Learning Science · Working memory and task demands
Task improvement and limits of broad training claims
Separate gains on a practised exercise from claims that training improves a broad mental ability.
Use the examples to connect this lesson with the earlier topics. No specialist equipment is needed.
Back to the chapterAfter reading, you can
- Distinguish near and far transfer
- Evaluate a broad training claim
A better score on the game
A learner repeatedly practises remembering coloured positions and improves. This is evidence of better performance on that game under those conditions. It may reflect knowledge of its patterns, a new strategy or more efficient execution. It does not alone show better school learning, general intelligence or everyday memory.
Transfer means an effect of prior learning on another task. Near transfer concerns closely related tasks; far transfer concerns more different activities. These are relative descriptions, so name the actual tasks rather than relying on the label alone.
Ask for the bridge
A broad claim needs outcomes beyond the trained exercise, relevant comparison groups and checks of persistence. An active comparison helps distinguish the specific training from time, engagement, expectation or repeated testing. A later unfamiliar outcome is more informative about generalisation than a second score from the same game.
Research reviews of working-memory training have found task-specific gains without convincing evidence for many broad benefits under stronger comparison conditions. This is a reason to require relevant evidence, not a proof that no training could ever transfer to any useful outcome.
| Claim | Relevant observation | What remains unresolved |
|---|---|---|
| Better at this game | Improved trained-game performance | Untrained tasks and durability |
| Better at a similar untrained task | A specified untrained outcome with a relevant comparison | More distant everyday capabilities |
| Easier work in daily life | Direct work-relevant outcomes and appropriate later checks | Other populations, tasks and settings not studied |
Choose practice for a real goal
If the target is reading a difficult subject, practise relevant reading, explanation and use with appropriate support. A cognitive game may be enjoyable, but enjoyment and a game score are separate from that target. Ask what benefit is promised and whether the evidence directly measures it.
Expectations can influence a training comparison. Participants who believe an activity will help may work differently on the outcome test. A relevant comparison and clear procedures help address that possibility. Also inspect whether the outcome was chosen before results were known and whether the promised benefit matches what was actually measured.
Worked example
An advertisement’s missing outcome
An app displays rising scores on its coloured-position game and advertises easier everyday work. What evidence is missing?
- Name what was actually measured: repeated performance on the trained game. Improvement could involve game-specific strategies or familiarity with its rules.
- Name the advertised outcome separately: an untrained work-like task that represents the claimed benefit. Ask whether a relevant active-comparison group spent similar time in another engaging activity, and how allocation and expectations were handled.
- Inspect whether the untrained outcome improves and whether any difference persists on a later check. A graph of game scores alone cannot answer those questions; equally, failure to demonstrate broad transfer does not erase useful game-specific gains.
The app graph supports improvement on the practised game under its recorded conditions. A wider work benefit needs evidence on that wider outcome with a comparison that addresses plausible alternatives.
Try it yourself
Think through each question before opening the answer.
Why is a no-training comparison sometimes insufficient?
Show the answer and reasoning
Differences could reflect attention, expectation or engagement rather than the specific exercise. A relevant active comparison can address some alternatives.
Compared with doing nothing, an engaging activity may change expectations, effort or familiarity with testing. An active comparison can address some alternatives, although its design still needs inspection.
Does weak evidence for far transfer make task-specific improvement worthless?
Show the answer and reasoning
No. Specific improvement can matter when the trained task is useful. The conclusion should match the measured outcome.
The value of a specific gain depends on the goal. If the task itself matters, becoming better at it is useful; the error is promising benefits that were not measured.
Training improves digit-span scores but an untrained reading task shows no clear improvement against an active comparison. Which conclusion fits?
Show the answer and reasoning
The evidence supports the measured task gains; it does not establish a reading benefit under that comparison.
Digit-span and reading outcomes are different capabilities. “No clear reading benefit here” is narrower than “no training could ever help reading”.
What to carry forward
- Training gains and broad benefits are different claims.
- Name the untrained outcome.
- Practise the capability that serves the goal.
References
Sources supporting the explanations and learning checks in this lesson.