Exercise Was Linked to a 0.47 Cognitive Effect in 12-Study Esports Performance Review

A new systematic review pooled 12 controlled studies of exercise and esports performance and found a measurable effect on cognitive and psychomotor tests. The evidence that those gains carry over into actual esports performance remained imprecise, and the authors graded the whole body of work as very low certainty.

Esports performance research has grown fast enough to need periodic re-pooling, and an updated review published in Scientific Reports on October 24, 2026 re-ran the numbers with each study treated as the independent unit of analysis.

The work comes from Wen Jie Wu, ZiMei Zhong and Rui Liu, based at Hezhou University in Hezhou, Guangxi, China, with the corresponding author also affiliated with Universiti Sains Malaysia in Penang. The team searched PubMed, Scopus and PsycINFO through May 2026, added supplementary searching on 3 August 2026, and re-ran the search during revision through 4 September 2026 without finding further eligible studies.

Twelve controlled study clusters met the inclusion criteria, covering 660 participants enrolled or randomized, of whom 608 were reported as analysed. Nine of the twelve evaluated exercise-only interventions, but only five of those were estimable for quantitative synthesis, contributing 184 enrolled and 182 analysed participants across 18 outcomes.

Pooled across those five studies, cognitive and psychomotor outcomes produced a Hedges g of 0.465, with a 95 percent confidence interval of 0.073 to 0.856. Outcomes measured on the game itself were larger but imprecise, at 0.574 with an interval running from minus 0.002 to 1.150 — an interval that crosses zero.

What the Esports Performance Review Actually Pooled

The distinction between the twelve included clusters and the five pooled ones matters for reading any claim about esports performance, because the two sets support very different strengths of statement. Seven of the twelve contributed to the narrative synthesis only: four further exercise trials, two trials of active virtual reality, and one multicomponent intervention.

That leaves a quantitative core of five studies and roughly 182 analysed participants on which every pooled esports performance estimate rests. Within each study, the authors aggregated multiple outcomes assuming a between-outcome correlation of 0.50, a standard approach when a single trial reports several related measures of esports performance.

The Gap Between Cognition and Gameplay

The headline split in this esports performance review is between what a laboratory test measures and what a match measures. Cognitive and psychomotor outcomes — reaction time, attention, processing speed and similar tasks — reached statistical significance, with the interval for Hedges g sitting entirely above zero at 0.073 to 0.856.

The outcomes drawn directly from esports performance in play did not. The point estimate was actually higher at 0.574, which is the kind of result that invites over-reading, but the interval of minus 0.002 to 1.150 includes no effect at all. On the measures closest to real esports performance, the review cannot rule out that exercise changes nothing.

The Composite That Looks Strongest and Means Least

The tidiest number in the paper is an exploratory cross-domain omnibus composite of 0.512, with a confidence interval of 0.214 to 0.810 across five studies. It is narrower than either of the domain-specific estimates and sits comfortably above zero.

It is also the estimate to lean on least. Combining cognitive tests and in-game measures into one composite answers a question nobody asks about esports performance, because a player does not care about an average of their reaction time and their match result. The authors label it exploratory for exactly that reason, and the domain-specific estimates remain the ones that map onto a real decision.

Read in order, the three numbers describe a familiar shape in a young literature: the broadest construct produces the most confident-looking estimate, and the estimate closest to what matters is the one still compatible with no effect on esports performance at all.

Why the Heterogeneity Statistics Read as Zero

Both I-squared and tau-squared were estimated as zero in the quantitative models, which in a larger review would suggest the included trials were telling the same story. The authors explicitly decline that interpretation.

They note that with only three to five independent studies these heterogeneity statistics have low power, and that the result should not be interpreted as evidence that effects are truly homogeneous. It is a reminder that in small pooled sets of esports performance trials, a clean-looking consistency figure can simply reflect a shortage of data.

The Seven Studies That Could Not Be Pooled

Seven of the twelve clusters sat outside the meta-analysis, and their composition shapes how much of the esports performance field this review really covers. Four were further exercise trials, two tested active virtual reality, and one used a multicomponent intervention.

Active virtual reality is a particularly awkward fit. It is physical activity and it is gaming at once, so a trial of it cannot cleanly separate the exercise from the screen time, which is one reason it was handled narratively rather than pooled with the conventional exercise arms.

What the Review Declines to Recommend

Readers looking for an esports performance protocol will not find one here. The authors state that the repeated-exercise subgroup, which contained just two studies, is descriptive only, and that no definitive performance prescription or acute-versus-repeated superiority claim is supported.

In other words, the review cannot say whether a single bout before competition or a sustained training block does more for esports performance, nor how long, how hard or what kind. That question is still open, and the review is candid that its data on esports performance cannot close it.

How Esports Performance Findings Sit Alongside Other Research

The cognition-versus-execution gap that runs through esports performance research shows up elsewhere in exercise science. A trial in 78 children found that coordination training beat cardio on working memory scores, which is a closer analogue to the cognitive outcomes pooled here than to anything measured in a game.

Research on skilled athletes shows how exacting the performance side of the equation is. One study put reaction timing errors in 312 combat sports athletes at an average of 32.4 milliseconds, and a tennis trial found that serve accuracy fell only once workload reached 80 percent. Esports performance measures face the same difficulty of isolating a real change from ordinary match-to-match variation.

The measurement problem is familiar too. A scoping review of countermovement jump testing found that protocols varied widely across 117 studies, which is roughly the stage that esports performance testing appears to be at now.

What This Does and Does Not Mean for Players

Nothing in this review argues against exercise for competitive gamers, and nothing in it supports exercise as an esports performance lever with a known dose. The honest summary is that the cognitive side has a signal worth following and the gameplay side has not yet been demonstrated.

There is also a simpler case that sits outside the esports performance question entirely. The general health benefits of regular activity apply to people who sit for long competitive sessions at least as much as to anyone else, and those benefits do not depend on whether a match result improves.

A few general points follow from how this review was built:

• A pooled estimate from five small studies is a starting point for further research, not a training prescription.
• Laboratory reaction-time gains and in-game results are different outcomes and should not be reported as one.
• Anyone adding a new exercise routine around long seated sessions should raise it with their own clinician, particularly if they have existing joint, back or cardiovascular concerns.