Brain Training and Intelligence: Does Cognitive Training Transfer?
Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy
Research cluster: Intelligence, Reasoning & Cognitive Abilities · Written by Artificial Sapiens
Brain training promises an appealing outcome: practice a few demanding mental exercises and become better at thinking in general. The central scientific question is whether improvements carry beyond the activity being practiced. Researchers call this cognitive transfer. After decades of experiments, the most defensible answer is that brain training reliably improves performance on trained exercises and can sometimes help closely related tasks, while convincing, durable improvements in general intelligence or independently measured IQ remain unestablished. Evidence varies by training method, study design, population, and outcome.
This review examines what studies actually measure, why positive early findings attracted attention, why controlled meta-analyses frequently reach more restrained conclusions, and how to evaluate a brain-training program without confusing game scores with general cognitive ability.
Does brain training increase IQ? The evidence-based answer
For healthy people, there is currently no convincing general evidence that commercial brain-training games or repeated working-memory exercises produce meaningful, lasting improvements in overall intelligence. The strongest recurring effect is learning to perform the exercises themselves. Some studies report benefits on untrained tests of working memory or specific executive functions, but robust far transfer to fluid reasoning, academic achievement, or everyday cognitive functioning is much harder to demonstrate.
In a review of 87 publications containing 145 experimental comparisons, Melby-Lervåg, Redick, and Hulme (2016) found short-term improvement on working-memory measures but no convincing gains in nonverbal ability, verbal ability, reading, or arithmetic when trained participants were compared with active controls. A later second-order meta-analysis by Sala and colleagues (2019) similarly concluded that distance from the practiced activity matters: apparent benefits shrink as the outcome becomes less like the exercise.
More recent research reinforces this distinction. A 2024 meta-analysis by Rodas, Asimakopoulou, and Greene found a small improvement in working memory overall, a much larger effect when tests resembled training exercises, and no overall improvement in fluid intelligence. An active-control 2025 randomized trial with 103 healthy older adults observed improvement across trained app exercises but found no corresponding advantage on independent tests of working memory, processing speed, attention, or language.
This is a conclusion about current evidence, populations, programs, and outcome measures. It leaves room for effective training of particular skills and for targeted interventions under particular conditions. It does not justify an unconditional claim that every activity marketed as brain training is useless.
What counts as brain training?
Brain training, also called cognitive training, refers to repeated, usually structured practice intended to improve a specified cognitive operation. The exercises may involve remembering a sequence, rapidly identifying patterns, dividing attention, switching rules, inhibiting automatic responses, or solving puzzles under time pressure. Computerized programs often adapt difficulty to performance and report progress through scores, levels, or personalized dashboards.
A useful distinction separates four kinds of activity. Task-specific practice is repetition aimed at getting better at a particular exercise. Cognitive training programs aim to improve an underlying process, such as working memory or attention, through repeated tasks. Strategy training explicitly teaches methods for handling a real task, such as remembering names with mnemonic cues. Cognitive remediation is a more individualized rehabilitation approach that may combine training, strategy learning, coaching, and real-world functional goals for a particular clinical population.
These categories can overlap, but their evidence must be evaluated against their intended outcomes. Becoming better at memorizing sequences is a legitimate learning result. An intervention claiming to improve everyday independence, school performance, or overall IQ carries a substantially larger burden of proof.
The underlying idea is plausible: repeated cognitive activity changes performance, and learning is associated with experience-dependent brain changes. The missing step in many popular claims is showing that the change generalizes to abilities and settings outside the trained task. A change in performance after practice is evidence of learning; broader generalization requires separate tests.
Cognitive transfer: training effects, near transfer, and far transfer
The scientific debate becomes clearer when researchers distinguish several outcomes.
Training effects: improvement on exactly what was practiced
If someone practices an n-back task repeatedly, they will often become faster or more accurate at recognizing whether a current stimulus matches one presented several steps earlier. Improvement may reflect a better task-specific strategy, greater familiarity with the rules, reduced uncertainty, motor fluency, motivation, or more efficient handling of the practiced stimuli. It can also involve changes in processes relevant to working memory.
The training score alone cannot determine which explanation is correct. It is an outcome tied to the activity used for practice.
Near transfer: improvement on a different but similar task
Near transfer means that gains appear on an untrained task sharing substantial demands with the practiced task. For example, someone trained on one sequence-memory exercise may do better on another sequence-memory exercise with a different format. Such findings can be meaningful, but their strength depends on how independent the tests truly are.
A task may be described as untrained yet use nearly the same stimuli, responses, timing, or cognitive strategy. The more similar the test and training exercise, the easier it is for familiar procedures to explain improvement.
Far transfer: improvement in broader, substantially different abilities
Far transfer refers to improvement in outcomes that extend beyond the exercised procedures: fluid reasoning measured by independent tests, academic learning, complex decision-making, everyday functioning, or other substantial real-world outcomes. A claim that a brain-training app increases IQ is a far-transfer claim.
Far transfer is precisely what makes general brain training commercially and scientifically consequential. If practicing one game improved reasoning across unfamiliar situations, an inexpensive exercise could have broad effects. The evidence does not currently establish that ordinary brain-training programs consistently deliver this outcome.
Near and far transfer are descriptions along a continuum, rather than fixed categories established by the name of a test. Researchers must justify how much two tasks share and whether the claimed improvement is useful outside the experiment.
Why brain training can improve a score without increasing intelligence
Intelligence is an individual-differences construct describing patterned variation across reasoning, learning, knowledge, and related cognitive performances. General intelligence, often represented by a latent factor called g, summarizes the correlations among multiple cognitive measures. Fluid intelligence describes reasoning in relatively novel situations; crystallized intelligence concerns knowledge and learned skills. Working memory is the ability to maintain and manipulate information for a short period, and executive functions include processes such as goal-directed control, inhibition, and switching.
These concepts are related but not interchangeable. A computer game is a performance task. An IQ result is an observed score obtained from a standardized assessment and interpreted against appropriate norms. A broad psychometric ability is inferred from patterns across multiple measures. A latent g factor is estimated statistically; it is not directly measured by the score on one app game.
If a person repeatedly performs a matrix-like puzzle and later receives a higher score on a similar puzzle, the change could partly reflect learned solution strategies and reduced novelty. To conclude that general reasoning improved, researchers need a suitable range of independent measures, ideally interpreted as reliable change across a latent construct, together with a controlled design and meaningful follow-up. The Standards for Educational and Psychological Testing provide the wider measurement framework for validity and interpretation.
Even a statistically detectable increase on one intelligence test does not automatically establish a durable increase in a person's broad cognitive capacity. Practice, measurement error, regression toward the mean, test familiarity, and the precision of a score can all matter.
Why researchers expected working-memory training to increase fluid intelligence
Working memory and reasoning are strongly associated in individual-differences studies. Solving an unfamiliar problem often requires maintaining information, comparing alternatives, managing interference, and updating intermediate results. This relationship inspired a causal proposal: if training expands working-memory capacity, reasoning may improve in tasks that were never practiced.
That is a testable theory, not a demonstrated identity between working memory and intelligence. Two abilities can correlate because they share several mechanisms, environmental influences, or measurement demands. Improving performance on one working-memory task therefore does not necessarily change all processes responsible for the association.
An influential early experiment by Jaeggi and colleagues (2008) reported improved fluid-intelligence test performance after working-memory training. The study helped establish adaptive dual n-back as a prominent research paradigm. Subsequent investigators examined replication, control conditions, outcome choice, and the durability of the result.
The most informative question has never been simply whether trained participants show better scores afterward. It is whether they improve more than an appropriately matched comparison group on valid, independent outcomes, and whether those gains persist.
What randomized experiments and meta-analyses found
Early positive findings and why conclusions differed
The research record includes positive reports. A meta-analysis by Au and colleagues (2015), examining 20 n-back training studies involving healthy adults, found a small positive average effect on fluid-intelligence tests. This finding matters because it shows why the scientific discussion persisted: some laboratory outcomes initially appeared to support wider transfer.
Interpretation changed when reviews paid closer attention to what comparison groups did, how closely assessment resembled training, small-study effects, and the choice of outcomes. Different reviews were sometimes answering different questions. A synthesis of studies comparing training against no activity estimates an effect that may include motivation, expectations, and routine testing practice. A synthesis emphasizing credible active controls asks what the specific training program adds over and above those nonspecific influences.
These methods do not guarantee that every positive finding is spurious. They explain why an initially encouraging average effect is insufficient evidence for an advertised increase in broad intelligence.
Randomized trials with active controls
A randomized placebo-controlled study by Redick and colleagues (2013) found improvement on practiced working-memory exercises without persuasive improvement in independently measured intelligence. Participants assigned to a demanding comparison activity helped researchers distinguish training-specific changes from general effects of taking part in a cognitive experiment.
An earlier, large online experiment by Owen and colleagues (2010) recruited more than 11,000 participants and showed substantial improvements on practiced brain-training tasks but little evidence of transfer to untrained cognitive tests. Its design and online sample have their own limitations, so it should not be treated as a universal verdict on every intervention. Its central lesson is that impressive within-game progress can coexist with limited generalization.
The 2016 and 2019 meta-analytic conclusions
Melby-Lervåg and colleagues (2016) analyzed a large research base and found that working-memory training did not convincingly improve more distant measures when appropriate control conditions were used. Crucially, the amount by which working-memory performance improved did not predict the magnitude of far-transfer effects across studies. This finding weakens a simple account in which successful working-memory training automatically expands general reasoning ability.
The 2019 second-order meta-analysis evaluated patterns across multiple previous meta-analyses, comparing near and far transfer. It found that training-related improvement was concentrated in similar tasks and that broad far-transfer claims were not supported by robust evidence. A second-order synthesis is a review of reviews; its conclusions inherit some limits of the underlying studies and how the original analyses categorized outcomes.
What newer studies add
The 2024 analysis by Rodas and colleagues evaluated 52 independent comparisons and explicitly assessed bias. The reported standardized mean difference for working-memory outcomes was approximately 0.18, while performance differences were much larger, approximately 1.15, when assessment tasks were closely similar to trained exercises. Fluid intelligence did not show an overall benefit. The review also reported possible improvement in selected executive-function outcomes, showing that the evidence is more nuanced than a blanket claim that all transfer is impossible.
These numbers are standardized group differences in the studies reviewed. They are not IQ points, and they should not be converted into promised individual gains.
A 2025 randomized study by Sutton and colleagues compared three months of an adaptive commercial program with an active control in 103 healthy older adults. Participants improved on trained games, but independent cognitive outcomes did not show corresponding transfer. This is especially informative for the question of whether improving app scores produces a broader benefit measurable outside the app.
Taken together, the best-supported general finding remains substantial task learning and much weaker evidence for far transfer. Particular intervention types and populations still require direct evaluation.
Why study design changes the answer
A good brain-training study needs more than a before-and-after chart. Several details determine how credible an apparent effect is.
Active controls and expectations
If one group practices stimulating games for several weeks and another does nothing, differences at retesting may be influenced by attention from researchers, enthusiasm, computer familiarity, confidence, or belief in the intervention. An active control group performs a credible alternative activity of similar time and engagement. Ideally participants in both groups have comparable expectations about cognitive benefit.
Expectations are difficult to match perfectly because people may recognize their activity. Nevertheless, active comparisons greatly strengthen causal interpretation.
Independent measures and test contamination
An untrained outcome must be meaningfully independent of the practiced activity. Researchers should avoid treating near-identical versions of the same game as proof of general intelligence. A stronger protocol includes several valid, independent measures of reasoning or cognition, ideally with alternate forms and blinded scoring.
Testing too many outcomes creates another problem. If researchers examine dozens of scores and emphasize only statistically significant improvements, chance results can acquire an exaggerated impression of certainty. Prespecified primary outcomes, correction for multiplicity, transparent reporting, and replication reduce this risk.
Randomization, sample size, and attrition
Random assignment makes groups more comparable, but small trials can still produce unstable estimates. Selective dropout can also make an intervention appear successful if participants who struggle are less likely to complete the study. Researchers should report sample sizes, attrition, analysis choices, and the uncertainty around observed effects.
A meaningful claim needs both statistical evidence and a clinically or practically important size of change. A small detectable difference is not automatically a useful improvement in everyday thinking.
Follow-up and real-world outcomes
An immediate posttest cannot tell whether a gain persists for months or years. Studies should distinguish immediate effects, delayed follow-up, and functional changes in daily life. For claims about dementia prevention, academic success, job performance, or independence, those outcomes need direct measurement. Better scores on a training dashboard cannot substitute for them.
The Stanford Center on Longevity's scientific consensus statement specifically warned against overinterpreting evidence used in commercial brain-training claims.
What different kinds of cognitive training can achieve
Dual n-back and working-memory games
Dual n-back asks users to track streams of information and identify repetitions at a specified distance. Adaptive difficulty can make practice demanding, and improvements in n-back performance are unsurprising. Transfer to untrained working-memory tasks is variable; evidence for a meaningful increase in fluid intelligence is especially weak under stricter evaluation.
The result should be described according to the measured outcome. Someone who becomes highly skilled at dual n-back has acquired a skill. That outcome may be personally enjoyable or motivating without constituting proof of higher overall IQ.
Processing-speed and attention exercises
Training rapid visual decisions or specific attention tasks can improve speed and accuracy in related circumstances. Some structured speed-of-processing interventions for older adults have been investigated in clinical and aging research, and their results should be assessed for those specific programs and outcomes.
A measured benefit in visual processing, driving-related testing, or another bounded activity should not be recast as a universal gain in intelligence. General-purpose mobile games and carefully evaluated interventions differ in dose, content, supervision, and outcome measures.
Puzzles, chess, and strategy games
Crosswords can expand familiarity with clues and vocabulary; chess practice can enhance chess knowledge and pattern recognition; strategy games can foster strategies needed within those environments. Broad claims that expertise reliably transfers to unrelated reasoning tests require evidence beyond the obvious gain in expertise itself.
A limited or absent effect on a distant IQ test does not diminish the value of recreation, learning, or cultural participation. The evaluation depends on the promised outcome.
Commercial apps and personalized scores
An app may display a calculated brain age, cognitive index, performance percentile, or composite score. These numbers are often based on the product's own exercises and scoring rules. They should not be interpreted as standardized IQ, clinical status, or an objective measure of overall brain health unless the developer provides strong independent validation for that specific inference.
It is particularly important to distinguish the app's progress metric from a validated external outcome. Reaching a higher level in a game shows that a user is getting better at that game; generalized benefit remains a separate hypothesis.
Does cognitive training work better for older adults?
Older adults are frequently the audience for brain-health marketing, and this raises distinct questions. A successful training program might target a practical goal such as learning a compensatory memory strategy, managing everyday routines, or performing a specific task more efficiently. A broader claim that routine computer games prevent age-related decline or dementia requires much stronger long-term evidence.
A Cochrane review of computerized cognitive training lasting at least 12 weeks found low-quality evidence for small immediate effects in certain outcomes among cognitively healthy older adults. It did not establish lasting benefits in global cognition at one year. The review also emphasized limitations in the evidence base and differences among intervention types.
The US National Academies report on preventing cognitive decline and dementia recognized promising but incomplete evidence for some interventions, including a particular speed-of-processing training approach. The report should be read as a cautious examination of selected interventions, not as scientific endorsement of every commercial cognitive app.
The active-control Sutton et al. (2025) trial provides an additional reminder that sustained practice in later life can increase task performance without measurable gains on other cognitive tests.
Maintaining meaningful activities, physical health, social participation, appropriate treatment of health conditions, and opportunities for learning can be worthwhile for many reasons. The value of these pursuits does not depend on claiming that they add a specific number of IQ points.
Brain training, cognitive remediation, and clinical treatment are different questions
Cognitive remediation for people with schizophrenia or other defined clinical needs may combine cognitive exercises, strategy instruction, a therapist or facilitator, and structured transfer to practical tasks. Success may be measured through functional capacity, achievement of individual goals, or particular cognitive outcomes. The appropriate evidence base is condition-specific clinical research.
That research cannot simply be imported into claims that consumer brain games increase general intelligence in healthy adults. Conversely, null far-transfer findings in healthy volunteers do not by themselves invalidate every targeted rehabilitation approach.
For a separate review of this clinical intervention, see Cognitive Remediation for Schizophrenia: Evidence and Practical Goals.
A person noticing new or worsening difficulties with memory, attention, or daily functioning needs an appropriate health evaluation rather than assuming that a game score identifies the underlying cause. Cognitive weaknesses, screen-based performance, and IQ scores do not independently establish a medical or psychiatric diagnosis.
Can brain training improve school achievement?
Improvement in a memory exercise does not prove that reading comprehension, mathematics, or grades will improve. School learning depends on domain knowledge, quality of teaching, practice with relevant material, feedback, motivation, language, and many contextual factors. An intervention designed to improve classroom outcomes needs studies measuring those outcomes directly.
The Melby-Lervåg et al. meta-analysis did not establish reliable far-transfer benefits for reading or arithmetic from working-memory training under credible control conditions. Claims about academic improvement therefore need program-specific evidence, age-appropriate measures, sufficient follow-up, and a comparison with effective direct instruction or other educational activities.
When the actual goal is mastering algebra, reading more fluently, or understanding a subject, training and feedback in the relevant domain offer a much clearer mechanism of improvement than practicing an unrelated puzzle and hoping for transfer.
Can brain training change the brain?
Learning often involves changes in neural activity and networks, and researchers can detect functional or structural correlates of practice. Neuroplasticity describes the nervous system's ability to change in response to experience; it does not guarantee any particular improvement in IQ or everyday cognition.
Changes in brain activation during a trained task can reflect increasing efficiency, shifting strategies, novelty, effort, or other aspects of task learning. A scan showing a different activation pattern after training is evidence of a neural difference under particular measurement conditions. It is not, by itself, proof of increased general intelligence, lower dementia risk, or better daily functioning.
The causal sequence must be tested at every stage: intervention, neural change, independently measured cognitive performance, and a meaningful real-world outcome. Researchers also need to rule out alternative explanations and establish whether any observed change persists.
Why more hours or greater difficulty do not guarantee far transfer
An adaptive program makes exercises harder as users improve. This helps keep the task challenging and can be useful for training a target skill. It does not follow that greater difficulty or more training sessions automatically raise general intelligence.
The 2024 Rodas meta-analysis reported that differences in program dosage and training type did not explain training effectiveness in the meta-regressions examined. This does not establish that duration never matters; it shows that a simple more-is-better rule lacks support from those data.
When people plateau, their improvements may become increasingly specialized. More practice might refine shortcuts that are effective within the app rather than recruit new capacities useful in unrelated settings. A good training plan therefore defines the desired transferable behavior at the start and checks it with an outcome that does not merely repeat the exercise.
What consumers should ask before paying for a brain-training app
A program can still be worth using as entertainment, a structured hobby, or a way to practice a particular skill. The purchase decision changes when its marketing promises higher IQ, general cognitive superiority, dementia prevention, or improved performance at school and work.
Ask what exact result the program promises and whether it was tested against an appropriate active comparison group. Look for published randomized studies conducted with the actual product or intervention, in a population similar to yours, using valid tests outside the app. Check whether the effect was still present after training ended, how large it was, whether the researchers preregistered their primary outcomes, and whether independent investigators reproduced the finding.
Promotional evidence deserves scrutiny when it relies on testimonials, improvement in proprietary game scores, uncontrolled before-and-after comparisons, isolated positive subtests, brain images without functional outcomes, or vague claims about neuroplasticity.
Be alert to the difference between a relative improvement and an absolute change. A large percentage increase on a low starting game score might be impressive within the game and still say nothing about standardized IQ points or everyday functioning. Claims about preventing dementia are health claims and demand particularly strong longitudinal evidence.
Better ways to pursue a specific cognitive goal
The most practical approach begins by defining the outcome. If the goal is to remember appointments, external reminders, routines, and deliberate organizational strategies may be more useful than accumulating points in a memory game. If the goal is to become more proficient in a language, mathematics, programming, or another field, sustained domain-specific learning with useful feedback provides direct opportunities to improve that skill.
If the goal is to learn material efficiently, active recall, appropriate spacing, checking understanding, and repeated application to varied examples offer a better-matched route than generic games. If the goal is to support cognitive functioning over the lifespan, attending to sleep, physical activity, meaningful engagement, medical care when indicated, and sustainable routines addresses a broader set of determinants.
Sleep is especially relevant to the conditions under which people learn, retrieve information, and perform attention-demanding tasks. For a more focused discussion, see Sleep and Memory: Consolidation, Learning, and Sleep Loss.
These activities should not be marketed as guaranteed IQ-raising treatments. They are better understood as ways to pursue demonstrable knowledge, functioning, health, and personally valued abilities.
What would count as convincing evidence that a program raises intelligence?
A persuasive demonstration would use random assignment; a credible, expectation-matched active control; enough participants to detect realistic effects; preregistered hypotheses; and outcome assessors who do not know who received which intervention. It would measure a well-defined cognitive construct using multiple reliable tests with minimal overlap with the training tasks.
The strongest analysis would distinguish change in observed individual test scores from change in a latent general ability, report uncertainty and effect sizes, account for multiple comparisons, and demonstrate a benefit that persists beyond the immediate posttest. Independent replication and evidence for an everyday consequence would strengthen a claim that the effect matters to people's lives.
Without these safeguards, a significant gain can mean that participants learned a task, became comfortable with the test, or benefited from aspects of participation that have little to do with the proposed training mechanism. A credible claim about intelligence must survive those alternative explanations.
Common myths about brain training
Myth: A higher game score means a higher IQ
Game scores generally reflect performance within that game and its scoring system. They are not interchangeable with norm-referenced intelligence assessments or broad latent cognitive abilities.
Myth: Any activity that exercises the brain strengthens all thinking abilities
Learning and adaptation are often specific. Transfer depends on shared skills and procedures, the teaching context, and how the new task is measured.
Myth: Brain plasticity proves that cognitive games must increase intelligence
Neuroplasticity makes learning possible. A general intelligence benefit still requires direct behavioral and psychometric evidence.
Myth: The lack of broad IQ gains means practice has no value
Training may be enjoyable, motivating, socially engaging, and effective for skills that are directly practiced. These are meaningful but distinct benefits.
Myth: A single favorable study is enough to establish a brain-training product
Clinical and scientific conclusions depend on the full body of evidence, replication, bias assessment, and results from credible comparisons. A small favorable pilot study is a beginning, not a settled conclusion.
Frequently asked questions
Does Lumosity or another brain-training app actually increase IQ?
Independent evidence has not established that routine use of commercial brain-training apps reliably raises general intelligence or standardized IQ in healthy users. Improved scores on proprietary tasks show learning within those tasks. A product-specific far-transfer claim requires independent, adequately controlled studies.
Can dual n-back training improve fluid intelligence?
Some early studies and an early meta-analysis reported small positive effects on fluid-reasoning tests. Larger and more methodologically stringent reviews have not established a consistent, meaningful improvement in fluid intelligence. Dual n-back practice is much more clearly associated with improvement on n-back itself.
Is working memory the same as intelligence?
Working memory and intelligence are related constructs. Working memory concerns maintaining and manipulating information; intelligence encompasses broader patterns of reasoning and cognitive performance. Their correlation does not mean that improving one practiced working-memory task must raise overall intelligence.
Does brain training improve memory in everyday life?
Some forms of practice improve specific memory tasks, and particular strategies can help with specific daily activities. General-purpose game training has not consistently demonstrated durable improvements in everyday memory. The outcome should be measured in the setting where the benefit is claimed.
Can brain games prevent dementia?
Evidence does not justify treating ordinary commercial brain games as a proven dementia-prevention intervention. The effectiveness of particular approaches, for specific populations and outcomes, remains a separate research question. Concerns about cognitive decline warrant appropriate health guidance and evaluation.
Are crosswords and chess useful for the brain?
They provide intellectually engaging practice and can improve skills connected to the activity. Their recreational and educational value is real. A reliable increase in broad IQ or protection against dementia cannot be inferred merely from doing them.
How much brain training is needed to improve IQ?
No scientifically established dosage of ordinary commercial cognitive training guarantees an IQ increase. More sessions can improve familiarity and skill on trained exercises without producing far transfer.
Can children increase IQ by playing memory games?
Programs should be evaluated using age-appropriate controlled research and independent developmental or educational outcomes. Generalized IQ or academic gains cannot be presumed from better performance on the games.
Does a higher score after training prove that intelligence has changed?
A score can rise because of learning, familiarity, fluctuating performance, or measurement effects. Establishing a change in intelligence requires reliable, independent assessment and appropriate interpretation of the construct measured.
What is the difference between near transfer and far transfer?
Near transfer is improvement on another task substantially similar to training. Far transfer involves improvement on more distant abilities or real-world outcomes, such as independent reasoning or academic performance. The farther the claimed benefit extends beyond the trained activity, the more demanding the evidence needs to be.
Conclusion
Brain training teaches the brain to perform the activities it practices, sometimes with benefits on closely related tasks. The best contemporary syntheses do not support broad promises that commercial cognitive games reliably increase general intelligence or IQ. The decisive scientific issue is transfer: whether improvement reaches an independently measured ability, persists, and matters beyond the exercise.
Consumers can judge programs more accurately by asking for the precise claimed outcome, appropriate active-control evidence, meaningful effect sizes, and long-term results. Practicing worthwhile skills, learning directly in a valued domain, and using targeted strategies remain sensible pursuits even when a generalized IQ increase has not been demonstrated.
Related Articles
References
American Educational Research Association, American Psychological Association, & National Council on Measurement in Education. (2014). Standards for educational and psychological testing. Official standards information.
Au, J., Sheehan, E., Tsai, N., Duncan, G. J., Buschkuehl, M., & Jaeggi, S. M. (2015). Improving fluid intelligence with training on working memory: A meta-analysis. Psychonomic Bulletin & Review, 22, 366–377. https://doi.org/10.3758/s13423-014-0699-x.
Gates, N. J., Rutjes, A. W. S., Di Nisio, M., Karim, S., Chong, L.-Y., March, E., Martínez, G., & Vernooij, R. W. M. (2020). Computerised cognitive training for 12 or more weeks for maintaining cognitive function in cognitively healthy people in late life. Cochrane Database of Systematic Reviews, CD012277. https://doi.org/10.1002/14651858.CD012277.pub3.
Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perrig, W. J. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences, 105, 6829–6833. https://doi.org/10.1073/pnas.0801268105.
Melby-Lervåg, M., Redick, T. S., & Hulme, C. (2016). Working memory training does not improve performance on measures of intelligence or other measures of far transfer: Evidence from a meta-analytic review. Perspectives on Psychological Science, 11, 512–534. https://doi.org/10.1177/1745691616635612.
National Academies of Sciences, Engineering, and Medicine. (2017). Preventing cognitive decline and dementia: A way forward. National Academies Press. https://doi.org/10.17226/24782.
Owen, A. M., Hampshire, A., Grahn, J. A., Stenton, R., Dajani, S., Burns, A. S., Howard, R. J., & Ballard, C. G. (2010). Putting brain training to the test. Nature, 465, 775–778. https://doi.org/10.1038/nature09042.
Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E., Hambrick, D. Z., Kane, M. J., & Engle, R. W. (2013). No evidence of intelligence improvement after working memory training: A randomized, placebo-controlled study. Journal of Experimental Psychology: General, 142, 359–379. https://doi.org/10.1037/a0029082.
Rodas, J. A., Asimakopoulou, A. A., & Greene, C. M. (2024). Can we enhance working memory? Bias and effectiveness in cognitive training studies. Psychonomic Bulletin & Review, 31, 1891–1914. https://doi.org/10.3758/s13423-024-02466-8.
Sala, G., Aksayli, N. D., Tatlidil, K. S., Tatsumi, T., Gondo, Y., & Gobet, F. (2019). Near and far transfer in cognitive training: A second-order meta-analysis. Collabra: Psychology, 5(1), 18. https://doi.org/10.1525/collabra.203.
Stanford Center on Longevity. (2014). A consensus on the brain training industry from the scientific community. Official consensus statement.
Sutton, E., Catling, J., Veldhuijzen van Zanten, J. J. C. S., & Segaert, K. (2025). Practice makes perfect, but to what end? Computerised brain training has limited cognitive benefits in healthy ageing. Psychological Research, 89, Article 75. https://doi.org/10.1007/s00426-025-02110-7.