Content uploaded by Ismael Muela
Author content
All content in this area was uploaded by Ismael Muela on Feb 04, 2025
Content may be subject to copyright.
Vol.:(0123456789)
Current Addiction Reports (2025) 12:9
https://doi.org/10.1007/s40429-025-00614-1
A Critical Review andMeta‑Analysis ofInterventions toReduce
Compulsivity inBehavioral Addictions andRelated Conditions
JoseLópez‑Guerrero1· MiguelA.Vadillo2· FranciscoJ.Rivero1· IsmaelMuela1· JuanF.Navas3· JoséC.Perales1
Accepted: 2 December 2024
© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025
Abstract
Purpose of Review This preregistered review aims to identify and assess the effectiveness of intervention tools designed to
reverse, counteract, or reduce compulsivity in behavioral addictions and related conditions. The review covers reconditioning
techniques, non-invasive brain stimulation, and mindfulness/non-judgmental observation interventions, and meta-analyzes
their effects using random-effect models. Moderation by tool type, behavioral domain, and methodological quality was
assessed, and several tests of publication and reporting bias were conducted.
Recent Findings Compulsivity is a core characteristic of addictive behaviors, as individuals with addiction feel increasingly
compelled to act in ways that go against their own best interests. Viewing compulsivity as a result of conditioning processes
causing loss of control over behavior has led to the development of tools to reverse or reduce the expression of those pro-
cesses, but their effectiveness in behavioral addictions and related conditions remains insufficiently researched.
Summary Although compulsivity-oriented interventions hold potential, the current evidence base is limited by small-study
effects and insufficient methodological rigor. A shift toward more robust, theory-driven studies is needed to effectively isolate
the techniques’ impact and improve therapeutic outcomes.
Keywords Behavioral addiction· Compulsivity· Bias reduction· Exposure therapy· Non-invasive brain stimulation·
Mindfulness
* Jose López-Guerrero
joselogue@ugr.es
1 Department ofExperimental Psychology. Mind, Brain,
andBehavior Research Center (CIMCYC), University
ofGranada, Granada, Spain
2 Department ofBasic Psychology, Autonomous University
ofMadrid, Madrid, Spain
3 Department ofClinical Psychology, Complutense University
ofMadrid (UCM), Madrid, Spain
Introduction
In prevalent psychobiological models, addictive behaviors
are considered as such because they have become compul-
sive. Still, and despite its importance in defining addiction,
compulsivity remains an ill-defined construct.
In general, it is said that someone’s behavior has become
compulsive when that individual feels “forced” to act against
their best interest, so there is a personal conflict between
acts and goals [1]. In some addiction models, this discon-
nection is explained as resulting from a transition between
goal-driven (instrumental) and stimulus-driven (habitual)
behavior [2]. However, there are behaviors customarily
defined as compulsive that have been convincingly shown
to remain voluntary and relatively flexible, and habit learn-
ing does not seem to be a sine-qua-non condition for com-
pulsivity acquisition [3, 4]. In line with this evidence, other
models conceptualize compulsive behaviors as instrumental,
but resulting from transitioning from being motivated by
some kind of identifiable extrinsic reinforcer (e.g. pleasure-
seeking, having fun, socializing, or reducing anxiety) to
being motivated by the need to satisfy an irresistible urge,
or to relief a craving or withdrawal state [5], so that control
difficulties do not directly arise from behavior automatic-
ity but from an abnormally intense motivational drive. In
the first family of theories, the process to be explained is
how pathological habits are formed, whereas in the second
it is the one by which urge, craving, or withdrawal states are
progressively strengthened by chronic exposure to an addic-
tive agent. The term “motivational habit” has been coined
to refer to behaviors that are instrumentally motivated by
craving relief (in a Response-Outcome fashion), but in which
contextual cues have acquired the power to trigger such a
Current Addiction Reports (2025) 12:9 9 Page 2 of 14
preexisting craving response (in a Stimulus-Response, auto-
matic manner) [6].
Here, we will partially skip this etiological controversy,
and will stick to a phenomenological and behavioral defini-
tion of compulsivity. In a detailed qualitative analysis of
self-report compulsivity measures in behavioral addiction
literature, Muela etal. [7, 8] identified six different opera-
tionalizations of compulsivity expressed in item wordings.
These were (a) automatic or habitual behavior occurring
in the absence of conscious instrumental goals, (b) over-
whelming urge or desire that impels the individual to initiate
the activity and jeopardizes control attempts, (c) inability to
stop or interrupt the activity once initiated, resulting in an
episode substantially longer or more intense than intended
(bingeing), (d) behavior insensitive to negative consequences
despite conscious awareness of them, (e) attentional capture
and cognitive hijacking, and (f) inflexible rules, stereotyped
behaviors, and rituals related to task completion or execu-
tion [7]. Muela etal.’s analysis regards sets of behavioral
features as perceived and reported by the individual that are
sufficient, but neither mutually exclusive nor necessary, for a
behavior to be labeled as compulsive, and that remains theo-
retically neutral. That is, the six different manifestations of
compulsivity may result from a common underlying etiolog-
ical construct or from different psychobiological processes.
Still, these and other behavioral and phenomenological
operationalizations of compulsive behavior share some fea-
tures (apart from the previously mentioned feeling of being
“forced” to do something against one’s will, and the ensuing
perceived conflict between acts and personal goals). Namely,
resistance to control attempts, being transitory (occurring
in discrete episodes) but repetitive, and being triggered by
contextual or internal events [7].
This domain-specific definition must be differentiated
from cross-domain compulsivity as a trait. Some individu-
als seem to be constitutionally prone to develop compul-
sive behaviors, so compulsivity as a trait can be regarded
as a transdiagnostic vulnerability factor for the acquisition,
chronification, and maintenance of addictive disorders [9].
Our interest here is however focused on compulsivity as an
acquired feature of some behaviors, by virtue of a process of
exposure and interaction with an addictive agent, regardless
of the ultimate nature of the learning mechanism responsible
for that acquisition process.
In this review, we have isolated and assessed the effect
of treatment techniques designed to interfere with compul-
sivity acquisition, to reverse it in some way, or to directly
counteract its expression. In line with the features outlined
above (resistance to control, transitoriness, and context
dependency), our review explicitly focuses on techniques
intending to reduce the reactive element of compulsivity,
and leaves out other techniques aimed at strengthening pro-
active control via belief modification and psychoeducation
[10], contingency management [11], goal management [12],
or training in explicit emotion regulation strategies [13]. It
encompasses, however, (a) exposure therapies based on
extinction or counterconditioning of emotional or physi-
ological responses, (b) cognitive attentional or approach
bias modification (CBM), and (c) consolidation-inspired
techniques. Along with these, our review will also include
(d) mindfulness/nonjudgmental observation, (e) biofeed-
back, and (f) non-invasive brain stimulation (NIBS). In
other words, our review focuses on testing whether directly
addressing compulsivity contributes to improving therapeu-
tic outcomes in behavioral addictions and related conditions.
Most intervention studies are designed to reduce addic-
tion symptoms, and so their outcomes are normally meas-
ures of severity. However, compulsivity and severity are not
interchangeable concepts, and neither changes in compulsiv-
ity necessarily result in general improvements in symptoma-
tology, nor do severity improvements necessarily require a
reduction in compulsivity. Consequently, our review will
encompass severity measures and other compulsivity-sensi-
tive outcomes, such as cue-reactivity and craving measures.
Different measures from the same study will be separately
considered, while modeling their statistical dependency.
Importantly, the scope of assessed techniques was con-
ceptually guided, and agreed a priori between the authors.
Tentatively, Fig.1 displays the mapping of the links between
the reviewed techniques, the underlying theoretical mecha-
nism accounting for their effects, and the operationalizations
that could manifest a reduction of compulsivity as a conse-
quence of such effects.
Exposure therapies are considered the standard approach
to treating abnormal conditioned emotional or psycho-
physiological responses, and have been extensively stud-
ied as craving extinction or counterconditioning tools in
substance use disorders [14], an approach that has been
recently revived by the emergence of virtual reality tech-
nology [15]. CBM techniques share some features with
exposure therapy, and consist of computer tasks in which
trainees are exposed to drug-related stimuli (e.g. pictures of
alcoholic drinks) and trained to disengage their attention,
or to withdraw from them (for instance, using a joystick to
zoom the stimulus out). This training has been reported to
be effective at counteracting automatic attentional biases
and approach tendencies in people suffering from substance
use disorders and behavioral addictions [16]. Consolida-
tion-inspired therapies are designed to supplement expo-
sure techniques, by taking advantage of preclinical findings
showing that associative memories return to a labile state
when reactivated [17]. Mindfulness was included in this
review in cases in which it was explicitly aimed at reduc-
ing the reactive emotional components of compulsivity.
Moreover, although the mechanisms of mindfulness effi-
cacy are a matter of discussion, one of its proposed active
Current Addiction Reports (2025) 12:9 Page 3 of 14 9
ingredients is controlled exposure to emotion-triggering
thoughts or imagined stimuli that incidentally enter the
mind during sessions in the absence of behavioral response
[18, 19]. And finally, biofeedback and NIBS have also been
specifically used to reduce cue reactivity. Brain stimula-
tion has preferentially targeted the prefrontal cortex [20],
whereas biofeedback has been aimed at reducing central
or peripheral psychophysiological responses to craving-
triggering events [21]. Again, we will focus on interven-
tions explicitly designed to reduce compulsive reactions or
behaviors to such events, and eventually symptom severity,
but will not take into consideration reports of interventions
used to facilitate training of more general control strategies
(e.g. biofeedback-enhanced training for emotion regulation
or stress reduction [22]).
Importantly, these techniques are typically not used in
isolation, but as parts of more comprehensive packages
(e.g. Cognitive Behavioral Therapy). In most interven-
tion studies, a comparison is made between a treatment
group receiving the whole package and a control group
(e.g. waitlist, treatment as usual, or active control), so that
the effect of the technique of interest is confounded with
the other components, and the added value of specifically
targeting compulsivity remains hidden. Hence, here we
will focus on intervention studies in which the compul-
sivity-oriented technique is used in isolation, or in a com-
pound pitched against a control group that is identical to
the experimental one except for the component of interest
(e.g. exposure + group therapy vs. group therapy alone).
We searched for all intervention studies in the field of
behavioral addictions and related conditions. For clarity,
these related conditions are those that have been proposed
as candidate behavioral addictions in the scientific litera-
ture, but have not yet been recognized by the main psychi-
atric nosologies. So, our search included gaming and gam-
bling disorders (and their different denomination variants)
along with the varied labels used for problematic internet,
social media and smartphone use, compulsive buying, and
compulsive sexual behavior or pornography use. Generic
behavioral addiction-related search terms were also added
to capture putative addictions not included in this list.
The aim of our review was two-fold. First, we selected the
studies in which compulsivity-oriented therapeutic ingre-
dients are isolatable, summarized their basic features, and
extracted and reported the size of their effects on symptom
severity, craving, and other behavioral, psychophysiologi-
cal, or self-report measures of cue-reactivity. And second,
we meta-analyzed those effects using a random-effects
model (that assumes an effect size distribution rather than
a single populational effect size). Given the multiplicity of
techniques under assessment, we also estimated the poten-
tial moderation effects of intervention type, as well as of
the risk of bias, the behavioral domain, and the type of
outcome measure.
Fig. 1 Tentative mapping of the links between intervention techniques and their behavioral manifestation in compulsivity reduction via their
potential underlying mechanisms
Current Addiction Reports (2025) 12:9 9 Page 4 of 14
Previewing the results, the techniques assessed jointly
showed therapeutic potential. However, this result is
overshadowed by concerns about study quality and reli-
ability, and the lack of proper control conditions or their
inadequate design. We conclude with some recommenda-
tions for addressing the methodological and conceptual
problems identified, and for improving causal inferences
and model formulation based on intervention studies’
data.
Method
This systematic review and meta-analysis was pre-registered
on PROSPERO and can be accessed via: https:// www. crd.
york. ac. uk/ prosp ero/ displ ay_ record. php? Recor dID= 490637.
The review followed the PRISMA 2020 guidelines [23], with
the flowchart (Fig.2) outlining the process for study identifica-
tion, screening, and selection. A comprehensive search was con-
ducted across four major electronic databases: PubMed, Scopus,
Web of Science, and ProQuest. The search strategy, detailed in
the supplementary file 1 provided via the OSF link included in
the Data Availability statement, was formulated to identify inter-
vention studies for behavioral addictions and related conditions
using therapeutic techniques aimed at reducing compulsivity,
craving, or automatic behavior, as justified earlier.
Studies eligible for inclusion were peer-reviewed primary
research articles describing both controlled and non-con-
trolled designs with clinical or at-risk population samples.
Primary outcomes included validated measures of addiction
severity or severity-related constructs (e.g., harm, quality
of life), craving, compulsivity, and cue-reactivity, assessed
through self-report instruments, behavioral tasks, or neuro-
physiological measures. Neurophysiological measures were
selected only if concerned specific indices of interest (e.g. an
ERP component, or activation observed in a predetermined
ROI). Effects for differences in brain activation detected
using whole-brain analyses were considered non-eligible,
as effect detection in these analyses is by definition based
on threshold values for difference statistics and cluster size,
and including them in the meta-analysis would imply cherry-
picking large effects.
Behavioral addictions of interest were those recognized
in major psychiatric classifications, namely Gambling
Disorder (GD) and (Internet) Gaming Disorder (IGD),
along with other behavioral patterns (related conditions)
proposed as addictive in the literature but not included
in these nosologies. There were no restrictions regard-
ing publication date, and the search encompassed studies
published in English, Spanish, French, or Portuguese. To
detect relevant studies not captured by our initial search, a
backward and forward citation analysis was also conducted
Fig. 2 PRISMA flowchart for article selection
Current Addiction Reports (2025) 12:9 Page 5 of 14 9
from systematic reviews retrieved during the database
search (e.g [24–26]).
Two team members (JLG and FR) independently con-
ducted the search on Jan. 22nd, 2024, followed by the
screening and selection of studies in the subsequent weeks.
Discrepancies between the two reviewers were discussed,
and when consensus could not be reached, the PI of the
project (JCP) was consulted to referee on the disagreement
until consensus was reached. The Excel file documenting
the selected and excluded studies, along with the reasons
for initial exclusions, is available in the supplementary
file 2 provided via the Open Science Framework (OSF)
link included in the Data Availability statement. To avoid
leaving out relevant studies that might have come out dur-
ing the submission, review, and revision process, and in
response to an anonymous reviewer’s suggestion, a second
restricted search with the same inclusion and exclusion
criteria was conducted for studies published between the
initial search and November 21st, 2024.
Data Extraction andRisk ofBias Assessment
Data were extracted independently by JLG and FR, using
a standardized extraction form (see the supplementary file
3 provided via the OSF link included in the Data Avail-
ability statement). For each study, we extracted data on
study characteristics (author, year, DOI, country), interven-
tion modality, type of control group (if applicable), sample
characteristics, and outcome measures. When reported in
the manuscripts or computable from the available infor-
mation, means and standard deviations were collected for
all outcome measures in all relevant conditions (pre-post
intervention for the experimental and the control condi-
tion). If these data were unavailable, t-values or F-values
for the main effect of condition (experimental vs. control)
were coded. If these were also missing, the correspond-
ing authors were contacted twice to request the necessary
information. If no response was obtained after two con-
tact attempts and the data remained unavailable, the study
or the outcome was excluded from further analyses. Out-
come measures of interest were immediate post-interven-
tion ones, or, if this was not available, the first follow-up
assessment. For the two additional studies identified in the
review process, effect sizes were computed from the sta-
tistics reported in the corresponding manuscripts, inferred
from reported mean and variability measures, or by apply-
ing an automated data extraction tool to descriptive graphs.
The risk of bias was assessed using the Cochrane Risk
of Bias tool for randomized controlled trials (ROB-2 [27]),
as pre-registered. This tool was chosen for its suitability
in evaluating intervention studies, even those without con-
trol groups, ensuring a consistent assessment across all
studies. Discrepancies between reviewers were resolved by
consensus, and another author (JFN) was consulted when
necessary.
Strategy forData Synthesis
In a first meta-analysis, for all valid post-treatment out-
comes, we calculated an unbiased estimate of the standard-
ized mean difference (Hedges’ gs) between experimental and
control groups (using Eq.4.18 and 4.19 from Borenstein
etal. [28]). In designs without a control group but with a
within-participant control condition (crossover designs), we
computed a comparable effect size measure (Hedges’ gIG)
with Eqs.13 and 23 from Morris and DeSchon [29], using
the standard deviation of the control condition to standardize
mean differences. In these cases, we computed the variance
of the effect sizes assuming a correlation of 0.70 between the
control and experimental conditions [30], unless this corre-
lation was reported or could be computed with reported data.
The R code used for these and the following calculations is
available in the supplementary file 4 provided via the OSF
link included in the Data Availability statement.
A second meta-analysis included studies with a control
group, and pre-treatment and post-treatment measures for all
groups, hence allowing the computation of a standardized
mean change difference between groups, i.e., Hedges’ gppc2
(see Eqs.8–10 from Morris [31]). Again, for the computa-
tion of the variances, a correlation of 0.70 between pre- and
post-test measures was imputed, unless the studies reported
sufficient information to estimate the actual correlation.
The pre-registered protocol established that studies
exploring pre-post effects in conditions without a valid con-
trol group would also be included in the first meta-analysis,
unless their number was sufficiently large to allow for a
separate analysis. As this condition was met, we conducted
a third meta-analysis including only pre-post comparisons
without a control group. Although in the protocol we stated
that we would compute effect sizes using Hedges’ gz, for the
sake of comparability of these effects with the Hedges’ gs
and gppc2 used in the previous analyses, we decided to use
gIG scores (Eqs.13 and 23 from Morris and DeSchon [29]).
In all cases, effect sizes were meta-analyzed using ran-
dom-effects models. Given that some studies contributed
with more than one valid effect size estimate, a standard
univariate meta-analysis, ignoring the multi-level nature of
the data, could potentially bias the results, by giving undue
weight to studies with multiple valid effect sizes. To avoid
this bias, a random intercept at the sample identity level
was added to account for dependencies between effect sizes
when necessary (e.g., different outcome measures for the
same study sample). This method has been shown to deal
satisfactorily with multi-level data like the one included in
our meta-analysis (see van den Noorgate etal. [32]).
Current Addiction Reports (2025) 12:9 9 Page 6 of 14
The small-study effect, potentially reflecting a publication
bias, was assessed using several methods, since no single
method has been identified to consistently outperform the
others. The distribution of effect sizes and standard errors
(SEs) was presented in funnel plots, and asymmetry was
tested by including SE as a continuous moderator to the pre-
vious random-effects model. Bias-corrected estimates were
calculated using PET, PEESE, PET-PEESE, the 3-parameter
selection model (3PSM), and trim-and-fill. To account for
dependencies among effect sizes, PET, PEESE, and PET-
PEESE were fitted adding a random-intercept at the sam-
ple level. We are not aware of any implementation of the
3PSM and trim-and-fill for multilevel data. Therefore, we
fitted these models twice: one entering all the individual
effect sizes as statistically independent effects (i.e., ignoring
the multilevel structure of the data), and then a second one
aggregating effect sizes from each study.
We conducted a series of moderator analyses to explore
potential sources of heterogeneity in the effect size pool.
The moderators tested included (1) the type of intervention,
with interventions grouped into reconditioning techniques
(e.g., Exposure Therapy and Cognitive Bias Modification),
mindfulness-based interventions, and brain stimulation tech-
niques (no studies using biofeedback or reconsolidation were
identified in the search). (2) Addiction type was classified as
either Gambling Disorder or other putative behavioral addic-
tions. (3) Outcome measures were categorized into severity
instruments or cue-reactivity measures. Additionally, (4) the
risk of bias was assessed using the ROB-2 tool, and modera-
tor analyses included each of the individual ROB-2 compo-
nents as well as the overall score to evaluate the impact of
study quality on the pooled effect sizes. Deviations from the
preregistered list of moderators and the final categories were
due to the considerable heterogeneity observed among the
included studies. The moderators as described in the pre-
registration, the modified moderators, and justifications for
changes are detailed in the supplementary file 5 provided
via the OSF link included in the Data Availability statement.
Results
Study Selection andCharacteristics
The initial search of four databases —PubMed, Scopus, Web
of Science, and ProQuest— yielded a total of 209 records.
After removing 133 duplicates, 76 records were screened
based on titles and abstracts. Of these, 53 were excluded
for not meeting the eligibility criteria, primarily because
they were review articles. Full-text assessments were con-
ducted for 23 studies, and 8 were excluded for reasons such
as mixed addictions or lack of valid outcome data. A total of
15 studies were ultimately included from database searches
[33–47], and 13 additional studies were identified through
backward and forward citation analysis [48–60]. The
detailed study selection process is outlined in the PRISMA
flow diagram (Fig.1) and the supplementary file 2 provided
via the OSF link included in the Data Availability state-
ment. As noted earlier, as part of the review and revision
process, an updated search was conducted, which identified
two additional studies that met the inclusion criteria and
were incorporated into the meta-analysis [61, 62].
We grouped the studies according to the specific meta-
analysis criteria outlined earlier. The first category included
studies that had an adequate control group or a control condi-
tion that allowed post-treatment comparisons [33, 35, 37, 39,
41, 44, 48–51, 53, 55, 56, 62]. The second category included
a subset of the first category with the studies having a control
group and both pre- and post-treatment measures [33, 37, 39,
44, 48–50, 53, 55, 56, 62]. Lastly, we grouped studies without
a control condition together with those that used an active
control group (not allowing the isolation of the component of
interest) in the third category [34, 36, 38, 40, 42, 43, 45–47,
52, 54, 57–61]. Importantly, none of the studies from the first
category was included in the third category.
Risk ofBias Assessment
The quality assessment revealed that all studies were affected
by quality-related concerns. The primary issues most fre-
quently identified were the absence of pre-registration and
the lack of control groups. Additionally, in some cases, only
partial results were reported, raising concerns about the pos-
sibility of selective outcome reporting. These factors contrib-
uted to an overall high risk of bias in many of the included
studies. The detailed results of the risk of bias assessment are
presented in the supplementary file 3 provided via the OSF
link included in the Data Availability statement.
Although not directly related to quality assessment, it is
important to assess the presence of outliers in the sample of
effects sizes under consideration. Effect sizes largely beyond
what is normally expected in the field are sometimes regarded
with some suspicion, although no flagrant errors are detected
in the methods or results sections of the corresponding manu-
scripts. In our case, the automated outlier detection method
implemented in the metafor package [63] identified 1 out-
lier in meta-analysis 1, no outliers in meta-analysis 2, and
1 outlier in meta-analysis 3. Given that outlier removal was
not included in our pre-registration, and we have no strong
reasons beyond the results of this automated method for dis-
regarding these studies, the results of meta-analyses 1 and 3
without the outliers are reported in the supplementary file 6
provided via the OSF link included in the Data Availability
statement and briefly considered in the Discussionsection.
Current Addiction Reports (2025) 12:9 Page 7 of 14 9
Meta‑Analysis 1: Mean Post‑Treatment Difference
Between Experimental andControl Conditions
The first meta-analysis included 9 [33, 35, 37, 39, 50, 51,
53, 56, 62] studies with 26 meta-analyzable measures, after
excluding 5 studies due to shared samples, the inability to
obtain valid data despite repeated attempts to contact the
authors, or excessive data attrition rendering the results
non-interpretable. The R code and a list of all studies
included in this meta-analysis are available in the supple-
mentary file 4 provided via the OSF link included in the Data
Availability statement. The results yielded a pooled effect
size of 0.72 [0.32, 1.12] (Fig.3). Heterogeneity was large,
I2 = 87.80% and significant, Q(25) = 164.01, p < .001. We did
not find significant differences between studies with separate
control groups and those with crossover within-participant
Fig. 3 Forest plot for Meta-
analysis 1
Fig. 4 Funnel plots for Meta-analyses 1, 2, and 3
Current Addiction Reports (2025) 12:9 9 Page 8 of 14
designs, Q(1) < 1. While the funnel plot (Fig.4) was signifi-
cantly asymmetric, p < .001, not all tests of publication bias
returned significant and consistent results. PET and PEESE
returned negative bias-corrected effect size estimates, gs =
−1.89 [−2.70, −1.08] and − 0.79 [−1.32, −0.27], respec-
tively. Trim-and-fill and the 3PSM did not detect significant
evidence of bias either when fitted to all individual effect
sizes or when fitted to aggregate effect sizes. None of the
tested moderators reached statistical significance.
Meta‑Analysis 2: Mean Change Difference Between
Experimental andControl Groups
The second meta-analysis included 7 studies (which are a
subset of the studies included in meta-analysis 1 [33, 37, 39,
50, 53, 56, 62]) with a total of 21 meta-analyzable measures,
after excluding 4 studies for the same reasons as in meta-anal-
ysis 1. The pooled effect size was 0.75 [0.20, 1.30] (Fig.5),
reflecting an improvement in addiction-related outcomes
post-intervention compared to control conditions. Heteroge-
neity was large and significant, I2 = 91.40%, Q(17) = 177.29,
p < .001. The funnel plot (Fig.4) was significantly asym-
metric, p < .001. PET and PET-PEESE returned a negative
bias-corrected estimate, −0.88 [−1.72, −0.05] and PEESE a
non-significant one, 0.18 [−0.34, 0.70]. Trim-and-fill and the
3PSM returned evidence of bias when fitted to all individual
effect sizes. The bias-corrected estimates remained positive
and significant with trim-and-fill, g = 0.59 [0.24, 0.94], but
not with the 3PSM, g = 0.40 [−0.34, 1.14]. Among all tested
moderators, only the type of instrument approached signifi-
cance, Q(1) = 3.74, p = .053, with instruments measuring
severity showing larger average effects, g = 0.83 [0.23, 1.44],
than instruments measuring cue reactivity, g = 0.56 [−0.05,
1.18]. Full details on the included studies and the R code used
for the analyses are provided in Supplementary Material 4.
Meta‑Analysis 3: Pre‑Post Effects inStudies Without
aControl Group oraControl Group Not Allowing
Isolation oftheIngredient ofinterest
Finally, the third meta-analysis included 15 [34, 36, 38, 40,
42, 43, 45–47, 54, 57–61] studies contributing with 16 inde-
pendent samples and a total of 34 meta-analyzable meas-
ures. One study initially included was excluded at a later
stage due to uncertainty about the intervention’s alignment
with the predefined inclusion criteria. This meta-analysis
was conducted to extract all the possible information from
the relatively numerous identified studies that could not be
included in the previous two analyses, but also to compare
studies with and without a carefully designed control group.
Fig. 5 Forest plot for Meta-
analysis 2
Current Addiction Reports (2025) 12:9 Page 9 of 14 9
The analysis revealed a pooled effect size of 0.92 [0.65,
1.19] (Fig.6), which is notably larger than the moderate
effect sizes observed in the previous two meta-analyses. The
funnel plot (Fig.3) was significantly asymmetric, p < .001,
with PET and PET-PEESE returning a non-significant bias-
corrected estimate of 0.04 [−0.44, 0.53], but PEESE return-
ing a larger and significant bias-corrected estimate of 0.78
[0.51, 1.05]. Trim-and-fill did not detect evidence of bias
either with all or with aggregated effect sizes. The 3PSM
failed to converge when fitted to all individual effect sizes,
but returned marginally significant evidence of bias, p = .078,
and a non-significant corrected estimate, 0.73 [−0.08, 1.54],
when fitted to aggregated effect sizes. Among all tested mod-
erators, only the “Deviations from intended interventions”
item from the risk-of-bias checklist approached significance,
Q(2) = 4.99, p = .082, with studies at high risk, g = 1.10 [0.48,
1.72], or some risk, g = 1.85 [0.91, 2.79], yielding larger
effect sizes than studies at low risk, g = 0.78 [0.49, 1.08].
Discussion
The present review focuses on intervention techniques
designed to reduce the level of compulsivity in poten-
tially addictive behaviors not involving substance use, and
identifies the studies in which those techniques are either
implemented in isolation (with or without a control group),
or as part of a package that is pitched against a control con-
dition, identical except for the absence of the ingredient of
interest. The bibliographic search yielded a heterogeneous
set of intervention studies encompassing reconditioning,
mindfulness-based, and non-invasive brain stimulation tech-
niques aimed at reducing cue dependency and reactivity,
craving, and automatic attentional and approach biases for
a variety of behavioral addictions and related conditions,
including Gambling Disorder, Gaming Disorder, Problem-
atic Internet Use and its derivatives (such as Problematic
Smartphone Use), and Compulsive Sexual Behavior Dis-
order. No studies using biofeedback or reconsolidation-
inspired techniques in these domains met our criteria to be
included in the review.
Three meta-analyses were carried out. The first of them
meta-analyzed post-treatment differences between an experi-
mental and a control group, provided that these groups dif-
fered only in the therapeutic ingredient of interest (i.e.,
comparisons between two different active interventions were
excluded). The second one encompassed the subset of the
previous ones for which pre and post-intervention measures
were available. These allow computing between-group dif-
ferences in pre-post treatment change measures. That is, they
Fig. 6 Forest plot for meta-
analysis 3
Current Addiction Reports (2025) 12:9 9 Page 10 of 14
potentially allow to test whether the improvement observed
in the experimental group is or not significantly larger than
the one in the control group. Finally, the third one included
studies where there was no control group, or only the experi-
mental groups in studies in which the control group included
a different active intervention (and thus it was not possible to
isolate the effect of the ingredient of interest). In these cases,
we meta-analyzed effect sizes for the within-participant pre-
post treatment difference. The two first meta-analyses (for
between-participant effects) yielded rather similar medium-
to-large effects, whereas the third one (for within-participant
effects) yielded a large effect. However, this shared set of
results must be interpreted with some caution.
First, despite the variety of techniques, outcome meas-
ures, and behavioral domains under initial consideration,
the number of studies that allow a proper assessment of
the efficacy of the techniques of interest is surprisingly
small. For problematic or disordered gambling, we col-
lected 6 studies using reconditioning techniques, 2 studies
using mindfulness-based interventions, and 5 studies using
brain stimulation. For other putative addictions (predomi-
nantly problematic or disordered gaming) we identified 10
studies implementing reconditioning techniques, 5 studies
using mindfulness, and 2 studies using brain stimulation
techniques. This scarcity is partly due to the fact that most
intervention studies and RCTs currently available in the lit-
erature implemented sophisticated treatment packages in
which different therapeutic ingredients remain confounded.
Although this strategy can have a clinical or commercial jus-
tification (as, for instance, empirically supporting the large-
scale use of a closed and fully developed product; e.g [64])
, it also reflects that intervention studies are rarely aimed
at rigorously testing whether fine-grained, theory-informed
add-ons to the available therapeutic options contribute or not
to increase their efficacy.
And second, most studies present some risk of bias, and
many of the funnel-plot asymmetry analyses yield substan-
tial evidence of small-study effects. This is not by itself a
definitive proof of the generalized existence of questionable
research practices, but calls for the need to raise methodolog-
ical standards in the field. As it happens in virtually any other
area in behavioral sciences, more well-powered and prereg-
istered studies are necessary to prevent file-drawer effects,
selective reporting of outcomes, and HARKing [65–69].
Although we have implemented several bias-correction
techniques (PET, PEESE, PET-PEESE, 3PSM, and trim-
and-fill), neither of them is free of criticism [70, 71]. Addi-
tionally, the small number of studies implies that asymmetry
tests are probably underpowered, and the level of uncertainty
in unbiased effect size estimates is too large for them to be
interpretable. This is also probably the reason why differ-
ent methods for bias correction yield so strikingly diverging
unbiased effect size estimates.
The small number of studies and the ensuing low power of
moderation analyses is also probably one of the reasons why
most of these analyses yielded null results. Different addic-
tion models predict effectiveness differences for the condi-
tions under consideration. For instance, habit learning and
conditioning models, which have been proposed to account
for compulsivity development in substance use and gambling
disorders, consider reconditioning techniques to be particularly
well-suited to treat these conditions. On the contrary, gratifica-
tion/compensation models specifically formulated to account
for problematic video gaming and social media use predict
compulsivity-oriented techniques to be less effective (relative,
for example, to belief modification or contingency management
techniques; see [72] for a review). The absence of effectiveness
differences attributable to the candidate disorder’s behavioral
domain or the type of intervention could be interpreted as sup-
porting a Dodo effect (namely, that any therapy is better than
no therapy, but no therapy is better than the other ones), but
is probably better explained by lack of sufficient high-quality
evidence to detect the differences. This insufficiently solid evi-
dence base goes against the long-held –yet modestly material-
ized– ambition of attaining higher levels of treatment tailoring
and personalization for addictive disorders [73, 74].
Additionally, it is important to note that heterogeneity
excess was large and significant in our meta-analyses, but
moderators mostly failed to account for it. Outliers seem to
explain, however, at least part of that heterogeneity (see the
supplementary file 6 provided via the OSF link included in
the Data Availability statement). Moreover, after automated
outlier removal (a) the pooled effect sizes of meta-analysis 1
and 3 shrank to 0.56 and 0.84, respectively, (b) in meta-anal-
ysis 2 the effect of the quality item “selection of the reported
result”, reached significance, and (c) the effect size for the
studies with low risk of bias fell below significance (g = 0.26
[−0.03, 0.56], relative to g = 0.80 [0.56, 1.04] for studies with
some concerns).
Regarding the differences between meta-analyses, the most
relevant result regards the diverging pooled effect sizes found
in studies with and without an adequate control group, with
the former yielding a medium-to-large pooled effect, and the
latter a large one. Although both are probably inflated, there
are two potential factors contributing to this difference. First,
the fact that pre-post treatment changes are substantially larger
than control-experimental group differences probably reveals
that spontaneous recovery from this family of conditions is
non-negligible. Therefore, causal interpretations of pre-post
differences should be avoided, especially in cases where the
condition under study has been shown to be markedly unstable
across time [75]. The second factor concerns the importance
of carefully selecting an adequate control group, which is a
crucial quality index, as it has been widely observed in many
fields that higher-quality studies tend to report smaller effects
(e.g [76]).
Current Addiction Reports (2025) 12:9 Page 11 of 14 9
Limitations
This study is not free of limitations. On the one hand, we are
aware that our selection criteria were overinclusive, as they
collapsed very different techniques and behavioral domains.
This was nonetheless an intentional strategy to allow for
a sufficiently ample evidence base regarding theoretically
related intervention tools, while acknowledging that the
sample of effect sizes would be more representative of a
distribution of them (as modeled by random-effects meta-
analyses), than of a single populational parameter.
On the other hand, and despite this overinclusiveness,
the sample of eligible studies remained small, and prob-
ably insufficient to detect significant moderation effects.
Thus far, as noted earlier we cannot establish the supe-
riority of certain tools over others, or more efficacy of
the assessed tools in some conditions than in others. In
the future, an ampler and more reliable base should allow
practitioners to identify which tools are more or less effec-
tive as a function of the patient’s profile.
Conclusion andFuture Directions
To sum up, our findings yield a promising prospect for
directly addressing compulsive behavior in behavioral
addictions and related conditions. However, this poten-
tial is currently based more on individual, well-conducted
studies rather than on the evidence accumulated across
multiple studies. Among the studies we reviewed, a spe-
cific subset, including [37] and [56], stands out for adher-
ing to higher quality standards, particularly in their use of
random participant assignment and handling of missing
data, and may provide stronger evidence for the interven-
tions’ potential. The first of them [37] used contingent
(experimental) and non-contingent (control) approach-
avoidance training and found problematic gambling
symptoms reduction in both of them (thus failing to show
evidence in favor of the target intervention), whereas
the second one [56] succeeded in finding a significant
reduction of compulsive gaming in a group of individu-
als treated with an emotional bias reduction technique
(relative to a sham treatment group). Interestingly, both
studies used tools directly inspired by rigorous preclini-
cal research on the basic mechanisms of bias learning
and unlearning in addictive processes [77], which shows
systematic theory development is closely linked to high
methodological standards, and both aspects of scientific
advancement should be considered inseparable.
More specifically, our review points out the importance
of implementing several measures for future research to
be able to inform evidence-based therapy for behavioral
addictions and related conditions. First, therapy devel-
opment must be theory-guided, and intervention studies
should be carefully designed to allow a step-by-step pro-
gression in the design of packages that can be tailored
to specific behavioral profiles. Second, and relatedly, the
selection of outcomes and the formulation of hypotheses
regarding them should be established a priori, and obey
the causal mechanisms theorized to account for the effect
of interest. Ideally, the current dominant verbal theories
making dichotomous predictions regarding the efficacy of
interventions should be replaced by formal ones, capable
of making differential quantitative predictions [78–80].
And third, the entire field should work to speed up the
adoption of more transparent, reliable, and collaborative
research practices. Without implementing these measures,
the advancement of psychological interventions will stall
at a Dodo-bird verdict stage where almost any therapy is
better than no therapy, but no single therapy appears to be
better than any other [81].
Key References
• Eben C, Bőthe B, Brevers D, Clark L, Grubbs JB, Heir-
ene R, etal. The landscape of open science in behavioral
addiction research: Current practices and future direc-
tions. J Behav Addict. 2023;https:// doi. org/ 10. 1556/ 2006.
2023. 00052.The first conceptual and position paper on the
importance of adopting open science practices in the field
of behavioral addictions.
•• Muela I, Navas JF, Ventura-Lucena JM, Perales JC.
How to pin a compulsive behavior down: A systematic
review and conceptual synthesis of compulsivity-sensitive
items in measures of behavioral addiction. Addict Behav.
2022;https:// doi. org/ 10. 1016/j. addbeh. 2022. 107410.This
systematic review is the first one rigorously trying to define
and operationalize compulsivity in the field of behavioral
addictions.
•• Wiers RW, Stacy AW, editors. Handbook of implicit
cognition and addiction. Sage; 2006.A sound compilation
of models and theory-informed interventions to understand
the mechanisms underlying the learning and unlearning of
incidental biases contributing to compulsivity in addictions.
Current Addiction Reports (2025) 12:9 9 Page 12 of 14
• Wittekind CE, Bierbrodt J, Lüdecke D, Feist A, Hand I,
Moritz S. Cognitive bias modification in problem and path-
ological gambling using a web-based approach-avoidance
task: A pilot trial. Psychiatry Res. 2019;https:// doi. org/ 10.
1016/j. psych res. 2018. 12. 075.This study stands out for its
theory-informed intervention and methodological quality. It
shows symptoms reduction following contingent and non-
contingent approach-avoidance training, thus failing to show
the effect of cue-response contingency in the reduction of
approach biases in problematic gambling.
• Wu L, Xu J, Song K, Zhu L, Zhou N, Xu L, etal. Emo-
tional bias modification weakens game-related compulsivity
and reshapes frontostriatal pathways. Brain. 2022;https://
doi. org/ 10. 1093/ brain/ awac2 67.This study stands out for its
theory-informed intervention and methodological quality.
It succeeded in finding a significant compulsivity reduction
in a group of people with problematic gaming treated with
an emotional bias reduction technique (relative to a sham
treatment group).
Author Contributions JLG and JCP conceptualized the study. JLG, FJR
and JFN conducted the literature review and JLG and MAV performed
the meta-analysis. JLG, JCP and MAV wrote the main manuscript text.
JLG and MAV prepared Figs.1, 2, 3, 4 and 5. IM, JFN and FJR contrib-
uted to data interpretation and manuscript revision. JCP provided revi-
sions and supervised the project. All authors reviewed and approved
the final manuscript.
Funding Work by JLG, JFN, JCP, FJR and IM is supported by a R&D
project (Proyecto I + D + i), funded by the Spanish Research Agency
(Agencia Española de Investigación), Spanish Ministry of Science and
Innovation (Ministerio de Ciencia e Innovación) (MCIN/AEI/https://
doi. org/ 10. 13039/ 50110 00110 33), with reference PID2020-116535
GB-I00.
JLG’s work is supported by an individual research grant (PRE2021-
100665), funded by MICIU/AEI/https:// doi. org/ 10. 13039/ 50110 00110
33 and by “ESF+”. MV’s work is supported by the project CNS2022-
135346 funded by the Spanish Research Agency. FJR’s work is sup-
ported by an individual research grant (FPU21/00462, Ministerio de
Ciencia e Innovación).
Data Availability Supplementary files are hosted on the Open Science
Framework (OSF) and can be accessed via the following link: https://
osf. io/ 5jg8v/? view_ only= a4f70 8d0f4 f542c 4ab6e 6983a d25d1 90.
Declarations
Human and Animal Rights and Informed Consent This meta-analysis is
based on previously published studies. All procedures involving human
or animal subjects in those studies were assumed to have followed ethi-
cal standards at the time of publication.
Competing Interests The authors declare no competing interests.
References
1. Luigjes J, Lorenzetti V, de Haan S, Youssef GJ, Murawski C,
Sjoerds Z, etal. Defining Compuls Behav Neuropsychol Rev.
2019. https:// doi. org/ 10. 1007/ s11065- 019- 09404-9.
2. Lüscher C, Robbins TW, Everitt BJ. The transition to compul-
sion in addiction. Nat Rev Neurosci. 2020. https:// doi. org/ 10.
1038/ s41583- 020- 0289-z.
3. Vandaele Y, Ahmed SH. Habit, choice, and addiction. Neu-
ropsychopharmacology. 2021. https:// doi. org/ 10. 1038/
s41386- 020- 00899-y.
4. Hogarth L. Addiction is driven by excessive goal-directed drug
choice under negative affect: translational critique of habit and
compulsion theory. Neuropsychopharmacology. 2020. https://
doi. org/ 10. 1038/ s41386- 020- 0600-8.
5. Koob GF. Neurobiological substrates for the Dark side of com-
pulsivity in addiction. Neuropharmacology. 2009. https:// doi.
org/ 10. 1016/j. neuro pharm. 2008. 07. 043.
6. Sjoerds Z, van den Luigjes J, Denys D, Yücel M. The role of
habits and Motivation in Human Drug Addiction: a reflection.
Front Psychiatry. 2014. https:// doi. org/ 10. 3389/ fpsyt. 2014.
00008.
7. Muela I, Navas JF, Ventura-Lucena JM, Perales JC. How to pin
a compulsive behavior down: a systematic review and concep-
tual synthesis of compulsivity-sensitive items in measures of
behavioral addiction. Addict Behav. 2022. https:// doi. org/ 10.
1016/j. addbeh. 2022. 107410.
8. Muela I, Navas JF, Barrada JR, López-Guerrero J, Rivero FJ, Bre-
vers D, etal. Operationalization and measurement of compulsiv-
ity across video gaming and gambling behavioral domains. BMC
Psychol. 2023. https:// doi. org/ 10. 1186/ s40359- 023- 01439-1.
9. Tiego J, Trender W, Hellyer PJ, Grant JE, Hampshire A, Cham-
berlain SR. Measuring compulsivity as a self-reported multidi-
mensional transdiagnostic construct: large-scale (N = 182,000)
validation of the Cambridge–Chicago Compulsivity Trait Scale.
Assessment. 2023. https:// doi. org/ 10. 1177/ 10731 91122 11490 83.
10. Magill M, Martino S, Wampold B. The principles and practices
of psychoeducation with alcohol or other drug use disorders: a
review and brief guide. J Subst Abuse Treat. 2021. https:// doi.
org/ 10. 1016/j. jsat. 2021. 108442.
11. Oluwoye O, Kriegel L, Alcover KC, McPherson S, McDonell MG,
Roll JM. The dissemination and implementation of contingency
management for substance use disorders: a systematic review.
Psychol Addict Behav. 2020. https:// doi. org/ 10. 1037/ adb00 00487.
12. Verdejo-Garcia A, Chong TTJ. Targeting goal-based decision-
making for addiction recovery. Pharmacol Biochem Behav.
2021. https:// doi. org/ 10. 1016/j. pbb. 2021. 173275.
13. Stasiewicz PR, Bradizza CM, Schlauch RC, Coffey SF, Gulliver
SB, Gudleski GD, etal. Affect regulation training (ART) for
alcohol use disorders: development of a novel intervention for
negative affect drinkers. J Subst Abuse Treat. 2013. https:// doi.
org/ 10. 1016/j. jsat. 2013. 05. 012.
14. Martin T, LaRowe SD, Malcolm R. Progress in cue expo-
sure therapy for the treatment of addictive disorders: a review
update. Open Addict J. 2010. https:// doi. org/ 10. 2174/ 18749
41001 00301 0092.
15. Wiebe A, Kannen K, Selaskowski B, Mehren A, Thöne AK,
Pramme L, etal. Virtual reality in the diagnostic and therapy for
mental disorders: a systematic review. Clin Psychol Rev. 2022.
https:// doi. org/ 10. 1016/j. cpr. 2022. 102213.
Current Addiction Reports (2025) 12:9 Page 13 of 14 9
16. Wiers RW, Van Dessel P, Köpetz C. ABC training: a new the-
ory-based form of Cognitive-Bias modification to foster autom-
atization of alternative choices in the treatment of addiction and
related disorders. Curr Dir Psychol Sci. 2020. https:// doi. org/
10. 1177/ 09637 21420 949500.
17. Kiefer F, Dinter C. New approaches to addiction treatment based
on learning and memory. In: Sommer W, Spanagel R, editors.
Behavioral neurobiology of alcohol addiction. Current topics in
behavioral neurosciences. Berlin, Heidelberg: Springer; 2011.
pp. 247–70. https:// doi. org/ 10. 1007/ 978-3- 642- 28720-6_ 147.
18. Treanor M. The potential impact of mindfulness on exposure
and extinction learning in anxiety disorders. Clin Psychol Rev.
2011. https:// doi. org/ 10. 1016/j. cpr. 2011. 02. 003.
19. Tapper K. Mindfulness and craving: effects and mechanisms.
Clin Psychol Rev. 2018. https:// doi. org/ 10. 1016/j. cpr. 2017. 11.
003.
20. Murphy E, Poudel G, Ganesan S, Suo C, Manning V, Beyer E,
etal. Real-time fMRI-based neurofeedback to restore brain func-
tion in substance use disorders: a systematic review of the lit-
erature. Neurosci Biobehav Rev. 2024. https:// doi. org/ 10. 1016/j.
neubi orev. 2024. 105865.
21. Lucas I, Solé-Morata N, Baenas I, Rosinska M, Fernández-Aranda
F, Jiménez-Murcia S. Biofeedback interventions for impulsivity-
related processes in addictive disorders. Curr Addict Rep. 2023.
https:// doi. org/ 10. 1007/ s40429- 023- 00499-y.
22. Ter Harmsel JF, Noordzij ML, Goudriaan AE, Dekker JJM, Van
Der Swinkels LTA, etal. Biocueing and ambulatory biofeedback
to enhance emotion regulation: a review of studies investigating
non-psychiatric and psychiatric populations. Int J Psychophysiol.
2021. https:// doi. org/ 10. 1016/j. ijpsy cho. 2020. 11. 009.
23. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC,
Mulrow CD, etal. Updating guidance for reporting systematic
reviews: development of the PRISMA 2020 statement. J Clin Epi-
demiol. 2021. https:// doi. org/ 10. 1016/j. jclin epi. 2021. 02. 003.
24. Smith DP, Dunn KI, Harvey PW, Battersby MW, Pols RG. Assess-
ing randomised clinical trials of cognitive and exposure therapies
for gambling disorders: a systematic review. Behav Change. 2013.
https:// doi. org/ 10. 1017/ bec. 2013. 15.
25. Concerto C, Signorelli MS, Chiarenza C, Ciancio A, Di Francesco
A, Mineo L, etal. Transcranial magnetic stimulation for the treat-
ment of gambling disorder: a systematic review. J Integr Neurosci.
2023. https:// doi. org/ 10. 31083/j. jin22 06164.
26. Chen Y, Lu J, Wang L, Gao X. Effective interventions for gaming
disorder: a systematic review of randomized control trials. Front
Psychiatry. 2023. https:// doi. org/ 10. 3389/ fpsyt. 2023. 10989 22.
27. Sterne JAC, Savović J, Page MJ, Elbers RG, Blencowe NS,
Boutron I, etal. RoB 2: a revised tool for assessing risk of bias in
randomised trials. BMJ. 2019. https:// doi. org/ 10. 1136/ bmj. l4898.
28. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduc-
tion to meta-analysis. 1st ed. Wiley; 2009. https:// doi. org/ 10. 1002/
97804 70743 386.
29. Morris SB, DeShon RP. Combining effect size estimates in meta-
analysis with repeated measures and independent-groups designs.
Psychol Methods. 2002. https:// doi. org/ 10. 1037/ 1082- 989X.7. 1. 105.
30. Rosenthal R. Meta-analytic procedures for social research. SAGE
Publications, Inc.; 1991. https:// doi. org/ 10. 4135/ 97814 12984 997.
31. Morris SB. Estimating effect sizes from pretest-posttest-control
group designs. Organ Res Methods. 2008. https:// doi. org/ 10. 1177/
10944 28106 291059.
32. n den Noortgate W, López-López JA, Marín-Martínez F, Sánchez-
Meca J. Three-level meta-analysis of dependent effect sizes. Behav
Res Methods. 2013. https:// doi. org/ 10. 3758/ s13428- 012- 0261-6.
33. Lan Y, Ding JE, Li W, Li J, Zhang Y, Liu M, etal. A pilot study of
a group mindfulness-based cognitive-behavioral intervention for
smartphone addiction among university students. J Behav Addict.
2018. https:// doi. org/ 10. 1556/ 2006.7. 2018. 103.
34. Holas P, Draps M, Kowalewska E, Lewczuk K, Gola M. A pilot
study of mindfulness-based relapse prevention for compulsive
sexual behaviour disorder. J Behav Addict. 2020. https:// doi. org/
10. 1556/ 2006. 2020. 00075.
35. Gay A, Boutet C, Sigaud T, Kamgoue A, Sevos J, Brunelin J, etal.
A single session of repetitive transcranial magnetic stimulation of
the prefrontal cortex reduces cue-induced craving in patients with
gambling disorder. Eur Psychiatry. 2017. https:// doi. or g/ 10. 1016/j.
eurpsy. 2016. 11. 001.
36. Salerno L, Grassi E, Makris N, Pallanti S. A Theta burst stimu-
lation on pre-SMA: proof-of-concept of transcranial magnetic
stimulation in gambling disorder. J Gambl Stud. 2022. https://
doi. org/ 10. 1007/ s10899- 022- 10129-3.
37. Wittekind CE, Bierbrodt J, Lüdecke D, Feist A, Hand I, Moritz S.
Cognitive bias modification in problem and pathological gambling
using a web-based approach-avoidance task: a pilot trial. Psychia-
try Res. 2019. https:// doi. org/ 10. 1016/j. psych res. 2018. 12. 075.
38. Smith DP, Battersby MW, Harvey PW, Pols RG, Ladouceur R.
Cognitive versus exposure therapy for problem gambling: Ran-
domised controlled trial. Behav Res Ther. 2015. https:// doi. org/
10. 1016/j. brat. 2015. 04. 008.
39. Lu C, Zou L, Becker B, Griffiths M, Yu Q, Chen S, etal. Com-
parative effectiveness of mind-body exercise versus cognitive
behavioral therapy for college students with problematic smart-
phone use: a randomized controlled trial. Int J Ment Health Pro-
mot. 2020. https:// doi. org/ 10. 32604/ ijmhp. 2020. 014419.
40. Melero Ventola AR, Yela JR, Crego A, Cortes-Rodriguez M.
Effectiveness of a mindfulness-based cognitive therapy group
intervention in reducing gambling-related craving. J Evid Based
Psychother. 2020;20(1):107–34.
41. Wu L, Potenza MN, Zhou N, Kober H, Shi X, Yip SW, etal.
Efficacy of single-session transcranial direct current stimulation
on addiction-related inhibitory control and craving: a randomized
trial in males with internet gaming disorder. J Psychiatry Neuro-
sci. 2021. https:// doi. org/ 10. 1503/ jpn. 190137.
42. Rémond JJ, Hernández-Mora M, Marseille C, Mesure G, Romo
L. Feasibility study of therapeutic exposure in virtual reality for
pathological gamblers. Encephale. 2020. https:// doi. org/ 10. 1016/j.
encep. 2019. 09. 002.
43. Blaszczynski A, Drobny J, Steel Z. Home-based imaginal desen-
sitisation in pathological gambling: short-term outcomes. Behav
Change. 2005. https:// doi. org/ 10. 1375/ bech. 22.1. 13. 66782.
44. Snippe L, Boffo M, Galvin H, Willemen R, Pronk T, Dom G,
Wiers RW. How we lost 90% of participants on a bad bet: results
from a pilot randomized controlled trial on cognitive bias modifi-
cation in problem gamblers. J Gambl Stud. 2023. https:// doi. org/
10. 1007/ s10899- 023- 10263-6.
45. Pettorruso M, Martinotti G, Montemitro C, De Risio L, Spag-
nolo PA, Gallimberti L, etal. Multiple sessions of high-frequency
repetitive transcranial magnetic stimulation as a potential treat-
ment for gambling addiction: a 3-month, feasibility study. Eur
Addict Res. 2020. https:// doi. org/ 10. 1159/ 00050 4169.
46. Lee S, Im J, Oh JK, Choi E, Yoon S, Bikson M, etal. Transcranial
direct current stimulation for online gamers: a prospective single-
arm feasibility study. J Behav Addict. 2018. https:// doi. org/ 10.
1556/ 2006.7. 2018. 107.
47. McIntosh CC, Crino RD, O’Neill K. Treating problem gambling
samples with cognitive behavioural therapy and mindfulness-
based interventions: a clinical trial. J Gambl Stud. 2016. https://
doi. org/ 10. 1007/ s10899- 016- 9602-1.
48. Liu L, Potenza MN, Lacadie CM, Zhang JT, Yip SW, Xia CC,
etal. Altered intrinsic connectivity distribution in internet gam-
ing disorder and its associations with psychotherapy treatment
outcomes. Addict Biol. 2021. https:// doi. org/ 10. 1111/ adb. 12917.
49. Zhang JT, Yao YW, Potenza MN, Xia CC, Lan J, Liu L, etal.
Altered resting-state neural activity and changes following a
Current Addiction Reports (2025) 12:9 9 Page 14 of 14
craving behavioral intervention for internet gaming disorder. Sci
Rep. 2016. https:// doi. org/ 10. 1038/ srep2 8109.
50. He J, Pan T, Nie Y, Zheng Y, Chen S. Behavioral modification
decreases approach bias in young adults with internet gaming dis-
order. Addict Behav. 2021. https:// doi. org/ 10. 1016/j. addbeh. 2020.
106686.
51. Sauvaget A, Bulteau S, Guilleux A, Leboucher J, Pichot A, Val-
rivière P, etal. Both active and sham low-frequency rTMS single
sessions over the right DLPFC decrease cue-induced cravings
among pathological gamblers seeking treatment: a randomized,
double-blind, sham-controlled crossover trial. J Behav Addict.
2018. https:// doi. org/ 10. 1556/ 2006.7. 2018. 14.
52. Yao YW, Chen PR, Li CR, Hare TA, Li S, Zhang JT, etal. Com-
bined reality therapy and mindfulness meditation decrease inter-
temporal decisional impulsivity in young adults with internet
gaming disorder. Comput Hum Behav. 2017. https:// doi. org/ 10.
1016/j. chb. 2016. 11. 038.
53. Zhang JT, Yao YW, Potenza MN, Xia CC, Lan J, Liu L, etal.
Effects of craving behavioral intervention on neural substrates
of cue-induced craving in internet gaming disorder. Neuroimage
Clin. 2016. https:// doi. org/ 10. 1016/j. nicl. 2016. 09. 004.
54. Zhao Q, Zhang Y, Wang M, Ren J, Chen Y, Chen X, etal. Effects
of retrieval-extinction training on internet gaming disorder. J
Behav Addict. 2022. https:// doi. org/ 10. 1556/ 2006. 2022. 00006.
55. Pallanti S, Camprodon JA, Di Ponzio M, Makris N. Efficacy of
continuous theta burst stimulation to the pre-supplementary motor
area in gambling disorder: a randomized double-blind controlled
trial. Int J Ment Health Addict. 2023. https:// doi. org/ 10. 1007/
s11469- 023- 01120-z.
56. Wu L, Xu J, Song K, Zhu L, Zhou N, Xu L, etal. Emotional bias
modification weakens game-related compulsivity and reshapes
frontostriatal pathways. Brain. 2022. https:// doi. org/ 10. 1093/
brain/ awac2 67.
57. Riley B, Smith D, Oakes J. Exposure therapy for problem gam-
bling in rural communities: a program model and early outcomes.
Aust J Rural Health. 2011. https:// doi. org/ 10. 1111/j. 1440- 1584.
2011. 01199.x.
58. Li W, Garland EL, McGovern P, O’Brien JE, Tronnier C, Howard
MO. Mindfulness-oriented recovery enhancement for internet gam-
ing disorder in U.S. adults: a stage I randomized controlled trial. Psy-
chol Addict Behav. 2017. https:// doi. org/ 10. 1037/ adb00 00269.
59. Rabinovitz S, Nagar M. Possible end to an endless quest? Cogni-
tive bias modification for excessive multiplayer online gamers.
Cyberpsychol Behav Soc Netw. 2015. https:// doi. org/ 10. 1089/
cyber. 2015. 0173.
60. Park SY, Kim SM, Roh S, Soh MA, Lee SH, Kim H, etal. The
effects of a virtual reality treatment program for online gaming
addiction. Comput Methods Programs Biomed. 2016. https:// doi.
org/ 10. 1016/j. cmpb. 2016. 01. 015.
61. Oka T, Kubo T, Kobayashi N, Murakami M, Chiba T, Cortese
A. Decoding and modifying dynamic attentional bias in gaming
disorder. Philos Trans R Soc Lond B Biol Sci. 2024. https:// doi.
org/ 10. 1098/ rstb. 2023. 0090.
62. Ni H, Wang H, Ma X, Li S, Liu C, Song X, etal. Efficacy and
neural mechanisms of mindfulness meditation among adults with
internet gaming disorder: a Randomized Clinical Trial. JAMA
Netw Open. 2024. https:// doi. org/ 10. 1001/ jaman etwor kopen.
2024. 16684.
63. Viechtbauer W. Conducting Meta-analyses in R with the metafor
Package. J Stat Softw. 2010. https:// doi. org/ 10. 18637/ jss. v036. i03.
64. Martín-Pérez C, Navas JF, Perales JC, López-Martín Á, Cordovilla-
Guardia S, Portillo M, etal. Brief group-delivered motivational
interviewing is equally effective as brief group-delivered cognitive-
behavioral therapy at reducing alcohol use in risky college drinkers.
PLoS ONE. 2019. https:// doi. org/ 10. 1371/ journ al. pone. 02262 71.
65. Hardwicke TE, Wagenmakers EJ. Reducing bias, increasing trans-
parency and calibrating confidence with preregistration. Nat Hum
Behav. 2023. https:// doi. org/ 10. 1038/ s41562- 022- 01497-2.
66. Lakens D. Sample size justification. Collabra Psychol. 2022.
https:// doi. org/ 10. 1525/ colla bra. 33267.
67. Eben C, Bőthe B, Brevers D, Clark L, Grubbs JB, Heirene R, etal.
The landscape of open science in behavioral addiction research:
current practices and future directions. J Behav Addict. 2023.
https:// doi. org/ 10. 1556/ 2006. 2023. 00052.
68. Louderback ER, Gainsbury SM, Heirene RM, Amichia K, Gross-
man A, Bernhard BJ, etal. Open science practices in gambling
research publications (2016–2019): a scoping review. J Gambl
Stud. 2022. https:// doi. org/ 10. 1007/ s10899- 022- 10120-y.
69. LaPlante DA, Louderback ER, Abarbanel B. Gambling research-
ers’ use and views of open science principles and practices: a brief
report. Int Gambl Stud. 2021. https:// doi. org/ 10. 1080/ 14459 795.
2021. 18912 72.
70. Carter EC, Schönbrodt FD, Gervais WM, Hilgard J. Correcting for
bias in psychology: a comparison of meta-analytic methods. Adv
Methods Pract Psychol Sci. 2019. https:// doi. org/ 10. 1177/ 25152
45919 847196.
71. Bartoš F, Maier M, Wagenmakers EJ, Doucouliagos H, Stanley
TD. Robust bayesian meta-analysis: model-averaging across
complementary publication bias adjustment methods. Res Synth
Methods. 2023. https:// doi. org/ 10. 1002/ jrsm. 1594.
72. Perales JC, King DL, Navas JF, Schimmenti A, Sescousse G,
Starcevic V, etal. Learning to lose control: a process-based
account of behavioral addiction. Neurosci Biobehav Rev. 2020.
https:// doi. org/ 10. 1016/j. neubi orev. 2019. 12. 025.
73. Carroll KM. The profound heterogeneity of substance use disor-
ders: implications for treatment development. Curr Dir Psychol
Sci. 2021. https:// doi. org/ 10. 1177/ 09637 21421 10269 84.
74. Oslin D. Personalized addiction treatment: how close are we?
Alcohol Alcohol. 2011. https:// doi. org/ 10. 1093/ alcalc/ agr030.
75. Ferguson CJ, Jeong EJ, Wang JCK. Pathological gaming: a lon-
gitudinal study from the perspectives of mental health problems
and social stress model. J Gen Psychol. 2023. https:// doi. org/ 10.
1080/ 00221 309. 2022. 20601 76.
76. Balk EM, Bonis PA, Moskowitz H, Schmid CH, Ioannidis JP,
Wang C, Lau J. Correlation of quality measures with estimates of
treatment effect in meta-analyses of randomized controlled trials.
JAMA. 2002. https:// doi. org/ 10. 1001/ jama. 287. 22. 2973.
77. Wiers RW, Stacy AW, editors. Handbook of implicit cognition and
addiction. Sage; 2006.
78. Haslbeck JMB, Ryan O, Robinaugh DJ, Waldorp LJ, Borsboom D.
Modeling psychopathology: from data models to formal theories.
Psychol Methods. 2022. https:// doi. org/ 10. 1037/ met00 00303.
79. Fried EI. Theories and models: what they are, what they are for,
and what they are about. Psychol Inq. 2020. https:// doi. org/ 10.
1080/ 10478 40X. 2020. 18540 11.
80. van Rooij I, Blokpoel M. Formalizing verbal theories: a tutorial by
dialogue. Soc Psychol. 2020. https:// doi. org/ 10. 1027/ 1864- 9335/
a0004 28.
81. Budd R, Hughes I. The Dodo Bird verdict—Controversial, inevi-
table and important: a commentary on 30 years of meta-analyses.
Clin Psychol Psychother. 2009. https:// doi. org/ 10. 1002/ cpp. 648.
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds
exclusive rights to this article under a publishing agreement with the
author(s) or other rightsholder(s); author self-archiving of the accepted
manuscript version of this article is solely governed by the terms of
such publishing agreement and applicable law.
A preview of this full-text is provided by Springer Nature.
Content available from Current Addiction Reports
This content is subject to copyright. Terms and conditions apply.
