Does Mental Math Improve Cognition? The Research
Every math app markets a sharper brain, and two decades of cognitive research say the honest picture is smaller and more specific than that pitch. Mental math practice reliably improves mental math itself, sometimes lifts working memory in children under twelve, and almost never transfers to unrelated cognitive tasks. This piece walks through the studies that survived replication, the ones that did not, and where the evidence still runs thin.
What Counts as Transfer, and Why It Matters
Cognitive scientists split training gains into three tiers. Near transfer means the skill improves on tasks that share structure with the drill, such as faster two-digit addition after two-digit addition practice. Intermediate transfer means related but distinct tasks improve, such as working memory span after weeks of arithmetic work. Far transfer, the marketing claim, means unrelated abilities like fluid intelligence or reading comprehension rise. The 2010 Owen et al. study in Nature tested 11,430 adults on six weeks of online brain training and found near transfer only, with zero measurable far transfer. That result set the ceiling for every commercial brain-training claim since.
What Mental Math Reliably Does
Arithmetic drills reliably speed up arithmetic. Randomized trials in children aged 6 to 12 show that daily ten-minute mental math sessions cut single-digit response times by 30 to 50 percent within eight weeks, and the effect holds four months after training stops if practice continues twice weekly. The same interventions lift standardized math achievement scores by roughly 0.4 standard deviations, which matches the effect size of one full grade level of extra classroom instruction. For adults, six weeks of daily practice on the kinds of drills covered in our two-digit multiplication guide rebuilds fluency lost after school, though gains plateau near prior peak rather than exceeding it.
Where Working Memory Enters the Picture
Working memory is the mental scratchpad that holds intermediate results, and it is the bottleneck for almost every mental calculation past two digits. Studies on Japanese abacus (soroban) students show that after two years of daily anzan practice, working memory span rises by 1 to 2 items on backward digit-span tests, with fMRI evidence of increased right parietal activation during mental calculation. The catch is that this gain shows up in children who trained for hundreds of hours before age 12. Adult studies at similar intensity show smaller working memory gains, typically 0.2 to 0.4 standard deviations, and only on numerical tasks. Formats that force you to hold three or four numbers at once, like the three-slot scratchpad drill in Mathness, tap the same mechanism without demanding two thousand hours of soroban.
The Far-Transfer Claims That Do Not Hold Up
The strongest brain-training claims lose their shape under replication. Melby-Lervåg and Hulme's 2013 meta-analysis of 23 working-memory training studies found reliable near transfer, small and short-lived intermediate transfer, and no far transfer to nonverbal ability, verbal ability, or arithmetic outside the trained format. Their 2016 follow-up covering 87 studies confirmed the pattern. A 2017 Cochrane review of computerized cognitive training in older adults reached the same conclusion: trained tasks improve, untrained cognition does not. The commercial claim that daily puzzles raise IQ, delay dementia, or lift school performance in non-math subjects has no replicated evidence. The tighter claim, that mental math practice improves mental math, is well supported.
- Single-digit arithmetic speed rises 30 to 50 percent after 8 weeks of daily drill.
- Standardized math scores rise about 0.4 SD in trained children aged 6 to 12.
- Working memory span rises 1 to 2 items in children after 2+ years of abacus training.
- Adult numerical working memory gains sit at 0.2 to 0.4 SD, and only on numerical tasks.
- All gains fade within 6 to 12 months once practice stops entirely.
- No replicated transfer to reading, verbal reasoning, or fluid intelligence.
Why Targeted Practice Beats Puzzle Apps for Most Adults
The mechanism behind every replicated gain above is retrieval practice on the specific arithmetic operations you want to speed up. Anders Ericsson's deliberate-practice framework predicts this precisely: skills improve where feedback is tight and repetition is targeted, and they do not improve elsewhere. A ten-minute daily session on a focused format like the Mathness daily puzzle produces measurable arithmetic gains inside a month if the operations pushed match the operations you want faster. Apps that shuffle across memory, attention, and reasoning tasks spread the retrieval too thin to produce reliable near transfer, and by the meta-analytic evidence they produce no far transfer either. Depth on one skill beats a scattered survey of ten.
The Honest Verdict for Parents, Students, and Adults
For children between 6 and 12, daily arithmetic practice is one of the highest-return uses of ten minutes, both for math achievement and for the modest working-memory bump that repeated calculation seems to produce. For students preparing for standardized tests, targeted mental math drills raise the arithmetic subscores where speed is graded, and nothing else. For adults chasing sharper thinking, the evidence supports faster arithmetic and a satisfying sense of fluency, not a broader cognitive lift. The right frame is training a specific skill, the way running trains running. If the goal is faster mental math, pick a format that pushes retrieval under a clock, such as the target-number puzzle mechanic that forces bidirectional search on every round.
