What practice can and cannot do for your thinking

Transfer depends on what another task requires

Learning to estimate the cost of a shopping basket might help you estimate another basket containing unfamiliar items. Both attempts involve quantities, prices and a way of checking whether the total is plausible. This is a candidate for near transfer, where practice carries into a task with similar demands. Far transfer means carrying improvement into something substantially different, such as resolving an argument at work after practising number puzzles. Where near transfer ends and far transfer begins depends on how closely the demands of the two tasks match, so the labels describe a spectrum more than two separate categories.

A familiar surface can conceal a different problem, because two puzzles that both contain numbers might call for unrelated procedures, while planning a meal and planning a meeting might both depend on tracking which step must come first. To judge whether transfer is plausible, describe the operation you learned and the operation the new task needs. Shared features give you a reason to investigate, although they cannot show that an improvement occurred, and only a separate test on the new task can do that.

Schematic, not data: the round distances illustrate increasingly different demands, without estimating the size or likelihood of transfer.

Illustrative positions used in the transfer schematic
Task	Distance from practice	Demand
Practised task	0 illustrative distance units	Same exercise
Near transfer	50 illustrative distance units	Similar demands
Far transfer	100 illustrative distance units	Different demands

Suppose your outside goal is judging whether a household bill is plausible. Record your estimate before opening several bills, then compare it with the amount due and note which information you missed. Practice on fresh bills gives you a direct measure of that judgement. A run of abstract arithmetic puzzles provides a less direct test, even when your puzzle score rises steadily. You might learn a useful checking procedure from the puzzles, although its value for bills remains a hypothesis until you apply it there. Keep the outside task stable enough to compare attempts, while changing the examples so you cannot answer from memory. This personal record can guide your next choice without establishing a general training effect.

Brain training improves its exercises more reliably than other skills

In Owen and colleagues' 2010 study, Putting brain training to the test, published in Nature, 11,430 participants trained several times each week for six weeks in an online study of tasks designed to improve reasoning, memory, planning, visuospatial skills and attention. Participants improved on every task they trained, yet the study found no evidence of transfer to untrained tasks, even when those tasks were cognitively closely related. The 11,430 were people who stayed with the study and completed its assessments, so they are a self-selected subset of the volunteers who signed up and may differ from them.

One trial cannot settle every training claim. Simons and colleagues' 2016 review, Do “Brain-Training” Programs Work?, in Psychological Science in the Public Interest, found extensive evidence of improvement on trained tasks, less evidence for closely related tasks and little evidence for distant tasks or everyday cognitive performance. Improvements must also survive comparison with a suitable control group, because expectations and repeated testing can influence results on their own.

Working-memory training has received particular attention because holding information in mind contributes to many activities. The 2016 meta-analysis by Melby-Lervåg, Redick and Hulme found reliable short-term gains on closely related memory measures, without convincing evidence of far transfer to intelligence or academic skills when trained groups were compared with control groups. Sala and Gobet's 2019 review, Cognitive Training Does Not Enhance General Cognition, in Trends in Cognitive Sciences, likewise concluded that training produces little transfer beyond closely related tasks. Small positive estimates deserve caution when they shrink under stronger controls, and mixed findings cannot support a promise of general cognitive improvement.

Deliberate practice matters without explaining every difference

Ericsson, Krampe and Tesch-Römer's 1993 paper, The Role of Deliberate Practice in the Acquisition of Expert Performance, in Psychological Review, described practice designed to improve specific weaknesses, with feedback and opportunities for correction. Repeating a comfortable exercise can leave the weakness untouched. A pianist who isolates a difficult transition has a clearer target than someone who repeatedly plays an entire familiar piece.

Macnamara, Hambrick and Oswald's 2014 meta-analysis, read with its 2018 corrigendum, gives a more bounded picture. Across 88 studies involving 11,135 participants, the corrected pooled estimate associated accumulated deliberate practice with 14% of the variance in performance between people. That percentage describes a statistical association, rather than a percentage improvement caused by training. Its contribution varied considerably between domains, and retrospective reports of practice cannot establish the causes of every difference. Definitions of deliberate practice also remain disputed.

Spacing, retrieval and feedback give practice a clearer target

The evidence for retaining particular material is firmer than the evidence for raising general intelligence. Cepeda and colleagues' 2006 review of distributed practice supports spreading learning across occasions, while Roediger and Karpicke's 2006 study of test-enhanced learning shows how recalling material can improve later retention. The suitable interval depends on when you need to remember it. These findings concern learning and remembering material under particular conditions.

Try reading a short explanation of tides, closing it and writing the causal chain from memory. Checking your account reveals whether you remembered the relationship or merely recognised the words. Feedback then supplies something specific to repair: perhaps you omitted the Moon's gravitational influence or confused tides with wind-driven waves. Hattie and Timperley's 2007 review, The Power of Feedback, in Review of Educational Research, explains why feedback varies in usefulness. Information about the task and your next correction gives practice a direction that praise alone cannot supply.

Sleep and exercise belong in the account

Learning continues after the exercise has ended. Rasch and Born's 2013 review, About Sleep's Role in Memory, in Physiological Reviews, describes evidence that sleep supports memory consolidation. Sacrificing sleep to accumulate more repetitions can therefore work against the material you hope to retain.

Exercise also belongs in the account, with care about which outcomes were measured. In Erickson and colleagues' 2011 randomised trial, 120 older adults were assigned either to aerobic training or to a stretching control, and the aerobic group showed an increase in the size of the anterior hippocampus. That region is involved in memory, although its size cannot establish a gain in every thinking skill. Evidence for broad cognitive gains from exercise remains mixed: Ciria and colleagues' 2023 umbrella review questioned the strength of the causal evidence in healthy populations. A 2024 response by Dupuy and colleagues argued that the review underestimated the benefits, illustrating the disagreement about interpreting these effects. Check the particular outcome before treating a result as support for a wider benefit.

The app makes practice visible without promising intelligence gains

Funga Wega provides a session of about eight minutes mixing pattern puzzles, a range estimate, comprehension and reasoning questions, a working-memory task, a field note and review of earlier misses. The memory task asks you to read a start state and some steps, hide them and answer. Field notes pair a short thinking move with a question requiring it, and each note labels its evidence as research-backed, mixed evidence or practitioner wisdom. Hints, when you ask for them, move from where to look, through an approach, to a worked first step.

Missed items return after intervals of 1, 3, 7, 14 and 30 days. Facts return as the same question, while generated puzzles return as fresh examples of the same kind. Short factual lessons in the reading room feed facts into that review. Before answering, you record your confidence; Insights compares it with your results and checks whether your 90% ranges contain the truth about nine times in ten. Progress stays on your device by default, and nothing is generated or graded by AI. These activities offer practice and a record of performance, and they make no claim to increase intelligence. If you hope the practice carries over to something outside the app, test that hope against the outside task itself.

Questions about applying the evidence to practice

Does improvement on a puzzle establish wider improvement?

A higher puzzle score establishes improvement on that measure under the conditions you tested. To assess a wider benefit, choose a separate task that represents the skill you care about and track that performance too.

Can working-memory exercises still have a useful purpose?

They can provide practice in holding and updating information within the exercise. A useful personal aim might be learning a checking procedure, while broader improvements in intelligence remain unsupported by the reviews discussed here.

How can I make practice relevant to daily life?

Choose a recurring task, such as estimating costs before shopping, and practise with varied examples followed by a check. Record both the estimate and the outcome so you can identify the particular error your next attempt should address.