Differential resting-state neurophysiological activity associated with game usage patterns and genres in Internet gaming disorder
Funding information: This research was funded by National Research Foundation of Korea, grant number 2021R1F1A1046081.
Abstract
We investigated differences in quantitative electroencephalogram (EEG) patterns associated with game usage patterns and genre among patients with Internet gaming disorder (IGD). Data from 140 participants (76 IGD patients and 64 healthy controls) were analysed. The IGD group was divided into subgroups based on game usage patterns (single game [SG] or multiple games [MGs]) and genre (multiplayer online battle arena, first-person shooter [FPS], or massively multiplayer online role-playing game [MMORPG; hereafter, MMG]). A resting-state, eye-closed quantitative EEG was recorded, and the absolute power and coherence of brain waves were analysed. IGD patients who played SGs showed increased beta activity compared with those who played MGs and controls. Increased absolute beta power was significantly associated with higher tendencies towards behavioural inhibition compared with controls. FPS gamers showed increased delta power in the frontal region compared with controls, which was related to the severity of IGD. Furthermore, decreased intrahemispheric coherence in the left frontoparietal region was observed in the MMG and FPS groups compared with controls. This decreased coherence was observed in the theta (MMG and FPS), delta (MMG), and beta (FPS) bands. These features were related to impairment in visuospatial working memory. Unique neurophysiological features related to preoccupation with an SG may be associated with the inhibition of behavioural changes. The present study suggests that the underlying neurophysiological networks in IGD differ according to game usage patterns and genre.
1 INTRODUCTION
Internet gaming disorder (IGD) is characterized by impaired functioning due to repeated use of Internet-based games. According to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), IGD is a behavioural addiction and requires further clinical research before being labelled a mental disorder.1 In addition, the World Health Organization recently included gaming disorder in the International Classification of Diseases because of its potential for addiction.2 During the coronavirus disease-2019 pandemic, people have been advised to isolate at home, which has led to a marked increase in the prevalence of IGD.3
There is growing interest in IGD, and many studies have explored the neurophysiological characteristics of this disorder. Some studies have focused on the resting-state brain activity of IGD patients using electroencephalography. Electroencephalography has several advantages over other neuroimaging techniques, including high temporal resolution, noninvasiveness, and a significantly low cost; thus, it is widely used to investigate the neurophysiological features of various disorders, including IGD. The two most frequently used methods of analysing an electroencephalogram (EEG) are power spectral and coherence analyses. Studies using power spectral analysis have reported that individuals with IGD have higher absolute power in the delta and theta bands4-6 and lower absolute power in the beta band.4, 6 Furthermore, altered EEG activity in the delta and beta bands is associated with the degree of impulsivity and the severity of IGD.4 Based on these findings, researchers have suggested that increased slow-frequency and decreased high-frequency activity are neurophysiological markers of IGD.7 In addition, coherence analysis has been used to study the functional interaction and connectivity between distinct regions of the brain. Several studies have used EEG coherence analyses to investigate the neural mechanisms underlying IGD. One study8 found higher intrahemispheric gamma and beta coherence in IGD patients compared with patients with alcohol use disorder. Another study9 also showed that IGD patients exhibited increased intrahemispheric gamma and beta coherence compared with healthy controls. These results suggest that increased fast-frequency coherence may be a neurophysiological marker of IGD. Recent studies have attempted to classify neurophysiological features in IGD patients according to coexisting symptoms (e.g., depressive symptoms and attentional problems) or psychological characteristics (e.g., resilience). For example, one study10 reported that IGD patients with low resilience had higher alpha coherence in the right hemisphere than those with high resilience, which suggests that the brain activity of IGD patients may differ according to individual characteristics.
The game genre may also affect the brain activity of IGD patients. Elements of contemporary Internet games, such as the game environment, playing method, and degree of social interaction with other users, vary by genre. There is accumulating evidence that the characteristics of gamers and the effects of gaming vary significantly by genre; therefore, different game genres require separate investigation.11-15 For example, massively multiplayer online role-playing games (MMORPGs; hereafter, MMGs), such as World of Warcraft and MapleStory, are some of the most popular games worldwide. In MMGs, gamers primarily develop and manage characters (or “avatars”) and complete various tasks (or “quests”) either alone or together with other gamers in an online virtual world.16 MMGs enable simultaneous interaction among thousands of gamers; thus, socialization plays an important role in these games. Their endless and highly social nature distinguishes MMGs from other game genres.
Two relatively little studied online games are first-person shooter (FPS; e.g., Dungeon and Fighter) and multiplayer online battle arena (MOBA; e.g., League of Legends) games. FPS games are included in the action genre and differ significantly from MMGs. These games depict three-dimensional environments viewed through the eyes of the character (i.e., a first-person view), with usually only the weapon being visible, and consist of fast-paced and short-duration individual rounds. In each round, gamers form teams to engage in combat with enemies and aim to complete tasks before their enemies. The shoot-to-kill style and visual display of FPS games are highly arousing and violent. MOBA games are a relatively new genre derived from real-time strategy games and are becoming increasingly popular worldwide because of the popularity of League of Legends. In MOBA games, gamers compete in real time in a virtual arena to obtain resources, confront other gamers, and defend their base.16 Unlike MMGs, which are based on an endless virtual world, MOBA games consist of shorter and more intensive game sessions. MOBA games, similar to FPS games, consist of individual sessions that can be played as a team; the sessions are longer and less violent than the usual FPS games. It has been argued that these genre-specific features differentially affect the behaviours of gamers; therefore, game genre should be considered in studies of the effects of games,17 and neural correlates of genre-specific differences in IGD patients should be investigated.
An important yet poorly studied factor that can influence IGD is game usage patterns. In particular, differences between gamers who play single games (SGs) or multiple games (MGs) are of interest. Internet games are dynamic cognitive tasks that require gamers to respond to various complex challenges in a virtual environment. Therefore, playing multiple different games requires the ability to switch cognitive sets and rapidly adapt to various cognitive tasks. One hypothesis is that addicted gamers who play MGs may have a higher level of cognitive flexibility than those who play SGs. Conversely, pathological preoccupation with one game may be associated with difficulties in shifting attention and cognitive sets. Another possibility is that playing MGs may signify impulsivity or sensation seeking. Addicted gamers may choose to alternate between MGs because they are easily fed up with performing the same tasks and seek novel, stimulating sensations. Therefore, it is possible that addiction to MGs may be related to deficits in inhibitory function and sensation-seeking tendencies. Insights into the neural correlates and/or contributors associated with playing SGs or MGs help in understanding the heterogeneity among individuals with IGD.
However, to the best of our knowledge, no studies have investigated differences within IGD patients. Although a few studies have investigated genre-specific differences, they have focused exclusively on sociodemographic and psychological features14, 15, 18; neurophysiological differences by genre and game usage patterns have yet to be explored. We observed neurophysiological similarities and differences among IGD patients according to game genre and usage patterns and compared them with the neurophysiological features of healthy controls (HCs). Data from the resting-state EEGs of treatment-seeking IGD patients and HCs were analysed using power spectral and coherence analyses. In addition, psychological and neurocognitive data were collected to investigate their relationships with neurophysiological features.
2 MATERIALS AND METHODS
2.1 Participants
We recruited young males from the outpatient clinic of SMG-SNU Boramae Medical Center, South Korea, and the local community. The IGD group consisted of individuals diagnosed by an experienced psychiatrist based on DSM-5 criteria. HCs were recruited from local community and universities by advertisements. They were required to have no history of psychiatric disorders, and play Internet games for less than 2 h per day. All participants were medication naïve and right-handed, with no history of psychotic or neurological disorders, significant head injury, seizures, or intellectual disability (intelligence quotient [IQ] ≤ 80). Data from 140 participants were used in the present study: 76 were patients with IGD (age: 25.04 ± 3.76) and 64 were healthy controls (age: 24.67 ± 3.44). EEG recordings disrupted by excessive movement or poor quality were not included in the final analyses (>100 μV2).
The IGD group was divided into subgroups based on game usage patterns (SG or MG) and genre (MOBA, FPS, or MMG). Data on game usage patterns were collected with the following question: “How many Internet games do you play?” Participants who responded with “I only play a single game for more than 6 months” were classified as SG users, and those who responded with “I play multiple games” were classified as MG users. After we excluded missing values, data from 72 IGD patients and 64 HCs were included in the analyses. Of the 72 IGD patients, 50% (n = 36) were SG users and 50% (n = 36) were MG users. The favourite Internet game of the participants was used to classify the participants by game genre. Based on the 2020 White Paper on Korean Games, a comprehensive summary of worldwide publications related to Internet games, Internet games were classified into three genres: MOBA, FPS, and MMG. After we excluded missing values and unclassified responses (n = 12), 59 IGD patients were included in the genre analyses: 57.6% (n = 34), 25.4% (n = 15), and 16.9% (n = 10) used MOBA, FPS, and MMG, respectively.
2.2 Psychological assessments
2.2.1 Young Internet Addiction Test (Y-IAT)
The Y-IAT was used to assess the severity of IGD.19, 20 It includes 20 items rated on 5-point scales, with higher scores reflecting higher levels of problematic use of Internet games. The total score ranges from 20 to 100 (Cronbach's α = 0.961).
2.2.2 Beck Depression Inventory-2 (BDI-II)
The BDI-II consists of 21 items that measure the severity of depressive symptoms experienced over the previous 2 weeks.21, 22 Items are scored on a 4-point Likert scale, with higher scores indicating more severe depressive symptoms (Cronbach's α = 0.947).
2.2.3 Beck Anxiety Inventory (BAI)
The BAI is a 21-item self-report measure designed to assess symptoms of anxiety during the previous week.23, 24 Items are scored on a 4-point Likert scale, with higher scores indicating more severe symptoms of anxiety (Cronbach's α = 0.944).
2.2.4 Behavioral Activation System (BAS) and Behavioral Inhibition System (BIS)
Dispositional sensitivity to rewards and punishments was evaluated with the BAS and BIS scales.25 The BAS and BIS contain 13 and 7 items, respectively, that are scored on a 4-point Likert scale. The BAS includes three subscales that evaluate fun seeking, reward responsiveness, and drive (Cronbach's α = 0.863).
2.2.5 Aggression Questionnaire (AQ)
The AQ consists of 29 items that assess tendencies toward aggression, including verbal and physical aggression, anger, and hostility.26 Items are scored on a 5-point Likert scale, with higher scores indicating higher levels of aggression (Cronbach's α = 0.934).
2.2.6 Barratt Impulsiveness Scale-11 (BIS-11)
The BIS-1127 was used to measure impulsiveness. It consists of 23 items that assess the degree of impulsivity on a 4-point Likert scale and includes three subscales: cognitive, motor, and nonplanning impulsiveness (Cronbach's α = 0.859).
2.2.7 Korean version of the Wechsler Adult Intelligence Scale
The Korean version of the Wechsler Adult Intelligence Scale, fourth edition,28 was used to measure IQ.
2.3 Neurocognitive assessments
Three neurocognitive tests were conducted to examine group differences in cognitive functioning and their relationship with EEG data. The Stroop Word and Color test29 was used to assess the ability to inhibit cognitive interference and selective attention. Verbal fluency was assessed with letter (e.g., retrieval of words that begin with a letter of the Korean alphabet) and category (e.g., retrieval of words from the category “animals”) fluency tasks.30 The Spatial Working Memory (SWM) section of the Cambridge Neuropsychological Test Automated Battery (CANTAB) was administered to measure the ability to maintain and manipulate visuospatial information (see <http://www.camcog.com> for details).
2.4 EEG data
2.4.1 EEG recording
Each participant was seated on a comfortable chair in an isolated sound-shielded room with dimmed lights, and then underwent EEG recording in a resting state that lasted for 10 min (5 min with eyes closed, 5 min with eyes open). All participants were instructed to avoid moving or becoming drowsy. All EEG activity was recorded using a 64-channel Quik-cap (Compumedics Neuroscan, El Paso, TX, USA) based on the modified international 10/20 system, in conjunction with vertical and horizontal electrooculograms (EOGs) and one bipolar reference electrode connected to the mastoid. All EEG acquisitions were done using SynAmps 2 (Compumedics, Abbotsford, Australia) and the Neuroscan system (Scan 4.5; Compumedics). EEG signals were amplified at a sampling rate of 1000 Hz using a 0.1- to 100-Hz online bandpass filter and a 0.1- to 50-Hz offline bandpass filter, while electrode impedance was kept below 5 kΩ. All acquired EEG data were processed with NeuroGuide software (ver. 2.6.1; Applied Neuroscience, St. Petersburg, FL, USA); 19 sites of 64 channels were driven by the following montage set of NeuroGuide software: FP1, F3, F7, Fz, FP2, F4, F8, T3, C3, Cz, T4, C4, T5, P3, O1, Pz, T6, P4, and O2. EEG recordings were examined to eliminate artefacts by using NeuroGuide software and were also visually inspected to remove eye muscle movements and other artefacts. Accepted epochs of EEG data for absolute (μV2) power were smoothed using fast Fourier transforms and averaged in five frequency bands by NeuroGuide's spectral analysis system: delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–30 Hz), and gamma (30–40 Hz). The activity at 19 sites was divided into three regions by averaging within each region: Frontal (FP1, F3, F7, Fz, FP2, F4, and F8), central (T3, C3, Cz, T4, and C4), and posterior (T5, P3, O1, Pz, T6, P4, and O2). To determine the variation in absolute power by hemisphere, the EEG signals were categorized into the left (FP1, F3, F7, T3, C3, T5, P3, and O1) and right hemisphere (FP2, F4, F8, T4, C4, T6, P4, and O2). Visualizations were performed with MATLAB software (Math-Works, Natick, MA, USA).
2.4.2 Coherence analyses
Coherence values were computed for all pairwise combinations of the 19 channels, for each of the five frequency bands (delta, theta, alpha, beta, gamma), using NeuroGuide software. The computation of coherence was based on previous studies.8, 31 A total of 171 intrahemispheric and interhemispheric pairwise combinations of electrodes were obtained and in and intrahemispheric coherence was examined using the F3-C3, F3-T3, F3-P3, C3-T3, C3-P3, and T3-P3 electrode pairs in the left hemisphere and the F4-C4, F4-T4, F4-P4, C4-T4, C4-P4, and T4-P4 electrode pairs in the right hemisphere. Interhemispheric coherence was calculated between electrode pairs F3-F4, C3-C4, T3-T4, and P3-P4.
2.5 Statistical analyses
We compared demographic, clinical, and neurocognitive variables among the groups ([1] SG, MG, and HC and [2] MOBA, FPS, and MMG) using one-way analysis of variance. Next, we used a generalized estimating equation (GEE) to analyse the EEG characteristics in each band. GEEs were used to consider unknown and possible correlations between repeated or multiple outcomes from the same subjects. In the absolute power analyses, group, region (frontal, central, and posterior), hemisphere (left and right), and their interaction effects were tested in each band with GEEs. In the coherence analyses, intrahemispheric and interhemispheric coherence values were assessed according to the following factors: (1) Intrahemispheric coherence: Group × region (frontocentral, frontotemporal, frontoparietal, centrotemporal, centroparietal, or temporoparietal) × hemisphere (left or right) and (2) interhemispheric coherence: Group × region (frontal, central, temporal, or parietal). Years of education, IQ, and BDI and BAI values were adjusted for among the groups. Furthermore, Pearson's correlation analyses were used to explore associations between EEG activity that showed significant main or interaction effects in GEE analyses and clinical characteristics including IAT, BAS/BIS, AQ, and BIS-11. Associations between EEG activity and neurocognitive performance on the Stroop Word and Color test, Verbal Fluency, and the SWM were also explored. All statistical analyses were performed with SPSS (version 26.0; IBM, Armonk, NY, USA). Correlations were visualized with R (version 3.2.3; R Foundation for Statistical Computing, Vienna, Austria).
2.6 Ethics
The institutional review board of SMG-SNU Boramae Medical Center approved the study protocol, and the study followed the guidelines of the Declaration of Helsinki. All participants understood the study procedure and provided written informed consent.
3 RESULTS
3.1 Group differences by gaming usage patterns
3.1.1 Demographic, clinical, and neurocognitive assessments
Table 1 shows the demographic and clinical characteristics of participants in the SG, MG, and HC groups. Age did not differ significantly among the groups. The HCs had more years of education than the MG group, but there was no significant difference in age between the SG group and HCs. Both the SG and MG groups spent more weekday and weekend hours on Internet gaming than the HCs. IGD symptoms were most severe in the SG group, followed by the MG group and finally the HCs. Both IGD groups showed higher scores on the BDI, BAI, BIS, and BIS-11 and lower IQs compared with the HCs. The BAS score differed significantly between the SG group and HCs but not between the MG group and HCs. AQ scores were higher in both IGD groups compared with the HCs, and scores were higher in the SG group compared with the MG group.
| Single game users (N = 36) | Multiple game users (N = 36) | Healthy controls (N = 64) | F | P | Post hoc | |
|---|---|---|---|---|---|---|
| Demographic data | ||||||
| Age (years) | 25.00 (6.07) | 24.47 (4.42) | 24.67 (3.44) | 0.126 | 0.882 | |
| Education (years) | 13.60 (1.91) | 13.433 (1.69) | 14.59 (2.16) | 5.63** | 0.005 | H > M |
| Game usage in weekdays (h) | 4.87 (3.77) | 4.93 (3.39) | 0.43 (0.84) | 46.35*** | <0.001 | S, M > H |
| Game usage in weekends (h) | 6.43 (4.24) | 8.15 (10.37) | 0.57 (1.20) | 22.13*** | <0.001 | S, M > H |
| Clinical data | ||||||
| IQ | 107.19 (16.29) | 100.33 (16.69) | 115.91 (10.86) | 14.68*** | <0.001 | H > S, M |
| IAT | 62.14 (14.43) | 54.14 (16.93) | 29.94 (9.54) | 81.08*** | <0.001 | S > M > H |
| BDI | 16.25 (10.54) | 15.64 (10.35) | 4.25 (4.54) | 34.56*** | <0.001 | S, M > H |
| BAI | 13.36 (10.58) | 14.50 (12.53) | 3.52 (4.66) | 22.81*** | <0.001 | S, M > H |
| BAS | 36.58 (5.91) | 34.22 (7.11) | 32.44 (5.77) | 5.21** | 0.007 | S > H |
| BIS | 21.58 (4.35) | 21.03 (4.90) | 17.64 (3.58) | 13.27*** | <0.001 | S, M > H |
| AQ | 78.86 (19.49) | 68.64 (15.03) | 54.69 (11.90) | 31.35*** | <0.001 | S > M > H |
| BIS-11 | 66.61 (9.66) | 65.86 (8.27) | 56.38 (7.69) | 23.43*** | <0.001 | S, M > H |
| Neurocognitive data | ||||||
| Stroop word | 57.14 (8.24) | 56.22 (12.27) | 51.53 (7.99) | 5.21** | 0.007 | S > H |
| Stroop colour | 100.94 (23.76) | 100.69 (27.87) | 89.09 (16.25) | 4.87** | 0.009 | S, M > H |
| Verbal fluency (letter) | 40.79 (12.60) | 35.25 (11.28) | 48.61 (13.59) | 13.38*** | <0.001 | S, M < H |
| Verbal fluency (category) | 38.11 (11.61) | 34.28 (10.43) | 44.34 (8.60) | 12.69*** | <0.001 | S, M < H |
| SWM (totala) | 10.39 (13.77) | 11.72 (13.00) | 14.06 (12.46) | 0.73 | 0.482 | n.s. |
| SWM (strategyb) | 28.94 (5.94) | 28.78 (5.71) | 30.39 (5.47) | 0.983 | 0.377 | n.s. |
- Note: Data are presented as mean (standard deviation). Post hoc comparisons were carried out using Bonferroni correction. S, Single game users; M, Multiple game users; H, healthy controls; IQ, Intelligence Quotient; IAT, Internet Addiction Test; BDI, Beck Depression Index; BAI, Beck Anxiety Inventory; BAS, Behavioral Activation Scale; BIS, Behavioral Inhibition Scale; AQ; Aggression Questionnaire; BIS-11; Barratt Impulsiveness Scale 11; SWM, Spatial Working Memory; n.s., not significant.
- a Total number of errors.
- b Estimate of the use of the best strategy. High score represents poor use of this strategy; low score represents effective use.
- * P < 0.05.
- ** P < 0.01.
- *** P < 0.001.
Comparisons of neurocognitive variables among the SG, MG, and HC groups are presented in Table 1. Stroop word reaction times were significantly longer in the SG group compared with the HCs. Both the SG and MG groups showed longer Stroop colour reaction times and fewer responses on the verbal fluency tests compared with the HCs. No significant differences among the groups were observed in errors or strategy use on the SWM.
3.1.2 Absolute power
Figure 1 illustrates the scalp topography of each group in terms of the absolute power in each band. The GEE results showed significant group (χ2 = 6.269; p = 0.044) and group × region interaction effects for absolute power in the beta band (χ2 = 15.387; p = 0.004) after adjusting for education, IQ, depression, and anxiety symptoms. The SG group exhibited higher absolute power in the beta band than the HCs, in particular in the frontal region. Although there was a significant group × hemisphere interaction effect in the delta band (χ2 = 7.487; p = 0.024), there were no group differences. No main or interaction effects were observed for the theta, alpha, or gamma bands (Table 2).
| Absolute power | Wald χ2 | df | P | Post hoc |
|---|---|---|---|---|
| Delta | ||||
| Group | 1.775 | 2 | 0.412 | |
| Group × Region | 8.780 | 4 | 0.067 | |
| Group × Hemisphere | 7.487* | 2 | 0.024 | n.s. |
| Group × Region × Hemisphere | 8.188 | 4 | 0.085 | |
| Theta | ||||
| Group | .750 | 2 | 0.687 | |
| Group × Region | 7.360 | 4 | 0.118 | |
| Group × Hemisphere | 4.854 | 2 | 0.088 | |
| Group × Region × Hemisphere | 8.470 | 4 | 0.076 | |
| Alpha | ||||
| Group | 1.326 | 2 | 0.515 | |
| Group × Region | 2.997 | 4 | 0.558 | |
| Group × Hemisphere | 4.149 | 2 | 0.126 | |
| Group × Region × Hemisphere | 5.406 | 4 | 0.248 | |
| Beta | ||||
| Group | 6.269* | 2 | 0.044 | SG > HC |
| Group × Region | 15.387** | 4 | 0.004 | Frontal: SG > HC |
| Group × Hemisphere | 2.988 | 2 | 0.224 | |
| Group × Region × Hemisphere | 6.928 | 4 | 0.140 | |
| Gamma | ||||
| Group | 3.012 | 2 | 0.222 | |
| Group × Region | 5.551 | 4 | 0.235 | |
| Group × Hemisphere | .102 | 2 | 0.950 | |
| Group × Region × Hemisphere | 3.941 | 4 | 0.414 | |
- Note: Post hoc comparisons were carried out using Bonferroni correction. SG, Single game users; HC, healthy controls; n.s., not significant.
- * P < 0.05.
- ** P < 0.01.
- *** P < 0.001.
3.1.3 Intrahemispheric coherence
Analyses of intrahemispheric coherence revealed significant group × region × hemisphere interaction effects in the delta (χ2 = 58.692; p < 0.001) and theta (χ2 = 51.409; p < 0.001) bands and group × region effects in the theta (χ2 = 19.178; p = 0.038), beta (χ2 = 22.203; p = 0.014), and gamma (χ2 = 20.322; p = 0.026) bands after controlling for demographic and clinical variables. However, no significant group differences were found for any of the frequency bands (Table 3).
| Intrahemispheric coherence | Wald χ2 | df | P | Post hoc |
|---|---|---|---|---|
| Delta | ||||
| Group | .706 | 2 | 0.703 | |
| Group × Region | 15.019 | 10 | 0.131 | |
| Group × Hemisphere | .298 | 2 | 0.862 | |
| Group × Region × Hemisphere | 58.692** | 15 | <0.001 | n.s. |
| Theta | ||||
| Group | .725 | 2 | 0.696 | |
| Group × Region | 19.178* | 10 | 0.038 | n.s. |
| Group × Hemisphere | 1.730 | 2 | 0.421 | |
| Group × Region × Hemisphere | 51.409** | 15 | <0.001 | n.s. |
| Alpha | ||||
| Group | .322 | 2 | 0.851 | |
| Group × Region | 17.946 | 10 | 0.056 | |
| Group × Hemisphere | 1.888 | 2 | 0.389 | |
| Group × Region × Hemisphere | 10.045 | 15 | 0.817 | |
| Beta | ||||
| Group | 1.094 | 2 | 0.579 | |
| Group × Region | 22.203* | 10 | 0.014 | n.s. |
| Group × Hemisphere | 1.458 | 2 | 0.482 | |
| Group × Region × Hemisphere | 14.114 | 15 | 0.517 | |
| Gamma | ||||
| Group | 2.060 | 2 | 0.357 | |
| Group × Region | 20.322* | 10 | 0.026 | n.s. |
| Group × Hemisphere | 1.202 | 2 | 0.548 | |
| Group × Region × Hemisphere | 16.099 | 15 | 0.376 | |
| Interhemispheric coherence | Wald χ2 | df | P | Post hoc |
|---|---|---|---|---|
| Delta | ||||
| Group | .910 | 2 | 0.634 | |
| Group × Region | 8.782 | 6 | 0.186 | |
| Theta | ||||
| Group | 1.668 | 2 | 0.434 | |
| Group × Region | 6.393 | 6 | 0.381 | |
| Alpha | ||||
| Group | 2.144 | 2 | 0.342 | |
| Group × Region | 10.234 | 6 | 0.115 | |
| Beta | ||||
| Group | 2.640 | 2 | 0.267 | |
| Group × Region | 15.879* | 6 | 0.014 | Central: SG > MG |
| Gamma | ||||
| Group | 1.330 | 2 | 0.514 | |
| Group × Region | 7.319* | 6 | 0.292 | |
- Note: Post hoc comparisons were carried out using Bonferroni correction. SG, Single game users; MG, Multiple game users; n.s., not significant.
- * P < 0.05.
- ** P < 0.01.
- *** P < 0.001.
3.1.4 Interhemispheric coherence
Data on interhemispheric coherence are presented in Table 3. A significant group × region interaction effect was observed in the beta band after adjusting for demographic and clinical variables, with the SG group showing higher beta coherence in the central region compared with the MG group (χ2 = 15.879; p = 0.014). No significant main group or interaction effects were observed for the other frequency bands.
3.1.5 Correlation analyses
Correlation analyses were conducted of clinical and neurocognitive variables with EEG features that exhibited significant group differences in the GEE analyses (absolute beta power in the frontal region). There was a significant positive correlation between BIS scores and absolute beta power in the frontal region in the SG group (r = 0.356; Figure 2).
3.2 Group differences among gaming genres
3.2.1 Demographic, clinical, and neurocognitive assessments
The demographic and clinical characteristics of participants included in the genre-specific groups are shown in Table 4. There were no significant differences in age among the groups; however, education was significantly lower in the MOBA group than the HCs. The time spent on Internet gaming was significantly greater in individuals with IGD than the HCs, and MMG users devoted the most time to Internet gaming on weekdays compared with the other users. No significant differences were observed among the genre groups in terms of the severity of IGD. All three genre groups had higher scores on the BAI, BIS-II, and AQ and lower IQs than the HCs; however, scores did not differ significantly by genre. Depressive symptoms were most severe in the MMG group, followed by the MOBA and FPS groups, and finally the HCs. The MOBA group showed higher BIS and BAS scores compared with the HCs, whereas the MMG group showed higher BIS scores than the HCs.
| MOBA (N = 34) | FPS (N = 15) | MMG (N = 10) | HC (N = 64) | F | P | Post hoc | |
|---|---|---|---|---|---|---|---|
| Demographic data | |||||||
| Age (years) | 22.68 (3.46) | 25.67 (5.74) | 26.10 (6.84) | 24.67 (3.44) | 3.11* | 0.029 | |
| Education (years) | 13.24 (1.58) | 13.73 (2.25) | 13.00 (1.63) | 14.59 (2.16) | 4.53** | 0.005 | H > M |
| Game usage in weekdays (h) | 5.06 (3.13) | 3.11 (2.34) | 8.15 (4.96) | 0.43 (0.84) | 47.46*** | <0.001 | R > M = F > H |
| Game usage in weekends (h) | 8.13 (10.62) | 5.79 (4.42) | 9.00 (4.76) | 0.57 (1.20) | 14.33*** | <0.001 | R = M = F > H |
| Clinical data | |||||||
| Age of onset | 14.55 (4.48) | 15.82 (3.37) | 19.43 (9.85) | 1.52 | 0.238 | ||
| IQ | 105.24 (15.56) | 97.87 (20.45) | 97.60 (17.23) | 115.91 (10.86) | 11.07*** | <0.001 | R = M = F < H |
| IAT | 58.94 (16.41) | 55.67 (16.33) | 58.50 (19.32) | 29.94 (9.54) | 44.49*** | <0.001 | R = M = F > H |
| BDI | 16.06 (9.46) | 12.07 (10.32) | 22.60 (15.18) | 4.25 (4.54) | 25.47*** | <0.001 | R > M = F > H |
| BAI | 14.82 (12.48) | 12.33 (12.06) | 14.00 (11.33) | 3.52 (4.66) | 14.20*** | <0.001 | R = M = F > H |
| BAS | 36.91 (7.42) | 33.33 (5.89) | 34.00 (5.58) | 32.44 (5.77) | 3.81* | 0.012 | M > H |
| BIS | 21.74 (4.68) | 19.40 (5.41) | 22.00 (4.45) | 17.64 (3.58) | 8.44*** | <0.001 | M = R > H |
| AQ | 73.18 (17.80) | 74.40 (20.23) | 77.10 (15.13) | 54.69 (11.89) | 17.17*** | <0.001 | R = M = F > H |
| BIS-11 | 68.03 (8.14) | 66.27 (9.61) | 65.00 (9.50) | 56.38 (7.69) | 17.78*** | <0.001 | R = M = F > H |
| Neurocognitive data | |||||||
| Stroop word | 57.71 (7.71) | 55.53 (7.91) | 57.30 (21.64) | 51.53 (7.99) | 3.57* | 0.016 | M > H |
| Stroop colour | 98.21 (22.63) | 104.60 (20.64) | 107.00 (48.76) | 89.09 (16.25) | 3.52* | 0.017 | n.s. |
| Verbal fluency (letter) | 38.21 (10.77) | 36.53 (10.74) | 33.90 (12.23) | 48.61 (13.59) | 9.16*** | <0.001 | M = F = R < H |
| Verbal fluency (category) | 35.62 (10.00) | 30.93 (8.57) | 39.40 (10.83) | 44.34 (8.60) | 12.23*** | <0.001 | M = F < H |
| SWM (totala) | 8.97 (9.40) | 13.73 (15.53) | 29.10 (16.39) | 11.72 (13.00) | 6.64*** | <0.001 | R > M = F=H |
| SWM (strategyb) | 28.35 (4.98) | 31.27 (5.57) | 34.80 (3.29) | 28.78 (5.71) | 4.72** | 0.004 | R > M = H |
- Note: Data are presented as mean (standard deviation). Post hoc comparisons were carried out using Bonferroni correction. MOBA, Multiplayer Online Battle Arena; FPS, First-Person Shooter; MMG, Massive Multiplayer Online Role-Playing Game; M, MOBA group; F, FPS group; R; MMG group; HC and H, healthy controls; IQ, Intelligence Quotient; IAT, Internet Addiction Test; BDI, Beck Depression Index; BAI, Beck Anxiety Inventory; BAS, Behavioral Activation Scale; BIS, Behavioral Inhibition Scale; AQ; Aggression Questionnaire; BIS-11; Barratt Impulsiveness Scale 11; SWM, Spatial Working Memory.
- a Total number of errors.
- b Estimate of the use of the best strategy. High score represents poor use of this strategy; low score represents effective use.
- * P < 0.05.
- ** P < 0.01.
- *** P < 0.001.
A comparison of neurocognitive data by genre is shown in Table 4. The Stroop word reaction time was longer in the MOBA group compared with the HCs. In the verbal fluency tests, performance on letter fluency was lower in all three genre groups compared with the HCs, whereas the MOBA and FPS groups showed lower scores on the category fluency test than the HCs. The MMG group had more errors on the SWM than the other groups and showed poorer use of strategy than the MOBA group and HCs. Although not statistically significant, the MOBA group showed more effective use of strategy and fewer errors on the SWM compared with the HCs.
3.2.2 Absolute power
Figure 3 presents the scalp topography of each group in terms of the absolute power in each band. The GEE results showed a significant group × region interaction effect for absolute power in the delta band after adjusting for education, IQ, depression, and anxiety symptoms (χ2 = 15.383; p = 0.017). The FPS group exhibited higher absolute power in the delta band than the HCs, and the difference was most significant in the frontal region (Table 5).
| Absolute power | Wald χ2 | df | P | Post hoc |
|---|---|---|---|---|
| Delta | ||||
| Group | 3.581 | 3 | 0.310 | |
| Group × Region | 15.383 * | 6 | 0.017 | Frontal: F > HC |
| Group × Hemisphere | 6.072 | 3 | 0.108 | |
| Group × Region × Hemisphere | 6.248 | 6 | 0.396 | |
| Theta | ||||
| Group | 3.106 | 3 | 0.376 | |
| Group × Region | 7.785 | 6 | 0.254 | |
| Group × Hemisphere | 4.416 | 3 | 0.220 | |
| Group × Region × Hemisphere | 7.722 | 6 | 0.259 | |
| Alpha | ||||
| Group | 3.154 | 3 | 0.369 | |
| Group × Region | 10.915 | 6 | 0.091 | |
| Group × Hemisphere | 4.391 | 3 | 0.222 | |
| Group × Region × Hemisphere | 7.153 | 6 | 0.307 | |
| Beta | ||||
| Group | 3.008 | 3 | 0.390 | |
| Group × Region | 7.555 | 6 | 0.273 | |
| Group × Hemisphere | .906 | 3 | 0.824 | |
| Group × Region × Hemisphere | 10.276 | 6 | 0.114 | |
| Gamma | ||||
| Group | .824 | 3 | 0.844 | |
| Group × Region | 2.229 | 6 | 0.890 | |
| Group × Hemisphere | 1.280 | 3 | 0.734 | |
| Group × Region × Hemisphere | 5.661 | 6 | 0.462 | |
- Note: Post hoc comparisons were carried out using Bonferroni correction. F, FPS group, HC, healthy controls.
- * P < 0.05.
- ** P < 0.01.
- *** P < 0.001.
3.2.3 Intrahemispheric coherence
Analyses of intrahemispheric coherence revealed significant group × region × hemisphere interaction effects in the delta, theta, and beta bands after controlling for covariates (Table 6). The MMG group exhibited lower delta (χ2 = 108.431; p < 0.001) and theta (χ2 = 65.628; p < 0.001) coherence in the left frontoparietal area compared with the HCs. The FPS group showed lower theta (χ2 = 65.628; p < 0.001) and beta (χ2 = 32.316; p = 0.040) coherence in the left frontoparietal area compared with the HCs. Although there were significant group × region (χ2 = 27.315; p = 0.028) and group × region × hemisphere (χ2 = 45.734; p < 0.001) interaction effects in the gamma band, there were no group differences.
| Intrahemispheric coherence | Wald χ2 | df | P | Post hoc |
|---|---|---|---|---|
| Delta | ||||
| Group | 6.741 | 3 | 0.081 | |
| Group × Region | 45.181 ** | 15 | <0.001 | |
| Group × Hemisphere | 1.440 | 3 | 0.696 | |
| Group × Region × Hemisphere | 108.431 ** | 20 | <0.001 | Left frontoparietal: R < HC |
| Theta | ||||
| Group | 6.486 | 3 | 0.090 | |
| Group × Region | 35.390 ** | 15 | 0.002 | |
| Group × Hemisphere | .186 | 3 | 0.980 | |
| Group × Region × Hemisphere | 65.628 ** | 20 | <0.001 | Left frontoparietal: R, F < HC |
| Alpha | ||||
| Group | 7.225 | 3 | 0.065 | |
| Group × Region | 21.705 | 15 | 0.116 | |
| Group × Hemisphere | 2.026 | 3 | 0.567 | |
| Group × Region × Hemisphere | 13.753 | 20 | 0.843 | |
| Beta | ||||
| Group | 5.388 | 3 | 0.145 | |
| Group × Region | 23.728 | 15 | 0.070 | |
| Group × Hemisphere | .721 | 3 | 0.868 | |
| Group × Region × Hemisphere | 32.316 * | 20 | 0.040 | Left frontoparietal: F < HC |
| Gamma | ||||
| Group | 7.589 | 3 | 0.055 | |
| Group × Region | 27.315 * | 15 | 0.028 | n.s. |
| Group × Hemisphere | 2.331 | 3 | 0.507 | |
| Group × Region × Hemisphere | 45.734 ** | 20 | <0.001 | n.s. |
| Interhemispheric coherence | Wald χ2 | df | P | Post hoc |
|---|---|---|---|---|
| Delta | ||||
| Group | 3.245 | 3 | 0.355 | |
| Group × Region | 11.519 | 9 | 0.242 | |
| Theta | ||||
| Group | 1.626 | 3 | 0.653 | |
| Group × Region | 9.043 | 9 | 0.433 | |
| Alpha | ||||
| Group | 2.319 | 3 | 0.509 | |
| Group × Region | 10.082 | 9 | 0.344 | |
| Beta | ||||
| Group | 7.910 * | 3 | 0.048 | n.s. |
| Group × Region | 6.542 | 9 | 0.685 | |
| Gamma | ||||
| Group | 5.671 | 3 | 0.129 | |
| Group × Region | 11.330 | 9 | 0.254 | |
- Note: Post hoc comparisons were carried out using Bonferroni correction. F, FPS group, R, MMG group, HC, healthy controls; n.s., not significant.
- * P < 0.05.
- ** P < 0.01.
- *** P < 0.001.
3.2.4 Interhemispheric coherence
Analyses of interhemispheric coherence showed a significant main effect of group in the beta band after adjustment for covariates; however, post hoc analyses revealed no significant group differences (Table 6). There were no main effects of group and no interaction effects for interhemispheric coherence in the other frequency bands.
3.2.5 Correlation analyses
Based on the GEE results, correlations between absolute delta power in the frontal region and clinical and neurocognitive variables were investigated (Figure 4). There was a significant positive correlation between Y-IAT score and absolute delta power in the frontal region in the FPS group (r = 0.524), which indicated that increased absolute delta power in the frontal region was associated with more severe addiction symptoms. A significant negative correlation was observed between performance on the SWM and theta intrahemispheric coherence in the FPS group, which indicated that decreased theta intrahemispheric coherence correlated with more errors (r = −0.531) and poorer use of strategy (r = −0.552). No significant correlations were found between delta and beta intrahemispheric coherence and clinical/neurocognitive features in the MMG group.
4 DISCUSSION
The present study investigated differences in eye-closed resting-state EEGs according to gaming usage patterns and genre within IGD patients. First, we found that IGD patients who played SGs had increased absolute beta power in the frontal region compared with the HCs and increased beta coherence in the central region compared with IGD patients who played MGs. In the SG group, absolute beta power was significantly associated with tendencies toward behavioural inhibition. Second, in terms of genre-specific differences, FPS gamers showed increased delta power in the frontal region compared with the HCs. In addition, decreased intrahemispheric coherence in the left frontoparietal region was found in the MMG and FPS groups compared with the HCs. These differences were observed in the theta and delta bands in the MMG group and in the theta and beta bands in the FPS group.
The SG group showed a significant increase in absolute beta power compared with HCs and an increased beta coherence compared with the MG group. In previous studies, increased beta power, in particular in the frontal region, was reported in patients with alcohol use disorder,32 patients with anxiety disorders (such as panic disorder and agoraphobia),33 and children with learning disabilities.34 Previous studies in patients with IGD, however, have reported lower absolute power in the beta band compared with healthy controls, which is contrary to the findings from the present studies.7 One possible explanation for the conflicting results is that the increased absolute beta power and coherence in beta band may be a neurophysiological marker specific to single game users. A previous study35 suggested that exacerbation of beta band activity and beta band coherence may indicate abnormally strong inhibition and pathological persistence of the status quo, which could interfere with flexible behavioural and cognitive control. In fact, in the present study, increased beta coherence in the frontal area in the SG group was associated with tendencies towards behavioural inhibition. Therefore, it is possible that the SG group exhibits abnormal excessive inhibition of behavioural and cognitive changes at the cortical level, which may interfere with flexible control of behavioural and cognitive sets. In addition, these neurophysiological features may increase vulnerability to pathological preoccupation with an SG. However, further research is required to determine this explanation.
In terms of genre-specific differences, only FPS gamers showed increased absolute delta power in the frontal region compared with the HCs, and this activity was significantly associated with the severity of IGD. Slow-wave activity, such as delta waves, is involved in a wide range of cognitive processes, including demand attention and decision making. Therefore, an increase in slow-wave activity leads to impairments in these higher order cognitive processes. In particular, slow-wave activity in the frontal region is associated with inhibitory control. For example, a previous study found that increased activity in the slow-frequency band in the lateral prefrontal cortex is linked to deficits in inhibitory control.36 Another study also reported increased delta power in IGD patients and suggested that this may be a neural marker of dysfunctional inhibitory control associated with addictive gaming.5 In the present study, increased delta power was only observed in the FPS group, which was related to the severity of IGD. These results suggest that the relationships between IGD addiction symptoms and problems with inhibitory control and their associated neural activity may be most pronounced in FPS gamers compared with other genre users. This is in line with previous reports of an association between FPS gaming addiction and impulsivity and disinhibition. For example, one study found that addicted FPS gamers had significantly higher self-reported impulsivity and disinhibition in a go/no-go task compared with HCs.37 The researchers suggested that the relationship between deficits in inhibitory control and addictive gaming may be unique to the FPS genre and may be associated with its fast-paced, violent, and stimulating nature. Our results also suggest that increased slow-wave activity associated with dysfunctional inhibitory control in IGD patients may be specific to FPS gamers.
Our findings can be interpreted in two ways. The increased delta power observed in the FPS group can be explained as a secondary change derived from excessive use of FPS games. Increased delta power in the frontal region indicates cortical hypoarousal, which may impact subsequent task-related brain activation. Because of the lack of EEG studies on the impact of gaming, we can only refer to previous functional magnetic resonance imaging (fMRI) studies. For example, a previous study38 found that adolescents who played FPS games immediately showed diminished BOLD response in the dorsolateral prefrontal cortex during a behavioural inhibition task compared with those who played racing games. Another fMRI study demonstrated that activity in the prefrontal cortex was reduced during cognitive inhibition after participants played an FPS game over a period of 1 or 2 weeks.39 In addition, these effects persisted after the second week of playing the game. The researchers suggested that playing FPS games not only has an immediate impact on prefrontal activity but also has long-term consequences that may impair inhibitory control. Although one should be cautious in attributing the changes in brain activities to specific aspects of games, the authors suggest that relatively high levels of violence and excitement of FPS games may be related to decreased inhibition-related brain activities. Based on the aforementioned findings, we speculate that the increased delta power in the frontal region in the FPS group may suggest cortical hypoarousal and dysfunctional inhibitory control related to addictive gaming, which is also suggested by the reduced activation of the prefrontal cortex seen in previous fMRI studies.
Conversely, increased delta power in the frontal region may increase the likelihood of becoming addicted to FPS games. FPS games are distinguished from other games, such as MMG and MOBA games, by their fast pace and short duration, which leads to high levels of excitement. FPS games require rapid reactions during violent and fast-paced interactions, so immediate responses to stimuli are more beneficial than planned and careful actions. Therefore, individuals with relatively poor behavioural inhibition abilities and a preference for immediate behavioural responses and rewards may be more likely to engage in FPS games, which may predispose them to FPS gaming addiction. In this regard, increased delta power in the frontal region, associated with dysfunctional inhibitory control, may be a neurophysiological vulnerability that increases the likelihood of addictive use of FPS games. It should be noted, however, that these explanations remain speculative and any causal inferences cannot be determined in the present study.
In the present study, decreased intrahemispheric coherence in the left frontoparietal area was found in the delta (MMG group), theta (FPS and MMG groups), and beta (FPS group) bands compared with the HCs. Frontoparietal connectivity plays a central role in working memory, in particular in processing visuospatial information. The involvement of a frontoparietal network in visuospatial working memory has been found with different methods, such as fMRI,40 transcranial magnetic stimulation,41 and lesion studies.42 Conversely, reduced frontoparietal EEG coherence has been observed in patients with deficits in working memory, such as Alzheimer's disease, schizophrenia, and Parkinson's disease.43-46 Such findings suggest that the decreased intrahemispheric coherence in the frontoparietal area found in the FPS and MMG groups may be associated with greater deficits in visuospatial working memory. In the present study, worse performance was observed on the visuospatial working memory task in the FPS and MMG groups compared with the HCs. In addition, intrahemispheric coherence in the left frontoparietal area for the theta band was negatively associated with visuospatial working memory performance in the FPS group. Similarly, in the MMG group, correlations were observed between decreased intrahemispheric coherence and errors on the working memory task, although they were not statistically significant. Taken together, our findings suggest that MMG and FPS gamers show decreased intrahemispheric coherence in the left frontoparietal region, which might be associated with dysfunction in visuospatial working memory. On the contrary, the MOBA group showed similar or even a trend toward better performance than the HCs in spatial working memory task. The absence of visuospatial working memory impairments in the MOBA group, unlike the other two genre groups, may explain why this group exhibited no notable differences in EEG activities compared with HCs.
Previous studies on the relationship between Internet or video game playing and working memory have produced inconsistent results. Some studies of the relationship between game playing and working memory have shown a positive relationship47, 48 or no relationship,49, 50 whereas others have found that excessive game playing is associated with deficits in working memory, including visuospatial working memory.51, 52 Although the evidence is not conclusive, the results of the current study suggest that these inconsistent findings may be attributable to the heterogeneity of the game genre. We speculate that the negative relationship between excessive game use and visual working memory may be specific to FPS and MMG gamers, as evidenced by the neural and behavioural abnormalities found in these groups. These findings are unexpected, in particular with regard to FPS gamers, because FPS gamers are expected to be good at attending to and tracking many visual stimuli simultaneously, given the action-intense nature of the genre. One possible explanation for this discrepancy is that excessive exposure to stimulating visual stimuli during FPS games may cause the neural networks related to the visual working memory system to become hypoaroused in the resting state. However, further studies are needed to elucidate the characteristics of FPS and MMG games that may be associated with these genre-specific differences.
The present study has several limitations. First, the sample size was relatively small, and only male participants were included. This may limit the generalizability of the results. Further research in a larger, more representative sample is needed to examine gender differences in neurophysiological activity linked to IGD. In addition, the small sample size of subgroups may have limited statistical power to detect between-group differences. This is reflected in the fact that only some of the significant results in GEE analyses were found significant in the post hoc multiple comparisons. Second, given the cross-sectional design of the study, the identified relationships may be interpreted in both directions, and no conclusions about causality can be made. Future studies using longitudinal designs may help establish the direction of the relationships among genre, game usage patterns, and neurophysiological markers identified in the present study. Finally, responses on game usage patterns were collected from a single question with a dichotomous response option. Therefore, we could not assess whether MG players were biased toward one particular game (making them similar to SG players). In addition, because the game genre was classified based on the most frequently played game, there may have been cases in which patients may have been addicted to MGs simultaneously, which may have confounded the results.
5 CONCLUSIONS
Despite these limitations, this is the first study to examine differences in the neurophysiological characteristics of IGD patients according to game usage patterns and genre. The present study demonstrated that IGD patients who played SGs had increased absolute beta power in the frontal region, which was significantly related to behavioural inhibition. We suggest that increased beta activity associated with an inability to switch behavioural and cognitive sets may be a neurophysiological marker of pathological preoccupation with an SG. In addition, FPS gamers in this study showed increased delta power in the frontal region, which was associated with the severity of IGD. These findings suggest that increased slow-wave activity associated with dysfunctional inhibitory control in IGD patients may be specific to FPS gamers. Decreased intrahemispheric coherence in the left frontoparietal area was found in MMG and FPS gamers, which was associated with significant impairment in visuospatial working memory. These findings provide important information regarding the heterogeneity of the neurophysiological networks underlying IGD with respect to game genre and usage patterns.
CONFLICTS OF INTEREST
The authors declare no conflict of interest.
AUTHOR CONTRIBUTIONS
DH, DYO, and JSC contributed to study conception and design and interpretation of the findings. DH contributed to writing the first draft of manuscript. DYO, SYY, and JSC contributed to study supervision and were involved with providing feedback on drafted work. SYP contributed to data collection. DH and LJY contributed to data analysis and visualization. All authors critically reviewed content and approved final version for publication.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
