Video Game Research in the 21st Century: A Critical Literature Review

Video games are a relatively new development in daily life, and research of video games is even more nascent. Since the mid 2000s, there has been increased interest in video games from an academic perspective due both to the widespread acceptance and availability of games, as well as the realism of modern games. Psychologists have begun to view video games (VGs) and video game players (VGPs) as a new or uninvestigated demographic and as such there is quite a bit of video game related psychology research. This literature review aims to analyse a wide array of VG psychology research from the perspective of an endemic gamer and psychologist. To standardize the set of data, all research used in this review will utilize imaging techniques such as fMRI, MRI, and EEG. In doing so, this review will try to understand and elucidate some of the issues that arise in VG research, as well as highlight research that is notably thorough and reflects an understanding both of VGs and VG culture. This review will not attempt to reinterpret the results or draw new conclusions from the discussed research, but it will consider the proposed conclusions in the greater context of other research as well as in the potential implications or consequences on a broader scope. The review will most heavily consider four major aspects of research: the posed research question and preconceptions and prior knowledge of games and gaming, sample selection and demographic analysis, collection and operationalization of data, and conclusions made and the broader consequences or implications of the research. While no variables are being tested in this review, it will conclude with a critical analysis of the quality and progression of VG research.

1. Introduction

Some of the most historically prominent VG research in psychology pertains to increased aggression in violent VGPs. An empirical literature review from 1998 by Karen E. Dill and Jody C. Dill considered 82 published papers dating back as far as 1983. The review states that the “preponderance of evidence from the existing literature suggests that exposure to video-game violence increases aggressive behavior and other aggression-related phenomena” (Dill & Dill 1998). This and other preconceptions of video game often influence the way in which the research is conducted. Similarly, misconceptions of the gender disparity often lead to experiments that are dominantly or entirely male. Sample selection for VG research often raises a new question for researchers: how is the population defined or determined for VGPs?

Once the sample is finalized, researchers must design experiments that collect data that can be operationalized and analyzed. Given the nascency of video game research as well as a limited understanding of brain function, the usability of data isn’t entirely understood and researchers often run the risk of false attribution. Properly collected and operationalized data generally tests limited or mutually exclusive variables. Conclusions made should be exclusively based on the collected data.

1.1 Research Question, Misconceptions, and Experiment Design

One of the core concerns with older video game research is misinformation or faulty preconceptions shaping the experiment. Researchers were often not endemic gamers and allowed their perception of gamers color the hypothesis or the design of experiments. In the review by Dill and Dill, begins with the preconception that video game violence universally increases aggression and violence, and used many papers that made similar claims. Many of the papers used in the Dill and Dill review were built on the question “Do violent video games increase aggression” rather than “Do video games change or modify construals and alter behavior”. While this review starts on the assumption that violent VGs increase aggression, it concludes with a discussion in which they admit that much of the research is inconclusive, that the research was in a “fledgling state” and that future research needed to focus on careful methodology. The review also states that while the demographic that commits the most violent crime is also most likely to play video games, a correlation between violence in video games and violent or aggressive behavior is not certain and needs more research (Dill & Dill 1998).

It is important to understand the preconceptions held by researchers. By investigating some of these prior held beliefs or understandings of VGs, one can critically analyse the methodology used as well how conclusions were made. For example, research by Dye, Green, and Bavalier testing attention skills in action VGPs used the list in Appendix 1 to define action and non-action VGs. The list of non-action VGs includes titles like Soul Calibur 2, Final Fantasy VII, Kingdom Hearts, and Spiderman all of which are not defined as exclusively action games, but are heavily influenced by the same factors that define action VGs (Dye, Green, and Bavalier 2009). By limiting their scope of action VGs to exclusively first-person shooters (FPS) and strictly action-adventure games, the study neglects to consider all action VGPs and only investigates a segment of the whole population. Many studies exist that were initiated on a misconception or incorrect stereotyping of VGPs and these studies often share this trait; by limiting or biasing the original scope of the study, they improperly choose a sample population for their experiments.

1.2 Sample Selection

A common misconception made in old VG research was that primarily males played VGs, and samples in many studies reflect this and are exclusively male. The table shows more recent data which indicates that on average, approximately 40% of gamers in most countries are female. While there have definitely been changes in VGs and VG culture that have contributed to this change, the Dill and Dill review from 1998 noted that though males preferred violent video games, the majority of both male and female adolescents played VGs on a regular basis (Dill & Dill 1998). Similarly, a regularly occurring problem that arises with general VG research is defining “the whole population.” For VG research, the researcher must decide whether or not to use the whole population or the population of VGPs and to choose a sample accordingly. There are merits to both methods. By using the whole set of VGPs, a researcher is able to make inferences about all VGPs but cannot measure the impact on non-VGPs (NVGPs). By using the whole population, researchers can understand the impact of video games as they impact all people, but may not be able to accurately measure any specific impacts on VGPs. Generally, this choice is made at the onset and the limitation is designed into the experiment. For example, a study by Kwak, Blackburn, and Han that analyzed toxic behavior in games, limited the study to League of Legends (LOL), immediately limiting the whole population to LOL players. Methods like this can be a convenient way of limiting the scope of a study to be more reasonable, but can also aid in increasing the specificity of research (Kwak, Lee, and Blackburn 2015) .

1.3 Data Collection and Operationalization

Though the understanding of brain function has improved significantly in the past, there is still much that is unknown about the brain. This is true of video games as well; while much has been learned about video games in culture and society, there is not a complete or thorough understanding of how video games impact the psychology of individuals. For example, in research by Boot et al., the cognitive battery used to test the effects of VGs on attention, memory, and executive control consisted of twelve tests. Each test had its own set of data collected, and all the data was operationalized using standardized ANOVA procedures. The research uses multiple games, and each set of participants did the same cognitive battery albeit with different frequency (cross-sectional participants took the battery once, while longitudinal participants took the battery three times). The data collected widely pointed to the same results: VGPs perform higher than their NVGP counterparts, but the difference is not statistically significant. While on the first battery, VGPs outperformed NVGPs, the NVGP group for each of the tests did show varying degrees of improvement over the span of the study (Boot, Kramer, Simons, Fabiani, & Gratton 2008). The vast amount of data collected leads to a natural concern: was any of the collected data extraneous or did it distract from more important or notable data?

When analyzing data collection and operationalization, this review will consider the amount of data collected, as well as the difficulty that arises when considering many sets of data. Using systems such as ANOVA and MANOVA can standardize data and make it easier to understand, but broad data collection may lead to a phenomenon known as kitchen sink regression in which too many independent variables are attributed to a dependent variable, making it difficult to understand how the independent variable effects the dependent variable. I will be looking for research that collects sufficient data, but more importantly, uses the data in a conscientious manner.

1.4 Conclusions and Broader Consequences

In the 1990s, there was a focus on research of violent VGs. This was meant to determine whether violent VGs increased aggression and lead to increased crime rates within violent VGPs. As early as 1983, the United States Surgeon General, Dr. Charles Everett Koop implicated video games in increased crime and violence rates in America. In the 1990s, especially in the aftermath of tragedies like the Columbine School Shooting, awareness of VG violence was up significantly, and the public discourse was heavily opposed to violence in VGs. Bolstered by fears of increased violence in our communities, research was carried out to conclude that VG violence translated to increased aggression and crime rates. The 1998 research “Seeing the World Through Mortal Kombat Colored Glasses” makes one such conclusion. The publication concludes by stating that “the results of the study offer some support for the contention that violent video games lead to the development of a short-term hostile attribution bias” (Kirsh 1998).

Studies and conclusions like this lead to legislation introduced in 2005 to limit the sale of “Mature” and “Adult-Only” video games (Vargas 2005). The law was never passed. The conclusions made in any research should solely take into account data collected and the analysis of that data. It can and should also use up-to-date demographic data for VGPs. The major risk in the conclusions is making too broad of a conclusion, as this could lead to more legislation or policy built on insufficient data.

2. Analysis of Research

For this analysis, I will consider publications that use MRI, fMRI, or other imaging techniques as a means of recording changes or alterations in brain activity. While this is not a comprehensive set of imaging studies, the studies span from 2006 to current and cover a variety of VG related topics. These studies often come with a severe limitation; it is logistically difficult to have a large number of participants due to the availability of the machines. That said, certain criteria for analysis will change depending on the limitations of the study.

2.1 Research Question and Preconceptions

The research using imaging use these techniques to answer general questions about brain activity and changes in brain function. Research by Klaus Mathiak and René Weber in 2006 attempted to understand brain correlates as VGPs played violent video games. The researchers posed a basic question: when playing violent video games, does the brain activity observed mirror or correlate to natural behavior? This question is clear and both simple and general allowing for the research to reflect the impact of violent VGs on brain activity but not on behavior or personality traits. Additionally, Mathiak and Weber initiated the research without the preconception or bias that violent VGs increase aggression. In doing so, they disregarded some of the previous evidence of violent VGs increasing aggression and instead compared their MRI results to better understood phenomena (explained further in Section 2.3). Furthermore, in designing the experiment, the researchers understood that different parts of games elicit different socioemotional responses, each with a unique neural response. This was evident in their experiment design in which they had participants play five controlled and preselected rounds of the violent VG “Tactical Ops: Assault on Terror” for a preselected time in an fMRI machine. Each round allowed researchers to test for unique brain patterns following certain events in game (Mathiak and Weber 2006).

Though there was a clear focus on the impact of violent VGs in the early and mid 2000s, there has been a paradigm shift in VG research in the past 10–15 years. This is highlighted by an increase in research designed to better understand complex problem solving and more recently, VG addiction, known as Internet Gaming Disorder (IGD). Research by Lee et al. from 2012 was initiated to “characterize the effect of training strategy on functional brain activation changes during a complex visuomotor task.” The goal of this work reflected a clear understanding of the growth and complexity of modern VGs. To this end, the researchers used Space Fortress, a game developed and designed to be a rigorous visuomotor task, for the experiment. Additionally, researchers allowed participants to familiarize themselves with the mechanics of the game before conducting MRI sessions, indicating a clear understanding of the importance of familiarity for VGPs. Similarly to the Mathiak research, each participant was tested multiple times with researchers often focusing on different zones of activation (Lee et al. 2012). Follow-up research published in 2014 by the same group tested parietal plasticity again using Space Fortress. This research was predicated on the idea that the differences in working memory performance are linked to differences in working memory task-based neural activation in the prefrontal and parietal lobes (as reported by Kane and Engle 2002). The experimental design was thorough, using two very different memory tasks after having participants learn and play Space Fortress. The researchers believed that playing the complex VG would improve participants’ scores on both memory tasks (Nikolaidis, Voss, Lee, Vo & Kramer 2014).

This acute interest in the effects of VGs on cognitive tasks was further studied in 2017 by Richlan et al. This study sought out action VGPs with the preconception that because action VGPs must track multiple objects in game, they could solve visual and verbal cognitive tasks with higher success rates than NVGPs. This hypothesis was bolstered by previous research that indicated that this may be the case. While the researchers entered into the experiment with this preconception, it did not heavily influence or flaw the experimental design. All participants were given a battery of preliminary behavioral tests and a training session and carried out three tasks within the fMRI machine. The fMRI procedure for experimentation as well as data collection was standardized in much the same way as the Weber experiment (Richlan, Schubert, Mayer, Hutzler, & Kronbichler 2017). This research will be more deeply discussed in section 2.4.

One of the more recent interests for researchers has been understanding potential IGD. Many studies exist that establish a possible modality by which IGD occurs, with research often looking for parallels to substance addiction and desire. Research by Han et al. in 2011 investigated brain activity associated with desire in VGPs. The posed question was very simple: will VGPs desire gameplay more than NVGPs and can these differences be measured via MRI? This is another fantastic example; the researchers hypothesize that VGPs will have stronger feelings of desire, but they seek out a clear difference in levels of desire in VGPs and NVGPs. Moreover, the experimental design was simple, but effective. All participants were told to play the same game for a fixed duration for 10 days, and they were all exposed to the same footage in the MRI. By standardizing the parameters that all the participants needed to meet, it was easier to directly compare the results from VGPs and NVGPs (Han, et al. 2011). This research did not mention or investigate IGD, but did predicate the experiment on possible similarities to drug or alcohol addiction. This was the basis of a widely cited study often used in defense of the inclusion of IGD in the Diagnostic and Statistical Manual of Mental Disorders (DSM). The study, by Wang et al., analyzed people who self-identified as having IGD by using the Young Internet Addiction Test kit (IAT kit). This study is problematic and will be discussed more heavily in later sections, but the posed research question as well as the preconceptions are sound. The study investigated changes in cortical thickness in people with IGD based on the premise that IGD mirrors substance addiction which can cause significant decreases in cortical thickness. The issues arise in experiment design as well as sample selection and conclusions. The experiment used exclusively VGPs and compared data within this set of participants. Additionally, by using people who self-identify as having IGD, the researchers did not qualitatively or quantitatively confirm the presence of IGD (Wang et al. 2018).

The Wang study built on research that indicated that there were strong similarities between IGD and substance abuse and addiction. One such study carried out in 2011 used chemical techniques to reduce video game cravings in the same way one would reduce substance cravings. This study, also by Han et al., used bupropion, a chemical used to aid in the cessation of smoking, to reduce cravings for VGs in patients with IGD. This study was quite novel as it proposed a new treatment for potential IGD, and the posed question was easy to test. Patients who were identified as exceeding the threshold for excessive gaming were treated with extended release bupropion, and exposed to games or game material in fMRI machines to test for decreases in brain activation at exposure to games. A standard procedure was used for fMRI data collection (Han, Hwang, & Renshaw 2010).

Studies have also been carried out using other imaging techniques such as EEG and NIRS to understand other facets of psychology as it pertains to VGs. Russoniello et al. investigated the positive effects of gaming on mood and emotion in 2009 using EEG. They sought also to understand if VGs can reduce stress and if these changes could be observed with psychological and physiological measures. The posed question of this study is very reasonable, as is the preconceptions about potential positive effects that VGs may have. The only concern that arises is a potential bias in the “positivity” of the paper: the authors mention that most previous VG research focuses on negative impacts of VGs, citing studies on violence and IGD, contrasting this research by focusing on the perceived “positivity” of the study. While this is not likely to have heavily impacted the experiment design, it likely influenced the conclusions made. The design itself was sound, using a standard protocol EEG lead arrangement to measure electrical changes most associated with mood. The researchers used casual, yet engaging games to elicit positive emotions in participants (Russoniello, O’Brien, & Parks 2009).

The final study for analysis uses NIRS to measure physiological changes, namely changes in the oxygenation of hemoglobin in the brain. This study, by Matsuda and Hiraki, measures changes in oxygenated hemoglobin in the dorsal prefrontal cortex of children while playing VGs. the research was predicated on previous research also by Matsuda and Hiraki that showed a sustained decrease in oxygenated hemoglobin in adults playing four kinds of video games. The present study is an extension to that previous work meant to understand if these physiological changes are mirrored in children as well. Once again, by simplifying the posed question, researchers are better able to investigate whether or not the expected phenomena is observed and easily explained. The study, conducted in Japan, used two very popular games at the time to test for these phenomena. The design of the experiment was again very simple: researchers had participants sit in front of a television and initially stare at a grey screen before beginning to play. This allowed researchers to measure the oxygenated hemoglobin levels pre-play, during play and post-play (Matsuda & Hiraki 2006).

2.2 Sample Selection

Due to the limitations involved in conducting extensive imaging studies, relatively small samples were used for every study. Additionally, most of the studies use people within the immediate vicinity of the institution where the study was completed. This leads to certain limitations in the broad consequences for these studies, but still the data collected is notable and can be used in the context of other studies.

The Mathiak and Weber research had a sample size of thirteen male volunteers from Germany. This sample is remarkably small and the exclusion of female VGPs is concerning, though this oversight can be attributed to a lack of gender specific demographic studies regarding video games at the time. The sample, while small, was appropriate for the experiment as Mathiak and Weber sought only to identify brain activity associated with semi-natural behavior as it exhibited during violent VG gameplay. The use of only German volunteers does further limit the larger implications for this study as culture may impact the perception of violent VGs. Additionally, this study does not use a control sample, and rather refers to other existing data as the control (Mathiak and Weber 2006).

The two publications from the Beckman Institute at the University of Illinois use similar methods for sample selection. The study by Lee et al. used 75 participants from the Urbana-Champaign community. The 75 people were primarily female (2:1 Female to Male ratio). The sample was then divided into three groups for the different aspects of the experiment. Without knowing specific details about the participants, it cannot be known if the sample is fully representative of all VGPs, but the inclusion of female participants helps affirm the legitimacy of the sample chosen. That said, having 66% of participants as females does not align with the demographic numbers on gender of VGPs as has been reported in the past (Lee, et al. 2012).

Similarly, the follow up study authored by Aki Nikolaidis chose its sample from the Urbana-Champaign area. The sample size was smaller at 45 participants, but the gender ratio was more representative of the currently accepted demographic analysis. Of the 45 total participants, 27 were female. The study also recruited 25 control participants, but these results were not included in the final publication because the study focused on individual differences between the VGP participants (Nikolaidis, Voss, Lee, Vo, & Kramer 2014).

The study by Richlan et al. used a different approach, seeking out “longtime action VGPs” and NVGPs for the study. The experiment used a sample of 28, with a 50:50 ratio of action VGPs and NVGPs. Gender and demographic analysis was not provided, but again, we can assume that the sample was taken from the area around the University of Salzberg where the research was conducted (Richlan, Schubert, Mayer, Hutzler, & Kronbichler 2017).

The studies investigating IGD and VG desire all used very small samples that share some of the same limitations as the Mathiak study. The studies by Han et al. and Wang et al., all used samples comprising almost of males within close proximity of the institutions represented. The publication on VG desire used 19 male participants who identified as casual or regular VGPs (Han, et al. 2011). Similarly, the Wang study used 38 VGPs (27 male VGPs), and 66 recreational VGPs (37 male VGPs). The recreational VGPs were used as a control (Wang, et al 2018). While on first glance, this sample is reasonable, it comes under criticism when considered in context with the results found. The bupropion study used 11 male participants who all identified as exceeding a threshold for IGD. While the exclusion of females from the Han studies is troublesome, it does not heavily detract from the results given the generality of the questions posed (Han, Hwang, & Renshaw 2010). Follow-up research should use larger samples that are more representative of a modern VGP demographic.

Of the final two, non-MRI/fMRI studies, the EEG study by Russoniello et al. used a large sample of 143 participants of which the data from 134 participants was used. Gender analysis was not made clear in the publication as it said there were 57 female participants and 44 male participants, which does not fully account for the 134 participants whose data was included in the final results. Again, these students were recruited via flyers on campus or in the town nearby, and most participants were students, faculty, and staff (Russoniello, O’Brien, & Parks 2009). The Matsuda and Hiraki study recruited 20 school children for their experiment of which 15 were male and 5 were female. If a follow-up was proposed, the sample could be increased, but for the topic investigated, 20 participants sufficed. Since the goal was exclusively to find parallels between the levels of oxygenated hemoglobin, the researchers simply needed to test for changes in the children that were akin to those in adults (Matsuda & Hiraki 2006).

2.3 Data Collection and Operationalization

One of the most notable benefits of using imaging techniques is the existence of standardization processes. For all MRI and fMRI studies, a number of widely available protocols built in programs like Matlab were used to operate the machinery for experimental purposes. Every use of MRI or fMRI utilized one of these protocols to initiate a multidimensional analysis of brain activity, often using 1–3 mm slices. The fidelity of each scan was dependent on the strength of the machine; two standard strengths of machine were reported in the research used in this review — 1.5 T and 3 T. Machines operating at 3 T are able to produce thinner slices with higher fidelity, while machines operating at 1.5 T must take thicker slices to retain this fidelity. Once imaging was completed, ANOVA and MANOVA were commonly used to operationalize the data in a way that is usable for analysis. Additionally, other statistical methods were used to standardize data between participants or sample groups.

Similar protocol was used for the Russoniello research, albeit without MRI. EEG leads were used to collect data from the brain and a heart rate monitor was employed as well. The researchers used widely accepted protocol regarding the location and placement of the leads, and used basic statistical analysis associated with EEG to standardize data for all groups tested. Once the data was collected and standardized, it was operationalized with Cohen’s Delta which allows for comparison of effect size on group means.

2.4 Conclusions and Broader Consequences

Though VG research in the 20th Century was often loaded to shine a negative light on VGs and VG companies, research in the 21st Century has been much more conscious of the possible consequences of research. Most research cited in this review is very aware of the limitations of VG research, as well as the limitations of interpretation of data. The only research cited that overstates the possible consequences of the results is the Wang study . The study states that the limited sample of self-identified VGPs with IGD can be representative of all VGPs, and that the functions involved in IGD mirror substance abuse entirely due to decreases in cortical thickness.

By comparison, every other study cited is fully aware of the possible limitations involved in their research. The Mathiak and Weber study states this very clearly, noting that if violent VGs can increase aggression on a societal level, this could have major social implications, but goes on to explain that their data does not support the hypothesis that violent VGs promote aggression in real life scenarios. They proceeded to propose possible modalities by which violence in VGs impacts us, and states that those ideas are for later investigation.

The task completion studies are also very clear with their results and do not make larger, unfounded implications. In the initial study by Lee et al., the researchers noted that having training on a complex task reduces the visuomotor processing needed for the tasks used, but Lee goes on to explain that there are clear limitations for the implications of the study because we do not fully understand the systems involved in training-related changes in the brain. This sense of the limitations of results is mirrored in the Nikolaidis publication; while the results confirm the hypothesis that frontal-parietal networks are crucial to working memory tasks, he also states that the sample was not idealized for the study and that future studies need to reflect the shortcomings of the chosen sample. Regardless, the broader implications of research like this are clear — the ability to complete complex tasks in VGs translates in some ways to the ability to complete visuomotor tasks out of game. Though this is important, without further research on a larger scale, currently these results cannot be applied to all VGPs.

The Richlan research which also focused on task completion had a drastically different outcome. While VGPs had a faster reaction time, this did not mean they were more able to successfully complete visual or verbal cognitive tasks. Additionally, they found no discernible differences between VGPs and NVGPs behavioral pattern, though they did observe certain physiological changes from test to test. The researchers understood that the experimental design used may not have been visually stimulating enough to elicit a neural response, but also that the blood oxygen imaging used in tandem with fMRI observed significant differences in VGPs and NVGPs. This is currently a topic for future investigation for the research group.

The two publications by Han et al. both confirm many of the expected results. As was hypothesized, brain activation when exposed to VGs is very similar to activation when exposed to addictive substances. Additionally, the publication on game desire pinpointed a few potential regions of the brain that activate heavily when exposed to video games. Again, because the sample was entirely male, the researchers state that future studies should use a more diverse sample, and that other unknown factors may have played a role in the increased desire for a VG experience. In the bupropion study, Han observed a clear reduction in the desire to play games in patients with IGD, but again notes that there are heavy limitations. First and foremost, the treatment method was only used for a relatively short period and researchers do not know the long term effects of the treatment. Additionally, IGD is very likely comorbid with other mental illnesses and any changes may also be due to factors outside of the control of the researchers.

The results of both the EEG and NIRS studies confirmed the initial hypothesis. In the EEG study, casual VGP can improve mood and reduce stress, and that different games have different yet complementary mood lifting effects. While the study used EEG and heart rate, it could have used respiration as well as other biometrics to monitor stress levels. That said, the paper made a clear connection between VG play and positive mood improvement without any caveats. This does not test for diminishing returns or increased stress during VG play. The NIRS study proposed that children experience the same decreases in oxygenated hemoglobin in the dorsal prefrontal cortex, noting also that age did not have an impact on the rate of change of oxygenated hemoglobin, thus giving us more insight into some of the modalities by which VGs impact brain physiology in the short term.

3. Conclusions and Final Thoughts on VG research

There is no question that VG research has evolved dramatically in the past 20 years. While VG research used to focus on the negative impacts of VGs on VGPs, modern research tries to understand how we interact with VGs both consciously and unconsciously. On their own, most of these studies only hold so much weight, but when considered in the greater context of each other as well as the full body of other VG research, collectively we are slowly getting a more full understanding of the impacts of VGs on physiology and behavior. With research focusing more on how our brains react and respond to VG stimuli, we are able to deepen our understanding of media studies as well as we are able to more concretely draw connections between various forms of media. Additionally, with research like that of Matsuda and Hiraki or Han et al., we are better able to understand some of the chemistry and biology involved in VG play. As we deepen our understanding of any one facet of VG research, we will inevitably deepen our understanding of other facets.

Though I firmly believe that VG research has made great strides, it is safe to say that there is still an immense amount of research needed to fully understand the intricacies of how VGs effect and impact human psychology and physiology. As more endemic gamers begin to find their passion in psychology, our ability to learn more about gaming will rise dramatically.

APPENDIX 1

Citations

1. Dill, Karen E., and Jody C. Dill. “Video Game Violence.” Aggression and Violent Behavior, vol. 3, no. 4, 1998, pp. 407–428., doi:10.1016/s1359–1789(97)00001–3.
2. Dye, M.w.g., et al. “The Development of Attention Skills in Action Video Game Players.” Neuropsychologia, vol. 47, no. 8–9, 2009, pp. 1780–1789., doi:10.1016/j.neuropsychologia.2009.02.002.
3. Han, Doug Hyun, et al. “Bupropion Sustained Release Treatment Decreases Craving for Video Games and Cue-Induced Brain Activity in Patients with Internet Video Game Addiction.” Experimental and Clinical Psychopharmacology, vol. 18, no. 4, 2010, pp. 297–304., doi:10.1037/a0020023.
4. Han, Doug Hyun, et al. “Brain Activity and Desire for Internet Video Game Play.” Comprehensive Psychiatry, vol. 52, no. 1, 2011, pp. 88–95., doi:10.1016/j.comppsych.2010.04.004.
5. Kane, Michael J., and Randall W. Engle. “The Role of Prefrontal Cortex in Working-Memory Capacity, Executive Attention, and General Fluid Intelligence: An Individual-Differences Perspective.” Psychonomic Bulletin & Review, vol. 9, no. 4, 2002, pp. 637–671., doi:10.3758/bf03196323.
6. Kirsh, Steven J. “Seeing the World Through Mortal Kombat-Colored Glasses.” Childhood, vol. 5, no. 2, 1998, pp. 177–184., doi:10.1177/0907568298005002005.
7. Kwak, Haewoon, et al. “Exploring Cyberbullying and Other Toxic Behavior in Team Competition Online Games.” Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems — CHI 15, 2015, doi:10.1145/2702123.2702529.
8. Lee, Hyunkyu, et al. “Videogame Training Strategy-Induced Change in Brain Function during a Complex Visuomotor Task.” Behavioural Brain Research, vol. 232, no. 2, 6 Apr. 2012, pp. 348–357., doi:10.1016/j.bbr.2012.03.043.
9. Mathiak, Klaus, and René Weber. “Toward Brain Correlates of Natural Behavior: FMRI during Violent Video Games.” Human Brain Mapping, vol. 27, no. 12, 2006, pp. 948–956., doi:10.1002/hbm.20234.
10. Matsuda, Goh, and Kazuo Hiraki. “Sustained Decrease in Oxygenated Hemoglobin during Video Games in the Dorsal Prefrontal Cortex: A NIRS Study of Children.” NeuroImage, vol. 29, no. 3, 2006, pp. 706–711., doi:10.1016/j.neuroimage.2005.08.019.
11. Nikolaidis, Aki, et al. “Parietal Plasticity after Training with a Complex Video Game Is Associated with Individual Differences in Improvements in an Untrained Working Memory Task.” Frontiers in Human Neuroscience, vol. 8, no. 169, 27 Mar. 2014, doi: 10.3389/fnhum.2014.00169.
12. Richlan, Fabio, et al. “Action Video Gaming and the Brain: FMRI Effects without Behavioral Effects in Visual and Verbal Cognitive Tasks.” Brain and Behavior, vol. 8, no. 1, 2017, doi:10.1002/brb3.877.
13. Russoniello, Carmen, et al. “The Effectiveness of Casual Video Games in Improving Mood and Decreasing Stress.” Journal of CyberTherapy and Rehabilitation, vol. 2, no. 1, 2009, pp. 53–66., www.researchgate.net/publication/289131468.
14. Vargas, Jose A. “On Capitol Hill: Blame Games.” The Washington Post, WP Company, 17 Dec. 2005, www.washingtonpost.com/wp-dyn/content/article/2005/12/16/AR2005121601883.html?tid=a_inl_manual&noredirect=on.
15. Wang, Ziliang, et al. “Cortical Thickness and Volume Abnormalities in Internet Gaming Disorder: Evidence from Comparison of Recreational Internet Game Users.” European Journal of Neuroscience, vol. 48, no. 1, 2018, pp. 1654–1666., doi:10.1111/ejn.13987.