Documente Academic
Documente Profesional
Documente Cultură
Background. Despite evidence of inhibitory control and visual processing impairment in attention deficit hyperactivity
disorder (ADHD), knowledge about its corresponding alterations in the brain is still evolving. The current study used
counting Stroop functional MRI and the Cambridge Neuropsychological Test Automated Battery (CANTAB) to investi-
gate if brain activation of inhibitory control and visual processing would differ in youths with ADHD relative to neuro-
typical youths.
Method. We assessed 25 youths with ADHD [mean age 10.9 (S.D. = 2.2) years] and 23 age-, gender- and IQ-matched
neurotypical youths [mean age 11.2 (S.D. = 2.9) years]. The participants were assessed by using the Wechsler
Intelligence Scale for Children, third edition, and two tests from the CANTAB: rapid visual information processing
(RVP) and pattern recognition memory (PRM) outside the scanner.
Results. Youths with ADHD showed more activation than neurotypical youths in the right inferior frontal gyrus
[Brodmann area (BA) 45] and anterior cingulate cortex, which were correlated with poorer performance on the RVP
test in the CANTAB. In contrast, youths with ADHD showed less activation than neurotypical youths in the left superior
parietal lobule (BA 5/7), which was correlated with the percentage of correct responses on the PRM test in the CANTAB.
Conclusions. Our findings suggest that youths with ADHD might need more inhibitory control to suppress interference
between number and meaning and may involve less visual processing to process the numbers in the counting Stroop task
than neurotypical youths.
Received 29 May 2013; Revised 6 December 2013; Accepted 18 December 2013; First published online 22 January 2014
Key words: Cambridge Neuropsychological Test Automated Battery (CANTAB), counting Stroop functional MRI,
inhibitory control, visual processing.
networks is the most frequently observed functional meaning (Bush et al. 1998), and the visual properties
imaging finding in children with ADHD; Nakao et al. of the numbers to evaluate visual processing
(2011) reported abnormal gray matter volumes in pre- (Rasmussen & Bisanz, 2005). In the current study, we
frontal, anterior cingulate cortex (ACC), and parietal used the counting Stroop task to explore the differen-
regions in children with ADHD. According to a tial brain activity related to visual processing load be-
meta-analysis of patients with ADHD (Hart et al. tween the ADHD and neurotypical groups, in
2013), most imaging studies have shown abnormal addition to inhibitory control that has been explored
neural activity for inhibitory control in the right in- in fMRI studies.
ferior frontal gyrus (IFG), ACC, and parietal regions. The Cambridge Neuropsychological Test Auto-
First, impaired IFG functions reduced the ability to mated Battery (CANTAB) was also used to examine
optimally recruit subsidiary brain regions and strate- these two processes. The CANTAB has been used to
gies to perform cognitive tasks in ADHD (Fassbender assess executive functions, including both inhibitory
& Schweitzer, 2006). The IFG has been implicated in control and visual processing. Previous studies have
higher-level cognitive functioning, including atten- shown that children with ADHD perform worse in re-
tional processes, inhibitory control, working memory sponse inhibitory control (Slaats-Willemse et al. 2007;
and planning (Fassbender & Schweitzer, 2006; Bush, Gau et al. 2009; Gau & Shang, 2010a, b; Gau &
2011). Particularly, the right IFG is thought to be Huang, 2014) and visual memory (Rhodes et al. 2005;
involved in response inhibitory control or target detec- Gau et al. 2009; Gau & Shang, 2010a; Shang & Gau,
tion (Durston, 2003; Hampshire et al. 2010; Cortese 2011), as compared with neurotypical children. Pre-
et al. 2012). Second, the ACC is also believed to play vious studies have also suggested that the deficits in
critical roles in complex and effortful cognitive proces- inhibitory control and visual processing are major
sing, target detection, response selection and inhibitory endophenotypes for ADHD genetic studies (Gau &
control, error detection, performance monitoring, and Shang, 2010a; Shang & Gau, 2011). Hence, combining
motivation (as reviewed in Bush, 2011). A number of the CANTAB and counting Stroop fMRI offers a better
studies have implicated that the ACC is involved in understanding of the relationships among these two
top-down attentional control and inhibitory control processes. We used the CANTAB to assess partici-
of competing responses to various stimuli (Pardo pants’ cognitive mechanisms of inhibitory control
et al. 1990). Studies have shown hypoactivity in the and visual processing load outside the scanner because
ACC and robust activity in the right IFG in adults the CANTAB has been reported to be used to measure
with ADHD using the counting Stroop task (Bush the endophenotypes for ADHD (Doyle et al. 2005;
et al. 1999). Third, parietal activations have been Gau & Shang, 2010a; Shang & Gau, 2011; Gau &
reported to be involved in attention and visual proces- Huang, 2014). In addition, previous studies have
sing (Fassbender & Schweitzer, 2006; Schneider et al. shown that the deficits in inhibitory control and visual
2010). Specifically, the superior parietal lobule (SPL) processing are major endophenotypes for ADHD gen-
in the left hemisphere has been suggested to be asso- etic studies (Gau & Shang, 2010a; Shang & Gau, 2011).
ciated with number processing on screen in healthy Despite the striking nature of the inhibitory control
adults (Dehaene et al. 2003; Cavanna & Trimble, and visual processing impairment in ADHD, knowl-
2006). Bush et al. (1998) and Lévesque et al. (2006) edge about its corresponding alterations in the brain
found robust activity in the left SPL/precuneus is still evolving. Previous studies have used the ap-
[Brodmann area (BA) 7] during a counting Stroop task. proach of correlating the structural integrity of fiber
Neuropsychological studies have combined several tracts based on a group difference with the CANTAB
constructs, namely executive functions and visual pro- to investigate executive function and inhibitory
cessing, into a single study to examine drug-related im- control in children with ADHD (Shang et al. 2012;
provement among children with ADHD (Coghill et al. Wu et al. 2014). In the findings of Konrad et al.
2007, Shang & Gau, 2012). Previous imaging studies (2012), adults with ADHD had significantly lower
have also combined inhibitory control and sustained structural integrity in the left inferior longitudinal
attention into a single task using the Go/No-Go task fasciculus based on a group difference, which was
in ADHD (Trommer et al. 1988; Schulz et al. 2004; negatively correlated with attentional performance of
Fassbender & Schweitzer, 2006). They found that com- the test of variables of attention. Therefore, the present
mission errors (failure to inhibit to the no-go response) study was designed to use the approach of correlating
were related to inhibitory control and impulsivity, and brain activity and the CANTAB to assess the differen-
omission errors (failure to respond to the go stimulus) tial inhibitory control and visual processing between
were associated with inattention. The counting Stroop the ADHD and neurotypical groups. Two tests of in-
task is designed to evaluate the underlying mechan- hibitory control and visual processing were chosen
ism of cognitive interference between number and from the CANTAB.
Inhibitory control and visual processing in ADHD 2663
Studies on brain activity in the right IFG and Wechsler, 1991), and had normal hearing and
parietal regions have revealed inconsistent results in normal or corrected-to-normal vision. Participants
terms of increased or decreased activation in ADHD who had a clinical diagnosis of any other psychiatric
(Fassbender & Schweitzer, 2006; Silk et al. 2008; disorders were excluded from the study. In addition,
Bush, 2011; Hart et al. 2013). These studies are participants with ADHD who currently took medi-
mainly limited by separate investigations of these cation affecting the central nervous system, and neuro-
two brain regions and the study of adult populations. typical participants who ever or currently took any
We therefore used counting Stroop fMRI to examine psychotropic drug and had a history of attention or
the neural correlates of inhibitory control and visual verbal-language deficits were excluded from the
processing, and correlated the brain activity with the study.
CANTAB in youths with ADHD and neurotypical The study was approved by the Research Ethics
youths. According to a priori hypotheses, we hypo- Committee at the National Taiwan University
thesized that the involvement of the right IFG and Hospital, Taiwan and all the participants and their
ACC may be related to the inhibitory control that parents provided written informed consent before
can be measured by the CANTAB in youths with study implementation (IRB no. 200903062R;
ADHD, and that the involvement of the parietal ClinicalTrials.gov no. NCT00916851). All the partici-
regions may be related to the visual processing that pants were assessed using the Chinese K-SADS-E
can be measured by the CANTAB in neurotypical interviews, followed by the WISC-III and the
youths. CANTAB. The six participants with ADHD,
who took psychotropic medication before, did not
take any medication for at least 1 week before the
Method assessments.
Participants and procedures
Fig. 1. The counting Stroop task in Chinese. Experimental stimuli were divided into congruent, incongruent and control
conditions. In the ‘congruent’ condition, the number of words (i.e. one) was consistent with the meaning of the word (i.e.
one). In the ‘incongruent’ condition, the number of words (i.e. three) was inconsistent with the meaning of the word (i.e.
four). In the ‘control’ condition, the Chinese words did not give any clue to number.
represent the index of sustained attention and verbal MRI image acquisition
working memory, respectively.
Participants lay in the scanner with their head position
Neuropsychological measures secured. The head coil was positioned over the partici-
pants’ head. Participants viewed visual stimuli pro-
The CANTAB is a standard, computerized, non- jected onto a screen via a mirror attached to the
linguistic and culturally blind test to assess a wide inside of the head coil. Each participant performed
range of executive functions and visual processing. two functional runs. Each run took 2.8 min. Images
Rapid visual information processing (RVP) was chosen were acquired using a 3-T Siemens Tim-Trio scanner
to assess inhibitory control (Sahakian et al. 1989; Gau & with the 32-channel head coil (Siemens Medical
Huang, 2014), while pattern recognition memory Solutions, Germany). Each functional run had 85
(PRM) was used to assess visual processing (Shang & image volumes that were acquired with the echo
Gau, 2011, 2012). planar imaging method to detect the blood oxygen-
In the RVP test, participants were asked to respond ation level-dependent (BOLD) signal.
to the specific sequence of digits, when a white box The scanning parameters were the following: rep-
was presented in the center of the screen with digits etition time (TR) = 2000 ms; echo time (TE) = 24 ms;
(ranging from 2 to 9) appearing one at a time flip angle = 90°; matrix size = 64 × 64; field of view =
(100 digits/min) in the center of the screen in a 25.6 cm; slice thickness = 3 mm; number of slices = 34.
pseudo-random order. Participants were instructed to A high-resolution, T1-weighted three-dimensional
respond to three specific number sequences of digits image was also acquired (magnetization prepared
(i.e. 2–4–6, 3–5–7, 4–6–8) by pressing the touch pad. rapid gradient-echo; TR = 2300 ms; TE = 2.98 ms;
Two measures were recorded: A’, a signal detection flip angle = 9°; matrix size = 256 × 256; field of view =
measure of sensitivity to the target, regardless of 25.6 cm; slice thickness = 1 mm). The orientation of
response tendency (Sahgal, 1987); and probability of the three-dimensional image was identical to the func-
false alarms (of the participant responding inappropri- tional slices. The task was administered in a pseudo-
ately), i.e. total false alarms divided by the sum of total random order for all participants for event-related
false alarms and total correct rejections. design (Burock et al. 1998). We used the Optseq script
The PRM test is designed to measure the capacity of for randomized event-related design (http://surfer.
visual processing in a two-choice forced discrimination nmr.mgh.harvard.edu/optseq, written by D. Greve,
paradigm (Sahakian et al. 1988). Participants were pre- Charlestown, MA, USA) that implemented the
sented with a series of visual geometric patterns, one Burock et al. (1998) approach.
at a time, in the center of the screen. In the recognition
phase, participants were presented with two geometric
Image and statistical analysis
patterns and the test patterns were presented in the
reverse order to the original order of presentation. Data analysis was performed using SPM5 (Statistical
The participants were instructed to choose the correct Parametric Mapping). The functional images were cor-
pattern between an already-seen pattern and a novel rected for the differences in slice-acquisition time to the
pattern. The percentage of correct responses was middle volume and were realigned to the first volume
recorded. in the scanning session using affine transformations.
Inhibitory control and visual processing in ADHD 2665
The exclusion criteria for motion were 3 mm for dis- PRM tests. All reported results were significant using
placement and 3° for rotations. No participant had p < 0.05.
more than 3 mm of movement in any plane. Co-
registered images were normalized to the Montreal
Results
Neurological Institute (MNI) average template.
Statistical analyses were calculated on the smoothed Behavioral results
data (10 mm isotropic Gaussian kernel, the concept of
There were no group differences in IQ profiles, or digit
kernel was defined the shape of function to calculate
span forward. Compared with neurotypical youths,
weighted average of each data point with its neighbor-
youths with ADHD had significantly fewer digits
ing data points), with a high-pass filter (128 s cut-off
recalled backward, and lower scores on A’ (target sen-
period) in order to remove low frequency artifacts.
sitivity), higher probability of false alarm in the RVP
Data from each participant were entered into a
test, and lower percentage of correct responses in the
general linear model using an event-related analysis
PRM test (Table 1).
procedure (Josephs & Henson, 1999). Stimuli were
Online Supplementary Table S1 presents the accu-
treated as individual events for analysis and modeled
racy and reaction time of the counting Stroop test in
using a canonical hemodynamic response function
the three conditions by the two groups. A two-way
(HRF). Parameter estimates from contrasts of the ca-
ANOVA on accuracy revealed significant main effects
nonical HRF in single-subject models were entered
of condition (F2,92 = 14.58, p < 0.01), with being more
into random-effects analysis using the one-sample
accurate in the congruent and control conditions than
t test across all participants to determine whether acti-
the incongruent condition (t47 = 4.91, p < 0.01; t47 = 3.77,
vation during a contrast was significant (i.e. parameter
p < 0.01, respectively), but no significant effect of
estimates were reliably greater than 0) in a whole
group (F1,46 = 1.59, p = 0.214). The interaction of group
brain analysis. There were three event types: congru-
and condition was not significant (F2,92 = 0.37, p = 0.692).
ent, incongruent, and control. To observe the neural
Regarding reaction time, a two-way ANOVA
correlates of inhibitory control, we compared the in-
showed a significant main effect of condition (F2,92 =
congruent with the congruent condition. To further
20.37, p < 0.01), with shorter reaction time in the
observe the neural correlates of visual processing, we
congruent and control conditions than in the incongru-
compared the larger number of words (i.e. ‘three’
ent condition (t47 = −5.33, p < 0.01; t47 = −5.40, p < 0.01,
and ‘four’) with the fewer number of words (i.e. ‘one’
respectively) but a marginally significant effect of
and ‘two’) in the incongruent condition and in the
group (F1,46 = 4.00, p = 0.052). The interaction of group
congruent condition. For the contrast within each
and condition was not significant (F2,92 = 2.85, p = 0.063).
group, all reported areas of activation were significant
To further observe visual processing in the congru-
using p < 0.05 for family-wise error (FWE) corrected at
ent and incongruent conditions, the larger number of
the voxel level with a cluster size greater than 50 voxels
words (i.e. ‘three’ and ‘four’) showed significant longer
in a whole brain analysis. For the contrasts between
reaction time than the fewer number of words (i.e.
groups, all reported areas of activation were significant
‘one’ and ‘two’) (t95 = −2.13, p < 0.05), suggesting that
using p < 0.05 for FWE corrected at the voxel level with
all participants spent more time in making judgments
a cluster size greater than 10 voxels, with the anatom-
as the number of words increased.
ical masks of right IFG, ACC, and left SPL due to our
a priori hypothesis. The three anatomical masks
fMRI results
were defined from the WFU PickAtlas, with the option
of right IFG, ACC, and left SPL (http://fmri.wfubmc. Table 2 presents activation of brain regions for the in-
edu/software/PickAtlas). We then extracted the β congruent versus congruent condition for the ADHD
values from peak voxels of significant brain regions be- and neurotypical groups. For the contrast of ‘incongru-
tween groups. ent versus congruent condition’ within group, the
We used SPSS to conduct statistical analysis ADHD group showed greater activation in the right
(IBM, USA). The descriptive results were displayed IFG (BA 45) and ACC (online Supplementary Fig. S1,
as frequency and percentage for categorical variables, Table 2). For group comparisons, youths with ADHD
and mean and S.D. for continuous variables. We also had greater activation in the right IFG (BA 45) and
conducted two-way analysis of variance (ANOVA) ACC than neurotypical youths (Fig. 2, Table 2). There
with group and condition interactions for behavioral was no significant activation for neurotypical youths
results of the counting Stroop test. Moreover, we per- as compared with youths with ADHD.
formed Pearson’s correlations between β values of For the contrast of ‘larger versus fewer numbers of
peak voxels of significant brain regions and the words’ within group, the ADHD group showed
scores of the parameters derived from the RVP and greater activation in the left postcentral gyrus (BA 3),
2666 L.-Y. Fan et al.
Table 1. Age, IQ scores, and performance in the RVP and PRM tests for all participants
Gender, n
Male 23 21
Female 2 2
Age, years 10.9 (2.2) 11.2 (2.9) 0.667
Range 8–16 8–16
WISC-III
Verbal IQ 109.8 (10.1) 111.1 (7.9) 0.628
Performance IQ 104.0 (11.1) 105.2 (9.5) 0.676
Full-scale IQ 107.2 (10.8) 109.1 (7.5) 0.496
Digit span
Forward 8.0 (1.2) 8.3 (0.7) 0.177
Backward 4.8 (1.8) 5.9 (1.8) 0.034
RVP
A’ 0.83 (0.06) 0.87 (0.08) 0.018
Probability of false alarm 0.025 (0.031) 0.010 (0.010) 0.044
PRM
Percentage of correct responses 90.2 (4.1) 93.0 (4.4) 0.049
IQ, Intelligence quotient; RVP, rapid visual information processing; PRM, pattern recognition memory; ADHD, attention
deficit hyperactivity disorder; WISC-III, Wechsler Intelligence Scale for Children, Third Edition; A’, signal detection measure
of sensitivity to the target.
Data are given as mean (standard deviation).
Table 2. Inhibitory control: greater activation for the incongruent compared with congruent condition within each group, and between the two
groupsa
MNI coordinatesb
MNI, Montreal Neurological Institute; H, hemisphere; BA, Brodmann area; ADHD, attention deficit hyperactivity disorder;
R, right; L, left; FWE, family-wise error.
a
Region of interest masks (anatomical masks of right inferior frontal gyrus, anterior cingulate cortex, and left superior
parietal lobule) were only used in between-group analysis.
b
Coordinates of activation peak(s) within a region based on a Z test are given in the MNI stereotactic space (x, y, z).
c
Number of voxels in cluster at p < 0.05 FWE corrected, only clusters greater than or equal to 50 are presented.
* p < 0.05 for FWE corrected with the use of an anatomical mask.
† p < 0.05 for uncorrected.
Inhibitory control and visual processing in ADHD 2667
Fig. 2. (a) Inhibitory control. Greater activation was found in the right inferior frontal gyrus [IFG; Brodmann area (BA) 45]
and anterior cingulate cortex (ACC) with region of interest (ROI)-masked analysis for the attention deficit hyperactivity
disorder (ADHD) group as compared with the neurotypical group during the contrast of ‘incongruent versus congruent
condition’. (b) Visual processing. Greater activation was found in the left superior parietal lobule (SPL; BA 5/7) with
ROI-masked analysis for the neurotypical group as compared with the ADHD group during the contrast of ‘larger versus
fewer numbers of words’. The color bar presents the strength of the activation.
Table 3. Visual processing: greater activation for the larger number of words compared with the fewer number of words within each group,
and between the two groupsa
MNI coordinatesb
MNI, Montreal Neurological Institute; H, hemisphere; BA, Brodmann area; ADHD, attention deficit hyperactivity disorder;
L, left; FWE, family-wise error.
a
Region of interest masks (anatomical masks of right inferior frontal gyrus, anterior cingulate cortex, and left superior
parietal lobule) were only used in between group analysis.
b
Coordinates of activation peak(s) within a region based on a Z test are given in the MNI stereotactic space (x, y, z).
c
Number of voxels in cluster at p < 0.05 FWE corrected, only clusters greater than or equal to 50 are presented.
d
Subcluster.
* p < 0.05 for FWE corrected with the use of an anatomical mask.
† p < 0.05 for uncorrected.
inhibitory control (Fassbender & Schweitzer, 2006; (Hampshire et al. 2010; Whelan et al. 2012), suggesting
Bush, 2011). Previous fMRI studies have suggested that youths with ADHD might need more activation in
that children and adults with ADHD might engage the right IFG and ACC to compensate their deficits
in alternative, compensatory brain regions with con- in inhibitory control. According to the meta-analysis
comitant cognitive strategies due to a selectively weak- in ADHD (Hart et al. 2013), adult patients with
ened neural system (for a review, see Fassbender & ADHD relative to controls showed decreased acti-
Schweitzer, 2006). They suggested that hyperactivity vation in the right frontal region for inhibitory control.
in a brain region can be considered as ‘inefficiency’ be- However, we found that youth participants with
cause an individual needs to use more energy than ADHD showed greater activations than neurotypical
should be to perform a given task. Moreover, extra ac- youths in the right IFG and ACC. The current study
tivity in the clinical group may be viewed as com- has added value to the effect of age.
pensatory activity of brain regions that the clinical In contrast, youths with ADHD showed less acti-
group is enrolling in order to compensate for under- vation in the left SPL than neurotypical youths.
activity in the ‘appropriate’ brain network Previous studies point out that ADHD is associated
(Fassbender & Schweitzer, 2006). With regards to with impaired visual processing (Shang & Gau, 2011,
ADHD studies, Schulz et al. (2004) have suggested 2012) and less activation in the left parietal region
the recruitment of additionally compensatory pre- may be attributed to impaired visual processing in
frontal regions to perform the Go/No-Go task in children with ADHD (Silk et al. 2008). Our finding of
ADHD. Moreover, an earlier study showed that indivi- a positive correlation between left SPL activation and
duals with ADHD might have an immature prefrontal the percentage of correct responses in the PRM test
cortex, so they may compensate with the recruitment in neurotypical youths suggests that neurotypical
of additional cortical areas in order to perform at the youths might have better visual processing than
same level as that of neurotypical subjects (Rubia youths with ADHD. Taken together, our findings
et al.1999). In our findings, greater activations in the lend important evidence to support that youths with
right IFG and ACC among youths with ADHD than ADHD might have worse visual processing of num-
neurotypical youths were consistent with previous bers as compared with neurotypical youths while per-
findings related to the inhibitory control process forming the counting Stroop task.
Inhibitory control and visual processing in ADHD 2669
Sahakian BJ, Jones G, Levy R, Gray J, Warburton D (1989). and executive dysfunction in children with attention
The effects of nicotine on attention, information processing, deficit hyperactivity disorder. Psychological Medicine 43,
and short-term memory in patients with dementia of the 1093–1107.
Alzheimer type. British Journal of Psychiatry 154, 797–800. Silk TJ, Vance A, Rinehart N, Bradshaw JL, Cunnington R
Sahakian BJ, Morris RG, Evenden JL, Heald A, Levy R, (2008). Dysfunction in the fronto-parietal network in
Philpot M, Robbins TW (1988). A comparative study of attention deficit hyperactivity disorder (ADHD): an fMRI
visuospatial memory and learning in Alzheimer-type study. Brain Imaging and Behavior 2, 123–131.
dementia and Parkinson’s disease. Brain 111, 695–718. Slaats-Willemse DI, Swaab-Barneveld HJ,
Sahgal A (1987). Some limitations of indices derived from de Sonneville LM, Buitelaar JK (2007). Family-genetic
signal detection theory: evaluation of an alternative index study of executive functioning in attention-deficit/
for measuring bias in memory tasks. Psychopharmacology 91, hyperactivity disorder: evidence for an endophenotype?
517–520. Neuropsychology 21, 751–760.
Schneider MF, Krick CM, Retz W, Hengesch G, Trommer BL, Hoeppner JA, Lorber R, Armstrong KJ (1988).
Retz-Junginger P, Reith W, Rösler M (2010). Impairment The Go-No-Go paradigm in attention deficit disorder.
of fronto-striatal and parietal cerebral networks correlates Annals of Neurology 24, 610–614.
with attention deficit hyperactivity disorder (ADHD) Wechsler D (1991). Wechsler Intelligence Scale for Children,
psychopathology in adults – a functional magnetic 3rd edn. The Psychological Corporation: San Antonio, TX.
resonance imaging (fMRI) study. Psychiatry Research: Whelan R, Conrod PJ, Poline J-B, Lourdusamy A,
Neuroimaging 183, 75–84. Banaschewski T, Barker GJ, Bellgrove MA, Büchel C,
Schulz KP, Fan J, Tang CY, Newcorn JH, Buchsbaum MS, Byrne M, Cummins TDR, Fauth-Bühler M, Flor H,
Cheung AM, Halperin JM (2004). Response inhibition in Gallinat J, Heinz A, Ittermann B, Mann K, Martinot J-L,
adolescents diagnosed with attention deficit hyperactivity Lalor EC, Lathrop M, Loth E, Nees F, Paus T, Rietschel M,
disorder during childhood: an event-related fMRI study. Smolka MN, Spanagel R, Stephens DN, Struve M,
American Journal of Psychiatry 161, 1650–1657. Thyreau B, Vollstaedt-Klein S, Robbins TW,
Shang C-Y, Gau SS (2011). Visual memory as a potential Schumann G, Garavan H, Imagen Consortium
cognitive endophenotype of attention deficit hyperactivity (2012). Adolescent impulsivity phenotypes
disorder. Psychological Medicine 41, 2603–2614. characterized by distinct brain networks. Nature
Shang C-Y, Gau SS (2012). Improving visual memory, Neuroscience 15, 920–925.
attention, and school function with atomoxetine in boys Wu YH, Gau SS, Lo YC, Tseng WY (2014). White matter
with attention-deficit/hyperactivity disorder. Journal of Child tract integrity of frontostriatal circuit in attention deficit
and Adolescent Psychopharmacology 22, 353–363. hyperactivity disorder: association with attention
Shang CY, Wu YH, Gau SS, Tseng WY (2012). Disturbed performance and symptoms. Human Brain Mapping 35,
microstructural integrity of the frontostriatal fiber pathways 199–212.