Working Memory Training: Does It Actually Transfer?

Written and researched by the Studio âge web research team / Last reviewed: / Editorial policy

Working memory training research and transfer evidence

The core question

Working memory training is built on an intuitive idea: if working memory helps us reason, read, calculate, and stay focused, then training working memory might improve those broader abilities. The scientific record is more complicated. Based on the evidence reviewed here through 2024, working memory training can improve the trained task and often improves closely related tasks, but strong transfer to intelligence or real-world skills is not well established.

This distinction matters for users of an n-back game. If you practice n-back, you should expect to get better at n-back. You may also improve at similar working-memory updating tasks. But it would be an overstatement to claim that any working memory game reliably raises general intelligence, grades, job performance, or everyday productivity. The main debate is not whether people learn from practice. They do. The debate is whether that learning generalizes far beyond the practiced task.

Near transfer vs. far transfer

Near transfer means that training improves performance on tasks that resemble the training task. For example, practicing a visual n-back game might improve performance on another n-back version, a similar updating task, or a task that also requires active maintenance and rapid comparison of recent information. Near transfer is plausible because the trained and tested tasks share processes.

Far transfer means improvement on outcomes that are more distant from the training task. Examples include fluid intelligence, reading comprehension, arithmetic, school achievement, work performance, or general cognitive ability. Far transfer is much harder to prove. It requires ruling out test familiarity, placebo effects, motivation differences, weak control groups, and publication bias. That is why researchers often put more weight on randomized studies with active control conditions than on studies using no-contact controls.

Studies that show positive transfer

The best-known positive study is Jaeggi, Buschkuehl, Jonides, and Perrig (2008), published in Proceedings of the National Academy of Sciences. Jaeggi et al. (2008) reported that adaptive dual n-back training transferred to measures of fluid intelligence. The study became influential because participants trained on a working-memory task but were tested on reasoning tasks, suggesting possible far transfer.

A later synthesis by Au, Sheehan, Tsai, Duncan, Buschkuehl, and Jaeggi (2015) examined n-back training and fluid intelligence in a meta-analysis. Au et al. (2015) included 20 studies and found a small but statistically significant positive effect of n-back training on fluid intelligence. This result supports a cautious pro-transfer position: there may be measurable gains on laboratory reasoning tests after n-back training, but the average effect is modest.

More recent second-order evidence also suggests that working memory itself can improve with training. Syed, Lum, Byrne, and Skvarc (2024) reviewed meta-analyses of working-memory interventions in healthy adults and concluded that working memory performance can improve, but the effect size is small and real-life cognitive utility may be limited. This fits the balanced view: training effects exist, but they should not be oversold.

Studies that challenge broad transfer

The strongest skeptical evidence comes from reviews focused on far transfer. Shipstead, Redick, and Engle (2012), in Psychological Bulletin, asked whether working memory training is effective and argued that many optimistic claims rested on weak methodology, poor controls, and ambiguous transfer measures. Their review helped shift the field toward stricter standards for active control groups and broader outcome batteries.

Melby-Lervåg, Redick, and Hulme (2016) published a major meta-analytic review in Perspectives on Psychological Science. They analyzed 87 publications with 145 experimental comparisons. Their conclusion was blunt: working memory training produced short-term improvements on similar working-memory outcomes, but there was no convincing evidence of reliable far transfer to nonverbal ability, verbal ability, word decoding, reading comprehension, or arithmetic when compared with treated control conditions.

Another important n-back-specific synthesis is Soveri, Antfolk, Karlsson, Salo, and Laine (2017). They conducted a multi-level meta-analysis of n-back training studies and found medium transfer to untrained n-back tasks, but very small effects for other working-memory tasks, fluid intelligence, and cognitive control. Their conclusion was that much of the transfer after n-back training is task-specific.

Why n-back specifically is studied

The n-back task is popular in research because it is simple to describe, easy to standardize, and adjustable in difficulty. In a 2-back task, the participant responds when the current stimulus matches the one two trials earlier. Researchers can vary the stimulus type, the timing, the value of n, and whether the task is single or dual modality.

N-back is also widely used in neuroimaging. Owen, McMillan, Laird, and Bullmore (2005) reviewed functional neuroimaging studies of n-back tasks and showed that the paradigm reliably engages frontoparietal brain regions associated with working memory and cognitive control. This does not prove that n-back training improves intelligence, but it explains why n-back is a useful research tool: it reliably taxes the systems researchers want to study.

For an n-back game, this means the task has a real scientific background. It is not random clicking. It is a simplified browser version of a family of tasks used in cognitive psychology and neuroscience. But research legitimacy is not the same as proof of broad benefits. The task can be scientifically meaningful even if its far-transfer effects are limited.

What this means for everyday users

If you use working memory training, set expectations carefully. The most likely improvement is better performance on the training task itself. That can still be useful if your goal is to practice focused attention, track progress, and challenge your working memory under controlled conditions. It is also reasonable to use n-back as a short mental exercise, much like a demanding puzzle.

Do not treat working memory training as a replacement for sleep, exercise, study strategies, reading practice, math practice, or targeted skill learning. If your goal is better reading, practice reading. If your goal is arithmetic, practice arithmetic. If your goal is general cognitive challenge, n-back can be part of the mix, but it should not be sold as a guaranteed intelligence intervention.

A practical routine is simple: start with 1-back or 2-back, train for a few focused minutes, track accuracy, and increase difficulty only when performance is stable. If you enjoy the challenge, continue. If you find it frustrating, reduce the difficulty or stop. The honest value of n-back is structured cognitive practice, not a promise of dramatic far transfer.

Try the tool

You can try the browser-based n-back task for free on this site. Use it as a focused working-memory exercise and watch how accuracy changes as the value of n increases.

Try the free N-Back tool

References