The Contribution of Neural Reward Signals to Development of Cognitive Control in Early Adolescence PDF Free Download
1 / 60/60
100%
The Contribution of Neural Reward Signals to Development of Cognitive Control in Early
Adolescence
A Master’s Thesis
Presented to
The Faculty of the Graduate School of Arts and Sciences
Brandeis University
Department of Psychology
James Howard, Advisor
In Partial Fulfillment
of the Requirements for the Degree
Master of Arts
by
Xukun Guo
August 2024
ii
ACKNOWLEDGMENTS
I would like to express my deepest gratitude to my academic advisor, Dr. James Howard,
who not only provided crucial guidance on neuroscience-related topics during my thesis
writing but also created an environment that helped me quickly adapt to the native culture
during my two years of study and research as a master's student at Brandeis University.
Guidance from Dr. Howard was a fundamental pillar of my academic success.
I want to extend my deepest appreciation to my Second Reader, Dr. Hannah Snyder. Her
patience and insightful feedback throughout my thesis writing process significantly enhanced
my research skills, which will benefit my life in the long term. Additionally, the Writing
Group she organized for graduate students was incredibly helpful in navigating the challenges
I faced in my research as a student.
I am also grateful to my instructors and the professors I collaborated with as a Course
Assistant, including Dr. Xiaodong Liu, Dr. Teresa Mitchell, Dr. Jutta Wolf, and Dr. Hannah
Clark. My growth would not have been possible without their support and dedication to
teaching.
Many thanks to the faculty and staff of the Psychology Department, my peers, lab members,
and everyone behind the scenes who created a supportive and loving environment for my
studies at Brandeis.
Lastly, I’d like to mention my parents, my husband, and my friends, who always provided me
with emotional support and shared my joy. Special thanks to my rabbit friend, who often
warmed my heart by showing how happy she was with a piece of banana.
My gratitude to all of you will last a lifetime.
iii
ABSTRACT
The Contribution of Neural Reward Signals to Development of Cognitive Control in Early
Adolescence
A thesis presented to the Faculty of the
Graduate School of Arts and Sciences of Brandeis University
Waltham, Massachusetts
By Xukun Guo
Reward sensitivity and cognitive control are crucial for making wise decisions, but their
developmental relationship during adolescence is not fully understood. The dual-system
model suggests that these processes develop independently, whereas the imbalanced model
claims that the robust motivational signals regarding reward provide essential inputs to build
cortical-subcortical connections that facilitate cognitive control. The central aim of this study
is to investigate the longitudinal relationship between neural reward sensitivity and cognitive
control across adolescent development, by examining hypotheses corresponding to three
research questions regarding adolescents from 9 to 14 years old: Does neural reward
sensitivity predict cognitive control during development (n=915)? Does the development of
neural reward sensitivity predict cortical-subcortical functional connectivity supporting
cognitive control (n=1513)? Does cognitive control-related cortical-subcortical connectivity
mediate the relationship between neural reward sensitivity and cognitive control (n=1207)?
To answer these questions, we specified latent growth curve models and mediation models
based on the longitudinal fMRI imaging and behavioral data provided by the Adolescent
Brain Cognitive Development SM Study (ABCD Study®) database. To quantify reward
sensitivity, cognitive control, and cortical-subcortical connectivity, we focused on
iv
measurements taken at three time points from ages 9-10 to 13-14. Reward sensitivity was
measured by the blood-oxygen-level-dependent (BOLD) signal evoked by reward
anticipation and receipt during the Monetary Incentive Delay (MID) task. Cognitive control
was measured by the behavioral performance during the Stop Signal Task (SST). Cortical-
subcortical functional connectivity measures were derived from resting state fMRI scanning
sessions. Two major findings were observed. First, individuals who maintained a relatively
high level of right nucleus accumbens (NAcc) neural reward sensitivity showed faster
improvement in cognitive control performance over time, which partially supported our
hypothesis that the slope of reward sensitivity measured during reward feedback would be
positively correlated with the slope of cognitive control. Second, a higher left NAcc baseline
reward sensitivity during reward anticipation was associated with a faster strengthening of
Frontoparietal network (FPN)-left NAcc functional connectivity, which partially supported
our hypothesis that the intercept of reward sensitivity measured during reward anticipation
would be positively correlated with the slope of resting state FPN-VS functional connectivity.
These findings suggest that reward sensitivity significantly predicts the longitudinal
development of cognitive control in adolescents. These findings enhanced our understanding
of the adaptive function of reward sensitivity and helped differentiate between the dual
system model and the imbalance model.
Keywords: Reward sensitivity, neural reward signal, cognitive control, adolescent
development
v
TABLE OF CONTENTS
List of Tables ...........................................................................................................................vii
Introduction ................................................................................................................................ 1
Potential Developmental Trajectories of Reward Sensitivity and Cognitive Control during
Adolescence ........................................................................................................................... 2
Relationship between Reward Sensitivity and Cognitive Control ......................................... 5
Current study .......................................................................................................................... 8
Methods...................................................................................................................................... 9
Participants ............................................................................................................................. 9
Data Availability .................................................................................................................. 10
Measures .............................................................................................................................. 10
Measures of reward sensitivity ........................................................................................ 10
Measures of cognitive control .......................................................................................... 11
Measures of cortical-subcortical functional connectivity ................................................ 12
Measures of demographic variables................................................................................. 12
Statistical Analysis ............................................................................................................... 13
Models.............................................................................................................................. 13
Covariates ........................................................................................................................ 14
Inclusion/exclusion criteria .............................................................................................. 15
Missing values ................................................................................................................. 15
Outliers ............................................................................................................................. 15
Results ...................................................................................................................................... 16
Demographic statistics ......................................................................................................... 16
Question 1: Does neural reward sensitivity positively predict cognitive control during
development from 9 to 14 years old? ................................................................................... 16
vi
Question 2: Does the development of neural reward sensitivity predict cortical-subcortical
functional connectivity supporting cognitive control? ........................................................ 19
Question 3: Does cognitive control-related cortical-subcortical connectivity mediate the
relationship between neural reward sensitivity and cognitive control? ............................... 21
Discussion ................................................................................................................................ 24
Question 1: Does neural reward sensitivity positively predict cognitive control during
development from 9 to 14 years old? ................................................................................... 24
Question 2: Does the development of neural reward sensitivity predict cortical-subcortical
functional connectivity supporting cognitive control? ........................................................ 27
Question 3: Does cognitive control-related cortical-subcortical connectivity mediate the
relationship between neural reward sensitivity and cognitive control? ............................... 29
Limitations ............................................................................................................................... 29
Conclusion ............................................................................................................................... 30
References ................................................................................................................................ 32
Appendix A .............................................................................................................................. 38
Appendix B .............................................................................................................................. 46
vii
LIST OF TABLES
Table 1 Demographic information of participants included in the current study .................... 16
Table 2 Descriptive statistics for sample 1 corresponding to Question 1 ................................ 17
Table 3 Means and variance of the unconditional LGCMs corresponding to Question 1 ....... 18
Table 4 Model fit statistics of the LGCMs corresponding to Question 1 ................................ 18
Table 5 Results of the LGCMs corresponding to Question 1 .................................................. 18
Table 6 Descriptive statistics for sample 2 corresponding to Question 2 ................................ 19
Table 7 Means and variance of the unconditional LGCMs corresponding to Question 2 ....... 20
Table 8 Model fit statistics of the LGCMs corresponding to Question 2 ................................ 20
Table 9 Results of the LGCMs corresponding to Question 2 .................................................. 21
Table 10 Descriptive statistics for sample 3 corresponding to Question 3 .............................. 22
Table 11 Path coefficients in the mediation models for Question 3 ........................................ 23
1
Introduction
Reward sensitivity refers to the degree to which an individual’s behavior is motivated by
reward-relevant stimuli (Kim et al., 2015). Sensitivity to rewards is essential for human
beings to learn from experience and acquire basic motivated behaviors such as obtaining food
(Kidd & Loxton, 2018). Through iterative processes of obtaining rewards and evaluating
outcomes, organisms can adaptively shape their behavior by using predictive information in
the environment to anticipate the best course of action (Rangel et al., 2008). However,
cognitive control systems are also an essential component of the decision process, in that they
serve as a check on reward-seeking behavior by enabling organisms to evaluate long-term
goals and avoid potential dangers (Bari & Robbins, 2013).
Thus, developing a balanced relationship between reward sensitivity and cognitive
control is an important task for the human brain. In the process of human development,
adolescence is a unique stage in which reward sensitivity peaks (Telzer, 2016), while
cognitive control ability is slowly developing (Ordaz et al., 2013). These changes might
contribute to behavioral alterations. Indeed, literature suggests adolescent behaviors are
affected more by motivational factors rather than by calculations of outcome probabilities
based on cognitive processing (Luciana & Collins, 2012), and adolescence marks the onset of
many risk-taking behaviors (Willoughby et al., 2021). Across the world, adolescents exhibit a
stronger propensity for risk-taking than other age groups (Duell et al., 2018). This trend even
extends to non-human primates, such that adolescent chimpanzees also have a greater
inclination towards risk-taking when anticipating rewards than adult chimpanzees (Rosati et
al., 2023). While risk-taking behavior is typically viewed as detrimental to health and well-
being, risk-taking in adolescence can be developmentally adaptative (Crone & Dahl, 2012;
Duell & Steinberg, 2021). However, overly high-level risk-taking such as dangerous driving
or firearm use can of course be harmful (Cunningham et al., 2018). It is therefore important
2
to understand how developmental trajectories of reward sensitivity and cognitive control
interact during human adolescence.
Potential Developmental Trajectories of Reward Sensitivity and Cognitive Control
during Adolescence
Reward sensitivity has been studied using various measures in prior studies. For
example, perceived reward sensitivity can be measured by self-report questionnaires such as
the Behavioral Inhibition System/Behavioral Activation System (BIS/BAS) scales (Carver &
White, 1994) or its adapted version (Barch et al., 2018). Reward sensitivity can also be
measured by performance in incentive-related tasks such as the Monetary Incentive Delay
(MID) task (Knutson et al., 2000). At the neural level, reward sensitivity can be defined as
the strength of blood oxygen level-dependent (BOLD) functional magnetic resonance
imaging (fMRI) signal measured during the reward anticipation or feedback stage of the MID
task. In the current research, we mainly focus on neural reward sensitivity measured in the
ventral striatum, specifically, the nucleus accumbens (NAcc). The NAcc is particularly
relevant for reward processing, as it undergoes dramatic dopaminergic development during
adolescence (Galvan, 2010), and dopaminergic changes are considered the biological
foundation of changes in reward sensitivity in adolescence (Van Leijenhorst et al., 2010).
Studies on the developmental trajectory of reward sensitivity revealed that adolescents tend to
show greater NAcc activity in reward processing compared to children and adults (Galvan et
al., 2006; Luciana et al., 2012).
Cognitive control includes diverse yet partially overlapping components such as
inhibition, cognitive shifting, and working memory updating (Friedman & Robbins, 2022).
Aligning with the viewpoint that there are shared factors across these components, a meta-
analysis of 193 fMRI studies showed broad patterns of activation across the prefrontal, dorsal
anterior cingulate, and parietal cortices for tasks designed to measure any component of
3
cognitive control (Niendam et al., 2012). In the context of motivation, research on cognitive
control often focuses on inhibitory control, measured using tasks such as the Go/No Go task
(Simmonds et al., 2008) and the Stop signal task (SST) (Boucher et al., 2007; Chaarani et al.,
2021). Although inhibitory control is not the only component of cognitive control, it is a
foundational one (Verbruggen et al., 2014; Weigard et al., 2023; Weiss & Luciana, 2022).
For example, updating working memory may require inhibition to ignore irrelevant
information (Zacks & Hasher, 1994). Also, a latent factor modeling approach that used a
higher order "common" cognitive control factor to predict inhibiting, updating, and shifting
also indicated an almost perfect prediction of the "common" factor on inhibiting but not on
the other two components (Friedman et al., 2008). This evidence suggests inhibition may be a
potential common factor in any cognitive control component (Friedman & Robbins, 2022).
However, it is important to note that the term "inhibition" used across various contexts may
not always relate to actual neural inhibition (Aron, 2007; Friedman & Miyake, 2004;
Friedman & Robbins, 2022). The developmental trajectory of cognitive control tends to
follow a slowly increasing pattern from childhood to early adulthood (Constantinidis & Luna,
2019; Ordaz et al., 2013). Regarding the neural foundation, frontoparietal network (FPN) and
cingulo-opercular network (CON) are considered major networks supporting cognitive
control (Dosenbach et al., 2008). A meta-analysis focused on 49 studies involving 6- to 18-
year-old individuals suggested that children and adolescents exhibit activation in the FPN and
CON when performing various cognitive control tasks. Notably, in the 8-12 age group, as
well as in the overall sample, no brain regions were found to exclusively support inhibition
without also supporting the putative unitary "common executive," and vice versa. (McKenna
et al., 2017) This suggests that inhibition and general cognitive control abilities are
indistinguishable from each other during late childhood and early adolescence. This
characteristic highlights the uniqueness of inhibitory control ability in adolescence, as it may
4
better reflect general cognitive control ability in this age range compared to other cognitive
control components.
There are two prominent models that focus on the dynamic relationship between reward
sensitivity and cognitive control during adolescence. They are the dual-system model
(Steinberg, 2008; Steinberg et al., 2018) and the imbalance model (Casey, 2015; Casey et al.,
2016). Both models attribute behavioral changes in self-control in adolescence to the
differential developmental trajectories of two brain systems: (1) The "bottom-up" incentive-
processing system, mainly consisting of subcortical reward and threat processing regions
such as the ventral striatum, matures relatively fast; (2) The "top-down" cognitive control
system, mainly consisting of lateral prefrontal cortex and lateral parietal cortex, matures
relatively slowly (Shulman et al., 2016). According to both models, reward sensitivity peaks
around mid-adolescence, and the cognitive control system continues to develop until
adulthood (Shulman et al., 2016).
However, there are also major differences between the two models. First, the dual
system model considers reward sensitivity and cognitive control as two competing,
orthogonal systems that follow divergent trajectories. Under this viewpoint, adolescents are
increasingly motivated to pursue rewards (i.e., adolescents have an increased reward
sensitivity), yet their developing cognitive control system does not have enough ability to
restrain potentially dangerous inclinations (Shulman et al., 2016; Steinberg, 2008; Steinberg
et al., 2018). This situation continues until reward sensitivity starts to decline and the
cognitive control system matures (Steinberg, 2008; Steinberg et al., 2018). On the other hand,
the imbalance model views the brain as distributed interactive networks that are continuously
refined by experience (Casey, 2015). This model stipulates that subcortical reward networks
and cortical cognitive control networks do not develop independently, but dynamically
interact (Casey, 2015; Casey et al., 2016; Somerville et al., 2010). The imbalance model
5
proposes that self-control during adolescence relies on the refining of connections between
subcortical limbic circuits and prefrontal network, and motivation signals have interactive
effects on cognitive control during adolescent development (Casey, 2015; Casey et al., 2008).
From this account, a heightened function of subcortical regions such as ventral striatum and
amygdala are necessary in adolescence, as the robust signals help build a connection in
cortical-subcortical circuits, such that individuals gradually obtain "top-down" control on
their behavior (Casey et al., 2019). That is, although heightened reward sensitivity at an early
age allows individuals to obtain new experiences through engaging in risky behaviors, the
robust motivational signals provide essential inputs to build cortical-subcortical connections
that facilitate cognitive control that benefits the individual in the long term.
Overall, the two models have differing views on when reward sensitivity peaks during
adolescence. More importantly, they differ fundamentally in how they view the relationship
between reward sensitivity and cognitive control: are reward sensitivity and cognitive control
abilities developing completely independently, or does the early development of reward
sensitivity facilitate the development of cognitive control abilities? To address this question,
it is meaningful to focus on the longitudinal relationship between reward sensitivity and
cognitive control in early adolescence. This approach allows us to capture the early changes
in reward sensitivity during this critical developmental period.
Relationship between Reward Sensitivity and Cognitive Control
Many cross-sectional studies on adolescents have shown changes in cognitive control in
contexts where rewards are present, though the directions of these changes vary (Luna et al.,
2013; Padmanabhan et al., 2011; Somerville et al., 2011). For example, one study using a
Go/No Go task with happy faces or calm faces showed that, although cognitive control to
calm faces (neutral cues) improves linearly in three age groups (children, adolescents, and
adults), it decreases in the adolescent group (aged 13-17) when happy faces (appetitive cues)
6
are presented instead (Somerville et al., 2011). This decrease in cognitive control was
observed together with increased activation in the ventral striatum. However, because reward
sensitivity and cognitive control were measured in the same task in this study, there is a
possibility that the measured brain activity or behavioral performance reflects a mixture of
reward sensitivity and cognitive control processes rather than separate systems interacting.
Using similar approaches, limited longitudinal studies suggest that cognitive control
gradually strengthens during adolescence, and the presence of rewards can enhance cognitive
control performance (Parr et al., 2022). These findings support the possibility that reward
sensitivity plays a positive role in cognitive control development.
On the other hand, some longitudinal non-imaging studies that separately measured
reward sensitivity and cognitive control suggest that reward sensitivity and cognitive control
follow independent developmental trajectories, although they used different terminologies.
For example, the dual-system model considers sensation seeking and self-regulation as the
behavioral level form of reward sensitivity and cognitive control (Steinberg, 2008; Steinberg
et al., 2018). One study found that sensation seeking is correlated with pubertal status, yet
self-regulation is only correlated with age but not pubertal status (Steinberg, 2008).
Consistent with this finding, another study examined the relationship between self-reported
impulsive control and sensation seeking in individuals ranging from 10-25 years old, and
found that heightened impulse control cannot predict the decrease of sensation seeking
(Shulman et al., 2014). With inconsistent research findings and sometimes the procedure of
measuring cognitive control and reward sensitivity within the same task, the precise
longitudinal relationship between these two constructs remains unclear.
Investigating whether the cortical-subcortical connections supporting cognitive control
change with the development of reward sensitivity could also provide insights into the
relationship between reward sensitivity and cognitive control. As previously mentioned, the
7
frontoparietal network (FPN) and the cingulo-opercular network (CON) are two functional
networks that support top-down cognitive control (Dosenbach et al., 2008). Studies using
data from the Human Connectome Project have found that cognitive control is related to FPN
activity and the thickness of the gray matter in the CON (Lerman-Sinkoff et al., 2017).
In prior literature, an increased functional coupling was found between the right
intraparietal sulcus and bilateral NAcc and putamen during reward anticipation (Padmala &
Pessoa, 2011). This suggests that facing rewards may provide the cortical-subcortical
pathway with opportunities to strengthen. From the perspective of the imbalance model, the
significant role of encountering incentives during adolescence, when reward sensitivity is
heightened, is that robust motivational signals provide essential inputs to build cortical-
subcortical connections that facilitate cognitive control (Casey et al., 2016). In other words, it
provides opportunities to "practice" establishing the top-down control pathway. In animal
research, a study using optogenetic stimulation on adolescent mice found that plastic
dopamine neuron activity in the ventral tegmental area induced the formation of mesofrontal
axonal boutons (Mastwal et al., 2014). Although dopamine signaling is not equivalent to
reward signaling, this result demonstrates the positive impact of subcortical signals on
cortical development during adolescence.
As a result of strengthened cortical-subcortical connection, one study revealed that the
development of top-down connectivity from frontal cognitive control regions to downstream
cortical and subcortical regions (although not specifically with NAcc) supports the
improvement of inhibitory control (Hwang et al., 2010). Under this context, a scientific
hypothesis could be that if the development of reward sensitivity, as indicated by NAcc
activity in the MID task, contributes to the development of cognitive control, then functional
connectivity between the NAcc and cognitive control-related networks may be strengthened
8
in response to strong signal inputs from the NAcc. This enhanced connectivity could serve as
a mediator in the relationship between reward sensitivity and cognitive control.
In summary, future studies that measure reward sensitivity and cognitive control
separately and collect a sufficient amount of longitudinal neuroimaging data reflecting neural
reward sensitivity (i.e., motivational signals toward incentives) and FPN/CON-NAcc
functional connectivity may shed light on the relationship between reward sensitivity and
cognitive control.
Current study
In the current research, we addressed these limitations by using a latent growth curve
modeling (LGCM) approach to longitudinal data to first examine whether neural reward
sensitivity positively predicts cognitive control during development from 9 to 14 years old
(Question 1). Additionally, we tested whether the development of neural reward sensitivity
predict FPN/CON-NAcc functional connectivity supporting cognitive control (Question 2),
and whether FPN/CON-NAcc connectivity mediates the relationship between neural reward
sensitivity and cognitive control (Question 3). To test these questions we analyzed data
generated by the Adolescent Brain Cognitive Development (ABCD) Study (Jernigan, n.d.),
which is the largest long-term study of brain development and child health in the US
(Adolescent Brain Cognitive Development, n.d.-b). The advantage of using this dataset is that
neural reward sensitivity and cognitive control are quantified separately and longitudinally.
Additionally, the dataset provides a large amount of neuroimaging data on adolescents, which
helps to increase statistical power to detect significant findings. Results of this study may
deepen our understanding of the adaptive function of reward sensitivity and help differentiate
between the dual-system model and the imbalance model.
9
Methods
Participants
The current study used the ABCD Data Release 5.1, derived from the Adolescent Brain
Cognitive Development (ABCD) Study, which focuses on adolescent development. This
dataset represents the entire ABCD cohort (n=11,880 at baseline) except for information of
12 participants who withdrew consent to share their data. Participants were initially recruited
at ages 9-10 years and their development would be followed for ten years using various
assessments across multiple time points. ABCD study follows a strict sampling strategy to
increase the representation of the baseline cohort sample of the U.S. population of children
(Garavan et al., 2018). Specifically, participants were recruited from public and private
primary schools, primarily using the multi-stage probability sampling approach across
schools within 21 ABCD study sites well-distributed in the US. To reduce the impact of
factors beyond the researchers' control on participant recruitment (such as some schools or
families refusing to participate), the ABCD study also monitored the accumulating sample
during recruitment and made dynamic adjustments to include more participants from under-
recruited populations, in order to create a demographic profile that matched national
proportions.
Our analysis concentrated on the data collected during three time points: the baseline
(n=11,868, collected during 2016-2018), 2-year follow-up (n=10,973, collected during 2018-
2020), and 4-year follow-up (n=4754, collected during 2020-2022). The 4-year follow-up
data collection was ongoing when data for Release 5.1 was frozen, so only a part of the
participants (n=4754) had their 4-year follow-up data documented in this dataset. Based on
the data inclusion/exclusion criteria (see the Inclusion/Exclusion Criteria section under
Statistical Analysis), the sample sizes for each research question were as follows: 915 for
Question 1, 1,513 for Question 2, and 1,207 for Question 3. Table 1 shows the demographics
10
of each sample.
Data Availability
Data used in the preparation of this article were obtained from the Adolescent Brain
Cognitive Development SM (ABCD) Study (https://abcdstudy.org), held in the NIMH Data
Archive (NDA). The ABCD data repository grows and changes over time. The ABCD data
used in this report came from http://dx.doi.org/10.15154/z563-zd24. This dataset is publicly
accessible to qualified researchers with a valid research use of the data. The ABCD Data
Release 5.1 provides information about the principle of naming and structure of data tables
(Adolescent Brain Cognitive Development, 2024b). Also, an interactive data dictionary
explorer application named ABCD Data Dictionary is available online to identify variables
measured in the ABCD Study (Adolescent Brain Cognitive Development, n.d.-a).
Measures
Measures of reward sensitivity
In the current study, reward sensitivity was measured by the BOLD signal evoked by
reward anticipation and receipt during the MID task. The MID task is widely used to measure
the brain's functional activation when individuals anticipate and respond to monetary
incentives. In the ABCD study, each trial in the MID task starts with a pseudorandomly
ordered incentive cue (2000 ms) indicating one of five outcomes: Win $0.20, Win $5, Lose
$0.20, Lose $5, or $0-no money at stake. After a varied anticipation event (1500-4000 ms), a
target appears (150-500 ms) for participants to respond and potentially win or avoid loss. In
this task, for a participant to win money or avoid losses, they must make quick key-press
responses to the visual stimuli on every trial, which would be a circle in the win trial, a square
in the lose trial, or a triangle in the neutral trial. They then receive feedback message based
on their response, informing the outcome of the trial (e.g., "You won $5.00"). The duration of
the message is 2000 ms minus the target's duration. The MID task conducted in the ABCD
11
study consists of 2 runs and each run has 50 sequential trials, with 10 for each type. Each run
lasts 5:42. In the ABCD study, this task takes about 12 minutes. As for the success rate, the
adaptive algorithm adjusts the length of the target duration to maintain a hit rate in the range
of 50-60% by making it harder or easier for a participant to respond in time.
Specifically, the variables we used to measure reward sensitivity were the standardized
beta weights for both "all anticipation of reward versus neutral contrast" and "all reward
positive versus negative feedback contrast" estimated from general linear models and
averaged across voxels in the ventral striatum. These beta weights were provided in the
ABCD Release 5.1. The corresponding regions of interest (ROIs) were the NAcc in each
hemisphere of the brain. In the ABCD Release 5.1, these ROIs were labeled using atlas-based
segmentation (Fischl et al., 2002). Participants were expected to complete the MID tasks at
all three time points (baseline, 2-year, and 4-year follow-up).
Measures of cognitive control
Due to our aim of exploring the basic components behind risk-taking, in the current
study, cognitive control was narrowly constrained to inhibitory control and measured by the
behavioral performance during the SST (Boucher et al., 2007). The SST is used to measure
impulse control by requiring participants to inhibit a button press response to a left or right-
facing arrow "Go" signal when the “Go” signal is followed by an unexpected upward-facing
arrow "Stop" signal. Both the "Go" trial and the "Stop" trial last 1000ms, and each inter-trial
interval lasts 700-2000ms. The SST used in the ABCD study consists of 2 runs and each run
has 180 sequential trials, with 30 in each run being "Stop" trials. Each run lasts 5:49. In the
ABCD study, this task takes about 13 minutes. The success rate of inhibition is controlled at
approximately 50%, as the adaptive algorithm adjusts the interval between the "Go" signal
and the "Stop" signal.
12
Specifically, the variable used to measure cognitive control is the Stop Signal Reaction
Time (SSRT) during the Stop Signal Task (SST). The SSRTs applied in the current study
were computed using an integration method (Adolescent Brain Cognitive Development,
2024a; Verbruggen et al., 2019). Regarding the reliability of SST, although adult samples
yielded variable reliability, adolescent studies showed more consistent results (Weiss &
Luciana, 2022). For example, in one study that included participants with similar ages in the
current analysis, the split-half reliability of SST for both early adolescence (ages 9–12, n = 41,
SB2r = 0.86) and mid-adolescence (ages 13–17, n = 50, SB2r = 0.91) was excellent (Williams
et al., 1999). Participants were expected to complete the SST at all three time points (baseline,
2-year, and 4-year follow-up). The order of participating in the SST and MID tasks for each
participant is randomized.
Measures of cortical-subcortical functional connectivity
In the current study, cortical-subcortical functional connectivity measures were derived
from resting state fMRI scanning sessions completed by the ABCD study participants.
Resting state fMRI was conducted before task-based fMRI in two 10-minute scanning
sessions. The brain network segmentation was based on prior literature (Gordon et al., 2016).
The variables used to measure resting-state cortical-subcortical functional connectivity
were "correlation between frontoparietal network and ventral striatum activations" and
"correlation between cingulo-opercular network and ventral striatum activations". These
correlation coefficients were provided in the ABCD Release 5.1. Specifically, the ventral
striatum consisted of the NAcc in each hemisphere of the brain. Participants were expected to
complete the resting state fMRI scanning at all three time points (baseline, 2-year, and 4-year
follow-up).
Measures of demographic variables
13
Baseline age was treated as the covariate in the current study, identified as the
participant's age in months at baseline. Due to model convergence issues, demographics such
as sex at birth, race, ethnicity, and socioeconomic status could not be included as covariates.
Statistical Analysis
Models
Based on our three research questions and data measured at three time points (baseline,
2-year-follow-up, and 4-year-follow-up), we used latent growth curve models to answer the
first two questions, and a mediation model to answer the third question. The statistical
toolbox R was used to clean the data and to output descriptive statistics, and Mplus was used
to generate the results of the specified models.
Hypotheses corresponding to Question 1 are as follows: 1a) The slope of reward
sensitivity measured during reward anticipation is positively correlated with the slope of
cognitive control. 1b) The intercept of reward sensitivity measured during reward anticipation
is positively correlated with the slope of cognitive control. 1c) The intercept of cognitive
control is positively correlated with the slope of reward sensitivity measured during reward
anticipation. 1d) The slope of reward sensitivity measured during reward feedback is
positively correlated with the slope of cognitive control. 1e) The intercept of reward
sensitivity measured during reward feedback is positively correlated with the slope of
cognitive control. 1f) The intercept of cognitive control is positively correlated with the slope
of reward sensitivity measured during reward feedback. See Figure A1 and Figure A2 in
Appendix A for the graphical presentations of modeling components.
Hypotheses corresponding to Question 2 are as follows: 2a) The slope of reward
sensitivity measured during reward anticipation is positively correlated with the slope of
resting state FPN-VS functional connectivity. 2b) The intercept of reward sensitivity
measured during reward anticipation is positively correlated with the slope of resting state
14
FPN-VS functional connectivity. 2c) The slope of reward sensitivity measured during reward
anticipation stages is positively correlated with the slope of resting state CON-VS functional
connectivity. 2d) The intercept of reward sensitivity measured during reward anticipation
stages is positively correlated with the slope of resting state CON-VS functional connectivity.
2e) The slope of reward sensitivity measured during reward feedback stages is positively
correlated with the slope of resting state FPN-VS functional connectivity. 2f) The intercept of
reward sensitivity measured during reward feedback stages is positively correlated with the
slope of resting state FPN-VS functional connectivity. 2g) The slope of reward sensitivity
measured during reward feedback stages is positively correlated with the slope of resting
state f CON-VS functional connectivity. 2h) The intercept of reward sensitivity measured
during reward feedback stages is positively correlated with the slope of resting state CON-VS
functional connectivity. See Figure A3-A6 in Appendix A for the graphical presentations of
modeling components.
Hypotheses corresponding to Question 3 are as follows: 3a) Resting-state FPN-VS /
CON-VS functional connectivity measured at the 2-year follow-up timepoint mediates the
relationship between neural reward sensitivity measured during reward anticipation at the
baseline timepoint and cognitive control measured at the 4-year follow-up timepoint. 3b)
Resting-state FPN-VS / CON-VS functional connectivity measured at the 2-year follow-up
timepoint mediates the relationship between neural reward sensitivity measured during
reward feedback at the baseline timepoint and cognitive control measured at the 4-year
follow-up timepoint. See Figure A7 and Figure A8 in Appendix A for the graphical
presentations of modeling components.
Covariates
Age at baseline was added as a covariate in the current study to evaluate the robustness
of our findings.
15
Inclusion/exclusion criteria
For analyses corresponding to Questions 1 and 2, we first excluded participants who had
MID contrast data labeled as unacceptable in any of the three time points. As a result, 1677
participants out of 11868 were included. Additional inclusion/exclusion criteria for each
research question were as follows:
Question 1 exclusion criteria: 1) Missing SST behavioral data at any of the three time
points; 2) SST behavioral data labeled as unacceptable performance. This resulted in a
sample size of 915.
Question 2 exclusion criteria: 1) Missing resting-state fMRI data at any of the three time
points; 2) Resting-state fMRI data labeled as unacceptable performance. This resulted in a
sample size of 1513.
Question 3 exclusion criteria: 1) Missing or unacceptable MID fMRI data at baseline; 2)
Missing or unacceptable resting-state fMRI data at 2-year follow-up; 3) Missing or
unacceptable SST behavioral data at 4-year follow-up. This resulted in a sample size of 1207.
Missing values
Partly due to the ongoing collection process, the rate of missing data in Release 5.1 was
quite high. Specifically, of the 11,868 subjects that participated in the baseline data collection,
10,973 participated in the 2-year follow-up, and 4754 participated in the 4-year follow-up by
the time the data was frozen for the ABCD Data Release 5.1. When testing Question 1 and 2,
we only included participants who had qualifying data across all three time points.
Outliers
In current research, statistical outliers were removed from the dataset. Statistical outliers
were defined as follows: (1) Reward sensitivity: MID fMRI contrast beta weight values ≥ 3
standard deviations away from the sample mean across three time points. (2) Cognitive
control: SSRT values ≥ 3 standard deviations away from the sample mean across three time
16
points. (3) Cortical-subcortical connectivity: resting state fMRI correlation values ≥1 or ≤ -1.
This exclusion criterion for correlation values was applied to detect any errors in recording
the data. No such values were found in the current dataset.
Results
Demographic statistics
Demographic information of participants is shown in Table 1, in which sample 1, 2, 3
represent samples corresponding to the analysis of Question 1, 2, and 3 in the current study.
Table 1
Demographic information of participants included in the current study
Sample 1
Sample 2
Sample 3
N
915
1513
1207
Female Percentage
47.54%
48.05%
47.97%
Race
White
74.54%
71.95%
71.97%
Black
7.43%
8.53%
9.12%
Asian
2.08%
1.92%
1.82%
Other
4.15%
5.69%
5.06%
Mixed
10.93%
10.98%
11.03%
Unrevealed
0.87%
0.93%
1.00%
Hispanic
Yes
17.81%
20.62%
18.89%
No
80.98%
78.06%
79.87%
Unrevealed
1.20%
1.32%
1.24%
SES
0.11 (SD=0.86)
0.09 (SD=0.86)
0.07 (SD=0.87)
Note. Female Percentage was based on the sex at birth data, which was collected at baseline from
participants' parents by asking " What sex was the child assigned at birth, on the original birth certificate?".
In the current study, sex was considered as a categorical variable with two levels: female and male. Race
was coded as a categorical variable with six levels: White, Black, Asian and NHPI, Mixed, Other, and
Unrevealed (i.e., the participant's parent refused to answer or did not know). Ethnicity was collected by the
answers of parents to the question "Do you consider the child Hispanic/Latino/Latina?" In current study,
ethnicity was coded as a categorical variable with three levels: Yes, No, and Unrevealed (i.e., the
participant's parent refused to answer or did not know). SES in the current analysis was a numeric variable,
represented by an existing latent factor in Release 5.1 (Gonzalez et al., 2019).
Question 1: Does neural reward sensitivity positively predict cognitive control during
development from 9 to 14 years old?
Models 1-4 were specified to test specific hypotheses related to Question 1. Table 2
shows descriptive statistics corresponding to Question 1. Table 3 shows means and variances
of latent variables in the unconditional LGCMs. Table 4 shows the model fit statistics. Table
17
5 shows the estimated intercept and slope of reward sensitivity and cognitive control in all
models and some important correlation coefficients. Figure B1 and Figure B2 in Appendix
B show the complete path diagram.
Similar results were observed in model 1, model 2 and model 4. In each model, after
adjusting for baseline age and fixing non-significant paths to zero, the final models showed
good model fit. The slope of reward sensitivity did not significantly correlate with the slope
of cognitive control, suggesting no significant linear developmental relationship between
them from 9-14 years old in this study sample.
Model 3 showed a different pattern, which was about the relationship between cognitive
control and reward sensitivity during reward feedback measured in right NAcc. In model 3,
after adjusting for baseline age and fixing non-significant paths to zero, the final model
showed a good model fit. In contrast to model 1, 2, and 4, model 3 showed a significant
negative correlation between the slope of reward sensitivity and the slope of cognitive control
(r=-0.436, p=0.012), suggesting a tendency that individuals with a smaller decrease in reward
sensitivity (i.e., those who maintained a relatively high level of reward sensitivity) tend to
show faster improvement in cognitive control performance (i.e., less SSRT).
Regarding cognitive control, in all four above models, the intercept of cognitive control
was negatively correlated with its slope, suggesting a tendency that the better the baseline
cognitive control is (i.e., lower SSRT), the faster it develops over time (Model1: r =-0.641,
p<0.001; Model 2: r=-0.641, p<0.001; Model 3: r=-0.644, p<0.001; Model 4: r =-0.641,
p<0.001).
Table 2
Descriptive statistics for sample 1 corresponding to Question 1
Baseline
2-year Follow-Up
4-year Follow-Up
Age in Months
119.73 (SD=7.4)
143.71(SD=7.61)
169.16 (SD=8.09)
Reward Sensitivity
(anticipation stage)
Right NAcc
0.057 (SD=0.223)
0.064 (SD=0.229)
0.053 (SD=0.209)
Left NAcc
0.066 (SD=0.218)
0.063 (SD=0.232)
0.056 (SD=0.214)
18
Reward Sensitivity
(feedback stage)
Right NAcc
0.126 (SD=0.260)
0.118 (SD=0.264)
0.116 (SD=0.245)
Left NAcc
0.154 (SD=0.256)
0.135 (SD=0.272)
0.125 (SD=0.242)
Cognitive Control
280.74 (SD=74.80)
253.00 (SD=63.68)
245.30 (SD=55.50)
Table 3
Means and variance of the unconditional LGCMs corresponding to Question 1
Unconditional
Model
Intercept
Slope
Mean
Variance
Mean
Variance
RS (Anti-Right)
0.06***
0.004
-0.002
0.002
RS (Anti-Left)
0.066***
0.004
-0.005
0.002
RS (Feed-Right)
0.125***
0.007
-0.005
0.001a
RS (Feed-Left)
0.153***
0.005
-0.015*
0.001
CC
277.078***
3010.183***
-16.98***
511.953***
Note. Means and variances of latent variables in the unconditional LGCMs for reward sensitivity (RS) indicators
and cognitive control (CC) indicators, analyzed separately, corresponding to Question 1. Anti: reward sensitivity
at the reward anticipation stage; Feed: reward sensitivity at the reward feedback stage; Right: in right NAcc;
Left: in left NAcc.
a. Variance was fixed at 0.001 for model convergence.
*p < .05. ***p < .001
Table 4
Model fit statistics of the LGCMs corresponding to Question 1
Model Number
χ2
df
CFI
RMSEA
SRMR
1 (Anti-Right)
37.916
13
0.947
0.046
0.025
2 (Anti-Left)
36.298
13
0.951
0.044
0.023
3 (Feed-Right)
37.803
13
0.950
0.046
0.024
4 (Feed-Left)
38.326
13
0.948
0.046
0.024
Table 5
Results of the LGCMs corresponding to Question 1
Model Number
Ir
Sr
Ic
Sc
rIc-Sc
rSr-Sc
1 (Anti-Right)
0.060
-0.002
277.099
-16.975
-0.641***
-0.169
2 (Anti-Left)
0.066
-0.005
277.095
-16.983
-0.641***
-0.259
3 (Feed-Right)
0.125
-0.005
277.113
-17.002
-0.644***
-0.436*
4 (Feed-Left)
0.153
-0.015
277.106
-16.989
-0.641***
-0.213
Note. The estimated intercept and slope of reward sensitivity and cognitive control as well as critical findings
in all models corresponding to Question 1. Anti: reward sensitivity at the reward anticipation stage; Feed:
reward sensitivity at the reward feedback stage; Right: in right NAcc; Left: in left NAcc; Ir: intercept of
reward sensitivity; Sr: slope of reward sensitivity; Ic: intercept of cognitive control; Sc: slope of cognitive
control. rIc-Sc: correlation coefficient between the intercept and the slope of cognitive control; rSr-Sc:
correlation coefficient between the slope of reward sensitivity and the slope of cognitive control.
*p < .05. ***p < .001
19
Question 2: Does the development of neural reward sensitivity predict cortical-
subcortical functional connectivity supporting cognitive control?
Models 5-12 were specified to test hypotheses for Question 2. Table 6 shows descriptive
statistics corresponding to Question 2. Table 7 shows means and variances of latent variables
in the unconditional LGCMs. Table 8 shows the model fit statistics. Table 9 shows the
estimated intercept and slope of reward sensitivity and functional connectivity in all models
and correlation coefficients between the slope of t reward sensitivity and functional
connectivity. Figure B3-B6 in Appendix B show the complete path diagram.
In all models, after adjusting for baseline age, fixing unsignificant paths to zero except
for paths that would result in misspecification if fixed, the final models showed good model
fit. The slope of reward sensitivity did not significantly correlate with the slope of functional
connectivity in any of the above models, suggesting no significant linear developmental
relationship between them from 9 to 14 years old in this study sample.
Notably, in model 6, the intercept of reward sensitivity was positively correlated with
the slope of FPN-left NAcc functional connectivity (r =0.531, p=0.006), suggesting a
tendency that the higher the baseline reward sensitivity during reward anticipation in left
NAcc, the faster the FPN-left NAcc functional connectivity strengthens.
Regarding an exploratory possibility that baseline FPN-NAcc functional connectivity
might in return predict the developmental rate of reward sensitivity, model 5 and model 10
showed a positive correlation between the intercept of baseline FPN connectivity and the
slope of reward sensitivity (Model 5: r =0.323, p=0.029; Model 10: r =0.392, p=0.048),
suggesting a tendency that the stronger the baseline FPN connectivity is, the faster the reward
sensitivity decrease.
Table 6
Descriptive Statistics for sample 2 Corresponding to Question 2
Baseline
2-year Follow-Up
4-year Follow-Up
20
Age in Months
119.43 (SD=7.58)
143.37 (SD=7.90)
168.76 (SD=8.32)
Reward Sensitivity
(anticipation stage)
Right NAcc
0.065 (SD=0.220)
0.068 (SD=0.230)
0.048 (SD=0.211)
Left NAcc
0.071 (SD=0.216)
0.064 (SD=0.230)
0.055 (SD=0.215)
Reward Sensitivity
(feedback stage)
Right NAcc
0.131 (SD=0.256)
0.122 (SD=0.265)
0.116 (SD=0.242)
Left NAcc
0.154 (SD=0.255)
0.132(SD=0.271)
0.129 (SD=0.243)
FPN-NAcc
Connectivity
Right side
-0.003 (SD=0.062)
-0.001 (SD=0.064)
-0.007 (SD=0.069)
Left side
-0.064 (SD=0.076)
-0.061 (SD=0.080)
-0.059 (SD=0.078)
CON-NAcc
Connectivity
Right side
0.248 (SD=0.099)
0.244 (SD=0.094)
0.241 (SD=0.095)
Left side
0.113 (SD=0.078)
0.096 (SD=0.081)
0.074 (SD=0.079)
Table 7
Means and variance of the unconditional LGCMs corresponding to Question 2
Unconditional Model
Intercept
Slope
Mean
Variance
Mean
Variance
RS (Anti-Right)
0.068***
0.005
-0.008*
0.002
RS (Anti-Left)
0.072***
0.006
-0.008*
0.001
RS (Feed-Right)
0.131***
0.001
-0.008
0.001a
RS (Feed-Left)
0.151***
-0.012**
-0.012**
0.001
FPN-NAcc (Right)
-0.002
<0.001*
-0.002
<0.001
FPN-NAcc (Left)
-0.064***
0.001**
0.003
<0.001
CON-NAcc (Right)
0.248***
0.003***
-0.004*
0.001*
CON-NAcc (Left)
0.114***
0.002***
-0.02***
<0.001
Note. Means and variances of latent variables in the unconditional LGCMs for reward sensitivity (RS) indicators
and cortical-subcortical connectivity (i.e., FPN-NAcc and CON-NAcc connectivity) indicators, analyzed
separately, corresponding to Question 2. Anti: reward sensitivity at the reward anticipation stage; Feed: reward
sensitivity at the reward feedback stage; Right: in right NAcc; Left: in left NAcc.
a. Variance was fixed at 0.001 for model convergence.
*p < .05. **p < .01. ***p < .001
Table 8
Model fit statistics of the LGCMs corresponding to Question 2
Model Number
χ2
df
CFI
RMSEA
SRMR
5 (FPN-Anti-Right)
10.840
12
1.000
<0.001
0.014
6 (FPN-Anti-Left)
11.186
12
1.000
<0.001
0.014
7 (CON-Anti-Right)
14.918
13
0.993
0.010
0.016
8 (CON-Anti-Left)
6.241
12
1.000
<0.001
0.010
9 (FPN-Feed-Right)
23.656
15
0.903
0.020
0.022
10 (FPN-Feed-Left)
37.321
12
0.848
0.037
0.026
21
11 (CON-Feed-Right)
54.811
15
0.868
0.042
0.036
12 (CON-Feed-Left)
9.861
12
1.000
<0.001
0.014
Table 9
Results of the LGCMs corresponding to Question 2
Model Number
Ir
Sr
Icor
Scor
rSr-Scor
5 (FPN-Anti-Right)
0.068
-0.008
-0.002
-0.002
-0.328
6 (FPN-Anti-Left)
0.072
-0.008
-0.064
0.003
-0.460
7 (CON-Anti-Right)
0.068
-0.008
0.248
-0.004
0.372
8 (CON-Anti-Left)
0.072
-0.008
0.114
-0.020
0.066
9 (FPN-Feed-Right)
0.131
-0.008
-0.002
-0.002
-0.389
10 (FPN-Feed-Left)
0.151
-0.012
-0.064
0.003
-0.260
11 (CON-Feed-Right)
0.131
-0.008
0.248
-0.004
0.112
12 (CON-Feed-Left)
0.151
-0.012
0.114
-0.020
-0.383
Note. The estimated intercept and slope of reward sensitivity and functional connectivity as well as critical
findings in all models corresponding to Question 2. FPN: FPN-NAcc connectivity; CON: CON-NAcc
connectivity; Anti: reward sensitivity at the reward anticipation stage; Feed: reward sensitivity at the reward
feedback stage; Right: reward sensitivity in right NAcc and functional connectivity between networks and
NAcc in the right hemisphere; Left: reward sensitivity in left NAcc and functional connectivity between
FPN/CON and NAcc in the left hemisphere; Ir: intercept of reward sensitivity; Sr: slope of reward sensitivity;
Icor: intercept of functional connectivity; Sc: slope of functional connectivity. rSr-Scor: correlation coefficient
between the slope of reward sensitivity and the slope of functional connectivity. All correlation shown in this
table were not statistically significant.
Question 3: Does cognitive control-related cortical-subcortical connectivity mediate the
relationship between neural reward sensitivity and cognitive control?
Models 13-16 were specified to test hypotheses for Question 3. Table 10 shows
descriptive statistics corresponding to Question 3. Table 11 shows the path coefficients in the
mediation models specified for Question 3. Figure B7 and Figure B8 in Appendix B show
the complete path diagram.
Models including the right NAcc activation showed similar results. Both model 13 and
model 15 showed good model fits after adjusting for baseline age (Model 13: (χ2=1.803,
df =2, CFI=1.000, RMSEA<0.001, SRMR=0.010; Model 15: χ2=1.821,
df =2, CFI=1.000, RMSEA<0.001, SRMR=0.010). In both models, reward sensitivity was not
significantly associated with FPN-NAcc connectivity or CON-NAcc connectivity. Two
mediators, FPN-NAcc connectivity and CON-NAcc connectivity, was significantly
22
negatively correlated (Model 13: std.β=-0.104, p=0.002; Model 15: std.β=-0.104, p=0.001).
However, reward sensitivity, FPN-NAcc connectivity, and CON-NAcc connectivity were not
significantly associated with cognitive control. The indirect effect from reward sensitivity to
cognitive control via FPN-NAcc connectivity or CON-NAcc connectivity was not significant.
Modeling on the left NAcc activation was more challenging. Model 14, which focused
on the reward sensitivity measured during anticipation stages and left hemisphere activation
did not converge. Model 16, which focused on the reward sensitivity measured during
feedback stages and left hemisphere activation did not converge after controlling for baseline
age. Thus, we specified model 16 without including baseline age as a covariate. When doing
so, this model showed a good model fit (χ2<0.001,
df =0, CFI=1.000, RMSEA<0.001, SRMR=<0.001). Path model reveals that reward sensitivity
was significantly positively associated with CON-NAcc connectivity (std.β=0.060, p=0.022)
but not with FPN-NAcc connectivity. Two mediators, FPN-NAcc connectivity and CON-
NAcc connectivity, were significantly negatively correlated (std.β=-0.067, p=0.030).
However, reward sensitivity, FPN-NAcc connectivity, and CON-NAcc connectivity were not
significantly associated with cognitive control. The indirect effect from reward sensitivity to
cognitive control via FPN-NAcc connectivity or CON-NAcc connectivity was not significant.
Table 10
Descriptive statistics for sample 3 corresponding to Question 3
Baseline
2-year Follow-Up
4-year Follow-Up
Age in Months
119.46 (SD=7.41)
143.36 (SD=7.65)
168.81 (SD=8.07)
Reward Sensitivity
(anticipation stage)
Right NAcc
0.060 (SD=0.229)
/
/
Left NAcc
0.066 (SD=0.224)
/
/
Reward Sensitivity
(feedback stage)
Right NAcc
0.132 (SD=0.262)
/
/
Left NAcc
0.153 (SD=0.260)
/
/
FPN-NAcc
Connectivity
Right side
/
-0.002 (SD=0.063)
/
23
Left side
/
-0.061 (SD=0.080)
/
CON-NAcc
Connectivity
Right side
/
0.245 (SD=0.093)
/
Left side
/
0.096 (SD=0.079)
/
Cognitive Control -
SSRT
/
/
245.30 (SD=55.50)
Table 11
Path coefficients in the mediation models for Question 3
Paths
β
std.β
95% C.I.
S.E.
p-value
Model 13
RS→FPN
0.006
0.021
-0.007~0.019
0.008
0.469
RS→CON
0.004
0.011
-0.016~0.025
0.356
0.722
RS→CC
3.501
0.013
-8.483~15.484
7.285
0.631
FPN→CC
-33.068
-0.035
-78.016~11.881
27.324
0.226
CON→CC
-6.725
-0.011
-38.708~25.258
19.443
0.729
FPN↔CON
-0.001
-0.104
-0.001~ 0.000
<0.001
0.002**
Model 14
Model did not converge
Model 15
RS→FPN
<0.001
0.001
-0.012~0.013
0.008
0.987
RS→CON
-0.018
-0.050
-0.035~-0.001
0.01
0.085
RS→CC
7.024
0.031
-3.192~17.240
6.21
0.258
FPN→CC
-33.261
-0.036
-77.478~10.956
26.88
0.216
CON→CC
-7.526
-0.012
-41.017~25.965
20.359
0.712
FPN↔CON
-0.001
-0.104
-0.001~0.000
<0.001
0.001**
Model 16 a
RS→FPN
0.008
0.026
-0.007, 0.023
0.009
0.375
RS→CON
0.018
0.060
0.005~ 0.032
0.008
0.022*
RS→CC
0.989
0.004
-10.171~12.148
6.784
0.884
FPN→CC
-32.066
-0.043
-66.786~2.655
21.107
0.129
CON→CC
3.971
0.005
-32.507~40.450
22.176
0.858
FPN↔CON
<0.001
-0.067
-0.001~0.000
<0.001
0.03*
Note. Results obtained from four mediation models used to test the mediation effect of resting-state FPN-VS
/ CON-VS functional connectivity on the relationship between neural reward sensitivity and cognitive
control. RS: reward sensitivity; FPN: FPN-NAcc connectivity; CON: CON-NAcc connectivity; CC:
cognitive control.
a This model was specified without including baseline age as a covariate
*p < .05. **p < .01
24
Discussion
In the current study, we investigated the associations between neural reward sensitivity
and cognitive control, specifically focusing on the inhibitory control component, as well as
between reward sensitivity and the FPN/CON-NAcc functional connectivity. Additionally,
we explored whether the FPN/CON-NAcc functional connectivity mediates the relationship
between neural reward sensitivity and cognitive control.
Our hypothesis testing yielded two major findings. First, we observed in this study
sample that individuals who maintained a relatively high level of right NAcc neural reward
sensitivity during reward feedback showed faster improvement in cognitive control
performance over time,. Second, we identified a relationship that the higher the left NAcc
neural reward sensitivity during reward anticipation at 9-10 years old, the faster the FPN-left
NAcc functional connectivity strengthens.
Additionally, our exploratory analysis revealed that the better the baseline cognitive
control (i.e., lower SSRT), the faster it develops over time. Also, our results suggested a
potential association between baseline FPN-NAcc functional connectivity and the
development rate of reward sensitivity.
Question 1: Does neural reward sensitivity positively predict cognitive control during
development from 9 to 14 years old?
Regarding the developmental trajectory of both cognitive control and reward sensitivity,
the results show that performance in a cognitive control task tended to gradually improve
between ages 9 to 14, which was consistent with results in prior literature (Ordaz et al., 2013).
Meanwhile, the results indicated that neural reward sensitivity showed a slight tendency to
decrease from ages 9 to 14, which contrasted with previous research findings (Galvan et al.,
2006; Luciana et al., 2012). However, these results partially aligned with two studies
specifically using MID tasks in adolescents and adults which found NAcc activation during
25
reward feedback stages did not differ between age groups, although adolescents showed
decreased NAcc activation during reward anticipation (Bjork et al., 2004; Bjork et al., 2010).
The observed tendency for a decrease in reward sensitivity may be attributed to the
following reasons. First, within the 9-14 age range, the change in reward sensitivity might be
limited, making it challenging to identify the direction of change. This possibility was
supported by the very small change rates in the model results, as indicated by the small slope
values of reward sensitivity across models specified for Question 1. Alternatively, the
decrease could be partly due to the difficulty for adolescents aged 9-14 to remain still during
fMRI scans, coupled with the relatively short duration assigned to the MID task in this
experiment (i.e., 12 minutes total). This might lead to a limited number of valid repeated
measurements of reward sensitivity, resulting in larger measurement errors. This possibility
aligned with concerns raised in literature discussing the quality of ABCD study data
(Kennedy et al., 2022).
One of our major findings is that the negative correlation between the slope of reward
sensitivity measured in the right NAcc during the reward feedback stage and the slope of
SSRT in cognitive control tasks. It suggested that individuals with a smaller decrease in
reward sensitivity tend to show more improvement in cognitive control performance (i.e., less
SSRT). This implied that maintaining higher reward sensitivity in early adolescence may be
beneficial for the development of cognitive control. On the other hand, a bigger decline in
reward sensitivity in early adolescence might limit the development of cognitive control
abilities. Other than the above possibility, this association between two components could
also result from third variables, such as other pubertal changes. This result implied a
possibility that reward sensitivity and cognitive control were not completely independent
traits during adolescent development. However, the significance observed only in the right
NAcc reminded us to interpret this finding cautiously. Lateralization of reward sensitivity
26
was also observed in prior literature (Martin-Soelch et al., 2011), although the reason remains
unknown. There could also be alternative explanations for this result. First, the relationship
between reward sensitivity and cognitive control might not be strong enough to be
consistently detected in both NAcc. Second, this statistically significant relationship could be
due to a Type II error (i.e., a false positive result).
Overall, the findings suggested a slight decline in reward sensitivity in early adolescence,
but maintaining higher reward sensitivity might benefit cognitive control development. This
result partially supported Hypothesis 1d, suggesting an adaptive function of reward
sensitivity that may enhance the development of cognitive control. However, the decreasing
developmental trajectory of reward sensitivity and the lateralization of the relation between
reward sensitivity and cognitive control did not perfectly align with the viewpoint of the
imbalanced model, especially because the model emphasized an increase in reward sensitivity
during adolescence.
In the exploratory analysis, we found that better baseline cognitive control (i.e., less
SSRT) was associated with its faster development over time. It was a meaningful finding
although this relationship was not the primary focus of this study. It suggested that
differences in cognitive control abilities observed at ages 9-10 were likely to widen by ages
13-14. This increasing difference in cognitive control performance might be influenced by a
combination of genetic and environmental factors and their interactions. The widening gap in
cognitive control as individuals age may be attributed to the significant influence of genetics
(Polderman et al., 2015). For instance, the analysis of a multivariate twin study suggests that
the correlation between different types of cognitive control is due to a highly heritable
common factor (Friedman et al., 2008). Considering the interaction of genetics and
environment in individual development (Tucker-Drob et al., 2013), one possible explanation
for the results of the current study is that Individuals with strong baseline cognitive control
27
are more likely to seek out and accumulate relevant experiences, thereby further enhancing
their cognitive control abilities. The reasons for their higher baseline levels and the faster
development of cognitive control abilities may overlap, potentially due to genetic influences.
This perspective aligns with the concept of the "rich get richer" effect (or magnification effect)
in cognitive development, which claims that individuals with higher baseline cognitive
abilities benefit more from training (Guye et al., 2017; Sauce et al., 2021). For example, a
study on undergraduate students indicated that prior knowledge was positively related to new
learning within the same domain (Witherby & Carpenter, 2022). Our findings highlight the
predictive role of early cognitive control on its future development during early adolescence.
Cognitive strategies and interventions to minimize this developmental gap worth further
investigation.
Question 2: Does the development of neural reward sensitivity predict cortical-
subcortical functional connectivity supporting cognitive control?
Regarding the development trajectory of NAcc’s functional connectivity with cognitive
control-related networks, most paths showed a tendency to decline in connectivity with age
from 9 to 14, except for the connectivity between the FPN and the left NAcc, which showed a
slight increase over time. This finding contradicted our predictions and there was a lack of
prior research in adolescents for comparison. Model convergence issues were worth attention
as well. Prior literature suggested limits to the generalizability of resting-state fMRI in the
ABCD Study. It found that the group of participants with low average motion in their resting-
state fMRI might be less representative of the general population, and head motion remained
a confound (Cosgrove et al., 2022).
Additionally, we observed that in resting-state fMRI, the FPN-NAcc functional
correlation was consistently negative, while the CON-NAcc functional correlation was
consistently positive. These contrasts reflected different modes of interaction between these
28
networks and the NAcc, and meaningfully aligned with prior literature, showing that despite
both the FPN and CON being considered networks associated with cognitive control, they
function in distinct ways (Dosenbach et al., 2007; Hausman et al., 2022). The negative FPN-
NAcc correlation indicated that when NAcc activity was strong, FPN activity tended to be
weak, and vice versa. This observation suggested a possibility that heightened FPN activity
might suppress NAcc’s activity, thereby reducing reward sensitivity. On the other hand, the
positive CON-NAcc correlation suggested that when NAcc activity was strong, CON activity
was also strong. This observation implied that heightened CON activity might enhance NAcc
activity, thereby increasing reward sensitivity.
The above observations suggested a relationship between the activity intensity of NAcc
and FPN/CON at the same time points. However, based on these observations alone, we
could not conclude if the activity of NAcc during adolescence could enhance its connection
with networks corresponding to cognitive control. In fact, models specified for Question 2
did not show significant developmental relationships between reward sensitivity and
CON/FPN-NAcc functional connectivity. This indicated that the development rate of reward
sensitivity between ages 9-14 could not predict the development rate of CON/FPN-NAcc
functional connectivity. As an alternative explanation, their relationship might not be
detectable using linear models and/or in resting-state fMRI.
However, we did find that higher baseline reward sensitivity during reward anticipation
stages in the left NAcc was associated with a faster strengthening of FPN-left NAcc
functional connectivity. This result aligned with Hypothesis 2b, suggesting that more
motivational signals from the NAcc might enhance its connectivity with cognitive control-
related networks during adolescence. However, as mentioned in the discussion of Question 1,
the lateralized nature of this finding needed cautious interpretation.
29
In exploratory analyses, two out of eight models indicated that stronger baseline FPN
connectivity was associated with a faster decrease in reward sensitivity. Overall, our results
suggest a potential association between NAcc and FPN-NAcc functional connectivity, yet
their relationship might require further research to precisely detect.
Question 3: Does cognitive control-related cortical-subcortical connectivity mediate the
relationship between neural reward sensitivity and cognitive control?
The results indicated that neither FPN-NAcc connectivity nor CON-NAcc connectivity
significantly mediated the relationship between reward sensitivity and cognitive control. This
finding was not surprising, given that in the analysis for Question 2, we did not observe a
significant developmental relationship between reward sensitivity and CON/FPN-NAcc
functional connectivity.
Additionally, as the results showed that reward sensitivity at ages 9-10 was not
significantly associated with cognitive control after four years, this suggested that the
intensity of reward sensitivity in early adolescence did not significantly predict improvements
in cognitive control abilities. However, the analysis for Question 3 considered only one
measurement time point of measurement for each variable. Thus, for a more comprehensive
understanding of their developmental trajectories, the analysis related to Question 1 was more
informative.
Limitations
First, for the samples used in Questions 1 and 2, we only included participants who had
qualifying data across all three time points. This approach limited the number of missing
values. However, excluding participants might reduce the representativeness of the sample
and could lead to selection bias. Future research should consider alternative methods for
handling missing data, such as full information maximum likelihood, instead of listwise
deletion.
30
Second, this study included measurements at only three time points, which limited the
observation of developmental trajectories. Specifically, we only explored whether there were
linear relationships between the variables, and the age range did not cover the entire
adolescence period. Future research should consider non-linear relationships and include
more time points as the ABCD study is actively ongoing.
Third, we used indicators of neural reward sensitivity and inhibitory control to explore
the relationship between reward sensitivity and cognitive control. These relationships may
not generalize to all contexts discussing the connection between reward sensitivity and
cognitive control. Future research should employ meta-analytic methods or incorporate
various measures of reward sensitivity and cognitive control into latent variable models, to
further explore these relationships.
Finally, due to model convergence issues, we only controlled for baseline age as a
covariate. Future research should attempt to replicate these findings using other datasets
while including more covariates, particularly gender, race, and SES.
Conclusion
The current study yielded two major findings. First, individuals with a smaller decrease
in reward sensitivity measured in the right nucleus accumbens (NAcc) showed more
improvement in cognitive control performance over time. Second, higher baseline reward
sensitivity during reward anticipation stages in the left NAcc was associated with a faster
strengthening of FPN-left NAcc functional connectivity. These findings suggested that neural
reward sensitivity can serve as a possible predictor for the development of cognitive control
in adolescents aged 9 to 14. This interdependent relationship between reward sensitivity and
cognitive control leaned to support the imbalance model than the dual system model.
However, it is worth mentioning that these relationships were observed only in either
right or left NAcc respectively, and the results also showed a slight tendency to decline in
31
reward sensitivity with age, rather than an increase. These characteristics suggested that our
findings did not perfectly align with the imbalance model. Future research should include
more time points and more covariates to further investigate the developmental relationship
between reward sensitivity and cognitive control.
32
References
Adolescent Brain Cognitive Development. (2024a). Estimates of Stop Signal Reaction Time.
Retrieved May 7, 2024 from https://wiki.abcdstudy.org/release-notes/imaging/task-
fmri-behavior.html#estimates-of-stop-signal-reaction-time
Adolescent Brain Cognitive Development. (2024b). Release Notes. Retrieved May 7, 2024
from https://wiki.abcdstudy.org/release-notes/start-page.html
Adolescent Brain Cognitive Development. (n.d.-a). ABCD Data Dictionary (Release 5.1).
Retrieved May 7, 2024 from https://data-dict.abcdstudy.org/?
Adolescent Brain Cognitive Development. (n.d.-b). About the Study. Retrieved April 29,
2024 from https://abcdstudy.org/
Aron, A. R. (2007, Jun). The neural basis of inhibition in cognitive control. Neuroscientist,
13(3), 214-228. https://doi.org/10.1177/1073858407299288
Barch, D. M., Albaugh, M. D., Avenevoli, S., Chang, L., Clark, D. B., Glantz, M. D.,
Hudziak, J. J., Jernigan, T. L., Tapert, S. F., Yurgelun-Todd, D., Alia-Klein, N., Potter,
A. S., Paulus, M. P., Prouty, D., Zucker, R. A., & Sher, K. J. (2018). Demographic,
physical and mental health assessments in the adolescent brain and cognitive
development study: Rationale and description. Dev Cogn Neurosci, 32, 55-66.
https://doi.org/10.1016/j.dcn.2017.10.010
Bari, A., & Robbins, T. W. (2013). Inhibition and impulsivity: behavioral and neural basis of
response control. Prog Neurobiol, 108, 44-79.
https://doi.org/10.1016/j.pneurobio.2013.06.005
Bjork, J. M., Knutson, B., Fong, G. W., Caggiano, D. M., Bennett, S. M., & Hommer, D. W.
(2004). Incentive-elicited brain activation in adolescents: similarities and differences
from young adults. J Neurosci, 24(8), 1793-1802.
https://doi.org/10.1523/JNEUROSCI.4862-03.2004
Bjork, J. M., Smith, A. R., Chen, G., & Hommer, D. W. (2010). Adolescents, adults and
rewards: comparing motivational neurocircuitry recruitment using fMRI. PLoS One,
5(7), e11440. https://doi.org/10.1371/journal.pone.0011440
Boucher, L., Palmeri, T. J., Logan, G. D., & Schall, J. D. (2007). Inhibitory control in mind
and brain: an interactive race model of countermanding saccades. Psychol Rev, 114(2),
376-397. https://doi.org/10.1037/0033-295X.114.2.376
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and
affective responses to impending reward and punishment: The BIS/BAS Scales.
Journal of Personality and Social Psychology, 67(2), 319-333.
https://doi.org/10.1037/0022-3514.67.2.319
Casey, B. J. (2015). Beyond Simple Models of Self-Control to Circuit-Based Accounts of
Adolescent Behavior. Annual Review of Psychology, 66(1), 295-319.
https://doi.org/10.1146/annurev-psych-010814-015156
Casey, B. J., Galvan, A., & Somerville, L. H. (2016). Beyond simple models of adolescence
to an integrated circuit-based account: A commentary. Dev Cogn Neurosci, 17, 128-
130. https://doi.org/10.1016/j.dcn.2015.12.006
Casey, B. J., Getz, S., & Galvan, A. (2008). The adolescent brain. Dev Rev, 28(1), 62-77.
https://doi.org/10.1016/j.dr.2007.08.003
Casey, B. J., Heller, A. S., Gee, D. G., & Cohen, A. O. (2019). Development of the emotional
brain. Neurosci Lett, 693, 29-34. https://doi.org/10.1016/j.neulet.2017.11.055
Chaarani, B., Hahn, S., Allgaier, N., Adise, S., Owens, M. M., Juliano, A. C., Yuan, D. K.,
Loso, H., Ivanciu, A., Albaugh, M. D., Dumas, J., Mackey, S., Laurent, J., Ivanova,
M., Hagler, D. J., Cornejo, M. D., Hatton, S., Agrawal, A., Aguinaldo, L., Ahonen, L.,
Aklin, W., Anokhin, A. P., Arroyo, J., Avenevoli, S., Babcock, D., Bagot, K., Baker,
F. C., Banich, M. T., Barch, D. M., Bartsch, H., Baskin-Sommers, A., Bjork, J. M.,
33
Blachman-Demner, D., Bloch, M., Bogdan, R., Bookheimer, S. Y., Breslin, F., Brown,
S., Calabro, F. J., Calhoun, V., Casey, B. J., Chang, L., Clark, D. B., Cloak, C.,
Constable, R. T., Constable, K., Corley, R., Cottler, L. B., Coxe, S., Dagher, R. K.,
Dale, A. M., Dapretto, M., Delcarmen-Wiggins, R., Dick, A. S., Do, E. K.,
Dosenbach, N. U. F., Dowling, G. J., Edwards, S., Ernst, T. M., Fair, D. A., Fan, C.
C., Feczko, E., Feldstein-Ewing, S. W., Florsheim, P., Foxe, J. J., Freedman, E. G.,
Friedman, N. P., Friedman-Hill, S., Fuemmeler, B. F., Galvan, A., Gee, D. G., Giedd,
J., Glantz, M., Glaser, P., Godino, J., Gonzalez, M., Gonzalez, R., Grant, S., Gray, K.
M., Haist, F., Harms, M. P., Hawes, S., Heath, A. C., Heeringa, S., Heitzeg, M. M.,
Hermosillo, R., Herting, M. M., Hettema, J. M., Hewitt, J. K., Heyser, C., Hoffman,
E., Howlett, K., Huber, R. S., Huestis, M. A., Hyde, L. W., Iacono, W. G., Infante, M.
A., Irfanoglu, O., Isaiah, A., Iyengar, S., Jacobus, J., James, R., Jean-Francois, B.,
Jernigan, T., Karcher, N. R., Kaufman, A., Kelley, B., Kit, B., Ksinan, A., Kuperman,
J., Laird, A. R., Larson, C., LeBlanc, K., Lessov-Schlagger, C., Lever, N., Lewis, D.
A., Lisdahl, K., Little, A. R., Lopez, M., Luciana, M., Luna, B., Madden, P. A., Maes,
H. H., Makowski, C., Marshall, A. T., Mason, M. J., Matochik, J., McCandliss, B. D.,
McGlade, E., Montoya, I., Morgan, G., Morris, A., Mulford, C., Murray, P., Nagel, B.
J., Neale, M. C., Neigh, G., Nencka, A., Noronha, A., Nixon, S. J., Palmer, C. E.,
Pariyadath, V., Paulus, M. P., Pelham, W. E., Pfefferbaum, D., Pierpaoli, C., Prescot,
A., Prouty, D., Puttler, L. I., Rajapaske, N., Rapuano, K. M., Reeves, G., Renshaw, P.
F., Riedel, M. C., Rojas, P., de la Rosa, M., Rosenberg, M. D., Ross, M. J., Sanchez,
M., Schirda, C., Schloesser, D., Schulenberg, J., Sher, K. J., Sheth, C., Shilling, P. D.,
Simmons, W. K., Sowell, E. R., Speer, N., Spittel, M., Squeglia, L. M., Sripada, C.,
Steinberg, J., Striley, C., Sutherland, M. T., Tanabe, J., Tapert, S. F., Thompson, W.,
Tomko, R. L., Uban, K. A., Vrieze, S., Wade, N. E., Watts, R., Weiss, S., Wiens, B.
A., Williams, O. D., Wilbur, A., Wing, D., Wolff-Hughes, D., Yang, R., Yurgelun-
Todd, D. A., Zucker, R. A., Potter, A., Garavan, H. P., & Consortium, A. (2021).
Baseline brain function in the preadolescents of the ABCD Study. Nat Neurosci, 24(8),
1176-1186. https://doi.org/10.1038/s41593-021-00867-9
Constantinidis, C., & Luna, B. (2019). Neural Substrates of Inhibitory Control Maturation in
Adolescence. Trends Neurosci, 42(9), 604-616.
https://doi.org/10.1016/j.tins.2019.07.004
Cosgrove, K. T., McDermott, T. J., White, E. J., Mosconi, M. W., Thompson, W. K., Paulus,
M. P., Cardenas-Iniguez, C., & Aupperle, R. L. (2022). Limits to the generalizability
of resting-state functional magnetic resonance imaging studies of youth: An
examination of ABCD Study(R) baseline data. Brain Imaging Behav, 16(4), 1919-
1925. https://doi.org/10.1007/s11682-022-00665-2
Crone, E. A., & Dahl, R. E. (2012). Understanding adolescence as a period of social-affective
engagement and goal flexibility. Nat Rev Neurosci, 13(9), 636-650.
https://doi.org/10.1038/nrn3313
Cunningham, R. M., Walton, M. A., & Carter, P. M. (2018). The Major Causes of Death in
Children and Adolescents in the United States. N Engl J Med, 379(25), 2468-2475.
https://doi.org/10.1056/NEJMsr1804754
Dosenbach, N. U., Fair, D. A., Cohen, A. L., Schlaggar, B. L., & Petersen, S. E. (2008). A
dual-networks architecture of top-down control. Trends Cogn Sci, 12(3), 99-105.
https://doi.org/10.1016/j.tics.2008.01.001
Dosenbach, N. U., Fair, D. A., Miezin, F. M., Cohen, A. L., Wenger, K. K., Dosenbach, R. A.,
Fox, M. D., Snyder, A. Z., Vincent, J. L., Raichle, M. E., Schlaggar, B. L., & Petersen,
S. E. (2007). Distinct brain networks for adaptive and stable task control in humans.
34
Proc Natl Acad Sci U S A, 104(26), 11073-11078.
https://doi.org/10.1073/pnas.0704320104
Duell, N., & Steinberg, L. (2021). Adolescents take positive risks, too. Developmental
Review, 62. https://doi.org/10.1016/j.dr.2021.100984
Duell, N., Steinberg, L., Icenogle, G., Chein, J., Chaudhary, N., Di Giunta, L., Dodge, K. A.,
Fanti, K. A., Lansford, J. E., Oburu, P., Pastorelli, C., Skinner, A. T., Sorbring, E.,
Tapanya, S., Uribe Tirado, L. M., Alampay, L. P., Al-Hassan, S. M., Takash, H. M. S.,
Bacchini, D., & Chang, L. (2018). Age Patterns in Risk Taking Across the World. J
Youth Adolesc, 47(5), 1052-1072. https://doi.org/10.1007/s10964-017-0752-y
Fischl, B., Salat, D. H., Busa, E., Albert, M., Dieterich, M., Haselgrove, C., van der Kouwe,
A., Killiany, R., Kennedy, D., Klaveness, S., Montillo, A., Makris, N., Rosen, B., &
Dale, A. M. (2002). Whole brain segmentation: automated labeling of
neuroanatomical structures in the human brain. Neuron, 33(3), 341-355.
https://doi.org/10.1016/s0896-6273(02)00569-x
Friedman, N. P., & Miyake, A. (2004). The relations among inhibition and interference
control functions: a latent-variable analysis. J Exp Psychol Gen, 133(1), 101-135.
https://doi.org/10.1037/0096-3445.133.1.101
Friedman, N. P., Miyake, A., Young, S. E., DeFries, J. C., Corley, R. P., & Hewitt, J. K.
(2008). Individual differences in executive functions are almost entirely genetic in
origin. J Exp Psychol Gen, 137(2), 201-225. https://doi.org/10.1037/0096-
3445.137.2.201
Friedman, N. P., & Robbins, T. W. (2022). The role of prefrontal cortex in cognitive control
and executive function. Neuropsychopharmacology, 47(1), 72-89.
https://doi.org/10.1038/s41386-021-01132-0
Galvan, A. (2010). Adolescent development of the reward system. Front Hum Neurosci, 4, 6.
https://doi.org/10.3389/neuro.09.006.2010
Galvan, A., Hare, T. A., Parra, C. E., Penn, J., Voss, H., Glover, G., & Casey, B. J. (2006).
Earlier development of the accumbens relative to orbitofrontal cortex might underlie
risk-taking behavior in adolescents. J Neurosci, 26(25), 6885-6892.
https://doi.org/10.1523/JNEUROSCI.1062-06.2006
Garavan, H., Bartsch, H., Conway, K., Decastro, A., Goldstein, R. Z., Heeringa, S., Jernigan,
T., Potter, A., Thompson, W., & Zahs, D. (2018). Recruiting the ABCD sample:
Design considerations and procedures. Dev Cogn Neurosci, 32, 16-22.
https://doi.org/10.1016/j.dcn.2018.04.004
Gonzalez, M. R., Palmer, C. E., Uban, K. A., Jernigan, T. L., Thompson, W. K., & Sowell, E.
R. (2019). Economic, social, and physiological resilience predict brain structure and
cognitive performance in 9 - 10-year-old children. https://doi.org/10.1101/852988
Gordon, E. M., Laumann, T. O., Adeyemo, B., Huckins, J. F., Kelley, W. M., & Petersen, S.
E. (2016). Generation and Evaluation of a Cortical Area Parcellation from Resting-
State Correlations. Cereb Cortex, 26(1), 288-303.
https://doi.org/10.1093/cercor/bhu239
Guye, S., De Simoni, C., & von Bastian, C. C. (2017). Do Individual Differences Predict
Change in Cognitive Training Performance? A Latent Growth Curve Modeling
Approach. Journal of Cognitive Enhancement, 1(4), 374-393.
https://doi.org/10.1007/s41465-017-0049-9
Hausman, H. K., Hardcastle, C., Albizu, A., Kraft, J. N., Evangelista, N. D., Boutzoukas, E.
M., Langer, K., O'Shea, A., Van Etten, E. J., Bharadwaj, P. K., Song, H., Smith, S. G.,
Porges, E., DeKosky, S. T., Hishaw, G. A., Wu, S., Marsiske, M., Cohen, R.,
Alexander, G. E., & Woods, A. J. (2022). Cingulo-opercular and frontoparietal
35
control network connectivity and executive functioning in older adults. Geroscience,
44(2), 847-866. https://doi.org/10.1007/s11357-021-00503-1
Hwang, K., Velanova, K., & Luna, B. (2010). Strengthening of top-down frontal cognitive
control networks underlying the development of inhibitory control: a functional
magnetic resonance imaging effective connectivity study. J Neurosci, 30(46), 15535-
15545. https://doi.org/10.1523/JNEUROSCI.2825-10.2010
Kennedy, J. T., Harms, M. P., Korucuoglu, O., Astafiev, S. V., Barch, D. M., Thompson, W.
K., Bjork, J. M., & Anokhin, A. P. (2022). Reliability and stability challenges in
ABCD task fMRI data. Neuroimage, 252, 119046.
https://doi.org/10.1016/j.neuroimage.2022.119046
Kidd, C., & Loxton, N. J. (2018). Junk food advertising moderates the indirect effect of
reward sensitivity and food consumption via the urge to eat. Physiol Behav, 188, 276-
282. https://doi.org/10.1016/j.physbeh.2018.02.030
Kim, S. H., Yoon, H., Kim, H., & Hamann, S. (2015). Individual differences in sensitivity to
reward and punishment and neural activity during reward and avoidance learning. Soc
Cogn Affect Neurosci, 10(9), 1219-1227. https://doi.org/10.1093/scan/nsv007
Knutson, B., Westdorp, A., Kaiser, E., & Hommer, D. (2000). FMRI visualization of brain
activity during a monetary incentive delay task. Neuroimage, 12(1), 20-27.
https://doi.org/10.1006/nimg.2000.0593
Lerman-Sinkoff, D. B., Sui, J., Rachakonda, S., Kandala, S., Calhoun, V. D., & Barch, D. M.
(2017). Multimodal neural correlates of cognitive control in the Human Connectome
Project. Neuroimage, 163, 41-54. https://doi.org/10.1016/j.neuroimage.2017.08.081
Luciana, M., & Collins, P. F. (2012). Incentive Motivation, Cognitive Control, and the
Adolescent Brain: Is It Time for a Paradigm Shift? Child Dev Perspect, 6(4), 392-399.
https://doi.org/10.1111/j.1750-8606.2012.00252.x
Luciana, M., Wahlstrom, D., Porter, J. N., & Collins, P. F. (2012). Dopaminergic modulation
of incentive motivation in adolescence: age-related changes in signaling, individual
differences, and implications for the development of self-regulation. Dev Psychol,
48(3), 844-861. https://doi.org/10.1037/a0027432
Luna, B., Paulsen, D. J., Padmanabhan, A., & Geier, C. (2013). Cognitive Control and
Motivation. Curr Dir Psychol Sci, 22(2), 94-100.
https://doi.org/10.1177/0963721413478416
Martin-Soelch, C., Szczepanik, J., Nugent, A., Barhaghi, K., Rallis, D., Herscovitch, P.,
Carson, R. E., & Drevets, W. C. (2011). Lateralization and gender differences in the
dopaminergic response to unpredictable reward in the human ventral striatum. Eur J
Neurosci, 33(9), 1706-1715. https://doi.org/10.1111/j.1460-9568.2011.07642.x
Mastwal, S., Ye, Y., Ren, M., Jimenez, D. V., Martinowich, K., Gerfen, C. R., & Wang, K. H.
(2014). Phasic dopamine neuron activity elicits unique mesofrontal plasticity in
adolescence. J Neurosci, 34(29), 9484-9496.
https://doi.org/10.1523/JNEUROSCI.1114-14.2014
McKenna, R., Rushe, T., & Woodcock, K. A. (2017). Informing the Structure of Executive
Function in Children: A Meta-Analysis of Functional Neuroimaging Data. Front Hum
Neurosci, 11, 154. https://doi.org/10.3389/fnhum.2017.00154
Niendam, T. A., Laird, A. R., Ray, K. L., Dean, Y. M., Glahn, D. C., & Carter, C. S. (2012).
Meta-analytic evidence for a superordinate cognitive control network subserving
diverse executive functions. Cogn Affect Behav Neurosci, 12(2), 241-268.
https://doi.org/10.3758/s13415-011-0083-5
Ordaz, S. J., Foran, W., Velanova, K., & Luna, B. (2013). Longitudinal growth curves of
brain function underlying inhibitory control through adolescence. J Neurosci, 33(46),
18109-18124. https://doi.org/https://doi.org/10.1523/JNEUROSCI.1741-13.2013
36
Padmala, S., & Pessoa, L. (2011). Reward reduces conflict by enhancing attentional control
and biasing visual cortical processing. J Cogn Neurosci, 23(11), 3419-3432.
https://doi.org/10.1162/jocn_a_00011
Padmanabhan, A., Geier, C. F., Ordaz, S. J., Teslovich, T., & Luna, B. (2011).
Developmental changes in brain function underlying the influence of reward
processing on inhibitory control. Dev Cogn Neurosci, 1(4), 517-529.
https://doi.org/10.1016/j.dcn.2011.06.004
Parr, A. C., Calabro, F., Tervo-Clemmens, B., Larsen, B., Foran, W., & Luna, B. (2022).
Contributions of dopamine-related basal ganglia neurophysiology to the
developmental effects of incentives on inhibitory control. Dev Cogn Neurosci, 54,
101100. https://doi.org/10.1016/j.dcn.2022.101100
Polderman, T. J., Benyamin, B., de Leeuw, C. A., Sullivan, P. F., van Bochoven, A., Visscher,
P. M., & Posthuma, D. (2015). Meta-analysis of the heritability of human traits based
on fifty years of twin studies. Nat Genet, 47(7), 702-709.
https://doi.org/10.1038/ng.3285
Rangel, A., Camerer, C., & Montague, P. R. (2008). A framework for studying the
neurobiology of value-based decision making. Nat Rev Neurosci, 9(7), 545-556.
https://doi.org/10.1038/nrn2357
Rosati, A. G., Emery Thompson, M., Atencia, R., & Buckholtz, J. W. (2023). Distinct
developmental trajectories for risky and impulsive decision-making in chimpanzees. J
Exp Psychol Gen, 152(6), 1551-1564. https://doi.org/10.1037/xge0001347
Sauce, B., Wiedenhoeft, J., Judd, N., & Klingberg, T. (2021). Change by challenge: A
common genetic basis behind childhood cognitive development and cognitive training.
npj Science of Learning, 6(1). https://doi.org/10.1038/s41539-021-00096-6
Shulman, E. P., Harden, K. P., Chein, J. M., & Steinberg, L. (2014). The Development of
Impulse Control and Sensation‐Seeking in Adolescence: Independent or
Interdependent Processes? Journal of Research on Adolescence, 26(1), 37-44.
https://doi.org/10.1111/jora.12181
Shulman, E. P., Smith, A. R., Silva, K., Icenogle, G., Duell, N., Chein, J., & Steinberg, L.
(2016). The dual systems model: Review, reappraisal, and reaffirmation. Dev Cogn
Neurosci, 17, 103-117. https://doi.org/10.1016/j.dcn.2015.12.010
Simmonds, D. J., Pekar, J. J., & Mostofsky, S. H. (2008). Meta-analysis of Go/No-go tasks
demonstrating that fMRI activation associated with response inhibition is task-
dependent. Neuropsychologia, 46(1), 224-232.
https://doi.org/10.1016/j.neuropsychologia.2007.07.015
Somerville, L. H., Hare, T., & Casey, B. J. (2011). Frontostriatal maturation predicts
cognitive control failure to appetitive cues in adolescents. J Cogn Neurosci, 23(9),
2123-2134. https://doi.org/10.1162/jocn.2010.21572
Somerville, L. H., Jones, R. M., & Casey, B. J. (2010). A time of change: behavioral and
neural correlates of adolescent sensitivity to appetitive and aversive environmental
cues. Brain Cogn, 72(1), 124-133. https://doi.org/10.1016/j.bandc.2009.07.003
Steinberg, L. (2008). A Social Neuroscience Perspective on Adolescent Risk-Taking. Dev
Rev, 28(1), 78-106. https://doi.org/10.1016/j.dr.2007.08.002
Steinberg, L., Icenogle, G., Shulman, E. P., Breiner, K., Chein, J., Bacchini, D., Chang, L.,
Chaudhary, N., Giunta, L. D., Dodge, K. A., Fanti, K. A., Lansford, J. E., Malone, P.
S., Oburu, P., Pastorelli, C., Skinner, A. T., Sorbring, E., Tapanya, S., Tirado, L. M.
U., Alampay, L. P., Al-Hassan, S. M., & Takash, H. M. S. (2018). Around the world,
adolescence is a time of heightened sensation seeking and immature self-regulation.
Dev Sci, 21(2). https://doi.org/10.1111/desc.12532
37
Telzer, E. H. (2016). Dopaminergic reward sensitivity can promote adolescent health: A new
perspective on the mechanism of ventral striatum activation. Dev Cogn Neurosci, 17,
57-67. https://doi.org/10.1016/j.dcn.2015.10.010
Tucker-Drob, E. M., Briley, D. A., & Harden, K. P. (2013). Genetic and Environmental
Influences on Cognition Across Development and Context. Curr Dir Psychol Sci,
22(5), 349-355. https://doi.org/10.1177/0963721413485087
Van Leijenhorst, L., Zanolie, K., Van Meel, C. S., Westenberg, P. M., Rombouts, S. A., &
Crone, E. A. (2010). What motivates the adolescent? Brain regions mediating reward
sensitivity across adolescence. Cereb Cortex, 20(1), 61-69.
https://doi.org/10.1093/cercor/bhp078
Verbruggen, F., Aron, A. R., Band, G. P., Beste, C., Bissett, P. G., Brockett, A. T., Brown, J.
W., Chamberlain, S. R., Chambers, C. D., Colonius, H., Colzato, L. S., Corneil, B. D.,
Coxon, J. P., Dupuis, A., Eagle, D. M., Garavan, H., Greenhouse, I., Heathcote, A.,
Huster, R. J., Jahfari, S., Kenemans, J. L., Leunissen, I., Li, C. R., Logan, G. D.,
Matzke, D., Morein-Zamir, S., Murthy, A., Pare, M., Poldrack, R. A., Ridderinkhof,
K. R., Robbins, T. W., Roesch, M., Rubia, K., Schachar, R. J., Schall, J. D., Stock, A.
K., Swann, N. C., Thakkar, K. N., van der Molen, M. W., Vermeylen, L., Vink, M.,
Wessel, J. R., Whelan, R., Zandbelt, B. B., & Boehler, C. N. (2019). A consensus
guide to capturing the ability to inhibit actions and impulsive behaviors in the stop-
signal task. Elife, 8. https://doi.org/10.7554/eLife.46323
Verbruggen, F., McLaren, I. P., & Chambers, C. D. (2014). Banishing the Control Homunculi
in Studies of Action Control and Behavior Change. Perspect Psychol Sci, 9(5), 497-
524. https://doi.org/10.1177/1745691614526414
Weigard, A., Matzke, D., Tanis, C., & Heathcote, A. (2023). A cognitive process modeling
framework for the ABCD study stop-signal task. Dev Cogn Neurosci, 59, 101191.
https://doi.org/10.1016/j.dcn.2022.101191
Weiss, H., & Luciana, M. (2022). Neurobehavioral maturation of motor response inhibition
in adolescence - A narrative review. Neurosci Biobehav Rev, 137, 104646.
https://doi.org/https://doi.org/10.1016/j.neubiorev.2022.104646
Williams, B. R., Ponesse, J. S., Schachar, R. J., Logan, G. D., & Tannock, R. (1999).
Development of inhibitory control across the life span. Dev Psychol, 35(1), 205-213.
https://doi.org/10.1037//0012-1649.35.1.205
Willoughby, T., Heffer, T., Good, M., & Magnacca, C. (2021). Is adolescence a time of
heightened risk taking? An overview of types of risk-taking behaviors across age
groups. Developmental Review, 61. https://doi.org/10.1016/j.dr.2021.100980
Witherby, A. E., & Carpenter, S. K. (2022). The rich-get-richer effect: Prior knowledge
predicts new learning of domain-relevant information. J Exp Psychol Learn Mem
Cogn, 48(4), 483-498. https://doi.org/10.1037/xlm0000996
Zacks, R. T., & Hasher, L. (1994). Directed ignoring: Inhibitory regulation of working
memory. In T. H. C. D. Dagenbach (Ed.), Inhibitory processes in attention, memory,
and language (pp. 241-264). Academic Press. https://psycnet.apa.org/record/1994-
97487-006
38
Appendix A
Note. Path diagram illustrating parallel process latent growth curve models evaluating the interplay between
reward sensitivity during reward anticipation stages and cognitive control, with baseline age as a covariate.
These models are designed to assess Hypotheses 1a, 1b, and 1c. Model 1 and Model 2 show reward sensitivity
measured in the right and left NAcc separately. Rectangular boxes represent observed variables, while ovals
denote latent variables. Unidirectional arrows signify the directional effects of one variable on another, while
bidirectional arrows denote correlations. Bold arrows highlight critical tests corresponding to the hypotheses.
39
Note. Path diagram illustrating parallel process latent growth curve models evaluating the interplay between
reward sensitivity during reward feedback stages and cognitive control, with baseline age as a covariate. These
models are designed to assess Hypotheses 1d, 1e, and 1f. Model 3 and Model 4 show reward sensitivity
measured in the right and left NAcc separately. For clarification on the symbolism of arrows, boxes, and ovals,
please refer to Figure A1.
40
Note. Path diagram illustrating parallel process latent growth curve models evaluating the interplay between
reward sensitivity during reward anticipation stages and FPN-NAcc functional connectivity, with baseline age
as a covariate. These models are designed to assess Hypotheses 2a and 2b. Model 5 and Model 6 show reward
sensitivity measured in the right and left NAcc separately. For clarification on the symbolism of arrows, boxes,
and ovals, please refer to Figure A1.
41
Note. Path diagram illustrating parallel process latent growth curve models evaluating the interplay between
reward sensitivity during reward anticipation stages and CON-NAcc functional connectivity, with baseline age
as a covariate. These models are designed to assess Hypotheses 2c and 2d. Model 7 and Model 8 show reward
sensitivity measured in the right and left NAcc separately. For clarification on the symbolism of arrows, boxes,
and ovals, please refer to Figure A1.
42
Note. Path diagram illustrating parallel process latent growth curve models evaluating the interplay between
reward sensitivity during reward Feedback stages and FPN-NAcc functional connectivity, with baseline age as a
covariate. These models are designed to assess Hypotheses 2e and 2f. Model 9 and Model 10 show reward
sensitivity measured in the right and left NAcc separately. For clarification on the symbolism of arrows, boxes,
and ovals, please refer to Figure A1.
43
Note. Path diagram illustrating parallel process latent growth curve models evaluating the interplay between
reward sensitivity during reward feedback stages and CON-NAcc functional connectivity, with baseline age as a
covariate. These models are designed to assess Hypotheses 2g and 2h. Model 11 and Model 12 show reward
sensitivity measured in the right and left NAcc separately. For clarification on the symbolism of arrows, boxes,
and ovals, please refer to Figure A1.
44
Note. Path diagram illustrating mediation models evaluating the mediational effect of FPN-NAcc and CON-
NAcc functional connectivity on the relationship between neural reward sensitivity during reward anticipation
stages and cognitive control, with baseline age as a covariate. These models are designed to assess Hypotheses
3a. Model 13 and Model 14 show reward sensitivity measured in the right and left NAcc separately. For
clarification on the symbolism of arrows, boxes, and ovals, please refer to Figure A1.
45
Note. Path diagram illustrating mediation models evaluating the mediational effect of FPN-NAcc and CON-
NAcc functional connectivity on the relationship between neural reward sensitivity during reward feedback
stages and cognitive control, with baseline age as a covariate. These models are designed to assess Hypotheses
3b. Model 15 and Model 16 show reward sensitivity measured in the right and left NAcc separately. For
clarification on the symbolism of arrows, boxes, and ovals, please refer to Figure A1.
46
Appendix B
Note. Path diagram with results illustrating parallel process latent growth curve models, evaluating the
interplay between reward sensitivity during reward anticipation stages and cognitive control, with baseline
age as a covariate. Solid lines among latent variables represent statistically significant correlations, while
dashed lines indicate non-significant correlations.
a To test the relationship between the slopes, we fixed the non-significant paths to zero, under the condition
that doing so would not lead to model non-convergence or misspecification.
***p < .001
47
Note. Path diagram with results illustrating parallel process latent growth curve models, evaluating the
interplay between reward sensitivity during reward feedback stages and cognitive control, with baseline
age as a covariate. Solid lines among latent variables represent statistically significant correlations, while
dashed lines indicate non-significant correlations.
a To test the relationship between the slopes, we fixed the non-significant paths to zero, under the condition
that doing so would not lead to model non-convergence or misspecification.
*p < .05. ***p < .001
48
Note. Path diagram with results illustrating parallel process latent growth curve models, evaluating the
interplay between reward sensitivity during reward anticipation stages and FPN-NAcc functional
connectivity, with baseline age as a covariate. Solid lines among latent variables represent statistically
significant correlations, while dashed lines indicate non-significant correlations.
a To test the relationship between the slopes, we fixed the non-significant paths to zero, under the condition
that doing so would not lead to model non-convergence or misspecification.
*p < .05. **p < .01
49
Note. Path diagram with results illustrating parallel process latent growth curve models, evaluating the
interplay between reward sensitivity during reward anticipation stages and CON-NAcc functional
connectivity, with baseline age as a covariate. Solid lines among latent variables represent statistically
significant correlations (none were found), while dashed lines indicate non-significant correlations.
a To test the relationship between the slopes, we fixed the non-significant paths to zero, under the condition
that doing so would not lead to model non-convergence or misspecification.
50
Note. Path diagram with results illustrating parallel process latent growth curve models, evaluating the
interplay between reward sensitivity during reward Feedback stages and FPN-NAcc functional
connectivity, with baseline age as a covariate. Solid lines among latent variables represent statistically
significant correlations, while dashed lines indicate non-significant correlations.
a To test the relationship between the slopes, we fixed the non-significant paths to zero, under the condition
that doing so would not lead to model non-convergence or misspecification.
*p < .05. ***p < .001
51
Note. Path diagram with results illustrating parallel process latent growth curve models, evaluating the
interplay between reward sensitivity during reward feedback stages and CON-NAcc functional
connectivity, with baseline age as a covariate. Solid lines among latent variables represent statistically
significant correlations, while dashed lines indicate non-significant correlations.
a To test the relationship between the slopes, we fixed the non-significant paths to zero, under the condition
that doing so would not lead to model non-convergence or misspecification.
*p < .05. ***p < .001
52
Note. Path diagram with correlation coefficients illustrating mediation models evaluating the mediational
effect of FPN-NAcc and CON-NAcc functional connectivity on the relationship between neural reward
sensitivity during reward anticipation stages and cognitive control, with baseline age as a covariate. Solid
lines among variables represent statistically significant correlations, while dashed lines indicate non-
significant correlations.
a Correlation coefficients in Model 14 were not displayed here because the model did not converge.
**p < .01
53
Note. Path diagram with correlation coefficients illustrating mediation models evaluating the mediational
effect of FPN-NAcc and CON-NAcc functional connectivity on the relationship between neural reward
sensitivity during reward feedback stages and cognitive control, with baseline age as a covariate. Solid
lines among variables represent statistically significant correlations, while dashed lines indicate non-
significant correlations.
a Model 16 was specified without including baseline age as a covariate
**p < .01
