Brain structural co-development is associated

with internalizing symptoms two years later in

the ABCD cohort

Journal of Behavioral YIHONG ZHAO1,2p , MARTIN P. PAULUS3,4 and

Addictions MARC N. POTENZA2,5,6,7,8,9pp
12 (2023) 1, 80–93
1
Columbia University School of Nursing, New York, NY, USA
DOI:
2
10.1556/2006.2023.00006 Department of Psychiatry, Yale University School of Medicine, New Haven, CT, USA
© 2023 The Author(s) 3
Laureate Institute for Brain Research, Tulsa, OK, USA
4
Department of Psychiatry, University of California San Diego, USA
5
Child Study Center, Yale School of Medicine, New Haven, CT, USA
6
Department of Neuroscience, Yale University, New Haven, CT, USA
7
FULL-LENGTH REPORT Wu Tsai Institute, Yale University, New Haven, CT, USA
8
Connecticut Mental Health Center, New Haven, CT, USA
9
Connecticut Council on Problem Gambling, Wethersfield, CT, USA

Received: November 2, 2022 • Revised manuscript received: January 23, 2023 • Accepted: February 13, 2023
Published online: March 20, 2023

ABSTRACT
Background and aims: About 1/3 of youth spend more than four hours/day engaged in screen media
activity (SMA). This investigation utilized longitudinal brain imaging and mediation analyses to examine
relationships among SMA, brain patterns, and internalizing problems. Methods: Data from Adolescent Brain
Cognitive Development (ABCD) participants with baseline and two-year follow-up structural imaging data
that passed quality control (N 5 5,166; 2,385 girls) were analyzed. Joint and Individual Variation Explained
(JIVE) identified a brain co-development pattern among 221 brain features (i.e., differences in surface area,
thickness, or cortical and subcortical gray-matter volume between baseline and two-year-follow-up data).
Generalized linear mixed-effect models investigated associations between baseline SMA, structural co-
development and internalizing and externalizing psychopathology at two-year follow-up. Results: SMA at
baseline was related to internalizing psychopathology at year 2 (β ¼ 0:020; SE ¼ 0:008; P ¼ 0:014) and
a structural co-development pattern (β ¼ 0:015; SE ¼ 0:007; P ¼ 0:029), where the co-development
pattern suggested that rates of change in gray-matter volumes of the brainstem, gray-matter volumes and/or
cortical thickness measures of bilateral superior frontal, rostral middle frontal, inferior parietal, and
inferior temporal regions were more similar than those in other regions. This component partially
mediated the relationship between baseline SMA and future internalizing problems (indirect effect 5 0.020,
P-value 5 0.043, proportion mediated: 2.24%). Discussion and conclusions: Greater youth engagement in
SMA at ages 9–10 years statistically predicted higher levels of internalizing two years later. This association
was mediated by cortical-brainstem circuitry, albeit with relatively small effect sizes. The findings may help
delineate processes contributing to internalizing behaviors and assist in identifying individuals at greater
risk for such problems.

KEYWORDS
p screen media activity, addictive behaviors, brain co-development pattern, internalizing behavior, substance use
Corresponding author.
E-mail: [email protected] problems, addiction circuit
pp
Corresponding author.
E-mail: [email protected]
INTRODUCTION
Screen media activity (SMA) is among the most common recreational behaviors of youth,
and its relationships to health and disease have been debated. Both better (Chaarani et al.,

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 81

2022; Sauce et al., 2022) and worse (Fors & Barch, 2019; function, as well as other developmental outcomes, in large-
Sanders, Parent, Forehand, Sullivan, & Jones, 2016; Twenge scale longitudinal studies.
& Campbell, 2018; Zhao et al., 2022) health measures have Individual variations in associations between certain behav-
been associated with SMA. Although the impact of SMA on iors and structural brain characteristics may reflect individual
mental health may be modest, the processes that contribute to differences in brain organization (i.e., how different parts of the
these associations remain poorly understood. Findings relating brain work in conjunction). From a developmental perspective,
the extent of SMA to psychopathology across different the brain is characterized by more synapses in infancy and early
developmental stages have been mixed (Eirich et al., 2022; childhood, followed by decreases related to active synaptic
Neville, McArthur, Eirich, Lakes, & Madigan, 2021). More pruning during later childhood, adolescence, and early adult-
recently, several studies (Kanai et al., 2012; Paulus et al., 2019; hood (Johnson, Blum, & Giedd, 2009; Lenroot & Giedd, 2006).
Su, Han, Yu, Wu, & Potenza, 2020; Von Der Heide, Vyas, & Synaptic pruning is a process of eliminating less-used synapses
Olson, 2014; Zhao et al., 2022) support the hypothesis that and strengthening frequently used connections. Despite multiple
SMA may alter brain structure and function. This view is in factors (e.g., genetics, nutrition, socioeconomic status) contrib-
line with the general observation (Alexander-Bloch, Giedd, & uting to brain development, it has been suggested that synaptic
Bullmore, 2013) that human brains have evolved to dynami- pruning occurring between early childhood and adulthood is
cally and adaptively reorganize network structures along largely activity-/experience-dependent and influenced by gene–
common developmental paths in responses to environmental environment interactions (Tierney & Nelson, 2009). Behavior-
characteristics. There is also some evidence that youth who ally, across different developmental stages, excessive SMA has
engage in high-frequency SMA may do so to the exclusion of been associated with negative health measures including obesity
other activities or schoolwork (Adelantado-Renau et al., 2019), (Robinson et al., 2017), depression (Boers et al., 2019; Madhav,
altering their developmental trajectories and leading to Sherchand, & Sherchan, 2017), poor academic performance
depression, anxiety and other concerns. Alternatively, in- (Adelantado-Renau et al., 2019), and delays in achieving
dividuals with mental health concerns may seek escape important developmental milestones in language, communica-
through SMA (Eden, Johnson, Reinecke, & Grady, 2020; Nabi, tion, and other domains (Madigan et al., 2019). Thus, there is a
Perez Torres, & Prestin, 2017; Wolfers & Schneider, 2021). need to investigate whether and how high-frequency SMA may
Therefore, more longitudinal studies are needed to examine directly impact brain development and potentially result in
these possibilities (Gonzalez-Bueso et al., 2018). worse health outcomes using longitudinal data.
Cross-sectional analyses of the ABCD data have associ- This investigation analyzed longitudinal data from the
ated various patterns of SMA with structural brain charac- Adolescent Brain Cognitive Development study (Volkow
teristics (Paulus et al., 2019). In addition, cortical-subcortical et al., 2018) to investigate whether covariation patterns of
structural covariation among regions in a thalamus-pre- changes in cortical and subcortical morphological measures
frontal-cortex (PFC)-brainstem circuit has been related to (i.e., brain co-development patterns) were related to SMA,
SMA and externalizing behaviors in children (Zhao et al., internalizing problems, and externalizing problems at
2022). Interestingly, a virtually identical thalamus-PFC- baseline. Based on the identification of such a covariation,
brainstem circuitry had been linked previously in adults to we next tested whether the brain co-development pattern
early onset of alcohol use (Zhao, Constable, Hien, Chung, & mediated the relationship between baseline total screen time
Potenza, 2021), which opens the possibility that potential and future psychopathological (internalizing, externalizing)
risk factors for extensive SMA share common structural concerns. We hypothesized that high-frequency SMA would
neural substrates with those for substance-use disorders relate to changes in cortical-subcortical co-development
(SUDs) (Zhao et al., 2022). Other data linking online social involving regions in a thalamus-PFC-brainstem circuit
networks to gray-matter (GM) density in the amygdala (Zhao et al., 2021), which in turn would lead to more
(Kanai et al., 2012; Von Der Heide et al., 2014) suggest that internalizing and externalizing problems two years later. We
popular non-gaming forms of SMA (e.g., social media use, a further explored gender-related differences related to brain
possible concern particularly for girls and women (Su et al., structural covariation patterns and SMA and internalizing
2020)) may impact developing brains (Crone & Konijn, and externalizing behaviors in these domains given the
2018). Certain types and patterns of SMA (e.g., high-fre- relevance of sex as a biological variable and gender-related
quency gaming that is typical of IGD) have been linked to differences in brain structure (Kaczkurkin, Raznahan, &
psychopathology (Gonzalez-Bueso et al., 2018) as well as Satterthwaite, 2019), SMA (Su et al., 2020) and internalizing
more positive measures of cognitive functioning (Chaarani and externalizing behaviors (Gutman & Codiroli McMaster,
et al., 2022; Sauce et al., 2022). While research criteria exist 2020; Hicks et al., 2007).
for IGD in DSM-5 and gaming disorder is included in the
ICD-11 (Billieux et al., 2017; King & Potenza, 2019; Potenza,
2018; Raffin et al., 2021; Saunders et al., 2017; Yao et al., METHODS
2017), other internet behaviors relating to other online be-
haviors (social networking, pornography viewing and
buying/shopping) have been considered as the bases for
Data source
possible disorders (Brand et al., 2020). Together, further This investigation used data from the ABCD study (Release
study is needed of SMA and its links to brain structure and 3.0) (Volkow et al., 2018), an ongoing large-scale longitudinal

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

82 Journal of Behavioral Addictions 12 (2023) 1, 80–93

cohort study following more than 11,000 participants aged asked to mark on a three-point Likert scale, with items rated
9–10 years at baseline, recruited from 21 sites across the in terms of how much the behavior is “true” for the
US. Data from both baseline (T0) and two-year follow-up respondent (0 5 not true, 1 5 somewhat true, and 2 5 very
(T2) were analyzed. Participants without passing structural true). Data were aggregated into eight empirically based
imaging quality control (QC) measures at both time points syndrome scales including withdrawn/depressed, somatic
and with missing age, sex/gender, race, screen time, and complaints, anxious/depressed, social problems, thought
psychopathology measures were excluded from the main problems, attention problems, rule-breaking behavior, and
analyses, resulting in a final sample size of 5,166 participants aggressive behavior. The internalizing problem score was the
(2,385 girls). average of the first four subscale scores, and the externalizing
problem score was the average of the rule-breaking and
Imaging preprocessing aggressive behavior subscales. The CBCL has been widely
used to measure youth general mental illness/psychopa-
The structural MRI (sMRI) data of all participants were
thology in both clinical and research settings (Ebesutani
collected on one of three types of 3T scanners (Siemens
et al., 2010). It has high test-retest reliability and excellent
Prisma, General Electric 750, and Philips) (Casey et al.,
internal consistency (Achenbach, 2009). The Cronbach’s
2018). The raw brain imaging data were pre-processed by
alphas for the CBCL syndrome scales range from 0.78 to
the ABCD investigative team using FreeSurfer with pipelines
0.94. In this study, the Cronbach’s alphas were 0.89 and 0.90
specifically designed to address several known challenges
for the internalizing and externalizing behaviors, respec-
(e.g., head motion, distortion, and intensity inhomogeneity)
tively. Raw scores were used, with higher scores reflecting
in MRI data pre-processing (Hagler et al., 2019). For
more severe psychopathology.
example, prospective motion correction (Tisdall et al., 2012;
White et al., 2010) has been implemented to reduce motion-
related image degradation (an issue particularly significant Statistical analyses
in image data acquisition with children). In the post-pre-
Prior to statistical modeling, group differences on de-
processing quality control (QC) step, for each cortical sur-
mographic characteristics and parental perception of par-
face reconstruction, trained technicians reviewed cortical
ticipants’ SMA behaviors were assessed by ANOVAs for
surface reconstruction and assigned a value ranging from
continuous measures, Chi-square tests for categorical mea-
0 (i.e., absent) to 3 (i.e., severe) to indicate the accuracy of
sures, and Cochran-Armitage trend tests for binary mea-
reconstruction in terms of motion, intensity inhomogeneity,
sures, respectively. Tests were adjusted for multiple
white-matter underestimation, pial overestimation, and
comparison problems by setting the False Discovery Rate at
magnetic-susceptibility artifacts. Participants were excluded
a 0.05 level.
from analyses if their sMRI data at T0 and/or T2 were rated
Cortical and subcortical structural co-development pat-
as ‘severe’ in any of the aforementioned QC criteria (Hagler
terns were characterized by the Joint and Individual Vari-
et al., 2019). Structural brain features, including surface area,
ance Explained (JIVE) approach (Lock et al., 2013). JIVE
thickness, and gray-matter volume (GMV), were measured
extends principal component analysis to detect common and
for 68 cortical regions based on the Desikan-Killiany atlas
distinct covariation patterns among multiple data sources.
(Desikan et al., 2006). Subcortical GMV (including bilateral
This method has been previously used to detect robust brain
amygdala, caudate, hippocampus, nucleus accumbens, pu-
structural covariation patterns across different develop-
tamen, pallidum, thalamus proper, ventral diencephalon,
mental stages (Zhao, Klein, Castellanos, & Milham, 2019),
and brainstem) were also included in analyses.
and detected cross-sectional structural covariation patterns
have been linked in adults to alcohol initiation prior to age
Measures 21 (Zhao et al., 2021) and in children to SMA (Zhao et al.,
SMA The primary SMA measure was baseline youth self- 2022). In this investigation, brain development measures
reported total number of hours spent on non-school-related were defined as differences in structural features between T2
SMAs on typical weekdays (i.e., Mondays through Fridays) and T0. JIVE analyses were performed using brain devel-
and weekends (i.e., Saturday and Sunday) in the past year. opment measures from different morphological features
Main types of SMA activities included watching TV shows/ including cortical and subcortical GM volume, thickness,
movies, watching videos, playing video games, texting on an and surface area. The joint components estimated by the
electronic device, visiting social networking sites, and video- JIVE stand for brain co-development patterns. Here, the
chatting. Data from the parent screentime survey at T2 were optimal numbers of joint and individual components were
used to provide a general view of types of devices partici- determined via permutation tests (Lock et al., 2013). Here,
pants owned, most frequently used screen media, and we aimed to understand the effects of baseline SMA on
parental perceptions of their children’s SMA. coordinated brain co-development and their associations
Psychopathology measures were based on internalizing with subsequent internalizing and externalizing behaviors.
and externalizing symptoms from the parent-reported Child Toward these goals, linear mixed models (LMMs) with
Behavior Checklist (CBCL) (Achenbach, 2009). Specifically, study site and family as random effects were used to assess
the CBCL measures include 118 item-level questions about whether: 1) SMA as measured by total screen time was
children’s behaviors in the past six months. Parents were associated with psychopathology two years later, and 2) the

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 83

effects of SMA were mediated by the identified brain co- 19.81 h; P-value < 0.001). The top three most popular elec-
development pattern (i.e., the JIVE component). Analyses tronic devices owned by participants were cell phones (n 5
were performed in R using functions from the lme4 package. 3,024, 62.0%), tablets (n 5 2,335, 45.2%), and laptops (n 5
Site and family were included as random effects. All mixed 1,650, 32.0%). The top three most frequently used screen
effects models included age, sex/gender, race (a four-level media included televisions (n 5 3,093, 59.9%), cell phones
variable), parental highest education level, family income, (n 5 2,867, 55.5%), and video games (n 5 2,437, 47.2%).
handedness, and the whole brain volume as covariates. R2 Lastly, most parents worried, to some extent, that their
reported in this study was the marginal R2 (Nakagawa & children spent too much time online (59.0%) and would
Schielzeth, 2013), representing the proportion of variance view inappropriate things online (61.0%), but only a small
explained by the variable of interest. A small portion of portion of parents worried that their children would post
participants had missing family income and/or parental inappropriate things online (17.7%). Finally, we note that
education level. The missing data were imputed using the basic demographic information between participants being
predictive mean matching method implemented in the R analyzed in this study were significantly different from these
package MICE (Zhang, 2016). Results from analyses of whose brain imaging data have yet to be released (Table A1
samples with listwise deletion agreed with the results using in Appendix).
imputed data.
Structural co-development patterns in early
Ethics adolescence
The current study involved analyses of deidentified data JIVE analysis of changes in brain features between two years
from the ABCD study and was exempted by the Yale IRB, following baseline (T2) and baseline (T0) identified one joint
the Yale Human Investigation Committee. Thus, the study is component that explained 11.9% of the total variation across
in accordance with the principles of the Declaration of all 221 brain features (Fig. 1). The components derived from
Helsinki. boys and girls correlated highly with those derived from the
entire sample (r 5 0.98–0.99). Specifically, this global brain
co-development pattern accounted for around 20% of vari-
RESULTS ation in both cortical thickness and GMV, 5% of variation in
subcortical GMV, and 1% of variation in surface area.
Sample characteristics and parental perceptions of Figure 2A highlights the top 10% of brain structures with
the largest loading magnitudes in the joint component.
participants’ SMA These regions included GMV of the brainstem, left putamen,
Demographic information of the 5,166 participants, among and bilateral rostral middle frontal, superior frontal, inferior
them 2,385 (46.17%) were female, showed no significant and superior parietal, inferior and middle temporal, left
effects of sex/gender on age, race/ethnicity, household in- precentral and left superior temporal cortices, and cortical
come, or parental education (Table 1). In terms of parental thickness of the bilateral entorhinal region and frontal and
perceptions of participants’ SMA (Table 2), on average, boys temporal poles. Pairwise correlations among brain features
reported more SMA (27.33 ± 21.04 h) than girls (23.04 ± with large magnitudes (mean 5 0.23, SD 5 0.12) were

Table 1. Demographic characteristics

Variable Girls Boys Test Statistics1 P-value2
Sample size (n, %) 2,385 (46.17%) 2,781 (53.83%)
Age (Months; mean ± SD) 119.01 ± 7.30 119.57 ± 7.45 2.744 0.075
Race (n, %) 10.109 1,000
White 1,607 (67.41%) 1,988 (71.49%)
Black 309 (12.96%) 314 (11.29%)
Asian 56 (2.35%) 56 (2.01%)
Other/Mixed 412 (17.28%) 423 (15.21%)
Parental highest education level (n, %) 0.789 1,000
Up to HS graduation/GED 274 (11.50%) 306 (11.02%)
Some College 621 (26.07%) 710 (25.57%)
Bachelor’s Degree 655 (27.50%) 762 (27.44%)
Post-Graduate Degree 832 (34.93%) 999 (35.97%)
Household Income (n, %) 2.623 1,000
[<$50,000] 610 (27.38%) 676 (26.18%)
[≥$50,000 & <$100,000] 691 (31.01%) 772 (29.90%)
[≥$100,000] 927 (41.61%) 1,134 (43.92%)
1
: T-statistic was reported for continuous measure, and Chi-square test statistic was reported for categorical outcomes.
2
: FDR adjusted P-values were reported.

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

84 Journal of Behavioral Addictions 12 (2023) 1, 80–93

Table 2. Parental reports of children’s screen media activity

Variable Girls Boys Test Statistics1 P-value2
Sample size (n, %) 2,385 (46.17%) 2,781 (53.83%)
Baseline total screen time 23.04 ± 19.81 27.33 ± 21.04 7.510 <0.001
(Hours; mean ± SD)
Internalizing scores at two-year 5.17 ± 5.82 4.75 ± 5.48 2.708 0.075
follow-up (mean ± SD)
Externalizing scores at two-year 3.47 ± 5.18 4.42 ± 5.74 6.196 <0.001
follow-up (mean ± SD)
Number of devices owned (n, %) 25.334 0.002
0 218 (9.14%) 333 (11.97%)
1 942 (39.51%) 1,187 (42.68%)
2 788 (33.05%) 853 (30.67%)
3 436 (18.29%) 408 (14.67%)
Number of frequently visited social 115.688 <0.001
media sites (n, %)
0 17 (0.71%) 13 (0.47%)
1 796 (33.39%) 662 (23.80%)
2 673 (28.23%) 650 (23.37%)
≥3 898 (37.67%) 1,456 (52.36%)
Parental concerns regarding their
children (n, %)
spends too much time online 1,372 (57.65%) 1,669 (60.19%) 3.313 1,000
will view inappropriate things online 1,374 (57.83%) 1,762 (63.68%) 18.745 0.003
will post inappropriate things online 392 (16.47%) 522 (18.81%) 4.651 1,000
Devices owned (n, %)
Cellular Phone 1,594 (66.86%) 1,610 (57.89%) 43.468 <0.001
Tablet 1,096 (45.97%) 1,239 (44.55%) 0.990 1,000
Laptop computer 779 (32.68%) 871 (31.32%) 1.025 1,000
Frequently used screen media (n, %)
Television 1,478 (62.00%) 1,615 (58.07%) 8.065 0.451
Video games 514 (21.56%) 1,923 (69.15%) 1164.5 <0.001
Cellular phone 1,498 (62.84%) 1,369 (49.23%) 95.706 <0.001
Tablet 913 (38.30%) 962 (34.59%) 7.461 0.511
Handheld video game devices 130 (5.45%) 440 (15.82%) 139.51 <0.001
Computer/laptop 846 (35.49%) 1,038 (37.32%) 1.793 1,000
1
: T-statistic was reported for continuous measure, and Chi-square test statistic was reported for categorical outcomes.
2
: FDR adjusted P-values were reported.

stronger than those among features with small loading

magnitudes (mean 5 0.08, SD 5 0.13) in the joint
component (Fig. 3), indicating brain features with large
magnitudes in the joint component co-develop together
more closely than the other brain features. JIVE analyses
supported the covariation of brain development among key
regions in our hypothesized thalamus-PFC-brainstem cir-
cuitry that has been previously linked in adults to early
initiation of alcohol use (Zhao et al., 2021) and in children to
SMA (Zhao et al., 2022) in cross-sectional studies, albeit
perhaps with less loading from the thalamus. In bivariate
association analyses, this brain co-development pattern
correlated with baseline total screen time and internalizing
Fig. 1. Proportion of the variance of brain-feature change scores
explained by JIVE components. The brain co-development pattern
behavior at two-year follow-up with Pearson correlations of
accounted for around 20% of variation in both cortical thickness 0.04 each (Table A2 in Appendix).
and GMV, 5% of variation in subcortical GMV, and 1% of variation Controlling for potentially confounding variables, higher
in surface area. A large portion of total variation was explained by baseline total screen time was associated with larger joint
the residuals, indicating significant individual variations in brain component scores (β ¼ 0:015; SE ¼ 0:007; P ¼ 0:029), and
development. Subcort 5 subcortical baseline total screen time explained 0.1% of variation in the

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 85

Fig. 2 (A) Plots are shown of brain regions in the brain co-development pattern with loadings larger than 0.105 (approximately corre-
sponding to the top 10% of regions with the largest loading magnitudes). This JIVE component was related to both baseline screen time
(β 5 0.015, P 5 0.029) and internalizing behaviors at two-year follow-up (β 5 0.020, P 5 0.014). (B) The brain co-developmental pattern
partially mediated the effects of baseline screen time on internalizing behavior at two-year follow-up, explaining approximately 2.2% of the
SMA effects on future internalizing behavior. The direct effect of baseline total screen time on internalizing behavior two years later was
estimated to be 0.82

joint component. To facilitate the interpretation, percentage high-frequency SMA demonstrated slower expansion in
of brain changes in regions with large magnitudes in the subcortical regions (e.g., brainstem, left putamen). Mean (SD)
joint component are presented in Table A3. We divided percentage of change in these regions was 2.11% (0.035) for
participants into those with low-frequency (≤1 h/day), high-frequency and 2.67% (0.036) for low-frequency SMA (P-
moderate-frequency (>1 h/day and ≤ 7 h/day), or high-fre- value 5 0.01). In contrast, individuals with high-frequency
quency screen use (> 7 h/day) based on findings from a SMA had greater GMV reduction (2.00% (0.034)) as
large-scale study (Twenge & Campbell, 2018). Those with opposed to 1.52% (0.027) for those with low-frequency SMA
high-frequency SMA were reported to have at least twice the (P-value 5 0.02) in key regions (e.g., superior frontal and
likelihood of experiencing a diverse range of psychological rostral middle frontal cortices) in the hypothesized thalamus-
problems including anxiety and depression compared to PFC-brainstem circuitry. This global structural co-develop-
those with no more than one hour/day SMA (Twenge & ment pattern provided a proxy measure of imbalanced brain
Campbell, 2018). In this study, compared to individuals with structural development (mainly in GMV and cortical thick-
low-frequency and moderate-frequency SMA, those with ness) among cortical and subcortical regions. Baseline

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

86 Journal of Behavioral Addictions 12 (2023) 1, 80–93

Fig. 3 Compared to features with small loading magnitudes, pairwise correlation of brain change scores were much larger among features
with large loading magnitudes in the JIVE joint component. The dotted line stands for the mean correlation coefficient across all pairwise
correlations

internalizing (β ¼ 0:008; SE ¼ 0:025; P ¼ 0:74) and exter- Girls reported more engagement with cellular phones (boys:
nalizing (β ¼ −0:004; SE ¼ 0:023; P ¼ 0:88) scores were not 49.23%, girls: 62.84%; P-value < 0.001). Boys and girls did
associated with the joint component score. not differ on internalizing scores (boys: 4.75 ± 5.48, girls:
5.17 ± 5.82; P-value 5 0.075). Parents were more likely to
Mediation models investigating baseline SMA have concerns that boys more so than girls would view
inappropriate material online (boys: 63.68%, girls: 57.83%;
relationships with internalizing and externalizing
P-value 5 0.003). The extent to which this may apply to
behaviors two years later certain forms of SMA more prevalent in males (e.g.,
Higher baseline total SMA was associated with higher pornography viewing) or other factors requires further
internalizing (β ¼ 0:013; SE ¼ 0:004; P ¼ 0:002) and exter- study.
nalizing (β ¼ 0:024; SE ¼ 0:004; P<0:001) behaviors two- We note there were no significant differences between
years later, suggesting that an increase in screen time by one boys and girls on SMA/future-psychopathology relation-
hour/day corresponded to an increase of internalizing scores ships and on brain co-development-pattern/future-psycho-
by 0.093 points and externalizing scores by 0.16 points two pathology relationships, respectively. SMA explained a small
years later. Controlling for baseline total screen time, higher amount of variation on future internalizing (R2 5 0.30% in
joint component scores were associated with higher inter- boys vs R2 5 0.10% in girls) and externalizing (R2 5 0.88%
nalizing scores two years later (β ¼ 0:020; SE ¼ 0:008; in boys vs R2 5 0.54% in girls) behaviors. A small (R2 5
P ¼ 0:014; Fig. 2B) but not with externalizing scores 0.13% in boys vs R2 5 0.14% in girls) but significant brain
(β ¼ 0:009; SE ¼ 0:008; P ¼ 0:26Þ: Subsequent mediation co-development pattern effect on future internalizing
analysis showed that the joint component partially mediated behavior was observed in both boys and girls.
the effects of total screen time on internalizing behavior two
years later (indirect effect 5 0.020, 95% CI: (0.0003, 0.05),
P-value 5 0.043, proportion mediated: 2.24%, Fig. 2B) but
not externalizing behavior (indirect effect 5 0.009, 95% CI:
DISCUSSION
(0.007, 0.03), P-value 5 0.29, proportion mediated: 0.51%).
This investigation, which examined in youth 9–10 years of
age, the relationships among baseline SMA, brain structural
Gender-related differences co-development, and psychopathological problems two
Tables 1 and 2 summarize differences in boys and girls. Boys years later, yielded three main results. First, more extensive
showed more externalizing behaviors than girls (boys: 4.42 ± baseline SMA was related to greater internalizing and
5.74, girls: 3.47 ± 5.18; P-value < 0.001), SMA involving externalizing behaviors two years later. Second, there was a
videogames (boys: 69.15%, girls: 21.56%; P-value < 0.001) pattern of structural brain co-development indicating that
and visitation of social media sites (boys: 52.36% frequently brain features within a medial PFC-brainstem circuit (Sici-
visited at least three sites, girls: 37.67%; P-value < 0.001). liano et al., 2019) previously linked to early alcohol use in

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 87

adult humans and SMA in children shared similar structural suggest that imbalanced brain structural co-development
features. Third, this pattern of structural brain co-develop- among cortical and subcortical regions may consequently
ment was associated with baseline SMA and mediated re- lead to heightened internalizing behavior, and not vice versa.
lationships with subsequent internalizing but not This result is not necessarily surprising given that brain
externalizing behaviors two years later. development is a complex process including genetics, envi-
This investigation focused on cortical and subcortical ronmental and lifestyle factors, sleep disturbances, and
covariation patterns among brain structural changes from mental health issues that may contribute to and shape brain
baseline (T0 at ages 9–10 years) to a two-year follow-up (T2 development in children and adolescents. For example, in
at ages 11–12 years). The data showed that in general cross-sectional studies, smaller brainstem GMVs among
subcortical GMV and surface area continued to expand from participants with mild traumatic brain injuries (Kim et al.,
baseline to two-year follow-up, while thickness and GMV in 2021) and among children with autism (Hashimoto et al.,
the cortical regions on average started to decline (Table A1). 1992) have been reported. In addition, higher levels of
The observed structural expansion/reduction pattern was in physical activity have been associated with increased cortical
line with neurodevelopmental milestones (Bethlehem et al., thickness in older adults (Lee et al., 2016; Raffin et al., 2021).
2022) in late childhood and early adolescence. Our JIVE However, in children and adolescents, cortical thickness
analyses indicated that 61.8% of total variation (Fig. 1) in and/or GMV reduction have been associated with both
brain structural changes cannot be explained by any sys- higher physical activity and better academic achievement
tematic variation patterns among the tested regions, sug- (Chaddock-Heyman et al., 2015), childhood abuse (Gold
gesting that there exists extensive individual variability in et al., 2016), and impulsive choice in adolescents (Pehliva-
brain structural development. Despite these individual dif- nova et al., 2018). In short, multiple findings suggest that
ferences, JIVE analyses identified a brain structural co- brain structures change dynamically across the life course
development pattern across various structural features and (Sowell et al., 1999) in response to environmental demands
brain regions. This structural co-development pattern sug- (Spear, 2000).
gested that participants with high-frequency SMA had a Children and adolescents experience critical develop-
greater level of brain development imbalance among cortical mental brain changes. According to the selective-elimination
and subcortical regions, especially those implicated in a hypothesis (Changeux & Danchin, 1976), synaptic pruning
medial-PFC-brainstem circuit (Siciliano et al., 2019) and key during adolescence is highly specific and activity-dependent.
regions in a thalamus-PFC-brainstem circuit (Zhao et al., Brain connections that are rarely used will be selectively
2021, 2022). This finding is important because these circuits pruned to make the brain work more efficiently, and con-
have been linked to compulsive drinking in mice (Siciliano nections that are frequently used will become strengthened.
et al., 2019), early initiation of alcohol use in young adult Therefore, children and adolescents have opportunities to
humans (Zhao et al., 2021), and SMA in human children make their brains healthier by spending their time wisely on
(Zhao et al., 2022) in cross-sectional studies. Our data sug- activities that promote good cognitive and mental health
gest abnormal imbalance in brain development (i.e., slower outcomes. The extent to which SMA may promote positive
subcortical GMV expansion and greater cortical GMV or negative outcomes during development has been debated.
reduction in individuals with high versus low SMA) in key However, high-frequency or excessive SMA has been more
regions involved in the aforementioned neural circuits. This consistently linked to negative health measures, and the
finding is in line prior data linking risky behavior with current investigation suggests a possible brain mechanism
imbalances in brain development. Based on previous studies accounting for some of this relationship.
(Casey, Getz, & Galvan, 2008; Constantinidis & Luna, 2019; In partial support of our hypothesis, we found that the
Kuss, Pontes, & Griffiths, 2018; Luna & Wright, 2016; brain structural co-variation pattern mediated the relation-
Somerville, Jones, & Casey, 2010), neurodevelopmental ship between SMA in 9-10-year-old children and internal-
changes during the second decade of life are characterized by izing but not externalizing psychopathology two years later.
greater cortical control over subcortical circuits. These However, SMA at baseline was linked to both internalizing
changes are accompanied by pruning of cortical neurons and externalizing psychopathology two years later. These
leading to thinning of cortical GM. Thus, alterations in these findings suggest that the relationship between SMA and
processes may lead to less cognitive control over subcortical externalizing behaviors may be linked through independent
circuits, which has been proposed to be a basis for an brain mechanisms at this developmental period, and future
imbalance in brain development related to risk behaviors studies should seek to identify such mechanisms. Nonethe-
and psychopathology. less, the finding of this link with subsequent internalizing
Using the ABCD longitudinal data, our findings psychopathology may be particularly clinically relevant
demonstrated that extensive SMA (i.e., more than seven given links between high-frequency SMA or internet
hours per day) may impact brain development processes, addiction and internalizing disorders like depression (Liu
leading to imbalanced brain development of specific cortical et al., 2022; Lozano-Blasco & Cortés-Pascual, 2020). Taken
and subcortical regions. Given that baseline SMA, but not together, individuals who engage in high-frequency SMA
internalizing and externalizing behaviors, was associated (i.e., >7 h a day) also showed cortical and subcortical
with the global structural co-development pattern, along developmental patterns that were associated with future
with the results from the mediation analyses, our findings internalizing behaviors. We acknowledge that the strength of

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

88 Journal of Behavioral Addictions 12 (2023) 1, 80–93

the SMA and brain co-development relationship was rela- gender-/sex-related differences by types of SMA warrant
tively weak, and the size of the mediation effect was small. further investigation.
Therefore, our results should be interpreted with caution. In conclusion, our findings provide a deeper under-
This investigation has limitations. First, SMA data were standing of how SMA may affect brain development with
based on youth self-report. Objective screentime data have respect to the development of internalizing psychopathol-
advantages, as the self-report data could be influenced by ogy. While providing a potential brain mechanism under-
individual differences in psychological and contextual fac- lying some of the relationship between SMA and subsequent
tors (Hodes & Thomas, 2021), resulting in overestimation or internalizing problems, more effort is needed to identify
underestimation of actual screentime. Importantly, the other sources promoting both internalizing and external-
amount of screentime is not equivalent to the extent of izing psychopathology, particularly those amenable to in-
problematic use (Twenge & Farley, 2021), especially in terventions that will serve to prevent or ameliorate mental
children and adolescents who may not have full control over health concerns related to SMA.
their access to media (e.g., related to different parenting
approaches or school rules). Some types of screen time may
be more problematic and disruptive and carry greater po- Funding sources: This study was supported by grants from
tential for future disordered use. Thus, it is important to Children and Screens (CSDMB001) and the NIH (RF1
consider different screen modalities beyond total screen use MH128614, R01 AA029611). The views presented in this
and additional potential influences in future studies. Second, manuscript represent those of the authors and not neces-
whether SMA-related brain co-development pattern in this sarily those of the funding agencies.
investigation can be generalized to other age groups or to the
full ABCD or other samples needs to be further investigated. Authors’ contribution: Y.Z. performed data analyses and wrote
The ABCD participants whose data have yet to be released at the original draft. Y.Z. and M.N.P obtained funding and
the time of this study were in general younger and had a conceptualized the study. M.N.P. and M.P. provided critical
higher portion of non-White and lower-income-family revisions and intellectual input. All authors have read and
participants (Table A1). Thus, future study of the ABCD agreed to the published version of the manuscript. All authors
data appears indicated. Additionally, the impact of screen had full access to all data in the study and take responsibility for
exposure on brain co-development pattern may be age- the integrity of the data and the accuracy of the data analysis.
dependent, as synaptic pruning in different brain regions
occurs at different age periods (Kolb & Gibb, 2011), also Conflict of interest: The authors declare no conflicts of in-
suggesting the need for additional studies. Third, in cross- terest. Dr. Potenza has consulted for Opiant Therapeutics,
sectional analyses, we have found that GMV of the thalamus Game Day Data, Baria-Tek, the Addiction Policy Forum,
proper covaries with GMV of the brainstem and prefrontal AXA and Idorsia Pharmaceuticals; has been involved in a
regions across different ages (Zhao et al., 2021, 2022). It is patent application with Yale University and Novartis; has
currently unclear why the rates of change in thalamic GMV received research support from Mohegan Sun Casino and the
deviate from those in the brainstem and superior and rostral National Center for Responsible Gaming; has participated in
middle frontal regions, and this should be further explored surveys, mailings or telephone consultations related to drug
in future studies. Fourth, we only explored the impact of addiction, impulse-control disorders or other health topics;
SMA on future psychopathological problems. How SMA has consulted for and/or advised gambling and legal entities
may interact with other potential risk factors, such as family on issues related to impulse-control/addictive disorders; has
conflict (Mathiesen, Sanson, Stoolmiller, & Karevold, 2009), provided clinical care in a problem gambling services pro-
harsh parenting style (Pinquart, 2017), and adverse child- gram; has performed grant reviews for research-funding
hood events (Bolger & Patterson, 2001) that may influence agencies; has edited journals and journal sections; has given
the development of internalizing behaviors, warrants further academic lectures in grand rounds, CME events and other
investigation. In addition, bidirectional relationships be- clinical or scientific venues; and has generated books or book
tween SMA and psychopathology may exist (Gamez-Gua- chapters for publishers of mental health texts. Dr. Potenza is
dix, 2014; Jeong et al., 2019), and these should be an associate editor of the Journal of Behavioral Addictions.
investigated using data from future releases. Fifth, consid- The other authors do not report disclosures.
erable gender-/sex-related differences in types of SMA were
identified in this study and have been reported elsewhere Acknowledgement: Data used in the preparation of this
(Twenge & Farley, 2021). However, the SMA-psychopa- article were obtained from the Adolescent Brain Cognitive
thology and the SMA-brain co-development pattern re- Development (ABCD) Study (https://abcdstudy.org), held in
lationships did not differ significantly between boys and the NIMH Data Archive (NDA). This is a multisite, longi-
girls. The ABCD data included participants aged 9–10 years tudinal study designed to recruit more than 10,000 children
at baseline; however, as reported elsewhere, significant age 9–10 and follow them over 10 years into early adult-
changes in SMA occur during the second decade of life hood. The ABCD Study® is supported by the National In-
(Orben et al., 2022). It is possible that sex differences may stitutes of Health and additional federal partners under
emerge in the ABCD data set as participants age and award numbers U01DA041048, U01DA050989, U01DA05
experience pubertal changes. Thus, brain correlates of 1016, U01DA041022, U01DA051018, U01DA051037,

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 89

U01DA050987, U01DA041174, U01DA041106, U01DA should be considered as disorders in the International Classi-
041117, U01DA041028, U01DA041134, U01DA050988, fication of Diseases (ICD-11) designation of "other specified
U01DA051039, U01DA041156, U01DA041025, U01DA disorders due to addictive behaviors. Journal of Behaviors and
041120, U01DA051038, U01DA041148, U01DA041093, Addiction. https://doi.org/10.1556/2006.2020.00035.
U01DA041089, U24DA041123, U24DA041147. A full list of Casey, B. J., Cannonier, T., Conley, M. I., Cohen, A. O., Barch, D. M.,
supporters is available at https://abcdstudy.org/federal- Heitzeg, M. M., … Workgroup, A. I. A. (2018, Aug). The
partners.html. A listing of participating sites and a com- adolescent brain cognitive development (ABCD) study: Imaging
plete listing of the study investigators can be found at acquisition across 21 sites [Multicenter study research support,
https://abcdstudy.org/consortium_members/. ABCD con- N.I.H., extramural research support, U.S. Gov’t, non-P.H.S.
sortium investigators designed and implemented the study Review]. Developmental Cognitive Neuroscience, 32, 43–54.
and/or provided data but did not necessarily participate in https://doi.org/10.1016/j.dcn.2018.03.001.
the analysis or writing of this report. This manuscript re- Casey, B. J., Getz, S., & Galvan, A. (2008). The adolescent brain.
flects the views of the authors and may not reflect the Developmental Review, 28(1), 62–77. https://doi.org/10.1016/j.
opinions or views of the NIH or ABCD consortium in- dr.2007.08.003.
vestigators. The ABCD data repository grows and changes Chaarani, B., Ortigara, J., Yuan, D., Loso, H., Potter, A., & Garavan,
over time. The ABCD data used in this report came from the H. P. (2022, Oct 3). Association of video gaming with cognitive
NDA collection #2573 (DOI: 10.15154/1527880). performance among children. JAMA Network Open, 5(10),
e2235721. https://doi.org/10.1001/jamanetworkopen.2022.35721.
Chaddock-Heyman, L., Erickson, K. I., Kienzler, C., King, M.,
Pontifex, M. B., Raine, L. B., … Kramer, A. F. (2015). The role
REFERENCES of aerobic fitness in cortical thickness and mathematics
achievement in preadolescent children. PLoS One, 10(8),
Achenbach, T. M. (2009). The achenbach system of empirically e0134115. https://doi.org/10.1371/journal.pone.0134115.
based assessment (ASEBA): Development, findings, theory, and Changeux, J. P., & Danchin, A. (1976, Dec 23–30). Selective sta-
applications. Burlington, VT: University of Vermont Research bilisation of developing synapses as a mechanism for the
Center for Children, Youth, & Families. specification of neuronal networks. Nature, 264(5588), 705–712.
Adelantado-Renau, M., Moliner-Urdiales, D., Cavero-Redondo, I., https://doi.org/10.1038/264705a0.
Beltran-Valls, M. R., Martinez-Vizcaino, V., & Alvarez-Bueno, C. Constantinidis, C., & Luna, B. (2019, Sep). Neural substrates of
(2019, Nov 1). Association between screen media use and inhibitory control maturation in adolescence. Trends in Neu-
academic performance among children and adolescents: A sys- rosciences, 42(9), 604–616. https://doi.org/10.1016/j.tins.2019.
tematic review and meta-analysis. JAMA Pediatrics, 173(11), 07.004.
1058–1067. https://doi.org/10.1001/jamapediatrics.2019.3176. Crone, E. A., & Konijn, E. A. (2018, Feb 21). Media use and brain
Alexander-Bloch, A., Giedd, J. N., & Bullmore, E. (2013, May). development during adolescence. Nature Communications,
Imaging structural co-variance between human brain regions 9(1), 588. https://doi.org/10.1038/s41467-018-03126-x.
[Research support, N.I.H., extramural research support, non- Desikan, R. S., Segonne, F., Fischl, B., Quinn, B. T., Dickerson, B. C.,
U.S. Gov’t research support, U.S. Gov’t, P.H.S. Review]. Nature Blacker, D., … Killiany, R. J. (2006, Jul 1). An automated
Reviews Neuroscience, 14(5), 322–336. https://doi.org/10.1038/ labeling system for subdividing the human cerebral cortex on
nrn3465. MRI scans into gyral based regions of interest [Research Sup-
Bethlehem, R. A. I., Seidlitz, J., White, S. R., Vogel, J. W., Anderson, port, N.I.H., Extramural Research Support, Non-U.S. Gov’t].
K. M., Adamson, C., … Alexander-Bloch, A. F. (2022, Apr). Neuroimage, 31(3), 968– 980. https://doi.org/10.1016/j.
Brain charts for the human lifespan. Nature, 604(7906), neuroimage.2006.01.021.
525–533. https://doi.org/10.1038/s41586-022-04554-y. Ebesutani, C., Bernstein, A., Nakamura, B. J., Chorpita, B. F., Higa-
Billieux, J., King, D. L., Higuchi, S., Achab, S., Bowden-Jones, H., McMillan, C. K., Weisz, J. R., &amp; The Research Network on
Hao, W., … Poznyak, V. (2017, Sep 1). Functional impairment Youth Mental, H. (2010, Sep). Concurrent validity of the child
matters in the screening and diagnosis of gaming disorder. behavior checklist DSM-oriented scales: Correspondence with
Journal of Behaviors and Addiction, 6(3), 285–289. https://doi. DSM diagnoses and comparison to syndrome scales. Journal of
org/10.1556/2006.6.2017.036. Psychopathology and Behavioral Assessment, 32(3), 373–384.
Boers, E., Afzali, M. H., Newton, N., & Conrod, P. (2019, Sep 1). https://doi.org/10.1007/s10862-009-9174-9.
Association of screen time and depression in adolescence. Eden, A. L., Johnson, B. K., Reinecke, L., & Grady, S. M. (2020).
JAMA Pediatrics, 173(9), 853–859. https://doi.org/10.1001/ Media for coping during COVID-19 social distancing: Stress,
jamapediatrics.2019.1759. anxiety, and psychological well-being. Frontiers in Psychology,
Bolger, K. E., & Patterson, C. J. (2001, Fall). Pathways from child 11, 577639. https://doi.org/10.3389/fpsyg.2020.577639.
maltreatment to internalizing problems: Perceptions of control Eirich, R., McArthur, B. A., Anhorn, C., McGuinness, C., Chris-
as mediators and moderators. Development and Psychopathol- takis, D. A., & Madigan, S. (2022, May 1). Association of screen
ogy, 13(4), 913–940. https://www.ncbi.nlm.nih.gov/pubmed/ time with internalizing and externalizing behavior problems in
11771914. children 12 years or younger: A systematic review and meta-
Brand, M., Rumpf, H. J., Demetrovics, Z., A, M. U., Stark, R., King, analysis. JAMA Psychiatry, 79(5), 393–405. https://doi.org/10.
D. L., … Potenza, M. N. (2020, Jun 30). Which conditions 1001/jamapsychiatry.2022.0155.

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

90 Journal of Behavioral Addictions 12 (2023) 1, 80–93

Fors, P. Q., & Barch, D. M. (2019, Dec). Differential relationships of multimodal neuroimaging [Review research support, N.I.H.,
child anxiety and depression to child report and parent report intramural research support, non-U.S. Gov’t research support,
of electronic media use. Child Psychiatry and Human Devel- N.I.H., extramural]. Neuropsychopharmacology, 44(1), 71–85.
opment, 50(6), 907–917. https://doi.org/10.1007/s10578-019- https://doi.org/10.1038/s41386-018-0111-z.
00892-7. Kanai, R., Bahrami, B., Roylance, R., & Rees, G. (2012, Apr 7).
Gamez-Guadix, M. (2014, Nov). Depressive symptoms and prob- Online social network size is reflected in human brain structure.
lematic internet use among adolescents: Analysis of the longi- Proceedings of the Biological Sciences, 279(1732), 1327–1334.
tudinal relationships from the cognitive-behavioral model. https://doi.org/10.1098/rspb.2011.1959.
Cyberpsychology, Behavior and Social Networking, 17(11), Kim, E., Seo, H. G., Lee, H. H., Lee, S. H., Choi, S. H., Yoo, R. E., …
714–719. https://doi.org/10.1089/cyber.2014.0226. Oh, B. M. (2021, May 1). Reduced brainstem volume after mild
Gold, A. L., Sheridan, M. A., Peverill, M., Busso, D. S., Lambert, H. K., traumatic brain injury. Archives of Physical Medicine and
Alves, S., … McLaughlin, K. A. (2016, Oct). Childhood abuse Rehabilitation, 100(5), 473–482. https://doi.org/10.1097/PHM.
and reduced cortical thickness in brain regions involved in 0000000000001580.
emotional processing. Clinical Psychology & Psychotherapy, King, D. L., & Potenza, M. N. (2019, Jan). Not playing around:
57(10), 1154–1164. https://doi.org/10.1111/jcpp.12630. Gaming disorder in the international classification of diseases
Gonzalez-Bueso, V., Santamaria, J. J., Fernandez, D., Merino, (ICD-11). Journal of Adolescent Health, 64(1), 5–7. https://doi.
L., Montero, E., & Ribas, J. (2018, Apr 3). Association between org/10.1016/j.jadohealth.2018.10.010.
internet gaming disorder or pathological video-game use and Kolb, B., & Gibb, R. (2011, Nov). Brain plasticity and behaviour in
comorbid psychopathology: A comprehensive review. Interna- the developing brain. Journal of the Canadian Academy of Child
tional Journal of Environmental Research and Public Health, and Adolescent Psychiatry, 20(4), 265–276. https://www.ncbi.
15(4). https://doi.org/10.3390/ijerph15040668. nlm.nih.gov/pubmed/22114608.
Gutman, L. M., & Codiroli McMaster, N. (2020, May). Gendered Kuss, D. J., Pontes, H. M., & Griffiths, M. D. (2018). Neurobio-
pathways of internalizing problems from early childhood to logical correlates in internet gaming disorder: A systematic
adolescence and associated adolescent outcomes. Journal of literature review. Frontiers in Psychiatry, 9, 166. https://doi.org/
Abnormal Psychology, 48(5), 703–718. https://doi.org/10.1007/ 10.3389/fpsyt.2018.00166.
s10802-020-00623-w. Lee, J. S., Shin, H. Y., Kim, H. J., Jang, Y. K., Jung, N. Y., Lee, J., …
Hagler, D. J., Jr., Hatton, S., Cornejo, M. D., Makowski, C., Fair, D. A., Seo, S. W. (2016, Apr 11). Combined effects of physical exercise
Dick, A. S., … Dale, A. M. (2019, Nov 15). Image processing and education on age-related cortical thinning in cognitively
and analysis methods for the adolescent brain cognitive devel- normal individuals. Scientific Reports, 6, 24284. https://doi.org/
opment study. Neuroimage, 202, 116091. https://doi.org/10.1016/ 10.1038/srep24284.
j.neuroimage.2019.116091. Lenroot, R. K., & Giedd, J. N. (2006). Brain development in chil-
Hashimoto, T., Tayama, M., Miyazaki, M., Sakurama, N., Yoshi- dren and adolescents: Insights from anatomical magnetic
moto, T., Murakawa, K., & Kuroda, Y. (1992, Mar). Reduced resonance imaging. Neuroscience and Biobehavioral Reviews,
brainstem size in children with autism. Brain & Development, 30(6), 718–729. https://doi.org/10.1016/j.neubiorev.2006.06.
14(2), 94–97. https://doi.org/10.1016/s0387-7604(12)80093-3. 001.
Hicks, B. M., Blonigen, D. M., Kramer, M. D., Krueger, R. F., Liu, M., Kamper-DeMarco, K. E., Zhang, J., Xiao, J., Dong, D., &
Patrick, C. J., Iacono, W. G., & McGue, M. (2007, Aug). Gender Xue, P. (2022, Apr 24). Time spent on social media and risk of
differences and developmental change in externalizing disor- depression in adolescents: A dose-response meta-analysis. In-
ders from late adolescence to early adulthood: A longitudinal ternational Journal of Environmental Research and Public
twin study. Journal of Abnormal Psychology, 116(3), 433–447. Health, 19(9). https://doi.org/10.3390/ijerph19095164.
https://doi.org/10.1037/0021-843X.116.3.433. Lock, E. F., Hoadley, K. A., Marron, J. S., & Nobel, A. B. (2013, Mar
Hodes, L. N., & Thomas, K. G. F. (2021). Smartphone Screen Time: 01). Joint and individual variation explained (Jive) for inte-
Inaccuracy of self-reports and influence of psychological and grated analysis of multiple data types. Annals of Applied Sta-
contextual factors. Computers in Human Behavior, 115, 106616. tistics, 7(1), 523–542. https://doi.org/10.1214/12-AOAS597.
https://doi.org/https://doi.org/10.1016/j.chb.2020.106616. Lozano-Blasco, R., & Cortés-Pascual, A. (2020). Problematic internet
Jeong, H., Yim, H. W., Lee, S. Y., Lee, H. K., Potenza, M. N., Jo, S. J., uses and depression in adolescents: A meta-analysis. Comunicar:
& Son, H. J. (2019, Dec 1). Reciprocal relationship between Media Education Research Journal, 28(63), 103–113.
depression and internet gaming disorder in children: A 12- Luna, B., & Wright, C. (2016). Adolescent brain development:
month follow-up of the iCURE study using cross-lagged path Implications to thejuvenile criminal justice system. In K.
analysis. Journal of Behaviors and Addiction, 8(4), 725–732. Heilbrun (Ed.), APA handbooks in psychology: APA handbook
https://doi.org/10.1556/2006.8.2019.74. of psychology and Juvenile Justice. American Psychological As-
Johnson, S. B., Blum, R. W., & Giedd, J. N. (2009, Sep). Adolescent sociation. https://doi.org/10.1037/14643-005.
maturity and the brain: The promise and pitfalls of neurosci- Madhav, K. C., Sherchand, S. P., & Sherchan, S. (2017, Dec). As-
ence research in adolescent health policy. Journal of Adolescent sociation between screen time and depression among US adults.
Health, 45(3), 216–221. https://doi.org/10.1016/j.jadohealth. Prague Medical Report, 8, 67–71. https://doi.org/10.1016/j.
2009.05.016. pmedr.2017.08.005.
Kaczkurkin, A. N., Raznahan, A., & Satterthwaite, T. D. (2019, Jan). Madigan, S., Browne, D., Racine, N., Mori, C., & Tough, S. (2019,
Sex differences in the developing brain: Insights from Mar 1). Association between screen time and children’s

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 91

performance on a developmental screening test. JAMA Pedi- Sanders, W., Parent, J., Forehand, R., Sullivan, A. D., & Jones, D. J.
atrics, 173(3), 244–250. https://doi.org/10.1001/jamapediatrics. (2016, May–Jun). Parental perceptions of technology and
2018.5056. technology-focused parenting: Associations with youth screen
Mathiesen, K. S., Sanson, A., Stoolmiller, M., & Karevold, E. (2009, time. Annals of Diagnostic Pathology, 44, 28–38. https://doi.org/
Feb). The nature and predictors of undercontrolled and inter- 10.1016/j.appdev.2016.02.005.
nalizing problem trajectories across early childhood. Journal of Sauce, B., Liebherr, M., Judd, N., & Klingberg, T. (2022, May 11).
Abnormal Child Psychology, 37(2), 209–222. https://doi.org/10. The impact of digital media on children’s intelligence while
1007/s10802-008-9268-y. controlling for genetic differences in cognition and socioeco-
Nabi, R. L., Perez Torres, D., & Prestin, A. (2017). Guilty pleasure nomic background. Scientific Reports, 12(1), 7720. https://doi.
no more: The relative importance of media use for coping with org/10.1038/s41598-022-11341-2.
stress. Journal of Media Psychology: Theories, Methods, and Saunders, J. B., Hao, W., Long, J., King, D. L., Mann, K., Fauth-
Applications, 29(3), 126–136. https://doi.org/10.1027/1864- Buhler, M., … Poznyak, V. (2017, Sep 1). Gaming disorder: Its
1105/a000223. delineation as an important condition for diagnosis, manage-
Nakagawa, S., & Schielzeth, H. (2013). A general and simple ment, and prevention. Journal of Behaviors and Addiction, 6(3),
method for obtaining R2 from generalized linear mixed-effects 271–279. https://doi.org/10.1556/2006.6.2017.039.
models. Methods in Ecology and Evolution, 4, 133–142. https:// Siciliano, C. A., Noamany, H., Chang, C. J., Brown, A. R., Chen, X.,
doi.org/doi:10.1111/j.2041-210x.2012.00261.x. Leible, D., … Tye, K. M. (2019, Nov 22). A cortical-brainstem
Neville, R. D., McArthur, B. A., Eirich, R., Lakes, K. D., & Madigan, S. circuit predicts and governs compulsive alcohol drinking. Sci-
(2021, Dec). Bidirectional associations between screen time ence, 366(6468), 1008–1012. https://doi.org/10.1126/science.
and children’s externalizing and internalizing behaviors. Clin- aay1186.
ical Psychology & Psychotherapy, 62(12), 1475–1484. https:// Somerville, L. H., Jones, R. M., & Casey, B. J. (2010, Feb). A time of
doi.org/10.1111/jcpp.13425. change: Behavioral and neural correlates of adolescent sensi-
Orben, A., Przybylski, A. K., Blakemore, S. J., & Kievit, R. A. (2022, tivity to appetitive and aversive environmental cues. Brain and
Mar 28). Windows of developmental sensitivity to social media. Cognition, 72(1), 124–133. https://doi.org/10.1016/j.bandc.
Nature Communications, 13(1), 1649. https://doi.org/10.1038/ 2009.07.003.
s41467-022-29296-3. Sowell, E. R., Thompson, P. M., Holmes, C. J., Batth, R., Jernigan, T. L.,
Paulus, M. P., Squeglia, L. M., Bagot, K., Jacobus, J., Kuplicki, R., & Toga, A. W. (1999, Jun). Localizing age-related changes in
Breslin, F. J., … Tapert, S. F. (2019, Jan 15). Screen media ac- brain structure between childhood and adolescence using sta-
tivity and brain structure in youth: Evidence for diverse tistical parametric mapping. Neuroimage, 9(6 Pt 1), 587–597.
structural correlation networks from the ABCD study. Neuro- https://doi.org/10.1006/nimg.1999.0436.
image, 185, 140–153. https://doi.org/10.1016/j.neuroimage. Spear, L. P. (2000, Jun). The adolescent brain and age-related
2018.10.040. behavioral manifestations. Neuroscience and Biobehavioral Re-
Pehlivanova, M., Wolf, D. H., Sotiras, A., Kaczkurkin, A. N., views, 24(4), 417–463. https://doi.org/10.1016/s0149-7634(00)
Moore, T. M., Ciric, R., … Satterthwaite, T. D. (2018). 00014-2.
Diminished cortical thickness is associated with impulsive Su, W., Han, X., Yu, H., Wu, Y., & Potenza, M. N. (2020). Do men
choice in adolescence. Journal of Neuroscience, 38, 2471–2481. become addicted to internet gaming and women to social
https://doi.org/10.1523/JNEUROSCI.2200-17.2018. media? A meta-analysis examining gender-related differences
Pinquart, M. (2017, May). Associations of parenting dimensions in specific internet addiction. Computers in Human Behavior,
and styles with externalizing problems of children and adoles- 113, 106480. https://doi.org/https://doi.org/10.1016/j.chb.2020.
cents: An updated meta-analysis. Developmental Psychology, 106480.
53(5), 873–932. https://doi.org/10.1037/dev0000295. Tierney, A. L., & Nelson, C. A., 3rd. (2009, Nov 1). Brain devel-
Potenza, M. N. (2018, Jun 1). Do gaming disorder and hazardous opment and the role of experience in the early years. Zero
gaming belong in the ICD-11? Considerations regarding the Three, 30(2), 9–13. https://www.ncbi.nlm.nih.gov/pubmed/
death of a hospitalized patient that was reported to have 23894221.
occurred while a care provider was gaming. Journal of Behav- Tisdall, M. D., Hess, A. T., Reuter, M., Meintjes, E. M., Fischl, B., &
iors and Addiction, 7(2), 206–207. https://doi.org/10.1556/2006. van der Kouwe, A. J. (2012, Aug). Volumetric navigators for
7.2018.42. prospective motion correction and selective reacquisition in
Raffin, J., Rolland, Y., Fischer, C., Mangin, J. F., Gabelle, A., neuroanatomical MRI. Magnetic Resonance in Medicine, 68(2),
Vellas, B., … Group, M. D. (2021, Feb 2). Cross-sectional 389–399. https://doi.org/10.1002/mrm.23228.
associations between cortical thickness and physical activity in Twenge, J. M., & Campbell, W. K. (2018, Dec). Associations be-
older adults with spontaneous memory complaints: The MAPT tween screen time and lower psychological well-being among
Study. Journal of Sport and Health Science. https://doi.org/10. children and adolescents: Evidence from a population-based
1016/j.jshs.2021.01.011. study. Prague Medical Report, 12, 271–283. https://doi.org/10.
Robinson, T. N., Banda, J. A., Hale, L., Lu, A. S., Fleming-Milici, F., 1016/j.pmedr.2018.10.003.
Calvert, S. L., & Wartella, E. (2017, Nov). Screen media expo- Twenge, J. M., & Farley, E. (2021, Feb). Not all screen time is
sure and obesity in children and adolescents. Pediatrics, created equal: Associations with mental health vary by activity
140(Suppl 2), S97–S101. https://doi.org/10.1542/peds.2016- and gender. Social Psychiatry and Psychiatric Epidemiology,
1758K. 56(2), 207–217. https://doi.org/10.1007/s00127-020-01906-9.

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

92 Journal of Behavioral Addictions 12 (2023) 1, 80–93

Volkow, N. D., Koob, G. F., Croyle, R. T., Bianchi, D. W., Gordon, toward ICD-11. The American Journal of Psychiatry, 174(5),
J. A., Koroshetz, W. J., … Weiss, S. R. B. (2018, Aug). The 486–487. https://doi.org/10.1176/appi.ajp.2017.16121346.
conception of the ABCD study: From substance use to a broad Zhang, Z. (2016, Jan). Multiple imputation with multivariate
NIH collaboration [Review]. Developmental Cognitive Neuro- imputation by chained equation (MICE) package. Annals of
science, 32, 4–7. https://doi.org/10.1016/j.dcn.2017.10.002. Translational Medicine, 4(2), 30. https://doi.org/10.3978/j.issn.
Von Der Heide, R., Vyas, G., & Olson, I. R. (2014, Dec). The social 2305-5839.2015.12.63.
network-network: Size is predicted by brain structure and Zhao, Y., Constable, R. T., Hien, D., Chung, T., & Potenza, M. N.
function in the amygdala and paralimbic regions. Social (2021). Brain anatomical covariation patterns linked to binge
Cognitive and Affective Neuroscience Electronic Resource, 9(12), drinking and age at first full drink. NeuroImage: Clinical, 29,
1962–1972. https://doi.org/10.1093/scan/nsu009. 102529. https://doi.org/10.1016/j.nicl.2020.102529.
White, N., Roddey, C., Shankaranarayanan, A., Han, E., Rettmann, D., Zhao, Y., Klein, A., Castellanos, F. X., & Milham, M. P. (2019, Aug
Santos, J., … Dale, A. (2010, Jan). PROMO: Real-time pro- 30). Brain age prediction: Cortical and subcortical shape
spective motion correction in MRI using image-based tracking. covariation in the developing human brain. Neuroimage, 202,
Magnetic Resonance in Medicine, 63(1), 91–105. https://doi.org/ 116149. https://doi.org/10.1016/j.neuroimage.2019.116149.
10.1002/mrm.22176. Zhao, Y., Paulus, M., Bagot, K. S., Constable, R. T., Yaggi, H. K.,
Wolfers, L. N., & Schneider, F. M. (2021). Using media for coping: Redeker, N. S., & Potenza, M. N. (2022, Jul 13). Brain struc-
A scoping review. Communication Research, 48(8), 1210–1234. tural covariation linked to screen media activity and exter-
https://doi.org/10.1177/0093650220939778. nalizing behaviors in children. Journal of Behaviors and
Yao, Y. W., Potenza, M. N., & Zhang, J. T. (2017, May 1). Internet Addiction, 11(2), 417–426. https://doi.org/10.1556/2006.2022.
gaming disorder within the DSM-5 framework and with an eye 00044.

Appendix

Table A1. Baseline demographic differences between the data currently analyzed and these to be released in the futurep
Variable Data Available (n 5 5,166) Data Not Available (n 5 6,691)
Age (Months; mean ± SD) 119.31 ± 7.38 118.73 ± 7.57
Total screen time (Baseline hours; mean ± SD) 25.35 ± 20.59 27.47 ± 22.15
Internalizing score (Baseline; mean ± SD) 5.09 ± 5.48 5.00 ± 5.56
Externalizing score (Baseline; mean ± SD) 4.44 ± 5.82 4.45 ± 5.89
Gender/sex (Female; n, %) 2,385 (46.17%) 3,289 (49.16%)
Race (n, %)
White 3,596 (69.61%) 3,914 (58.50%)
Black 623 (12.06%) 1,243 (18.58%)
Asian 112 (2.17%) 163 (2.44%)
Other/Mixed 835 (16.16%) 1,371 (20.49%)
Education Level (n, %)
Up to HS graduation/GED 580 (11.24%) 1,138 (17.03%)
Some College 1,331 (25.79%) 1,742 (26.06%)
Bachelor’s Degree 1,417 (27.46%) 1,595 (23.86%)
Post-Graduate Degree 1,832 (35.50%) 2,209 (33.05%)
Family Income (n, %)
[<$50,000] 1,286 (26.73%) 1,944 (32.11%)
[≥$50,000 & <$100,000] 1,463 (30.41%) 1,612 (26.63%)
[≥$100,000] 2,062 (42.86%) 2,498 (41.26%)
p
Samples being analyzed in this study are statistically different from these whose data yet to be released on all key baseline demographic
information listed in the table.

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC

Journal of Behavioral Addictions 12 (2023) 1, 80–93 93

Table A2. Bivariate associations among SMA, brain co-development pattern, and psychopathology measures
Internalizing
Brain Problem at
Baseline Total Co-development Internalizing Externalizing Two Year
Screen Time Pattern Problem at Baseline Problem at Baseline Follow Up
Baseline Total Screen Time
Brain Co-development Pattern 0.04p
Internalizing Problems at Baseline 0.07pppp <0.01
Externalizing Problems at Baseline 0.16pppp 0.01 0.59pppp
Internalizing Problems at 0.04pp 0.04pp 0.65pppp 0.39pppp
Two-Year Follow-Up
Externalizing Problems at 0.14pppp 0.01 0.43pppp 0.69pppp 0.57pppp
Two-Year Follow-Up
1) FDR-adjusted P-values were reported here.
2) pppp P < 0.0001, pp P < 0.01, p P < 0.05

Table A3. Mean percentages of changes in brain features between baseline and two-year follow-up by total screen time usage. Participants in
Low, Moderate, and High use groups on average had up to one hour per day, between one hour and less than seven hours per day, and at
least seven hours per day screen time, respectively
Feature Loading Low (n 5 632) Moderate (n 5 3,914) High (n 5 620)
GMV_brain.stem 0.24 4.4 4.0 3.9
GMV_superiorfrontal.rh 0.19 1.0 1.2 1.5
GMV_superiorfrontal.lh 0.18 0.9 1.2 1.7
GMV_rostralmiddlefrontal.lh 0.17 1.6 1.9 2.3
Thick_temporalpole.rh 0.17 0.2 0.1 0.4
Thick_temporalpole.lh 0.17 0.3 0.4 0.7
GMV_rostralmiddlefrontal.rh 0.15 2.0 2.2 2.6
GMV_superiorparietal.rh 0.15 3.0 3.0 3.5
Thick_entorhinal.lh 0.14 0.6 0.2 0.4
GMV_inferiorparietal.rh 0.14 2.5 2.6 2.8
GMV_putamen.lh 0.13 0.9 0.6 0.3
GMV_inferiortemporal.lh 0.13 1.0 1.3 1.3
Thick_entorhinal.rh 0.13 1.0 0.9 0.1
GMV_superiorparietal.lh 0.12 2.6 2.9 3.4
GMV_inferiortemporal.rh 0.12 1.0 1.3 1.2
GMV_middletemporal.rh 0.12 1.1 1.1 1.4
GMV_middletemporal.lh 0.11 0.6 0.9 1.2
GMV_inferiorparietal.lh 0.11 2.6 2.7 3.0
Thick_frontalpole.rh 0.11 2.1 1.6 2.0
GMV_precentral.lh 0.11 0.2 0.3 0.4
Thick_frontalpole.lh 0.11 2.0 1.7 1.9
GMV_superiortemporal.lh 0.11 1.1 1.4 1.8
GMV_thalamus.proper.lh 0.07 3.1 2.9 3.0
GMV_thalamus.proper.rh 0.05 1.9 1.7 1.5
p
Note: Both left and right Thalamus Proper GMV were not ranked as top features in the co-development pattern and are included here given
their hypothesized involvement. Other listed measures include the brain regions contributing most substantially to the identified brain co-
development pattern.

Open Access. This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (https://
creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted use, distribution, and reproduction in any medium for non-commercial purposes, provided the
original author and source are credited, a link to the CC License is provided, and changes – if any – are indicated.

Unauthenticated | Downloaded 09/10/24 01:09 AM UTC