The authors have declared that no competing interests exist.
Conceived and designed the experiments: BDG WTM BCL. Performed the experiments: BDG. Analyzed the data: BDG. Contributed reagents/materials/analysis tools: BDG BCL. Wrote the paper: BDG WTM BCL.
Training in action video games can increase the speed of perceptual processing. However, it is unknown whether video-game training can lead to broad-based changes in higher-level competencies such as cognitive flexibility, a core and neurally distributed component of cognition. To determine whether video gaming can enhance cognitive flexibility and, if so, why these changes occur, the current study compares two versions of a real-time strategy (RTS) game. Using a meta-analytic Bayes factor approach, we found that the gaming condition that emphasized maintenance and rapid switching between multiple information and action sources led to a large increase in cognitive flexibility as measured by a wide array of non-video gaming tasks. Theoretically, the results suggest that the distributed brain networks supporting cognitive flexibility can be tuned by engrossing video game experience that stresses maintenance and rapid manipulation of multiple information sources. Practically, these results suggest avenues for increasing cognitive function.
Neuroplasticity is the ability of the adult brain to not only learn new behaviors and form new memories but to alter the underlying neural structures responsible for such learning
In 2008, 72% of the general population and 97% of teenagers aged 12–17 reported playing video games
Prior experimental investigation of the cognitive consequences of video gaming provides evidence that cognitive and perceptual changes occur in those who transition from non-gamers to gamers. Specifically, training on action games (e.g., first-person, fast paced, kill-or-be-killed situations) has been linked to enhanced core perceptual processing
One key question is whether video game play can alter aspects of higher-level cognition
This array of studies is a promising indication that video game training can lead to improvements in cognitive test performance. However, another video game training study carried out on younger adults with little gaming experience did not find improved performance on cognitive testing after 23.5 video game training on either a real-time strategy game, an action video game, or a puzzle game
In the present study, we add to an emerging body of literature suggesting video gaming can alter higher-level competencies by devising a gaming intervention and specifying diagnostic measures that are optimized to assess the effects of gaming on cognitive flexibility. Cognitive flexibility is the essential ability to assess and adapt ongoing psychological operations and to coordinate the allocation of cognitive processes appropriately in dynamic decision making environments
Real-time strategy (RTS) game training is an excellent candidate for tuning these cortical networks due to sustained maintenance and rapid switching across multiple information sources at a high workload for long periods of time over several weeks. Our main prediction is that our RTS gaming manipulation will complement previous work in action video games by promoting “fast thinking”, while not strongly affecting “fast perception.” To assess whether video game training can alter cognitive flexibility, the current study utilizes an RTS game, StarCraft (published by Blizzard Entertainment, Inc. in 1998). RTS gaming involves the creation, organization, and command of an army against an enemy army in a real-time map-based setting (see
To further elucidate the role of video gaming on cognitive flexibility, gaming is examined with a within-game and between-game comparison. A life-simulator game is used as a control gaming condition against two versions of the RTS game: a full-map version and a half-map version. The full-map version (SC-2) involves two friendly bases and two enemy bases, whereas the half-map version (SC-1) involves one friendly and one enemy base and half the available gaming space. In the SC-2 version, the player is commanding and controlling two separate bases in multiple battles against two separate opponent bases. For this reason, the SC-2 subcondition promotes more switching and coordination of cognitive resources, hallmarks of cognitive flexibility. Both versions were also modified with a reactive difficulty level in order to maintain a win rate near 50%. Thus, the SC-1 version was designed to be as engrossing and difficult as the SC-2 version, but did not emphasize maintaining awareness of and switching between two spatially separated (out of view) bases. Importantly, game feature and behavior recording within the RTS game allowed for verification of the amount of attended information between the two versions. The life-simulator video game, The Sims 2 (published by Electronic Arts, Inc. in 2004), has been shown to be a useful control for experimental video game research
To determine changes in cognitive flexibility that occurred as a result of video gaming, participants completed a battery of psychological tasks at pre-test and post-test (at 40 hours of gaming). The battery included measures that address cognitive flexibility as well as measures of unrelated constructs. Measures of flexibility included the Attention Network Test (ANT)
Undergraduate participants were recruited from the University of Texas at Austin. Respondents to an advertisement were screened and selected for inclusion on the basis of reported video game use. Participants were randomly assigned to one of the three experimental conditions. Those who reported <2 hours per week of video game use were included (SC-1, n = 26; SC-2, n = 20; Sims, n = 26). Mean age was 20.3 years (SD = 1.1) for SC-1, 20.4 years (SD = 1.1) for SC-2, and 19.9 years (SD = 0.8) for the Sims condition. All participants were female due to the small number of non-gaming males (see
The experimental procedure included the completion of a pre-test task battery, followed by gaming condition assignment (SC-1, SC-2, or The Sims). Each participant engaged in 40 total hours of video gaming. This video game play occurred outside the lab on the participants' own laptops. Psychological testing occurred over sessions in the laboratory.
The video game software along with experiment running software was installed on each participant's laptop. The software provided a portal which controlled the settings for each gaming session and prevented the participant from playing more than 40 hours before post-test. Players were instructed to play for roughly one hour per day. For the SC subconditions, gaming sessions alternated between two StarCraft map formats that differed by terrain theme and layout. Additionally, a titration procedure raised or lowered the difficulty of the following game in response to whether the previous game was won or lost. Difficulty varied between 15 levels and was defined as the amount of production resources available to the opponent entity in the game. This procedure successfully honed win rate (42.6%, SD = 8.8% for SC-1; 43.0%, SD = 8.7% for SC-2). For the Sims condition, participants controlled and developed a single “family household” in a virtual neighborhood. Participants were given 7 weeks (49 days) to complete the study, with the average completion occurring after 43.7 days (SD = 6.24). Further information on materials and methods is available in the
All participants gave informed written consent to participate in this study. Approved by the Institutional Review Board of the University of Texas at Austin.
To compare across the various cognitive tasks, and account for the effects of re-test learning, we contrasted post-test enhancements in the RTS subconditions against the control life-simulator condition using diffusion modeling when appropriate, and applying a meta-analytical Bayesian technique
To achieve a meta-analytical Bayes factor approach, we computed a single performance metric for each task. For tasks in which there is accuracy and response time measures, we use diffusion modeling to combine these two measures into a single measure of performance. Diffusion modeling is an established modeling technique for tasks which permit a speed/accuracy tradeoff
Using a meta-analytical Bayesian technique that aggregated across multiple
The
Task Group | Task | SC | SC-1 | SC-2 |
Cognitive Flexibility | Stroop | 3.06 (0.53) | 1.32 (0.31) | 2.84 (0.70) |
ANT | 2.28 (0.52) | 0.86 (0.32) | 2.24 (0.65) | |
Task Switching | 0.92 (0.18) | 0.98 (0.31) | 0.23 (0.08) | |
Multi-Location Switching | 0.98 (0.20) | 0.69 (0.17) | 0.68 (0.19) | |
Ospan | 0.81 (0.13) | 0.56 (0.15) | 0.61 (0.12) | |
Other | BART | −0.56 (−0.08) | −0.39 (−.08) | 0.61 (−0.08) |
Visual Search | −0.58 (−0.13) | 0.42 (0.09) | −1.29 (−0.29) | |
Information Filtering | −1.00 (−0.20) | −0.29 (−0.08) | −1.34 (−0.34) | |
Digit span | −1.76 (−0.22) | −2.66 (−0.47) | −0.25 (−0.04) |
Effect sizes (Cohen's
Task | Session |
Control | SC-1 | SC-2 |
Stroop | Pre | 0.905 (0.183) | 0.871 (0.164) | 0.667 (0.119) |
Post | 0.943 (0.193) | 1.106 (0.190) | 1.229 (0.169) | |
ANT | pre | 0.019 (0.002) | 0.017 (0.002) | 0.015 (0.002) |
post | 0.024 (0.002) | 0.025 (0.002) | 0.024 (0.002) | |
Task Switching | pre | 0.011 (0.001) | 0.011 (0.001) | 0.011 (0.001) |
post | 0.013 (0.001) | 0.013 (0.001) | 0.012 (0.001) | |
Multi-Location Switching | pre | 0.006 (0.000) | 0.006 (0.001) | 0.006 (0.001) |
post | 0.006 (0.001) | 0.007 (0.001) | 0.007 (0.001) | |
Ospan | pre | 51.60 (3.49) | 40.65 (3.85) | 44.30 (3.90) |
post | 55.70 (4.77) | 47.35 (4.69) | 50.00 (3.96) | |
BART | pre | 28.66 (2.89) | 34.98 (2.17) | 27.78 (1.94) |
post | 33.05 (2.45) | 38.52 (3.18) | 31.24 (2.03) | |
Visual Search | pre | 0.014 (0.001) | 0.014 (0.001) | 0.012 (0.001) |
post | 0.018 (0.001) | 0.019 (0.001) | 0.016 (0.001) | |
Information Filtering | pre | 0.009 (0.001) | 0.009 (0.001) | 0.009 (0.001) |
post | 0.011 (0.001) | 0.010 (0.001) | 0.009 (0.001) | |
Digit span | pre | 0.391 (0.176) | 0.183 (0.211) | 0.065 (0.154) |
post | 0.948 (0.205) | 0.367 (0.193) | 0.587 (0.189) |
There were no significant pre-test differences for the StarCraft groups combined versus the Sims control group.
All main predictions held. For the cognitive flexibility task grouping, the Bayes factor for RTS game play (StarCraft) versus the control game (The Sims) was 40.76. As predicted, benefits for RTS gaming only held for the cognitive flexibility task grouping. For the other tasks, the Bayes factor for RTS game play (StarCraft) versus the control game (The Sims) was 0.02, which signifies very strong evidence for the null hypothesis. When comparing the SC-1 and SC-2 subconditions directly, no strong evidence for differences was found. However, when considering the cognitive flexibility task grouping, the SC-2 subcondition did differ from the Sims control game (Bayes factor of 6.77), whereas the evidence for SC-1 subcondition differing from the Sims control game was inconsequential (Bayes factor of 1.17). Overall, these results are highly supportive of our predictions – RTS gaming selectively promotes cognitive flexibility, particularly under conditions in which players must rapidly switch between contexts while maintaining memory for both contexts.
One key question is whether an underlying dimension of cognitive flexibility emerges or strengthens as a result of video game experience. A principal component analysis (PCA) on the absolute cognitive task scores (as opposed to post-test minus pre-test difference scores as compared to the Sims control condition) revealed the main underlying performance components within each game group. At pre-test, neither the Sims nor the SC group (SC-1 and SC-2) had a pattern of primary principle components, which aligned with the hypothesized cognitive flexibility task distinction. However, at post-test, the SC group's first principle component organized along the hypothesized cognitive flexibility task distinction. To demonstrate the emergence of this pattern of focused improvement on cognitive flexibility, the vector of PCA component contributions for each task were correlated with the hypothesized contribution factor vector (1 for cognitive flexibility tasks, 0 for unrelated tasks). For pre-test, the correlations for the Sims and SC conditions were not significant,
Taken together, the PCA and meta-analytical BF results reveal that the RTS video game training paradigm successfully enhanced and aligned the cognitive performance of the group along a new and unified dimension of cognitive flexibility. Forty hours of RTS video game training was sufficient to create dramatic changes in players' cognitive flexibility.
Our previous analyses indicate that those in the SC-2 subcondition, but not necessarily those in the SC-1 subcondition, benefited from enhanced cognitive flexibility as a result of RTS training. We originally hypothesized that RTS gaming involved the practice of maintaining, assessing, and coordinating between multiple information and action sources simultaneously. In order to verify that the SC-2 subcondition stresses these operations more than the SC-1 subcondition, we analyzed players' behavior within the RTS game. In particular, we examined players' use of game features to test the hypotheses that SC-2 players simultaneously process more within game information than SC-1 players. See the
During RTS game play, game state features and user behavior were recorded every 250 milliseconds. Game features (e.g., whether a unit is being attacked) are used to predict which unit is selected by the player. By constructing the selection/feature analysis for time step lags into the past, it became possible to determine which features in the near past drove user selection in the present. In this way, we implement Bayesian model selection to determine how many features a player entertains.
As time progresses, fewer gaming features from the past influence current game play behavior. The SC-2 group attended to more features and at post-test exhibited enhanced cognitive flexibility performance relative to the Sims group.
The present study finds that cognitive flexibility is a trainable skill. Forty hours of training within an RTS game that stresses rapid and simultaneous maintenance, assessment, and coordination between multiple information and action sources was sufficient to affect change. As a result of RTS game experience, an underlying dimension of cognitive flexibility emerged and characterized individual differences in performance on a variety of laboratory tasks.
Secondary analyses involved comparing SC-1 and SC-2 subconditions. Comparing the SC-1 and SC-2 individually to the baseline control game, SC-2, which involved maintaining awareness of and switching between two spatially separated (out of view) bases, was particularly effective in boosting players' cognitive flexibility. By recording real-time gaming data, we were able to compare the two StarCraft groups in terms of the number of significant features attended during game play. This has revealed that in fact the SC-2 gaming setup led to more attended features overall than SC-1.
This delineation of gaming environment and the resulting gaming behavior is a novel demonstration that begins to clarify which gaming attributes are important for meaningful cognitive change. We have shown that training with a sufficient level of simultaneous information and action coordination in a real-time video game leads to the specific enhancement of higher-level cognition along a clear unified component. With the ability to control and quantify specific video game parameters and behavior, we have shown that it is possible to alter cognitive flexibility, a core component with broad influence on the psychological abilities and well-being of an individual. We have also shown that only one version of RTS gaming led to cognitive flexibility enhancements while another did not. The SC-2 subcondition was designed to require even more switching and maintenance of information. Modeling of the gaming behavior indicated that those in the SC-2 subcondition were attending to more gaming features. These aspects of RTS gaming may have been the critical elements of the video gaming training regimen required to elicit a cognitive flexibility enhancement.
One key direction for future research is to detail the brain basis for these dramatic changes in behavior. Promising avenues include structural magnetic resonance imaging (MRI) to explore volumetric changes to brain regions following video game experience, as well as diffusion tensor imaging (DTI) to explore changes in connectivity between brain regions as a result of training. Given the nature of cognitive flexibility, we expect that changes will not be localized to one region. Understanding the brain basis for the behavior change may suggest how to best pair varieties of game experience with populations seeking to improve cognitive flexibility. Applications of this innovation include the development of clinical regimens to target deficits in populations with specific cognitive flexibility or executive functioning dysfunction. These include attention deficit hyperactivity disorder (ADHD)
(DOC)
(TIF)
Latin square task counterbalancing (MMI = Multimedia Multitasking Index, BART = Balloon Analog Risk Taking, TS = Task Switching, DS = WAIS-IV Digit Span, ANT = Attention Network Test, Ospan = Operating Span, IF = Information Filtering, VS = Visual Search, MLM = Multi-location Memory).
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
(DOCX)
The authors thank Marc Tomlinson, Devon Greer, Rusty Gomez, Reid Tissing, and Kristen Crabtree for their help in making this project possible.