DIALOGUES IN CLINICAL NEUROSCIENCE 2023, VOL. 25, NO. 1, 14–23 https://doi.org/10.1080/19585969.2023.2168135 REVIEW Independent component analysis for internet gaming disorder Gemma Mestre-Bacha,b , Roser Graneroc,d,e , Fernando Fernandez-Arandad,e,f,g , Susana Jimenez-Murciad,e,f,g and Marc N. Potenzah,i,j,k,l,m a Facultad de Ciencias de la Salud, Universidad Internacional de la Rioja, La Rioja, Spain; bInstitute for Culture and Society (ICS), University of Navarra, Pamplona, Spain; cDepartament de Psicobiologia i Metodologia de les Ciencies de la Salut, Universitat Autonoma de Barcelona, Barcelona, Spain; dCiber Fisiopatologıa Obesidad y Nutricion (CIBERObn), Instituto de Salud Carlos III, Madrid, Spain; ePsychoneurobiology of Eating and Addictive Behaviors Group, Neurosciences Programme, Bellvitge Biomedical Research Institute (IDIBELL), Barcelona, Spain; fBehavioral Addictions Unit, Department of Psychiatry, Bellvitge University Hospital, Barcelona, Spain; gDepartment of Clinical Sciences, School of Medicine and Health Sciences, University of Barcelona, Barcelona, Spain; hDepartment of Psychiatry, Yale University School of Medicine, New Haven, CT, USA; iDepartment of Neuroscience, Yale University School of Medicine, New Haven, CT, USA; jChild Study Center, Yale University School of Medicine, New Haven, CT, USA; k Connecticut Council on Problem Gambling, Wethersfield, CT, USA; lConnecticut Mental Health Center, New Haven, CT, USA; mWu Tsai Institute, Yale University School of Medicine, New Haven, CT, USA ABSTRACT ARTICLE HISTORY Introduction: There is a growing interest in the study of the neurobiological correlates of inter- Received 21 September 2022 net gaming disorder (IGD), and new techniques are beginning to be implemented for this pur- Revised 24 October 2022 pose, such as independent component analysis (ICA). Accepted 17 November 2022 Aims: The present narrative review aimed to explore the studies that had used ICA for the study KEYWORDS of the different brain networks possibly associated with IGD. Internet gaming disorder; Methods: We specifically focussed on three of the main networks: default-mode network, execu- independent component tive-control and salience networks. analysis; fMRI; default-mode Results: Most studies have identified alterations in these three brain networks in individuals network; executive-control with IGD, which may be involved in the development and maintenance of this disorder. network; salience network Conclusion: More studies are needed to deepen an understanding of the specific role of each in the symptomatology and treatment of IGD. Introduction between (a) an increase in incentive-oriented urges and desires; and (b) a reduction in situation-specific Internet Gaming Disorder (IGD) was recommended as a inhibitory control over these urges and desires (Brand mental disorder in Section III of the Diagnostic and et al. 2019). Therefore, individuals with reward deficien- Statistical Manual of Mental Disorders (DSM-5) cies are more likely to develop incentive sensitisation (American Psychiatric Association 2013), and gaming dis- (as a result of conditioning processes), which may be order has been recognised in the 11th revision of the associated with attentional biases, cue-reactivity and International Classification of Disease (ICD-11) (World craving, factors that may contribute to engagement in Health Organization 2020) as a ‘disorder due to addict- IGD and other addictions (Brand et al. 2019). ive behaviors’. IGD has been understood as a persistent The neural correlates of IGD involved in this imbal- and recurrent pattern of internet gaming involvement ance have been explored in different functional mag- despite physical/psychological impairment. netic resonance imaging (fMRI) studies, and some The Interaction of Person-Affect-Cognition-Execution have used independent component analysis (ICA). The (I-PACE) model has proposed that IGD, as well as other exploration of neural correlates or biomarkers (under- addictive behaviours, is the consequence of interac- stood as any structure, process, or substance that can tions that occur between core characteristics of an be measured objectively) is important in IGD due to individual and multiple moderating and mediating vari- the increasing prevalence of, and morbidity related to, ables (Brand et al. 2016, 2019). At a neurobiological IGD (Strimbu and Tavel 2010; Kashif et al. 2021). The level, this model suggests that the development and identification of biomarkers could facilitate improved maintenance of IGD are associated with an imbalance understanding that could lead to better prevention CONTACT Marc N. Potenza marc.potenza@yale.edu ß 2023 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. DIALOGUES IN CLINICAL NEUROSCIENCE 15 and treatment strategies for IGD (Kashif et al. 2021). and IGD (Bin et al. 2022; Zeng et al. 2022). The present In the present review, the main findings of these stud- narrative review aims to examine the main studies that ies using ICA will be discussed (Table 1), taking into have used ICA in individuals with IGD in order to gain account three main brain connectivity networks: the insight into the neurobiology of this mental disorder. default-mode network (DMN), the executive-control network (ECN) and the salience network (SN). Default-mode network The brain’s DMN encompasses different networks pre- Independent component analysis in internet sent in different areas of the association cortex gaming disorder (Buckner and DiNicola 2019). Specifically, the main brain regions involved in the DMN are the posterior Data-driven methodologies have advantages over cingulate cortex (PCC), the medial prefrontal cortex some other approaches in that they may facilitate the (MPFC) and the inferior parietal lobule (IPL) (Di and identification of novel, unanticipated findings and thus Biswal 2014). The DMN has been implicated in high- propel fields of research. When applied to neuroimag- level cognitive processes, such as emotional-regulation ing data, ICA is a blind-source separation technique processes, autobiographical memory and self-related that may allow the identification of distinct groups of cognition, social cognition, risky decision-making, brain regions that exhibit the same temporal pattern impulsivity and future-oriented thinking (Buckner and of homodynamic signal change. Furthermore, it does DiNicola 2019; Yan et al. 2021; Figure 1). not rely on any experimental design matrix or any In the case of IGD, individuals may show less con- other prior information about the temporal response nectivity with DMN brain areas (Liu, Potenza, et al. (Wang, Zhang et al. 2018). Therefore, it is a method 2021), and an altered functional interaction between that belongs to blind signal-separation methods that the DMN and the ventral-attentional network. This assume that there is statistical independence of the may increase incentive salience, that is, attentional source signals, and which allows the isolation of these bias, and reduce cognitive control (Yan et al. 2021). different sources (McKeown et al. 2003). In order to Although the specific role of DMN in the development identify these source signals, covariance measures are and maintenance of IGD remains unclear (Yan et al. necessary and higher-order statistic algorithms (such as 2021), it has been suggested that less connectivity in FastICA, Infomax, or JADE) or intra-source correlation the DMN may be a neural marker of the disorder, methods (such as Molgedey and Schuster, and Ziehe which contributes importantly to the differentiation and Muller) are often used (McKeown et al. 2003). between individuals with IGD and recreational game Some authors support its usefulness given the limi- use (Dong et al. 2020), as well as between IGD and tations of general linear modelling (GLM), a widely non-gaming control participants (Dong et al. 2017). used index to explore task-specific brain responses. Additionally, although IGD was not included in this This is done by analysing a time series of blood- study, a craving network identified across conditions oxygen-level dependent signals. However, this index suggests that craving may involve difficulties redirect- requires an experimental design matrix and does not ing attention from internal self-related processes as allow the identification of a specific brain region that linked to the DMN (Garrison et al. 2021). Given the is functionally connected to another (Wang, Zhang relevance of craving to IGD, additional research should et al. 2018). The same applies to the more classical explore this possible role for the DMN in IGD. Some methods of factor analysis and principal component authors have suggested that the DMN may be the analysis, which, as ICA, can decompose the data, but most informative network in predicting the presence may not find underlying sources (Hyv€arinen 2013). of IGD (Song et al. 2021). In recent years, ICA has begun to be more widely Some studies have explored a specific role for the used in the mental health field, in disorders such as DMN in IGD using ICA. Lee et al. (2020) used ICA to schizophrenia (Brandt et al. 2015; Gupta et al. 2017), explore the DMN (focusing on the PCC) in 17 male depression (Maglanoc et al. 2020; Liu, Jiao, et al. 2021; adolescents with IGD and 18 control participants. The Luo et al. 2022), anxiety (Ni et al. 2021) and bipolar authors found aberrant functional connectivity of (Yip et al. 2014; Tang et al. 2020) disorders. In the DMN (specifically involving the PCC) with the left pos- addictions field, it has been used in the study of both terior superior temporal sulcus (pSTS) only in the substance-use disorders and addictive behaviours, and group of adolescents with IGD. The DMN, specifically mostly in cocaine-use disorder (Zhang et al. 2016; the anterior DMN (aDMN), was also explored through Wang, Worhunsky et al. 2018; Worhunsky et al. 2021), ICA by Ma et al. (2021) in a sample of 21 male young 16 Table 1. Main studies using ICA to explore the neurobiology of IGD. Reference Country Aims Sample IGD assessment Task Results Lee et al. (2020) South Korea To investigate the functional 17 male adolescents with Korean Internet Addiction Resting state Differential FC of the SN and DMN connectivity between large-scale IGD; 18 male HCs Proneness scale; 9 DSM-5 with the left pSTS was observed intrinsic networks including the criteria for IGD in male adolescents with IGD. FC default-mode, executive-control between the SN and pSTS and salience networks. correlated with severity of internet addiction and self-reported cognitive problems. ICA revealed that pSTS was involved in a social brain network. G. MESTRE-BACH ET AL. Ma et al. (2019) China To investigate the activity of 29 male individuals with IGD; Scores of the CIAS 67; Cue-reactivity task involving 4 FNs were identified that were temporally coherent, large-scale 23 male HCs Spent more than half of internet-gaming stimuli related to the response to game FNs during cue-reactivity in IGD. the time online on games; (i.e., game cues) and cues relative to control cues and Spent 14 h on internet general internet surfing- that showed altered gaming per week (i.e., related stimuli (i.e., control engagement/disengagement in with averagely at least 2 h cues) IGD in comparison with HCs. spent on internet gaming These FNs included temporo- every day) occipital and temporo-insula networks (related to sensory processing), a frontoparietal network (involved in memory and executive functioning) and a dorsal-limbic network (involved in reward and motivation processing). Within IGD participants, game versus control engagement of the temporo- occipital and frontoparietal networks were positively linked with IGD severity. Similarly, disengagement of the temporo- insula network was negatively correlated with higher game- craving. Ma et al. (2021) China (1) To examine fMRI data of the 21 male young adults with Meeting at least 5 of the Stop-signal task Diminished engagement of the stop-signal task and compare IGD; 21 male HCs 9 DSM-5 criteria for IGD; fronto-parietal network during individuals with IGD and HCs in Scores 50 on the IAT; post-error slowing was observed, neural responses during post-error Engagement in internet as well as weaker activity in the behavioural adjustment. gaming for >20 h per ventral attention and anterior (2) To investigate the neural week for a minimum of 1 DMN in response to stop signals representation of probabilistic year; internet gaming as in IGD compared to HCs. expectations of the stop signal the primary online activity during the stop-signal task. Wang et al. (2016) China To identify the neural mechanism of 19 male individuals with IGD; IAT; 9 DSM-5 criteria for IGD Probability discounting task The individuals with IGD showed risky decision-making in IGD using 21 male HCs greater task-related activity in a probability discounting task. DMN and less engagement in the ECN in comparison with HCs during risky decisions. The engagement of the DMN was negatively associated with the reaction time and the ECN (continued) Table 1. Continued. Reference Country Aims Sample IGD assessment Task Results engagement was positively linked with the probability discounting rates. Wang et al. (2017) China To explore FC in male participants 18 male individuals with IGD; IAT Delay-discounting task Two networks were linked with IGD: with IGD. 21 male HCs (a) the ECN containing the anterior cingulate cortex and the medial and superior frontal gyrus; and, (b) the basal ganglia network containing the lentiform nucleus. Individuals with IGD showed stronger FC when selecting small and now options, in comparison with HCs. In addition, the delay-discounting rates were positively associated with the modulation of the 2 networks and reaction times. The results suggest that individuals with IGD may have enhanced sensitivity to reward and decreased ability to control their impulsivity effectively, which may lead to myopic decision-making. Wang, Zhang et al. (2018) China To explore alterations in related 18 male individuals with IGD; IAT; 9 DSM-5 criteria for IGD Addiction Stroop task ICA identified 4 functional networks functional brain networks 19 male HCs that showed differences between underlying attentional biases in both groups, which were related individuals with IGD. to the right ECN and visual networks. Right ECN: individuals with IGD showed increased FC in the temporal gyrus and frontal gyrus, and reduced FC in the posterior cingulate cortex, temporal gyrus and frontal gyrus, in comparison with HCs. Wang et al. (2019) China To explore differences between the 18 male individuals with IGD; IAT; 9 DSM-5 criteria for IGD Probability discounting task ICA analysis showed that individuals engagement of neuronal networks 20 male RGUs with IGD have stronger FC in of individuals with IGD and that reward circuits and their ECN, as of RGUs. well as lower FC in the ASN than RGUs. Xing et al. (2014) China To investigate whether the structural 17 adolescents with IGD; 17 IAT; LOL as their primary Stroop colour-word task Impaired cognitive control in IGD connectivity and FC within SN adolescent HCs mainly Internet activity was suggested by more errors were altered in adolescents with during the incongruent condition IGD, and to assess impaired in the colour-word Stroop task. cognitive control in adolescents The right SN tract showed the with IGD. decreased FA in adolescents with IGD, though no significant differences in FC were detected. The FA values of the right SN tract were negatively correlated DIALOGUES IN CLINICAL NEUROSCIENCE with the errors during the (continued) 17 18 Table 1. Continued. Reference Country Aims Sample IGD assessment Task Results incongruent condition in IGD adolescents. Yip et al. (2018) China To compare neural responses 28 male individuals with IGD; 9 DSM-5 criteria for IGD Emotio-regulation task Individuals with IGD showed between drug-naïve young adults 28 male HCs significantly blunted neural with and without IGD during responses within distributed performance of an emotion- subcortical and cortical regions regulation task using both whole- including the striatum, insula, brain GLM and ICA approaches. lateral prefrontal cortex and anterior cingulate in response to G. MESTRE-BACH ET AL. negative affective cues, as well as during emotion regulation, in comparison with HCs. ICA further identified between-group differences in engagement of a fronto-cingulo-parietal network, involving decreased engagement in individuals with IGD, compared to HCs. Yuan et al. (2016) China To investigate the abnormal brain 28 adolescents with IGD; 25 YDQ; LOL as their primary Stroop colour-word task Abnormal FC within CENs and network interactions using adolescent HCs mainly internet activity effective connectivity within the multimodal imaging between SN in IGD adolescents were adolescents with IGD and HCs. detected, and inefficient interactions between both brain networks were observed. Reduced fractional anisotropy in the SN and right CEN tracts, and between-network (the anterior cingulate cortex-right dorsolateral prefrontal cortex tracts) pathways were identified in individuals with IGD. A significant correlation between the effective and structural connection from SN to CENs and the number of errors during the incongruent condition in the Stroop task were identified in individuals with IGD and HCs. Zeng et al. (2022) China To investigate the effective 148 individuals with IGD; 169 IAT; 9 DSM-5 diagnostic Resting-state fMRI data Individuals with IGD showed connectivity patterns differences RGUs criteria for IGD inhibitory effective connectivity of frontostriatal circuits between from the right OFC to the right individuals with IGD and HCs. caudate and from the right dorsolateral prefrontal cortex to the left OFC, compared with RGUs. Excitatory effective connectivity was observed from the thalamus to the left OFC. Correlation analyses showed that the directional connection from the right OFC to the right caudate (continued) DIALOGUES IN CLINICAL NEUROSCIENCE 19 adults with IGD and 21 control participants. The IAT: Internet addiction test; IC: independent component; ICA: independent component analysis; IGD: Internet gaming disorder; LOL: League of Legends; MELODIC: Multivariate Exploratory Linear Decomposition into ECN: executive-control network; FA: fractional anisotropy; FC: functional connectivity; fMRI: functional magnetic resonance imaging; FNs: functional brain networks; GLM: general-linear-model; HCs: healthy controls; Independent Components; mPFC: medial prefrontal cortex; OFC: orbitofrontal cortex; pSTS: posterior superior temporal sulcus; RAI: resource allocation index; RGU: recreational Internet game users; TUDs: tobacco Abbreviations. ASN: anterior salience network; CENs: central executive networks; CIAS: Chinese Internet Addiction Scale; DMN: default mode network; DSM: Diagnostic and Statistical Manual of Mental Disorders; subcortical network and cerebellar Individuals with IGD had significantly measures in IGD participants were and decreased RAI connectivity in negatively associated with their network were found using ICA. increased SN-DMN connectivity authors observed that the aDMN showed greater dis- compared with HCs. The rRAI the right hemisphere (rRAI), was negatively linked with engagement during stop-signal anticipation. This Significant differences in the could reflect a lower attentional and monitoring proc- Results essing of unexpected cues and, consequently, a sub- addiction severity. craving scores. optimal disposition to withhold response in the case of individuals with IGD. Wang et al. (2016) through ICA, observed that 19 individuals with IGD included in their study showed higher task-related activity in DMN, compared to control participants. Score 67 of the CIAS; Spent Resting-state fMRI data DSM-5 criteria for IGD; Scores Resting-state fMRI data Executive-control network Task The ECN involves fronto-parietal regions comprising the dorsolateral prefrontal cortex (dlPFC), the ventro- lateral prefrontal cortex (VLPFC) and the posterior par- ietal cortex (PPC). This network has been implicated in top-down cognitive-control functions, such as the games; 14 h or more spent spent on internet gaming maintenance and manipulation of information in work- on internet gaming per online time on internet week (with at least 2 h more than half of the more than 50 in IAT IGD assessment ing memory, as well as the cognitive control of thoughts, emotions and behaviours, in rule-based problem-solving and decision-making in goal-directed every day) behaviours (Menon 2011; Cole et al. 2013; Figure 1). Some studies have explored specific roles for the ECN in IGD through ICA. Ma et al. (2019) explored, through ICA, the activity of temporally coherent, large- 92 individuals with IGD; 96 scale functional networks during cue reactivity in a individuals with TUD\; 39 males with IGD; 34 sample of 29 male individuals with IGD and 23 control Sample participants. The authors examined, among others, the ECN (especially dlPFC, middle and inferior frontal cortex, male HCs 107 HCs middle temporal gyrus and inferior parietal lobe), high- lighting an association between ECN and IGD severity use disorders; YDQ: Young Diagnostic Questionnaire for Internet addiction criteria. (mainly hours spent gaming). The findings suggest that TUD and HC groups through intra- network interactions of large-scale to make comparisons among IGD, relevant components of the brain networks in IGD, and to associate individuals with IGD who spend more time gaming may the alterations with IGD-related To assess alterations in the inter- have more memories of previous gaming experiences ICA was used to determine the network analysis and inter- when exposed to internet-gaming cues or may show a greater identification with the virtual world given that Aims network analysis. regions of the ECN have been implicated in theory of mind and self-other distinctions. behaviours. Ma et al. (2021) described less engagement of the left and right fronto-parietal networks during post- error slowing in individuals with IGD. Both the left and right fronto-parietal networks may be involved in dif- Country ferent dimensions of cognitive control. Therefore, China China decreased activity in these networks may imply that individuals with IGD have difficulty using them to Table 1. Continued. adjust response strategies to balance the opposing Zhang et al. (2017) demands of stop-and-go trials and inhibit motor Zeng et al. (2022) responses after errors. Wang, Zhang et al. (2018) in their ICA study of 18 Reference male individuals with IGD and 19 control participants identified differences between both groups with 20 G. MESTRE-BACH ET AL. Figure 1. Main brain areas of the different networks. Executive-control network. dlPFC: dorsolateral prefrontal cortex; vlPFC: ventrolateral prefrontal cortex; dmPFC: dorsomedial pre- frontal cortex. Salience network. IPL: inferior parietal lobe; dACC: dorsal anterior cingulate cortex; Ins: insula. Default mode network. IPL: inferior parietal lobe; mPFC: medial prefrontal cortex; PCC: posterior cingulate cortex. respect to functional networks, especially in the right towards the new options. Yip et al. (2018) identified ECN during an addiction-related Stroop task. through ICA a decreased engagement of a fronto-cin- Specifically, IGD individuals, compared to control par- gulo-parietal network in individuals with IGD, in com- ticipants, presented different functional connectivity in parison to controls. this network, with increased connectivity involving the middle temporal gyrus, superior temporal gyrus and Salience network middle frontal gyrus. This finding may suggest that The SN mainly comprises the anterior insula, the individuals with IGD present abnormally enhanced dorsal anterior cingulate cortex (dACC) and the fronto- cognitive-control processing towards gaming-related insular cortex (FIC). Moreover, the amygdala and stri- cues, perhaps to try to exert cognitive control over atum may contribute to this network (Menon and responses. In this vein, the ICA results by Wang et al. Uddin 2010; Borsook et al. 2013). The SN is involved in (2019), who explored ECN in 18 male individuals with the detection of salient stimuli, both internal and IGD and 20 males with recreational internet game use, external, to direct behaviour and maintain homeosta- showed enhanced functional connectivity in the ECN sis (Toga 2015). The insula is responsible for detecting in individuals with IGD, in comparison with the recre- salient stimuli and initiating opportune control proc- ational-game-use group in probability and certain con- essing in response (Craig 2009; Menon and Uddin ditions. However, individuals with IGD had difficulty 2010). The SN also contributes to switching between controlling their risky behaviours, tending to prefer the other two networks mentioned (DMN and ECN), lower probability and higher risky choices, possibly thus facilitating orientation towards external versus due to their higher levels of impulsivity. Similarly, by internal stimuli (Sridharan et al. 2008; Figure 1). using ICA and a delay-discounting task, Wang et al. Regarding IGD, it has been suggested that individu- (2017) observed that individuals with IGD exhibited als with this disorder show reduced functional con- greater functional connectivity of the ECN, compared nectivity between the dACC and other brain areas to control participants when selecting options related within the SN (Chun et al. 2020). Furthermore, in IGD to now (versus later). Thus, individuals with IGD may versus control participants, a positive association have experienced greater demands on the behavioural between reward sensitivity and effective connectivity inhibition system, demonstrating an additional cogni- in the ventral striatum of the SN has been observed tive effort to select among possible options. However, (Chun et al. 2020). Other authors have highlighted the the increased functional connectivity of the ECN in role of the SN in error processing, both in individuals these individuals could not inhibit their impulses, with IGD and in control participants, given that both given that they typically exhibit difficulties in behav- groups seem to activate the bilateral insula and the ioural control, so they showed a greater tendency anterior cingulate cortex (Ko et al. 2014). DIALOGUES IN CLINICAL NEUROSCIENCE 21 Some studies have explored the specific role of the ICA to interrogate data from IGD participants from SN in IGD using ICA. Lee et al. (2020) explored, using multiple other jurisdictions are needed to address pos- ICA, the SN, taking into account the bilateral anterior sible cultural influences and other factors that may insular cortex. In a sample of 17 male adolescents with limit the generalisability of the findings that exist to IGD and 18 control participants, the authors observed date. Fifth, most studies have used the Internet addic- aberrant functional connectivity of the SN with the left tion test (IAT) (Young 2009) to assess IGD, but it is not pSTS only in adolescents with IGD. Furthermore, a specific psychometric tool for IGD. Similarly, most functional connectivity between the SN and pSTS was studies use the nine diagnostic criteria for IGD that associated with proneness to internet addiction and were suggested in the DSM-5 (American Psychiatric self-reported cognitive problems. The pSTS is a rele- Association 2013). Future studies should also use the vant part of the social brain network. Therefore, exces- diagnostic criteria accepted by ICD-11 (World Health sive use of social stimuli linked to video games may Organization 2020). Sixth, the cross-sectional design is alter the interaction between the SN and the social prevalent in included studies, so it is not possible to brain network (generating stronger functional connect- determine whether the alterations in brain connectiv- ivity between both networks), consequently possibly ity observed in individuals with IGD are a consequence producing cognitive problems and executive altera- of the disorder or a predisposing factor. Therefore, tions. In this vein, Wang et al. (2019) explored by ICA longitudinal studies exploring these factors at different the anterior SN in 18 male individuals with IGD and 20 developmental stages of participating individuals are males with recreational internet game use. Differential needed. Seventh, numerous included studies do not connectivity of the anterior SN was identified in indi- examine with precision the presence of psychiatric viduals with IGD, and this may be related to alterations comorbidities, which could be influencing the in the regulation of risky behaviour and, consequently, described findings. Finally, task-based studies are sub- to impaired cognitive control in IGD. Xing et al. (2014) ject to possible biases attributable to behavioural observed a decreased fractional anisotropy in the SN, assessments, which should be taken into account specifically in ACC/right-insula tracts, in individuals when interpreting the results. Moreover, many of the with IGD, in comparison with control participants. Yuan tasks differed among studies, and there appears to be et al. (2016) found using ICA an interaction between no uniformly agreed-upon approach for studying the the SN and right ECN (an influence of the ACC to neural correlates of IGD. Thus, while ICA studies have rDLPFC) during resting state, reduced in the 28 individ- advanced the understanding of the neurobiology of uals with IGD, in comparison with control participants. IGD, additional research is needed. Finally, Zhang et al. (2017) explored the three networks together and found increased SN-DMN connectivity in individuals with IGD. Conclusion The DMN, ECN and SN have been explored using ICA Limitations and future research in individuals with IGD. Most studies have identified alterations in these three brain networks in individuals The studies using ICA included in this narrative review with IGD, and these networks may be involved in the have several limitations that should be considered. development and maintenance of IGD. However, more First, the sample sizes of several investigations were studies are needed in order to deepen an understand- relatively small, which may weaken the rigour of the ing of the specific roles of each in the symptomatol- results. Therefore, studies with larger sample sizes are ogy and treatment of IGD. needed to draw stronger conclusions. Second, most studies include exclusively males with IGD. Although IGD appears to be a more prevalent disorder in males, Disclosure statement further studies examining the neurobiology of IGD in The authors report no conflicts of interest with the topic of females, as well as possible gender-related differences, this work. Marc N. Potenza has consulted for and advised are needed. Third, studies often included individuals Opiant Pharmaceuticals, Idorsia Pharmaceuticals, Baria-Tek, with IGD recruited from educational centres, so the AXA, Game Day Data and the Addiction Policy Forum; has results may not generalise to broader populations. been involved in a patent application with Yale University and Novartis; has received research support from the More studies focussed on treatment-seeking individu- Mohegan Sun Casino and Connecticut Council on Problem als with IGD are needed. Fourth, all studies included Gambling; has participated in surveys, mailings or telephone in the present narrative review have been conducted con-sultations related to drug addiction, impulse control dis- in Asia, mostly in China. Therefore, more studies using orders or other health topics; and has consulted for law 22 G. MESTRE-BACH ET AL. offices and gambling entities on issues related to impulse Affect-Cognition-Execution (I-PACE) model. Neurosci control or addictive disorders. Fernando Fernandez-Aranda Biobehav Rev. 71:252–266. and Susana Jimenez-Murcia received consultancy honoraria Brandt CL, Doan NT, Tønnesen S, Agartz I, Hugdahl K, Melle from Novo Nordisk, and Fernando Fernandez-Aranda editor- I, Andreassen OA, Westlye LT. 2015. Assessing brain struc- ial honoraria as EIC from Wiley. No potential conflict of inter- tural associations with working-memory related brain pat- est was reported by the rest of the authors. terns in schizophrenia and healthy controls using linked independent component analysis. Neuroimage Clin. 9: 253–263. Funding Buckner RL, DiNicola LM. 2019. The brain’s default network: CERCA Programme / Generalitat de Catalunya gave institu- updated anatomy, physiology and evolving insights. Nat tional support. This work was additionally supported by two Rev Neurosci. 20(10):593–608. grants from the Delegacio n del Gobierno para el Plan Chun JW, Park CH, Kim JY, Choi J, Cho H, Jung DJ, Ahn KJ, Nacional sobre Drogas (2019I47 and 2021I031), and Instituto Choi JS, Kim DJ, Choi IY. 2020. Altered core networks of de Salud Carlos III (ISCIII) (P I20/00132) and cofunded by brain connectivity and personality traits in internet gam- FEDER funds/European Regional Development Fund (ERDF), a ing disorder. J Behav Addict. 9(2):298–311. way to build Europe. CIBERobn is an initiative of ISCIII. Dr. Cole MW, Reynolds JR, Power JD, Repovs G, Anticevic A, Gemma Mestre-Bach was supported by a postdoctoral grant Braver TS. 2013. Multi-task connectivity reveals flexible of FUNCIVA and by the 2022–2023 Institute for Culture and hubs for adaptive task control. Nat Neurosci. 16(9):1348– Society (ICS) challenge on “Youth, relationships and psycho- 1355. logical well-being” of the University of Navarra. Roser Granero Craig AD. 2009. How do you feel – now? The anterior insula was supported by The Catalan Institution for Research and and human awareness. Nat Rev Neurosci. 10:59–70. Di X, Biswal BB. 2014. Identifying the default mode network Advanced Studies (ICREA-2021 Academia Program). Dr. Marc structure using dynamic causal modeling on resting-state N. Potenza’s involvement was supported by the Connecticut functional magnetic resonance imaging. Neuroimage. 86: Council on Problem Gambling and by the NIMH (National 53–59. Institute of Mental Health) grant RF1 MH128614. Dong G, Li H, Wang L, Potenza MN. 2017. The correlation between mood states and functional connectivity within ORCID the default mode network can differentiate Internet gam- ing disorder from healthy controls. Prog Gemma Mestre-Bach http://orcid.org/0000-0001-5345- Neuropsychopharmacol Biol Psychiatry. 77:185–193. 0484 Dong GH, Wang Z, Dong H, Wang M, Zheng Y, Ye S, Zhang Roser Granero http://orcid.org/0000-0001-6308-3198 J, Potenza MN. 2020. More stringent criteria are needed Fernando Fernandez-Aranda http://orcid.org/0000-0002- for diagnosing internet gaming disorder: evidence from 2968-9898 regional brain features and whole-brain functional con- Susana Jimenez-Murcia http://orcid.org/0000-0002-3596- nectivity multivariate pattern analyses. J Behav Addict. 8033 9(3):642–653. Marc N. Potenza http://orcid.org/0000-0002-6323-1354 Garrison K, Sinha R, Potenza M, Gao S, Liang Q, Lacadie C, Scheinost D. 2021. Transdiagnostic, connectome-based References prediction of craving. Am J Psychiatry. Gupta CN, Castro E, Rachkonda S, van Erp TGM, Potkin S, American Psychiatric Association. 2013. Diagnostic and statis- Ford JM, Mathalon D, Lee HJ, Mueller BA, Greve DN, et al. tical manual of mental disorders. 5th ed. (DSM-5). 2017. Biclustered independent component analysis for Washington (DC): American Psychiatric Association. complex biomarker and subtype identification from struc- Bin ZY, Dong HH, Wang M, Zhou W, Lin X, Dong GH. 2022. tural magnetic resonance images in schizophrenia. Front Similarities and differences between internet gaming dis- Psychiatry. 8:179–10. order and tobacco use disorder: a large-scale network Hyv€arinen A. 2013. Independent component analysis: recent study. Addict Biol. 27(2):1–13. advances subject areas. Philos Trans A Math Phys Eng Sci. Borsook D, Edwards R, Elman I, Becerra L, Levine J. 2013. 371(1984):20110534. Pain and analgesia: the value of salience circuits. Prog Kashif S, Pandey S, Warriach ZI. 2021. Neurophysiological markers of internet gaming disorder: a literature review of Neurobiol. 104:93–105. €lfling K, Robbins €ller A, Wo electroencephalography studies. Cureus. 13(9):e17866. Brand M, Wegmann E, Stark R, Mu Ko CH, Hsieh TJ, Chen CY, Yen CF, Chen CS, Yen JY, Wang TW, Potenza MN. 2019. The Interaction of Person- PW, Liu GC. 2014. Altered brain activation during response Affect-Cognition-Execution (I-PACE) model for addictive inhibition and error processing in subjects with Internet behaviors: update, generalization to addictive behaviors gaming disorder: a functional magnetic imaging study. beyond internet-use disorders, and specification of the Eur Arch Psychiatry Clin Neurosci. 264(8):661–672. process character of addictive behaviors. Neurosci Lee J, Lee D, Namkoong K, Jung YC. 2020. Aberrant posterior Biobehav Rev. 104:1–10. superior temporal sulcus functional connectivity and Brand M, Young KS, Laier C, Wo €lfling K, Potenza MN. 2016. executive dysfunction in adolescents with internet gaming Integrating psychological and neurobiological considera- disorder. J Behav Addict. 9(3):589–597. tions regarding the development and maintenance of Liu G, Jiao K, Zhong Y, Hao Z, Wang C, Xu H, Teng C, Song specific Internet-use disorders: an Interaction of Person- X, Xiao C, Fox PT, et al. 2021. The alteration of cognitive DIALOGUES IN CLINICAL NEUROSCIENCE 23 function networks in remitted patients with major depres- Wang L, Zhang Y, Lin X, Zhou H, Du X, Dong G. 2018. Group sive disorder: an independent component analysis. Behav independent component analysis reveals alternation of Brain Res. 400:113018. right executive control network in Internet gaming dis- Liu L, Potenza MN, Lacadie CM, Zhang JT, Yip SW, Xia CC, order. CNS Spectr. 23(5):300–310. Lan J, Yao YW, Deng LY, Park SQ, et al. 2021. Altered Wang W, Worhunsky PD, Zhang S, Le TM, Potenza MN, Li intrinsic connectivity distribution in internet gaming dis- CSR. 2018. Response inhibition and fronto-striatal-thalamic order and its associations with psychotherapy treatment circuit dysfunction in cocaine addiction. Drug Alcohol outcomes. Addict Biol. 26(2):1–12. Depend. 192:137–145. Luo L, Lei X, Zhu C, Wu J, Ren H, Zhan J, Qin Y. 2022. Wang Y, Wu L, Zhou H, Lin X, Zhang Y, Du X, Dong G. 2017. Decreased connectivity in precuneus of the ventral atten- Impaired executive control and reward circuit in Internet tional network in first-episode, treatment-naïve patients gaming addicts under a delay discounting task: independ- with major depressive disorder: a network homogeneity ent component analysis. Eur Arch Psychiatry Clin Neurosci. and independent component analysis. Front Psychiatry. 267(3):245–255. 13:925253–925258. Wang Z, Liu X, Hu Y, Zheng H, Du X, Dong G. 2019. Altered Ma SS, Li CSR, Zhang S, Worhunsky PD, Zhou N, Zhang JT, brain functional networks in Internet gaming disorder: Liu L, Yao YW, Fang XY. 2021. Altered functional network independent component and graph theoretical analysis activities for behavioral adjustments and Bayesian learning under a probability discounting task. CNS Spectr. 24(5): in young men with Internet gaming disorder. J Behav 544–556. Addict. 10(1):112–122. Worhunsky PD, Angarita GA, Zhai ZW, Matuskey D, Gallezot Ma SS, Worhunsky PD, Xu JS, Yip SW, Zhou N, Zhang JT, Liu JD, Malison RT, Carson RE, Potenza MN. 2021. Multimodal L, Wang LJ, Liu B, Yao YW, et al. 2019. Alterations in func- investigation of dopamine D2/D3 receptors, default mode tional networks during cue-reactivity in Internet gaming network suppression, and cognitive control in cocaine-use disorder. J Behav Addict. 8(2):277–287. disorder. Neuropsychopharmacology. 46(2):316–324. Maglanoc LA, Kaufmann T, Jonassen R, Hilland E, Beck D, World Health Organization. 2020. International classification Landrø NI, Westlye LT. 2020. Multimodal fusion of struc- of diseases 11th revision. Geneva (Switzerland): WHO. tural and functional brain imaging in depression using Xing L, Yuan K, Bi Y, Yin J, Cai C, Feng D, Li Y, Song M, linked independent component analysis. Hum Brain Wang H, Yu D, et al. 2014. Reduced fiber integrity and Mapp. 41(1):241–255. cognitive control in adolescents with internet gaming dis- McKeown MJ, Hansen LK, Sejnowsk TJ. 2003. Independent order. Brain Res. 1586:109–117. component analysis of functional MRI: what is signal and Yan H, Li Q, Yu K, Zhao G. 2021. Large-scale network dys- what is noise? Curr Opin Neurobiol. 13(5):620–629. function in youths with Internet gaming disorder: a meta- Menon V. 2011. Large-scale brain networks and psychopath- analysis of resting-state functional connectivity studies. ology: a unifying triple network model. Trends Cogn Sci. Prog Neuropsychopharmacol Biol Psychiatry. 109:110242. 15(10):483–506. Yip SW, Gross JJ, Chawla M, Ma SS, Shi XH, Liu L, Yao YW, Zhu Menon V, Uddin LQ. 2010. Saliency, switching, attention and L, Worhunsky PD, Zhang J. 2018. Is neural processing of control: a network model of insula function. Brain Struct Funct. 214(5–6):655–667. negative stimuli altered in addiction independent of drug Ni MF, Zhang BW, Chang Y, Huang XF, Wang XM. 2021. effects? Findings from drug-naïve youth with Internet gam- Altered resting-state network connectivity in panic dis- ing disorder. Neuropsychopharmacology. 43(6):1364–1372. order: an independent component analysis. Brain Imaging Yip SW, Mackay CE, Goodwin GM. 2014. Increased temporo- Behav. 15(3):1313–1322. insular engagement in unmedicated bipolar II disorder: an Song KR, Potenza MN, Fang XY, Gong GL, Yao YW, Wang ZL, exploratory resting state study using independent compo- Liu L, Ma SS, Xia CC, Lan J, et al. 2021. Resting-state con- nent analysis. Bipolar Disord. 16(7):748–755. nectome-based support-vector-machine predictive model- Young K. 2009. Internet addiction: the emergence of a new ing of internet gaming disorder. Addict Biol. 26(4):1–10. clinical disorder. CyberPsychology Behav. 1(3):237–244. Sridharan D, Levitin DJ, Menon V. 2008. A critical role for the Yuan K, Qin W, Yu D, Bi Y, Xing L, Jin C, Tian J. 2016. Core right fronto-insular cortex in switching between central- brain networks interactions and cognitive control in executive and default-mode networks. Proc Natl Acad Sci Internet gaming disorder individuals in late adolescen- USA. 105(34):12569–12574. ce/early adulthood. Brain Struct Funct. 221(3):1427–1442. Strimbu K, Tavel JA. 2010. What are biomarkers? Curr Opin Zeng N, Wang M, Dong H, Du X, Dong GH. 2022. Altered HIV Aids. 5(6):463–466. dynamic interactions within frontostriatal circuits reflect Tang F, Yang H, Li L, Ji E, Fu Z, Zhang Z. 2020. Fusion analysis disturbed craving processing in internet gaming disorder. of gray matter and white matter in bipolar disorder by CNS Spectr. 27(1):109–117. multimodal CCA-joint ICA. J Affect Disord. 263(1066):80–88. Zhang S, Hu S, Sinha R, Potenza MN, Malison RT, Li C, Shan Toga AW. 2015. Brain mapping: an encyclopedic reference. R. 2016. Cocaine dependence and thalamic functional Washington (DC): Academic Press. connectivity: a multivariate pattern analysis. Neuroimage Wang L, Wu L, Lin X, Zhang Y, Zhou H, Du X, Dong G. 2016. Clin. 12:348–358. Dysfunctional default mode network and executive con- Zhang JT, Ma SS, Yan CG, Zhang S, Liu L, Wang LJ, Liu B, trol network in people with Internet gaming disorder: Yao YW, Yang YH, Fang XY. 2017. Altered coupling of independent component analysis under a probability dis- default-mode, executive-control and salience networks in counting task. Eur Psychiatry. 34:36–42. Internet gaming disorder. Eur Psychiatry. 45:114–120.