Clinical Neurophysiology 113 (2002) 579585

www.elsevier.com/locate/clinph

EEG coherence in attention-deficit/hyperactivity disorder:

a comparative study of two DSM-IV types
Robert J. Barry a,*, Adam R. Clarke a, Rory McCarthy b, Mark Selikowitz b
a
Brain & Behaviour Research Institute and Department of Psychology, University of Wollongong, Wollongong 2522, Australia
b
Private Paediatric Practice, Sydney, Australia
Accepted 23 January 2002

Abstract
Objectives: This study investigated differences in intrahemispheric and interhemispheric electroencephalographic (EEG) coherences
between attention-deficit/hyperactivity disorder (ADHD) and control children, and between children with the Combined (ADHDcom) and
Inattentive (ADHDin) types of ADHD.
Methods: Three age- and sex-matched groups of 40 children, aged 812 years, diagnosed with ADHDcom, ADHDin, and normal control
children, participated in this study. EEG was recorded from 21 sites during an eyes-closed resting condition and Fourier transformed. Wave-
shape coherence was calculated for 8 intrahemispheric electrode pairs (4 in each hemisphere), and 8 interhemispheric electrode pairs, within
each of the delta, theta, alpha and beta bands.
Results: At shorter inter-electrode distances, ADHD children had elevated intrahemispheric coherences in the theta band and reduced
lateral differences in the theta and alpha bands. At longer inter-electrode distances, ADHD children had lower intrahemispheric alpha
coherences than controls. Frontally, ADHD children had interhemispheric coherences elevated in the delta and theta bands, and reduced in
the alpha band. An alpha coherence reduction in temporal regions, and a theta coherence enhancement in central/parietal/occipital regions,
were also apparent. ADHDcom had greater intrahemispheric theta and beta coherences than ADHDin. Frontally, ADHDcom had higher
levels of interhemispheric coherences than ADHDin for the delta and theta bands. In central/parietal/occipital regions, beta coherences were
elevated in ADHDcom.
Conclusions: EEG coherences suggest reduced cortical differentiation and specialisation in ADHD, particularly in cortico-cortical circuits
involving theta activity. Generally, ADHDcom children displayed greater anomalies than ADHDin children. q 2002 Elsevier Science
Ireland Ltd. All rights reserved.
Keywords: Attention deficit/hyperactivity disorder; Children; Electroencephalography; Coherence; DSM-IV types

1. Introduction ods of both synaptic proliferation and pruning. These

Both cognition and behaviour depend on the integrated
activity of different brain regions, and hence study of the
coupling between regions would seem useful in understand-
ing both normal brain function and the important processes
involved in a range of brain dysfunctions. Most electroen-
cephalographic (EEG) studies utilise power estimates
within the traditional frequency bands, providing little infor-
mation about the coupling of brain activity between differ-
ent recording sites. The coherence of the EEG activity
between two sites, conceptualised as the correlation in the
time domain between two signals in a given frequency band
(Shaw, 1981), provides just that information. Normal brain
development from birth to the pre-adult years involves peri-
* Corresponding author. Tel./fax: 161-2-4221-4421.
E-mail address: robert_barry@uow.edu.au (R.J. Barry).
processes presumably underlie the observed systematic
waxing and waning in coherence levels (e.g. Thatcher et
al., 1987; Thatcher, 1994). From about 4 to 6 years, a growth
spurt mainly involves increasing coherence in the frontal
regions and left frontaloccipital coupling. From 8 to 10,
another spurt involves fronto-temporal connections in the
right hemisphere. Further spurts occur from about 11 to 14
years and from 15 years to adulthood, and are interpreted as
reflecting the sequencing of development of different anato-
mical systems. This cyclic pattern complicates the interpre-
tation of observed changes or differences in coherence.
Gasser et al. (1987) investigated coherence at rest and in a
visual-matching task in normal and mildly retarded children
aged 1013 years. Coherences were generally higher and
more variable in the mentally retarded group than in normal
children. There were only slight age effects, with small

1388-2457/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved. CLINPH 2001168
PII: S 1388-245 7(02)00036-6
580 R.J. Barry et al. / Clinical Neurophysiology 113 (2002) 579585
increases in coherence with age in the normals. In younger
(3 months to 7 years) mentally retarded children (mostly
mildly and moderately retarded), Shibagaki et al. (1982)
were unable to find evidence that coherence values were
age-related. Marosi et al. (1995) examined EEG coherences
in 3 groups of children differing on readingwriting ability.
Generally, poor performance was associated with higher
coherences in the delta, theta and beta bands, and reduced
coherences in the alpha band. In older children, these differ-
ences were reduced, particularly in the theta, alpha and beta
bands. In a follow-up study of these same children over a 2
3 year interval (Marosi et al., 1997), group differences
remained, and a general increase in coherence was noted,
except in the theta band, with most changes occurring in the
alpha band.
Thatcher et al. (1986) advanced a two-process model of
cortico-cortical associations in which short and long neuro-
nal fibres contribute differentially to coherence as a function
of inter-electrode distance. At longer distances, coherence is
mainly dependent on the longer fibres alone, increases with
their density/development, and falls off systematically with
increasing inter-electrode distance. In contrast, increased
density/development of short fibres in specialised neuronal
populations reduces coherence by increasing the complexity
and competition of interactions within the cell population.
This two-compartment model appears to accommodate
much of the existing coherence data, and has wide currency
in the literature. For example, the increased coherence
reported in children with intellectual impairment and read-
ing disability (Gasser et al., 1987; Marosi et al., 1995), may
be understood as reflecting decreased cortical differentiation
compared with normal controls.
One of the most common disabilities of childhood listed
in the DSM-IV (APA, 1994) is attention-deficit/hyperactiv-
ity disorder (ADHD). There are numerous EEG studies of
ADHD using absolute and relative power estimates, mean
frequencies from the traditional frequency bands, and power
ratios (e.g. Satterfield et al., 1972; Dykman et al., 1982;
Callaway et al., 1983; Matousek et al., 1984; Lubar et al.,
1985; Lubar, 1991; Mann et al., 1992; Janzen et al., 1995;
Chabot and Serfontein, 1996; Lazzaro et al., 1998). Work
from our laboratory has confirmed that ADHD subjects have
increased levels of absolute and relative theta, and a
decrease in posterior relative beta and alpha (Clarke et al.,
1998, 2001b,c; Bresnahan et al., 1999). Children with
ADHD were also found to have higher delta/theta, theta/
alpha and theta/beta ratios, higher mean frequency in the
delta band and lower mean frequencies in the alpha and beta
bands (Clarke et al., 2001c). In our studies, children with the
Inattentive type of ADHD (ADHDin) had some EEG
abnormalities that were similar to those with the Combined
type (ADHDcom), but less extreme, while topographic
differences suggested that the Combined type had abnorm-
alities in the frontal regions that were not present in the
Inattentive type. Further investigation of maturational
changes indicated that there was an EEG component asso-
ciated with hyperactivity/impulsivity that matured with age
and an inattentive component that was more pervasive
(Clarke et al., 2001a).
In contrast to these studies, EEG coherence has not been
investigated thoroughly in the ADHD context. An early
study of hyperkinetic children by Montagu (1975) examined
interhemispheric and intrahemispheric (right hemisphere
only) coherences in 2 Hz bands (to 10 Hz) relative to normal
children. Interhemispheric coherences were slightly (non-
significantly) reduced in the hyperkinetic children, while
intrahemispheric coherences were generally significantly
elevated (the 10 Hz elevation was not significant). Chabot
and Serfontein (1996) and Chabot et al. (1996) reported on a
mixed group of attention disorder patients (43.9% ADHD,
40.5% ADD by DSM-III criteria, and 15.6% not meeting
those criteria) aged 617 years, some of whom had learning
disabilities. They found that attention disorders were asso-
ciated with interhemispheric and intrahemispheric hyperco-
herence in frontal and central regions, based on comparisons
with a normal control group from the John et al. (1980)
database. There was also generally reduced coherence parie-
tally. Chabot et al. (1999) reported pre-medication coher-
ence data from a similar patient mix, using a subset of
patients from their previous studies, and again noted
increased frontal interhemispheric coherence, particularly
in the theta and alpha bands, and increased intrahemispheric
coherence bilaterally in fronto-temporal regions. Unfortu-
nately, many of the coherence differences noted in these
reports do not specify the frequency ranges involved.
This study examined interhemispheric and intrahemi-
spheric coherences in the standard EEG frequency bands
in two types of ADHD children, and age-matched normal
controls. The aim was to investigate the usefulness of coher-
ence measures in clarifying brain function in ADHD, and to
examine the nature of the EEG differences between the two
most common DSM-IV types of the disorder.
2. Methods
2.1. Subjects
Three groups of 40 children (32 boys and 8 girls, repre-
senting the approximate gender ratio in ADHD) partici-
pated. The EEG power data from these children were
presented in Clarke et al. (2001c). All were aged 812
years, were right handed and footed, and had a full-scale
WISC-III IQ score of 85 or higher. The groups used were
children diagnosed with ADHDcom or ADHDin and a
control group. Both clinical groups of children were
drawn from new patients presenting at a Sydney-based
paediatric practice for an assessment of ADHD. The
ADHD subjects had not been diagnosed as having ADHD
previously, had no history of medication use for the disor-
der, and were tested before being prescribed any medica-
tion.
 R.J. Barry et al. / Clinical Neurophysiology 113 (2002) 579585
Inclusion in the ADHD groups was based on clinical
assessments by a paediatrician and a psychologist; children
were included only when both agreed on the diagnosis.
DSM-IV criteria were used and children were included
only if they met the full diagnostic criteria for the diagnosis
of ADHD, Combined type, or ADHD, Inattentive type. A
structured clinical interview was used which incorporated
information from as many sources as were available. The
interview included a description of the presenting problem
and a medical history given by a parent or guardian, a
physical examination, assessment for neurological soft
signs, review of school reports for the past 12 months seek-
ing behavioural/learning problems, reports from any other
health professionals, and behavioural observations during
the assessment. Children were also assessed using the
WISC-III, Neale Analysis of Reading and the Wide Range
Achievement Test (WRAT) spelling test. Children were
excluded from the clinical groups if they had a history of
a problematic prenatal, perinatal or neonatal period, a disor-
der of consciousness, a head injury with cerebral symptoms,
a history of central nervous system diseases, convulsions or
a history of convulsive disorders, paroxysmal headaches or
tics.
The control group consisted of children from local
schools and community groups. Inclusion was based on:
an uneventful prenatal, perinatal and neonatal period; no
disorders of consciousness, head injury with cerebral symp-
toms, history of central nervous system diseases, obvious
somatic diseases, convulsions, history of convulsive disor-
ders, paroxysmal headache, enuresis or encopresis after the
4th birthday, tics, stuttering, pavor nocturnes or excessive
nailbiting, obvious mental diseases, conduct disorders, and
no deviation with regard to mental and physical develop-
ment. Control subjects had to also score in the normal range
on the measures of accuracy and comprehension on the
Neale Analysis of Reading, and have a standard score of
90 or above on the WRAT spelling test. Assessment for
inclusion as a control was based on a clinical interview
with a parent or guardian similar to that of the ADHD
subjects, utilising the same sources of information, and
the same psychometric assessment as was used for the clin-
ical subjects.
Any child who showed signs of depression, anxiety,
oppositional behaviour or syndromal disorders was
excluded from this study. Children were also excluded if
spike wave activity was present in the EEG.
2.2. Procedures
All subjects were tested in a single session lasting
approximately 2.5 h. Subjects were first assessed by a
paediatrician (physical examination and clinical history),
then had a psychometric assessment consisting of a
WISC-III, Neale Analysis of Reading and WRAT-R spel-
ling, followed by an electrophysiological assessment
consisting of evoked potentials and an EEG. The EEG
581
was recorded in an eyes-closed resting condition, with
subjects seated on a reclining chair. Electrode placement
followed the international 1020 system, using an electrode
cap. A single electro-oculogram (EOG) electrode refer-
enced to Fpz was placed beside the right eye and a ground
lead was placed on the left cheek. A linked ear reference
was used with all EEG. Impedance levels were set at less
than 5 kV. The EEG was recorded and Fourier transformed
by a Cadwell Spectrum 32, software version 4.22, using test
type EEG, montage Q-EEG. The sensitivity was set at
150 mV/cm, low frequency filter 0.53 Hz, high frequency
filter 70 and 50 Hz notch filter. The sampling rate of the
EEG was 200 Hz and the Fourier transformation used
2.5 s epochs.
Thirty 2.5 s epochs were selected from the live trace and
stored to floppy disk. Epoch rejection was based on both
visual and computer selection. Computer reject levels were
set using a template recorded at the beginning of the session
and all subsequent epochs were compared to this. The EOG
rejection was set at 50 mV. The technician also visually
appraised every epoch and decided to accept or reject it.
These were further reduced to 24 epochs (1 min) for Fourier
analysis by a second technician. The EEG was analysed in 4
frequency bands: delta (1.53.5 Hz), theta (3.57.5 Hz),
alpha (7.512.5 Hz) and beta (12.525 Hz). Coherence esti-
mates were derived for each band for 8 intrahemispheric
(F3O1, F4O2, FP1F3, FP2F4, T3T5, T4T6, C3P3,
C4P4) and 8 interhemispheric (FP1FP2, F7F8, F3F4,
C3C4, T3T4, T5T6, P3P4, O1O2) electrode pairs.
2.3. Statistical analysis
Analyses of variance were used to examine the effects of
region and group upon coherences in each band. Prior to
analysis, each coherence value was transformed using Fish-
ers z-transform. For the intrahemispheric coherences, the
means within hemisphere were separately compared for (i)
short/medium inter-electrode distances (left: FP1F3, T3
T5, C3P3 versus right: FP2F4, T4T6, C4P4) and (ii)
long inter-electrode distances (left: F3O1 versus right: F4
O2), and within these analyses, laterality was examined 1.
The interhemispheric coherences were separately examined
within (iii) the frontal (FP1FP2, F7F8, F3F4), (iv)
temporal (T3T4, T5T6) and (v) central/parietal/occipital
(C3C4, P3P4, O1O2) regions. Within the Group factor,
planned contrasts compared the patient groups with the
control group (to establish ADHD differences from
normals) and the ADHDcom group with the ADHDin
group (to investigate type differences). As all these contrasts
are planned, and there are no more of them than the degrees
1
Because volume conduction may artificially inflate EEG coherence
more at short inter-electrode distances than at long distances, no compar-
isons involving distance as a variable were made. However, any such
coherence inflation due to volume conduction would not be expected to
affect laterality or group differences at either long or short inter-electrode
distance.
582 R.J. Barry et al. / Clinical Neurophysiology 113 (2002) 579585

Table 1 the theta band (F 8:77, P , 0:005), was not evident in

Mean ages and IQ scores for the ADHDcom, ADHDin and control groups
Mean Control ADHDcom
group group
Age (months) 124.0 125.0
Full scale IQ (WISC-III) 110.7 96.7
of freedom for effect, no Bonferroni-type adjustment to a is
required (Tabachnick and Fidell, 1989). Unless otherwise
specified, all F values reported have (1,117) degrees of free-
dom.
3. Results
As shown in Table 1, the groups did not differ on age.
Across types, the ADHD children had a significantly lower
mean IQ than the control group (F1; 118 50:36,
P , 0:001), but the types did not differ on IQ.
The mean coherences from each pair of electrodes are
shown in Table 2 for each group within each frequency
band. These data were grouped into regions and analysed
as outlined above.
3.1. Intrahemispheric coherences
With short-medium inter-electrode distances, there were
higher coherences in the left hemisphere in the delta
(F 215:92, P , 0:001), theta (F 8:31, P , 0:005),
alpha (F 587:72, P , 0:001) and beta (F 343:50,
P , 0:001) bands. These, and all other effects, are
summarised in Table 3. Across hemispheres, the ADHD
groups had higher coherences than the controls in the
delta band, but the difference only approached significance
(F 3:57, P 0:061). This difference was significant in
alpha, and approached significance in the beta band
ADHDin (F 3:16, P 0:078). The laterality effect was reduced
group in ADHD groups compared with controls in the theta
(F 8:31, P , 0:005) and alpha (F 6:19, P , 0:05)
124.5
bands. The ADHD types differed in intrahemispheric coher-
96.4
ences, with ADHDcom showing greater coherences than
ADHD in children in the theta (F 9:20, P , 0:005) and
beta bands (F 11:93, P , 0:001). With the long inter-
electrode distances, the ADHD groups demonstrated lower
coherences than the controls in the alpha band (F 4:60,
P , 0:05). Few topographic effects were apparent, although
there was a suggestion of higher coherences in the left hemi-
sphere in the beta band (F 2:83, P 0:095). This effect
was somewhat smaller in the ADHDcom than ADHDin
group (F 3:21, P 0:076).
3.2. Interhemispheric coherences
In the frontal regions, the ADHD groups displayed
greater coherences than controls in the delta (F 4:23,
P , :05) and theta (F 19:74, P , 0:001) bands, and
lower coherences in the alpha band (F 4:50, P , 0:05).
The ADHDcom had higher levels of coherence than
ADHDin in the delta (F 4:09, P , 0:05) and theta
bands (F 7:47, P , 0:01). In temporal regions, the only
group effect was that alpha coherences were lower in ADHD
children than in controls (F 7:80, P , 0:01). In the
central/parietal/occipital regions, interhemispheric coher-
ences in the delta band were somewhat higher in ADHD
children than controls (F 3:12, P 0:080). In the theta
band, this effect was significant (F 9:62, P , 0:005). In
the beta band, the ADHDcom group had higher coherences
than the ADHDin group (F 4:40, P , 0:05).

Table 2
Mean coherence level across subjects for each electrode pair a

Group Delta Theta Alpha Beta

Control ADHDin ADHDcom Control ADHDin ADHDcom Control ADHDin ADHDcom Control ADHDin ADHDcom

FP1FP2 0.882 0.893 0.910 0.872 0.897 0.915 0.893 0.886 0.879 0.726 0.731 0.721
F7F8 0.375 0.364 0.424 0.352 0.409 0.476 0.493 0.436 0.417 0.202 0.140 0.193
F3F4 0.776 0.789 0.797 0.770 0.796 0.827 0.805 0.783 0.772 0.603 0.583 0.605
C3C4 0.707 0.733 0.754 0.675 0.717 0.747 0.588 0.576 0.591 0.518 0.504 0.539
T3T4 0.210 0.183 0.227 0.155 0.144 0.202 0.210 0.175 0.157 0.113 0.118 0.121
T5T6 0.425 0.418 0.439 0.270 0.327 0.347 0.303 0.241 0.220 0.142 0.130 0.170
P3P4 0.730 0.753 0.768 0.705 0.736 0.750 0.612 0.592 0.613 0.553 0.544 0.587
O1O2 0.804 0.817 0.803 0.744 0.776 0.782 0.701 0.679 0.688 0.658 0.652 0.668
FP1F3 0.743 0.743 0.755 0.792 0.809 0.837 0.877 0.844 0.839 0.714 0.685 0.714
FP2F4 0.719 0.744 0.755 0.771 0.809 0.845 0.854 0.838 0.843 0.656 0.670 0.706
T3T5 0.585 0.590 0.593 0.596 0.579 0.607 0.563 0.580 0.596 0.468 0.459 0.529
T4T6 0.552 0.546 0.588 0.575 0.552 0.601 0.552 0.555 0.583 0.436 0.472 0.531
C3P3 0.768 0.778 0.785 0.777 0.786 0.806 0.706 0.721 0.735 0.692 0.681 0.726
C4P4 0.763 0.778 0.783 0.773 0.784 0.799 0.697 0.721 0.737 0.694 0.693 0.721
F3O1 0.217 0.182 0.189 0.221 0.212 0.224 0.231 0.206 0.177 0.161 0.130 0.146
F4O2 0.213 0.179 0.197 0.214 0.199 0.218 0.255 0.202 0.180 0.153 0.111 0.149
a
Data are shown for each frequency band for each group.
 R.J. Barry et al. / Clinical Neurophysiology 113 (2002) 579585 583

Table 3
Summary of obtained effects a

Effect Frequency band

Delta Theta Alpha Beta

Intrahemispheric coherences
(i) Short-medium
L vs. R L " **** L " *** L " **** L " ****
ADHD vs. controls ADHD " 0.061 ADHD " *** ADHD " 0.078
ADHDcom vs. ADHDin ADHDcom " *** ADHDcom " ****
L vs. R ADHD vs. controls ADHD # *** ADHD # *
L vs. R ADHDcom vs. ADHDin
(ii) Long
L vs. R L " 0.095
ADHD vs. controls ADHD # *
ADHDcom vs. ADHDin
L vs. R ADHD vs. controls
L vs. R ADHDcom vs. ADHDin ADHDcom # 0.076

Interhemispheric coherences
(iii) Frontal
ADHD vs. controls ADHD " * ADHD " **** ADHD # *
ADHDcom vs. ADHDin ADHDcom " * ADHDcom " **
(iv) Temporal
ADHD vs. controls ADHD # **
ADHDcom vs. ADHDin
(v) Central/parietal/occipital
ADHD vs. controls ADHD " 0.080 ADHD " ***
ADHDcom vs. ADHDin ADHD " com *
a
" , significantly increased; # , significantly decreased; *P , 0:05, **P , 0:01, ***P , 0:005, ****P , 0:001; probability levels are shown for effects
approaching significance.

3.3. Effects of IQ
Because of the significant IQ difference between the
control group and the two ADHD groups, further analyses
were carried out to investigate IQ effects upon coherence.
The control group was divided into two groups designated
as low IQ (containing children with IQs below the mean of
110.7) and high IQ (children with IQs above the mean).
These groups did not differ on age but had a significant
difference of 19 IQ points (F1; 38 32:62, P , 0:001).
The EEG coherences of these groups were examined using
the same planned contrasts (i)(v) as for the other group
comparisons. The only effect or interaction involving IQ
which approached significance was found in the intrahemi-
spheric comparison involving long inter-electrode distance.
The high IQ group appeared to have somewhat lower theta
coherence than the low IQ group (F1; 38 3:86,
P 0:087).
4. Discussion
The intrahemispheric coherences from short-medium
inter-electrode distances showed significant lateralisation,
pointing to the existence of substantial hemispheric differ-
entiation. The higher levels noted in all frequency bands in
the left hemisphere suggests that general differentiation has
progressed further in the right hemisphere than the left.
Although this was not expected from previous suggestions
of coherences being larger in the right than left hemisphere
(e.g. Thatcher et al., 1986; Tucker et al., 1986), it is unclear
from the literature what affect the age distribution of the
sample might have had in this regard, indicating the need
for further developmental data. The short-distance laterality
effect was reduced significantly in the theta and alpha bands
for ADHD children, suggesting that the development of
hemispheric differentiation, at least for neuronal circuits
involving theta and alpha activity, is reduced in children
with ADHD. Few lateralisation differences were apparent
at large inter-electrode distances, with a suggestion of
higher coherence on the left only in the beta band. This
independence between effects at short and long inter-elec-
trode distance supports Thatchers two-compartment model
of coherence (Thatcher et al., 1986).
With shorter inter-electrode distances, the ADHD groups
had significantly higher levels of intrahemispheric coher-
ence in the theta band, and the difference approached signif-
icance in the delta and alpha bands. These results are
broadly compatible with Montagu (1975). Chabot and
Serfontein (1996) also reported increased frontal/temporal
584 R.J. Barry et al. / Clinical Neurophysiology 113 (2002) 579585
coherence in ADHD, without specifying the frequency
bands involved, while Chabot et al. (1996) indicates that a
major variable emerging in their discriminant analysis was
elevated frontal coherence in the theta band. Chabot et al.
(1999) also mention frontal theta/alpha hypercoherence in
ADHD. The present data suggests that the ADHD groups
have less cortical differentiation than controls, particularly
in circuits involving theta activity. Although the group
differences in delta and beta coherences suggested this
effect extended to neuronal circuits beyond those involved
with theta activity, no difference was apparent here in the
alpha band. The results for long inter-electrode distances
were in marked contrast to the group differences noted at
shorter inter-electrode distances, with the ADHD groups
showing lower coherences in the alpha band than age-
matched controls. Chabot and Serfontein (1996) reported
reduced coherence from these electrodes in their ADHD
children, but with no frequency bands specified. If we inter-
pret our finding in terms of Thatchers two-compartment
model of coherence (Thatcher et al., 1986), our results
would imply that the long-distance neuronal circuits invol-
ving alpha activity are less differentiated in ADHD than in
age-matched controls.
Interhemispheric coherences in the frontal regions were
greater in the ADHD groups than controls for both delta and
theta activity. In the central/parietal/occipital regions,
greater coherences for the ADHD groups were also found
in theta activity, and this difference approached significance
for the delta band. Again, these data are compatible with
differences reported by Chabot and Serfontein (1996)
marked frontal and posterior hypercoherence (suggested in
Chabot et al., 1996 as particularly involving theta activity, at
least in the frontal regions) and moderate central and parie-
tal incoherence. The present findings signal widespread
deficiencies in cortical specialisation in ADHD, at least
for neuronal circuits involving slow wave activity. In the
frontal and temporal regions, the ADHD groups had lower
levels of alpha coherence than controls. Together, these data
indicate that ADHD children show atypical cortical differ-
entiation and specialisation in circuits involving both slow
wave (delta and theta) and alpha activity, although the direc-
tion of this difference differs with frequency. From Thatch-
ers model (Thatcher et al., 1986), the ADHD children may
be distinguished from controls in terms of deficits in non-
alpha neuronal circuits primarily involving short fibres, and
alpha circuits primarily involving long fibres.
None of these effects appear to be attributable to the IQ
differences between the groups, as data from the control
group indicated that higher IQ was associated only with a
non-significant reduction in theta coherence for the intrahe-
mispheric comparison involving long inter-electrode
distance. No effects were found with this analysis in the
main ADHDcom/ADHDin/control comparisons.
There were a number of intrahemispheric effects suggest-
ing differences in the cortical functioning of the different
ADHD types. At shorter inter-electrode distances,
ADHDcom children had higher coherence levels than
ADHDin for the theta and beta bands. ADHDcom children
also had higher levels of interhemispheric coherence than
ADHDin children. This was apparent frontally in both slow
wave bands, and in central/parietal/occipital regions in beta
activity. These interhemispheric data are compatible with
the intrahemispheric differences found between the types at
shorter inter-electrode distance. Together, these differences
support the view that ADHDin children suffer similar, but
less extreme, deviations from normality than ADHDcom
children.
This study found a range of differences in both intrahemi-
spheric and interhemispheric coherences between ADHD
patients and age-matched controls. These suggest reduced
cortical differentiation and specialisation in ADHD, parti-
cularly in cortico-cortical circuits involving theta activity. It
is interesting to note that this band exhibits the most consis-
tent power differences between ADHD children and
controls, with theta power being abnormally elevated in
ADHD (Clarke et al., 1998, 2001c). Disturbance in such
cortico-cortical circuits may be the fundamental dysfunc-
tion underlying the cognitive symptoms of inattention asso-
ciated with ADHD. The interpretation of these effects in
terms of the two dominant theories of ADHD as resulting
from a maturational lag or developmental deviation is
difficult in the absence of clear age-norms for coherences
between different brain regions. The paucity of data in this
field suggests the need for further basic developmental
research. The present data also confirm differences in the
extent of cortical dysfunction between the two most
common types of ADHD. It appears from the present data
that the differences in coherence patterns between the types
reflect a simple difference in the degree of departure from
normality. While this is compatible with some EEG power
studies of these ADHD types (e.g. Chabot and Serfontein,
1996; Clarke et al., 1998), others using a wider array of
measures (e.g., Kuperman et al., 1996; Clarke et al.,
2001c) have noted subtle differences which argue against
a unidimensional view of the types. The present results
indicate the need for further investigations of EEG coher-
ence in regard to both normal development, particularly in
relation to alpha coherences, and the reliability of these
group differences. Together, such studies should aid the
clarification of abnormal cortical functioning in ADHD.
References
American Psychiatric Association. DSM-IV. Diagnostic and statistical
manual of mental disorders, 4th ed. Washington DC: American
Psychiatric Association, 1994.
Bresnahan S, Anderson J, Barry R. Age-related changes in quantitative
EEG in attention deficit disorder. Biol Psychiatry 1999;46:16901697.
Callaway E, Halliday R, Naylor H. Hyperactive childrens event-related
potentials fail to support underarousal and maturational-lag theories.
Arch Gen Psychiatry 1983;40:12431248.
Chabot R, Serfontein G. Quantitative electroencephalographic profiles of
 R.J. Barry et al. / Clinical Neurophysiology 113 (2002) 579585
children with attention deficit disorder. Biol Psychiatry 1996;40:951
963.
Chabot R, Merkin H, Wood L, Davenport T, Serfontein G. Sensitivity and
specificity of QEEG in children with attention deficit or specific devel-
opmental learning disorders. Clin Electroencephalogr 1996;27:2634.
Chabot R, Orgill A, Crawford G, Harris M, Serfontein G. Behavioural and
electrophysiological predictors of treatment response to stimulants in
children with attention disorders. J Child Neurol 1999;14:343351.
Clarke A, Barry R, McCarthy R, Selikowitz M. EEG analysis in attention-
deficit/hyperactivity disorder: a comparative study of two subtypes.
Psychiatry Res 1998;81:1929.
Clarke A, Barry R, McCarthy R, Selikowitz M. Age and sex effects in the
EEG: differences in two subtypes of attention-deficit/hyperactivity
disorder. Clin Neurophysiol 2001a;112:815826.
Clarke A, Barry R, McCarthy R, Selikowitz M. EEG-defined subtypes of
children with attention-deficit/hyperactivity disorder. Clin Neurophy-
siol 2001b;112:20982105.
Clarke A, Barry R, McCarthy R, Selikowitz M. EEG differences in two
subtypes of attention-deficit/hyperactivity disorder. Psychophysiology
2001c;38:212221.
Dykman R, Holcomb P, Oglesby D, Ackerman P. Electrocortical frequen-
cies in hyperactive, learning-disabled, mixed, and normal children. Biol
Psychiatry 1982;17:675685.
Gasser T, Jennen-Steinmetz C, Verleger R. EEG coherence at rest and
during a visual task in two groups of children. Electroenceph clin
Neurophysiol 1987;67:151158.
Janzen T, Graap K, Stephanson S, Marshall W, Fitzsimmons G. Differences
in baseline EEG measures for ADD and normally achieving preadoles-
cent males. Biofeedback Self-Regul 1995;20:6582.
John E, Ahn H, Princhep L, Trepetin M, Brown D, Kaye H. Developmental
equations of the electroencephalogram. Science 1980;210:12551258.
Kuperman S, Johnson B, Arndt S, Lindgren S, Wolraich M. Quantitative
EEG differences in a nonclinical sample of children with ADHD and
undifferentiated ADD. J Am Acad Child Adolesc Psychiatry 1996;35:
10091017.
Lazzaro I, Gordon E, Whitmont S, Plahn M, Li W, Clarke S, Dosen A,
Meares R. Quantified EEG activity in adolescent attention deficit hyper-
activity disorder. Clin Electroencephalogr 1998;29:3742.
Lubar J. Discourse on the development of EEG diagnostics and biofeed-
back for attention-deficit/hyperactivity disorders. Biofeedback Self-
Regul 1991;16:201225.
585
Lubar J, Bianchini K, Calhoun W, Lambert E, Brody Z, Shabsin H. Spectral
analysis of EEG differences between children with and without learning
disabilities. J Learn Disabil 1985;18:403408.
Mann C, Lubar J, Zimmerman A, Miller C, Muenchen R. Quantitative
analysis of EEG in boys with attention deficit hyperactivity disorder:
controlled study with clinical implications. Pediatr Neurol 1992;8:30
36.
Marosi E, Harmony T, Becker J, Reyes A, Bernal J, Fernandez T, Rodri-
guez M, Silva J, Guerrero V. Electroencephalographic coherences
discriminate between children with different pedagogical evaluation.
Int J Psychophysiol 1995;19:2332.
Marosi E, Harmony T, Reyes A, Bernal J, Fernandez T, Guerrero V, Rodri-
guez M, Silva J, Yanez G, Rodriguez H. A follow-up study of EEG
coherences in children with different pedagogical evaluations. Int J
Psychophysiol 1997;25:227235.
Matousek M, Rasmussen P, Gilberg C. EEG frequency analysis in children
with so-called minimal brain dysfunction and related disorders. Adv
Biol Psychiatry 1984;15:102108.
Montagu J. The hyperkinetic child: a behavioural, electrodermal and EEG
investigation. Dev Med Child Neurol 1975;17:299305.
Satterfield J, Cantwell D, Lesser M, Podosin R. Physiological studies of the
hyperkinetic child: 1. Am J Psychiatry 1972;128:103108.
Shaw J. An introduction to the coherence function and its use in EEG signal
analysis. J Med Eng Technol 1981;5:279288.
Shibagaki M, Klyono S, Nakamura K, Fujisawa K. Developmental studies
on EEG spectral analysis in mentally retarded children. Clin Electro-
encephalogr 1982;13:257263.
Tabachnick B, Fidell L. Using multivariate statistics, New York, NY:
Harper Collins, 1989.
Thatcher R. Cyclic cortical reorganisation, origins of human cognitive
development. Dawson G, Fischer K, editors. Human behavior and the
developing brainNew York, NY: The Guildford Press, 1994. pp. 232
266.
Thatcher R, Krause P, Hrybyk M. Cortico-cortical associations and EEG
coherence: a two-compartmental model. Electroenceph clin Neurophy-
siol 1986;64:123143.
Thatcher R, Walker R, Giudice S. Human cerebral hemispheres develop at
different rates and ages. Science 1987;236:11101113.
Tucker D, Roth D, Bair T. Functional connections among cortical regions:
topography of EEG coherence. Electroenceph clin Neurophysiol
1986;63:242250.