollection. (self-reportvs.multipleinformants,andusingstructuredorsemi- Inaddition,severalstudiesdidnotincludeahealthycontrolgroup structureddiagnosticinterviews,questionnaires,orpolicerecords). orusedonlyhealthychildrenorclinicalcasesasparticipants(and A large number of studies made only a single measurement of therefore may have looked for variability within a narrowly de- NEUROBIOLOGYINANTISOCIALCHILDREN 157 fined group differing only moderately in terms of behavioral patients (CD/ODD) and is not related to aggressive behavior per deviance).Giventheselimitationsandmethodologicaldifferences, se. Indeed, Snoek, Van Goozen, Matthys, Buitelaar, and Van itisnoteworthythatstudieshaveneverthelessreportedconvergent Engeland (2004) found that child psychiatric patients suffering findings of an inverse relationship between cortisol levels and from ADHD showed a typical stress-induced cortisol response, antisocialbehavior. whereaschildrenwithDBDdidnot. Wherever possible, we calculated effect sizes for the studies The effect size value for cortisol reactivity across the four cited above investigating the relationship between basal cortisol studiesthathaveinvestigatedthis(VanGoozen,Matthys,Cohen- levelsandDBDoraggressivesymptoms,andwefoundthemean Kettenis, Gispen-de Wied, et al., 1998; Van Goozen, Matthys, et effectsizeacrossstudiestobed(cid:6)(cid:4)0.79.However,thisvalueis al.,2000;Mossetal.,1995,1999)isd(cid:6)0.57,whichcorresponds biasedbyanumberofsmallstudieswithdisproportionatelylarge to a moderate effect size (after correcting for sample size the effectsizes;theweightedaverageeffectsize,takingintoaccount weighted mean effect size is d (cid:6) 0.42). The effect sizes for the theeffectofeachstudy’ssamplesize,wasconsiderablylowerat groupeffectsinVanGoozenandcolleagues’(VanGoozen,Mat- d (cid:6) (cid:4)0.40. This indicates that there is, on average, a small to thys,Cohen-Kettenis,Gispen-deWied,etal.,1998;VanGoozen, moderateeffectsize(Cohen,1988)acrossstudiesinthedirection Matthys, et al., 2000) studies, which used a stressor involving of an inverse relationship between basal cortisol levels and DBD competition, social evaluation, and achievement stress were 0.98 symptoms.Therangeofeffectsizesforthesestudieswasbetween (large effect) and 0.61 (moderate effect), respectively, whereas 0.48and(cid:4)3.70,indicatingthattherewasagreatdealofvariability Moss et al. (1995, 1999) had smaller effect sizes (ds (cid:6) 0.32 and intermsofthedirectionandsizeofeffectsobserved.Thisappears 0.36, respectively), probably because the studies used a large, consistent with the suggestion that there were numerous method- heterogeneoussample(thesonsoffatherswhohaddrugproblems) ological differences and variations in sample characteristics be- andarelativelyweakstressor. tween studies. Unfortunately, for three studies (McBurnett et al., ItisknownthatchildrenwithDBDhaveoftenbeenexposedto 1991,2000;Susmanetal.,1999)wecouldnotcalculatetheeffect sizesbecausetheydidnotprovidevaluesforthemeansorstandard adverse rearing circumstances involving neglect, abuse, and do- deviationsoftherespectivegroups’basalcortisolvalues,orthey mesticviolence,butitisofcoursealsotruethattheirownproblem did not report a correlation (or R2 equivalent) value for the rela- behavior elicits negative responses from peers, siblings, and par- tionshipbetweencortisolandODD/CDsymptoms.Itshouldalso ents, which might be experienced as stressful (Kazdin, 1995). It benotedthatthecorrelationvalueusedtoyieldaneffectsizefor couldbethatfrequentexposuretostressfulsituationshasresulted Scerbo and Kolko (1994) was that for the relationship between inahabituationamongthesechildrento(sometypes)ofstress,and cortisol and staff-rated oppositional behavior; the other correla- as a result, they show low stress reactivity. In this context, it is tionsbetweencortisolandmeasuresofaggressivebehavior(e.g., important to study the individual interpretation of the stressful parentratings)werenotsignificant. eventtofindoutwhetherthesubjectiveexperienceoftheantiso- Findings of reduced basal levels of cortisol in antisocial indi- cialchildrenisinlinewiththeirphysiologicalexperience.Intwo vidualscouldsupportthestimulation-seekingtheory(seebelow). studies,antisocialchildren,whodidnotshowacortisolresponse Thissuggeststhatantisocialchildrenseekoutstressfulsituations, during stress, reported and showed intense emotions (i.e., they for example fights, to increase their aversive, low basal cortisol reacted more aggressively towards their opponent), suggesting a levels. On the other hand, the more often such children get in- mismatchbetweensubjectiveandphysiologicalarousal(Snoeket volvedindangerousorstressfulsituations,themorelikelytheyare al.,2004;VanGoozen,Matthys,etal.,2000). eventually to habituate to these stimuli and subsequently show a Arecentstudyshowedthatantisocialchildrenwithabluntedor bluntedcortisolresponsetostress.Theeffectoffrequentexposure absent cortisol response to stress showed the least improvement (i.e.,habituation)butalsoalackofanticipatoryfear,assuggested following a therapeutic intervention (Van de Wiel et al., 2004). by the fearlessness theory (see below), would be better studied Children with similar levels of externalizing behaviors to the understressfulconditions. cortisolnonresponders,butwhoexhibitedrelativelytypicalcorti- Moss and colleagues found that at-risk sons of fathers with a solresponsesinastressfulsituation,werefoundtorespondmore psychoactive substance use disorder secreted less cortisol in an- favorably to treatment. These findings show that, despite mani- ticipation of stress (Moss, Vanyukov, & Martin, 1995; Moss, festingsimilarlevelsofaggressiveoroppositionalbehaviorasdid Vanyukov, Yao, & Kirillova, 1999). In two psychological chal- their disruptive peers, the children with attenuated HPA axis lenge studies, Van Goozen, Matthys, Cohen-Kettenis, Gispen-de reactivityhadapoorerprognosis.Thesedatacanbeinterpretedin Wied,etal.(1998)andVanGoozen,Matthys,etal.(2000)found twoways:First,childrenwithHPAaxisdysfunctionsimplyhave thatchildrenwithODDhadlowercortisollevelsthandidhealthy amoreseriousandingrainedformofthedisorder,despiteshowing controlparticipantswhenexposedtofrustrationandprovocation. similar rates of externalizing behavior as their peers, or second, Specifically,thelatterstudyreportedthatchildrenwithODDand healthycontrolsubjectsdidnotdifferintermsofbaselinecortisol impairment of HPA axis functioning may prevent the types of levels,butthestress-inducedincreaseinsalivacortisolobservedin cognitive or emotional processing that play a critical role in the the control group was absent in the ODD group. Although in- therapeuticprocess.Thelatterinterpretationsuggeststhatchildren creasedcortisolstressreactivityhasalsobeenfoundinrelationto with early-onset antisocial behavior and either low basal cortisol aggressivebehavior,thishasbeenreportedinonlyhealthysubjects levels or attenuated cortisol reactivity may be more effectively andcommunitysamples(Gerraetal.,1997;Scarpa,Fikretoglu,& treatedusingpharmacologicallybasedtherapiesthatreinstatetyp- Luscher,2000).Theseresultssuggestthatapatternoflowcortisol icalHPAaxisfunctioning,perhapsasaprecursororanadjunctto reactivity during stress is a specific characteristic of antisocial psychologicalformsoftreatment. 158 VANGOOZEN,FAIRCHILD,SNOEK,ANDHAROLD Interactions Between Cortisol and DHEA formsofaggressionseemtoberelatedtoareducedfunctioningof in Antisocial Children the HPA axis. How can these apparently opposing findings be reconciled? Haller and colleagues (Haller, Hala´sz, Mikics, & Although there is evidence that baseline cortisol levels and Kruk,2004;Haller,vandeSchraaf,&Kruk,2001)havedeveloped cortisolreactivityarereducedandDHEA-Slevelsareelevatedin a hypoarousal-driven aggression model in animals, which is in- children with DBD (Van Goozen, Matthys, et al., 2000; Van tended to mimic human pathological aggression, specifically the Goozen, Matthys, Cohen-Kettenis, Gispen-de Wied, et al., 1998; consequencesofchronicallylowlevelsofcortisolasobservedin Van Goozen, Matthys, Cohen-Kettenis, Thijssen, & Van Enge- highlyviolentorpsychopathicadultsandaggressivechildrenwith land, 1998; Van Goozen, Van den Ban, et al., 2000), it is worth CD.Inthishypoarousalmodel,theanimalbehavesinapredatory noting that there appear to be important functional interactions fashion and appears to interpret the presence of any conspecific, between these adrenal steroids in the CNS. In particular, DHEA even a much smaller animal that does not represent an equal appears to moderate some of the effects of cortisol, and it may opponent,asathreat.Hallerandcolleaguesshowedthatalthough even antagonize its effects in some brain areas, such as the hip- the central amygdala is not involved in aggression relating to pocampus (Kimonides, Spillantini, Sofroniew, Fawcett, & Her- hyperarousal (i.e., escalation of aggression in healthy, intact ani- bert, 1999). It is therefore important to examine the interactions mals),itisstronglyactivatedintheglucocorticoiddeficientanimal betweenthesehormonesinantisocialchildren.Thisapproachhas during fighting situations (see Figure 2). This may explain such paid dividends in the study of adolescent depression, in which it animals’aberrantbehaviorandtheirevaluationofsocialsituations has been found that an elevated cortisol/DHEA ratio predicts the asmorethreateningthanisjustified. onset of persistent depression (Goodyer, Herbert, & Tamplin, Specifically, Haller et al. (2001) removed the adrenal glands of 2003). The absolute levels of cortisol were found to be less rodents and implanted slow-release pellets to provide a continuous, important than the ratio between levels of cortisol and DHEA-S, low level of corticosterone (the main glucocorticoid hormone in leadingtosuggestionsthatdepressionmightbeassociatedwitha rodents). Using this model of glucocorticoid deficiency, Haller and syndrome of functional hypercortisolaemia, in which moderate colleagues showed that rats with low concentrations of circulating increases in cortisol levels are amplified by reductions in the corticosteroneexhibitabnormallevelsofaggressivenessandattempt hormones that antagonize some of the effects of cortisol. DBDs to inflict the maximum amount of damage to a conspecific in a maybeassociatedwiththereversesyndrome,afunctionalhypo- nonresident-intruder fighting situation (Hala´sz, Liposits, Kruk, & cortisolaemia, in which the effects of reduced cortisol levels are Haller,2002;Halleretal.,2001,2004).Innormalcircumstances,a exacerbatedbyincreasesinDHEA-Sconcentrations. residentanimalconfrontedwithasmallerintruderwillprovidethreat cuestodiscouragetheintruderfromattemptingtofightandwilldirect Stress-Induced and Pathological Aggression: itsblowstononvulnerableareasofthebody(Blanchard,Hori,Rodg- Animal Research ers,Hendrie,&Blanchard,1989;Halleretal.,2001).Incontrast,rats withglucocorticoiddeficiencywereshowntoprovidefarfewerthreat Stressisaprimaryfactorinpromotingaggressionandviolencein cues to the intruder and to direct the majority of their attacks to humans. Indeed, recent studies have suggested that aggression is vulnerablepartsoftheopponent’sbody,suchasthehead,stomach,or associatedwithbothstressandincreasedcortisollevels(Gerraetal., neck(Halleretal.,2001).Acuteinjectionsofcorticosteronepriorto 1997). However, how stress mechanisms and the mechanisms in- exposuretothefightingsituationwereshowntopreventthisabnor- volved in aggression interact on a neurobiological level has only mal pattern of aggressive behavior. The authors concluded that an recentlyreceivedattention.Kruk,Halasz,Meelis,andHaller(2004) acuteincreaseincortisolisimportantinmakingthecorrectinterpre- demonstratedafastpositivefeedbackloopbetweentheadrenocortical tationofthetypeofsocialconflictoneisdealingwith.Ifonefailsto stressresponseandthebrainareathatcontrolsaggression,thatis,the initiateanacutecortisolresponse,itmaybemoredifficulttoevaluate hypothalamus.Specifically,stimulationoftheaggressiveareainthe thetruenatureoftheconflict. hypothalamusrapidlyactivatedtheHPAaxis,evenintheabsenceof In a subsequent study, Haller’s group (Hala´sz et al., 2002) an opponent or in the context of physical fighting. Hypothalamic investigated the neural background of glucocorticoid dysfunction aggressionwasalsoquicklyfacilitatedbyaninjectionofcorticoste- indefensiveaggression.Theyfoundthattheabnormalbehavioras rone.Theauthorsconcludedthatthisfast,mutual,positivefeedback shown by adrenalectomized rats did not involve activation of the system could contribute to the precipitation and the escalation of brainareasthatarenormallyinvolvedinthecontrolofaggression violentbehaviorunderstressfulconditions(Kruketal.,2004).These (cortex, amygdala, septum, hypothalamus, periaquaductal grey, findingshaveclearimplicationsforaggressionregulationandtreat- andlocuscoeruleus).Instead,thebrainareasinvolvedinmediat- ment.Hypothalamicaggressionisselectivelysensitivetoserotonergic ing the stress response (the parvocellular part of the PVN of the medicationandthebeta-blockerpropranolol.Alternatively,regulation hypothalamusorpPVN)andfearreactions(centralnucleusofthe oftheadrenocorticalstressresponsewithCRFantagonistsorcertain amygdalaorCeA)wereactivated.Hala´szetal.(2002)concluded anxiolytics,whichreducedifferentstress-inducedbehaviors,maybe that abnormal aggressiveness due to glucocorticoid hypofunction effectiveincounteractingacutestress-precipitatedviolence(Kruket is related to increased sensitivity to stressors and fear-eliciting al.,2004). stimuli, possibly in the sense that signals coming from the oppo- Thusinhealthyanimalsandhumans,aggressionispotentiated nentsaremisinterpreted,resultinginabehavioralresponsethatis by high levels of arousal, an effect that is partly mediated by characteristicofmorecriticalsituations. elevationsincorticosteroidconcentrations.Inthisrespectthistype Howcanthesefindingsbeappliedtothehumansituation?First, of aggression can be thought of as reactive. However, we earlier we speculate that glucocorticoid hypofunction in antisocial chil- reviewed evidence that pathological or persistent and deviant dren might enhance their fear reactions, resulting in an overreac-