Digital mahjong nudged memory in older adults

Gamified digital training is often sold as a way to keep ageing brains sharp, and games that are culturally familiar may hold attention far better than abstract brain-training drills. The problem, as the authors point out, is that randomised evidence linking behavioural change to convergent neuroimaging outcomes remains limited: most claims rest either on test scores with no biological corroboration, or on imaging with no functional payoff. This trial asked whether a home-based digital mahjong programme, bundled with an incentivised walking component, could improve cognition in community-dwelling older adults and whether any change was visible in the brain.

This was a single-blind, parallel-group randomised controlled trial run in Taipei between July 2020 and November 2021. Sixty community-dwelling adults aged 55 years and older were randomised 1:1 to six months of a digital mahjong intervention — two 30-minute sessions per week on an iPad app, with Mi Band step tracking earning in-game tokens — or to no intervention. Primary outcomes were multimodal MRI measures: grey matter volume, regional homogeneity, fractional amplitude of low-frequency fluctuations and global brain connectivity. Secondary outcomes covered cognition, self-reported physical activity, resilience and demoralisation, quality of life, fitness, body composition and blood biomarkers, analysed per protocol.

Completion was high, at 29 of 30 in the intervention group and 27 of 30 in controls. The intervention group improved more on Montreal Cognitive Assessment delayed recall, a between-group difference of 0.79 points (95% CI -1.538 to -0.042; P=.04). Imaging showed increased regional homogeneity in the right insula, decreased grey matter volume in the left frontal pole and orbitofrontal cortex, and reduced fALFF in the left frontal pole. Change in delayed recall tracked right insular regional homogeneity, and insular ReHo correlated with total practice sessions (r=0.48; P=.009). Diastolic blood pressure fell 3.50 mm Hg more in the intervention group; thyroid-stimulating hormone dropped but stayed within the normal range.

The study extends the cognitive-training literature by pairing a modest behavioural gain with convergent functional imaging and a dose relationship, which is more than most app studies offer. It also complicates a simple story: grey matter volume decreased in frontal regions rather than increasing, which is not the direction a naive 'use it and grow it' model would predict. The authors themselves call for large-scale trials using active control conditions and factorial designs to isolate which component — the game, the social framing or the walking incentive — is doing the work. Whether a sub-point change on a screening test matters for everyday memory is unresolved.

This is a reason to think that a culturally familiar, enjoyable cognitive challenge played regularly at home is at least feasible for older adults and may be associated with small memory and blood pressure changes over six months. It is not evidence that a mahjong app prevents dementia, and the trial cannot tell you whether the game, the step-tracking incentive or simply having a structured activity produced the effect. The dose correlation is intriguing, suggesting more practice sessions went with greater functional change in the insula, but correlation within a small trial is fragile. Read it as an encouraging feasibility signal awaiting a bigger, better-controlled test.