Content uploaded by Emily Barkley-Levenson
Author content
All content in this area was uploaded by Emily Barkley-Levenson on Feb 05, 2018
Content may be subject to copyright.
REGULAR ARTICLE
Prefrontal Cortical Activity During the Stroop Task: New Insights
into the Why and the Who of Real-World Risky Sexual Behavior
EmilyBarkley-Levenson, PhD1,2 • FengXue, PhD1,3 • VitaDroutman, PhD1 • Lynn C.Miller, PhD1,4 •
Benjamin J.Smith, MA1 • DavidJeong, PhD4 • Zhong-LinLu, PhD5 • AntoineBechara, PhD1 • Stephen J.Read, PhD1
Published online: XX XXXX 2018
© Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
Abstract
Background Research suggests that deficits in both ex-
ecutive functioning and trait impulsivity may play a role
in risky sexual behavior. At the neural level, differences
in regulation of the prefrontal cortex have been linked
to impulsivity, measured neurocognitively and through
self-report. The relationship between neurocognitive
measures of executive control and trait impulsivity in pre-
dicting risky sexual behavior has not been investigated.
Purpose To investigate the relationship between neural
functioning during the Stroop task and risky sexual
behavior, as well as the effect of individual differences
in urgent (positive and negative) impulsivity on this
relationship.
Methods A total of 105 sexually active men who have
sex with men completed the Stroop task during func-
tional magnetic resonance imaging scanning. They also
completed impulsivity inventories and self-reported their
risky sexual behavior (events of condomless anal sex in
the last 90days).
Results Risky participants had greater activation than
safe participants during the color congruent condition
of the Stroop task in anterior cingulate cortex/dorso-
medial prefrontal cortex, dorsolateral prefrontal cortex,
left frontal pole, and right insula. Across these regions,
this neural activation mediated the link between (posi-
tive and/or negative) urgent impulsivity and risky sexual
behavior.
Conclusions Findings suggest that the brains of men who
engage in risky sexual behavior may employ a different
distribution of cognitive resources during tasks of execu-
tive functioning than men who practice safe sex, and that
this may relate to differences in the prefrontal cortical/
fronto-insular system responsible for impulse control.
Keywords Impulsivity • Executive functioning • Stroop
task • Negative urgency • Positive urgency
Introduction
Risky sexual behavior (i.e., condomless anal sex [CAS])
among men who have sex with men (MSM) remains
prevalent [1], despite its role in increasing the risk
of sexually transmitted infections (STIs), including
HIV infection [2]. Many psychological, sociocultural,
and environmental factors contribute to risky sexual
decision-making in HIV-negative MSM, including high
sensation seeking, low socioeconomic status, drug and
alcohol use, number of partners, stigma, family re-
jection, and residing in an urban environment [3–6].
Emily Barkley-Levenson
Emily.barkleylevenson@hofstra.edu
1 Department of Psychology, University of Southern
California, Los Angeles, CA, USA.
2 Department of Psychology, Hofstra University, Hempstead,
New York
3 Department of Radiology, University of California San
Diego, San Diego, California
4 USC Annenberg School for Communication and Journalism,
University of Southern California, Los Angeles, California
5 Department of Psychology, The Ohio State University,
Columbus, Ohio
ann. behav. med. (2018) XX:1–13
DOI: 10.1093/abm/kax019
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
However, neurobiological contributions to risky sexual
decision-making are not well understood [7]. One neuro-
biological theory of risky decision-making [8, 9] posits
that the risky behavior emerges from the interaction of
three neural systems: an amygdala-striatal system that
promotes habitual appetitive behaviors, a prefrontal
cortex system facilitating executive functioning and im-
pulse control, and an insular cortex system that responds
to homeostatic and interoceptive signals and facilitates
switching among neural networks (default mode and
executive control) [10]. Although this model was devel-
oped in the context of addiction, it has been applied to
other types of decision-making that pit appetitive drives
against self control, such as overeating high-calorie
foods [11]. Like desirable foods and addictive substances,
sexual stimuli have also been shown to reduce impulse
control [12]. Furthermore, neural activity in the pre-
frontal cortex during response inhibition correlates with
both substance use and sexual risk in adolescents [13,
14] suggesting the efficacy of the neurobiological model
in the domain of risky sex as well. Similarly, the amyg-
dala and striatum have been implicated in the processing
of sexual stimuli [15, 16]. Stronger striatal reactivity in
response to sexual cues has been shown to predict sub-
sequent sexual behavior [17], and striatal sensitivity to
sexual cues has been observed in individuals with com-
pulsive sexual behavior relative to controls [16, 18]. Based
on this model and the existing work on the neurobiology
of sexual risk, one contributor to risky sexual behavior
in MSM may be individual variability within the pre-
frontal cortex system, leading to individual differences in
executive functioning.
The Stroop task [19, 20] is commonly used as a
measure of executive functions that are mediated by
the prefrontal cortex [21]. The classic Stroop task pre-
sented participants with a series of color words printed
in various ink colors and observed an interference effect
such that naming the ink color of incongruent word-
ink stimuli (e.g., the word “blue” printed in red ink) was
slower and less accurate than naming the ink color of
congruent stimuli (e.g., the word “red” printed in red ink)
or neutral stimuli (e.g., a series of Xs printed in red ink).
Stroop performance depends on successful response in-
hibition, interference resolution, and behavioral conflict
resolution [22] because the incongruent condition of the
task requires the suppression of a habitual prepotent re-
sponse (i.e., word reading) in favor of a more difficult
response (i.e., ink color naming of an incongruent color
word). In an event-related Stroop design, which we em-
ploy here, participants are asked to switch among the
conditions of the task (color naming or word reading for
congruent, neutral or incongruent combinations) on a
trial-by-trial basis, such that performance also depends
on other components of executive function (e.g., task
monitoring and switching, cognitive control).
Despite its prevalence in cognitive psychology and
neuroscience, relatively little neuroimaging research has
applied the Stroop task to risk-taking populations or
those with impulse control deficits. Among the studies
that have done so, results have been mixed. Abstinent
methamphetamine (MA) abusers, pathological gamblers,
and adults with attention-deficit hyperactivity disorder
(ADHD) have been found to have less activation in pre-
frontal cortical regions than controls during incongruent
Stroop conditions [23–25]. Other studies have found
that adolescents with ADHD [26], abstinent cocaine
users [27], and patients with schizophrenia [28] showed
greater activation than controls in similar regions during
incongruent Stroop conditions. In two studies examin-
ing the relationship between risky sexual behavior and
neuropsychological assessment of executive functioning,
Stroop interference reaction time (reaction time on in-
congruent trials minus reaction time on neutral trials)
was unrelated to sexually risky decision-making for
drug-dependent men who were HIV+ [29] or HIV+ and
HIV− men [30] for whom sexual preference is unknown.
None of these studies, however, examined whether any
patterns of neural activation exhibited during the Stroop
task are associated with sexual risk-taking.
Other research has explored relationships between
impulsivity, executive functioning, and risky sexual be-
havior from a personality trait perspective. Self-report
scales measuring impulsivity [31–33] have been shown to
correlate positively with one another but not with behav-
ioral task measures of impulsive disinhibition, including
the Stroop task, suggesting that laboratory tasks and
self-report scales may capture distinct facets of impulse
control and impulsivity [34]. Among self-report inven-
tories of impulsivity, positive urgency [35], negative ur-
gency, and sensation seeking [36] correlated positively
with risky sexual behavior in adults. Thus, a comprehen-
sive investigation of the role of impulse control in risky
sexual behavior should incorporate personality measures
as well as neural and behavioral ones. One relevant study
[37] has integrated the measurement of negative urgency
with behavioral and neural responses to the Go/No-Go
response inhibition task and found that neural activity
during response inhibition predicted subsequent sub-
stance abuse in individuals high in negative urgency, but
no research has synthesized all these components in rela-
tion to real-world sexual risk-taking behavior.
The heterogeneity of the extant neuroimaging and
neuropsychological findings, and the gaps in the prior
literature, underscore the exploratory nature of the
current study, which is the first to investigate the rela-
tionship between neural correlates of the Stroop task,
individual differences in impulsivity, and risky sexual
behavior. In this study, we employed a large sample size
(N=105) of risky (i.e., engaging in CAS over the past
90 days) and safe (i.e., never engaging in CAS) MSM
2 ann. behav. med. (2018) XX:1–13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
to investigate whether the neural signature of executive
function elicited by the Stroop task differs on the basis
of real-world sexual risk-taking behavior, in keeping
with the theorized role of the prefrontal cortex’s execu-
tive control system in risky decision-making. Specifically,
we hypothesized that risky participants would display
differential recruitment of the prefrontal cortex relative
to safe participants during the Stroop task condition
requiring the greatest inhibition, and that behavioral
differences would parallel these neural differences, with
risky participants exhibiting lower accuracy or longer re-
action times. We further sought to understand the extent
to which individual differences in impulsivity influence
real-world sexual risk-taking in this population, in the
hope that understanding the relationship between neural
executive control and underlying trait impulsivity can
produce a clearer picture of which individuals may be
more inclined toward sexual risk-taking and why. We
hypothesized that risky participants would self-report
higher impulsivity than safe participants and theorized
that impulsivity would correlate positively with neural
activity in the prefrontal cortex during the Stroop task.
Methods
Participants
The current analysis is part of a larger study that used
Internet advertisements to recruit a community sample
in Southern California of 177 sexually active nonmonog-
amous men MSM. The research protocol was approved
by the Institutional Review Board at the University of
Southern California, and informed consent was obtained
from all participants included in the study. All qualified
participants were between the ages of 18 and 30 years
and self-reported engaging in anal sex in the last 90days,
being HIV negative, were free of neurological diagnoses,
were not binge drinkers, and metall safety requirements
for magnetic resonance imaging (MRI) scanning. Data
collection took place between January 2012 and April
2014. Participants completed the tasks, and surveys
reported here as part of a two-session data collection
procedure that comprised two 1.25-hr MRI scan sessions
and approximately 1hr of self-report measures. Sessions
were typically separated by no more than 1 week; test-
ing was divided to avoid fatigue and discomfort during
extended MRI data collection. Participants were paid
$100 following each session, with an additional payment
of $1–$20 based on performance on certaintasks.
To categorize participants based on their sexual risk
for data analysis, participants self-reported the number
of times they had engaged in anal sex with and without
a condom in the past 90days; individuals who reported
more than zero events of CAS were categorized as “risky”,
whereas those with zero CAS were categorized as “safe”.
From this larger sample, 146 participants ended the study
with complete behavioral and functional MRI (fMRI)
data for two runs of the Stroop task. We excluded 27
participants on the basis of poor task performance using
two metrics: 13 participants who scored less than 70% in
any run of the congruent or neutral Stroop conditions,
which reflects 2 SDs below the mean accuracy for con-
gruent and neutral conditions (M=90.5%, SD=9.3%),
and 14 participants who scored 0% in any run of the in-
congruent Stroop conditions because neuroimaging data
analysis of the incongruent condition would not be pos-
sible for these participants. These restrictions attempt
to control for participants who failed to understand
the task (despite prior training), who fell asleep, or who
chose to respond at random, while still including par-
ticipants who attempted to complete the task correctly
but produced errors. These cutoffs were chosen specific-
ally for this version of the Stroop, which is more difficult
than the classic version of the task (which uses verbal
responses) because it requires the prior memorization
of a color-key mapping in addition to the task itself;
the thresholds were selected to maximize the number
of participants in analyses while ensuring enough tri-
als to facilitate fMRI data analysis for the incongruent
conditions. An additional 14 participants were excluded
for exceeding motion of 3 mm in any direction on any
run of the fMRI task. A total of 105 participants (33
safe, 72 risky) were included in the analysis. Risky par-
ticipants were oversampled relative to safe participants
in the larger study to facilitate subdivision of this popu-
lation on the basis of lifetime MA use for future analyses
(of the initial 177 participants, 56 reported zero CAS,
69 reported CAS but no MA use, 44 reported CAS and
lifetime MA use, and 8 reported zero CAS and lifetime
MA use). However, no analyses reported here were sig-
nificant when participants were grouped on the basis of
MA use versus nonuse (33 MA, 72 non-MA), suggest-
ing that neither the imbalance in sample size nor hetero-
geneity within the risky population drives our findings.
Demographic characteristics of the sample are reported
in Table1.
fMRI StroopTask
Our version of the Stroop task, adapted from the ori-
ginal [19] for use in the fMRI scanner, asked participants
to press one of four buttons (red, blue, yellow, or green)
in response to one of two conditions, which were cued
before the presentation of stimuli. In the “color” condi-
tion, participants responded to the ink color of stimuli,
which were congruent, incongruent, or neutral (a series
of Xs). In the “word” condition, participants responded
to the text of the word stimuli, which were congruent,
incongruent, or neutral (white text). In both conditions,
participants responded within 2,500 ms with a jittered
ann. behav. med. (2018) XX:1–13 3
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
inter-trial interval (mean = 5,310 ms; range 5,000–
8,000ms). Participants completed two runs of the task,
with 16 pseudorandomized stimuli from each of the six
conditions in each run, for a total of 192 trials.
Personality Measures
Participants completed a battery of personality inven-
tories, including the UPPS-P (urgency, premeditation,
persevaration, sensation-seeking, positive urgency) im-
pulsivity scale [38]. The UPPS-P consists of five impul-
sivity subscales: negative urgency (engaging in impulsive
behavior in response to negative affect), lack of pre-
meditation, lack of perseveration, sensation-seeking,
and positive urgency (engaging in impulsive behavior in
response to positive affect). Because attrition from the
study between the two sessions of data collection led to
some instances of missing data, UPPS-P scores were col-
lected for 90 of the 105 participants.
Functional Imaging Procedure
Participants completed the Stroop task as part of a
larger battery of structural and fMRI data collection,
which included a Go/No-Go task measuring response
inhibition, reversal learning tasks measuring learning
from reward and punishment, a cups gambling task
measuring risk-taking, and a virtual interactive narra-
tive game exploring naturalistic sexual behavior; results
from these tasks are being analyzed and reported sep-
arately. Neuroimaging data collection totaled approxi-
mately 2.5hr, divided across two sessions, such that the
Stroop task was completed during a 1.25-hr scan ses-
sion. The order of tasks was counterbalanced across
participants.
Participants lay supine in the scanner, and viewed
visual stimuli back-projected onto a screen through a
mirror attached to the head coil. Foam pads were used
to minimize head motion. Stimulus presentation and
timing of all stimuli and response events were controlled
by Matlab with Psychtoolbox (http://www.psychtool-
box.org) extensions on a MacBook Pro. Participants’
responses were collected online using an MRI-compatible
buttonbox.
MRI data for the Stroop task were collected in one
session in a 3T Siemens MAGNETOM Tim/Trio
scanner. A T1-weighted anatomical image, a set of diffu-
sion-weighted images, and several task-related functional
image sequences were collected. Task-related fMRI data
were acquired using T2*-weighted (repetition time [TR]
= 2,000 ms, echo time [TE] = 25 ms, 64 × 64 matrix size
with a resolution of 3 mm2, using 41 3.0-mm axial slices)
imaging.
fMRI Data Preprocessing and Statistical Analysis
Image preprocessing and statistical analysis were carried
out using FEAT (FMRI Expert Analysis Tool) version
6.00, part of the FSL package (FMRIB software li-
brary, version 4.1.8, www.fmrib.ox.ac.uk/fsl). The data
analysis pipeline was managed using the XFSL package
(http://xfsl.fmri.cn). The data were temporally filtered
using a nonlinear high-pass filter with a 100s cutoff and
spatially smoothed using a 5-mm full-width-half-maxi-
mum Gaussian kernel. A two-step registration procedure
was used images were first registered to the MPRAGE
(magnetization-prepared rapid gradient-echo) struc-
tural image and then into the standard MNI (Montreal
Neurological Institute) space, using affine transforma-
tions with FLIRT (FMRIB’s Linear Image Registration
Tool) [39, 40] to the MNI-152 T1-template brain.
Registration from MPRAGE (magnetization-prepared
rapid gradient-echo) structural images to standard space
was further refined using FNIRT (FMRIB’s Nonlinear
Image Registration Tool) [41]. Statistical analyses were
performed in the native image space, with the statistical
maps normalized to the standard space before higher
level analysis. Melodic independent components analysis
(ICA) was used to denoise the preprocessed functional
data [42]. The FIX software package was used to auto-
matically identify noise components [43, 44].
Data were modeled at the first level using a general
linear model within FSL’s FILM module. Correct trials
of the six conditions (color congruent [CC], color neu-
tral [CN], color incongruent [CI], word congruent, word
neutral, and word incongruent) were modeled separately.
Error trials from the six conditions were also added to
the model as separate nuisance covariates. We modeled
the main effects of each individual condition as well as
several contrasts, including comparisons of incongruent
> congruent and incongruent > neutral (collapsing
across color and word condition).
Table1 Participant Demographics
N
Age
M (SD)
White
N (%)
Black
N (%)
Latino
N (%)
CAS90
M (SD)
AS90
M (SD)
MA90
M (SD)
Safe 33 23.1 (2.7) 8 (24.2%) 15 (45.5%) 10 (30.3%) 0 (0) 8.3 (13.4) 0.1 (0.5)
Risky 72 24.9 (3.3) 26 (36.1%) 18 (25%) 28 (38.9%) 7.3 (9.3) 14.3 (12.2) 1.5 (4.4)
AI90 all events of anal sex in the past 90days; CAS90 events of condomless anal sex in the past 90days; M mean; MA90 uses of meth-
amphetamine in the past 90days; N number of participants; SD standard deviation.
4 ann. behav. med. (2018) XX:1–13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
At the group level, we tested whether any lower level
effects differed on the basis of risky sex condition (risky
vs. safe). In this model, we used a random-effects model
for group analysis using the FLAME (FMRIB’s Local
Analysis of Mixed Effects) stage 1 simple mixed-effect
model [45–47]. Group images were thresholded using
cluster detection statistics with a height threshold of
z > 2.3 and a cluster probability of p < .05, corrected
for whole-brain multiple comparisons using Gaussian
Random Field Theory. In addition, to address the recent
observation [48] that many fMRI analysis procedures in-
flate false-positive rates (although FLAME1 was noted
to be less inflated than other common analyses), we
confirmed the group-level results using nonparametric
permutation tests through FSL’s randomise function
[49], with 500 permutations for each analysis, applying
threshold-free cluster enhancement corrected for mul-
tiple comparisons; the nonparametric analysis results are
reported throughout.
Results
Behavioral Results
The participants did not differ significantly in accuracy
or reaction time on any of the task conditions on the
basis of sexual risk-taking category. Across risk catego-
ries, performance on both the color and word conditions
of the Stroop task differed on the basis of task difficulty.
Comparisons of accuracy results by condition are pre-
sented in Table2. Based on these behavioral results, we
focused subsequent fMRI analyses on the CC and CI
trials because they represent the least and most difficult
conditions of the Strooptask.
Risky and safe participants differed on several dimen-
sions of the UPPS-P scale. Risky participants had higher
scores than the safe participants on negative urgency
[t(88)=−3.837, p < .001], positive urgency [t(88)=−2.404,
p = .018], and lack of premeditation [t(88) = −2.124,
p=.036], with higher scores on these subscales reflecting
greater behavioral impulsivity. Negative and positive ur-
gency were highly positively correlated with one another
[r(88)=.716, p < .001], while lack of premeditation cor-
related positively but less strongly with negative urgency
[r(88)=.339, p=.001] and positive urgency [r(88)=.297,
p=.004]. We focus on negative and positive urgency for
subsequent analyses because of their reported significant
associations with risky sexual behavior [28, 29].
fMRI Results
Activation across task conditions
We observed a consistent pattern of significant activa-
tion during the CC, CN, and CI conditions in the dorsal
anterior cingulate cortex (dACC)/dorsomedial pre-
frontal cortex (DMPFC), dorsolateral prefrontal cortex
(DLPFC), anterior insula, superior parietal lobule, and
occipital cortex. Peak voxel activation in these regions
from nonparametric analyses for each of the color con-
ditions is shown in Table3.
We next investigated whether there were differences
in activation across the task conditions by identifying
regions where activation was greater in incongruent than
in congruent conditions and where activation was greater
in incongruent than neutral conditions. These contrasts
overlapped in several regions identified previously, includ-
ing the dACC/DMPFC, left DLPFC, bilateral insula, and
left frontal pole (Fig.1; Table3). Thus, activation in these
regions appears to increase with increasing task difficulty.
Neural differences by sexual risk category
We next investigated whether the increases in neural acti-
vation on the basis of task difficulty differed on the basis
of sexual risk category. To explore this, we created a “task
difficulty” mask of the overlap between the incongruent
Table2 Stroop Task Percent Accuracy by Condition
Condition M SD Contrast t p Cohen’s d
Color congruent (CC) 94.97% 5.49% CC vs. CN 4.88 <.001 0.48
Color neutral (CN) 91.85% 6.58% CI vs. CN 18.52 <.001 2.56
Color incongruent (CI) 62.95% 19.16% CI vs. CC 18.93 <.001 2.54
Word congruent (WC) 93.21% 6.43% WC vs. WN 0.67 .508 0.06
Word neutral (WN) 92.86% 6.79% WI vs. WN 9.83 <.001 1.12
Word incongruent (WI) 82.14% 13.36% WI vs. WC 10.16 <.001 1.19
CC vs. WC 3.00 .003 0.30
CN vs. WN 1.65 .103 0.16
CI vs. WI 10.77 <.001 1.09
M mean percent accuracy; SD standard deviation.
Signicance threshold=p < .006, applying Bonferroni correction for multiple comparisons.
ann. behav. med. (2018) XX:1–13 5
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
> congruent and incongruent > neutral contrasts (as
shown in Fig.1). We then created five discrete regions of
interest (ROIs) for each of the major anatomical regions
contained in that mask (ACC/DMPFC, DLPFC, right
insula, left insula, and left frontal pole), by selecting the
overlap of the task difficulty mask with anatomically
defined masks from the Harvard-Oxford cortical and
subcortical atlas for each region (Fig. 2A). Finally, we
extracted the percent signal change from each ROI and
compared percent signal change values for risky and safe
participants using independent-samples t-tests.
Risky participants had greater activation than safe
participants during CC trials in DMPFC [t(103)=2.712,
p = .008, d = .55], DLPFC [t(103) = 2.431, p = .017,
d=.51], right insula [t(103)=3.220, p=.002, d=.67], and
left frontal pole [t(103)=2.687, p=.008, d=.54]. Risky
participants also had greater activation than safe partic-
ipants during CN trials in right insula [t(103) = 2.689,
p = .008, d = .56] and left frontal pole [t(103) = 2.191,
p=.031, d = .47]. The patterns of activation in each of
these ROIs are shown in Fig.2B–F. When performing the
same analyses including only participants with no history
of MA use (safe N=28, risky N=44), all comparisons re-
main significant, despite the reduction in statistical power;
this suggests that stimulant use does not play a strong role
in this particular measure of neurocognitive function.
Table 3 Coordinates (in MNI space) and Maximum Intensity
(z-statistic) of Peak Voxel Activation for Nonparametric Whole-
Brain Contrasts
mm
Condition/
contrast Location Z-Max X Y Z
Color
congruent
dACC/DMPFC R 12.3 4 16 46
L 12.0 −4 14 48
DLPFC R 8.39 22 2 52
L 9.77 −26 6 52
Insula R 9.39 34 26 6
L 9.96 −30 22 6
Sup. parietal lobule/
Supramarginal
gyrus
R 9.73 34 −46 42
L 12.6 −38 −48 44
Occipital cortex R 10.5 32 −92 2
L 11.2 −38 −84 −10
Cerebellum L 5.87 −28 −60 −28
Color neutral dACC/DMPFC R 12.3 4 16 46
L 12.3 −2 14 50
DLPFC R 9.27 36 0 64
L 10.5 −26 2 62
Insula R 10.1 34 24 6
L 9.91 −30 18 8
Sup. parietal lobule/
Supramarginal
gyrus
R 9.55 42 −46 56
L 12.2 −46 −34 42
Occipital cortex R 11.9 32 −92 2
L 11.8 −30 −94 −4
Cerebellum L 6.35 −28 −62 −28
Color
incongruent
dACC/DMPFC R 14.6 8 16 42
L 14.5 −4 14 48
DLPFC R 10.5 28 −4 60
L 13.4 −48 10 32
Insula R 13.7 34 26 0
L 14.7 −32 24 0
Sup. parietal lobule/ R 11.5 36 −52 42
Supramarginal gyrus L 14.4 −32 −58 44
Occipital cortex R 11.4 32 −90 −8
L 13.0 −36 −84 −10
Cerebellum L 6.35 −28 −62 −28
Incongruent >
congruent
dACC/DMPFC R 9.22 8 30 26
L 6.31 −4 32 38
DLPFC R 4.99 40 10 32
L 7.33 −50 22 30
Insula R 3.79 46 12 −6
L 6.74 −32 24 −2
mm
Condition/
contrast Location Z-Max X Y Z
Frontal pole R 5 46 42 −10
L 6.53 −40 60 8
Angular gyrus R 5.9 56 −44 26
L 7.04 −32 −56 40
Caudate R 6.0 10 12 2
L 5.68 −12 14 6
VLPFC R 7.42 42 22 −4
L 7.52 −42 20 0
Cerebellum R 5.91 8 −78 −28
Incongruent >
neutral
dACC/DMPFC R 9.44 6 34 28
L 8.75 −2 20 50
DLPFC L 8.36 −50 22 28
Insula R 7.79 34 24 −4
L 8.31 −30 24 −2
Frontal pole L 6.23 −44 52 0
Angular gyrus L 7.37 −34 −54 38
dACC dorsal anterior cingulate cortex; DLPFC dorsolateral pre-
frontal cortex; DMPFC dorsomedial prefrontal cortex; VLPFC
ventrolateral prefrontal cortex.
continued
Table 3 continued
6 ann. behav. med. (2018) XX:1–13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
Fig.1. Signicant activation (p <.05, corrected for multiple comparisons) for the contrasts incongruent > congruent (in yellow) and in-
congruent > neutral (in blue), overlaid on MNI-152 template brain. The overlapping region between the two contrasts is shown in green.
MNI Montreal Neurological Institute.
Fig.2. (A) ROI masks for ACC/DMPFC (yellow), left DLPFC (red), left and right insula (blue), and left frontal pole (green). B–F.
Difference in percent signal change for the CC, CN, and CI conditions between safe and risky participants in ACC/DMPFC (B), left
DLPFC (C), right insula (D), left insula (E), and left frontal pole (F). *p < .05. ACC anterior cingulate cortex; CC color congruent; CI
color incongruent; CN color neutral; DLPFC dorsolateral prefrontal cortex; DMPFC dorsomedial prefrontal cortex; ROI region of
interest.
ann. behav. med. (2018) XX:1–13 7
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
No other differences between risky and safe partici-
pants were significant for the CN trials, and no differ-
ences were significant for CI trials in any of theROIs.
Across all participants, activation was greater during
the CI than the CC condition (as seen in the whole-brain
contrast); however, the increase in activation from the CC
condition to the CI condition was greater for safe partici-
pants than risky participants in DMPFC [t(103)=2.524,
p =.013, d= .52], DLPFC [t(103) = 2.237, p = .027,
d=.46], and right insula [t(103)=2.220, p=.029, d=.47].
Mediational analyses
Because both positive and negative urgency significantly
predicted whether our participants were risky or safe, we
investigated whether patterns of neural activation during
the Stroop mediate the link between urgency and hav-
ing engaged in risky sexual behavior. We focused on the
CC condition, controlling for activation during the CI
condition, to explore what might be driving the observed
difference between risk groups in this condition shown
in the ROI analyses. We assessed this in the five ROIs
(ACC/DMPFC, DLPFC, left frontal pole, right insula,
left insula) during the CC condition, controlling for ac-
tivation during the CI condition. To do this, we tested
a series of bias-corrected, bootstrapped (at 10,000 sam-
ples) mediation models using logistic regressions with
model 4 of the PROCESS macro for SPSS [50]. In these
mediational models, the a path denotes the effect of the
trait (x) on brain activations (m), the b path denotes the
effect of brain activations (m) on risk category (y), and
the c’ path denotes the effect of the trait positive or nega-
tive urgency (x) on risk category (y), controlling for the
indirect effect. Alack of a zero in the confidence interval
reported in Table4 indicates a significant indirect effect.
As indicated in Table 4, for negative urgency, the in-
direct (mediational) paths were significant for the dACC/
DMPFC, DLPFC, the left frontal pole, and the right
Table4 Bootstrapped Mediation Effects of Negative and Positive Urgency on Sexual Risk-Taking Category
Mediational component
95% CI
Indirect effect
Trait Location a b
Boot
LLCI
Boot
ULCI
Negative urgency dACC/DMPFC 0.0436 5.648** 0.0009 0.8127
DLPFCa0.0703* 4.22* 0.0253 0.8915
Left frontal pole 0.0563* 3.6149* 0.0004 0.6543
Right insula 0.0514* 5.0525* 0.0190 0.8096
Left insula 0.0031 3.171 −0.1278 0.2210
Positive urgency dACC/DMPFC 0.0485* 5.348** 0.0247 0.7270
DLPFCb0.0644* 4.477* 0.0406 0.8286
Left frontal pole 0.0494 3.872* 0.0018 0.6079
Right insula 0.0406 5.122* −0.0045 0.6322
Left insula 0.0104 2.7703 −0.0709 0.2793
Mediational component “a” is effect from trait to brain location, “b” is from brain location to dichotomous risk (safe [0], risky [1]).
Values listed under the indirect effect are 95% condence intervals of each indirect effect. *p < .05, **p < .01, ***p < .001.
dACC dorsal anterior cingulate cortex; DLPFC dorsolateral prefrontal cortex; DMPFC dorsomedial prefrontal cortex.
aAfter accounting for multivariate outliers via Mahalanobis distance, one outlier was removed. This had little inuence on the mediation
effect (a=0.0649, p < .05; b=3.949, p < .05; 95% CI =0.0059, 0.8081).
bAfter accounting for multivariate outliers via Mahalanobis distance, one outlier was removed. This had little inuence on the mediation
effect (a=0.0587, p < .05; b=4.2076, p < .05; 95% CI=0.0245, 0.7746).
Fig.3. Diagram of the mediational relationship among negative urgency, DMPFC brain activation, and risk category (risky vs. safe) (A)
and the mediational relationship among negative urgency, DLPFC brain activation and risk category (B). The same structure is also used
for the remaining eight analyses shown in Table4. DLPFC dorsolateral prefrontal cortex; DMPFC dorsomedial prefrontal cortex.
8 ann. behav. med. (2018) XX:1–13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
insula. For positive urgency, the indirect (mediational)
paths were significant for the ACC/DMPFC, DLPFC,
and left frontal pole. No other indirect effects were sig-
nificant. Figure3 shows the mediational relationships of
negative urgency, dACC/DMPFC and risk category, and
of negative urgency, DLPFC and risk category, for illus-
trative purposes. The same structure is also used for the
remaining eight analyses shown in Table4.
The direct effect of negative urgency on risk cat-
egory, controlling for the indirect effect, was signifi-
cant for each brain region: (i) dACC/DMPFC= 1.485
(SE = 0.469), CI: 0.5648, 2.406; (ii) DLPFC=1.2837
(SE = 0.4397), CI: 0.4219, 2.1454; (iii) left frontal
pole=1.4429 (SE = 0.4612), CI: 0.4391, 2.3468; (iv)
right insula =1.3887 (SE = 0.4561), CI: 0.4948, 2.2827
and; (v) left insula= 1.51626 (SE = .460), CI: 0.6609,
2.4643. For positive urgency, the direct effect was only
significant for the left insula and the left frontal pole:
(i): dACC/DMPFC= 0.7714 (SE=0.403), CI: −0.0178,
1.5606; (ii) DLPFC=0.6874 (SE=0.399), CI: −0.0961,
1.4710; (iii) left frontal pole=0.7938 (SE =0.402), CI:
0.0059, 1.5817; (iv) right insula=0.7728 (SE=0.399), CI:
−0.0097, 1.5554; and (v) left insula= 0.9114 (SE=0.392),
CI: 0.1435, 1.6792.
Discussion
The pattern of neural activation elicited by the Stroop
task in this study is consistent with other Stroop neu-
roimaging research. An automated meta-analysis of 113
studies categorized by the term “Stroop task” from the
NeuroSynth database (neurosynth.org) [51, 52] shows
activation in ACC, DMPFC, DLPFC, insula, and par-
ietal cortex, matching the regions we identified. These
regions are active during Stroop performance not only in
healthy control studies but also in studies of populations
associated with behavioral risk and impulsivity, such as
pathological gamblers [24], MA users [23], marijuana
users [53], and adolescents with ADHD [54]. This indi-
cates that even though the Stroop task has not previously
been used with sexually risky participants, our subjects
engaged in the expected cognitive processing when per-
forming thetask.
We observed no behavioral differences in Stroop
performance between participant groups in terms
of accuracy and reaction time, contrary to our ini-
tial expectations, despite observing individual differ-
ences in impulsivity and differences at the neural level.
However, findings of neural group differences in the
Stroop task in the absence of behavioral differences
have been shown in other studies, even those with im-
pulsive/high-risk populations [24, 26, 27]. This suggests
that performance effects are separable from processing
efficiency or neural recruitment in the context of the
Stroop task, and individual differences at the neural
level may alter only the latter while still producing
equivalent task performance.
Across several regions where the Stroop task elicits ac-
tivation (ACC/DMPFC, DLPFC, insula), the magnitude
of activation increased between the CC (easy) and CI
(difficult) condition. However, this increase was driven
by safe participants, with risky participants’ elevated ac-
tivation during the CC condition leading to less change
on the basis of task difficulty. These results deviate from
our hypothesis, as we expected neural differences to be
more prevalent in the CI condition, which necessitates
the most pronounced inhibitory control. These findings
suggest that risky men may modulate their recruitment
of the neural resources necessary for inhibitory control
less than safe men do in the face of changing task diffi-
culty. Asimilar pattern of activation was seen in a Stroop
study of MA abusers, wherein control participants
showed increased DLPFC activation during incongruent
trials that followed other incongruent trials versus those
that followed congruent trials, while MA participants
did not [55]. It is worth noting, however, that the pattern
reported by Salo etal. was due to a reversed pattern of
activation on the part of MA participants (greater acti-
vation during incongruent trials that followed congruent
trials), as opposed to the consistently elevated activation
we observe. This difference may relate to neurological
effects of MA dependence that are not present in our
risky sample, despite the fact that both groups are the-
orized to differ from control participants in cognitive
control. It has also been noted that unlike neutral trials,
the CC trials still provide a competing response-eligible
stimulus (e.g., participants must still process color infor-
mation and word information) and that participants may
engage in an attempt to suppress the word as distract-
ing information, leading to increased recruitment of the
left DLPFC during congruent trials [56]. The fact that
we observed this increased recruitment in risky but not
safe participants opens up the possibility that they are
engaging in different cognitive strategies to successfully
execute thetask.
The pattern of elevated cortical activity insensitive
to task difficulty that we observed in risky participants
may be explained by cortical inefficiency. A study of
unipolar depressed individuals [57] found that they dis-
played hyperactivity in the ACC and DLPFC during
incongruent Stroop conditions and theorized that the
altered affective state produced by depression elevates
neural activity, leading to inefficiencies in cognitive
processing during tasks. Similar work contrasting emo-
tional and nonemotional Stroop conditions found de-
activation in ACC during the nonemotional conditions,
suggesting that this deactivation allocates resources for
effective cognitive performance in healthy participants
[58]. Based on this theory, risky participants’ hyper-
activity relative to safe participants in the CC condition
ann. behav. med. (2018) XX:1–13 9
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
represents an inability to reduce activation during easy
trials, leading to altered neural activity but not neces-
sarily impaired behavioral performance. Several studies
[23, 24] have observed a blunted neural Stroop effect (i.e.,
smaller difference in activation between incongruent and
congruent conditions) in high-impulsivity populations
(gamblers, abstinent MA users) relative to control pop-
ulations and have attributed this finding to neural inef-
ficiency. It is worth noting, however, that other studies
[59, 60] have observed opposite findings—a larger neural
Stroop effect in patient populations relative to control
populations—and have also attributed these results to
neural inefficiency in patients. This highlights the need
for studies of the Stroop task to investigate not only
the interaction Stroop effect contrast but also activa-
tion during individual conditions (e.g., CC, CI) to dis-
entangle differences in the myriad underlying cognitive
processes utilized in Stroop performance. More broadly,
caution should be exercised in conclusively interpreting
larger and smaller magnitudes of activation as represent-
ing either efficiency or inefficiency without supporting
evidence of differences in the underlying process or net-
work engagement [61].
Several of the regions identified here have been
associated with trait impulsivity as measured by psy-
chometric inventories. Larger gray-matter volumes in
MPFC and DLPFC, as well as subcortical regions,
correlated positively with the Barratt Impulsiveness
Scale 11 (BIS-11) [24] in healthy adults [62]; for nega-
tive urgency (as measured by the UPPS-P), greater
impulsivity correlated positively with lateral PFC gray-
matter volumes in cocaine users [63] and negatively with
DMPFC gray-matter volumes in healthy participants
[64]. Glucose metabolism (a proxy measure of neural
activity) in the fronto-insular network, including the
right insula, ACC, and orbitofrontal cortex, correlated
positively with impulsivity as measured by the BIS-11 in
patients with Parkinson’s disease [65]. Similarly, fMRI
studies have found that the BIS-11 correlated positively
with DLPFC activation in 3,4-methylenedioxymeth-
amphetamine (MDMA, “ecstasy”) users and controls
during a delayed memory task [66] and with activation
in the ACC [67] and middle frontal gyrus [68] elicited
by a Go/No-Go task with healthy participants, despite
observing no differences in task performance. These
studies suggest increased fronto-insular brain activity
independent of task performance may be a marker of
impulsivity, which is consistent with our finding that
risky participants are higher in behavioral impulsivity
(negative and positive urgency and lack of premedita-
tion) than safe participants and could also explain the
elevated levels of neural activity across task conditions
observed for risky participants on the Strooptask.
Further support for the role of fronto-insular activity
in impulsivity comes from the mediation relationship that
we observed, which suggests that neural responsiveness
during low-difficulty cognitive control underlies the rela-
tionship between urgent impulsivity and risky sexual be-
havior. This parallels the model put forth by Chester etal.
[37] of negative urgency’s relationship with response in-
hibition (as measured by neural responsiveness during an
emotionally valenced Go/No-Go task) and real-world sub-
stance abuse. They observed greater recruitment of PFC
regions during negatively-valenced response inhibition for
participants high in negative urgency (without behavioral
performance differences), and this same pattern observed
in the anterior insula predicted greater substance abuse
measured 1month and 1year after scanning. Chester etal.
suggest that the elevated fronto-insular activity may be a
compensatory mechanism to achieve satisfactory response
inhibition for impulsive individuals, but that such demands
may contribute to real-world impulse control failures such
as substance use. Similarly, the elevated neural response
during the Stroop task observed in our participants may
reflect a need for greater cognitive control recruitment
among more impulsive individuals even during less chal-
lenging control tasks; this elevated neural effort may prove
insufficient under real-world circumstances requiring
greater impulse control, leading to real-world risk-taking
including risky sexual behavior. This interpretation is pred-
icated on the idea that condom use is desirable from a cog-
nitive control perspective (i.e., that using condoms during
sex prevents the spread of HIV and STIs) but not from
an affective perspective (i.e., that condomless sex is more
pleasurable or desirable for oneself and/or one’s partner)
and that the high-arousal state of a sexual encounter inter-
feres with cognitive control mechanisms.
This study faces several limitations. While the dichot-
omous definition of sexual risk employed here facilitates
neural comparisons (and allows for the possibility that
there are qualitative differences between individuals who
always use condoms and those who do not), it erases much
of the complexity in conceptualizing and measuring
sexual risk. Subsequent studies may find that incorporat-
ing a continuous measure of the number of instances of
CAS, coupled with other sexual behavior variables (i.e.,
number of partners, use of pre-exposure prophylaxis),
will provide a more nuanced understanding of the neu-
rocognition underlying sexual risk taking. Another limi-
tation is that the lack of behavioral or neural differences
between the risky and safe group on the CI condition
makes it difficult to interpret these results in the context
of response inhibition because accurate performance on
congruent conditions does not require inhibition per se.
Moreover, the Stroop task captures only some aspects of
interference resolution, which may not reflect the cog-
nitive processes most salient to real-world sexual situa-
tions. For example, Stroop is a less apt measure of the
ability to restrain an inappropriate response during re-
sponse execution than the Go/No-Go and Stop Signal
10 ann. behav. med. (2018) XX:1–13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
tasks, although all tasks share some common circuitry
[56]. Subsequent analyses from this research project will
be able to address this question more thoroughly, as par-
ticipants also completed a Go/No-Go task as a part of
the larger study. Integrating individual responses across
executive function tasks requiring various aspects of in-
hibition will paint a clearer picture of the role of various
neurocognitive constructs on risky sexual behavior.
Taken together, our findings suggest that the brains of
men who engage in risky sexual behavior may distribute
cognitive resources in a distinct manner during tasks
of executive function compared with men who practice
safe sex, and that this may be due to differences in the
prefrontal cortical/fronto-insular system responsible for
aspects of executive functioning, task monitoring, and
cognitive control. Although the observed neural dif-
ferences between risky and safe men did not negatively
impact behavioral performance on the Stroop task, this
may be due to the demand effects of experiment par-
ticipation or the nonarousing nature of the laboratory
setting. In a real-world sexual situation, individual dif-
ferences in executive functioning may interact with other
neural systems, such as elevated urge states or habitual/
reward-seeking behavior that are not present in a la-
boratory context, to increase the likelihood of engag-
ing in actual risky sexual behavior. Subsequent research
exploring the relationship between risky sexual behavior
and cognitive control under more affectively arousing
contexts can explore this theory. Furthermore, the medi-
ational relationship with urgent impulsivity suggests that
this self-report scale could be useful in identifying indi-
viduals at higher likelihood of engaging in CAS (with its
concomitant risk of contracting HIV and other STIs),
who could then receive targeted interventions such as
pre-exposure prophylaxis. These findings offer a first
look into the ability of the Stroop task to distinguish
risky and safe participants on the basis of their neural
patterns during a complex executive function task.
Acknowledgments This study was funded by the National
Institute on Drug Abuse (NIDA) under R01DA031626 awarded
to Stephen Read (PI) and LCM (Co-I) received research funding
from NIDA. Authors VD and BJS were supported as research
assistants by NIDA funding.
Compliance with Ethical Standards
Authors’ Statement of Conflict of Interest and Adherence to Ethical
Standards The authors declare there is nothing to disclose.
References
1. Centers for Disease Control and Prevention. HIV testing and
behaviors among gay, bisexual, and other men who have sex
with men – United States. Morbidity and Mortality Weekly
Report. 2013; 62: 958–962. Available at http://www.cdc.gov/
mmwr/pdf/wk/mm6247.pdf. Accessibility verified December
12, 2017.
2. Bearinger LH, Sieving RE, Ferguson J, Sharma V. Global
perspectives on the sexual and reproductive health of ado-
lescents: Patterns, prevention, and potential. Lancet. 2007;
369(9568): 1220–1231.
3. Dudley MG, Rostosky SS, Korfhage BA, Zimmerman
RS. Correlates of high-risk sexual behavior among young
men who have sex with men. AIDS Educ Prev. 2004; 16(4):
328–340.
4. Frye V, Latka MH, Koblin B, etal. The urban environment
and sexual risk behavior among men who have sex with men.
J Urban Health. 2006; 83(2): 308–324.
5. Mustanski BS, Newcomb ME, Du Bois SN, Garcia SC, Grov
C. HIV in young men who have sex with men: Areview of
epidemiology, risk, and protective factors, and interventions.
J Sex Res. 2011; 48: 218–253.
6. Myers HF, Javanbakht M, Martinez M, Obediah S.
Psychosocial predictors of risky sexual behaviors in African
American men: Implications for prevention. AIDS Educ
Prev. 2003; 15(1 Suppl): 66.
7. Ross JM, Duperrouzel J, Vega M, Gonzalez R. The neuro-
psychology of risky sexual behavior. Int Neuropsychol Soc.
2016; 22(6): 586–594.
8. Bechara A. Decision making, impulse control and loss of
willpower to resist drugs: Aneurocognitive perspective. Nat
Neurosci. 2005; 8(11): 1458–1463.
9. Noël X, Brevers D, Bechara A. A neurocognitive approach
to understanding the neurobiology of addiction. Curr Opin
Neurobiol. 2013; 23(4): 632–638.
10. Droutman V, Read SJ, Bechara A. Revisiting the role of the
insula in addiction. Trends Cogn Sci. 2015; 19(7): 414–420.
11. He Q, Xiao L, Xue G, etal. Poor ability to resist tempting
calorie rich food is linked to altered balance between neural
systems involved in urge and self-control. Nutr J. 2014;
13(1): 1.
12. Macapagal KR, Janssen E, Fridberg DJ, Finn PR, Heiman
JR. The effects of impulsivity, sexual arousability, and ab-
stract intellectual ability on men’s and women’s Go/No-Go
task performance. Arch Sex Behav. 2011; 40(5): 995–1006.
13. Feldstein Ewing SW, Ryman SG, Gillman AS, Weiland BJ,
Thayer RE, Bryan AD. Developmental cognitive neurosci-
ence of adolescent sexual risk and alcohol use. AIDS Behav.
2016; 20: S97–S108.
14. Goldenberg D, Telzer EH, Lieberman MD, Fulgini A,
Galván A. Neural mechanisms of impulse control in sexually
risky adolescents. Dev Cogn Neurosci. 2013; 6: 23–29.
15. Cyders MA, Dzemidzic M, Eiler WJ, Kareken DA. An fMRI
study of responses to sexual stimuli as a function of gender
and sensation seeking: A preliminary analysis. J Sex Res.
2016; 53(8): 1020–1026.
16. Voon V, Mole TB, Banca P, etal. Neural correlates of sexual
cue reactivity in individuals with and without compulsive
sexual behaviours. PLos One. 2014; 9(7): e102419.
17. Demos KE, Heatherton TF, Kelley WM. Individual dif-
ferences in nucleus accumbens activity to food and sexual
images predict weight gain and sexual behavior. J Neurosci.
2012; 32(16): 5549–5552.
18. Seok JW, Sohn JH. Neural substrates of sexual desire in indi-
viduals with problematic hypersexual behavior. Front Behav
Neurosci. 2015; 9: 206.
19. Stroop JR. Studies of interference in serial verbal reactions. J
Exp Psychol. 1935; 18: 643–62.
ann. behav. med. (2018) XX:1–13 11
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
20. MacLeod CM. Half a century of research on the Stroop
effect: An integrative review. Psychol Bull. 1991; 109: 163–203.
21. Gruber SA, Rogowska J, Holcomb P, Soraci S, Yurgelun-
Todd D. Stroop performance in normal control subjects: An
fMRI study. NeuroImage. 2002; 16: 349–360.
22. Adleman NE, Menon V, Blasey CM, etal. A developmental
fMRI study of the Stroop color-word task. NeuroImage.
2002; 16(1): 61–75.
23. Nestor LJ, Ghahremani DG, Monterosso J, London ED.
Prefrontal hypoactivation during cognitive control in early
abstinent methamphetamine-dependent subjects. Psychiatry
Res. 2011; 194(3): 287–295.
24. Potenza MN, Leung HC, Blumberg HP, et al. An fMRI
Stroop task study of ventromedial prefrontal cortical func-
tion in pathological gamblers. Am J Psychiatry. 2003; 160(11):
1990–1994.
25. Bush G, Frazier JA, Rauch SL, et al. Anterior cingulate
cortex dysfunction in attention-deficit/hyperactivity disorder
revealed by fMRI and the counting Stroop. Biol Psychiatry.
1999; 45(12): 1542–1552.
26. Fan LY, Gau SSF, Chou TL. Neural correlates of inhibitory
control and visual processing in youths with attention deficit
hyperactivity disorder: A counting Stroop functional MRI
study. Psychol Med. 2014; 44(12): 2661–2671.
27. Bolla K, Ernst M, Kiehl K, etal. Prefrontal cortical dysfunc-
tion in abstinent cocaine abusers. J Neuropsychiatry Clin
Neurosci. 2004; 16(4): 456–464.
28. Nordahl TE, Carter CS, Salo RE, etal. Anterior cingulate me-
tabolism correlates with Stroop errors in paranoid schizophrenia
patients. Neuropsychopharmacology. 2001; 25(1): 139–148.
29. Wardle MC, Gonzalez R, Bechara A, Martin-Thormeyer
EM. Iowa gambling task performance and emotional distress
interact to predict risky sexual behavior in individuals with
dual substance and HIV diagnoses. J Clin Exp Neuropsychol.
2010; 32: 1110–1121.
30. Gonzalez R, Vassileva J, Bechara A, etal. The influence of
executive functions, sensation seeking, and HIV serostatus on
the risky sexual practices of substance-dependent individuals.
J Int Neuropsychol Soc. 2005; 11(2): 121–131.
31. Patton JH, Stanford MS, Barratt ES. Factor structure of
the Barratt Impulsiveness Scale. J Clin Psychol. 1995; 51:
768–774.
32. Eysenck SBG, Pearson PR, Easting G, Allsopp JF. Age
norms for impulsiveness, venturesomeness and empathy in
adults. Pers Individ Dif. 1985; 6: 613–619.
33. Patrick CJ, Curtin JJ, Tellegen A. Development and valid-
ation of a brief form of the multidimensional personality
questionnaire. Psychol Assess. 2002; 14: 150–163.
34. Reynolds B, Ortengren A, Richards JB, de Wit H. Dimensions
of impulsive behavior: Personality and behavioral measures.
Pers Individ Dif. 2006; 40: 305–315.
35. Zapolski TCB, Cyders MA, Smith GT. Positive urgency
predicts illegal drug use and risky sexual behavior. Psychol
Addict Behav. 2009; 23(2): 348–354.
36. Deckman T, DeWall NC. Negative urgency and risky sexual
behaviors: Aclarification of the relationship between impul-
sivity and risky sexual behavior. Pers Individ Dif. 2011; 51(5):
674–678.
37. Chester DS, Lynam DR, Milich R, Powell DK, Andersen AH,
DeWall CN. How do negative emotions impair self-control?
Aneural model of negative urgency. NeuroImage. 2016; 132:
43–50.
38. Lynam DR, Smith GT, Whiteside SP, Cyders MA. The
UPPS-P: Assessing five personality pathways to impul-
sive behavior. Technical report. West Lafayette, IN: Purdue
University; 2006.
39. Jenkinson M, Bannister P, Brady M, Smith S. Improved op-
timization for the robust and accurate linear registration and
motion correction of brain images. NeuroImage. 2002; 17(2):
825–841.
40. Jenkinson M, Smith S. A global optimisation method for
robust affine registration of brain images. Med Image Anal.
2001; 5(2): 143–156.
41. Andersson JLR, Jenkinson M, Smith S. Non-linear regis-
tration, aka spatial normalisation. FMRIB technical report
TR07JA2. Oxford, UK: FMRIB Centre; 2010.
42. Beckmann CF, Smith SM. Probabilistic independent com-
ponent analysis for functional magnetic resonance imaging.
IEEE Trans Med Imaging. 2004; 23(2): 137–152.
43. Griffanti L, Salimi-Khorshidi G, Beckmann CF, et al. ICA-
based artefact removal and accelerated fMRI acquisition for
improved resting state network imaging. NeuroImage. 2014; 95:
232–247.
44. Salimi-Khorshidi G, Douaud G, Beckmann CF, Glasser MF,
Griffanti L, Smith SM. Automatic denoising of functional
MRI data: Combining independent component analysis
and hierarchical fusion of classifiers. NeuroImage. 2014; 90:
449–468.
45. Beckmann CF, Jenkinson M, Smith SM. General multilevel
linear modeling for group analysis in FMRI. NeuroImage.
2003; 20(2): 1052–1063.
46. Woolrich M. Robust group analysis using outlier inference.
NeuroImage. 2008; 41(2): 286–301.
47. Woolrich MW, Behrens TEJ, Beckmann CF, Jenkinson M,
Smith SM. Multilevel linear modelling for FMRI group
analysis using Bayesian inference. NeuroImage. 2004; 21(4):
1732–1747.
48. Eklund A, Nichols TE, Knutsson H. Cluster failure: Why
fMRI inferences for spatial extent have inflated false-positive
rates. PNAS. 2016; 113(28): 7900–7905.
49. Winkler A, Ridgway G, Webster M, Smith S, Nichols
T. Permutation inference for the general linear model.
NeuroImage. 2014; 92: 381–397.
50. Hayes AF. Introduction to Mediation, Moderation and
Conditional Process Analysis: A Regression-Based Approach.
New York, NY: Guilford Press; 2013.
51. Yarkoni T. Neurosynth core tools v0.3.1 [Data set]. Zenodo.
2014. Available at http://doi.org/10.5281/zenodo.9925.
Accessibility verified December 12, 2017.
52. Yarkoni T, Poldrack RA, Nichols TE, Van Essen DC, Wager
TD. Large-scale automated synthesis of human functional
neuroimaging data. Nat Methods. 2011; 8: 665–670.
53. Sagar KA, Dahlgren MK, Gönenç A, Racine MT, Dreman
MW, Gruber SA. The impact of initiation: Early onset ma-
rijuana smokers demonstrate altered Stroop performance
and brain activation. Dev Cog Neurosci. 2015; 16: 84–92.
54. Peterson BS, Potenza MN, Wang Z, etal. An fMRI study of
the effects of psychostimulants on default-mode processing
during Stroop task performance in youths with ADHD. Am
J Psychiatry. 2009; 166: 1286–1294.
55. Salo R, Ursu S, Buonocore MH, Leamon MH, Carter C.
Impaired prefrontal cortical function and disrupted adaptive
cognitive control in methamphetamine abusers: Afunctional
magnetic resonance imaging study. Biol Psychiatry. 2009;
65(8): 706–709.
56. Nee DE, Wager TD, Jonides J. Interference resolution:
Insights from a meta-analysis of neuroimaging tasks. Cogn
Affect Behav Neurosci. 2007; 7(1): 1–17.
57. Wagner G, Sinsel E, Sobanski T, etal. Cortical inefficiency
in patients with unipolar depression: An event-related fMRI
study with the Stroop task. Biol Psychiatry. 2006; 59(10):
958–965.
12 ann. behav. med. (2018) XX:1–13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
58. Whalen PJ, Bush G, McNally RJ, etal. The emotional count-
ing Stroop paradigm: Afunctional magnetic resonance im-
aging probe of the anterior cingulate affective division. Biol
Psychiatry. 1998; 44: 1219–1228.
59. Azizian A, Nestor LJ, Payer D, Monterosso JR, Brody
AL, London ED. Smoking reduces conflict-related anterior
cingulate activity in abstinent cigarette smokers perform-
ing a Stroop task. Neuropsychopharmacology. 2010; 35:
775–782.
60. Basten U, Stelzel C, Fiebach CJ. Trait anxiety modulates
the neural efficiency of inhibitory control. J Cogn Neurosci.
2011; 23(10): 3132–3145.
61. Poldrack RA. Is “efficiency” a useful concept in cognitive
neuroscience? Dev Cog Neurosci. 2015; 11: 12–17.
62. Cho SS, Pellecchia G, Aminian K, etal. Morphometric cor-
relation of impulsivity in medial prefrontal cortex. Brain
Topogr. 2013; 26: 479–487.
63. Moreno-López L, Catena A, Fernández-Serrano MJ, etal.
Trait impulsivity and prefrontal gray matter reductions in
cocaine dependent individuals. Drug Alcohol Depend. 2012;
125: 208–214.
64. Muhlert N, Lawrence AD. Brain structure correlates of
emotion-based rash impulsivity. NeuroImage. 2015; 115:
138–146.
65. Tahmasian M, Rochhausen L, Maier F, etal. Impulsivity is
associated with increased metabolism in the fronto- insular
network in Parkinson’s disease. Front Behav Neurosci. 2015;
9: 317.
66. Valdes IH, Steinberg JL, Narayana PA, etal. Impulsivity and
BOLD fMRI activation in MDMA users and healthy control
subjects. Psychiatry Res Neuroimaging. 2006; 147(2–3): 239–242.
67. Brown SM, Manuck SB, Flory JD, Hariri A. Neural basis of
individual differences in impulsivity: Contributions of corti-
colimbic circuits for behavioral arousal and control. Emotion.
2006; 6(2): 239–245.
68. Horn NR, Dolan M, Elliott R, Deakin JF, Woodruff PW.
Response inhibition and impulsivity: An fMRI study.
Neuropsychologia. 2003; 41(14): 1959–1966.
ann. behav. med. (2018) XX:1–13 13
Downloaded from https://academic.oup.com/abm/advance-article-abstract/doi/10.1093/abm/kax019/4837294
by guest
on 05 February 2018
