J. Child Psychol. Psychiat. Vol. 40, No. 8, pp. 1197–1208, 1999
Cambridge University Press
? 1999 Association for Child Psychology and Psychiatry
Printed in Great Britain. All rights reserved
Executive Functions and Physical Aggression after Controlling for
Attention Deficit Hyperactivity Disorder, General Memory, and IQ
Jean R. Se?guin and Bernard Boulerice
Universite? de Montre?al, Canada
Philip W. Harden
Montreal Children’s Hospital, Canada
Richard E. Tremblay
Universite? de Montre?al, Canada
Robert O. Pihl
McGill University, Montre?al, Canada
This study examined the role of ADHD in the association between physical aggression and
two types of executive functions. Boys received a cognitive-neuropsychological test battery
was obtained at age 15. Three groups, differing in stability and level of physical aggression
since kindergarten, were formed: Stable Aggressive, Unstable Aggressive, and Non-
aggressive. Composite scores of validated executive function tests of working memory
representing subjective ordering and conditional association learning were formed. A
MANCOVA (N?149) using ADHD status, teacher-rated negative emotionality, general
memory abilities, and IQ as covariates was performed on the two composite scores. ADHD
and teacher-rated emotionality did not provide significant adjustment to the dependent
variables. Number of ADHD symptoms was negatively associated only with general
memory and IQ. General memory contributed significantly to adjusting for conditional
association test scores. Group differences indicated lower conditional association scores for
Unstable Aggressive boys relative to the other groups. Both IQ and general memory abilities
interacted with subjective ordering within the groups. Specifically, Stable Aggressive boys
performed poorly on this measure and did not benefit from increases in IQ whereas
Nonaggressive boys performed best and were not disadvantaged by lower general memory
abilities. This suggests a relationship exists between aspects of working memory and a
history of physical aggression regardless of ADHD and IQ.
Keywords: ADD?ADHD, aggression, executive function, intelligence, cognition, hyper-
Abbreviations: CD: Conduct Disorder; DISC: Diagnostic Interview Schedule for Children;
NSCA: Non-Spatial Conditional Association test; SCA: Spatial Conditional Association
test; SOP: Self-Ordered Pointing test.
The role of cognitive factors in psychopathology has
been the object of a renewed interest. Studies of
externalising disorders have led several authors to focus
Requests for reprints to: Jean R. Se?guin, PhD, Research Unit
on Childhood Psychosocial Maladjustment, Universite? de
Montre?al, 3050 E?douard Montpetit, PO Box 6128, Succ.
Centre Ville, Montre?al, Que?bec, Canada H3T 1J7 (E-mail:
on one cognitive domain called executive function
(Barkley, 1997; Pennington & Ozonoff, 1996; Zelazo,
Carter, Reznick, & Frye, 1997). In general, current
definitions of executive functions comprise those abilities
implicated in goal-oriented processes such as the in-
itiation and maintenance of efficient strategies (Lezak,
1985), and the programming and planning of motor
behaviour skills. They also encompass learning and
applying contingency rules, abstract reasoning, problem
solving, sustained attention, and concentration. Execu-
tive functions (1) require effort and active ‘‘on-line’’
1198J. R. SE?GUIN et al.
monitoring resources in contrast with automatic pro-
cessing, are called (2) particularly to process novel
information (Zelazo et al., 1997) or (3) when high
demands for cognitive resources are solicited (Stuss,
& Pennington, 1988). They are largely but not exclusively
associated with the frontal lobes (Pennington & Ozonoff,
1996; Stuss, 1992).
An earlier report (see Se?guin, Pihl, Harden, Tremblay,
& Boulerice, 1995) concluded that in the context of a
large number of cognitive-neuropsychological domains,
executive functions were particularly impaired in boys
with a history of physical aggression more than any other
group of abilities. This conclusion raised an important
question regarding the issue of specificity of executive
if the relation between poor executive function and a
history of physical aggression could have been mediated
by the presence of Attention Deficit Hyperactivity Dis-
order (ADHD) (Pennington & Ozonoff, 1996). Physical
aggression is a diagnostic symptom of Conduct Disorder
(CD) and an associated feature of ADHD. Further, CD
and ADHD are often reported in the same individuals
and have both been established as risk factors for
alcoholism, most likely via a history of physical ag-
functions, Pennington and Ozonoff (1996) concluded
that impairments in executive functions are found in
ADHD and autism and that the evidence for poor
executive function involvement in CD without ADHD
was weak. Further, when studies of CD do report
executive function deficits, it is unclear whether ADHD
may still be confounding the results or not. One recent
study showed an association between aggressive CD in
for ADHD (Giancola, Mezzich, & Tarter, 1998). How-
ever, the authors did not control for nonexecutive
processes that may affect executive function tasks. The
importance of examining the role of ADHD in any
analysis of executive functions in externalising disorders
is also brought to the forefront by other current theor-
etical work. For example, Barkley (1997) recently de-
veloped a theory of the hyperactive-impulsive type of
ADHD in which he attributes a prominent role to poor
executive functions and behavioural inhibition. Since
none of the studies reviewed by Pennington and Ozonoff
(1996) (including our earlier report) supported an un-
equivocal conclusion about the executive function?CD
association when CD was not comorbid with ADHD, we
have sought to control for ADHD in this study of
executive functions and physical aggression. Thus our
first question examines the robustness of the executive
function?physical aggression relationship while con-
trolling for ADHD.
It has become clear that there are several aspects to
executive functions and various models use different
categories to account for these. For example, they have
been classified into categories such as set-shifting, plan-
ning, working memory, contextual memory, inhibition,
and fluency (Pennington & Ozonoff, 1996). Using a
problem-solving framework, the executive function has
(1) problem representation, (2) planning, (3) execution
(including both intending and rule use), and (4) evalu-
ation (including both error detection and error cor-
rection) (Zelazo et al., 1997). In another framework, they
have also been grouped into categories that include
working memory, self-regulation of affect-motivation-
arousal, internalisation of speech, and behavioural
analysis-synthesis (Barkley, 1997). Using the frontal lobe
metaphor, one particular subgroup of functions involved
frontal lobe activity (Petrides, Alivisatos, Evans, &
Meyer, 1993a). Within the dorsolateral frontal lobe there
are two dissociable areas, the posterior and mid-
dorsolateral areas (Petrides et al., 1993a). The posterior
dorsolateral front lobe appears to be specifically involved
in conditional association tasks, whereas the mid-
dorsolateral frontal lobe would be involved in tasks of
subjective ordering (Petrides et al., 1993a).
A number of studies suggest that performance of
physically aggressive individuals may differ on tests
representing these two domains of functions. Giancola’s
(1995)reviewsuggests that poorConditional Association
abilities, but not Subjective Ordering abilities, are
involved in laboratory aggression in normal males. One
study by Giancola and Zeichner (1994) found that
laboratory aggression (shock intensity and duration in
response to provocation) was related to poor perform-
ance on the Spatial Conditional Association test; (SCA;
(SOP; subjective ordering, Petrides & Milner, 1982). This
speculated that it might be an artefact of the high level of
education of their participants. In a similar study by Lau,
Pihl,andPeterson(1995),groupswere dividedinto upper
and lower quartiles on a composite score constructed
from tests of conditional association and subjective
to provocation) were higher in participants with poorer
executive functions. However, the composite score was
made of two variables extracted from the SCA and the
SOP. It is likely that the SCA contributed more than the
SOP to the composite by inflating the correlation of SCA
performance to the composite (Tabachnick & Fidell,
1989), a correlation that the authors did not report. We
also reported findings implicating a composite score
including Subjective Ordering and Conditional Associ-
ation Learning tests, and its negative association with a
childhood history of physical aggression (Se?guin et al.,
1995). In that study, the highest loading was for a test of
Non-Spatial Conditional Association (NSCA; Petrides,
1985, 1990), although the loadings for two different
subjective ordering tests (the SOP and a number
randomisation task) were also above .50. The NSCA is
related to, but slightly different from, the SCA used in
the above-mentioned studies. Composite scores, useful
as they may be, do not allow an assessment of the tests’
discriminant ability. It is thus possible that the two
executive function domains are not equally associated
with physical aggression. Following Giancola’s con-
clusion it would seem that performance on Conditional
Association Learning tests might have more relevance to
laboratory and physical aggression than performance on
Subjective Ordering tests. This comparison constitutes
our second question.
1199EXECUTIVE FUNCTIONS AND PHYSICAL AGGRESSION
In order to proceed with both questions, thorough
examination of executive functions requires essential
controls for cognitive abilities that may affect them
(Pennington & Ozonoff, 1996; Zelazo et al., 1997). Using
the frontal lobe metaphor, this issue addresses the role of
nonfrontal functions on measures of frontal executive
function. Theoretically, in order to conclude that a group
of individuals is impaired in working memory (i.e. more
specifically dorsolateral frontal lobe), one has to ensure
that variations in General Memory functions (i.e. those
often associated with the hippocampi and medial tem-
poral lobe) have been controlled (Petrides, 1995). The
model developed by Milner and Petrides postulates that
impairments in General Memory will necessarily affect
executive function (Luria & Homskaya, 1964; Petrides
& Milner, 1982), at least in individuals with lesions. This
neuropsychological point of view speaks directly to two
issues: (1) pre-frontal functions are affected by brain
changes in regions outside but closely connected to the
pre-frontal cortex (Petrides, 1995), and (2) nonexecutive
components are involved in executive function tasks
(Pennington & Ozonoff, 1996). This important notion is
similar to a methodological dissociation relevant to the
assessment of working memory functions as outlined
by Barkley (1997). This requirement specifies that the
involvement of poor executive functions must be
assessed in the context of proper information storage
(knowledge). Problems should only be found in infor-
mation manipulation and organisation over time (see
also Delis, Squire, Bihrle, & Massman, 1992; Stuss &
Benson, 1986; Zelazo et al., 1997). Finally, from a
psychological point of view, this speaks to characteristics
of complex functions that are inherently hierarchical in
nature (Zelazo et al., 1997). First, there may be a
hierarchy within the executive function domain and,
second, there are no executive function tasks exclusively
sensitive to variations in executive function or that
are not ‘‘contaminated’’ by non-executive-function
processes (Zelazo et al., 1997).
Aside from General Memory abilities, another non-
executivecomponentinvolvedinexecutive function tasks
may be IQ. There is overlap between IQ and executive
functions and one would not want to conclude that poor
executive function is a characteristic of behaviour
problems without ensuring that intelligence is not con-
founding the results. Although intelligence is often
reported as being independent of executive function
(Milner & Petrides, 1984), the capacity for readily
understanding and remembering the fairly complex rules
involved in some tests could presumably be affected by
poor intelligence. Intelligence, in addition to affecting
executive function, may mediate aggressive behaviour.
Indeed, intelligence is considered to be a protective factor
against the development of delinquency and criminality
(Kandel et al., 1988; Lynam, Moffitt, & Stouthamer-
Loeber, 1993; White, Moffitt, & Silva, 1989). A control
forIQin thecurrent studyisevenmore importantin light
of the studies of the role of the dorsolateral frontal lobe
and laboratory aggression reviewed above. Indeed, the
groups with high and low executive functions formed
by Lau et al. (1995) differed not only in laboratory
aggression but also in IQ. Thus it is not clear whether IQ
relationship they reported. On the other hand, Giancola
and Zeichner (1994) did control for IQ. It is possible that
the influence of IQ in their analyses may also have been
minimised by their high average levels of education
(mean?16.1 years, SD?1.9) and of IQ (mean?110.5,
SD?11.1). This suggests a problem of range. The role
of IQ in the low SES community sample we propose
to examine may be more important for the executive
function?physical aggression relationship, which may be
mediated, if not explained, by IQ.
In sum, we examined if poor performance on tests of
executive functions remained a feature of physical ag-
gression once ADHD status, General Memory abilities,
and intelligence were controlled. We expected that a
history of physical aggression would be characterised
more by poorer performance on conditional association
tests than on subjective ordering tests. To answer these
questions, data on an additional conditional association
test for working memory, on intelligence, and on ado-
lescence psychiatric diagnoses were collected on most of
the participants that were the object of our initial report
(Se?guin et al., 1995) in the 2 years that followed. We
added previously collected data on General Memory
abilities and executive function to complete our model.
portions of the cognitive-neuropsychological test battery have
been published elsewhere (see Se?guin et al., 1995), and will be
summarised where relevant.
Participants were obtained from a large low socioeconomic
status community sample of 1037 boys that has been followed
longitudinally since kindergarten. Teacher ratings of aggressive
behaviour at ages 6, 10, 11, and 12 (i.e. fights with other
children, kicks, bites and hits other children, and bullies or
intimidates other children; alpha?.86; Tremblay et al., 1991)
were used to form three nonoverlapping categories of boys who
differed in stability and severity of physically aggressive
behaviour.The Stable Aggressive group was composed of those
boys who were rated consistently high in their aggression at age
6 and later (cutoff for the scale score at the 70th percentile).
Those who were consistently below the cutoff point over the
same period formed a Nonaggressive group. The overwhelming
majority of Nonaggressive boys had a score of 0 across all
assessment points. Finally, those who did not meet the criteria
above and who were occasionally highly aggressive (at one or
two measurement points) formed an Unstable Aggressive
group?. In our initial report 177 boys came at both ages 13 and
dimension: negative emotionality. Efforts had been made at the
time to recruit participants who would show high levels of
aggression but low levels of negative emotionality in order to
identify individuals who would show characteristics of low-
anxious psychopaths. This purpose was independent of the
current study. Although there are too few such individuals to
perform statistical analyses, the sampling procedure resulted in
a change of the bivariate relation between physical aggression
to the large sample. This effect was controlled.
1200J. R. SE?GUIN et al.
14 (Se?guin et al., 1995). At age 15, 161 of them returned to our
laboratory for more cognitive-neuropsychological testing and
other scientific activities.
cluded the Digit Span and Paired Associates subtests from the
Wechsler Memory Scales-Revised (Wechsler, 1987), three
frontal lobe tests validated by lesion and brain imaging studies,
the NSCA (posterior dorsolateral) and SOP (subjective
ordering, mid-dorsolateral) tasks (Petrides et al., 1993a), and
the number randomisation (Petrides, Alivisatos, Meyer, &
Evans, 1993b) (a.k.a. subjective ordering; mid-dorsolateral;
Wiegersma, van der Scheer, & Human, 1990). This portion of
the battery was administered over two laboratory visits 1 year
apart (ages 13 and 14). These tests were included in two factors
defined by items loading above .50 and had been derived from
a larger cognitive-neuropsychological battery of tests after
oblique rotation(Se?guin et al., 1995). For this study we retained
elements of two factors representing verbal learning (Digit
Span, Paired Associates, respective loadings .55 and .55), and
executive functions (NSCA, SOP, and number randomisation,
respective loadings .82, .55, and .69). Digit Span and Paired
Associates theoretically assess General Memory processes such
as those subsumed by the hippocampus and medial temporal
lobe (see Petrides, Alivisatos, & Evans, 1995; Squire & Zola-
For the age 15 laboratory visit we collected the SCA task
(posterior dorsolateral frontal lobe; Petrides, 1985) that was
used in Giancola and Zeichner (1994) and Lau et al. (1995). We
also collected IQ estimates derived from the Vocabulary and
Block Design subtests of the Wechsler Intelligence Scale for
of these subtests correlates positively at more than .90 with Full
Scale IQ (Sattler, 1988). Although the measure of IQ was not
administered concurrently with the bulk of the cognitive-
neuropsychological assessment, its score should be reliable
since IQ is relatively stable across time. We will verify this claim
with the use of IQ estimates collected on a subsample of these
boys at ages 10, 11, and 12.
Task instructions are summarised as follows. The Digit Span
involves repeating digits in increasing spans in forward and
backward orders (Wechsler, 1987). The Paired Associates
requires listening to pairs of easy or difficult words. The first
word of the pair is then given in order to prime recall of the
second word of the pair (Wechsler, 1987). In the number
randomisation task a range of numbers is provided (e.g. from 1
to 10) (Se?guin et al., 1995). All numbers in the range must then
The SOP consists of two sets (one abstract and one concrete) of
12 arrays of the same 12 stimuli (Milner, Petrides, & Smith,
1985). The position of the stimuli changes from one array to the
next and subjects must select a different stimulus on each array.
Repetitions are counted as errors. In conditional association
learning tasks one must uncover six pairs of stimuli. In the
NSCA, six hand signals are initially taught (Petrides, 1990).
Then six colours are presented in a predetermined random
order, one at a time, until correct identification of the colour-
hand signal combination is achieved. For the SCA, six abstract
symbols need to be paired with small lights in six locations
(Petrides, 1985). Instead of colours, abstract symbols are
presented one at a time. In both cases, trials are repeated until
a criterion of 18 consecutive errorless trials is met. The
Vocabulary subtest of the WISC-R requires the definition of
words of gradually increasing difficulty (Wechsler, 1974).
Similarly, for the Block Design subtest of the WISC-R, subjects
must reproduce two-dimensional patterns of red and white
The test battery in-
for Children (DISC) was used to obtain childhood psychiatric
diagnoses (Costello, Edelbrock, & Costello, 1985). The DISC is
an assessment tool based on the Diagnostic and statistical
manual of mental disorders from the American Psychiatric
Association. The version employed in this study was based on
DSM-III-R criteria (American Psychiatric Association, 1987),
and is a French adaptation of the DISC 2.25 using child (-C)
and parent (-P) forms (Valla et al., 1994). In its original form,
theDISC-P isthoughttobemore sensitivethan theDISC-Cfor
several disorders (Fisher et al., 1993). Test–retest reliability of
the DISC 2.1 was moderate (i.e. CD, oppositional defiant
disorder [ODD]) to substantial (i.e. for ADHD), and found to
be best using a combination of both child and parent reports
(Jensen et al., 1995). More specifically, DISC-P reliability was
good to excellent for ADHD and fair for other externalising
disorderswithchildrenaged6–11,whereasit waspoorfor other
disorders using the DISC-Revised version (Schwab-Stone,
Fallon, Briggs, & Crowther, 1994). The relative validity of the
parent or child versions is not well known, but the combination
of parent and child reports would better approximate clinical
practice (Jensen et al., 1995).
Using symptom scores from our larger sample, reliabilities
for the child and parent as informants were respectively .88 and
.92 for ADHD. Spearman correlations for the child and parent
as informants were respectively of .42 and .43 for ADHD and
ODD and .32 and .34 for ADHD and CD. Correlations
between ADHD and internalising disorders were all below .10
for both informants. Results of a large random survey of
children aged 6 to 14 years old and using both the French and
English versions of the DISC 2.25 were published in an official
report (Valla et al., 1994). In that report, issues of validation
problems that hardly affected factual data. However, overall
between-informant agreement was low and, specifically,
teacher–parent agreement were higher than child–parent or
child–teacher agreements. These results nevertheless remain
consistent with those of several other studies (Achenbach,
McConaughy,& Howell,1987; Offord,Boyle, & Racine,1989).
Specifically, and as expected, externalising disorders were more
prevalent in boys and internalising disorders were more
were respectively 6.9% and 2.8% by child report, 7.3% and
5.1% by parent report, and 13.2% for age 9–11 by teacher
report. The boys’ prevalence rate for CD was around 3% for
child report between ages 9–14. Children with mental health
problems tended to have parents with such problems as well,
higher familial adversity. Poor social competence and low
father education were associated with externalising disorders.
In contrast to other epidemiological studies of the kind,
academic problems seemed to be exclusively associated with
The Diagnostic Interview Schedule
a letter followed up by telephone contacts served to solicit
participation. After setting an appointment, two trained inter-
viewers visited the boy’s residence in order to obtain in-
formation simultaneously from both informants individually
(the boy and one guardian, mainly the mother). The interview
lasted between 1 and 1??hours and was given in paper-pencil
form. DISC scores were obtained for 739 boys from the large
longitudinal sample. From the laboratory sample of 161,
The cognitive neuropsychological
administered over three visits to the laboratory at ages 13, 14,
test battery was
1201EXECUTIVE FUNCTIONS AND PHYSICAL AGGRESSION
object of a previous report (Se?guin et al., 1995). At age 15, the
IQ assessment was administered following the regular pro-
full IQ score estimate was obtained following Wechsler’s (1987)
instructions. The SCA test was administered during that same
ADHD, IQ, SCA and other cognitive-neuropsychological
data were available for 149 of the 161 participants. Analyses
were carried out on SPSS version 4.0 for UNIX.
No multivariate outliers could be identified (Tabachnick &
Fidell, 1989). In order to meet statistical assumptions of
received square-root and reciprocal transformations, data for
both the NSCA and SCA received square root transformation,
and a few outliers (three for NSCA and six for SCA) were
moved into the distribution next to the highest nonoutlying
value while respecting their ranking, data for Paired Associates
received logarithmic and reciprocal transformations, and nega-
tive emotionality received a square root transformation
(Tabachnick & Fidell). Signs for SOP, NSCA, and SCA were
changed to reflect better performance with higher scores. Test
scores of SOP abstract and concrete errors, and number
randomisation number of successes, were standardised and
averaged to form one composite score of Subjective Ordering.
This procedure was repeated with NSCA trials with errors, and
SCA trials with errors to form the Conditional Association
Learning composite score. Digit Span and Paired Associates
easy and difficult correct responses were standardised and
averaged to produce a General Memory control. Examination
of the variables revealed normal distributions within the groups
as measured by skewness and kurtosis following trans-
formations (Tabachnick & Fidell).
ADHD status using combined parent and child report can be
to boys who met diagnostic criteria based on parent or child
report.In thelaboratorysample(N?149),therewerea total of
19 participants (12.8% of the laboratory sample) who met
diagnostic criteria of ADHD by either parent or self-report.
None of the self-diagnoses and parent diagnoses of ADHD
coincided. Missing data within this sample’s ADHD diagnoses
were found for seven parent reports. Two of them met
diagnostic criteria according to self-reports. Breakdown by
groups for the participants who received a diagnosis of ADHD
was 10 Stable Aggressive, 8 Unstable Aggressive, 1 Non-
aggressive. Thus, 95% of ADHD participants were classified as
aggressive in one form or another.
Given the lack of overlap between informants (self- and
parent) regarding the ADHD diagnosis in the laboratory
sample, we examined the overlap in the larger sample for
ADHD symptoms. In the larger sample, ADHD diagnoses
using self-report or parent report could be established for 749
boys. Thus 78 boys met criteria for an ADHD diagnosis,
or Unstable Aggressive accounted for 92.3% of ADHD boys.
Thus the proportion of ADHD cases in the laboratory sample
cannot be attributed to the selection of the subsample from the
large one. Overall, parents (N?54) more than children (N?
20) reported ADHD symptoms in the large sample; four cases
met diagnostic criteria by both informants. Within those boys
who met diagnostic criteria by either parent or child report,
between-informant average agreement across the 14 symptoms
regarding (a) presence of symptoms was 18.8%, (b) absence of
symptoms was 23.7%, (c) symptom(s) identified by parent only
was 46.5%, and (d) symptom(s) identified by boy only was
11%. Thus, the agreement concerning the presence of
symptoms is much more important than it appears to be when
presenting ADHD diagnoses.
Teacher-rated negative emotionality was used as another
covariate since it had been a criterion for the selection of the
laboratory sample for reasons that were not the object of this
We predicted earlier that IQ at age 15 could serve as a
reliable control for neuropsychological test performance
IQ estimates at those ages, we were able to examine IQ
stabilityovertime ina subsampleoftheseboys.Thesame
IQ estimates had been collected at ages 10, 11, and 12 in
other components of this longitudinal study. An average
of those 3 years’ data collection correlated at .92 with age
Physical Aggression and Executive Functions
The main questions involved a control for ADHD, IQ,
and General Memory. These variables were entered as
covariates in a MANCOVA comparing the three groups
(independent variable) on the two executive function
composite scores (dependent variables). ADHD status
was entered as a dummy variable. We also controlled for
of cognitive neuropsychological variables and composite
scores prior to the analyses are presented in Table 1 for
descriptive purposes. Table 1 also includes Spearman
correlations for ADHD status and number of symptoms
with the cognitive neuropsychological variables. An
examination of Table 1 reveals a nonsignificant cor-
relation between the two Subjective Ordering tests, and a
high correlation between the two Conditional Associ-
ation Learning tests. The two composite scores were also
moderately correlated with each other and with the
General Memory composite and IQ. Since tests included
in the composite scores were averaged, this attributed
them equal weights as indicated by the fact they correlate
equally with the composite score. The correlations be-
for number of ADHD symptoms with IQ, the General
Memory composite, and the Paired Associates learning
test. These latter correlations were negative, as expected.
The MANCOVA was then performed on 149 cases: 57
in the Stable Aggressive group, 45 in the Unstable
Aggressive group, and 47 in the Nonaggressive group.
The initial step consisted of verifying the general
assumptions for the model. This examination revealed
significant heteroscedasticity of the variance?covariance
matrix across aggression groups. We postulated that this
phenomenon could be due to differential effects of the
covariates between the aggression groups (i.e. inter-
actions). The first covariate to be examined was ADHD.
1202 J. R. SE?GUIN et al.
Correlations among Cognitive-neuropsychological Tests and ADHD Prior to MANCOVA (N?149)
NRSOP SOc NSPSPA CAc DSPANPASS MEM IQ-ESADHD
ADHD: Attention Deficit Hyperactivity Disorder; ADHDS: ADHD symptom count; CAc: Conditional Association composite;
DSPAN: Digit Span; IQ-ES: Intelligence Quotient Estimate; MEM: General Memory; NR: Number Randomization; NSP: Non-
Spatial Conditional Association; PASS: Paired Associates; SPS: Spatial Conditional Association; SOc: Subjective Ordering
composite; SOP: Self-Ordered Pointing.
***p?.001; **p?.01; *p?.05.
Multivariate Statistics for Interactions and Main Effects on Subjective Ordering and
Conditional Association Learning Composite Scores
General memory by groups
IQ by groups
Univariate F Statistics for Subjective Ordering and Conditional Association Composite Scores
Subjective OrderingConditional Association
General memory by groups
IQ by groups
a η? (eta square) was used by SPSS to calculate the effect size.
However, the very low number of ADHD participants in
the nonaggressive group precluded direct verification of
the interaction term. Instead, we removed ADHD from
the model and observed that this did not resolve the
heteroscedasticity problem. A finer examination of the
correlations between number of ADHD symptoms and
the dependent variables in the groups revealed non-
significant correlations. We then reintroduced ADHD in
the model and examined if the assumption of hom-
ogeneity of regression slopes (i.e. interactions) for nega-
tive emotionality, IQ, and General Memory was met
(Tabachnick & Fidell, 1989). The interaction for negative
from the model. The two remaining interactions were
score. The effect for the Conditional Association Learn-
at p?.05 and power was low. The main effect for the
Conditional Association Learning composite score was
emotionality and ADHD status were not. Hotellings T
and F statistics for the multivariate effects are presented
in Table 2 for main effects and interactions. Corre-
sponding univariate statistics for Subjective Ordering
and Conditional Association Learning composite scores
are presented in Table 3. Results for Conditional As-
sociation Learning test scores will be presented first
followed by those for Subjective Ordering test scores.
1203 EXECUTIVE FUNCTIONS AND PHYSICAL AGGRESSION
Means and Standard Errors for Cognitive-neuropsychological Variables per Group
azSO: final adjusted standardised Subjective Ordering composite score; azCA: final adjusted
standardised posterior Conditional Association score; gMEM: General Memory; IQ-ES:
Intelligence Quotient estimate; zSO: nonadjusted standardised Subjective Ordering composite
score; zCA: nonadjusted standardised Conditional Association composite score.
a Means adjusted for the covariates negative emotionality and ADHD status.
b Means also adjusted for General Memory and IQ.
Conditional Association Learning test score main effect.
Since there were no interactions with the Conditional
Association Learning composite score, the main effects
reported in Table 3 may be considered for that variable.
The effect of aggressive groups on the Conditional
Association Learning score was significant. Table 3
indicates that General Memory brought significant cor-
rection to the Conditional Association Learning com-
posite score whereas IQ did not. Means and standard
errors for all the cognitive-neuropsychological variables
are presented in Table 4. The effects of the covariates on
the group means can be examined in Table 4 by
comparing nonadjusted to adjusted means for the execu-
tive function test scores. Planned contrasts for the main
effect of aggressive groups revealed differences between
the Stable Aggressive and Unstable Aggressive groups,
t (100)??2.41, p?.02, and between the Unstable Ag-
gressive and Nonaggressive groups, t (90)?2.34, p?
.02. This indicates that performance on the Conditional
Association Learning composite score was significantly
lower for the Unstable Aggressive group.
to the Conditional Association Learning test score, the
adjusted means and standard errors for the Subjective
Ordering score were ordered by level of aggression with
the Stable Aggressive group showing the poorest per-
formance. However, in MANOVA interactions preclude
any straightforward conclusion from the significant main
effect. In examining both interactions, we also verified if
these have not been caused by different ranges in the
aggression groups, ruling out the possibility of confound
between aggression groups and the covariate in their
relationship to the Subjective Ordering score.
General Memory and aggressive groups interaction.
This interaction for the Subjective Ordering composite
indicates that the regression slopes for both aggressive
groups do not differ significantly though they differ
significantly for the Unstable Aggressive and Non-
aggressive groups (parameter coefficient ?0.57, SE?
0.19, t??3.02, p?.003). In other words, we found that
an increment of General Memory that is associated with
an increase on the Subjective Ordering composite of 0.57
SD in the Unstable Aggressive group will be associated
is no relation between General Memory and the Sub-
jective Ordering composite score in the nonaggressive
groups. No range problems were observed.
IQ and aggressive groups interaction.
General Memory, this interaction for the Subjective
Ordering composite indicates that the slopes differed for
the aggressive groups (parameter coefficient 0.44, SE?
0.15,t?2.92,p?.004). Inotherwords, the increment of
IQ score that is associated with an increase of 1 SD in the
Stable Aggressive group is associated with an increase of
1.44 SD in the Unstable Aggressive group. However, as
opposed to General Memory, boxplots indicated that the
in overlap and in skewness. Therefore, teasing the effects
of aggression groups apart from IQ in their relation
with the Subjective Ordering composite is seriously
Interactions conclusion.In sum, the first interaction
indicates that, overall, the Nonaggressive group’s per-
formance is not as impaired as that of both aggressive
groups at lower levels of General Memory. The second
interaction indicates that higher IQ scores do not provide
the advantage to the Stable Aggressive group that it does
for the two other groups.
In contrast to
Specificity of Executive Function Test Performance
to Physical Aggression vs. ADHD
It is clear that in the current sample ADHD does not
account in any way for the observed association between
executive functions and a history of physical aggression.
Although ADHD was strongly associated with a history
of physical aggression, it was not associated at all with
poor executive functions as the literature generally seems
to indicate (Barkley, 1997; Pennington & Ozonoff, 1996).
Several factorsmay account for these observations in this
1204 J. R. SE?GUIN et al.
First, we chose to control for ADHD by using a
composite of child and informant diagnoses. Diagnostic
then combined. This method has limitations. Two other
methods have been used as well. One of them consists of
combining child and informant reports at the symptom
level (Jensen et al., 1995; Schwab-Stone et al., 1993). In
this situation, a symptom is given a 1 if the child or the
informant has reported it. Symptoms are then added
from this list to form a total symptom score. The total
symptom score can then be used by itself or following the
DSM diagnostic criteria. In the latter situation a positive
diagnosis is determined when the combined informant
symptom count reaches the symptom count criterion. We
Whether we used one or the other method, our results
were essentially the same as those already reported.
Interestingly, the second diagnostic approach increased
groups only. Nonetheless, the ADHD symptom count
was negatively associated with IQ and General Memory
scores. This indicates sensitivity of non-executive func-
tion cognitive abilities to ADHD as measured in this
Second, there may be differences between clinical and
of a study by McGee, Williams, Moffitt, and Anderson
(1989), who failed to find executive function deficits in
ADD-H in a community sample. However, the com-
parison with that study remains limited as ADD-H as
well as the neuropsychological assessments were differ-
ent and many of our non-ADHD participants could
not really qualify as controls since they had other
behaviour problems. Nonetheless, it is possible that
ADHD and poor neuropsychological function are more
strongly associated in referred samples (Pennington &
Ozonoff, 1996). Further, test–retest reliabilities of di-
agnostic instruments used in community vs. clinical
settings tend to be poorer in the former (Jensen et al.,
1995). Third, there is a possibility that our measures of
executive functions are different from most measures
employed in studies of ADHD. Since there are many
subdomains of EFs, even within working memory, as we
fourth possibility is that our measure of ADHD is
inappropriate. We are aware of the controversies sur-
Solovitz, Chen, & Casat, 1996; Shaffer, 1994). Although
far from perfect, version 2.25 represented improvements
2.25 in childhood has been questioned. It was found that
less than half of 9–11-year-old children really understood
the interview questions (Breton et al., 1995; Schwab-
Stone, 1995). However, we administered the DISC to
boys age 14–16, where this problem is less of an issue
(Edelbrock, Costello, Dulcan, Kalas, & Conover, 1985).
Further, and as indicated earlier, total number of ADHD
symptoms did correlate negatively with General Memory
and IQ, and ADHD status was reported specifically in
aggressive groups. A fifth possibility has to do with the
fact that our assessment of ADHD took place in
adolescence rather than in childhood. This methodo-
logical point, though, should facilitate the detection of
boys whose hyperactivity is truly chronic since ADHD
tends to be of early onset and stable (Biederman et al.,
1996). Further, ADHD appearsto show the samepattern
whether it is assessed in childhood or adolescence
(Biederman et al., 1998). However, we did not benefit
from a full prospective history of ADHD in the partici-
pants from this study.
The Dorsolateral Perspective on Executive
Functions and Physical Aggression
In this study executive functions were defined from a
neuropsychological perspective as performance on tests
theoretically assessing dissociable functions that cor-
respond to the mid-dorsolateral and posterior dorso-
lateral frontal lobe (Petrides et al., 1993a). The respective
abilities subsumed by these areas are Subjective Ordering
and Conditional Association Learning. Unlike a pre-
viously reported finding that indicated poorer perform-
ance in Stable and Unstable Aggressive groups than in a
Nonaggressive group on a composite of Subjective
Ordering and Conditional Association Learning test
scores, disaggregating these scores allowed us to uncover
two different patterns. The most straightforward pattern
concerned Conditional Association Learning test per-
formance. It is the Unstable Aggressive group that
showed the poorest performance relative to the Stable
and Nonaggressive groups. Although no specific contrast
had been planned to examine the differences between
Stable and Nonaggressive boys, an examination of
adjusted means and standard errors in Table 4 seems
consistent with the absence of differences between those
two groups. The second pattern revealed a relationship
between Subjective Ordering test performance and physi-
cal aggression that is not straightforward and requires
The question examined the relationship between non-
executive cognitive components and dorsolateral test
scores within the groups. We found interactions between
the Subjective Ordering composite score and the
covariates General Memory and IQ and must exert
indicate that the slopes describing the relationship for
General Memory and IQ with the Subjective Ordering
composite are not parallel. These interactions also mean
that General Memory does not bring any correction to
Subjective Ordering test performance in the Non-
aggressive group and that IQ fails to correct the same
scores in the Stable Aggressive group. From the point of
view of psychopathology, it appears that the Stable
Aggressive and Nonaggressive groups still differ. Not
only are their means ordered in the direction previously
reported, i.e. significantly better performance for Non-
with covariates seem consistent with that report.
Specifically, the Nonaggressive group does not appear to
be disadvantaged by lower General Memory abilities,
and the Stable Aggressive group does not gain the
advantage provided by higher IQ in the other groups.
Disaggregating the Subjective Ordering and Con-
ditional Association Learning test scores supports the
notion of involvement of poor executive function in boys
witha historyof physicalaggression. Theoutcome of this
1205 EXECUTIVE FUNCTIONS AND PHYSICAL AGGRESSION
process clearly indicates two profiles of executive
functions. Poor Conditional Association Learning abili-
of physical aggression whereas poor Subjective Ordering
of physical aggression from Nonaggressive boys. Thus
Unstable Aggressive boys are most clearly impaired in
cognitive functions that require learning new rules of
association, whereas Stable Aggressive boys do not differ
from Nonaggressive boys in these abilities once IQ and
General Memory have been controlled for. Stable Ag-
gressive boys, and Unstable Aggressive boys to a lesser
extent, have difficulties in actively organising greater
amounts of information in their environment by them-
selves than do Nonaggressive boys. Factors other than
cognitive abilities may be invoked in explaining why
variations in IQ in the Stable Aggressive group are not
associated with variations in Subjective Ordering test
scores, and why variations in General Memory are not
associated with variations in Subjective Ordering test
scores in Nonaggressive boys.
Our results may help clarify the apparent incon-
sistencies between studies of laboratory aggression that
were reported by Giancola and Zeichner (1994) and Lau
et al. (1995). Whereas these authors have examined the
reactions to provocation of individuals classified on the
basis of their EF test scores, we sought to examine the
executive function test performance of boys with known
histories of physically aggressive behaviour. Both lab-
oratory aggression studies have used mid- and posterior
dorsolateral tests but only Giancola and Zeichner looked
at nonaggregated scores. Further, Lau et al. biased their
composite executive function score by including two
highly correlated measures of the same Conditional
Association Learning test, i.e. total numbers of trials to
criterion and number of incorrect guesses. Their com-
posite score, which also included number of SOP errors,
was therefore disproportionately representative of Con-
ditional Association Learning test performance.
Giancolaand Zeichner’s(1994)finding that laboratory
aggression was specifically related to poor Conditional
Association Learning test performance and not to Sub-
jective Ordering test performance may indicate that
Unstable Aggressive boys have a history of aggression
because they are more likely to be easily provoked. This
is also consistent with the fact that these boys have a
tendency to be more sensitive to pain (Se?guin, Pihl,
Boulerice, Tremblay, & Harden, 1996), and have high
levels of neuroticism, which may contribute to failures
in self-regulation (Se?guin, Arseneault, Boulerice, &
Tremblay, 1999). Increased sensitivity to pain would
theoretically imply pain avoidance behaviour and there-
fore the avoidance of fights. However, poor performance
on Conditional Association Learning tasks may indicate
that these boys may not be very sophisticated when it
comes to understanding the rules of novel situations,
particularly those that may involve cues for provocation.
They may thus get stuck in fight-flight types of situations
for which they fail to see any escape (Gray, 1987).
Alternatively, these boys may attribute hostile intent
more readily by default because of their poor cognitive
teristic may thus render them more reactively aggressive
(Dodge & Coie, 1987; Dodge, Lochman, Harnish, Bates,
& Pettit, 1997). In sum, the poor cognitive abilities and
sensitivity to provocation could help explain the history
of physical aggression and the pain sensitivity could
explain the fact that their history of physical aggression is
The relative proficiency of Stable Aggressive boys on
Conditional Association Learning test performance may
appear as a surprise. However, one must keep in mind
that the similarity between Stable Aggressive and Non-
aggressive boys only appears once the significant
Withoutcovariates, the StableAggressive groupdoesnot
differ significantly from the Unstable Aggressive group in
It is also important to state that our analysis does not
Although the ‘‘frontal lobe metaphor’’ is a relevant
heuristic, it has several limitations when tests purported
to measure neuropsychological function are applied in
developmental psychopathology (Spreen, Risser, &
Edgell, 1995). Some of these limitations are psychometric
whereas others are theoretical. For example, from a
purely neuropsychological point of view one would have
expected a significant correlation between the two Sub-
jective Ordering tests. That was not the case. However,
two separate studies (Petrides et al., 1993a, b). One
problem may be that we administered both tasks on two
different visits. Although this methodological procedure
did not seem to affect the two Conditional Association
Learning tasks, it may have affected the Subjective
Ordering tasks. The former tasks are also more similar in
administration than the latter. It would be useful to have
fullvalidationstudiesusing both tasks in a samesitting in
order to strengthen the theoretical proposition. Further,
we do not know much about the stability of cognitive
abilities during adolescent development. While we have
shown that IQ is relatively stable, little is known about
the stability of executive functions. Nonetheless, our
prior factor analysis indicated that both tests of Sub-
jective Ordering loaded highly on the same executive
function factor (Se?guin et al., 1995).
Second, there may be conceptual questions regarding
the use of IQ subtests to control for executive function.
of the WISC-R as well as the Paired Associates test of the
Wechsler Memory Scale may reflect some executive
processing. However, from a neuropsychological per-
spective, circumscribed dorsolateral lesions should not
affect the Digit Span (Petrides et al., 1995). Although this
principle received support in lesion studies, it may not
hold for nonlesioned individuals since reciprocal con-
nections between otherwise dissociable brain areas re-
main intact. In fact the Digit Span and Paired Associates
tests loaded respectively at .39 and .21 on the executive
function factor in our former study (Se?guin et al., 1995).
This correlation may well be the fact that executive
functions theoretically require abilities such as those
involved in the Digit Span and Paired Associates tests.
On the other hand, this could be interpreted in favour of
conceptualising the Digit Span as an executive task. The
backwards version of the Digit Span test requires indeed
1206 J. R. SE?GUIN et al.
a little more on-line effort than its forward version, yet
the Paired Associates
remembering pairs of words in an immediate recall
format, whereas an associational learning executive
function task such as the Conditional Association Learn-
ing task requires figuring out the pairs of stimuli in a
much longer trial and error process with feedback. The
Digit Span and Paired Associates learning task are much
more passive memory tasks than the executive function
tasks presented here. With regards to the Block Design
task, the conceptual overlap with executive functions can
be seen in the requirements for planning, monitoring
performance, and error detection and correction. How-
ever, the results of the performance do not need to be
maintainedand manipulated on-linesince the participant
EF tasks, only the Conditional Association tasks have a
feedback component. However, that feedback needs to
be maintained in memory in a context of interference. In
sum, the ‘‘frontal lobe metaphor’’ clearly dissociates the
General Memory components from those of executive
function. However, the line between these types of tasks
is not drawn easily in a normal population; the difference
between executive functions and nonexecutive processes
necessary for them may appear as one of level, rather
thanquality, of processing. Nonetheless, the fact that our
results were obtained despite a rigorous control for these
IQ subtests and General Memory supports even more
strongly the importance of poor executive functions in a
history of physical aggression.
This focus on nonexecutive processes that may affect
executive functions and the notion of effortful processing
brings this discussion to a second set of factors that needs
to be examined as alternatives or complementary hy-
potheses to the ‘‘frontal lobe metaphor’’. The current
assessment did not measure any form of energy ex-
penditure. It is possible that those participants who
performed more poorly have done so because they did
not feel like putting any effort in achieving success. It is
also possible that they tried hard but failed. Current
testing procedures are not well designed to partial out
those processes. It would be very useful to identify the
conditions that will affect performance on executive
function tests. For example, Newman and colleagues
(1998) have formulated a theory to explain response
modulation deficits in psychopaths. These authors have
repeatedly shown that, once a dominant response is
established, psychopaths have difficulty automatically
attending to peripheral cues that would enable non-
is not attended to, it may not be processed at higher
case, the process at fault would not be due to a lack of
effortful processing and may theoretically occur in
individuals with otherwise adequate executive functions
(Se?guin et al., 1999). Newman’s position is akin to that of
Barkley (1997), who indicates that unless inhibition
occurs there will not be adaptive use of executive
functions. Finally, another factor at play may be
impulsivity. Only carefully crafted automated test bat-
teries designed to measure reaction times with well-
defined information processing phases will be of help in
examining impulsivity. These batteries could also be
designed to put constraints on performance such as
forcing a participant to pace her or his responses. Pacing
the responses could improve the capacity for reflectivity.
Newman and others (Backorowski & Newman, 1990;
Newman, Kosson, & Patterson, 1992; Patterson &
Newman, 1993) have shown improved performance
under certain conditions when delays to responding were
imposed and when contingencies were manipulated.
Acknowledgements—The National Science and Engineering
Research Council of Canada and the Fonds pour la Formation
des Chercheurs et l’Aide a ? la Recherche (FCAR: Que?bec) made
this collaborative work possible by scholarships to the first
author. The Social Sciences and Humanities Research Council
of Canada, FCAR and the Conseil Que?be?cois de la Recherche
Sociale funded the project and the research centre. Thanks are
extended to the boys and their families, to the Commission des
E?coles Catholiques de Montre?al, to the Research Unit on
Children’s Psychosocial Maladjustment, to the Research Team
on Prevention and treatment of Substance Abuse, to Denis
Larocque, to Michael Petrides, and to three anonymous
Achenbach, T. M., McConaughy, S. H., & Howell, C. T.
problems: Implications of cross-informant correlations for
situational specificity. Psychological Bulletin, 101, 213–232.
American Psychiatric Association. (1987). Diagnostic and stat-
istical manualof mental
Washington, DC: Author.
Bachorowski, J.-A., & Newman, J. P. (1990). Impulsive
behavior: Effects of personality and goal salience. Journal of
Personality and Social Psychology, 58, 512–518.
Barkley, R. A. (1997). Behavioral inhibition, sustained at-
tention, and executive functions: Constructing a unifying
theory of ADHD. Psychological Bulletin, 121, 65–94.
Biederman, J., Faraone, S. V., Milberger, S., Curtis, S., Chen,
L., Marrs, A., Ouellette, C., Moore, P., & Spence, T. (1996).
Predictors of persistence and remission of ADHD into
adolescence: Results from a four-year prospective follow-up
study. Journal of the American Academy of Child and
Adolescent Psychiatry, 35, 343–351.
Biederman, J., Faraone, S. V., Taylor, A., Sienna, M.,
Williamson, S., & Fine, C. (1998). Diagnostic continuity
between child and adolescent ADHD: Findings from a
longitudinal clinical sample. Journal of the American Acad-
emy of Child and Adolescent Psychiatry, 37, 305–313.
Gaudet, N. (1995). Do children aged 9 through 11 years
understand the DISC version 2.25 questions? Journal of the
American Academy of Child and Adolescent Psychiatry, 34,
Costello, E. J., Edelbrock, C. S., & Costello, A. J. (1985).
Validity of the NIMH Diagnostic Interview Schedule for
Children: A comparison between psychiatric and pediatric
referrals.Journalof AbnormalChild Psychology, 13, 579–595.
Delis, D. C., Squire, L. R., Bihrle, A., & Massman, P. (1992).
Componential analysis of problem-solving ability: Perform-
ance of patients with frontal lobe damage and amnesic
Dodge, K. A., & Coie, J. D. (1987). Social-information-
processing factors in reactive and proactive aggression in
children’s peer groups. Journal of Personality and Social
Psychology, 53, 1146–1158.
1207 EXECUTIVE FUNCTIONS AND PHYSICAL AGGRESSION
Dodge, K. A., Lochman, J. E., Harnish, J. D., Bates, J. E., &
Pettit, G. S. (1997). Reactive and proactive aggression in
school children and psychiatrically impaired chronically
assaultive youth. Journal of Abnormal Psychology, 106,
Edelbrock, C. S., Costello, A. J., Dulcan, M. K., Kalas, R., &
Conover, N. (1985). Age differences in the reliability of the
psychiatric interview of the child. Child Development, 56,
Fisher, P. W., Shaffer, D., Piacentini, J. C., Lapkin, J.,
Kafantaris, V., Leonard, H., & Herzog, D. B. (1993).
2nd ed. (DISC 2.1) for specific diagnoses of children and
adolescents. Journal of the American Academy of Child and
Adolescent Psychiatry, 32, 666–673.
Giancola, P. R. (1995). Evidence for dorsolateral and orbital
behavior. Aggressive Behavior, 21, 431–450.
Giancola, P. R., Mezzich, A. C., & Tarter, R. E. (1998).
behavior conduct-disordered adolescent females. Journal of
Abnormal Psychology, 107, 629–641.
Giancola, P. R., & Zeichner, A. (1994). Neuropsychological
performance on tests of frontal-lobe functioning and ag-
gressive behavior in men. Journal of Abnormal Psychology,
Gray, J. A. (1987). The psychology of fear and stress (2nd ed.).
Cambridge: Cambridge University Press.
Hodges, K. (1993). Structured interviews for assessing children.
Journal of Child Psychology and Psychiatry, 34, 49–68.
Jensen, P. S., Roper, M., Fisher, P. W., Piacentini, J. C.,
Canino, G., Richters, J. E., Rubio-Stipec, M., Dulcan, M.,
Goodman, S., Davies, M., Rae, D., Schaffer, D., Bird, H.,
Lahey, B. B., & Schwab-Stone, M. (1995). Test–retest re-
liability of the Diagnostic Interview Schedule for Children
of General Psychiatry, 52, 61–71.
Kandel,E., Mednick, S. A.,
Hutchings, B., Knop, J., Rosenberg, R., & Schulsinger, F.
(1988). IQ as a protective factor for subjects at high risk for
antisocial behavior. Journal of Consulting and Clinical Psy-
chology, 56, 224–226.
Lau, M. A., Pihl, R. O., & Peterson, J. B. (1995). Provocation,
acute alcohol intoxication, cognitive performance, and ag-
gression. Journal of Abnormal Psychology, 104, 150–155.
Lezak, M. D. (1985). Neuropsychological assessment. In
J. A. M. Frederiks (Ed.), Handbook of clinical neurology,
Vol. 1 (45): Clinical neuropsychology (pp. 515–530). New
Luria, A. R., & Homskaya, E. D. (1964). Disturbance in the
regulative role of speech with frontal lobe lesions. In J. M.
Warren & K. Akert (Eds.), The frontal granular cortex and
behavior (pp. 353–371). New York: McGraw-Hill.
Lynam,D. R.,Moffitt,T. E.,&Stouthamer-Loeber,M.(1993).
Explaining the relation between IQ and delinquency: Class,
race, test motivation, school failure, or self-control? Journal
of Abnormal Psychology, 102, 187–196.
McGee, R., Williams, S., Moffitt, T. E., & Anderson, J. (1989).
A comparison of 13-year-old boys with attention deficit
and?or reading disorder on neuropsychological measures.
Journal of Abnormal Child Psychology, 17, 37–53.
lobe lesion in man. Trends in Neuroscience, 7, 403–407.
Milner, B., Petrides, M., & Smith, M. L. (1985). Frontal lobes
and the temporal organization of memory. Human Neuro-
biology, 4, 137–142.
Newman, J. P. (1998). Psychopathic behavior: An information
processing perspective. In D. J. Cooke, A. E. Forth, & R. D.
Hare (Eds.), Psychopathy: Theory, research, and implications
for society (pp. 81–104). Dordrecht, The Netherlands:
Newman, J. P., Kosson, D. S., & Patterson, C. M. (1992).
Delay of gratification in psychopathic and nonpsychopathic
offenders. Journal of Abnormal Psychology, 101, 630–636.
Offord, D. R., Boyle, M. H., & Racine, Y. (1989). Ontario
Child Health study: Correlates of disorder. Journal of the
American Academy of Child and Adolescent Psychiatry, 28,
Patterson, C. M., & Newman, J. P. (1993). Reflectivity and
learning from aversive events: Toward a psychological
mechanism for the syndromes of disinhibition. Psychological
Review, 100, 716–736.
Pennington, B. F., & Ozonoff, S. (1996). Executive functions
and developmental psychopathology. Journal of Child Psy-
chology and Psychiatry, 37, 51–87.
tasks after frontal- and temporal-lobe lesions in man.
Neuropsychologia, 23, 601–614.
Petrides, M. (1990). Nonspatial conditional learning impaired
in patients with unilateral frontal but not unilateral temporal
lobe excisions. Neuropsychologia, 28, 137–149.
Petrides, M. (1995). Functional organization of the human
frontal cortex for mnemonic processing. Evidence from
neuroimaging studies. Annals of the New York Academy of
Sciences, 769, 85–96.
Petrides, M., Alivisatos, B., & Evans, A. C. (1995). Functional
activation of the human ventrolateral frontal cortex during
mnemonic retrieval of verbal information. Proceedings of the
Petrides, M., Alivisatos, B., Evans, A. C., & Meyer, E. (1993a).
Dissociation of human mid-dorsolateral from posterior
Proceedingsofthe NationalAcademyof Sciencesofthe United
States of America, 90, 873–877.
Petrides, M., Alivisatos, B., Meyer, E., & Evans, A. C. (1993b).
Functional activation of the human frontal cortex during the
performanceofverbalworking memory tasks. Proceedings of
the National Academy of Sciences of the United States of
America, 90, 878–882.
Petrides, M., & Milner, B. (1982). Deficits on subject-ordered
tasks after frontal- and temporal-lobe lesions in man.
Neuropsychologia, 20, 249–262.
Pihl, R. O., & Peterson, J. B. (1991). Attention-deficit hyper-
activity disorder, childhood conduct disorder, and alcohol-
ism: Is there an association? Alcohol Health and Research
World, 15, 25–31.
Roberts, R. E., Solovitz, B. L., Chen, Y.-W., & Casat, C.
(1996). Retest stability of DSM-III-R diagnoses among
adolescents using the Diagnostic Interview for Children
(DISC-2.1C). Journal of Abnormal Child Psychology, 24,
Sattler, J. M. (1988). Assessment of children (3rd ed.). San
Diego, CA: Sattler.
Schwab-Stone, M. E. (1995). Discussion of: ‘‘Do children aged
9 through 11 years understand the DISC version 2.25
questions?’’. Journal of the American Academy of Child and
Adolescent Psychiatry, 34, 954–956.
Schwab-Stone, M. E., Fallon, T., Briggs, M., & Crowther, B.
(1994). Reliability of diagnostic reporting for children aged
6–11 years: A test–retest study of the Diagnostic Interview
Schedule for Children-Revised. American Journal of Psy-
chiatry, 151, 1048–1054.
Schwab-Stone, M. E., Fisher, P. W., Piacentini, J. C., Shaffer,
D., Davies, M., & Briggs, M. (1993). The Diagnostic
Interview Schedule for Children-Revised version (DISC-R):
1208 J. R. SE?GUIN et al. Download full-text
II. Test–retest reliability. Journal of the American Academy of
Child and Adolescent Psychiatry, 32, 651–657.
Se?guin, J. R., Arseneault, L., Boulerice, B., & Tremblay, R. E.
(1999). Response perseveration in adolescent boys with a
history of physical aggression: A test of the response modu-
lation hypothesis. Manuscript submitted for publication.
Se?guin, J. R., Pihl, R. O., Boulerice, B., Tremblay, R. E., &
Harden, P. W. (1996). Pain sensitivity and stability of
physical aggression in boys. Journal of Child Psychology and
Psychiatry, 37, 823–834.
Se?guin, J. R., Pihl, R. O., Harden, P. W., Tremblay, R. E., &
Boulerice, B. (1995). Cognitive and neuropsychological
characteristics of physically aggressive boys. Journal of
Abnormal Psychology, 104, 614–624.
Shaffer, D. (1994). Debate and argument: Structured inter-
views for assessing children. Journal of Child Psychology
and Psychiatry, 35, 783–784.
Spreen, O., Risser, A. H., & Edgell, D. (1995). Developmental
neuropsychology. New York: Oxford.
Squire, L. R., & Zola-Morgan, S. M. (1991). The medial
temporal lobe memory system. Science, 253, 1380–1386.
Stuss, D. T. (1992). Biological and psychological development
of executive functions. Brain and Cognition, 20, 8–23.
Stuss, D. T., & Benson, D. F. (1986). The frontal lobes. New
Tabachnick, B. G, & Fidell, L. S. (1989). Using multivariate
statistics (2nd ed.). New York: Harper Collins.
Tremblay, R. E., Loeber, R., Gagnon, C., Charlebois, P.,
Larive?e, S., & LeBlanc, M. (1991). Disruptive boys with
stable and unstable high fighting behavior patterns during
junior elementary school. Journal of Abnormal Child Psy-
chology, 19, 285–300.
Valla, J.-P., Breton, J.-J., Bergeron, L., Berthiaume, C., Saint-
L., & Le?pine, S. (1994). Enque?te que?be?coise sur la sante?
mentale des jeunes de 6 a ? 14 ans 1992: Rapport de synthe ? se
[Quebec mental health survey of youth aged 6 to 14 years old
1992: Synthesis report]. Quebec: Ho ? pital Rivie ? re-des-
Prairies et Sante? Que?bec, en collaboration avec le Ministe ? re
Wechsler, D. (1974). Manual for the Wechsler Intelligence Scale
for Children-Revised. San Antonio, TX: The Psychological
Wechsler, D. (1987). Wechsler Memory Scale-Revised. New
York: Psychological Corporation.
Welsh, M. C., & Pennington, B. F. (1988). Assessing frontal
lobe functioning in children: Views from developmental
psychology. Developmental Neuropsychology, 4, 199–230.
White, J. L., Moffitt, T. E., & Silva, P. A. (1989). A prospective
replication of the protective effects of IQ in subjects at high
risk for juvenile delinquency. Journal of Consulting and
Clinical Psychology, 57, 719–724.
Wiegersma, S., van der Scheer, E., & Human, R. (1990).
Subjective ordering, short-term memory, and the frontal
lobes. Neuropsychologia, 28, 95–98.
Zelazo, P. D., Carter, A., Reznick, J. S., & Frye, D. (1997).
Early development of executive function: A problem-solving
framework. Review of General Psychology, 1, 198–226.
Manuscript accepted 1 June 1999