A more causal question, with a small evidence base
Many gaming studies compare existing players with non-players. A 2019 systematic review by Brilliant T, Nouchi, and Kawashima focused instead on intervention studies, where participants trained on video games and researchers measured brain structure or function.
That focus gets closer to causal evidence. Only nine studies met the review’s criteria, and their differences made a pooled effect estimate inappropriate.
What qualified for the review
The review searched PubMed and Google Scholar for randomized controlled trials published after 2000. It included healthy participants and non-cognitive commercial-style games, then required MRI evidence related to brain structure or function.
The nine retained articles covered:
- 3D adventure games;
- first-person shooters;
- puzzle games;
- rhythm dance games;
- strategy games;
- training periods from 16 to 90 hours;
- participants from different age groups;
- different structural and functional MRI outcomes.
The authors rated the included evidence as fair quality using their selected criteria. They chose a systematic narrative review rather than a meta-analysis because the participants, games, and outcomes varied.
What the studies reported
Across the included interventions, studies reported changes in selected structural or functional brain measures after training. The affected regions and direction of change differed by game and protocol.
That variation is not noise to hide. It is part of the answer. “Video games” is not one intervention, just as “exercise” is not one exercise programme.
The full Brain Sciences review provides the inclusion rules and study table.
Why 16 hours is not a prescription
The paper concluded that observed effects appeared after at least 16 hours in the included studies. That is a description of this selected literature, not a proven minimum dose.
It does not establish that:
- every game produces a beneficial change after 16 hours;
- more training is always better;
- a structural or functional MRI change improves daily performance;
- effects persist after training ends;
- children and adults respond in the same way;
- gaming is appropriate treatment for a health condition.
A threshold claim would require comparable interventions across several doses, consistent outcomes, and replication.
How this review differs from the 116-study review
The broader Palaus review combined 116 articles and mapped a wide field of neural correlates, including experimental and correlational work. This 2019 review asks a narrower question about interventions in healthy participants and retains nine studies.
The two articles therefore serve different reader needs:
- use the 116-study review to understand the range and heterogeneity of gaming neuroimaging;
- use this nine-study review to understand why training studies offer stronger causal designs but a much smaller, still varied evidence base.
What it means for Saiki
None of the nine interventions validates Saiki’s product rules. The review does not test League match history, self-report personality domains, champion pools, vision score, assists, deaths, or concordance between questionnaire and gameplay.
Brain plasticity after a game-training intervention and personality inference from routine match statistics are different claims. Evidence for the first cannot be borrowed to prove the second.
Saiki should therefore avoid describing its gameplay indicators as neuroscience-backed measures. The accurate statement is narrower: Saiki applies transparent, experimental rules to recent match statistics and displays them beside self-report scores.
The practical conclusion
Intervention studies are the right direction when the question is whether game training changes a measured outcome. This 2019 review found promising MRI results across nine studies, alongside substantial variation and methodological limits.
The next useful evidence would use preregistered outcomes, active control groups, adequate samples, consistent follow-up, and tests of whether any measured change transfers to daily tasks. It would still not replace separate validation for a personality product.