Back to the on-screen lesson ·
What span and n-back practice can and cannot do, and how to read a claim about it: near or far, and compared with what.
Paper packet. Every task here also exists on screen, where it is checked automatically; answers written on paper are not assessed by Nydus. When you are back at a device, enter your answers there.
You will tell a near-transfer claim from a far-transfer one, read a trial's gain against its control group, and try one optional letter-span trial as practice.
Every lesson in this course has changed the task rather than the person: bigger chunks, working lines, markers, one next step, instructions on paper. None of them promised a bigger desk.
You will meet that promise elsewhere, in apps and adverts for memory games. This lesson looks at what those games are, what the evidence says they do, and how to read a claim about them.
| Term | What it means |
|---|---|
| Span task | A task that shows a string of items and asks for it back in order. |
| n-back task | A stream of items where you say whether each one matches the one n places before it. |
| Near transfer | Getting better at the practiced task, or at one very like it. |
| Far transfer | Getting better at something distant from the practice, such as reading or reasoning. |
| Active control group | A comparison group that does a different activity for the same time, so effort and expectation are matched. |
| Meta-analysis | A study that pools the results of many studies to see the overall pattern. |
Practice a span task or an n-back task for a few weeks, and you will very likely get better at it. That is not in dispute. It is near transfer: the gain lands on the task you practiced, and a little on tasks that are almost the same.
The bigger promise is that the gain spreads: to reading, arithmetic, reasoning, grades, work, or remembering appointments. That is far transfer, and it is where the evidence runs out.
So this course offers span trials as optional practice of one task. You may enjoy it. It is not a treatment, it is not a measure of you, and finishing it is never described as a better memory.
The useful skill is reading claims. Ask two questions of any claim: where does the promised gain land? (near or far), and compared with what? (a group that did nothing, or a group that did something else for the same time).
Another way: diagram
Draw a dot for the practiced task, with rings around it. The first ring holds near-identical tasks, and gains are found there. The outer rings hold reading, arithmetic and everyday remembering, and pooled studies found no convincing gains out there.
Another way: story
Marcus plays an n-back game every evening for a month. His level climbs from 2-back to 4-back. The app says his memory has improved 80 percent. What has improved is his n-back level. Whether his reading, work or memory for errands changed is a separate question the app's number cannot answer.
Monica Melby-Lervåg and Charles Hulme (2013) pooled 23 studies of working-memory training. People got better at memory tasks like the ones they practiced, but those gains were short-lived, and there was no convincing evidence that the training carried over to distant abilities such as verbal ability, nonverbal reasoning, word reading or arithmetic.
A larger review by Melby-Lervåg, Thomas Redick and Hulme (2016) pooled 87 publications and 145 comparisons. Again: reliable gains, just after the practice period, on memory tasks like the practiced ones, and no convincing far transfer — to nonverbal ability, verbal ability, word decoding, reading comprehension or arithmetic — when trained groups were compared with groups that did another activity.
Daniel Simons and colleagues (2016) reviewed brain-training programs more widely and reached the same view: improvement on the trained tasks, little evidence of improvement on distant tasks or everyday life.
Anyone who takes a test twice tends to do a little better the second time. People who expect to improve try harder. So a gain from before to after is not, on its own, evidence that the training did anything.
A good trial compares the trained group with an active control group that spent the same time on something else, such as trivia quizzes. The honest number is the trained group's gain minus the control group's gain. Some early studies that looked promising, such as a 2008 n-back study that reported gains in reasoning, had a control group that did nothing. When Redick and colleagues (2013) repeated the idea with an active control group, the reasoning gains did not appear.
In a famous study, K. Anders Ericsson, William Chase and Steve Faloon (1980) trained a college student, SF, on digit span for more than 230 hours. His span rose from about 7 digits to 79. He had not grown a bigger desk. He was a keen runner, and he learned to chunk digits as running times — 3492 became 3 minutes 49.2 seconds, near a world-record mile. The skill was a chunking skill built for digits, and it did not carry over to letters.
That is near transfer at its most striking, and it shows why the claim better at the game never becomes better at everything without separate evidence.
In January 2016 the US Federal Trade Commission announced that Lumos Labs, the maker of the Lumosity brain-training games, would pay 2 million dollars to settle charges of deceptive advertising. The FTC said the company had claimed, without adequate scientific support, that its games would help users do better at work and in school, and would delay age-related decline and reduce cognitive impairment from conditions such as stroke and brain injury.
Read with this lesson's two questions, every one of those claims is far transfer. The order required the company to have competent and reliable scientific evidence before making such claims again. Getting better at the games themselves was never the issue.
Many apps show a line going up: your memory score has risen 35 percent. Ask where the gain lands. If the score is the app's own games, it is near transfer, and it is expected. Ask what it is compared with. If there is no comparison group, part of the rise is simply getting used to the games. A fair way to describe it is: I have got better at these games. That is a true sentence, and it is enough if you enjoy them.
My score went up, so my memory improved. Your score on that task went up. That is real, and it is near transfer. It says nothing on its own about reading, work or daily remembering.
A study found a gain, so it works. Compared with what? A gain over a group that did nothing may be practice at taking tests, or effort from expecting to improve.
If practice cannot enlarge the desk, nothing helps. The earlier lessons are the answer: changing the task — chunks, paper, markers, one next step — helps every day, with no promise needed.
List each claim in the advert.
A: reach level 10 in our span game. B: remember longer number strings in our game. C: raise your grades. D: never forget a name.
Each claim is judged separately.
Ask where each promised gain lands.
A, B → the practiced game. C → school work. D → everyday memory.
Distance from the practiced task decides near or far.
Label near and far.
A near · B near · C far · D far
Near is the task or one very like it; far is anything distant.
Match each label to the evidence.
Near: gains found. Far: no convincing gains in pooled studies.
Melby-Lervåg, Redick and Hulme (2016).
Write an honest summary.
You will likely get better at this game. Nothing shows it will raise grades or fix names.
The summary keeps only what the evidence supports.
Write each group's scores before and after.
Trained span: 5.0 → 6.2. Active control span: 5.0 → 5.3.
Both groups took the same tests twice.
Work out each group's gain.
Trained: 6.2 − 5.0 = 1.2. Control: 5.3 − 5.0 = 0.3.
The control's gain is what retesting alone gives.
Subtract the control's gain.
1.2 − 0.3 = 0.9 beyond the control
This is the gain the training can claim on span.
Do the same for a far measure.
Reading: trained +2, control +2 → 2 − 2 = 0
No difference beyond the control on reading.
State the finding honestly.
Near gain of 0.9 on span; no far gain on reading
The same pattern as the pooled studies.
Work out each group's reasoning gain.
Trained +3 · did-nothing group +1 · active control +3
Each gain is after minus before.
Compare with the group that did nothing.
3 − 1 = 2: looks like a gain
Retesting and expectation are not matched.
Compare with the active control.
An app for a letter-span game makes four claims. Which one is a near-transfer claim that the evidence supports?
A trial of an n-back game reports reasoning scores. The reading of the trial is half done. Fill in the later rows.
| Reading | |
|---|---|
| Trained group | 20 → 24 |
| Trained gain | 4 |
| Control group | 20 → 24 |
| Control gain | |
| Beyond the control | |
| Finding |
Match each term to what it means.
| Getting better at the practiced task or one very like it | Getting better at a distant ability, like reading | People who do a different activity for the same time | A group whose memory grew from practice | |
|---|---|---|---|---|
| Near transfer | ||||
| Far transfer | ||||
| Active control group |
On a span test, a trained group's average rose by 6 items and an active control group's rose by 2. Fill in each gain and the gain beyond the control.
Trained gain: a. Control gain: b. Gain beyond the control: c.
Optional letter-span practice. Six letters appear for ten seconds; write them back in order. This is practice of one task. It is not a test of your memory, and doing it does not improve memory in general. You can skip it.
Ask someone to read you the list, then cover it and write what you remember:
In 2016 the FTC settled with the maker of Lumosity over claims about its games. Label each claim near or far from the practiced games.
| Claim | Near or far? |
|---|---|
| Users score higher on the games | |
| Users perform better at work | |
| Users perform better in school | |
| Play delays age-related mental decline |
Lesson test: one question per skill, one attempt each, no hints. Your answers are checked when you submit.
A span-training program reports three results against an active control group. Label each near or far, and write the gain beyond the control.
| Measure (trained gain, control gain) | Near or far? | Gain beyond control |
|---|---|---|
| Span test (3, 1) | ||
| Reading test (2, 2) | ||
| Arithmetic test (1, 1) |
Find one claim about a memory app or game. Ask where the gain lands and what it was compared with, and say the honest version aloud. That closes the course: the moves that help every day change the task, not the desk. A trial in this course is practice. What is kept is what you wrote on this device today; it is not a score, it is not a test of your memory, and it is never compared with anyone. If you notice trouble with memory or thinking that is new or changing, tell your own doctor or clinician; this course is general practice, not an assessment or treatment. Stopping takes nothing more than closing the page: your place is already saved, and the lesson opens where you left it.
13. Your turn: the same trial, two comparison groups, step 3
3 − 3 = 0: no gain beyond the control
Matched effort and expectation remove the apparent gain.