How to read a study on this page
Eight things without which a result says nothing.
- Method — What exactly is compared: the same test before the classes and after them, in the same procedure and at a similar time of day.
- Sample — How many students took part and how they were selected. A result on a handful of people is a story, not a study: below ten people we do not report group results.
- Age — The age range of the participants. A result for nine-year-olds does not carry over automatically to fifteen-year-olds.
- Duration — How many weeks passed between measurements and how many classes took place in that time.
- Measurement tool — Which task we measured with, named, so that what it tests can be checked.
- Results — The numbers before and after, with the unit and the number of people in each row.
- Limitations — What the study does not show. Most often: no comparison group, the effect of repeating the test, and the fact that participants enrolled themselves.
- Source — Where the data come from and when we collected them.
Studies we go through
Scientific papers on the techniques we teach — each broken down into the same eight parts as our own studies, with figures only from the publication itself. None of them studied Space Memory; what we take from each into our classes is written separately at the end.
Six weeks of the memory palace in adults with no prior training
Dresler M., Shirer W. R., Konrad B. N., Müller N. C. J., Wagner I. C., Fernández G., Czisch M., Greicius M. D., 2017
After a short training in the method of loci, does an ordinary person remember like a memory championship competitor?
| Method | Participants were assigned pseudo-randomly to three groups: method of loci training, working memory training (n-back), or no training. The same test before and after six weeks; separately, 23 athletes from the top fifty of the memory sports ranking were compared with matched controls. |
|---|---|
| Sample | 51 · adult men with no experience of mnemonic techniques; about 17 people per group |
| Age | adults, 24 years on average (standard deviation 3 years) |
| Duration | 6 weeks: 40 sessions of 30 minutes a day, at home, on an online platform; retest after 4 months |
| Measurement tool | free recall of a list of 72 nouns after 20 minutes and after 24 hours |
Results
- The group training the method of loci improved significantly more than both control groups — both after 20 minutes and after 24 hours (η² above 0.4 in every comparison).
- The advantage held in the test 4 months after the training ended (η² = 0.39).
- 16 of the 17 trainees reached the level of 40 words at a time in the training programme.
- For comparison: memory athletes recalled on average 70.8 of 72 words, matched untrained adults — 39.9.
Limitations
- The participants were adult men, not school-age children.
- The groups were small — about 17 people each.
- What was measured was lists of unrelated words, not school material or grades.
- Training lasted 30 minutes a day for 40 days; shorter training was not studied.
What we take from this into our classes
This is where we take the assumption that the memory palace is a skill that can be taught in weeks, and that is why practice between classes is short and daily. How much of this carries over to a child and to school material, this study does not say; we check that with our own measurement.
Source: Neuron 93(5), 2017 · open
A test instead of rereading: what is left after a week
Roediger H. L., Karpicke J. D., 2006
Does quizzing on material teach, or only check?
| Method | Two experiments. Students read texts and then either reread them or recalled them from memory in writing without prompts — the same number of times. Final test after 5 minutes, after 2 days or after a week. |
|---|---|
| Sample | two experiments with students (numbers given in the full text, not in the abstract) · students at the authors' university, random assignment to conditions |
| Age | students, adults |
| Duration | from 5 minutes to 1 week between studying and the final test |
| Measurement tool | free written recall of the text; additionally, a rating of one's own confidence in remembering |
Results
- After 5 minutes, those who reread did better.
- After 2 days and after a week, those who recalled from memory did clearly better.
- Rereading raised students' confidence that they had remembered — despite a worse result later.
Limitations
- The participants were students, not children.
- Short texts in laboratory conditions, not a textbook or a school test.
- Recall was without feedback — in class it usually is different.
What we take from this into our classes
That is why homework with us is quizzing, not copying out, and why we tell parents that "read it three times and knows it" often turns out to be an illusion a few days later.
Source: Psychological Science 17(3), 2006 · open
Recall from memory versus concept maps — on science texts
Karpicke J. D., Blunt J. R., 2011
Does "deeper" work with a text (a concept map) give more than simple recall from memory?
| Method | Students studied textbook-like texts. Some drew a concept map with the text in front of them, others recalled the text from memory. A week later, a test — including questions requiring inference; in one variant the final test was drawing a concept map. |
|---|---|
| Sample | students, two experiments (numbers in the full text) · students at the authors' university, random assignment to conditions |
| Age | students, adults |
| Duration | 1 week between studying and the test |
| Measurement tool | a test with factual and inference questions; in the second experiment also a concept map drawn from memory |
Results
- Recall from memory produced more learning than a concept map — including on questions requiring inference.
- The advantage held even when the final test was drawing a concept map.
- The result was repeated on texts of the kind used in teaching science subjects.
Limitations
- University students, not primary school pupils.
- One week — it is not known what is left after months.
- Science texts; other subjects were not studied.
What we take from this into our classes
That is why in class a child answers in front of the group with the notebook closed instead of taking neat notes. Maps and notes are not bad — they are simply weaker than recall when it comes to what stays.
Source: Science 331(6018), 2011 · open
Short quizzes in sixth grade — in real lessons and grades
Roediger H. L., Agarwal P. K., McDaniel M. A., McDermott K. B., 2011
Does the testing effect also work with children, in class, on lesson material?
| Method | Three experiments in social studies lessons. Part of the material was quizzed in three short "low-stakes" multiple-choice quizzes, part was not quizzed or only presented twice. Comparison within the same class on chapter and end-of-semester exams that counted towards the grade. In the third experiment there was one quiz in class, and students were encouraged to quiz themselves online; the final test required typing in the answer. |
|---|---|
| Sample | sixth-grade students at one suburban school (numbers in the full text) · whole classes during regular lessons |
| Age | sixth grade, about 11–12 years |
| Duration | from the quizzes to the chapter exam and to the end-of-semester exam |
| Measurement tool | school chapter and end-of-semester exams; in the third experiment a test with typed-in answers |
Results
- Material quizzed in the quizzes did better on the chapter and end-of-semester exams than material not quizzed and than material presented twice.
- The effect held when the final test required typing in the answer independently, not just choosing one.
Limitations
- One school and one subject.
- Comparison within the class: quizzed and non-quizzed parts of the same material, not two groups of children.
- Effect sizes are given in the full text; we do not copy them here from the abstract.
What we take from this into our classes
This study is the closest to our work: children the age of our younger groups and real lesson material. Hence the rule that part of every session goes to what the child has just been assigned, and that checking is short and frequent.
Source: Journal of Experimental Psychology: Applied 17(4), 2011 · open
Five days of attention exercises in four- and six-year-olds
Rueda M. R., Rothbart M. K., McCandliss B. D., Saccomanno L., Posner M. I., 2005
Can a young child's executive attention be trained, and how does that compare with ordinary growing up?
| Method | Three experiments. Children were randomly divided into a training group and a control group; in two of the three experiments the control group watched cartoons during that time. The same measurement before and after; four-year-olds were also compared with six-year-olds to separate the effect of the exercises from maturation. |
|---|---|
| Sample | 73 · 49 four-year-olds and 24 six-year-olds; in each experiment, 12 children in the training group |
| Age | 4 years (52 months on average) and 6 years (77 months on average) |
| Duration | 5 exercise sessions over 2–3 weeks |
| Measurement tool | the child version of the attention network test (Child ANT) with 128-electrode EEG recording; the K-BIT intelligence test; a temperament questionnaire filled in by parents |
Results
- Children after the exercises — both four- and six-year-olds — showed a more mature pattern of executive attention than the control groups: in task results and in the EEG recording.
- Four-year-olds after the exercises improved their K-BIT intelligence test score significantly more than the control group; in six-year-olds the advantage was smaller — in vocabulary, and in the overall score at the edge of significance.
- Overall, from 4 to 6 years, attention and test scores grew strongly on their own — that is growing up, not training.
Limitations
- Groups of 12 children — very small.
- Only five sessions; whether the effect lasted for months was not checked.
- Children younger than ours (ours start at 8), computer exercises, not school material.
- Part of the EEG data was lost to recording artefacts.
What we take from this into our classes
Hence spreading the exercises out: a few short sessions over 2–3 weeks gave young children a measurable effect. The paper does not give the session length — our 10 minutes a day is a methodological choice, not a result of this study. We do not take from it a promise of "better concentration in general" — the study does not give one.
Source: Proceedings of the National Academy of Sciences 102(41), 2005 · open
Does working memory training carry over to learning? A review of 23 studies
Melby-Lervåg M., Hulme C., 2013
Do programmes that "train working memory" improve anything beyond the exercise itself?
| Method | Meta-analysis: studies with a control group (randomised and quasi-experimental) were collected and a pooled effect was calculated separately for working memory and for the skills the training was supposed to transfer to. |
|---|---|
| Sample | 23 · 23 studies, 30 group comparisons; children and adults, clinical and typically developing samples |
| Age | children and adults (various samples) |
| Duration | measurement right after training and — where there was one — at follow-up |
| Measurement tool | tests of verbal and visuospatial working memory; tests of nonverbal and verbal ability, attention, word decoding, arithmetic |
Results
- Shortly after training, working memory improved clearly.
- For verbal memory the improvement did not hold at follow-up; for visuospatial memory there was limited evidence that it did.
- There was no convincing evidence of transfer to other skills: nonverbal and verbal ability, inhibitory attention, word reading, arithmetic.
- A later, larger review by the same authors (Melby-Lervåg, Redick, Hulme, 2016: 87 publications, 145 comparisons) reached the same conclusion.
Limitations
- The authors themselves point out the age differences between samples and the mix of different clinical conditions.
- The review concerns working memory training programmes, not mnemonic techniques.
What we take from this into our classes
This is the limit we state plainly to parents: our trainers reinforce techniques, but they do not "train the brain", and we do not promise better grades or higher intelligence from them.
Source: Developmental Psychology 49(2), 2013 · open
Repetitions spread over time: a synthesis of 317 experiments
Cepeda N. J., Pashler H., Vul E., Wixted J. T., Rohrer D., 2006
Do the same repetitions give more when spread over time, and how long should the gaps be?
| Method | Meta-analysis of studies on recall of verbal material: learning "in one go" was compared with learning at several intervals, and shorter intervals with longer ones, depending on how long after learning the result was checked. |
|---|---|
| Sample | 839 · 839 measurements from 317 experiments described in 184 articles |
| Age | various samples, mainly adults |
| Duration | intervals between repetitions and time to the test — from minutes to months, depending on the study |
| Measurement tool | tasks of recalling verbal material |
Results
- Learning at intervals gave better retention than the same learning in one go.
- The interval between repetitions and the time to the test work together: the further away the test, the longer the interval between repetitions that gave the best result.
Limitations
- Verbal material, mainly in the laboratory; not school tests.
- Mainly adults; school-age children make up a minority of the samples.
What we take from this into our classes
Hence the method's third assumption: consolidation short and daily, not long and once a week, and returning to old material instead of piling on new.
Source: Psychological Bulletin 132(3), 2006 · open
Questions
Do these studies prove that Space Memory classes work?
No. They study techniques — the memory palace, quizzing, spaced repetition — on other people and in other conditions. They explain why we teach this way and not another. Whether it works for our students, we check with our own measurement and publish in the second part of the page once we have the data.
Why do some studies have no number of participants?
Because the text we have access to did not give it, and we do not write in numbers from memory. Where it says "in the full text", you can check it via the link with the study.
Why are there no studies of our own in the second part yet?
Because we only publish studies that have all eight things described, including limitations and the source of the data. We would rather leave the part empty than show a percentage nobody can check.
How is this different from the results page?
The measurement page describes what we measure with one student, and how. This page is for studies on groups — other researchers' and ours — with a description of the sample and the limitations.
Does a study without a comparison group mean anything?
It shows a change in the participants, but it does not prove the classes caused it: part of the change comes from simply repeating the test and from the child growing up. That is why we always state this in the limitations.
Is children's data published anywhere?
No. We only publish aggregate group results, without names and without any data from which a child could be recognised.
Measurement for one student, the method, and the tool.
Results and measurement
What we measure with every student, and how.
The Space Memory method
What we teach and in what order.
Space Memory cognitive tests
The tasks we measure with, and why these ones.
Want to start from your own measurement?
In a free trial lesson the coach takes the first measurement and shows where to start.
Request a callFor results of an individual student, see the page about measurement — this page is for studies on a group.