Your brain gets quieter when it gets sharper

You might assume that a better-performing brain is a more active brain. More neurons firing, more regions lighting up. A meta-analysis of 45 neuroimaging studies suggests the opposite. Li et al. (2026) found that computerized working memory training produces moderate cognitive gains — with a specific brain region becoming less active, not more. The effect size across studies was consistent and measurable. And the neural signature of improvement looks more like efficiency than effort.

The mechanism

Li et al. (2026) conducted a meta-analysis published in npj Digital Medicine. They pooled data from 45 studies that combined behavioral working memory assessments with neuroimaging. The studies used computerized training programs — structured digital exercises designed to tax and expand working memory capacity.

The behavioral result was clear:

→ Effect size g = 0.503 — a moderate, statistically reliable improvement in working memory performance.

But the neuroimaging data revealed something more specific. After training, participants showed decreased activation in the left angular gyrus. This region sits at the junction of the temporal and parietal lobes. It is involved in semantic processing, attentional reorientation, and the integration of information from different modalities.

Decreased activation here does not mean the region stops working. It suggests increased neural efficiency. The brain accomplishes the same task — or a harder one — with fewer resources. Think of it as the difference between a novice driver who grips the wheel with full concentration and an experienced driver who navigates automatically. The task is the same. The neural cost drops.

This pattern aligns with what neuroscientists call the "neural efficiency hypothesis." As a cognitive skill becomes more practiced, the brain shifts from effortful, widespread activation to streamlined, localized processing. The prefrontal and parietal networks still engage during working memory tasks. But training appears to reduce unnecessary recruitment of supporting regions like the angular gyrus.

A second meta-analysis strengthens this interpretation. Li et al. (2025), published in npj Aging, analyzed 24 neuroimaging studies on cognitive training in older adults. They found a comparable effect size:

→ g = 0.38 — moderate cognitive improvement.

That study also identified a neural change — but in the opposite direction for a different region. Older adults showed increased activation of the left inferior frontal gyrus after training. This region is central to executive control and verbal working memory. The finding suggests the brain compensates for age-related decline by recruiting frontal resources more effectively after structured practice.

Two meta-analyses. Different populations. Different neural signatures. But the same behavioral outcome: structured computerized training produces measurable working memory improvement.

The biological mechanism that connects these findings likely involves synaptic strengthening in frontoparietal networks. Repeated engagement of working memory circuits promotes long-term potentiation — the cellular process through which synaptic connections become more efficient. Over training sessions, this means faster signal transmission and reduced metabolic cost for the same cognitive operation.

Chai et al. (2025), in a meta-analysis of 20 studies published in Scientific Reports, adds a relevant layer. They identified consistent neural substrates through which physical exercise enhances executive function in healthy populations. The key regions overlap with those modified by cognitive training — particularly frontal networks involved in attentional control and inhibition. This convergence suggests that physical exercise and digital cognitive training may act on the same neural infrastructure through different pathways. Combined interventions could produce additive or synergistic effects on working memory.

This finding has been replicated across multiple studies. The primary meta-analysis aggregates 45 independent neuroimaging studies, and the confirmation from Li et al. (2025) across 24 additional studies with a comparable effect size (g=0.38) indicates the effect is robust and reproducible across different populations and training paradigms.

Training your working memory where you actually work

Knowledge workers operate in cognitively fragmented environments. A study published in the Journal of Public Health (2026) documented that professionals show differential mental workload and cognitive performance across activity-based offices, trains, and remote setups. Working memory is taxed differently depending on where you sit, who interrupts you, and how often you switch contexts. This means the environment where training matters most is also the environment where training is most accessible. A structured 20-minute digital session during a commute or between deep work blocks fits precisely into the gaps that already exist in a knowledge worker's day. The data from Li et al. (2026) does not specify minimum effective dose — but the studies pooled in the meta-analysis typically used sessions of 20-40 minutes, 3-5 times per week, across 4-8 weeks.

The neural efficiency finding has a specific implication for high-stakes decision-making. If training reduces the metabolic cost of holding information in working memory, it frees up cognitive bandwidth for higher-order processing — evaluation, comparison, strategic thinking. For someone running a team or a company, this translates to clearer thinking under pressure. Not because the problems get simpler. Because the brain spends fewer resources on the foundational operation of keeping relevant information active. The g=0.503 effect size is moderate. It will not double cognitive capacity. But in decisions where the margin between a good call and a bad call is narrow, a moderate improvement in the underlying hardware is not trivial.

The convergence between cognitive training and physical exercise identified by Chai et al. (2025) suggests a practical protocol. Someone who already exercises regularly may amplify the cognitive returns by adding structured digital working memory training. The two interventions appear to target overlapping frontal networks through different mechanisms — neurochemical via exercise, synaptic via repeated cognitive engagement. A 30-minute workout followed by a 20-minute working memory session could represent a combined investment of less than an hour with compounding neural returns. This is not proven as a combined protocol — no study has tested this specific pairing yet. But the mechanistic overlap is strong enough to warrant self-experimentation with measurable tracking.

The source

Li et al. (2026) — npj Digital Medicine

DOI: 10.1038/s41746-026-02478-9

Li et al. (2025) — npj Aging

Chai et al. (2025) — Scientific Reports

(2026) — Journal of Public Health

This content is informational and does not constitute medical advice.