Chapter

in Crime~
Public
Policies
J.QWilson
and J.Pet~rsilia
ICS Press:
Oakland,
California

The
Basis

for
Crime
Control
(Editors)
2002.
(pp.
4374)

Biological
of Crime

ADRIAN

RAINE

ecognition is increasing that biological processesare at some level

implicated in the development of criminal behavior. There is certainly debateabout the precisecontribution of such factors to crime
outcome, and there is considerabledebateabout the precisemechanismsthat
thesebiological factors reflect. Yet few serious scientistsin psychology and
psychiatry would deny that biological factors are relevant to understanding
crime, and public interest in and understanding of this perspective are
increasing.The discipline of criminology, on the other hand, has been reluctant to embrace this new body of knowledge. Part of the reason may be
interdisciplinary rivalries, part may simply be a lack of understanding, and
part may be due to deep-seatedhistorical and moral suspicionsof a biological approach to crime causation. For whatever reason,these data have been
largely ignored by criminologists and sociologists.
It is hoped that this chapter will go some way to allaying these suspicions. Certainly many reasons exist to take this body of knowledge
seriously.Biology is not destiny, and we can benig~ly change many of the
biological predispositions that shape the violent offender. One of the reasonswhy we have been so unsuccessfulin preventing adult crime is because
interventions to date have systematically ignored the biological side of the
biosocial equation that produces crime. If we are to be truly serious about
tackling crime and violence in society, we need to give more attention to
the biological factors that causecrime.
This chapter will first outline the evidence for agenetic predisposition
to crime. If genetic factors are indeed involved, then there has to be some
biological basis to crime (although it should also be made clear that
environmental factors in addition give rise to biological risk factors for
crime). It then turns to a discussion of psychophysiological factors that

43

44

Adrian

Raine

predispose to crime and how one heritable influence (low physiological

arousal) is thought to be the best-replicated biological correlate of antisocial behavior in child and adolescentsamples.Technical advanceshave led
to the ability to look directly at the brains of violent and criminal offenders, and the next section reviews the area of brain imaging and findings
of prefrontal functional and structural deficits in adult offenders. Other
biological processesare then briefly reviewed, including birth complications, minor physical anomalies, nutrition, hormones, neurotransmitters,
and molecular genetics.l Finally, policy implications of this researchwill be
outlined, covering intervention and prevention programs aimed at reducing the effect of biological risk factors, and also implications of brain
imaging researchfor the criminal justice system.
GENETICS
Twin

Studies

The twin method for ascertaining whether a given trait is to any extent
heritable makes useof the fact that monozygotic (MZ) or "identical" twins
are genetically identical, having lOO percent of their genes in common
with one another. Conversely, dizygotic (DZ) or "fraternal" twins are less
genetically alike than MZ twins, and are in fact no more alike genetically
than non-twin siblings.
When the trait being measured is a dichotomy (for example, criminal/noncriminal), "concordance" rates are calculated for MZ and DZ
twins separately.A 70 percent concordance for crime in a set of MZ twins,
for example, would mean that if one of the MZ pair is criminal, then the
chance of the co-twin being criminal is 70 percent. Similar concordance
ratescan be calculated for DZ twins. IfMZ twins have higher concordance
rates for crime than DZ twins, then this constitutes some evidence for the
notion that crime has a heritable component. The difference between these
correlation coefficients, when doubled, gives an estimate of heritability, or
the proportion of variance in criminality that can be attributed to genetic
influences (Falconer 1965).
Are identical twins more concordant for criminality than fraternal
twins? The answer from many reviews conducted on this expanding field
is undoubtedly yes. As one example, a review of all the twin studies of
crime conducted up to 1993 showed that although twin studies vary
widely in terms of the age, sex, country of origin, sample size, determination of zygosity, and definition of crime, neverthelessall thirteen studies of

The Biological

Basis

of

Crime

45

crime show greater concordance rates for criminality in MZ as opposed

to PZ twins (Raine 1993). If one averagesconcordance rates across all
studies (weighting for sample sizes), these thirteen studies result in concordancesof 51.5 percent for MZ twins and 20.6 percent for DZ twins.
Furthermore, the twin studies that have been conducted since 1993 have
confirmed the hypothesisthat there is greaterconcordancefor antisocial and
aggressivebehavior in MZ relative to DZ twins (for example, SlutSkeet al.
1997; Eley, lichenstein, and Stevenson1999).
Twin studies have methodological limitations that restrict the conclusions that can be drawn from individual studies. A very common criticism
of twi~ studies is that MZ twins may share a more common environment
than bz twins. For example, parents may treat MZ twins in a more similar fashion than DZ twins, thus artificially raising concordance rates in
MZ twins. If this were true, the greater concordance for crime in MZ
twins may be due more to environmental than genetic factors. Some evidence indicates that this may be the case(Allen 1976).
Criticisms such as these tend to lead researchersto discount results
from twin studies as showing evidence for heritability, but there is also
counterevidence. Grove et al. (1990) studied thirty-two sets of monozygotic twins who were separated and reared apart shortly after birth, and
found statistically significant heritabilities for antisocial behavior in both
childhood (0.41) and adulthood (0.28). Such evidence for heritability
cannot be due to being raised in the same environment. Furthermore, one
hasto c(>nsiderthe methodological problems with twin studies, which can
decrease
estimatesof heritability as opposed to artificially increasing them.
For example, there is evidence that some twins make attempts to "deidentify" or be different from orie another (Schacter and Stone 1985),
while other twin pairs develop opposite (for example, dominant-submissive) role relationships (Moilanen 1987). These effects are expected to be
greatcr in MZ pairs, with the result of artificially reducing heritability
estimates.Though MZ twins are genetically identical, identical twinning
can result in biological differences that can accentuatehuman differences.
For example, there is a greater discrepancy in the birthweights ofMZ twins
relative to DZ twins, and birth complications have been linked to differcnces in behavior and cognition. This nongenetic, biological factor will
result in an exaggeration of behavioral differences in MZ twins and a
reduction in heritability estimates.The methodological problems of twin
studies are just as likely to decreaseheritability estimates as opposed to
i'1ffi!ting them; in all probability these effects tend to cancel each other out.

46

Adoption

Adrian

Raine

Studies

Adoption studies also overcome the problem with tWin studies because
they more cleanly separateout genetic and environmental influences. We
can examine offspring who have been separatedfrom their criminal, biological parents early in life and sent out to other families. If these offspring
grow up to become criminal at greater rates than foster children whose biological parents were not criminal, this would indicate a genetic influence
with its origin in the subject's biological parents.
A variation of this type of study is the "cross-fostering" technique that
hasbeen usedextensivelyin experimental genetic studiesof animals. Applied
to humans, the offspring whose biological parents are criminal or noncriminal are raisedby parentswho themselvesare either criminal or noncriminal.
This 2 x 2 design capitalizeson what is effectively a natural experiment, and
allows for a more systematicexploration of genetic and environmental influences.As will be seenlater, this method also allows an assessmentof possible
interactions between genetic and environmental influences.
A good example of a cross-fosteringadoption study is a classic study
conducted by Mednick et al. (1984), illustrated in Table 3.1. These
researchersbased their analyseson 14,427 adoptions that took place in
Denmark between 1927 and 1947. Infants were adopted out immediately in
25.3 percent of cases,50.6 percent within one year, 12.8 percent in the
secondyear,and 11.3 percent after agetwo. Court recordswere obtained on
65,516 biological parents, adoptive parents, and adopteesin order to assess
which subjectshad convictions. When both adoptive and biological parents
were noncriminal (neither genetic nor environmental predispositions present), 13.5 percent of the adopteeshad a criminal record. This increasedto
14.7 percent When adoptive parents only were criminal, meaning that an
environmental but not genetic effect was operating. When only the biological parents were criminal, the conviction rate in the adoptees increasedto
20.0 percent. When both adoptive and biological parents were criminal
(both genetic and environmental predispositions present), the conviction
rate increasedto 24.5 percent.The effect of an adopted child having a criminal biological parent was associatedwith a statistically significant increasein
the likelihood of the adoptee becoming criminal.
While this is but one example, a review of fifteen other adoption studies conducted in Denmark, Sweden,and the United Statesshows that all but
one find a genetic basisto criminal behavior (Raine 1993). Importandy, evidence for this genetic predisposition has been found by severalindependent
researchgroups in severaldifferent counuies. These data, therefore, provide
evidencethat the basic finding is robust. Interestingly, the three studies that

The Biological

Basis

of

47

Crime

Table 3.1
ResuJtsof Cross-Fostering Analyses (Percentagesrefer to
the proportion of adoptees who had court convictions.)

ARE
PARENTS

BIOLOGICAL
CRIMINAL?

SOURCE: Mednick, Gabrielli, and Hutchings (1984).

had a large enough sample size to separateviolent from nonviolent, petty

property crime found that there is heritability for petty property crimes but
not for violent crimes (Bohman et al. 1982; Mednick et al. 1984; Sigvardsson et al. 1982). On the other hand, an adoption study by van den Oord,
Boomsma, and Verhulst (1994) found heritability of 70 percent for aggressivebehavior compared to 39 percent for delinquency. Consequently, while
there is very clear evidence for a genetic basis to adult criminal offending,
there is currently some question as to whether adoption studies of violent
adult offending in particular show a genetic basis.
One of the key themes of this chapter is the notion that the interaction
betWeenbiological and social factors may be particularly important. This
concept is well illustrated in a cross-fostering analysis of petty criminality
(Cloninger et al. 1982), results of which are illustrated in Figure 3.1.
Swedish adoptees (N = 862) were divided into four groups depending on
the presenceor absenceof (1) a congenital predisposition (that is, whether
biological parents were criminal) and (2) a postnatal predisposition (how
the children were raised by their adoptive parents). When both heredity
arid environmental predispositional factors were present, 40 percent of the
adopteeswere criminal compared to 12.1percent with only genetic factors
present,6.7 percent for those with only a bad family environment, and 2.9
percent when both genetic and environmental factors were absent. The
fact that the 40 percent rate for criminality when both biological and environmental factors are present is greater than the 18.8 percent rate given by
a combination of "congenital only" and "postnatal only" conditions indicatesthat genetic and environmental factors are interacting.

48

Adrian

Raine

Further analysesindicated that the occupational status of both biological and adoptive parents were the main postnatal variables involved in this
nonadditive interaction. Cloninger and Gottesman (1987) later analyzed
data for femaleswith larger sample sizes.As would be expected,these crime
rates in female adopteesare much lower than for males, but the sameinteractive pattern is present: Crime rates in adoptees are greatest when both
heritable and environmental influences are present, with this interaction
accounting for twice as much crime as is produced by genetic and environmental influences taken alone.
Evidence for gene x environment interaction is also provided by
Cadoret et al. (1983), who presented data from three adoption studies.
When both genetic and environmental factors are present, they account for
a greater number of antisocial behaviors than either of these two factors
acting alone. Crowe (1974) also found some evidencefor a gene x environment interaction in his analysis of adopted-away offspring of female
prisoners, though this trend was only marginally significant (p < .10).
Cadoret et al. (1995) in an adoption study of 95 male and 102 female
adoptees whose parents had either antisocial personality and/or alcohol
abuse showed that parental antisocial personality predicted increased
aggressionand conduct disorders in the offspring-illustrating evidence for
genetic processes.But in addition, adverseadoptive home environment was
found to interact with adult antisocial personality in predicting increased
aggressionin the offspring, that is, a gene x environment interaction effect.
A related but different concept is that of gene-environment correlations. An interesting example of this is a study by Ge et al. (1996), who
showed that the adopted-awayoffspring of biological parents who had antisocial personality I substanceabusewere more likely to show antisocial and
hostile behaviors in childhood. This helps establish genetic transmission of
childhood antisocial behavior. In addition, an association was found
between antisocial behavior in the biological parent and the parenting
behaviors of the adoptive parents. This can be explained by a transmission
pathway in which the biological parent contributes a genetic predisposition
toward antisocial behavior in the offspring. The antisocial offspring then in
turn elicit negative parenting behaviors in the adoptive parents. This study
provides direct evidence of an "evocative" gene-environment correlation,
and suggeststhat the association between negative parenting in the adoptive parent and antisocial behavior in the child is mediated by genetic
processes.One of the goals of future behavior genetic studies should be to
further examine the interplay between genesand environment in this fashion. More generally,there are likely to be future exciting developmentswith

The

Bi%gica/

Basis

of

Crime

49

Genetic x Environmental
Interaction
in Male Adoptees
(Cloninger et al. 1982)

CRIME
IN

ADOPTEES

Postnatal

None

Both

Congenital

Level of Predisposition

in Female

Genetic x Environmental
Interaction
Adoptees
(Cloninger
and Gottesman,

1982)

CRIME
IN

ADOPTEES

Postnatal

None

Congenital

Both

Level of Predisposition

Figure 3.1
Results of a Cross-Fostering

Analysis

between Genetic and Environmental

Indicating

Evidence for an Interaction

Factors in (a) Males and (b) Females.

SOURCES: Cloninger and Gottesman (1987); Cloninger et al. (1982).

50

Adrian

Raine

respect to identifying the specific geneswhich give rise to the risk factors
that shapecriminal behavior (seesection on neurogenetics).

PSYCHOPHYSIOLOGY
Since the 1940s an extensive body of researchhas been built up on the
psychophysiological basis of antisocial, delinquent, criminal, and psychopathic behavior. For example, there have been at least 150 studies on
electrodermal (sweat.rate) and cardiovascular (heart rate) activity in such
populations, and in electroencephalographic (EEG) research alone there
have been hundreds of studies on delinquency and crime (Gale 1975).
This body of researchhas received little attention in the broader field of
criminology, and is rarely referred to in textbooks on crime. One purpose
of this chapter is to bring this body of knowledge to the attention of this
more general audience.
Definitions of psychophysiology vary, but one useful perspective outlined by Dawson (1990) is that it is "concerned with understanding the
relationships berween psychological states and processeson the one hand
and physiological measureson the other hand" (p. 243). Psychophysiology
is uniquely placed to provide important insights into criminal behavior
becauseit rests at the interface berween clinical science, cognitive science,
and neuroscience (Dawson 1990). Thus, it is sometimes easier to see the
relevance of this research for crime relative to biochemistry research
becauseconcepts in psychophysiology are more easily linked to broader
concepts such as learning, emotion, arousal, and cognition.
There are many psychophysiological correlates of antisocial, criminal,
and psychopathic behavior.2 The focus here will lie with one particular
psychophysiological construct, low arousal, because-as will become
clear-it is the strongest psychophysiological finding in the field of antisocial and criminal behavior.
EEG Underarousal
One influential psychophysiological theory of antisocial behavior is that
antisocial individuals are chronically underaroused.Traditional psychophysiological measuresof arousal include heart rate, skin conductance activity,
and electroencephalogram(EEG) measured during a "resting" state. Low
heart rate and skin conductance activity, and more excessiveslow-waveEEG
(delta activity with a frequency of 1-4 cyclesper second [cps], 4-7 cps theta
activity, and 8-10 cps slow alpha) indicate underarousal, that is, less than

The Biological

Basis

of

Crime

51

averagelevels of physiological arousal. Most studies tend to employ single

measuresof arousal, although studies that employ multiple measuresare in
a stronger position to test an arousaltheory of antisocial behavior.
EEG is recorded from scalp electrodesthat measurethe electrical activity of the brain. Literally hundreds of studies assessingEEG in criminals,
delinquents, psychopaths,and violent offendershavebeendone over the past
sixty years,and it is clear that a large number of them implicate EEG abnormalitiesin violent recidivistic offending. As examples, Bach-y-Rita et al.
(19n)and Hill and Pond (1952) examined large samplesof violent offenders and observedEEG abnormalities in about 50 percent of the cases,with
the most common abnormality being excessiveslow-wave EEG (underarousal).These findings have been supported by studies of murderers and
other violent offenders (for example,Mark and Ervin 1970; Williams 1969).
Fishbein et al. (1989), in a sample of 124 adult male drug abusers,found
that aggressionwas associatedwith increasedslow-wave theta activity and
decreasedalpha, indicating underarousal. Convit, Cwbor, and Volavka
(1991) observedthe samepattern within a sample of psychiatric inpatients;
the number of instancesof violence on wards was related to increasedlevels
of delta activity and lower levels of alpha. Drake, Hietter, and Pakalnis
(1992) found that, while none of 24 depressedpatients and only I of 20
headachecontrol patients had abnormal EEGs, 7 of23 patients with either
intermittent explosive disorder or episodic dyscontrol (brief periods where
the individual lacks control) had diffuse or focal slowing in EEG. Murderers
havemore recently been shown to have more EEG deficits in the right than
the left hemisphereof the brain, with multiple abnormalities being especially
presentin the right temporal cortex (Evansand Park 1997). On the other
hand, Pillmann et al. (1999) showed greater abnormalities in the left temporal region of repeatviolent offenders.
Generally speaking, the prevalence of EEG abnormalities in violent
individuals in this large literature rangesfrom 25 percent to 50 percent, with
the rate of abnormalities in normals estimated as ranging from 5 percent
to 20 percent. The bulk of this researchimplicated the more frontal regions
of 1he brain, areasthat regulate executive functions such as planning and
decision making. Similar conclusions are drawn by Volavka (1987) and
Milstein (1988) for crime in generaland violent crime in particular.

Cardiovascular

Underarousal

Data on resting heart rate provides striking support for underarousal in

antisocials. Indeed, the findings for heart rate level (HRL) on non-

52

Adrian

Raine

institutionalized antisocials are believed to represent the strongest and

best replicated biological correlate of antisocial behavior. A detailed review
of these studies and full theoretical and methodological considerations
are given in Raine (1993) and Raine (1996a). Briefly, twenty-four studies
of resting heart rate and conduct-disordered, delinquent, and antisocial
children and adolescents performed between 1971 and 1996 yielded
twenty-nine independent samplesof antisocials and a total of thirty-eight
effect sizes (a measure of the strength of the effect). Of the thirty-eight
effect sizes,thirty-two were significant and positive (that is, low hean rate
associatedwith antisocial behavior), while only one was significant and in
the negative direction. This latter, unexpected finding (Zahn and Kruesi
1993) was interpreted by its authors as possibly due to biased referral to
the clinic by anxious, easily stressedparents who worried about their child
and who may genetically transmit high HRL to their child.
A low resting heart rate is the best-replicatedbiological marker of antisocial and aggressivebehavior in childhood ;md adolescent community
samples.Resting HRL was measured in a wide variety of ways, including
polygraphs, pulse meters, and stopwatches.A wide number of definitions of
antisocial behavior are used, ranging from legal criminality and delinquency
to teacher ratings of antisocial behavior in school, self-report socialization
measures,diagnostic criteria for conduct disorder, and genetically inferred
law breaking (i.e. offspring of criminals). Subjects were also assessedin a
wide variety of settings, including medical interview, study office, school,
university laboratory, and hospital. In the light of such variability, it is sucprising that consistency in findings have been obtained, attesting to the
robustness of the observed effects. Importantly, there has also been good
cross-laboratoryreplication of the finding, and it has also been found in six
different countries-England, Germany, New Zealand, the United States,
Mauritius, and Canada-illustrating invariance to cultural context.
The link between low heart rate and crime is not the result of such
things asheight, weight, body bulk, physical development, and muscle tone
(Raine, Venables, and Mednick 1997; Raine, Buchsbaum and La Casse
1997; Wadsworth 1976; Farrington 1997); scholasticability and IQ (Raine,
Venables,and Williams 1990; Farrington 1997); excessmotor activity and
inattention (Raine, Venables,and Mednick 1997; Farrington 1997); drug
and alcohol use (Raine, Venables,and Mednick 1997); engagementin physical exerciseand sports (Wadsworth 1976; Farrington 1997); or low social
class,divorce, family size, teenagepregnancy,and other psychosocialadversity (Raine, Venables, and Williams 1990; Wadsworth 1976; Farrington
1997). Intriguingly, an unusual and important feature of the relationship is

The

Biological

Basis

of

Crime

53

its diagnostic specificity. No other psychiatric condition has been linked to

low resting heart rate. Other psychiatric conditions, including alcoholism,
depression, schizophrenia, and anxiety disorder, have, if anything, been
linked to higher (not lower) resting heart rate.
Low heart rate has been found to be an independent predictor of
violence. Out of forty-eight psychosocial and individual measures, only
tWo risk factors were related to violence independently of all other risk
factors in all six analyses:low resting heart rate and poor concentration
(Farrington 1997). Indeed, low heart rate was more strongly related to
both self-report and teacher measuresof violence than having a criminal
parent. These findings led Farrington (1997) to conclude that low heart
rate may be one of the most important explanatory factors for violence (p.
99). There is also a substantial heritability for resting heart rate, suggesting
that it may be a genetic marker for antisocial and criminal behavior (Raine
et al. 1990). Furthermore, the offspring of criminal parents have been
shown to have low resting heart rate in tWo separatestudies (Farrington
1987; Venables 1987). Low heart rate characterizesfemale as well as male
antisocial individuals. Severalstudies, including tWo that are prospective,
have now establishedthat, within females, low heart rate is linked to antisocial behavior (Raines et al. 1990; Maliphant, Hume, and Furnham
1990; Raine, Venables, and Mednick 1997; Moffitt and Caspi In press).
Prospective

Studies

of

Underarousal

One of the major difficulties in trying to draw conclusions on the psychophysiological basisof criminal behavior is that most studies conducted
to date have been nonprospective and have utilized institutionalized populations. In addition, most studies report results from only one of the three
most commonly measured psychophysiological responsesystems (electrodermal, cardiovascular,and cortical). Prospectivelongitudinal research-that
is; researchthat follows people forward through their lives-a1lows for much
more powerful statements to be made about predispositions for criminal
behaviorand to elucidate causeand effect relationships;but becauseprospective researchis more difficult to execute,there are few such studies.
Regarding heart rate levels, five prospective studies of heart rate alone
haveconfirmed that low heart rate is predictive of later antisocial behavior,
while five additional prospective studies also show significant effects for
c)ectrodermal and electrocortical arousal. Wadsworth (1976) found that
tower resting heart rate in unselectedeleven-year-old schoolboys predicted
de)inquency measured from ages eight to twenty-one. The very lowest

54

Adrian

Raine

HRU were found in those who committed nonsexual violent criminal

offensesas adults. Similarly, Farrington (1987) found that resting heart rate
measured at age eighteen to nineteen in noninstitutionalized males predicted to violent criminal offending at age tWenty-five. With respect to
EEG, tWo separatestudies haveshown that slow alpha frequency predicts to
adult thievery in a sample of 129 Danish thirteen-year-old boys (Mednick
et al. 1981) and 571 Swedish one- to fifteen-year-old boys (Petersenet al.
1982). Low heart rate characterizeslife-course-persistentantisocial individuals in particular; Moffitt and Caspi (In press) have recently found that
low resting heart rate assessed
at agesseven,nine, and eleven is particularly
characteristic of life-course-persistent offenders, a group who have been
hypothesized as having early neurobiological deficits (Moffitt 1993).
As with most other studies, evidence in these prospective studies for
prediction to antisocial behavior is based on only one measure of arousal.
One nine-year prospective study of crime by Raine, Venables, and
Williams (1990) has shown, however, that low resting heart rate, low resting skin conductance activity, and excessive slow-wave theta EEG
(indicating underarousal) measured at age fifteen in normal unselected
schoolboys predicted criminal behavior at age tWenty-four. These three
measurescorrectly classified 74.7 percent of all subjects as criminal/noncriminal, a rate significantly greater than chance (50 percent). In the total
population, the three arousal measureswere statistically independent; the
fact that they all independently predicted to criminal behavior indicates
strong support for an arousal theory of criminal and antisocial behavior
(although this finding also cautions against the use of a simplistic, unitary
arousal concept in explaining crime). Group differences in social class,
academic ability, and area of residence were not found to mediate the
link betWeenunderarousal and antisocial behavior.

Interpretations
Fearlessness

of
and

Low Arousal:
Stimulation-Seeking

Theories

Why should low arousal and low heart rate predispose to antisocial and
criminal behavior? There are two main theoretical interpretations.
Fearlessnesstheory indicates that low levels of arousal are markers of low
levels of fear (Raine 1993; Raine 1997). For example, particularly fearless
individuals such as bomb disposal experts who have been decorated for
their bravery have particularly low HRU and reactivity (Cox et al. 1983;
O'Connor, Hallam, and Rachman 1985), as do British paratroopers decorated in the Falklands War (McMillan and Rachman 1987). A fearlessness

The

Bi o logi

cat

B asis

of

Grim

55

interpretation oflow arousal levels assumesthat subjectsare not actually at

"rest," but that instead the rest periods of psychophysiological testing represent a mildly stressful paradigm and that low arousal during this period
indicates lack of anxiety and fear. Lack of fear would predisposeto antisocial and violent behavior becausesuch behavior (for example, fights and
assaults)requires a degreeof fearlessness
to execute,while lack of fear, especially in childhood, would help explain poor socialization since low fear of
punishment would reduce the effectivenessof conditioning. Fearlessness
theory receives support from the fact that autonomic underarousal also
provides the underpinning for a fearlessor uninhibited temperament in
infancy and childhood (Scarpaet al. 1997; Kagan 1994).
A second theory explaining reduced arousal is stimulation-seeking
theory (Eysenck 1964; Quay 1965; Raine 1993; Raine, Reynolds,
Venables, et al. 1998). This theory argues that low arousal represents an
unpleasant physiological state; antisocials seek stimulation in order to
increasetheir arousal levels back to an optimal or normal level. Antisocial
behavior is thus viewed as a form of stimulation-seeking, in that committing a burglary, assault, or robbery could be stimulating for some
individuals. Before leaving this theory, the possibility has to be considered
that fearlessnesstheory and stimulation-seeking theory may be complementary rather than competing theories. That is, low levels of arousal may
predisposeto crime becauseit produces some degree of fearlessness,and
a]soencou!ages antisocial stimulation-seeking. Indeed, behavioral measl}reSof stimulation-seeking and fearlessness,both taken at age three in a
large sample, predict to aggressivebehavior at age eleven (Raine, et al.
1998). The combined effect of these two influences may be more imporrant in explaining antisocial behavior than either influence taken alone.
Psychophysiological
Factors
against

Crime

Protective
Development

Until recently, there had been no researchon biological factors that protect
against crime development, but that is changing. We are discovering that
higher autonomic activity during adolescencemay act as a protective factor
against crime development. Raine, Venables, and Williams (1995, 1996)
rcport on a fourteen-year prospective study in which measuresof arousal,
orienting (sweat rate and heart rate responsesto tone stimuli), and classical
conditioning (learning through association) were taken in 101 unselected
fifteen-year-old males. Of these, seventeen adolescent antisocials who
desistedfrom adult crime (Desistors)were matched on adolescentantisocial

Adrian

56

Raine

behavior and demographic variables with seventeenadolescent antisocials

who had become criminal by age tWenty-nine (Criminals), and seventeen
non-antisocial, noncriminals (Controls). Oesistorshad significantly higher
HRU as well as higher scoreson the other psychophysiological measures
than did the Criminals. Findings suggest that boys who are antisocial
during adolescencebut who do not go on to adult criminal offending may
be protected from such an outcome by their high arousal levels.
Findings from a second study on adults provide some support for this
initial finding in adolescents.Brennan et al. (1997) report on a study of protective factors in fifty men predisposed to crime by virtue of having a
seriously criminal father who had been imprisoned. Of these men, tWentyfour developeda criminal record and were imprisoned themselves,while the
other tWenty-six did not show any criminal offending. Heart rate and skin
conductance measuresof reactivity to fourteen orienting tones were measured at agethirty-five. The group who desistedfrom crime was found to have
significantly higher levels of physiological orienting relative to those who
exhibited criminal behavior, and to a noncriminal control group who had
noncriminal fathers. Desistors seem to have a nervous system particularly
sensitiveto forming associationsbetWeensignalsof punishment and the punishment itself In a similar fashion, higher resting heart ratesin Oesistorsmay
be interpreted as indicating higher levels of fearfulnessin these individuals.
Overall, the initial profile that is being built up on the psychophysiological characteristics of the Oesistor is one of heightened information
processing (better orienting), greater responsivity to environmental stimuli
in general (fast recovery), greater sensitivity to cuespredicting punishment
in particular (better classical conditioning), and higher fearfulness (high
HRU). The importance of researchon psychophysiological protective factors such as these is that they offer suggestions for possible intervention
and prevention strategies.
BRAIN

IMAGING

Advances in brain imaging techniques in the past fifteen years have provided the opportunity to gain dramatic new insights into the brain
mechanisms that may be dysfunctional in violent, psychopathic offenders.
In the past, the idea of peering into the mind of a murderer to gain insights
into his or her acts was the province of pulp fiction or space-agemovies.
Yet now we can literally look at, and into, the brains of murderers using
functional and structural imaging techniques that are currently revolutionizing our understanding of the causesof clinical disorders.

The Biological

Basis

of

Crime

57

Brain imaging studies of violent and psychopathic populations have

been reviewed by Raine (1993), Raine and Buchsbaum (1996), and
Henry and Moffitt (1997). These reviews (that cover studies up to 1994),
while showing variability in findings acrossstudies, concur in indicating
that violent offenders have structural and functional deficits to the frontal
lobe {behind the forehead) and the temporal lobe (near the ears). Between
1994 and 1997, six more studies support this key finding of anterior brain
dysfunction (Goyer et al. 1994; Volkow et al. 1995; Kuruoglu et al. 1996;
Seidenwurm et al. 1997; Intrator et al. 1997; Soderstrom et al. 2000).
Taken together, these later studies show continued support of the notion
that poor functioning of the frontal and temporal regions may predispose
to crime. Out of the six more-recent studies, five showed evidence for
frontal dysfunction and four showed evidence for temporal lobe
dysfunction. Despite some discrepancies, the first generation of brain
imaging studies supports earlier contentions from animal and neurological
studiesimplicating the frontal (and to some extent temporal) brain regions
in the regulation and expressionof aggression.
Prefrontal

i
~
!':
~
f!
r:,
}
f
I;
!\:

""

C;t

Dysfunction

in Murderers

In the first published brain imaging study of murderers (Raine, Buchsbaum, Stanleyet al. 1994), we scannedthe brains of twenty-two murderers
pleading not guilty by reason of insanity (or otherwise found incompetent
t~ ,stand trial) and compared them to the brains of twenty-two normal
controls who were matched with the murderers on sex and age.The technique we used was positron emission tomography (PET), which allowed
usto measurethe metabolic activity of many different regions of the brain
'ihcluding the prefrontal cortex, the frontalmost part of the brain. We had
subjectsperform a task that required them to maintain focusedattention and
bevigilant for a continuous period of time, and it is the prefrontal region of
the brain that in part subservesthis vigilance function.
,The key finding was that the murderers showed significantly poorer
,functioning of the prefrontal conex, that part of the brain lying above the
eyesand behind the forehead. It is thought that poorer functioning of the
,prefrontal cortex predisposespeople to violence for a number of reasons.
,Reducedprefrontal functioning can result in a loss of the ability of this
part of the brain to control deeper and more primitive subconical structures, such as the amygdala, which are thought to give rise to aggressive
feelings. Prefrontal damage also encourages risk-taking, irresponsibility,
rule breaking, emotional and aggressive outbursts, and argumentative

Adrian

58

Raine

behavior that can also predispose to violent criminal acts. Loss of selfcontrol, immaturity, lack of tact, inability to modify and inhibit behavior
appropriately, and poor social judgment could predispose to violence as
well. This loss of intellectual flexibility and problem-solving skills, and
reduced ability to use information provided by verbal cues can impair
social skills essential for formulating nonaggressivesolutions to fractious
encounters. Poor reasoning ability and divergent thinking that results from
prefrontal damage can lead to school failure, unemployment, and economic deprivation, thereby predisposing to a criminal and violent way of
life. A further study showed that it was especially the impulsive, emotionally undercontrolled murdererswho were especiallylikely to show prefrontal
deficits (Raine, Meloy et al. 1998). Nevertheless, it should be recognized
that, while there is an association betWeenpoor prefrontal function and
impulsive violence, this brain dysfunction may be essentially a predisposition only, requiring other environmental, psychological, and social factors
to enhance or diminish this tendency.
Corpus
Callosum,
and the Subcortex

Left

Angular

Gyrus,

What other brain deficits, apart from prefrontal dysfunction, characterize

murderers?We took this imaging researcha step further by expanding our
sample from tWenty-tWoto forty-one murderers, and also by increasing the
size of our control group to forty-one. This increasein sample size gave us
more statistical power to detect group differences, and in 1997 we reported
our updated findings (Raine, Buchsbaum, and La Casse1997). The results
were interesting for a variety of reasons. First, we confirmed that there
was a significant reduction in the activity of the prefrontal region in
murderers.
Second, we now found in this larger sample that the left angular gyrus
was functioning more poorly in the murderers. The angular gyrus lies
at the junction of the temporal (side of head), parietal (top and back of
head), and occipital (very back of head) regions of the brain and plays a
key role in integrating information from these three lobes. Reductions in
activity of the left angular gyrus have been correlated with reduced verbal
ability, while damage to this region has been linked to deficits in reading
and arithmetic. Such cognitive deficits could predispose to educational
and occupational failure which in turn predisposesto crime and violence.
The fact that learning deficits have been found to be common in violent
offenders lends further support to this interpretation.

j
~
i~
J

The

Biological

Basis

of

Crime

Third, we found reductions in the functioning of the corpus callosum,

the band of white nerve fibers that provide lines of communication betWeen
the left and right hemispheres.Although we can only speculateat the present time, we think that poor connection betWeen the hemispheres may
mean that the right hemisphere,which is involved in the generation of negative emotion {Davidson and Fox 1989), may experiencelessregulation and
control by the inhibitory processesof the more dominant left hemisphere,
a factor that may contribute to the expression of violence. Furthermore,
researchershave commented on the inappropriate nature of emotional
expressionand the inability to grasp long-term implications of a situation
in split-brain patients who have had their corpus callosum surgically severed. This implies that the inappropriate emotional expression of violent
offenders and their lack of long-term planning may be partly accounted
for by poor functioning of the corpus callosum. Nevertheless, callosal
dysfunction by itself is unlikely to causeaggression.Instead, it may only
contribute to violence in those who also have other brain abnormalities.
Effect
of the Home
Environment
on Brain-Violence
Relationships
How do psychosocialdeficits moderate the relationship betWeenprefrontal
dysfunction and violence? We addressedthis question by dividing our
sample of murderers into those who came from relatively good home backgrounds and those who came from relatively bad ones (Raine, Stoddard, et
al. 1998). In this stUdy, ratings of psychosocial deprivation took into
account early physical and sexual abuse, neglect, extreme poverty, foster
home placement, a criminal parent, severefamily conflict, and a broken
home. The results of the stUdy showed that, while the deprived murderer
showsrelatively good prefrontal functioning, it is the nondeprived murderer
who showsthe characteristiclack of prefrontal functioning. In particular, we
found that murderers from good homes had a 14.2 percent reduction in the
functioning of the right orbitofrontal cortex, a brain area that is of particularinterest. Damage to this brain area results in personality and emotional
deficits that paralld criminal psychopathic behavior, or what Damasio and
colleagueshave termed "acquired sociopathy" (Damasio 1994).
These findings are at one level counterintuitive, but from another
perspective they make some sense.If a seriously violent offender comes
from a bad home environment, then it seemslikely that the causeof the
offender's violence is due to that bad environment. But if they come from
a good home background, then environmental causation seemsless likely

60

Adrian

Raine

and instead biological deficits may be a better explanation. Consistent with

these brain imaging findings, previous researchhas shown that poor fear
conditioning in schoolchildren is related to antisocial behavior especially
in those from a good home environment (Raine and Venables 1981). That
is, the biological deficit (poor conditioning) is found in those who lack a
social predisposition to antisocial behavior. Perhapsnot surprisingly, it is
the right orbitofrontal cortex (situated above the eye orbits) that has been
found to play an important role in the development of fear conditioning.
Reduced
Prefrontal
Gray
in Antisocial
Personality

Matter
Disorder

Ranging from single casestudies (Damasio et al. 1994) to seriesof neurological patients (Damasio, Tranel, and Damasio 1990; Stuss and Benson
1986), those Who have suffered demonstrable damage to both gray and
white matter within the prefrontal region of the brain proceed to acquire an
antisocial, psychopathic-Iike personality. These patients alSo ShoW auronomic arousaland attention deficits to socially meaningful events (Damasio
1994, Damasio, Tranel, and Damasio 1990), a finding consistent with the
role played by the prefrontal cortex in modulating emotion, arousal, and
attention (Srussand Benson 1986; Davidson 1993; Raine, Reynolds, and
Sheard 1991).
We recently conducted a structural magnetic resonance imaging
(MRI) study on volunteers from the community with Antisocial Personality Disorder and made volumetric assessments
of prefrontal gray and white
matter (Raine et al. 2000). Skin conductance and heart rate activity during
a social stressor was also assessedin addition to psyChOSOCial
and demographic risk factors for violence. Subjects were drawn from temporary
employment agencies in Los Angeles and consisted of rwenty-one males
with Antisocial Personality Disorder, a normal control group of thirry-four
males, and a psychiatric control group of rwenty-seven males with substance dependence. Antisocials had significantly lower prefr9ntal gray
volumes than both Controls and SubstanceDependents. In contrast, groups
did not differ on white prefrontal volume, indicating specificity of the deficit
to gray matter (neurons). Furthermore, Antisocials also showed reduced
autonomic reactivity during the social stressorcompared to both Controls
and SubstanceDependents.
When prefrontal gray matter was expressedas a function of whole
brain volume, groups were again found to differ significantly, so the results
are not due to general difference in brain size but a specific difference in

The

Biological

Basis

of

Crime

61

prefrontal size. Further analysesindicated that the three prefrontal and

autonomic variables (prefrontal gray/whole brain, heart rate, skin conductance) predicted group membership with an accuracy of 76.9 percent.
These prefrontal and autonomic deficits were independent of psychosocial
deficits in the Antisocial group. After ten demographic and psychosocial
risk factors for antisocial personality were statistically controlled for, the
prefrontal and autonomic deficits added substantially to the prediction
of Antisocial vs. Control group membership. As such, the brain deficits
cannot be easily accounted for by psychosocialfactors, and instead appear
to reflect a different risk process.When both biological and social measureswere used together, they correctly classifiedto 88.5 percent, indicating
the importance of a biosocial perspective that integrates biological and
social factors.
Brain
with

Deficits
a History

in Violent
Offenders
of Childhood
Abuse

While the relationship between physical child abuse and violence is well
established (Lewis et al. 1988; Tarter et al. 1984; Widom 1997), there
appearsto be little or no research,biological or social, on factors that differentiate abused victims who go on to perpetrate violence from those who
refrain from adult violence. We recently conducted a study (Raine, Park, et
al. In press) in which we used functional magnetic resonance imaging
(tMRI) to addresstwo important gapsin our knowledge of brain functioning and violence: (1) What are the brain correlates of adults in the
community who have suffered severephysical abuseearly in life and who go
on to perpetrate seriousviolence in adulthood? (2) What characterizesthose
who experienceseverephysical abusebut who refrain from seriousviolence?
We recruited four groups of participants from the community: (1)
nonviolent controls who had not suffered abuse; (2) participants with
severephysical child abuse only (that is, had suffered severephysical or
sexual abuse in the first eleven years, but were not violent); (3) ones with
serious violence only (violence that causedeither bodily injury or trauma,
or that were life-threatening acts); and (4) severelyabused, seriously violent offenders. All underwent tMRI while performing a visual/verbal
working memory task. Functional magnetic resonance imaging measures
blood flow within brain tissue, and thus allows one to assessthe functional
properties of the brain. In this sense,it is like PET, but unlike PET there
is no exposure to radioactivity. Furthermore, it detects activity in brain
regions that are as small as 1 millimeter. The task involved subjects hold-

62

Adrian

Raine

ing in short-term memory pictures shown very briefly and pressing a

responsebutton any time a picture was repeated, a task known to activate
the temporal and frontal regions of the brain.
Violent offenders who had suffered severechild abuse show reduced
right hemisphere functioning, particularly in the right temporal cortex.
Abused individuals who refrain from serious violence showed relatively
lower left, but higher right, activation of the temporal lobe. Abused individuals, irrespective of violence status, showed reduced cortical activation
during the working memory task, especially in the left hemisphere. These
findings constitute the first fMRI study of brain dysfunction in violent
offenders and indicate that initial right hemisphere dysfunction, when
combined with the effects of severeearly physical abuse,may predispose to
serious violence. They also suggestthat relatively good right hemisphere
functioning protects against violence in physically abused children.
OTHER

BIOLOGICAL

BIRTH

COMPLICATIONS,

PHYSICAL
AND

Birth

ANOMALIES,

PROCESSES:
MINOR
NUTRITION,

NEUROCHEMISTRY

Complications

Severalstudies have shown that babies who suffer birth complications are
more likely to develop conduct disorder, delinquency, and impulsive crime
and violence in adulthood (seeRaine 1993 for a detailed review). Birth complications such as anoxia (getting too little oxygen), forceps delivery, and
preeclampsia(hypenension leading to anoxia) are thought to contribute to
brain damage, and this damage in turn may predispose to antisocial and
criminal behavior. On the other hand, birth complications may not by
themselvespredispose to crime, but may require the presenceof negative
environmental circumstance to trigger later adult crime and violence.
One example of this "biosocial interaction" is a study of birth compli"
cations and maternal rejection in all 4,269 live male births that took place
in one hospital in Copenhagen, Denmark (Raine, Brennan, and Mednick
1994). Birth complications were assessedby obstetricians and midwives.
When the baby was one year old, the mother was interviewed by a social
worker and the degree of maternal rejection was assessedby three measures: mother did not want the pregnancy, mother made an attempt to
abort the fetus, and the baby was institutionalized for at least four months
in the first year of life. Babies were then followed for eighteen years, when

The

Bi / ogica/

B as is

of

Grim

63

their arrests for violent crimes were assessed.

A highly significant interaction was found betWeen birth complications and maternal rejection.
Babieswho only suffered birth complications or who only suffered maternal rejection were no more likely than normal controls to become violent
in adulthood. On the other hand, those who had both risk factors were
much more likely to become violent. Only 4 percent of the sample had
both birth complications and maternal rejection, but this small group
accounted for 18 percent of all the violent crimes committed by the entire
sample. This finding from Denmark was replicated by Piquero and
Tibbetts (1999) in a prospective longitudinal study of 867 males and
femalesfrom the Philadelphia Collaborative Perinatal Project; those with
both prenatal/perinatal disturbances and a disadvantagedfamilial environment were much more likely to become adult violent offenders.
This sample was studied again when they were age thirty-four (Raine,
Brennan, and Mednick 1997) .The results indicate that the biosocial
interaction previously observed holds for violent but not nonviolent criminal offending. Furthermore, the interaction was found to be specific to
more serious forms of violence and not threats of violence. The interaction
held for early onset but not late onset violence and was not accounted for
by psychiatric illness in the mothers. Rearing in a public care institution in
the first year of life and an attempt to abort the fetus were the key aspects
of maternal rejection found to interact with birth complications in predisposing to violence.

Minor

Physical

Anomalies

Minor physical anomalies (MPAs) have been associatedwith disorders of

pregnancy and are thought to reflect maldevelopment of the fetus (including brain maldevelopment) toward the end of the first three months of
pregnancy. MPAs are relatively minor physical abnormalities consisting
of such features as low seated ears, adherent ear lobes, furrowed tongue,
curved fifth finger, single transversepalmar crease,gaps betWeenthe first
and second toes, unusually long third toes, and fine hair that doesn't easily
comb down. They are not stigmatizing as they are not obvious unless a
careful physical examination is done. While MPAs may have a genetic
basis, they may also be caused by environmental factors of the fetus such
asanoxia, bleeding, and infection (Guy et al. 1983).
Minor physical anomalies have also been found to characterize preadult antisocial behavior and temperament. Paulus and Martin (1986)
found more MPAs in aggressive and impulsive preschool boys, while
~,
f
..
t

64

Adrian

Raine

Halverson and Victor (1976) also found higher levels ofMPAs in elementary male schoolchildren with problem school behavior. MPAs have even
been linked to peer aggressionas earlyas age three (Waldrop et al. 1978).
Although MPAs have generally characterized behavior disorders in children drawn from the normal population (see Pomeroy, Sprafkin, and
Gadow 1988 for a review), at least one study failed to observe a link
betWeenMPAs and conduct disorder within a mixed group of emotionally
disturbed children (Pomeroy, Sprafkin, and Gadow 1988).
Mednick and Kandel (1988) assessed
MPAs in a sample of 129 tWelveyear-old boys examined by an experiencedpediatrician. MPAs were found
to be related to violent offending as assessednine years later when the
subjects were age tWenty-one, though not to property offenses without
violence. However, when subjects were divided into those from unstable,
nonintact homes versusthose from stablehomes, a biosocial interaction was
observed. MPAs only predicted violence in those individuals raised in
unstable home environments. A similar interactive relationship was also
observedfor birth weight and family stability (Kandel and Mednick, 1991) .
These findings are quite similar to those on birth complications reported
above; in both studies the presence of a negative psychosocial factor is
required to "trigger" the biological risk factor, and in both casesthe effects
are specific to violent offending. In a stUdy confirming specificity to violence, Arseneault et al. (2000) found that MPAs predicted to violent
delinquency during adolescencein 170 teenagers,but not to nonviolent
delinquency. Furthermore, effects were independent of family adversity.

Nutrition
Although deficiency in nutrition itself has been rarely stUdied in relation
to childhood aggression,several stUdieshave demonstrated.the effects of
related processesincluding food additives, hypoglycemia, and more recently
cholesterol on human behavior (Rutter, Giller, and Hagell 1998; Raine
1993; Fishbein and Pease1994). In addition, somestudies have shown associations between overaggressive
behavior and vitamin and mineral deficiency
(Breakey 1997; Werbach 1995). Furthermore, one stUdy (Rosen 1996)
claimed that nearly a third of a population of juvenile delinquents (mostly
males) showed evidence of iron deficiency. Nevertheless, these findings
remain both conflicting and controversial (Rutter, Giller, and Hagell1998).
One intriguing study illustrates the potentially causal role of malnutrition as early as pregnancy in predisposing to antisocial behavior. Toward
the end ofWorld War II when Germany was withdrawing from Holland,

The

Biological

Basis

of

Crime

65

they placed a food blockade on the country that led to major food shortagesand near starvation in the cities and towns for severalmonths. Women
who were pregnant at this time were exposedto severemalnutrition at different stages of pregnancy. The male offspring of these women were
followed up into adulthood to ascertain rates of Antisocial Personality
Disorder and were compared to controls who were not exposedto malnutrition. Pregnant women starved during the blockade had 2.5 times the
rates of Antisocial Personality Disorder in their adult offspring compared
to controls (Neugebauer,Hoek, and Susser1999).
Initial evidencealso shows relationships between both protein and zinc
deficiency and aggressionin animals (Tikal, Benesova,and Frankova 1976;
Halas, Reynolds, and Sanstead,1977). Recent studies of humans support
theseanimal findings. Protein and zinc deficiency may lead to aggressionby
negatively impacting brain functioning. There is extensiveexperimental evidence in animals that the offspring of rats fed a diet containing marginal
levels of either zinc or protein throughout pregnancy and lactation showed
impaired brain development (Oteiza et al. 1990; Bennis-Taleb et al. 1999).
In humans, zinc deficiency in pregnancyhas been linked to impaired DNA,
RNA, and protein synthesisduring brain development, and congenital brain
abnormalities (Pfeiffer and Braverman 1982). Similarly, protein provides
essentialamino acids for the rapid growth of fetal tissue.PET studiesof violent offenders have revealed deficits to the prefrontal cortex and corpus
callosum (Raine, Buchsbaum, and La Casse1997; Volkow et al. 1995), and
it is of interest that the offspring of rats fed a low-protein diet during pregnancy show a specific impairment to the corpus callosum (Wainwright and
Stefanescu 1983) and reduction in DNA concentration in the forebrain
(Bennis-Taleb et al. 1999). The amygdala,which also shows abnormal functioning in PET imaging of violent offenders (Raine, Buchsbaum, and La
Casse 1997; Raine, Meloy, et al. 1998; Davidson, Pumam, and Larson
2000), is densely innervated by zinc-containing neurons (Christensen and
Frederickson 1998), and males with a history of assaultivebehavior were
found to have lower zinc relative to copper ratios in their blood compared to
nonassaultivecontrols (Walsh et al. 1997). Consequently, protein and zinc
deficiency may contribute to the brain impairments shown in violent
offenders which in turn are thought to predisposeto violence.

Environmental

Pollutants

and

Neurotoxicity

It haS long been suspected that exposure to pollutants, particularly heavy

metals that have neurotoxic effects, can lead to mild degrees of brain

66

Adrian

Raine

impairment which in turn predisposesto antisocial and aggressivebehavior. One of the best studies to date is that of Needleman et al. ( 1996) who
assessedlead levels in the bones of 301 eleven-year-old schoolboys. Boys
with higher lead levels were found to have significantly higher teacher
ratings of delinquent and aggressivebehavior, higher parent ratings of
delinquent and aggressivebehavior, and higher self-report delinquency
scores.These findings do not occur in isolation: Similar links between lead
levels and antisocial, delinquent behavior and aggressionhave been found
in at least six other studies in severaldifferent countries (seeNeedleman et
al. 1996 for a review). Furthermore, experimental exposure to lead during
development increases aggressivebehavior in hamsters (Delville 1999),
thus suggestinga causal link.
Less strong to date, but neverthelessprovocative, are findings with
respect to manganese.At high levels, manganesehas toxic effects on the
brain and can damage the brain so much that it can even lead to
Parkinson-like symptoms. Furthermore, it reduceslevels of serotonin and
dopamine, neurotransmitters that playa key role in brain communication
(see later "Neurotransmitters" section). One study by Gottschalk et al.
(1991) found that three different samplesof violent criminals compared to
controls had higher levels of manganesein their hair. On the other hand,
not all studies have found this association (Schauss1981).It may be that
low levels of calcium intake interact with high manganesein predisposing
to violent behavior becauseanimal researchindicates that the neurotoxic
effect of manganeseis particularly strong when the animal is deficient in
calcium.
Hormones
Testosterone. Excellent reviews and discussions of the potential role
played by testosterone in both animals and man can be found in Olweus
(1987), Brain (1990), Archer (1991), and Susman and Ponirakis (1997).
Animal researchsuggeststhat the steroid hormone testosterone plays an
important role in the genesisand maintenance of some forms of aggressive
behavior in rodents, and early exposure to testosterone has been found to
increaseaggressionin a wide range of animal species(Brain 1990).
A key question generated by this literature is whether testosterone is
involved in aggressionand violent crime in man. Studies correlating questionnaire measures of aggression in normals to testosterone levels have
generally produced weak or nonsignificant findings (Rubin 1987). Studies
of violent incarcerated inmates on the other hand have been more consis-

The

Biological

Basis

of

Crime

67

tent in producing significant effects of moderate to large strength (Rubin

1987). This theme is reiterated by Archer (1991) in a rigorous and critical
review of the literature, arguing effects are small or negligible when aggression is measured using personality inventories, but strong and significant
when groups high and low on behavioral measuresof aggressionare compared. Five studies of prisoners reviewed by Archer resulted in substantial
effects. Furthermore, female prisoners also show high testosterone levels,
and interestingly this effect was found to be specific to femaleswho committed unprovoked assault but not those who reacted violently when
physicallyassaulted (Dabbs et al. 1988).
The critical question in this literature concerns whether testosteroneviolence relationships are causal. Little doubt exists that castration
decreases aggression in animals and administration of testosterone
increasesaggression. Few experimental studies have been conducted in
humans, but there is neverthelessevidenceof a causalrelationship. Olweus
et al. (1988) assessedtheir finding of higher testosterone in male adolescents with high levels of self-report aggression using path analysis and
concluded that testosterone had causal effects on both provoked and
unprovoked aggressivebehavior. One study that comes close to such an
ideal experiment is that of Wille and Beier (1989), who showed that
ninety-nine castrated German sex offenders had a significantly lower
recidivism rate eleven years postrelease(3 percent) compared to thirtyfive non castrated sex offenders (46 percent). There is in addition some
limited evidence that less drastic methods of reducing testosterone levels
such as administration of anti-androgens and progesterone derivatives
have some effect in lowering violence and sexual aggression(Rubin 1987;
Brain 1990; Archer 1991). One double-blind, crossoverhormone replacement study administered testosterone to male adolescentswith delayed
puberty and found that medium doses increased aggressive behavior
twenty-one months later (Susman and Ponirakis, 1997). It could be argued
that extreme alterations in testosterone are not a good model for the less
severe variability found in the general population, and that moderate
changesof this hormone do not significantly influence aggression.On the
other hand, Loosen, Purdon, and Pavlou (1994) found that mild reductions in testosterone in men were associatedwith reductions in outwardly
directed anger, thus suggestingthat mild changesin testosteronecan modulate aggression.
Testosterone levels are in part heritable (Turner et al. 1986), and it is
conceivable that the genetic predisposition to crime may in part be
expressedthrough the hormonal system. On the other hand, it is also

68

Adrian

Raine

known that environmental influences such as successin competition, the

perception of winning, exposure to erotic films, and social dominance
can increase circulating levels of testosterone (Brain 1990; Archer 1991).
Clearly, links between testosterone and aggressionare complex, and simplistic explanations of this link are probably incorrect. By the same token,
it would be equally erroneous to discount the evidence for the role of hormones in influencing aggressionmerely becausehormones are influenced
by the environment. In this context, theoretical perspectivesshould take
into account reciprocal influences between behavior and aggression and
the roles hormones play in the regulation of arousal (Susman et al. 1996).
Cortisol. Cortisol (also called a glucocorticoid) is a hormone produced
by a corticotropin-releasing factor (CRF), a peptide, which is in turn
produced by a brain area called the hypothalamus, which regulates autonomic functions (like heart rate and skin conductance) and emotional
responses.Individuals who are aroused or stressedshowan increasein cortisol. As such, one might expect reduced cortisol levels in antisocials who,
as we saw earlier, are thought to be relatively underaroused and fearless.
A number of studies in a wide variety of contexts have now shown that
there is indeed a link betWeenlow resting cortisol and antisocial, aggressive
behavior. McBurnett et al. (1991) found that boys with conduct disorder
without comorbid anxiety disorder showed significantly lower concentrations of cortisol than did boys with CD and comorbid anxiety disorder.
Cortisol has similarly been reported to be low in habitually violent incarcerated offenders (Virnkunnen 1985), in aggressive schoolchildren
(Tennes and Kreye 1985), in adolescentswith conduct problems (Susman
and Petersen 1992), boys with disruptive behavior (McBurnett et al.
2000), disinhibited children, and boys with oppositional defiant disorder
(van Goozen et al. 1998).
Taken together, these studies suggestthat resting cortisol may playa
nontrivial role in mediating antisocial, violent, and criminal behavior. In
this context it supports and extends the findings described earlier for resting heart rate, confirming the finding of autonomic arousal deficits in
antisocial populations. On the other hand, the link betWeen antisocial
behavior and cortisol reactivity to a stressor(that is, the change in cortisol
before and after a stressful event) is less clear, with some studies finding
decreasedreactivity in antisocial children (for example, van Goozen et al.
1998) while other studies find increasedreactivity (for example, Susman et
al. 1997). Even accepting that there are differences in antisocial children
in basal cortisol levels, it must be remembered that, as with many other

The

Biological

Basis

of

Crime

69

biological measures,cortisol is part of a dynamic system that is responsive

to environmental changes and demands (Susman, Dorn, and Chrousos
1991; Susman and Ponirakis 1997). As such, integration with social and
environmental factors must be a primary aim of future studies in this area.

Neurotra

nsmitters

Basic neurotransmitters such as dopamine, norepinephrine, and serotonin

form the basis to information processing and communication within the
brain, and in this senseunderlie all types of behavior, including sensation,
perception, learning and memory, eating, drinking and, more controversially, antisocial behavior. Neurotransmitters are chemicals stored in the
synaptic vesicles(small globules) of a communicating cell's axon; this axon
carries the nerve impulse of the cell body to other cells. During cellular
communication, these chemicals are discharged into the synaptic cleft
(the spacebetween two neurons) and are taken up by special receptors in
the postsynaptic membrane of the recipient cell, initiating what is termed
a postsynaptic potential. Thus, neurotransmitters form the basis to the
transmission of information throughout the brain.
Paradoxically, the well-known and best-studied neurotransmitters
(serotonin, dopamine, norepinephrine) account for only a small proportion of cell firing in the brain. For example, serotonin is thought to be
the transmitter substance in fewer than 0.1 percent of brain synapses.
Nevertheless, they are thought to be highly imponant in the context
of brain and behavior; for example, serotonin, norepinephrine, and
dopamine have been implicated in the etiology of major disorders such as
schizophrenia and depression.
A meta-analysis of twenty-nine studies examined the relationship
between norepinephrine, dopamine, and serotonin, and antisocial behavior (Raine 1993). Results indicated a relatively large effect size (-0.75)
between reduced central serotonin and antisocial behavior, and a medium
effect size of -0.41 between reduced norepinephrine and antisocial behavior, with no effect for dopamine. Subanalyseswere conducted in an
attempt to specify to which subgroups of antisocials these findings pertain.
These analysesindicated that serotonin was lowest in antisocials with a history of alcohol abuse,borderline personality disorder, and violence, while
cerebrospinal fluid norepinephrine was lowest in those with alcohol abuse,
borderline personality disorder, and depression.
Increased impulsivity has been associatedwith lower serotonin, and
there is some evidence that aggressiveindividuals who are impulsive have

70

Adrian

Raine

particularly lower serotonin. Humans low in social class have also been
found to have reduced serotonin, and it is possible that reductions in serotonin produced by a fall in dominance triggers impulsive aggressionas a
way of raising the individual in the dominance hierarchy. In evolutionary
terms this would be adaptive as increaseddominance givesgreater accessto
food and sex, thus increasing the individual's ability to reproduce and pass
on their genes.Yet again, poor diet may playa role. Diets low in, or otherwise blocking, the uptake of tryptophan or tyrosine (the precursors of
serotonin and norepinephrine respectively) have been found to lower the
levels of these transmitters in the brain (seeWeisman 1986) .In addition,
even when returned to a normal diet, brain serotonin levels are never fully
compensated (Timiras, Hudson, and SegalI1984). Poor nutrition possibly
occurring in individuals of lower socioeconomic status (including poor
dietary care during pregnancy) could conceivably influence neurotransmitter levels throughout life.
Neurogenetics
As outlined earlier, genetic studies are now beginning to progress from
simply demonstrating that there is indeed a genetic basis to crime. A new
generation of neurogenetic studies is now beginning to identify specific
genesthat give rise to abnormally aggressivebehavior, often via abnormal
functioning of specific neurotransmitters.
Some of the first clues have come from animal researchwhere "knockout" mice have been genetically engineered to lack individual genes that
normally give rise to specific neurotransmitters. One of the neurotransmitter systemsimplicated to date is nitric oxide. Mice that lack the gene
essential for nitric oxide in the brain have been found to become highly
aggressive,and furthermore agents which inhibit nitric oxide also lead to
very aggressivebehavior in normal mice (Demas et al. 1997).
Particularly interesting are findings, on the knockout of the gene that
codes for monoamine oxidase-A (MAOA), an enzyme that metabolizes
brain neurotransmitters including serotonin. Adult mice who are deficient
in MAOA are abnormally aggressiveto other males and are much rougher
and aggressivein their mating behavior with females (Caseset al. 1995;
Nelson et al. 1995). These mice also showed very high levels of serotonin.
What makesanimal MAOA findings such as these more provocative is that
they have also been found in humans. Brunner et al. (1993) have shown
that a family in the Netherlands with a history of highly aggressivebehav-'
ior have a mutation in the MAOA gene.They also showed impaired IQs,
and like the MAOA mice, had very high levels of serotonin.

The

Biological

Basis

of

Crime

71

Animal findings such as these which are also found in humans are
particularly interesting. Nevertheless, such genetic mutations affecting
the MAOA gene are rare in humans and therefore it will be difficult to
test whether Brunner's findings will be replicated in other samples.
Furthermore, deletion of this produces high, not low, serotonin levels, a
finding not consistent with the finding of lower serotonin in the cerebrospinal fluid of impulsive violent offenders. Nevertheless, it is not
unlikely that the field of neurogenetics will produce multiple breakthroughs in the next few years on the molecular genetic basis of crime and
violence.
POLICY

IMPLICATIONS

One of the biggest and widely held myths in criminology researchis that
biology is destiny. Instead, the reality is that the biological basesof crime
and violence are amenableto change through benign interventions. In the
past fifty years, intervention programs have not been assuccessfulin reducing crime and violence as had been hoped, and it is possible that part of
their failure has been due to the fact that they have systematically ignored
the biological component of the biosocial equation.
Brain damage and poor brain functioning have been shown to predispose to violence, and one possible source of this brain damage could be
birth complications (Raine, Brennan, and Mednick 1997). The implication is that providing better pre- and postbirth health care to poor mothers
may help reduce birth complications and thus reduce violence.
Alternatively, rather than attempting to reduce birth complications, interventions could focus on the psychosocialhalf of the biosocial equation and
attempt to reduce early maternal rejection which intensifies the effect of
birth complications (Raine, Brennan, and Mednick 1994; Raine, Brennan,
and Mednick 1997). Consideration might be given to multiple efforts
acrosstime to reduce maternal rejection by, for example, making parenting skills classescompulsory in high school to the next generation of
mothers with unwanted pregnancies;providing more pre-birth visits from
nurses to monitor both the pregnancy and the parent's attitude to the
unborn baby; and providing home visits from pediatricians specifically to
mothers who suffered birth complications to monitor the mother-infant
bonding process, assessthe physical and cognitive development of the
infant, and provide appropriate remediation of cognitive and physical
deficits which are known to follow from perinatal complications (Liu and
Raine 1999; Raine and Liu 1998).

72

Adrian

Raine

Another source of brain damage could be poor nutrition; and as has

been seenearlier, there is evidence for a link betWeenpoor nutrition during
pregnancy and later crime. Furthermore, cigarette and alcohol usage
during pregnancy have been linked to later antisocial behavior (for example' Brennan, Grekin, and Mednick [1999]). In this context, Olds et al.
(1998) in a methodologically strong randomized controlled trial showed
that improving the quality of pregnant mothers' prenatal diet (among
other factors) reduced offspring criminal and antisocial behavior fifteen
years later. The intervention was also shown to improve the quality of the
mothers' prenatal diet. Similarly, Lally, Mangione, and Honig (1988)
showed that advice to pregnant women on good nutrition, health, and
child rearing leads to a reduction in juvenile delinquency at age fifteen.
These studies provide more support to the notion that nutrition plays a
causal role in the development of childhood aggression,but future prevention trials that focus explicitly on the specific role of nutrition are
required to further support the specific role of malnutrition.
It has been shown that low physiological arousal is the best-replicated
biological correlate of antisocial behavior in child and adolescent samples.
An important question from a prevention perspectiveconcernswhether low
arousal is amenable to change using noninvasive procedures. Recent findings from Mauritius suggestthat it is. A nutritional, physical exercise,and
educational enrichment from ages three to five resulted in increased psychophysiological arousaland orienting at ageelevencompared to a matched
control group (Raine et al. In press).Furthermore, a longer-term follow-up
of these children to age seventeenshowed a reduction in conduct disorder
(p < .01) and motor excessscores(p < .03) in the experimental group compared to the control group (Raine et al. 1999). It should be noted, however,
that these latter behavioral effects are sleepereffects; no effect of the intervention was noted on age-elevenaggressionscores.Some initial reports have
shown the possible efficacy of using biofeedback to increasephysiological
arousal in hyperactive children (Lubar 1989), while more recent pilot work
has indicated that this technique shows some short-term behavioral
improvement in children with conduct problems. Biofeedback training, as
part of a larger multimodal treatment package, could conceivably help to
reduce antisocial and violent behavior in adolescents.
The policy implications of biological researchon crime also extend to
the criminal justice system. One question raised by these and other studies is whether any of us have freedom of will in the strict senseof the term.
Ifbrain deficits make it more likely that a person will commit violence, and
if the causeof the brain deficits was not under the control of the individ-

The

Biological

Basis

of

Crime

73

ual, then the question becomeswhether or not that person should be held
fully responsible for the crimes. Of coursewe have to protect society, and
unless we can treat this brain dysfunction, we may need to keep violent
offenders in secureconditions for the rest of their lives; but do they deserve
to be executed given the early constraints on their free will? It could be
argued that if an individual possessesrisk factors that make him disproportionately more likely to commit violence, then he has to take responsibility for these predispositions. Just like an alcoholic who knows he suffers
from the diseaseof alcoholism, the person at risk for violence needs to recognize his risk factors and take preventive steps to ensure that he does not
harm others. These persons have risk factors, but they still have responsibility and they have &ee will.
This makes good, practical sense,but responsibility and self-reflection
are not disembodied, ethereal processesbut are rooted firmly in the brain.
Patients who have damage to the ventromedial (lower) prefrontal cortex
are known to become irresponsible, lack self-discipline, and fail to reflect
on the consequencesof their actions. The ability to take responsibility for
one'sactions is damagedin violent offenders. It can be argued that they are
no longer able to reflect on their behavior as others do and take responsibility for their predispositions. It is not just that the brain mechanisms
subserving responsibility and internal soul-searching are damaged in the
violent offender and prevent him from taking action to rectify the causes
of his violence.
Brain scan information is increasingly being used in capital cases,most
frequently in the punishment stageof a trial as a mitigating circumstance
against the death penalty, but it has also been used successfullyin the guilt
phase of a trial (D'Agincourt 1993). Nevertheless, use of brain imaging
data in law courts is hotly debated, with some arguing against their use. Of
course, brain scansare not diagnostic in that they do not perfectly predict
who is violent and who is not. On the other hand, brain imaging data such
as PET and fMRI constitute more direct indexes of brain function than
EEG and neuropsychological test data, which have frequently been used in
courts. Still, others worry that brain imaging data will be inappropriately
usedto "excuse"violent crimes.
Yet perhaps brain imaging researchon violence is most troubling to
somebecauseit challengesthe way we conceptualizecrime. It questions our
treatment of murderers in just the same way that we now look back 200
yearsand question the way in which the mentally ill were kept in shackles
and chains, treated little better than animals. The history of civilization has
shown that as time progresses,society becomeswiser and more humane.

74

Adrian

Raine

Two hundred years from now, we may have reconceptualized recidivistic

serious criminal behavior as a clinical disorder with its roots in early social,
biological, and genetic forces beyond the individual's control (Raine 1993).
Will we look back aghast at our current practices of execution and inhumane treatment of seriously violent offenders?Will we view this execution
of prisoners as barbaric and unjustified, as we now view the burning of
witches?
Biological researchis beginning to give us new insights into what makes
a violent criminal offender. It is hoped that theseearly findings may lead us
to rethink our approach to violence and goad us into obtaining new answers
to the causesand cures of crime while we continue to protect society.