Learning · Sources

Learning and remembering

Also called study skills · remember what I read · learning a skill · retrieval practice · spaced practice · retention · understanding · ChatGPT study · AI tutor

Named sources. May contain inaccuracies or be incomplete. Not medical, legal, financial, or other professional advice.

Learning is a lasting change in what you know or can do. To remember and use an idea, learn what it means, bring it to mind without the answer in view, check the explanation, and return after a gap. Practise choosing and applying the idea as well as recalling its definition; later independent use tells you more than how smooth practice feels.

What does it mean to learn something?

Choose something you want to understand or use: a book's argument, a feature at work, or a skill someone is teaching you. These are related learning jobs, but they need different checks.

Name what you want to do with the knowledge

Being able to recognize a term, explain a relationship and perform a task are different achievements. Recognition means identifying something when it is presented; recall means producing information with fewer cues. Understanding includes explaining how ideas fit together, and transfer means using prior learning in another task. Soderstrom and Robert Bjork's review distinguishes lasting learning from performance during practice. Decide whether your goal is to remember a definition, explain a process or choose an action. That decision makes your practice question useful: a correctly recalled definition alone cannot establish that you can carry out a whole skill.

Learning Versus Performance: An Integrative Review Retrieval-based learning

Use reading to build meaning and recall to practise access

You need something intelligible to remember. Read or listen until you can identify the main idea and the unfamiliar terms, then try producing an explanation with the material set aside. Retrieval practice is this act of bringing learned knowledge to mind; it can contribute to learning rather than merely measure it. Karpicke's 2025 review distinguishes direct retrieval benefits from indirect benefits, such as discovering a gap that makes later restudy more useful. Checking an answer therefore has a second job: helping you repair what your attempt reveals. Reading, recall and correction can serve one connected purpose.

Retrieval-based learning

Keep facts connected to the decisions they support

Knowing a word's meaning can help you follow an explanation, but a collection of definitions may leave you unsure when to use them. Michelle Miller's teaching guidance asks what facts are for and recommends actively answering flashcards. Learn the vocabulary needed to distinguish the available choices, then ask which choice serves the current goal. A question about a definition and a question about using it reveal different gaps. In the reading-list example below, sorting arranges entries while filtering selects entries. Knowing those definitions helps you decide which operation a request needs; practice should check that decision too.

Helping Students Memorize: Tips from Cognitive Science Organizing Instruction and Study to Improve Student Learning

Where are you getting stuck?

Use the difficulty you notice to choose the kind of practice. The worked reading-list example later in this guide carries these choices through one task. Everyday examples and practice arrangements here are editorial illustrations, not validated personal training programs.

Bring back one idea from a page you have read

You finish an article, recognize its sentences and still cannot tell someone what it said. Close it and explain one point in your own words before reopening it. Karpicke and Janell Blunt studied undergraduates learning science texts; their recall-and-restudy condition supported better answers to conceptual questions a week later than the comparison activities. The useful lesson is to include producing an idea, not to stop reading or drawing diagrams. Notice whether you missed the meaning of a term, a relationship or a detail. Check that specific gap instead of starting the entire article again.

Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping

Give an important idea another place in a busy life

If you will need a procedure later, a single long session today may leave you little evidence about what remains. Spacing means separating learning opportunities in time. Cepeda and colleagues varied study gaps and final test delays in a fact-learning experiment; the useful gap changed with the desired retention interval. Pick one worthwhile question and attach its correction source. Put it somewhere you will encounter it again, such as a calendar reminder. Try answering before opening the source. The reminder is a practical arrangement, not an optimal timetable or a test of your commitment.

Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention Spacing and retrieval practice guide

Explain why an idea fits the new example

A tutorial makes sense, but a slightly different problem leaves you uncertain where to start. Transfer includes choosing a relevant principle and using it under changed conditions. Butler's experiments found advantages for repeated testing on later inference questions, but the most distant condition supplied a hint identifying the earlier concept. Your new task may provide no hint. State what you think applies and why, then check that link with a worked solution or instructor. If the principle is unfamiliar, ask for a clearer example before making the task harder. Identifying relevance is itself something to practise.

Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying Organizing Instruction and Study to Improve Student Learning

Use a safe mistake to locate the missing explanation

An incorrect practice answer can point to a specific misunderstanding, such as confusing a selection rule with an ordering rule. Metcalfe's review examines how errors followed by attended corrective feedback can support learning. Compare the explanation with what you actually did: which decision changed, and why? Restate the corrected idea before attempting another safe example. A red mark or a confident guess is less informative than the reason. Choose practice where errors can be corrected without harming someone; this educational principle is not permission to experiment with a real person's safety or a memory-treatment protocol.

Learning from Errors

What has learning research found?

The evidence is strongest when the method, comparison and later outcome are named. These findings support useful practice choices without promising the same effect for every learner or subject.

Include remembering in the learning activity

A practice question can do more than reveal what was learned. Agarwal, Nunes and Blunt's 2021 systematic review examined 50 classroom experiments involving 5,374 learners and found retrieval benefits across varied educational settings and materials. Its geographic coverage was narrow: only 6% of experiments occurred outside countries described as Western, educated, industrialized, rich and democratic. That strengthens the classroom case while leaving limits on universal claims. Use a small low-stakes question that requires some remembering and make correction available. Calling it a quiz does not require grades, public comparison or pressure.

Retrieval Practice Consistently Benefits Student Learning: a Systematic Review of Applied Research in Schools and Classrooms Spacing and retrieval practice guide

Spread important learning across occasions

Spacing has substantial support as a way to retain knowledge beyond the original session. Dunlosky's research-based teaching synthesis rated practice testing and distributed practice highly among the ten techniques reviewed. Cepeda's timing experiment adds an important qualification: longer gaps do not improve retention indefinitely, and useful timing depends on the later test. In ordinary life, another attempt on a separate occasion is a reasonable start. Choose material you actually expect to use. Return to its meaning as well as its wording, and use the later answer to decide whether you need more teaching.

Strengthening the Student Toolbox: Study Strategies to Boost Learning Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention

Give the correction enough attention to change your reasoning

Feedback is information about your attempt that helps improve it; it is more useful than a bare correct-or-incorrect mark when you need to understand an error. Metcalfe's review emphasizes processing the correct answer and its explanation. Read the correction, identify what you thought, and describe the difference. If you arrived at a correct result through a mistaken rule, keep working on the rule. If you remain unsure why the source is right, ask a knowledgeable person or consult a dependable worked example. Correction needs meaning, not just another answer to rehearse.

Learning from Errors

Ask why the step follows from the goal

An explanation becomes more usable when you can connect the given information, the action and the result. The Institute of Education Sciences practice guide recommends deep explanatory questions and connecting abstract concepts with concrete examples. Ask what a step is trying to achieve, which information makes it appropriate and how you would recognize a wrong choice. These questions expose the reasoning that copying a completed answer can hide. A definition and a why-question check different parts of understanding. Keep both when both matter, and use a worked example to examine the connection before expecting an independent solution.

Organizing Instruction and Study to Improve Student Learning

Alternate a worked example with an attempt of your own

A worked example displays a completed solution and the steps that produced it. The IES guide recommends combining these examples with problems learners solve. Inspect a step, explain its job, then attempt a comparable step using suitable material. You are making the decision visible enough to check, rather than copying the solution's appearance. Gurung and McCann's teaching synthesis also recommends verbalizing solution steps. If the model remains opaque, request a simpler example or instruction; removing all support does not automatically create worthwhile practice. These are educational recommendations, not proof of competence in a hazardous task.

Organizing Instruction and Study to Improve Student Learning How Should Students Study? Tips, Advice, and Pitfalls

Check learning later with the support level that matters

Performance is what you can demonstrate during or immediately after practice; learning concerns what remains available later and can be used under relevant conditions. Soderstrom and Robert Bjork's review shows why the two can diverge. A recent example, visible answer or instructor prompt may help you finish now without showing what you can do afterward. Try an important question after a gap and with fewer answer cues. Keep accessibility tools that enable participation. Removing an answer hint and removing an accommodation are different decisions. Difficulty today alone is no more a guarantee than ease today.

Learning Versus Performance: An Integrative Review Spacing and retrieval practice guide

Which practice choices depend on the task?

There is useful disagreement about implementation. Keep the comparison specific instead of turning one study into a rule for every learning session.

Fit equal or expanding gaps to your later use

Equal spacing keeps the gaps between attempts similar; expanding spacing begins with shorter gaps and widens them. Expanding practice can support early success, but that does not establish superior long-term retention. Karpicke and Roediger's vocabulary experiments found an expanding advantage after ten minutes and an equal-spacing advantage after two days. A further experiment highlighted the first recall delay rather than expansion itself. Their gaps used intervening trials, not a calendar prescription. Choose an arrangement you can carry out and assess later recall; an app's doubling pattern is one design choice, not a universal optimum.

Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention

Choose feedback timing that lets you use the explanation

Checking immediately keeps your attempt fresh; checking after a short set creates a separate review. Metcalfe describes findings that differ between laboratory and classroom settings and emphasizes attention to the correction. Ryan and colleagues studied 41 medical students receiving explanations after each question or after nine questions. They found no demonstrated timing difference on the transfer outcomes; the delayed block took a median of about thirteen minutes. This does not show that waiting weeks works equally well or that all timings are equivalent. For safe self-study, choose a manageable correction point and actually work through the explanation.

Timing’s not everything: Immediate and delayed feedback are equally beneficial for performance in formative multiple-choice testing Learning from Errors

Choose question formats with the correction process in view

Short answers require producing an answer; multiple choice offers alternatives that may support comparison as well as recognition. Kang and colleagues found that multiple-choice practice outperformed short answers without corrective feedback, while short answers produced the best delayed performance when feedback was supplied in their second experiment. In van Wijk's later study of 45 health-science students without supplied feedback, the harder short-answer format did not demonstrate better retention after three to six weeks. That small study does not prove equivalence, and the studies differed in more than feedback. Choose the thinking you need to practise and a reliable way to check it.

Test format and corrective feedback modify the effect of testing on long-term retention The battle of question formats: a comparative study of retrieval practice using very short answer questions and multiple choice questions

Mix categories when choosing between them is part of the skill

Blocked practice groups one kind of example; interleaving alternates kinds. The latter can require deciding which category or procedure applies. Brunmair and Richter synthesized 59 studies and found material-dependent results: interleaving favored paintings and some mathematical tasks, word-learning results favored blocking, and text results were uncertain. Learn enough about a category to make a meaningful comparison, then consider mixing similar types you need to distinguish. That is a practical interpretation, not a universal sequence proven best for every beginner. Interrupting unrelated tasks simply creates variety; it does not reproduce a category-learning experiment.

How do you make practice teach understanding?

A worked example: you want a reading list to show only unread books, oldest publication first. The list currently reads Cedar (2018, read), Harbor (2021, unread), Meadow (2016, unread), River (2023, read). Follow the same task from the first attempt to a later check. These invented titles and dates make the decisions visible.

Choose an operation by the result you need

First try to explain the two operations without looking at their definitions. Sorting arranges entries according to a rule; filtering selects which entries are shown. The request has two parts, so naming either operation alone is incomplete. Filtering for unread books leaves Harbor and Meadow. Ordering those two by publication year puts Meadow before Harbor. Sorting the whole list by year would put Meadow, Cedar, Harbor and River in order, but would still include books already read.

A flashcard asking 'What is filtering?' checks one fact. A question asking 'Why is sorting the whole list insufficient?' checks whether that fact supports the decision. Miller's guidance supports active, purposeful use of flashcards; it does not establish that every card should be a scenario or that factual cards have no role. Choose the question from the gap your attempt actually revealed.

Helping Students Memorize: Tips from Cognitive Science

Worked exampleSelect the books, then arrange the result

The request: unread books, oldest publication first. Follow which rows remain and which order changes. The titles and dates are invented, as in the worked example above.

1 · Start with the complete list
Book	Year	Status
Cedar	2018	Read
Harbor	2021	Unread
Meadow	2016	Unread
River	2023	Read

2 · Filter: keep only unread books
Book	Year	Status
Harbor	2021	Unread
Meadow	2016	Unread

Harbor stays ahead of Meadow at this stage. Filtering chose the rows; it did not put their years in order.

3 · Sort that selection: oldest first
Book	Year	Status
Meadow	2016	Unread
Harbor	2021	Unread

Both remaining books are unread, and 2016 comes before 2021. The result now satisfies both parts of the request.

This illustrates the distinction between filtering and sorting. The named sources in the surrounding article support the teaching guidance; they did not test this fictional reading list.

Compare examples through the rule they share

Now change the objects: a photo collection needs to show holiday photographs, oldest first. Filtering by the holiday label selects the relevant photographs; sorting by date orders that selection. The objects and fields changed, while the selection-and-ordering distinction stayed the same. Sumeracki's public teaching recommends examining the underlying principle across examples, rather than remembering only their surface details.

The new example also exposes a limit. If the photographs have no reliable holiday labels, the filter cannot identify the desired set from that missing information. More confident use of the same command will not fix it. The task needs a usable selection rule or better data before the procedure can succeed. Explaining both the shared principle and the missing condition is stronger evidence of understanding than noticing that books and photographs can both appear in lists.

Episode 15 - How Students Can Use Interleaving, Elaboration, Dual Coding, and Concrete Examples Organizing Instruction and Study to Improve Student Learning

Use a correction to change the next decision

Suppose your first reading-list attempt sorted by year and you called that 'filtering.' An accurate help page identifies two issues: the word was wrong, and the output still contained read books. Merely replacing the word in your notes corrects the vocabulary while leaving the task unfinished. Explain that the request needs an unread selection as well as an ordering rule, then check that the result contains Meadow followed by Harbor. Metcalfe's feedback review emphasizes attending to the correction, including what it means for your reasoning.

For a fresh decision, ask for all four books newest first. This time no subset was requested: sorting alone gives River, Harbor, Cedar, Meadow. If you filter merely because the last problem used filtering, the new result will be incomplete. A corrected answer becomes useful when it changes your choice under the new conditions. Check conflicting definitions with a dependable source or instructor before rehearsing them.

Learning from Errors

Return later and explain the choice before opening the answer

On a later occasion, keep the list and the request available but put the worked solution aside. Explain which operations the request needs and why. Leaving the input visible keeps the task about choosing and explaining operations, rather than unexpectedly turning it into a test of remembering fictional book dates. If recalling the definitions was also your goal, ask for those separately.

Patrice Bain's public classroom interview illustrates asking learners to produce the recap instead of supplying it for them. It is a teaching account, not an independent test of this adult routine. You can recap aloud, in writing or in another accessible form. If you can name the operations but cannot choose them, return to the comparison between requests; if the definitions themselves are missing, review those first. The later attempt has now identified a specific teaching need rather than simply produced a pass or fail.

Patrice Bain: Teaching with 'Power Tools'Spacing and retrieval practice guide

How does knowledge become usable elsewhere?

Practise recognizing when the principle is relevant

Near transfer changes relatively little from the learned task; farther transfer changes more of the context or use. Butler's 2010 experiments included later inference questions in another knowledge domain. In that condition, participants were explicitly told which learned concept was relevant, removing the need to discover the connection themselves. The result supports recall and application under those cues, not automatic relevance detection everywhere. For an everyday exercise, explain what features make a principle fit a new problem and what features would make it inappropriate. Ask for feedback on that choice as well as on the final answer.

Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying

Match the evidence to the skill you need

Larsen, Butler and Roediger randomized medical residents to testing with feedback or studying matched information on two clinical topics. Forty residents completed the study, and later knowledge tests favored repeated testing after about six months. This adds an adult professional setting to the evidence, but knowledge recall was the outcome, not patient benefit or certification. If you need a practical skill, use appropriate instruction, observed practice and the relevant competence standard. If you want to explain a book, a checked explanation may be an appropriate learning target. Keep the population and measured outcome attached to the finding.

Repeated testing improves long-term retention relative to repeated study: A randomised controlled trial Learning Versus Performance: An Integrative Review

How can you check learning when AI helps?

Finishing a supported task and learning to do it independently need separate evidence.

Keep assisted practice and independent results separate

Bastani and colleagues' 2025 PNAS study randomized classrooms at a Turkish high school to usual resources, GPT Base or GPT Tutor; nearly a thousand students participated. Both tools improved scores while available, but later unaided exams showed lower scores for GPT Base users than controls. GPT Tutor did not significantly improve or reduce unaided scores relative to controls. Its design included teacher-provided solutions, common mistakes and hints, not merely a generic request to ask questions. The trial concerns high-school mathematics and those particular systems. It cannot establish the effect of every AI tool, adult subject or safety-critical qualification.

Generative AI without guardrails can harm learning: Evidence from high school mathematics

Use help to investigate your attempt, then try without its answers

For a safe learning task, make your own attempt before asking a chatbot for a hint. Compare the explanation with an authoritative solution and attempt a similar item with the chat closed. This is a cautious editorial adaptation of the learning-versus-performance distinction, not a tested universal prompt sequence. If the explanation remains unclear or conflicts with the source, get appropriate teaching rather than treating fluency as proof. Follow your school or workplace's permitted-use rules and privacy requirements; omit private or high-stakes material from tools you do not control. No tool's confident answer is a qualification.

Generative AI without guardrails can harm learning: Evidence from high school mathematics Learning Versus Performance: An Integrative Review Guidance for generative AI in education and research

How do you keep the practice useful?

Keep only the records and supports that help you answer the next learning question. The reading-list example shows what each part is for.

Keep the explanation beside the question

Save the question with the source or worked explanation that lets you check its reasoning. For the reading list, 'Meadow, Harbor' alone tells you little: it omits the request and why those books qualify. Keep the unread-and-oldest-first condition with the distinction between selecting and ordering. Metcalfe's review explains why processing a correction matters; this is one practical way to keep it available. If the source is uncertain or disputed, retain that limit too. A bookmark and a short note may be enough.

Learning from Errors

Give the important gap a specific next action

After a check, record what needs teaching: a definition, a connection or a decision. Choosing the right operation but misreading a date needs a different response from believing that sorting removes read books. The IES guide recommends using judgments and quizzes to guide study allocation, while rating those monitoring recommendations lower than some instructional recommendations. Use the result modestly: inspect the input, revisit the distinction or ask for a clearer example. It is information about this task, not a diagnosis of ability or proof that study time is optimally allocated.

Organizing Instruction and Study to Improve Student Learning Strengthening the Student Toolbox: Study Strategies to Boost Learning

Keep the principle visible across different examples

Keep a comparison that names both what transfers and what changes. 'Books and photographs both use selection and ordering, but each needs the relevant fields' says more than collecting another screenshot. Sumeracki's public teaching emphasizes the underlying idea across concrete examples. Verify the connection with a reliable explanation; an appealing analogy can hide a crucial difference. This deliberate comparison is a practice choice, not a guarantee of transfer. You need enough context to reconstruct the reasoning, not an ever-growing collection of examples to maintain.

Episode 15 - How Students Can Use Interleaving, Elaboration, Dual Coding, and Concrete Examples Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying

Include the kind of use you care about

If you want to explain an idea to colleagues, practise a short checked explanation. If you want to perform a skill, seek the instruction and observable practice that the skill requires. The resident retention trial measured retained knowledge and therefore cannot establish every kind of professional performance. That boundary makes the practical choice clearer: use recall questions for knowledge gaps and the appropriate task for performance gaps. A successful quiz can be useful evidence without becoming permission to act beyond your competence. Keep the purpose beside the practice so that the exercise continues to match it.

Repeated testing improves long-term retention relative to repeated study: A randomised controlled trial

Let the later attempt change the support you use

A delayed attempt may show that you can recall an idea but need help selecting it, or can begin a task but cannot explain a decision. Soderstrom and Robert Bjork's review explains why practice-day smoothness cannot resolve those questions. Choose the next adjustment from what happened: a clearer model, a smaller step, another example or further instruction. Keep accommodations that make the material accessible. A missed reminder is information about the arrangement, not evidence of personal failure. Evaluate a useful capability over time instead of using difficulty, confidence or a streak as its substitute.

Learning Versus Performance: An Integrative Review Spacing and retrieval practice guide

What makes a study tip worth trusting?

Popular sayings often take one useful idea further than the evidence allows. Ask what the activity requires you to do and what later outcome supports the claim.

Use a relevant later check alongside enjoyable reading

The saying that a familiar page must be learned confuses the feeling of recognizing it with what you can later produce. Soderstrom and Robert Bjork review evidence that these can diverge. You can enjoy reading, use it to understand an argument and still decide to check an important idea later. Choose a check that matches the purpose: explain a relationship or apply a rule, rather than memorizing arbitrary wording. The check informs further learning. It is not a demand that every enjoyable book become an examination.

Learning Versus Performance: An Integrative Review

Choose representations for the subject and your access needs

A fixed learning-style label claims that teaching should match a preferred sensory category, such as visual or auditory, to improve learning. Pashler and colleagues found inadequate evidence for that matching prescription. Their review does not make individual preferences or disability access irrelevant. Choose a diagram for a spatial relationship, a clear verbal explanation for its meaning, and accommodations you need to participate. You can use different representations without adopting a fixed learner identity. The useful question is whether the material conveys the idea and lets you practise the intended use.

Learning Styles: Concepts and Evidence Spacing and retrieval practice guide

Make the challenge achievable enough to learn from

A task can be difficult because it asks for valuable thinking or because the words, instructions and background are inaccessible. Those are different problems. Karpicke's 2025 review discusses balancing retrieval effort with retrieval success; van Wijk's study also found greater immediate difficulty without a demonstrated later format advantage. Seek enough instruction or a cue to make a meaningful attempt, then reduce answer support when appropriate. A blank-page struggle is not a badge of good learning. The relevant check is what improves later, while preserving access and safe practice.

Retrieval-based learning The battle of question formats: a comparative study of retrieval practice using very short answer questions and multiple choice questions

Give mixed practice a clear comparison to make

The saying that mixing everything improves learning overlooks what was mixed and what learners needed to distinguish. Brunmair and Richter's review found variation by material, including results favoring blocked word practice and uncertain findings for text. Choose mixing when comparing related categories or selecting procedures is part of the goal. Keep unrelated interruptions from being mistaken for this learning method. Spacing concerns when you return; interleaving concerns which categories you alternate. You can use either or both, but a varied calendar alone does not show that you learned to make the needed distinction.

010

Check what a learning claim actually promises

These judgments distinguish supported claims from broader promises. A verdict applies to the claim as stated; read its explanation before turning it into a practice rule.

Use recall as one supported way to learn

Claim: recalling learned information can improve later learning. Verdict: established. Experiments and the classroom review support retrieval practice across the outcomes examined. That does not promise perfect memory, and successful recall is not the same as every kind of real-world skill. Choose a useful question, attempt it and check its meaning against a dependable explanation.

Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping Retrieval Practice Consistently Benefits Student Learning: a Systematic Review of Applied Research in Schools and Classrooms

Check retention beyond the feeling of ease

Claim: if reading feels easy, I will remember it later. Verdict: not. Immediate fluency does not establish later recall or use. Try the relevant explanation or task after a gap, with the answer removed where appropriate. Reading can still be useful and enjoyable without supporting that stronger prediction.

Learning Versus Performance: An Integrative Review

Treat a spacing schedule as a choice to evaluate

Claim: one spacing schedule is optimal for everyone. Verdict: not. Cepeda's results varied with the desired retention interval, and laboratory timing does not supply a personal calendar law. A workable reminder can help you return without being mathematically optimal. Check what remains and adjust for the material and intended use.

Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention

Select question formats by the thinking and feedback

Claim: short-answer practice always teaches more than multiple choice. Verdict: not. Format comparisons depend on questions, feedback and measured outcomes; the unconditional promise has no dependable universal winner. The small van Wijk study did not detect a reliable advantage; that does not prove the formats are equivalent. Choose meaningful questions with correction rather than making visible difficulty the deciding criterion.

Keep useful formats open to you

Claim: teaching must match my fixed learning style. Verdict: talk. The preferred-style matching prescription lacks adequate evidence in Pashler and colleagues' review. Preferences, accessible formats and subject-appropriate representations still matter. Use what communicates the idea clearly and allows the relevant practice, without letting a fixed category exclude useful alternatives.

Learning Styles: Concepts and Evidence

Mix material for the distinction you need to learn

Claim: mixing every subject always improves learning. Verdict: not. Interleaving results vary by material and task. Comparing similar categories may help a specific choice, while switching among unrelated activities is a different intervention. Name the comparison and assess the later use instead of treating variety as proof.

Check your own use after AI assistance

Claim: if a chatbot solves it during practice, I will know it later. Verdict: not. The high-school mathematics trial separated improved assisted scores from unaided results, with different outcomes for Base and guarded Tutor. It does not establish every adult AI use. Attempt a relevant task without the tool's answer and use reliable feedback to identify what you learned.

Generative AI without guardrails can harm learning: Evidence from high school mathematics

011

Try one idea you will want to use again

Choose one safe learning task and a dependable explanation. These flexible practice suggestions can be used in an accessible format; they are not a course or a required weekly schedule.

Explain one useful idea and repair one gap

Choose an idea you want to use. Set its explanation aside, state it in your own words, then reopen the source and repair one missing or mistaken connection. Karpicke and Blunt support including retrieval in learning; this short routine is an editorial adaptation of that principle. In the reading-list example, the useful correction explains which operation answers which part of the request. Sounding confident is not the check.

Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping Spacing and retrieval practice guide

Try another example and make the link explicit

Try a comparable low-stakes request that changes one meaningful condition. Explain which principle fits, why it fits and what result would reveal a wrong choice. Check the link and answer against a dependable example or instructor. Butler's transfer study helps distinguish applying a supplied concept from independently recognizing its relevance. If that link is unclear, ask for teaching rather than treating the difficulty as a verdict on intelligence.

Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying Organizing Instruction and Study to Improve Student Learning

Leave one question for a later attempt

Save that question with its correction source and choose a workable later opportunity. AERO recommends spaced retrieval in education; the reminder is an everyday adaptation, not an optimal calendar. Keep needed accessibility tools, remove answer cues where appropriate, and use the attempt to choose what needs further explanation. If AI helped earlier, try without its answer. Simplify any arrangement that costs more effort than the useful learning it supports.

Spacing and retrieval practice guide Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention Generative AI without guardrails can harm learning: Evidence from high school mathematics

012

Choose the support appropriate to the task

Keep practice safe and memory concerns in appropriate care

Learning research does not provide a memory assessment, treatment or a general brain-enhancement promise. Much of this evidence comes from students and constrained tasks; adult medical knowledge results do not automatically extend to every age, memory domain or practical outcome. NIH MedlinePlus distinguishes ordinary forgetfulness from concerning difficulty with familiar activities and advises seeing a health care provider if you are worried. Seek that assessment for noticeable changes, rather than substituting flashcards. Safety-critical skills require appropriate instruction, supervision and demonstrated competence; practising a recalled explanation alone cannot supply those safeguards.

Memory Repeated testing improves long-term retention relative to repeated study: A randomised controlled trial Learning from Errors

Use this as a reading companion

Named sources. Honest paraphrase of the finding. Not medical, legal, or financial advice. This is a reading companion for understanding learning methods and their boundaries. It does not prescribe an individualized training schedule or replace a teacher, supervisor or clinician where their support is needed. Choose accessible material and a safe way to check an attempt, and keep uncertainty visible when the source does not settle your question.

Organizing Instruction and Study to Improve Student Learning Memory

References for this topic

Sources and context

Learning Versus Performance: An Integrative Review ↗
id
learning--performance-2015 — Learning Versus Performance: An Integrative Review
kind
review
accessed at
2026-10-01
supports
Soderstrom and Robert A. Bjork distinguish temporary performance from retention and transfer, including differences across practice arrangements.
limitations
Review spans verbal/motor and foundational animal work; no animal-to-human prescription. Neither ease nor difficulty alone establishes lasting learning.
original id
performance-2015
Retrieval-based learning ↗
id
learning--retrieval-2025 — Retrieval-based learning
kind
review
accessed at
2026-10-01
supports
Reviews direct and indirect retrieval benefits, correction-guided restudy and the balance between retrieval success and effort.
limitations
Narrative review, not a new trial of the proposed adult routine. No claim that one theoretical mechanism is settled or all difficult attempts help.
original id
retrieval-2025
Helping Students Memorize: Tips from Cognitive Science ↗
id
learning--miller-memorize — Helping Students Memorize: Tips from Cognitive Science
kind
professional
accessed at
2026-10-01
supports
Miller explains the purpose and context of facts and active flashcard answering. Published online in 2019 from a 2014 teaching article.
limitations
Teaching guidance, not app or AI validation. Its generic mixing language is narrowed using the interleaving review. Sort/filter illustrations are independently authored.
original id
miller-memorize
Organizing Instruction and Study to Improve Student Learning ↗
id
learning--ies-learning — Organizing Instruction and Study to Improve Student Learning
kind
guideline
accessed at
2026-10-01
supports
Recommends alternating examples with attempts, connecting concrete and abstract ideas, explanatory questions and delayed monitoring. Its historical panel includes Metcalfe, McDaniel and Bain.
limitations
School/postsecondary guidance; everyday adult examples are adaptations. Monitoring recommendations rated Minimal, example/concrete-abstract recommendations Moderate, explanatory questions Strong in this guide; not uniformly high evidence.
original id
ies-learning
Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping ↗
id
learning--retrieval-2011 — Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping
kind
study
accessed at
2026-10-01
supports
Karpicke and Blunt’s undergraduate science-text experiments involved 80 and 120 learners, recall/restudy and delayed conceptual tests, including concept maps constructed from memory.
limitations
Specific study activities and one-week outcomes, not reading versus no reading, universal skill transfer or clinical memory improvement.
original id
retrieval-2011
Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention ↗
id
learning--spacing-2008 — Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention
kind
study
accessed at
2026-10-01
supports
Cepeda and colleagues’ fact-learning experiment found that useful study gaps depended on the final retention interval and that benefits did not grow without limit.
limitations
Abstract-level verification. No personal schedule, exact ratio, age-specific guidance or app algorithm validated.
original id
spacing-2008
Spacing and retrieval practice guide ↗
id
learning--aero-spacing — Spacing and retrieval practice guide
kind
guideline
accessed at
2026-10-01
supports
Official guide describes questions, correction, routines, repeated opportunities and reasonable adjustments for disability access; last updated September 8, 2026.
limitations
Primary/secondary school-oriented professional translation; adult reminders and exercises are editorial adaptations. Not all examples apply to every context.
original id
aero-spacing
Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying ↗
id
learning--transfer-2010 — Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying
kind
study
accessed at
2026-10-01
supports
Repeated testing with corrective feedback improved later inference performance, including new-domain use when the relevant concept was supplied as a hint.
limitations
Based on Butler’s author-uploaded primary paper. The farther-transfer hint removed spontaneous relevance detection; retrieval and feedback contributions were not isolated. Undergraduate tasks do not establish automatic transfer everywhere.
original id
transfer-2010
Learning from Errors ↗
id
learning--errors-2017 — Learning from Errors
kind
review
accessed at
2026-10-01
supports
Reviews attended correction, explanations of why an answer is right, feedback timing across settings and exceptions including amnesia.
limitations
Educational errorful practice is not a clinical rehabilitation protocol or justification for errors that could harm someone. General review, not validation of the sort/filter example.
original id
errors-2017
Retrieval Practice Consistently Benefits Student Learning: a Systematic Review of Applied Research in Schools and Classrooms ↗
id
learning--classroom-retrieval-2021 — Retrieval Practice Consistently Benefits Student Learning: a Systematic Review of Applied Research in Schools and Classrooms
kind
review
accessed at
2026-10-01
supports
Agarwal, Nunes and Blunt coded 50 classroom experiments with 5,374 learners, reporting benefits across settings and limited geographic coverage.
limitations
Publisher abstract, not the subscription full text. The synthesis does not establish individual subgroup estimates or benefits in every adult occupation.
original id
classroom-retrieval-2021
Strengthening the Student Toolbox: Study Strategies to Boost Learning ↗
id
learning--toolbox-2013 — Strengthening the Student Toolbox: Study Strategies to Boost Learning
kind
professional
accessed at
2026-10-01
supports
Dunlosky’s public article explains ten evaluated study methods, practice-testing and distributed-practice ratings, and the purposes of memory and comprehension.
limitations
Public translation of the strategy review, not a new independent trial or a fresh review of that review’s full text.
original id
toolbox-2013
How Should Students Study? Tips, Advice, and Pitfalls ↗
id
learning--gurung-study — How Should Students Study? Tips, Advice, and Pitfalls
kind
professional
accessed at
2026-10-01
supports
Gurung and McCann describe generating examples and verbalizing steps as part of effective study.
limitations
Teaching synthesis, not experimental validation of every checklist suggestion.
original id
gurung-study
Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention ↗
id
learning--schedule-2007 — Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention
kind
study
accessed at
2026-10-01
supports
Expanding vocabulary retrieval favored ten-minute results and equal spacing favored two-day results; experiment 3 varied initial recall timing.
limitations
Undergraduate vocabulary experiments using intervening trials; not prescribed days or proof that equal spacing is always superior.
original id
schedule-2007
Timing’s not everything: Immediate and delayed feedback are equally beneficial for performance in formative multiple-choice testing ↗
id
learning--feedback-timing-2024 — Timing’s not everything: Immediate and delayed feedback are equally beneficial for performance in formative multiple-choice testing
kind
study
accessed at
2026-10-01
supports
Among 41 medical students, explanations followed each item or a nine-item block taking a median of 13.11 minutes; no transfer timing difference was demonstrated.
limitations
Nonsignificance is not equivalence. Short block delay is not weeks without correction; no claim that every feedback system works equally well.
original id
feedback-timing-2024
Test format and corrective feedback modify the effect of testing on long-term retention ↗
id
learning--question-format-2007 — Test format and corrective feedback modify the effect of testing on long-term retention
kind
study
accessed at
2026-10-01
supports
Multiple-choice practice outperformed short answers without feedback; in the second experiment, feedback favored short answers on three-day final tests.
limitations
Abstract only; not every question, learner or time horizon. Differences from 2024 study cannot be attributed solely to feedback.
original id
question-format-2007
The battle of question formats: a comparative study of retrieval practice using very short answer questions and multiple choice questions ↗
id
learning--question-format-2024 — The battle of question formats: a comparative study of retrieval practice using very short answer questions and multiple choice questions
kind
study
accessed at
2026-10-01
supports
Among 45 health-science students receiving no supplied practice feedback, higher immediate short-answer effort did not demonstrate later superiority after three to six weeks.
limitations
Small health-science student sample, low-stakes extracurricular course, no no-practice control. No proof of equivalence or blanket argument against short answers.
original id
question-format-2024
Similarity matters: A meta-analysis of interleaved learning and its moderators ↗
id
learning--interleaving-2019 — Similarity matters: A meta-analysis of interleaved learning and its moderators
kind
review
accessed at
2026-10-01
supports
Brunmair and Richter review 59 studies, 238 effects and 158 samples: results depend on material, with a blocked-word advantage and uncertain text outcomes.
limitations
Accepted manuscript corroborated by the indexed published abstract, rather than the final copy of record. No mixing-everything rule or universally best beginner sequence follows.
original id
interleaving-2019
Episode 15 - How Students Can Use Interleaving, Elaboration, Dual Coding, and Concrete Examples ↗
id
learning--sumeracki-examples — Episode 15 - How Students Can Use Interleaving, Elaboration, Dual Coding, and Concrete Examples
kind
professional
accessed at
2026-10-01
supports
Sumeracki explains how to connect concrete examples through their underlying principle.
limitations
Professional translation, not a new trial of these adult illustrations. No universal endorsement of every analogy or representation.
original id
sumeracki-examples
Patrice Bain: Teaching with 'Power Tools' ↗
id
learning--bain-interview — Patrice Bain: Teaching with 'Power Tools'
kind
professional
accessed at
2026-10-01
supports
Bain’s public interview illustrates replacing a teacher’s recap with learner-generated recall and low-stakes classroom practice.
limitations
Historical practice account, not independent verification of classroom anecdotes or universal schedules.
original id
bain-interview
Repeated testing improves long-term retention relative to repeated study: A randomised controlled trial ↗
id
learning--resident-retention-2009 — Repeated testing improves long-term retention relative to repeated study: A randomised controlled trial
kind
study
accessed at
2026-10-01
supports
Randomized counterbalanced clinical-topic practice with feedback after tests; 40 residents completed the study, with a knowledge-retention advantage after approximately six months.
limitations
Primary abstract only. The outcome was retained knowledge, not patient benefit, certification or general memory health.
original id
resident-retention-2009
Generative AI without guardrails can harm learning: Evidence from high school mathematics ↗
id
learning--ai-math-2025 — Generative AI without guardrails can harm learning: Evidence from high school mathematics
kind
study
accessed at
2026-10-01
supports
Classroom-randomized study of nearly 1,000 Turkish high-school students compares GPT Base, GPT Tutor and controls: assisted gains differed from unaided exam outcomes.
limitations
One school, high-school mathematics and specific GPT-4 tools. Tutor used teacher inputs, not just a prompt slogan. No universal adult tool prescription or claim Tutor improved unaided exams.
original id
ai-math-2025
Guidance for generative AI in education and research ↗
id
learning--unesco-genai — Guidance for generative AI in education and research
kind
guideline
accessed at
2026-10-01
supports
Publication summary, updated January 2026, addresses human agency, pedagogical validation and privacy protection in educational AI.
limitations
Publication summary read, not full report. Guidance is not a causal learning trial, workplace-specific policy or a claim about current law.
original id
unesco-genai
Learning Styles: Concepts and Evidence ↗
id
learning--learning-styles-2008 — Learning Styles: Concepts and Evidence
kind
review
accessed at
2026-10-01
supports
Journal summary reports inadequate evidence for preferred-style matching as an instructional prescription.
limitations
Summary, not full paper. Does not reject accessibility, sensory needs, preferences or every possible untested version of learning styles.
original id
learning-styles-2008
Memory ↗
id
learning--memory-medlineplus — Memory
kind
official
accessed at
2026-10-01
supports
NIH/NLM summary attributed to NIA advises discussing worrying forgetfulness and difficulty with familiar activities with a health care provider.
limitations
Health-literacy guidance, not diagnosis or treatment.
original id
memory-medlineplus
Jeffrey Karpicke — official role record ↗
id
learning--professional-jeffrey-karpicke — Jeffrey Karpicke — official role record
kind
official
accessed at
2026-10-01
supports
Purdue Cognition and Learning Lab principal investigator; learning and memory researcher. Defines retrieval as a learning event and distinguishes direct benefits from correction-guided restudy. University-hosted research papers and the 2025 retrieval chapter explain learning for educators and readers.
limitations
Knowledge experiments and synthesis do not validate every adult skill or clinical memory program.
original id
professional-jeffrey-karpicke
Pooja Agarwal — official role record ↗
id
learning--professional-pooja-agarwal — Pooja Agarwal — official role record
kind
official
accessed at
2026-10-01
supports
PhD in cognitive psychology; Berklee associate professor. Coauthored the 2021 classroom review; makes low-stakes retrieval intelligible to teachers. Official profile documents RetrievalPractice.org, public teaching, workshops, books and media explanations.
limitations
Primary research supports the learning claims; books, reach and consulting are not independent validation.
original id
professional-pooja-agarwal
Janet Metcalfe — official role record ↗
id
learning--professional-janet-metcalfe — Janet Metcalfe — official role record
kind
official
accessed at
2026-10-01
supports
Columbia professor of psychology; PhD, University of Toronto. Explains attended correction, reasoning about errors and monitoring what is known. Public explanatory error-learning review and membership of the official IES practice-guide panel.
limitations
Educational feedback and metacognition are not individualized clinical rehabilitation guidance.
original id
professional-janet-metcalfe
Mark McDaniel — official role record ↗
id
learning--professional-mark-mcdaniel — Mark McDaniel — official role record
kind
official
accessed at
2026-10-01
supports
WashU professor emeritus of psychological and brain sciences; research scientist and CIRCLE director. Connects learning strategies with instructional choices; differentiates memory and use. Official practice-guide panel work and university teaching research translate memory science to instruction.
limitations
Prospective memory, content recall and transfer have different outcomes; an institutional role does not settle every technique.
original id
professional-mark-mcdaniel
Doug Rohrer — official role record ↗
id
learning--professional-doug-rohrer — Doug Rohrer — official role record
kind
official
accessed at
2026-10-01
supports
University of South Florida professor of psychology; PhD in psychology. Coauthored Cepeda's timing study; makes spacing and category-selection purposes distinct. Official profile links publicly available pedagogical research and explanatory materials.
limitations
No universal spacing timetable or endorsement of indiscriminate task mixing.
original id
professional-doug-rohrer
Elizabeth Ligon Bjork — official role record ↗
id
learning--professional-elizabeth-ligon-bjork — Elizabeth Ligon Bjork — official role record
kind
official
accessed at
2026-10-01
supports
UCLA psychologist with PhD in psychology; official lab documents former professor and distinguished research professor roles. Helps distinguish desirable challenge from a blanket celebration of difficulty. Official lab describes teaching awards and translation of memory research into instructional practices.
limitations
The lab verifies historical roles; no current administrative appointment inferred. Robert A. Bjork, not Elizabeth, coauthored the cited Soderstrom review.
original id
professional-elizabeth-ligon-bjork
John Dunlosky — official role record ↗
id
learning--professional-john-dunlosky — John Dunlosky — official role record
kind
official
accessed at
2026-10-01
supports
Kent State professor emeritus of psychological sciences. Evaluates actual study activities by their learning purpose instead of popularity. Public American Educator strategy guide and Kent teaching-institute outreach.
limitations
Strategy ratings are not individualized prescriptions or a universal number of repetitions.
original id
professional-john-dunlosky
Katherine Rawson — official role record ↗
id
learning--professional-katherine-rawson — Katherine Rawson — official role record
kind
official
accessed at
2026-10-01
supports
Kent State professor of psychological sciences; Comprehension and Memory Lab researcher. Coauthored the learning-technique evaluation; keeps comprehension, practice and retention separate. Kent’s Building a Better Student account describes her public teaching-institute discussion.
limitations
The source does not establish an exact number of successive-relearning attempts.
original id
professional-katherine-rawson
Michelle Miller — official role record ↗
id
learning--professional-michelle-miller — Michelle Miller — official role record
kind
official
accessed at
2026-10-01
supports
Northern Arizona University professor of psychological sciences. Connects facts to purpose and distinguishes active flashcard answering from passive reviewing. Public authored memorization guidance addresses teachers' questions about facts and higher-order thinking.
limitations
Practical teaching article is not a trial validating apps, every mixing procedure or AI tutors.
original id
professional-michelle-miller
Regan A. R. Gurung — official role record ↗
id
learning--professional-regan-a-r-gurung — Regan A. R. Gurung — official role record
kind
official
accessed at
2026-10-01
supports
Oregon State professor of psychological science; associate vice provost and executive director of the Center for Teaching and Learning; PhD. Makes reasoning visible by generating examples and verbalizing solution steps. Teaching workshops and an APS coauthored guide explain useful student practice.
limitations
Teaching synthesis includes nonexperimental evidence; not all-context causal proof for its recommendations.
original id
professional-regan-a-r-gurung
Patrice Bain — official role record ↗
id
learning--professional-patrice-bain — Patrice Bain — official role record
kind
official
accessed at
2026-10-01
supports
Classroom educator and author; named teacher member of the 2007 IES practice-guide panel. Shows how a teacher's recap can become a learner's remembered answer. Her St. Louis Fed educator interview gives examples of low-stakes retrieval.
limitations
Historical panel role, not an assumed current school appointment. Interview anecdotes and commercial books are not independent outcome evidence.
original id
professional-patrice-bain
Megan Sumeracki — official role record ↗
id
learning--professional-megan-sumeracki — Megan Sumeracki — official role record
kind
official
accessed at
2026-10-01
supports
Rhode Island College professor of psychology; PhD in cognitive psychology. Links different examples through a shared principle rather than memorable surface features. The Learning Scientists podcast explains the use of examples and representations.
limitations
Public teaching translation is not direct validation of every analogy or individual implementation.
original id
professional-megan-sumeracki
Andrew Butler — official role record ↗
id
learning--professional-andrew-butler — Andrew Butler — official role record
kind
official
accessed at
2026-10-01
supports
WashU professor of education and psychological and brain sciences by courtesy; PhD. Tests learning transfer while exposing the feedback and hints used in specific designs. Official profile describes teaching, translational education work and public research explanations.
limitations
Cued new-domain use is not spontaneous relevance recognition or automatic real-world transfer.
original id
professional-andrew-butler
Douglas Larsen — official role record ↗
id
learning--professional-douglas-larsen — Douglas Larsen — official role record
kind
official
accessed at
2026-10-01
supports
WashU professor of neurology and pediatrics; vice chair for educational affairs. Extends retention evidence to residents and distinguishes knowledge outcomes from clinical performance. Public institutional publications and medical-education research provide an adult professional learning example.
limitations
Retained answers do not establish patient benefit, safe independent practice or professional certification.
original id
professional-douglas-larsen
Denise Park — official role record ↗
id
learning--professional-denise-park — Denise Park — official role record
kind
official
accessed at
2026-10-01
supports
Late UT Dallas cognitive-aging researcher; official memorial lists distinguished university chair, professor and Center for Vital Longevity research director. Keeps age and the particular memory domain visible when considering how evidence generalizes. Official research profile presents lifespan cognitive research and linked public center resources.
limitations
Memorial record, not a current practitioner. No dementia-prevention or memory-treatment promise inferred.
original id
professional-denise-park
Building a Better Student ↗
id
learning--rawson-public-teaching — Building a Better Student
kind
official
accessed at
2026-10-01
supports
Kent’s November 17, 2014 account describes Rawson’s spring teaching-institute discussion and Dunlosky’s public teaching work.
limitations
Historical outreach record, not current appointment verification or workshop impact evaluation.
original id
rawson-public-teaching

Professional perspectives

Jeffrey Karpicke
Purdue Cognition and Learning Lab principal investigator; learning and memory researcher
Public usefulness
University-hosted research papers and the 2025 retrieval chapter explain learning for educators and readers.
Contribution
Defines retrieval as a learning event and distinguishes direct benefits from correction-guided restudy.
Scope and limits
Knowledge experiments and synthesis do not validate every adult skill or clinical memory program.
Reviewed
2026-10-01
Pooja Agarwal
PhD in cognitive psychology; Berklee associate professor
Public usefulness
Official profile documents RetrievalPractice.org, public teaching, workshops, books and media explanations.
Contribution
Coauthored the 2021 classroom review; makes low-stakes retrieval intelligible to teachers.
Scope and limits
Primary research supports the learning claims; books, reach and consulting are not independent validation.
Reviewed
2026-10-01
Janet Metcalfe
Columbia professor of psychology; PhD, University of Toronto
Public usefulness
Public explanatory error-learning review and membership of the official IES practice-guide panel.
Contribution
Explains attended correction, reasoning about errors and monitoring what is known.
Scope and limits
Educational feedback and metacognition are not individualized clinical rehabilitation guidance.
Reviewed
2026-10-01
Mark McDaniel
WashU professor emeritus of psychological and brain sciences; research scientist and CIRCLE director
Public usefulness
Official practice-guide panel work and university teaching research translate memory science to instruction.
Contribution
Connects learning strategies with instructional choices; differentiates memory and use.
Scope and limits
Prospective memory, content recall and transfer have different outcomes; an institutional role does not settle every technique.
Reviewed
2026-10-01
Doug Rohrer
University of South Florida professor of psychology; PhD in psychology
Public usefulness
Official profile links publicly available pedagogical research and explanatory materials.
Contribution
Coauthored Cepeda's timing study; makes spacing and category-selection purposes distinct.
Scope and limits
No universal spacing timetable or endorsement of indiscriminate task mixing.
Reviewed
2026-10-01
Elizabeth Ligon Bjork
UCLA psychologist with PhD in psychology; official lab documents former professor and distinguished research professor roles
Public usefulness
Official lab describes teaching awards and translation of memory research into instructional practices.
Contribution
Helps distinguish desirable challenge from a blanket celebration of difficulty.
Scope and limits
The lab verifies historical roles; no current administrative appointment inferred. Robert A. Bjork, not Elizabeth, coauthored the cited Soderstrom review.
Reviewed
2026-10-01
John Dunlosky
Kent State professor emeritus of psychological sciences
Public usefulness
Public American Educator strategy guide and Kent teaching-institute outreach.
Contribution
Evaluates actual study activities by their learning purpose instead of popularity.
Scope and limits
Strategy ratings are not individualized prescriptions or a universal number of repetitions.
Reviewed
2026-10-01
Katherine Rawson
Kent State professor of psychological sciences; Comprehension and Memory Lab researcher
Public usefulness
Kent's opened Building a Better Student account documents her public teaching-institute discussion.
Contribution
Coauthored the learning-technique evaluation; keeps comprehension, practice and retention separate.
Scope and limits
The source does not establish an exact number of successive-relearning attempts.
Reviewed
2026-10-01
Michelle Miller
Northern Arizona University professor of psychological sciences
Public usefulness
Public authored memorization guidance addresses teachers' questions about facts and higher-order thinking.
Contribution
Connects facts to purpose and distinguishes active flashcard answering from passive reviewing.
Scope and limits
Practical teaching article is not a trial validating apps, every mixing procedure or AI tutors.
Reviewed
2026-10-01
Regan A. R. Gurung
Oregon State professor of psychological science; associate vice provost and executive director of the Center for Teaching and Learning; PhD
Public usefulness
Official biography documents teaching workshops; opened APS coauthored guide explains student practice.
Contribution
Makes reasoning visible by generating examples and verbalizing solution steps.
Scope and limits
Teaching synthesis includes nonexperimental evidence; not all-context causal proof for its recommendations.
Reviewed
2026-10-01
Patrice Bain
Classroom educator and author; named teacher member of the 2007 IES practice-guide panel
Public usefulness
Opened St. Louis Fed educator interview offers simple low-stakes retrieval illustrations.
Contribution
Shows how a teacher's recap can become a learner's remembered answer.
Scope and limits
Historical panel role, not an assumed current school appointment. Interview anecdotes and commercial books are not independent outcome evidence.
Reviewed
2026-10-01
Megan Sumeracki
Rhode Island College professor of psychology; PhD in cognitive psychology
Public usefulness
Opened Learning Scientists podcast gives public practical explanations of examples and representations.
Contribution
Links different examples through a shared principle rather than memorable surface features.
Scope and limits
Public teaching translation is not direct validation of every analogy or individual implementation.
Reviewed
2026-10-01
Andrew Butler
WashU professor of education and psychological and brain sciences by courtesy; PhD
Public usefulness
Official profile describes teaching, translational education work and public research explanations.
Contribution
Tests learning transfer while exposing the feedback and hints used in specific designs.
Scope and limits
Cued new-domain use is not spontaneous relevance recognition or automatic real-world transfer.
Reviewed
2026-10-01
Douglas Larsen
WashU professor of neurology and pediatrics; vice chair for educational affairs
Public usefulness
Public institutional publications and medical-education research provide an adult professional learning example.
Contribution
Extends retention evidence to residents and distinguishes knowledge outcomes from clinical performance.
Scope and limits
Retained answers do not establish patient benefit, safe independent practice or professional certification.
Reviewed
2026-10-01
Denise Park
Late UT Dallas cognitive-aging researcher; official memorial lists distinguished university chair, professor and Center for Vital Longevity research director
Public usefulness
Official research profile presents lifespan cognitive research and linked public center resources.
Contribution
Keeps age and the particular memory domain visible when considering how evidence generalizes.
Scope and limits
Memorial record, not a current practitioner. No dementia-prevention or memory-treatment promise inferred.
Reviewed
2026-10-01

Keep this topic in view

Preserved original record

The earlier public profile is preserved separately from the current article. It may contain superseded wording and is not newly checked material.

Read the full original profile

slug

learning

name

Learning and remembering

browse Line

Practice testing and spaced review have the strongest study evidence.

photo

/learn/assets/topics/learning.jpg

photo Alt

An adult working through an example with a notebook at a library table

aliases

study skills
remember what I read
learning a skill
retrieval practice
spaced practice
retention
understanding
ChatGPT study
AI tutor

family

learning

heat

mid

talk

Sources

most Discussed

true

first Line

Learning is a lasting change in what you know or can do through study, experience and practice. Remembering facts, understanding ideas and applying a skill each need practice that matches the intended use.

why Line

feel Words

read
forget
remember
familiar
book
work
time
busy

common Picture

lead
Remember more of what you read
why
A page can look familiar while its main idea is difficult to explain without looking. Retrieval practice means trying to bring learned information to mind, rather than simply viewing it again. Jeffrey Karpicke and Janell Blunt found better delayed comprehension of science passages after recall and restudy than after their comparison study activities. That does not make reading or diagrams useless. For an ordinary article, try closing it, explaining one idea, then reopening it to check what you missed. This is a practical adaptation, not a promise to remember every detail.
lead
Keep learning around a busy life
why
If you need an idea next month, doing all your practice today may not be the best fit. Spacing means returning to material across separate sessions. Nicholas Cepeda and colleagues found that useful gaps between fact-learning sessions depended on how much later people were tested. There is no single interval that fits every subject or person. Choose one important question you can revisit, keep its correction source attached, and adjust after seeing what you can explain later. An ordinary calendar reminder is enough; an elaborate streak system is not required.
lead
Use an idea in a new example
why
Following an explanation and choosing the right idea in a fresh problem are different jobs. Researchers call applying prior learning to another task transfer. Andrew Butler found advantages from repeated testing with feedback on later inference questions; in the most distant transfer condition, students were explicitly told which earlier concept was relevant. The study did not demonstrate automatic recognition everywhere. Try a new, low-stakes example and say which principle you are using and why. If that link is unclear, ask for a worked example or explanation before adding more difficulty.
lead
Turn a practice mistake into a useful correction
why
A wrong answer during safe practice gives you something specific to investigate; it is not a verdict on your ability. Janet Metcalfe’s review explains how attempted answers followed by corrective feedback can support learning. The correction matters: simply rehearsing a guess is a different activity. Compare your reasoning with an accurate explanation, identify where it diverged, and try explaining the corrected idea. Keep this experiment away from real-world situations where an error could hurt someone. Learning support and the right difficulty matter more than proving you can struggle alone.

situation Quick Answers

- label
  Try this
  value
  Close the article, explain one idea from memory, then reopen it to check what you missed.
- label
  What you are practicing
  value
  Retrieval: bringing learned information to mind, alongside reading and restudy.
- label
  Return across sessions
  value
  Choose one important question to revisit and keep its correction source attached.
- label
  Choose the spacing
  value
  Adjust after seeing what you can explain later; there is no single interval for every subject or person.
- label
  Use a fresh example
  value
  Try a low-stakes problem and say which principle you are using and why.
- label
  When the link is unclear
  value
  Ask for a worked example or explanation before adding difficulty.
- label
  Use the correction
  value
  Compare your reasoning with an accurate explanation, find where it diverged, and explain the corrected idea.
- label
  Choose safe practice
  value
  Use this experiment where a mistake cannot hurt someone.

picture Lead

Remember more of what you read

watch For

These are learning ideas, not a memory assessment or treatment. Use low-stakes examples and an accurate way to check your work. Safety-critical skills need appropriate instruction, supervision and demonstrated competence; a recalled explanation is not a qualification. Choose accessible formats. The National Institute on Aging advises talking with a doctor about noticeable memory changes, especially difficulty with familiar daily tasks. A study technique is not a substitute for that assessment.

what People Say

Bring an idea to mind before checking it. Pooja Agarwal’s public teaching makes retrieval approachable: the learner does some remembering instead of having the answer continuously available. The Australian Education Research Organisation recommends this with correction and repeated opportunities in its school guidance. An adult adaptation might be to explain a spreadsheet feature on paper before opening its help page. Keep the question small enough to answer meaningfully. You are practising access to knowledge, not taking a high-pressure examination or measuring your worth.
Return to knowledge after a gap. Cepeda’s study varied the time between two learning sessions and the delay before a final test. Its central practical lesson is that timing depends on how long knowledge needs to last, not on a universal schedule. Dunlosky, Rawson and their coauthors evaluated study methods separately across learners, materials and outcomes such as memory and problem solving. That helps frame a useful choice: decide what you need to do later, then choose a manageable revisit. A correct definition and a successful new example tell you different things about what is ready to use.
Check the reasoning as well as the answer. Metcalfe identifies corrective feedback and attention to why an error occurred as important parts of learning from mistakes. That might mean noticing that you chose the right number for the wrong reason, or confused two definitions. Compare the explanation with the steps you actually took. Keep access to a dependable teacher, manual or worked solution so you can locate the difference. Then try the corrected reasoning on a safe example. The useful question is what changed in your understanding, not simply whether a mark changed from incorrect to correct.
Practise explaining connections, not only isolated facts. The Institute of Education Sciences recommends asking explanatory questions such as why a process works, alongside connecting concrete examples with abstract ideas. Mark McDaniel contributed to that educational guidance. For a safe everyday example, explain why sorting a list differs from filtering it, then describe when you would use each. This example is an editorial adaptation, not a tested STUDlearn exercise. Definitions remain useful; connecting them makes the goal clearer than collecting more disconnected flashcards.
Move between a worked example and your own attempt. The IES guide recommends alternating completed solutions with problems to solve. A worked example shows the decisions between the question and its answer, rather than leaving you to guess every step at once. Inspect one step, explain its purpose, and try a comparable step yourself. If the example is still unclear, ask for clarification or find a simpler one. The guide covers school through postsecondary education; an everyday adult adaptation needs suitable materials and a way to check the result. Help can be part of practice rather than a sign that practice failed.
Check what remains after the helpful cues are gone. Nicholas Soderstrom and Robert Bjork distinguish learning from performance during practice. Easy access to an example, recent exposure or instructor help can make immediate performance look stronger than later independent use. Elizabeth Ligon Bjork’s research makes the same distinction useful for planning practice. Revisit the important task after some time and with fewer cues. Better later performance is relevant evidence; difficulty today, by itself, is not proof that learning occurred.

disputed Notes

Choose a spacing pattern around the later task. One approach starts with a quick recall attempt and gradually widens the gaps; another keeps the gaps equal. The first aims to make early success easier while increasing the challenge. Karpicke and Roediger found a short-term advantage for expanding practice, but better two-day retention with equal spacing in their vocabulary experiments. A further experiment pointed to the timing of the first recall attempt rather than expansion itself. These gaps were measured in intervening trials, not prescribed days. Choose a workable revisit, then check what remains at the time you actually need the knowledge.
Choose a correction point you can give attention to. Immediate feedback repairs an answer while it is fresh; delayed feedback can create a separate opportunity to revisit it. Metcalfe’s review describes laboratory and classroom findings that sometimes favor different timings, with attention to the correction helping explain the difference. Ryan and colleagues compared explanations after each question with explanations after nine questions in 41 medical students. They found no demonstrated timing advantage on their transfer tests, not proof that all delays are equivalent. For safe self-study, checking each answer or reviewing a short set can both be workable; use the explanation rather than merely reveal the mark.
Choose the question format together with its feedback. Short answers ask you to produce an answer; multiple choice offers alternatives that can support recognition and comparison. In Kang and colleagues’ experiments, multiple-choice practice outperformed short answers without corrective feedback, while short answers produced the best delayed performance when feedback was supplied. Van Wijk’s later small study found no demonstrated retention advantage for its harder short-answer format. The questions and designs differed, so feedback alone cannot explain every result. Choose questions that practise the intended thinking and include a dependable correction process, rather than treating the harder-looking format as automatically better.
Group examples to learn them; mix them to compare. Blocked practice keeps one category together, while interleaving mixes categories. Grouping can help establish what an item or rule is; mixing asks you to distinguish which one applies. Brunmair and Richter’s meta-analysis favored interleaving for some materials, including paintings and mathematics, but word-learning results favored blocking and text findings were uncertain. The useful choice depends on the material and task, not a contest with one universal winner. For a new subject, notice whether you need clearer examples within a category or practice choosing between similar categories. Mixing unrelated work simply for variety does not answer either need.

talk Notes

Use later performance to judge a familiar page. Recognizing a paragraph can feel reassuring without showing that you can explain its point later. Soderstrom and Bjork’s review is a reason to add a delayed check, not to abandon reading. Make the check relevant: a definition for a definition, an explanation for an explanation, or a supervised task for a practical skill. The aim is better information about what needs attention, not a new reason to criticize an enjoyable reading session.
Match the format to the material and access needs. Harold Pashler and colleagues found inadequate evidence for assigning instruction by fixed preferred learning-style categories. That does not mean preferences, hearing, vision, language or accessibility needs are irrelevant. A diagram may be useful for a spatial relationship; an audio explanation may be easier to access on a commute. Choose a format that conveys the idea and lets you practise the intended task. A label such as “visual learner” should not close off useful alternatives.
Build challenge on a usable foundation. A task being uncomfortable is not evidence that it is improving memory. Van Wijk’s comparison gives a concrete example of greater immediate difficulty without a demonstrated later advantage. Review what made the task hard: unfamiliar vocabulary, unclear instructions, missing background or the actual reasoning. Simplifying an irrelevant barrier can make room for the worthwhile challenge. Support is compatible with learning; neither effortless familiarity nor struggle alone supplies a dependable quality measure.
Mix examples when comparison has a purpose. Interleaving means mixing categories or problem types during practice. Brunmair and Richter’s meta-analysis found that benefits varied with the material and category similarity; the result was not “mix everything.” Some tasks benefited, text findings were uncertain, and word-learning results favored blocking. Compare examples when telling them apart is part of the skill. Shuffling unrelated tasks or interrupting concentration is not the same intervention merely because both feel varied.

how People Run It

Explain one useful idea, then check it. Choose a short explanation you genuinely want to understand. Put it aside and state its main idea in your own words, aloud or in an accessible note. Then return to the source, correct one missing or mistaken part, and explain the revised version. Karpicke and Blunt’s work supports recall as part of learning; this small everyday sequence is an editorial adaptation rather than their exact study procedure. Use an accurate correction source, not a score for how confident you sounded.
Try a new example with the connection made explicit. Pick a safe example and write, “This uses the idea of … because …”. Check both the connection and the result with a reliable explanation or knowledgeable instructor. Butler’s transfer work suggests a reason to make the conceptual link visible; it does not prove that this sentence is a universal technique. If the link fails, return to a clearer example. The useful result is knowing which part you can apply and which part still needs help.
Leave one question worth returning to. Keep the source and a question that asks for the idea, not a sentence copied from it. Choose a later opportunity to try the answer before reopening your notes. AERO’s educational guidance supports combining revisits with retrieval; this is a flexible adult adaptation, not a fixed timetable. Make the question meaningful enough that its answer will be useful in your life. If the arrangement becomes burdensome, simplify it instead of treating a missed reminder as failure.
Attempt the problem before asking a chatbot. Use the first attempt to reveal what you understand and where you get stuck. Then request a hint, compare the explanation with a trusted source and try a similar item with the chat closed. This is a practical adaptation of the distinction between assisted performance and later independent learning, not a tested script for every subject. Keep private or high-stakes material out of tools you do not control, and follow the relevant school or workplace rules. The longer note explains what the high-school mathematics trial actually found.

how People Keep It

Keep the correction source attached. An answer worth practising needs a way to be checked. Note the relevant page, example or instructor explanation beside your question so you can revisit the reasoning, not merely a remembered verdict. Metcalfe’s correction work helps explain why that matters. This can be one short bookmark rather than a new filing system. Where sources disagree, keep the disagreement visible; do not turn an uncertain interpretation into an unquestioned answer simply because it fits on a card.
Review important gaps, not everything equally. Dunlosky’s strategy-evaluation work encourages asking what your study activity actually achieves. Use a later attempt to notice one important missing connection, rather than rereading an entire collection by habit. The IES guidance treats learner monitoring as useful but gives its specific monitoring recommendations a lower evidence rating than some other recommendations. Keep the method modest: it helps choose attention, not diagnose ability or guarantee an optimal allocation of every minute.
Ask what two examples have in common. Megan Sumeracki explains that learners can remember the surface of an example without grasping its underlying principle. Her public teaching suggests explicitly connecting different examples through that principle. For instance, sorting a reading list and sorting a collection of photographs both arrange items by a chosen rule, although the items look different. Ask what changes and what stays the same. This is an illustrative application, not a tested exercise. Keep enough context to make the prompt understandable, and check your connection against a reliable explanation.
Include the kind of use that matters to you. If your goal is to explain an idea to a group, practise a short explanation; if it is a practical skill, seek appropriate instruction and observable practice. Douglas Larsen’s medical-education research found benefits for later knowledge recall, not proof of improved patient outcomes. That distinction protects the useful idea from becoming an exaggerated promise. Build from knowledge to the relevant performance without pretending that a successful quiz establishes competence in every setting.
Let a later attempt guide the next adjustment. Soderstrom and Bjork’s distinction between temporary performance and more lasting learning helps make a review practical. Ask what you can now do without the original support and what still needs a cue or explanation. Record the next useful action if one is needed, not a score for personal worth. A learner can change materials, ask for teaching or reduce the task without abandoning the goal. Progress is specific evidence about a task, not a claim that memory is now permanently stronger.

claims Notes

Practise recalling, then check your answer. Claim: “Recalling learned information from memory can improve later learning” — established. Karpicke and Blunt found benefits when learners practised recalling information from memory and studied the material again. The useful claim concerns the knowledge and outcomes tested, not a guarantee of photographic memory. Choose a meaningful question, try answering without looking, then check an accurate source.
Check understanding after the cues are gone. Claim: “If reading feels easy, I will remember it later” — not what the research found. Reading easily now does not establish how much you will remember later. After a delay, try explaining the idea without the page in front of you or using it in a relevant task. Enjoyable reading can still be worthwhile without making that stronger memory claim.
Fit revisits to the task and time horizon. Claim: “One spacing schedule is optimal for everyone” — not what the research found. Cepeda’s timing results varied with the intended retention interval. A practical reminder can be useful without being optimal, and an app schedule should not be presented as a universal law.
Choose a question format by its purpose. Claim: “Short-answer practice always teaches more than multiple choice” — disputed. Format comparisons depend on questions, feedback and outcomes. Van Wijk found no demonstrated practice-format advantage in a small sample without feedback; that is neither proof of equivalence nor a universal winner.
Choose useful and accessible representations. Claim: “Teaching must match my fixed learning style” — popular talk. The preferred-style matching claim lacks the necessary evidence in Pashler’s review. A useful diagram or accessible audio explanation can still help convey a particular idea. Choose a representation for the subject and your access needs, without assigning a fixed learner identity.
Use mixed practice when comparison helps. Claim: “Mixing every subject always improves learning” — not what the research found. Brunmair and Richter found material-dependent results, not a universal mixing benefit. Their synthesis favored interleaving for some materials and blocking for word learning, with uncertain text results. Comparing similar categories has a learning purpose; repeatedly interrupting unrelated work is a different activity.
Check what you can do after the chatbot is closed. Claim: “If a chatbot can solve it during practice, I will know it later” — not what the research found. Bastani and colleagues found that a ChatGPT-style GPT-4 helper improved high-school math performance while it was available, then left students worse off than a no-AI group once it was removed. A more constrained tutor reduced that drop. Soderstrom and Bjork’s performance-versus-learning distinction names the same gap. Use the tool after your own attempt, then check unaided work; this is not a ban on AI or a claim about every subject.

deep Dive

Distinguish recognition from producing an explanation. Recognition asks whether something looks known; recall asks you to produce information with less support. Both have legitimate uses. Karpicke and Blunt’s science-passage experiments tested later conceptual questions, not just whether a sentence looked familiar. A practical contrast is seeing the definition of a term versus explaining it to someone who has not read it. Match the practice to the job rather than declaring one format good and another worthless. Choosing among alternatives can still involve meaningful reasoning.
Make a flashcard do a useful job. Michelle Miller’s teaching article asks why particular facts matter and distinguishes active answering from passive card review. Imagine learning the terms sort and filter. A card saying “Sort means arrange” may check a definition; a card asking “How would I show only unread books without changing their order?” asks you to select a use. Answer before turning it over, then check the explanation. These are different learning jobs, so the second card does not replace every definition card. Choose the one that addresses the gap you are trying to close.
Use spacing and interleaving for different purposes. Spacing separates practice over time; interleaving varies which type of item you practise. They can coexist, but they are not interchangeable. Brunmair and Richter show why the category and task matter when evaluating mixed practice. Distinguishing two similar diagram types is a different demand from memorizing unrelated word pairs. When choosing a practice arrangement, name its purpose: keeping an idea available later, learning when to choose it, or both. A varied schedule by itself is not evidence that either purpose was met.
Turn a correction into a clearer explanation. Suppose you say that filtering a list rearranges its entries. The help page explains that a filter selects which entries are shown; sorting changes their order. The useful correction is more than crossing out your sentence. Restate the distinction, choose a fresh example, and check whether you now choose the right operation. Metcalfe’s review supports attending to the reasoning behind errors; this list example is an everyday illustration. If the explanation still seems inconsistent, ask for help rather than practising a statement you do not understand.
Use a short recap to bring knowledge back into use. In a public teaching interview, Patrice Bain describes replacing a teacher-delivered recap with a question learners answer themselves. The change is small: remembering becomes part of the activity instead of something the teacher does for everyone. Before returning to a book or safe practice task, you might state the last useful idea you learned, then check it. Keep the recap short and low pressure so it leads into the next activity. Bain’s account illustrates classroom practice; it is not independent evidence for every adult adaptation or her reported classroom outcomes.
Make the bridge to a new task visible. Near transfer concerns a relatively similar task; farther transfer changes more of the context or required use. Butler deliberately supplied hints in his most distant condition so participants could identify the relevant earlier concept. Real life may not supply that hint. When teaching yourself, compare what stays the same and what changes between examples, then seek feedback on your choice of principle. This is a reason to practise the bridge, not to promise that a memory technique solves any problem.
Explain the decision between one step and the next. Gurung and McCann’s public teaching guidance recommends generating examples and verbalizing the steps used to solve a problem. The IES practice guide also pairs completed examples with attempts. To use that idea, take a safe worked example and ask why the next step follows: what is the goal, what information is available, and why is this action appropriate? Then try a comparable example. When you get stuck, return to that decision rather than copying the whole solution again. The aim is to make the reasoning visible enough to discuss and correct.
Keep the population and outcome in view. Much learning research comes from student samples and constrained tasks. Larsen’s resident study adds an adult-professional setting, but its outcome was knowledge recall, not global memory health or patient benefit. Denise Park’s cognitive-aging work also addresses questions different from a student’s immediate study strategy. Read a finding alongside who participated and what was measured. General learning guidance can be useful without becoming memory-disorder treatment, a brain-enhancement promise or a substitute for individualized instruction and care.
Check what remains when AI help is removed. Bastani and colleagues' 2025 PNAS study randomized classrooms at a Turkish high school to usual resources, GPT Base or GPT Tutor. Nearly 1,000 students took part. Both AI tools improved assisted practice scores. On later unaided exams, GPT Base users scored lower than controls; GPT Tutor users did not differ significantly from controls. The guarded tutor used teacher-supplied solutions, common mistakes and hints, so it was more than a generic request to ask questions. The trial shows why immediate task performance and independent learning need separate checks. It does not establish the best tool for every adult skill or safety-critical qualification.

related

attention
habits
career

watch For Long

Good to know. These are learning ideas, not a memory assessment or treatment. Use low-stakes examples and an accurate way to check your work. Safety-critical skills need appropriate instruction, supervision and demonstrated competence; a recalled explanation is not a qualification. Choose accessible formats. The National Institute on Aging advises talking with a doctor about noticeable memory changes, especially difficulty with familiar daily tasks. A study technique is not a substitute for that assessment.
Not advice. Named sources. Honest paraphrase of the finding. Not medical, legal, or financial advice.

sources

Jeffrey Karpicke. Purdue retrieval researcher. His science-text experiments help distinguish remembering with less support from rereading. They do not establish a complete learning system for every adult skill.
Pooja Agarwal. Berklee associate professor and public retrieval-practice educator. Her contribution is making low-stakes remembering usable in teaching; classroom translation and commercial books are not independent evidence for every adaptation.
Janet Metcalfe. Columbia psychologist studying memory, errors and metacognition. Her review helps explain why correction matters. This educational work does not supply a clinical rehabilitation protocol.
Mark McDaniel. Washington University emeritus professor and research scientist associated with CIRCLE. His strategy and instructional work helps connect research to teaching; different memory tasks still need different evidence.
Doug Rohrer. University of South Florida learning researcher. His spacing and interleaving work helps identify what a practice schedule is meant to accomplish. It does not endorse mixing every task indiscriminately.
Elizabeth Ligon Bjork. UCLA professor emeritus whose work examines learning and adaptive memory. Her contribution is distinguishing useful challenge from smooth immediate performance, not treating every difficult experience as beneficial.
John Dunlosky. Kent State professor emeritus and learning-strategy researcher. His evaluations encourage comparing actual methods and outcomes instead of impressive study routines. They are not an individual learning prescription.
Katherine Rawson. Kent State professor studying comprehension, memory and relearning. Her coauthored strategy review compares evidence across learners, materials and tasks. That is useful when deciding whether a study method addresses remembering a fact, understanding a concept or solving a problem.
Michelle Miller. Northern Arizona University professor and public educator on learning and technology. Her memorization article asks what facts will be used for and explains active flashcard answering. It offers practical teaching guidance rather than a blanket endorsement of study apps.
Regan Gurung. Oregon State professor and teaching leader. His article with Lee McCann recommends generating examples and verbalizing solution steps. It is a teaching synthesis with useful applications, not a trial validating each suggestion in every setting.
Patrice Bain. Experienced classroom teacher, public author and member of the 2007 IES practice-guide panel. In her public interview she turns a recap into a low-stakes question learners answer. That concrete classroom example helps explain retrieval without making it a large production.
Megan Sumeracki. Rhode Island College professor and public science-of-learning educator. Her podcast explanation connects different examples through their shared principle. It helps a learner look beyond the memorable setting of an example to the idea they need to use.
Andrew Butler. Washington University professor of education and, by courtesy, psychological and brain sciences. His work tests retrieval and transfer. Feedback and explicit cues in particular experiments limit claims about spontaneous transfer.
Douglas Larsen. Washington University professor of neurology and pediatrics and educational leader. His adult medical-learning research makes outcome distinctions concrete: retained knowledge does not by itself prove better clinical care.
Denise Park. The late University of Texas at Dallas cognitive-aging researcher. Her work brings age, memory and everyday function into view. Those questions are distinct from choosing a study strategy; clinical memory concerns need appropriate assessment rather than a self-directed learning exercise.

source Links

LEARN-RETRIEVAL-2011

name

Karpicke and Blunt.

kind

evidence

links

label
Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping, Science, February 11, 2011
href
https://learninglab.psych.purdue.edu/downloads/2011/2011_Karpicke_Blunt_Science.pdf

LEARN-SPACING-2008

name

Cepeda and colleagues.

kind

evidence

links

label
Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention, Psychological Science, 2008
href
https://digitalcommons.usf.edu/psy_facpub/1766/

LEARN-FEEDBACK-2017

name

Janet Metcalfe.

kind

evidence

links

label
Learning from Errors, Annual Review of Psychology, 2017
href
https://files.eric.ed.gov/fulltext/ED574569.pdf

LEARN-RESIDENTS-2009

name

Larsen, Butler and Roediger.

kind

evidence

links

label
Repeated testing improves long-term retention relative to repeated study: a randomised controlled trial, Medical Education, December 2009
href
https://pubmed.ncbi.nlm.nih.gov/19930508/

LEARN-TRANSFER-2010

name

Andrew Butler.

kind

evidence

links

label
Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying, September 2010
href
https://www.researchgate.net/publication/46094320_Repeated_Testing_Produces_Superior_Transfer_of_Learning_Relative_to_Repeated_Studying

LEARN-IES-2007

name

Institute of Education Sciences.

kind

evidence

links

label
Organizing Instruction and Study to Improve Student Learning, September 2007
href
https://ies.ed.gov/ncee/wwc/PracticeGuide/1

LEARN-PERFORMANCE-2015

name

Soderstrom and Bjork.

kind

evidence

links

label
Learning Versus Performance: An Integrative Review, 2015
href
https://bjorklab.psych.ucla.edu/wp-content/uploads/sites/13/2016/11/soderstorm_ra_learningvsperformance.pdf

LEARN-INTERLEAVING-2019

name

Brunmair and Richter.

kind

evidence

links

label
Similarity Matters: A Meta-analysis of Interleaved Learning and Its Moderators, 2019
href
https://pubmed.ncbi.nlm.nih.gov/31556629/

LEARN-FORMAT-2024

name

van Wijk and colleagues.

kind

evidence

links

label
The battle of question formats: a comparative study of retrieval practice using very short answer questions and multiple choice questions, December 30, 2024
href
https://link.springer.com/article/10.1186/s12909-024-06538-0

LEARN-STYLES-2008

name

Pashler and colleagues.

kind

evidence

links

label
Learning Styles: Concepts and Evidence, Psychological Science in the Public Interest
href
https://www.psychologicalscience.org/journals/pspi/j.1539-6053.2009.01038.x/

LEARN-AERO-2024

name

Australian Education Research Organisation.

kind

evidence

links

label
Spacing and retrieval practice guide, updated June 10, 2024
href
https://www.edresearch.edu.au/guides-resources/practice-guides/spacing-and-retrieval-practice-guide-full-publication

BASTANI-LEARNING-2024

name

Bastani H, Bastani O, Sungu A, Ge H, Kabakcı Ö, Mariman R. Generative AI Can Harm Learning. SSRN working paper, posted July 15, 2024.

kind

evidence

links

label
SSRN abstract
href
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4895486
label
DOI
href
https://doi.org/10.2139/ssrn.4895486

KARPICKE-LEARNING-2026

name

Jeffrey Karpicke

kind

professional

links

label
Purdue laboratory people
href
https://learninglab.psych.purdue.edu/people/

AGARWAL-LEARNING-2026

name

Pooja Agarwal

kind

professional

links

label
Berklee faculty profile
href
https://college.berklee.edu/faculty/pooja-k-agarwal

METCALFE-LEARNING-2026

name

Janet Metcalfe

kind

professional

links

label
Columbia profile
href
https://psychology.columbia.edu/content/janet-metcalfe

MCDANIEL-LEARNING-2026

name

Mark McDaniel

kind

professional

links

label
University profile
href
https://psych.washu.edu/people/mark-mcdaniel

ROHRER-LEARNING-2026

name

Doug Rohrer

kind

professional

links

label
USF faculty profile
href
https://www.usf.edu/arts-sciences/departments/psychology/people/faculty/doug-rohrer.aspx

BJORK-LEARNING-2026

name

Elizabeth Ligon Bjork

kind

professional

links

label
UCLA faculty record
href
https://www.psych.ucla.edu/faculty-page/ebjork/

DUNLOSKY-LEARNING-2026

name

John Dunlosky

kind

professional

links

label
Kent State profile
href
https://www.kent.edu/psychology/profile/john-dunlosky
label
His public teaching guide
href
https://www.aft.org/ae/fall2013/dunlosky

RAWSON-LEARNING-2026

name

Katherine Rawson

kind

professional

links

label
Laboratory record
href
https://www.kent.edu/psychology/comprehension-memory-lab
label
Coauthored strategy review
href
https://www.psychologicalscience.org/journals/pspi/1529100612453266/

MILLER-LEARNING-2026

name

Michelle Miller

kind

professional

links

label
NAU expert profile
href
https://experts.nau.edu/en/persons/michelle-miller-2/
label
Her memorization teaching article
href
https://www.facultyfocus.com/articles/teaching-and-learning/helping-students-memorize-tips-from-cognitive-science/

GURUNG-LEARNING-2026

name

Regan Gurung

kind

professional

links

label
OSU teaching-center profile
href
https://ctl.oregonstate.edu/directory/regan-ar-gurung
label
Coauthored study guidance
href
https://www.psychologicalscience.org/observer/how-should-students-study-tips-advice-and-pitfalls

BAIN-LEARNING-2026

name

Patrice Bain

kind

professional

links

label
Official historical panel
href
https://ies.ed.gov/ncee/wwc/PracticeGuide/1
label
Her public teaching interview
href
https://www.stlouisfed.org/education/teach-economics-podcast-series/teaching-with-power-tools-patrice-bain

SUMERACKI-LEARNING-2026

name

Megan Sumeracki

kind

professional

links

label
RIC directory
href
https://www.ric.edu/people-directory/megan-sumeracki
label
Public examples explanation
href
https://www.learningscientists.org/learning-scientists-podcast/2018/3/21/episode-15-how-students-can-use-interleaving-elaboration-dual-coding-and-concrete-examples

BUTLER-LEARNING-2026

name

Andrew Butler

kind

professional

links

label
Washington University Education profile
href
https://education.washu.edu/people/andrew-c-butler

LARSEN-LEARNING-2026

name

Douglas Larsen

kind

professional

links

label
University research profile
href
https://profiles.wustl.edu/en/persons/douglas-larsen/

PARK-LEARNING-2026

name

Denise Park

kind

professional

links

label
UT Dallas record
href
https://profiles.utdallas.edu/denise

LEARN-SCHEDULE-2007

name

Karpicke and Roediger.

kind

evidence

links

label
Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention, July 2007
href
https://learninglab.psych.purdue.edu/downloads/2007/2007_Karpicke_Roediger_JEPLMC.pdf

LEARN-TIMING-2024

name

Ryan and colleagues.

kind

evidence

links

label
Timing’s not everything: Immediate and delayed feedback are equally beneficial for performance in formative multiple-choice testing, July 2024
href
https://asmepublications.onlinelibrary.wiley.com/doi/10.1111/medu.15287

LEARN-FORMAT-2007

name

Kang, McDermott and Roediger.

kind

evidence

links

label
Test format and corrective feedback modify the effect of testing on long-term retention, July 2007
href
https://profiles.wustl.edu/en/publications/test-format-and-corrective-feedback-modify-the-effect-of-testing-/

LEARN-TECHNIQUES-2013

name

Dunlosky, Rawson, Marsh, Nathan and Willingham.

kind

evidence

links

label
Improving Students’ Learning With Effective Learning Techniques, January 2013
href
https://www.psychologicalscience.org/journals/pspi/1529100612453266/

LEARN-TOOLBOX-2013

name

John Dunlosky.

kind

evidence

links

label
Strengthening the Student Toolbox: Study Strategies to Boost Learning, American Educator, Fall2013
href
https://www.aft.org/ae/fall2013/dunlosky

LEARN-MILLER-2019

name

Michelle Miller.

kind

evidence

links

label
Helping Students Memorize: Tips from Cognitive Science, April12,2019
href
https://www.facultyfocus.com/articles/teaching-and-learning/helping-students-memorize-tips-from-cognitive-science/

LEARN-STUDY-2011

name

Regan Gurung and Lee McCann.

kind

evidence

links

label
How Should Students Study? Tips, Advice, and Pitfalls, April1,2011
href
https://www.psychologicalscience.org/observer/how-should-students-study-tips-advice-and-pitfalls

LEARN-BAIN-INTERVIEW

name

Patrice Bain, interviewed by Scott Wolla.

kind

evidence

links

label
Teaching with Power Tools, St.Louis Fed Teach Economics podcast
href
https://www.stlouisfed.org/education/teach-economics-podcast-series/teaching-with-power-tools-patrice-bain

LEARN-EXAMPLES-2018

name

Megan Sumeracki and Yana Weinstein.

kind

evidence

links

label
Learning Scientists Podcast Episode15: How Students Can Use Interleaving, Elaboration, Dual Coding, and Concrete Examples, March21,2018
href
https://www.learningscientists.org/learning-scientists-podcast/2018/3/21/episode-15-how-students-can-use-interleaving-elaboration-dual-coding-and-concrete-examples

LEARN-MEMORY-HELP-2023

name

NIH National Institute on Aging.

kind

evidence

links

label
Memory Problems, Forgetfulness, and Aging, reviewed November22,2023
href
https://www.nia.nih.gov/health/memory-loss-and-forgetfulness/memory-problems-forgetfulness-and-aging

BASTANI-PNAS-2025-CHECKED

name

Bastani et al. Generative AI without guardrails can harm learning: Evidence from high school mathematics. Source

kind

evidence

links

label
Source
href
https://doi.org/10.1073/pnas.2422633122

source Claims

situations.remember

LEARN-RETRIEVAL-2011
KARPICKE-LEARNING-2026

situations.return

LEARN-SPACING-2008

situations.apply

LEARN-TRANSFER-2010
BUTLER-LEARNING-2026

situations.mistakes

LEARN-FEEDBACK-2017
METCALFE-LEARNING-2026

split.established.0

LEARN-AERO-2024
AGARWAL-LEARNING-2026

split.established.1

LEARN-SPACING-2008
LEARN-TECHNIQUES-2013
RAWSON-LEARNING-2026
DUNLOSKY-LEARNING-2026
ROHRER-LEARNING-2026

split.established.2

LEARN-FEEDBACK-2017

split.established.3

LEARN-IES-2007
MCDANIEL-LEARNING-2026

split.established.4

LEARN-IES-2007

split.established.5

LEARN-PERFORMANCE-2015
BJORK-LEARNING-2026

split.disputed.0

LEARN-SCHEDULE-2007

split.disputed.1

LEARN-FEEDBACK-2017
LEARN-TIMING-2024

split.disputed.2

LEARN-FORMAT-2007
LEARN-FORMAT-2024

split.disputed.3

LEARN-INTERLEAVING-2019

split.talk.0

LEARN-PERFORMANCE-2015

split.talk.1

LEARN-STYLES-2008

split.talk.2

LEARN-FORMAT-2024

split.talk.3

LEARN-INTERLEAVING-2019

move

LEARN-RETRIEVAL-2011

try.0

LEARN-TRANSFER-2010

try.1

LEARN-AERO-2024

try.2

BASTANI-LEARNING-2024
LEARN-PERFORMANCE-2015
AGARWAL-LEARNING-2026
BJORK-LEARNING-2026
BASTANI-PNAS-2025-CHECKED

keep.0

LEARN-FEEDBACK-2017

keep.1

LEARN-IES-2007
LEARN-TOOLBOX-2013
DUNLOSKY-LEARNING-2026

keep.2

LEARN-EXAMPLES-2018
SUMERACKI-LEARNING-2026

keep.3

LEARN-RESIDENTS-2009
LARSEN-LEARNING-2026

keep.4

LEARN-PERFORMANCE-2015

deep.0

LEARN-RETRIEVAL-2011

deep.1

LEARN-MILLER-2019
MILLER-LEARNING-2026

deep.2

LEARN-INTERLEAVING-2019

deep.3

LEARN-FEEDBACK-2017

deep.4

LEARN-BAIN-INTERVIEW
BAIN-LEARNING-2026

deep.5

LEARN-TRANSFER-2010

deep.6

LEARN-IES-2007
LEARN-STUDY-2011
GURUNG-LEARNING-2026

deep.7

LEARN-RESIDENTS-2009
PARK-LEARNING-2026

deep.8

BASTANI-LEARNING-2024
LEARN-PERFORMANCE-2015
BJORK-LEARNING-2026
MILLER-LEARNING-2026
BASTANI-PNAS-2025-CHECKED

claims.0

LEARN-RETRIEVAL-2011

claims.1

LEARN-PERFORMANCE-2015

claims.2

LEARN-SPACING-2008

claims.3

LEARN-FORMAT-2024

claims.4

LEARN-STYLES-2008

claims.5

LEARN-INTERLEAVING-2019

claims.6

BASTANI-LEARNING-2024
LEARN-PERFORMANCE-2015
BJORK-LEARNING-2026

watch For

LEARN-FEEDBACK-2017
LEARN-RESIDENTS-2009
LEARN-MEMORY-HELP-2023

source Situation Keys

situations.remember
situations.return
situations.apply
situations.mistakes

professional Source Ids

KARPICKE-LEARNING-2026
AGARWAL-LEARNING-2026
METCALFE-LEARNING-2026
MCDANIEL-LEARNING-2026
ROHRER-LEARNING-2026
BJORK-LEARNING-2026
DUNLOSKY-LEARNING-2026
RAWSON-LEARNING-2026
MILLER-LEARNING-2026
GURUNG-LEARNING-2026
BAIN-LEARNING-2026
SUMERACKI-LEARNING-2026
BUTLER-LEARNING-2026
LARSEN-LEARNING-2026
PARK-LEARNING-2026

house Pointers

positive Guidance

status: DONE
reviewed At: 2026-09-05
professional Count: 15
scope: content-and-positive-titles-only

learn Article

article

format version

presentation

complete-card

overview

sections

id
understanding-learning
role
explanation
title
What does it mean to learn something?
intro
Choose something you want to understand or use: a book's argument, a feature at work, or a skill someone is teaching you. These are related learning jobs, but they need different checks.
items
- label
  Name what you want to do with the knowledge
  text
  Being able to recognize a term, explain a relationship and perform a task are different achievements. Recognition means identifying something when it is presented; recall means producing information with fewer cues. Understanding includes explaining how ideas fit together, and transfer means using prior learning in another task. Soderstrom and Robert Bjork's review distinguishes lasting learning from performance during practice. Decide whether your goal is to remember a definition, explain a process or choose an action. That decision makes your practice question useful: a correctly recalled definition alone cannot establish that you can carry out a whole skill.
  source ids
  performance-2015
  retrieval-2025
- label
  Use reading to build meaning and recall to practise access
  text
  You need something intelligible to remember. Read or listen until you can identify the main idea and the unfamiliar terms, then try producing an explanation with the material set aside. Retrieval practice is this act of bringing learned knowledge to mind; it can contribute to learning rather than merely measure it. Karpicke's 2025 review distinguishes direct retrieval benefits from indirect benefits, such as discovering a gap that makes later restudy more useful. Checking an answer therefore has a second job: helping you repair what your attempt reveals. Reading, recall and correction can serve one connected purpose.
  source ids
  retrieval-2025
- label
  Keep facts connected to the decisions they support
  text
  Knowing a word's meaning can help you follow an explanation, but a collection of definitions may leave you unsure when to use them. Michelle Miller's teaching guidance asks what facts are for and recommends actively answering flashcards. Learn the vocabulary needed to distinguish the available choices, then ask which choice serves the current goal. A question about a definition and a question about using it reveal different gaps. In the reading-list example below, sorting arranges entries while filtering selects entries. Knowing those definitions helps you decide which operation a request needs; practice should check that decision too.
  source ids
  miller-memorize
  ies-learning
id
ordinary-learning-situations
role
application
title
Where are you getting stuck?
intro
Use the difficulty you notice to choose the kind of practice. The worked reading-list example later in this guide carries these choices through one task. Everyday examples and practice arrangements here are editorial illustrations, not validated personal training programs.
items
- label
  Bring back one idea from a page you have read
  text
  You finish an article, recognize its sentences and still cannot tell someone what it said. Close it and explain one point in your own words before reopening it. Karpicke and Janell Blunt studied undergraduates learning science texts; their recall-and-restudy condition supported better answers to conceptual questions a week later than the comparison activities. The useful lesson is to include producing an idea, not to stop reading or drawing diagrams. Notice whether you missed the meaning of a term, a relationship or a detail. Check that specific gap instead of starting the entire article again.
  source ids
  retrieval-2011
- label
  Give an important idea another place in a busy life
  text
  If you will need a procedure later, a single long session today may leave you little evidence about what remains. Spacing means separating learning opportunities in time. Cepeda and colleagues varied study gaps and final test delays in a fact-learning experiment; the useful gap changed with the desired retention interval. Pick one worthwhile question and attach its correction source. Put it somewhere you will encounter it again, such as a calendar reminder. Try answering before opening the source. The reminder is a practical arrangement, not an optimal timetable or a test of your commitment.
  source ids
  spacing-2008
  aero-spacing
- label
  Explain why an idea fits the new example
  text
  A tutorial makes sense, but a slightly different problem leaves you uncertain where to start. Transfer includes choosing a relevant principle and using it under changed conditions. Butler's experiments found advantages for repeated testing on later inference questions, but the most distant condition supplied a hint identifying the earlier concept. Your new task may provide no hint. State what you think applies and why, then check that link with a worked solution or instructor. If the principle is unfamiliar, ask for a clearer example before making the task harder. Identifying relevance is itself something to practise.
  source ids
  transfer-2010
  ies-learning
- label
  Use a safe mistake to locate the missing explanation
  text
  An incorrect practice answer can point to a specific misunderstanding, such as confusing a selection rule with an ordering rule. Metcalfe's review examines how errors followed by attended corrective feedback can support learning. Compare the explanation with what you actually did: which decision changed, and why? Restate the corrected idea before attempting another safe example. A red mark or a confident guess is less informative than the reason. Choose practice where errors can be corrected without harming someone; this educational principle is not permission to experiment with a real person's safety or a memory-treatment protocol.
  source ids
  errors-2017
id
findings-to-build-on
role
findings
title
What has learning research found?
intro
The evidence is strongest when the method, comparison and later outcome are named. These findings support useful practice choices without promising the same effect for every learner or subject.
items
- label
  Include remembering in the learning activity
  text
  A practice question can do more than reveal what was learned. Agarwal, Nunes and Blunt's 2021 systematic review examined 50 classroom experiments involving 5,374 learners and found retrieval benefits across varied educational settings and materials. Its geographic coverage was narrow: only 6% of experiments occurred outside countries described as Western, educated, industrialized, rich and democratic. That strengthens the classroom case while leaving limits on universal claims. Use a small low-stakes question that requires some remembering and make correction available. Calling it a quiz does not require grades, public comparison or pressure.
  source ids
  classroom-retrieval-2021
  aero-spacing
- label
  Spread important learning across occasions
  text
  Spacing has substantial support as a way to retain knowledge beyond the original session. Dunlosky's research-based teaching synthesis rated practice testing and distributed practice highly among the ten techniques reviewed. Cepeda's timing experiment adds an important qualification: longer gaps do not improve retention indefinitely, and useful timing depends on the later test. In ordinary life, another attempt on a separate occasion is a reasonable start. Choose material you actually expect to use. Return to its meaning as well as its wording, and use the later answer to decide whether you need more teaching.
  source ids
  toolbox-2013
  spacing-2008
- label
  Give the correction enough attention to change your reasoning
  text
  Feedback is information about your attempt that helps improve it; it is more useful than a bare correct-or-incorrect mark when you need to understand an error. Metcalfe's review emphasizes processing the correct answer and its explanation. Read the correction, identify what you thought, and describe the difference. If you arrived at a correct result through a mistaken rule, keep working on the rule. If you remain unsure why the source is right, ask a knowledgeable person or consult a dependable worked example. Correction needs meaning, not just another answer to rehearse.
  source ids
  errors-2017
- label
  Ask why the step follows from the goal
  text
  An explanation becomes more usable when you can connect the given information, the action and the result. The Institute of Education Sciences practice guide recommends deep explanatory questions and connecting abstract concepts with concrete examples. Ask what a step is trying to achieve, which information makes it appropriate and how you would recognize a wrong choice. These questions expose the reasoning that copying a completed answer can hide. A definition and a why-question check different parts of understanding. Keep both when both matter, and use a worked example to examine the connection before expecting an independent solution.
  source ids
  ies-learning
- label
  Alternate a worked example with an attempt of your own
  text
  A worked example displays a completed solution and the steps that produced it. The IES guide recommends combining these examples with problems learners solve. Inspect a step, explain its job, then attempt a comparable step using suitable material. You are making the decision visible enough to check, rather than copying the solution's appearance. Gurung and McCann's teaching synthesis also recommends verbalizing solution steps. If the model remains opaque, request a simpler example or instruction; removing all support does not automatically create worthwhile practice. These are educational recommendations, not proof of competence in a hazardous task.
  source ids
  ies-learning
  gurung-study
- label
  Check learning later with the support level that matters
  text
  Performance is what you can demonstrate during or immediately after practice; learning concerns what remains available later and can be used under relevant conditions. Soderstrom and Robert Bjork's review shows why the two can diverge. A recent example, visible answer or instructor prompt may help you finish now without showing what you can do afterward. Try an important question after a gap and with fewer answer cues. Keep accessibility tools that enable participation. Removing an answer hint and removing an accommodation are different decisions. Difficulty today alone is no more a guarantee than ease today.
  source ids
  performance-2015
  aero-spacing
id
choose-practice-details
role
perspectives
title
Which practice choices depend on the task?
intro
There is useful disagreement about implementation. Keep the comparison specific instead of turning one study into a rule for every learning session.
items
- label
  Fit equal or expanding gaps to your later use
  text
  Equal spacing keeps the gaps between attempts similar; expanding spacing begins with shorter gaps and widens them. Expanding practice can support early success, but that does not establish superior long-term retention. Karpicke and Roediger's vocabulary experiments found an expanding advantage after ten minutes and an equal-spacing advantage after two days. A further experiment highlighted the first recall delay rather than expansion itself. Their gaps used intervening trials, not a calendar prescription. Choose an arrangement you can carry out and assess later recall; an app's doubling pattern is one design choice, not a universal optimum.
  source ids
  schedule-2007
  spacing-2008
- label
  Choose feedback timing that lets you use the explanation
  text
  Checking immediately keeps your attempt fresh; checking after a short set creates a separate review. Metcalfe describes findings that differ between laboratory and classroom settings and emphasizes attention to the correction. Ryan and colleagues studied 41 medical students receiving explanations after each question or after nine questions. They found no demonstrated timing difference on the transfer outcomes; the delayed block took a median of about thirteen minutes. This does not show that waiting weeks works equally well or that all timings are equivalent. For safe self-study, choose a manageable correction point and actually work through the explanation.
  source ids
  feedback-timing-2024
  errors-2017
- label
  Choose question formats with the correction process in view
  text
  Short answers require producing an answer; multiple choice offers alternatives that may support comparison as well as recognition. Kang and colleagues found that multiple-choice practice outperformed short answers without corrective feedback, while short answers produced the best delayed performance when feedback was supplied in their second experiment. In van Wijk's later study of 45 health-science students without supplied feedback, the harder short-answer format did not demonstrate better retention after three to six weeks. That small study does not prove equivalence, and the studies differed in more than feedback. Choose the thinking you need to practise and a reliable way to check it.
  source ids
  question-format-2007
  question-format-2024
- label
  Mix categories when choosing between them is part of the skill
  text
  Blocked practice groups one kind of example; interleaving alternates kinds. The latter can require deciding which category or procedure applies. Brunmair and Richter synthesized 59 studies and found material-dependent results: interleaving favored paintings and some mathematical tasks, word-learning results favored blocking, and text results were uncertain. Learn enough about a category to make a meaningful comparison, then consider mixing similar types you need to distinguish. That is a practical interpretation, not a universal sequence proven best for every beginner. Interrupting unrelated tasks simply creates variety; it does not reproduce a category-learning experiment.
  source ids
  interleaving-2019
id
design-useful-questions
role
application
title
How do you make practice teach understanding?
intro
A worked example: you want a reading list to show only unread books, oldest publication first. The list currently reads Cedar (2018, read), Harbor (2021, unread), Meadow (2016, unread), River (2023, read). Follow the same task from the first attempt to a later check. These invented titles and dates make the decisions visible.
items
- label
  Choose an operation by the result you need
  text
  First try to explain the two operations without looking at their definitions. Sorting arranges entries according to a rule; filtering selects which entries are shown. The request has two parts, so naming either operation alone is incomplete. Filtering for unread books leaves Harbor and Meadow. Ordering those two by publication year puts Meadow before Harbor. Sorting the whole list by year would put Meadow, Cedar, Harbor and River in order, but would still include books already read. A flashcard asking 'What is filtering?' checks one fact. A question asking 'Why is sorting the whole list insufficient?' checks whether that fact supports the decision. Miller's guidance supports active, purposeful use of flashcards; it does not establish that every card should be a scenario or that factual cards have no role. Choose the question from the gap your attempt actually revealed.
  source ids
  miller-memorize
- label
  Compare examples through the rule they share
  text
  Now change the objects: a photo collection needs to show holiday photographs, oldest first. Filtering by the holiday label selects the relevant photographs; sorting by date orders that selection. The objects and fields changed, while the selection-and-ordering distinction stayed the same. Sumeracki's public teaching recommends examining the underlying principle across examples, rather than remembering only their surface details. The new example also exposes a limit. If the photographs have no reliable holiday labels, the filter cannot identify the desired set from that missing information. More confident use of the same command will not fix it. The task needs a usable selection rule or better data before the procedure can succeed. Explaining both the shared principle and the missing condition is stronger evidence of understanding than noticing that books and photographs can both appear in lists.
  source ids
  sumeracki-examples
  ies-learning
- label
  Use a correction to change the next decision
  text
  Suppose your first reading-list attempt sorted by year and you called that 'filtering.' An accurate help page identifies two issues: the word was wrong, and the output still contained read books. Merely replacing the word in your notes corrects the vocabulary while leaving the task unfinished. Explain that the request needs an unread selection as well as an ordering rule, then check that the result contains Meadow followed by Harbor. Metcalfe's feedback review emphasizes attending to the correction, including what it means for your reasoning. For a fresh decision, ask for all four books newest first. This time no subset was requested: sorting alone gives River, Harbor, Cedar, Meadow. If you filter merely because the last problem used filtering, the new result will be incomplete. A corrected answer becomes useful when it changes your choice under the new conditions. Check conflicting definitions with a dependable source or instructor before rehearsing them.
  source ids
  errors-2017
- label
  Return later and explain the choice before opening the answer
  text
  On a later occasion, keep the list and the request available but put the worked solution aside. Explain which operations the request needs and why. Leaving the input visible keeps the task about choosing and explaining operations, rather than unexpectedly turning it into a test of remembering fictional book dates. If recalling the definitions was also your goal, ask for those separately. Patrice Bain's public classroom interview illustrates asking learners to produce the recap instead of supplying it for them. It is a teaching account, not an independent test of this adult routine. You can recap aloud, in writing or in another accessible form. If you can name the operations but cannot choose them, return to the comparison between requests; if the definitions themselves are missing, review those first. The later attempt has now identified a specific teaching need rather than simply produced a pass or fail.
  source ids
  bain-interview
  aero-spacing
id
bridge-to-another-task
role
application
title
How does knowledge become usable elsewhere?
items
- label
  Practise recognizing when the principle is relevant
  text
  Near transfer changes relatively little from the learned task; farther transfer changes more of the context or use. Butler's 2010 experiments included later inference questions in another knowledge domain. In that condition, participants were explicitly told which learned concept was relevant, removing the need to discover the connection themselves. The result supports recall and application under those cues, not automatic relevance detection everywhere. For an everyday exercise, explain what features make a principle fit a new problem and what features would make it inappropriate. Ask for feedback on that choice as well as on the final answer.
  source ids
  transfer-2010
- label
  Match the evidence to the skill you need
  text
  Larsen, Butler and Roediger randomized medical residents to testing with feedback or studying matched information on two clinical topics. Forty residents completed the study, and later knowledge tests favored repeated testing after about six months. This adds an adult professional setting to the evidence, but knowledge recall was the outcome, not patient benefit or certification. If you need a practical skill, use appropriate instruction, observed practice and the relevant competence standard. If you want to explain a book, a checked explanation may be an appropriate learning target. Keep the population and measured outcome attached to the finding.
  source ids
  resident-retention-2009
  performance-2015
id
learning-with-ai
role
findings
title
How can you check learning when AI helps?
intro
Finishing a supported task and learning to do it independently need separate evidence.
items
- label
  Keep assisted practice and independent results separate
  text
  Bastani and colleagues' 2025 PNAS study randomized classrooms at a Turkish high school to usual resources, GPT Base or GPT Tutor; nearly a thousand students participated. Both tools improved scores while available, but later unaided exams showed lower scores for GPT Base users than controls. GPT Tutor did not significantly improve or reduce unaided scores relative to controls. Its design included teacher-provided solutions, common mistakes and hints, not merely a generic request to ask questions. The trial concerns high-school mathematics and those particular systems. It cannot establish the effect of every AI tool, adult subject or safety-critical qualification.
  source ids
  ai-math-2025
- label
  Use help to investigate your attempt, then try without its answers
  text
  For a safe learning task, make your own attempt before asking a chatbot for a hint. Compare the explanation with an authoritative solution and attempt a similar item with the chat closed. This is a cautious editorial adaptation of the learning-versus-performance distinction, not a tested universal prompt sequence. If the explanation remains unclear or conflicts with the source, get appropriate teaching rather than treating fluency as proof. Follow your school or workplace's permitted-use rules and privacy requirements; omit private or high-stakes material from tools you do not control. No tool's confident answer is a qualification.
  source ids
  ai-math-2025
  performance-2015
  unesco-genai
id
keep-learning-manageable
role
application
title
How do you keep the practice useful?
intro
Keep only the records and supports that help you answer the next learning question. The reading-list example shows what each part is for.
items
- label
  Keep the explanation beside the question
  text
  Save the question with the source or worked explanation that lets you check its reasoning. For the reading list, 'Meadow, Harbor' alone tells you little: it omits the request and why those books qualify. Keep the unread-and-oldest-first condition with the distinction between selecting and ordering. Metcalfe's review explains why processing a correction matters; this is one practical way to keep it available. If the source is uncertain or disputed, retain that limit too. A bookmark and a short note may be enough.
  source ids
  errors-2017
- label
  Give the important gap a specific next action
  text
  After a check, record what needs teaching: a definition, a connection or a decision. Choosing the right operation but misreading a date needs a different response from believing that sorting removes read books. The IES guide recommends using judgments and quizzes to guide study allocation, while rating those monitoring recommendations lower than some instructional recommendations. Use the result modestly: inspect the input, revisit the distinction or ask for a clearer example. It is information about this task, not a diagnosis of ability or proof that study time is optimally allocated.
  source ids
  ies-learning
  toolbox-2013
- label
  Keep the principle visible across different examples
  text
  Keep a comparison that names both what transfers and what changes. 'Books and photographs both use selection and ordering, but each needs the relevant fields' says more than collecting another screenshot. Sumeracki's public teaching emphasizes the underlying idea across concrete examples. Verify the connection with a reliable explanation; an appealing analogy can hide a crucial difference. This deliberate comparison is a practice choice, not a guarantee of transfer. You need enough context to reconstruct the reasoning, not an ever-growing collection of examples to maintain.
  source ids
  sumeracki-examples
  transfer-2010
- label
  Include the kind of use you care about
  text
  If you want to explain an idea to colleagues, practise a short checked explanation. If you want to perform a skill, seek the instruction and observable practice that the skill requires. The resident retention trial measured retained knowledge and therefore cannot establish every kind of professional performance. That boundary makes the practical choice clearer: use recall questions for knowledge gaps and the appropriate task for performance gaps. A successful quiz can be useful evidence without becoming permission to act beyond your competence. Keep the purpose beside the practice so that the exercise continues to match it.
  source ids
  resident-retention-2009
- label
  Let the later attempt change the support you use
  text
  A delayed attempt may show that you can recall an idea but need help selecting it, or can begin a task but cannot explain a decision. Soderstrom and Robert Bjork's review explains why practice-day smoothness cannot resolve those questions. Choose the next adjustment from what happened: a clearer model, a smaller step, another example or further instruction. Keep accommodations that make the material accessible. A missed reminder is information about the arrangement, not evidence of personal failure. Evaluate a useful capability over time instead of using difficulty, confidence or a streak as its substitute.
  source ids
  performance-2015
  aero-spacing
id
evaluate-popular-study-advice
role
cautions
title
What makes a study tip worth trusting?
intro
Popular sayings often take one useful idea further than the evidence allows. Ask what the activity requires you to do and what later outcome supports the claim.
items
- label
  Use a relevant later check alongside enjoyable reading
  text
  The saying that a familiar page must be learned confuses the feeling of recognizing it with what you can later produce. Soderstrom and Robert Bjork review evidence that these can diverge. You can enjoy reading, use it to understand an argument and still decide to check an important idea later. Choose a check that matches the purpose: explain a relationship or apply a rule, rather than memorizing arbitrary wording. The check informs further learning. It is not a demand that every enjoyable book become an examination.
  source ids
  performance-2015
- label
  Choose representations for the subject and your access needs
  text
  A fixed learning-style label claims that teaching should match a preferred sensory category, such as visual or auditory, to improve learning. Pashler and colleagues found inadequate evidence for that matching prescription. Their review does not make individual preferences or disability access irrelevant. Choose a diagram for a spatial relationship, a clear verbal explanation for its meaning, and accommodations you need to participate. You can use different representations without adopting a fixed learner identity. The useful question is whether the material conveys the idea and lets you practise the intended use.
  source ids
  learning-styles-2008
  aero-spacing
- label
  Make the challenge achievable enough to learn from
  text
  A task can be difficult because it asks for valuable thinking or because the words, instructions and background are inaccessible. Those are different problems. Karpicke's 2025 review discusses balancing retrieval effort with retrieval success; van Wijk's study also found greater immediate difficulty without a demonstrated later format advantage. Seek enough instruction or a cue to make a meaningful attempt, then reduce answer support when appropriate. A blank-page struggle is not a badge of good learning. The relevant check is what improves later, while preserving access and safe practice.
  source ids
  retrieval-2025
  question-format-2024
- label
  Give mixed practice a clear comparison to make
  text
  The saying that mixing everything improves learning overlooks what was mixed and what learners needed to distinguish. Brunmair and Richter's review found variation by material, including results favoring blocked word practice and uncertain findings for text. Choose mixing when comparing related categories or selecting procedures is part of the goal. Keep unrelated interruptions from being mistaken for this learning method. Spacing concerns when you return; interleaving concerns which categories you alternate. You can use either or both, but a varied calendar alone does not show that you learned to make the needed distinction.
  source ids
  interleaving-2019
  aero-spacing
id
check-a-learning-claim
role
cautions
title
Check what a learning claim actually promises
intro
These judgments distinguish supported claims from broader promises. A verdict applies to the claim as stated; read its explanation before turning it into a practice rule.
items
- label
  Use recall as one supported way to learn
  text
  Claim: recalling learned information can improve later learning. Verdict: established. Experiments and the classroom review support retrieval practice across the outcomes examined. That does not promise perfect memory, and successful recall is not the same as every kind of real-world skill. Choose a useful question, attempt it and check its meaning against a dependable explanation.
  source ids
  retrieval-2011
  classroom-retrieval-2021
- label
  Check retention beyond the feeling of ease
  text
  Claim: if reading feels easy, I will remember it later. Verdict: not. Immediate fluency does not establish later recall or use. Try the relevant explanation or task after a gap, with the answer removed where appropriate. Reading can still be useful and enjoyable without supporting that stronger prediction.
  source ids
  performance-2015
- label
  Treat a spacing schedule as a choice to evaluate
  text
  Claim: one spacing schedule is optimal for everyone. Verdict: not. Cepeda's results varied with the desired retention interval, and laboratory timing does not supply a personal calendar law. A workable reminder can help you return without being mathematically optimal. Check what remains and adjust for the material and intended use.
  source ids
  spacing-2008
  schedule-2007
- label
  Select question formats by the thinking and feedback
  text
  Claim: short-answer practice always teaches more than multiple choice. Verdict: not. Format comparisons depend on questions, feedback and measured outcomes; the unconditional promise has no dependable universal winner. The small van Wijk study did not detect a reliable advantage; that does not prove the formats are equivalent. Choose meaningful questions with correction rather than making visible difficulty the deciding criterion.
  source ids
  question-format-2007
  question-format-2024
- label
  Keep useful formats open to you
  text
  Claim: teaching must match my fixed learning style. Verdict: talk. The preferred-style matching prescription lacks adequate evidence in Pashler and colleagues' review. Preferences, accessible formats and subject-appropriate representations still matter. Use what communicates the idea clearly and allows the relevant practice, without letting a fixed category exclude useful alternatives.
  source ids
  learning-styles-2008
- label
  Mix material for the distinction you need to learn
  text
  Claim: mixing every subject always improves learning. Verdict: not. Interleaving results vary by material and task. Comparing similar categories may help a specific choice, while switching among unrelated activities is a different intervention. Name the comparison and assess the later use instead of treating variety as proof.
  source ids
  interleaving-2019
- label
  Check your own use after AI assistance
  text
  Claim: if a chatbot solves it during practice, I will know it later. Verdict: not. The high-school mathematics trial separated improved assisted scores from unaided results, with different outcomes for Base and guarded Tutor. It does not establish every adult AI use. Attempt a relevant task without the tool's answer and use reliable feedback to identify what you learned.
  source ids
  ai-math-2025
id
one-useful-next-step
role
next_steps
title
Try one idea you will want to use again
intro
Choose one safe learning task and a dependable explanation. These flexible practice suggestions can be used in an accessible format; they are not a course or a required weekly schedule.
items
- label
  Explain one useful idea and repair one gap
  text
  Choose an idea you want to use. Set its explanation aside, state it in your own words, then reopen the source and repair one missing or mistaken connection. Karpicke and Blunt support including retrieval in learning; this short routine is an editorial adaptation of that principle. In the reading-list example, the useful correction explains which operation answers which part of the request. Sounding confident is not the check.
  source ids
  retrieval-2011
  aero-spacing
- label
  Try another example and make the link explicit
  text
  Try a comparable low-stakes request that changes one meaningful condition. Explain which principle fits, why it fits and what result would reveal a wrong choice. Check the link and answer against a dependable example or instructor. Butler's transfer study helps distinguish applying a supplied concept from independently recognizing its relevance. If that link is unclear, ask for teaching rather than treating the difficulty as a verdict on intelligence.
  source ids
  transfer-2010
  ies-learning
- label
  Leave one question for a later attempt
  text
  Save that question with its correction source and choose a workable later opportunity. AERO recommends spaced retrieval in education; the reminder is an everyday adaptation, not an optimal calendar. Keep needed accessibility tools, remove answer cues where appropriate, and use the attempt to choose what needs further explanation. If AI helped earlier, try without its answer. Simplify any arrangement that costs more effort than the useful learning it supports.
  source ids
  aero-spacing
  spacing-2008
  ai-math-2025
id
appropriate-learning-support
role
cautions
title
Choose the support appropriate to the task
items
- label
  Keep practice safe and memory concerns in appropriate care
  text
  Learning research does not provide a memory assessment, treatment or a general brain-enhancement promise. Much of this evidence comes from students and constrained tasks; adult medical knowledge results do not automatically extend to every age, memory domain or practical outcome. NIH MedlinePlus distinguishes ordinary forgetfulness from concerning difficulty with familiar activities and advises seeing a health care provider if you are worried. Seek that assessment for noticeable changes, rather than substituting flashcards. Safety-critical skills require appropriate instruction, supervision and demonstrated competence; practising a recalled explanation alone cannot supply those safeguards.
  source ids
  memory-medlineplus
  resident-retention-2009
  errors-2017
- label
  Use this as a reading companion
  text
  Named sources. Honest paraphrase of the finding. Not medical, legal, or financial advice. This is a reading companion for understanding learning methods and their boundaries. It does not prescribe an individualized training schedule or replace a teacher, supervisor or clinician where their support is needed. Choose accessible material and a safe way to check an attempt, and keep uncertainty visible when the source does not settle your question.
  source ids
  ies-learning
  memory-medlineplus

related

attention
habits
career

sources

id
retrieval-2011
title
Retrieval Practice Produces More Learning than Elaborative Studying with Concept Mapping
url
https://learninglab.psych.purdue.edu/downloads/2011/2011_Karpicke_Blunt_Science.pdf
kind
study
accessed at
2026-10-01
supports
Karpicke and Blunt’s undergraduate science-text experiments involved 80 and 120 learners, recall/restudy and delayed conceptual tests, including concept maps constructed from memory.
limitations
Specific study activities and one-week outcomes, not reading versus no reading, universal skill transfer or clinical memory improvement.
id
retrieval-2025
title
Retrieval-based learning
url
https://learninglab.psych.purdue.edu/downloads/2025/2025_Karpicke_Retrieval_Based_Learning_Review.pdf
kind
review
accessed at
2026-10-01
supports
Reviews direct and indirect retrieval benefits, correction-guided restudy and the balance between retrieval success and effort.
limitations
Narrative review, not a new trial of the proposed adult routine. No claim that one theoretical mechanism is settled or all difficult attempts help.
id
classroom-retrieval-2021
title
Retrieval Practice Consistently Benefits Student Learning: a Systematic Review of Applied Research in Schools and Classrooms
url
https://link.springer.com/article/10.1007/s10648-021-09595-9
kind
review
accessed at
2026-10-01
supports
Agarwal, Nunes and Blunt coded 50 classroom experiments with 5,374 learners, reporting benefits across settings and limited geographic coverage.
limitations
Publisher abstract, not the subscription full text. The synthesis does not establish individual subgroup estimates or benefits in every adult occupation.
id
spacing-2008
title
Spacing Effects in Learning: A Temporal Ridgeline of Optimal Retention
url
https://digitalcommons.usf.edu/psy_facpub/1766/
kind
study
accessed at
2026-10-01
supports
Cepeda and colleagues’ fact-learning experiment found that useful study gaps depended on the final retention interval and that benefits did not grow without limit.
limitations
Abstract-level verification. No personal schedule, exact ratio, age-specific guidance or app algorithm validated.
id
errors-2017
title
Learning from Errors
url
https://files.eric.ed.gov/fulltext/ED574569.pdf
kind
review
accessed at
2026-10-01
supports
Reviews attended correction, explanations of why an answer is right, feedback timing across settings and exceptions including amnesia.
limitations
Educational errorful practice is not a clinical rehabilitation protocol or justification for errors that could harm someone. General review, not validation of the sort/filter example.
id
ies-learning
title
Organizing Instruction and Study to Improve Student Learning
url
https://ies.ed.gov/ncee/wwc/PracticeGuide/1
kind
guideline
accessed at
2026-10-01
supports
Recommends alternating examples with attempts, connecting concrete and abstract ideas, explanatory questions and delayed monitoring. Its historical panel includes Metcalfe, McDaniel and Bain.
limitations
School/postsecondary guidance; everyday adult examples are adaptations. Monitoring recommendations rated Minimal, example/concrete-abstract recommendations Moderate, explanatory questions Strong in this guide; not uniformly high evidence.
id
performance-2015
title
Learning Versus Performance: An Integrative Review
url
https://bjorklab.psych.ucla.edu/wp-content/uploads/sites/13/2016/11/soderstorm_ra_learningvsperformance.pdf
kind
review
accessed at
2026-10-01
supports
Soderstrom and Robert A. Bjork distinguish temporary performance from retention and transfer, including differences across practice arrangements.
limitations
Review spans verbal/motor and foundational animal work; no animal-to-human prescription. Neither ease nor difficulty alone establishes lasting learning.
id
interleaving-2019
title
Similarity matters: A meta-analysis of interleaved learning and its moderators
url
https://www.uni-wuerzburg.de/fileadmin/06020400/2019/Brunmair_Richter_in_press__2019_META-ANALYSIS_OF_INTERLEAVED_LEARNING.pdf
kind
review
accessed at
2026-10-01
supports
Brunmair and Richter review 59 studies, 238 effects and 158 samples: results depend on material, with a blocked-word advantage and uncertain text outcomes.
limitations
Accepted manuscript corroborated by the indexed published abstract, rather than the final copy of record. No mixing-everything rule or universally best beginner sequence follows.
id
schedule-2007
title
Expanding Retrieval Practice Promotes Short-Term Retention, but Equally Spaced Retrieval Enhances Long-Term Retention
url
https://learninglab.psych.purdue.edu/downloads/2007/2007_Karpicke_Roediger_JEPLMC.pdf
kind
study
accessed at
2026-10-01
supports
Expanding vocabulary retrieval favored ten-minute results and equal spacing favored two-day results; experiment 3 varied initial recall timing.
limitations
Undergraduate vocabulary experiments using intervening trials; not prescribed days or proof that equal spacing is always superior.
id
feedback-timing-2024
title
Timing’s not everything: Immediate and delayed feedback are equally beneficial for performance in formative multiple-choice testing
url
https://asmepublications.onlinelibrary.wiley.com/doi/10.1111/medu.15287
kind
study
accessed at
2026-10-01
supports
Among 41 medical students, explanations followed each item or a nine-item block taking a median of 13.11 minutes; no transfer timing difference was demonstrated.
limitations
Nonsignificance is not equivalence. Short block delay is not weeks without correction; no claim that every feedback system works equally well.
id
question-format-2007
title
Test format and corrective feedback modify the effect of testing on long-term retention
url
https://profiles.wustl.edu/en/publications/test-format-and-corrective-feedback-modify-the-effect-of-testing-/
kind
study
accessed at
2026-10-01
supports
Multiple-choice practice outperformed short answers without feedback; in the second experiment, feedback favored short answers on three-day final tests.
limitations
Abstract only; not every question, learner or time horizon. Differences from 2024 study cannot be attributed solely to feedback.
id
question-format-2024
title
The battle of question formats: a comparative study of retrieval practice using very short answer questions and multiple choice questions
url
https://link.springer.com/article/10.1186/s12909-024-06538-0
kind
study
accessed at
2026-10-01
supports
Among 45 health-science students receiving no supplied practice feedback, higher immediate short-answer effort did not demonstrate later superiority after three to six weeks.
limitations
Small health-science student sample, low-stakes extracurricular course, no no-practice control. No proof of equivalence or blanket argument against short answers.
id
transfer-2010
title
Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying
url
https://www.researchgate.net/publication/46094320_Repeated_Testing_Produces_Superior_Transfer_of_Learning_Relative_to_Repeated_Studying
kind
study
accessed at
2026-10-01
supports
Repeated testing with corrective feedback improved later inference performance, including new-domain use when the relevant concept was supplied as a hint.
limitations
Based on Butler’s author-uploaded primary paper. The farther-transfer hint removed spontaneous relevance detection; retrieval and feedback contributions were not isolated. Undergraduate tasks do not establish automatic transfer everywhere.
id
resident-retention-2009
title
Repeated testing improves long-term retention relative to repeated study: A randomised controlled trial
url
https://profiles.wustl.edu/en/publications/repeated-testing-improves-long-term-retention-relative-to-repeate/
kind
study
accessed at
2026-10-01
supports
Randomized counterbalanced clinical-topic practice with feedback after tests; 40 residents completed the study, with a knowledge-retention advantage after approximately six months.
limitations
Primary abstract only. The outcome was retained knowledge, not patient benefit, certification or general memory health.
id
learning-styles-2008
title
Learning Styles: Concepts and Evidence
url
https://www.psychologicalscience.org/journals/pspi/j.1539-6053.2009.01038.x/
kind
review
accessed at
2026-10-01
supports
Journal summary reports inadequate evidence for preferred-style matching as an instructional prescription.
limitations
Summary, not full paper. Does not reject accessibility, sensory needs, preferences or every possible untested version of learning styles.
id
aero-spacing
title
Spacing and retrieval practice guide
url
https://www.edresearch.edu.au/guides-resources/practice-guides/spacing-and-retrieval-practice-guide-full-publication
kind
guideline
accessed at
2026-10-01
supports
Official guide describes questions, correction, routines, repeated opportunities and reasonable adjustments for disability access; last updated September 8, 2026.
limitations
Primary/secondary school-oriented professional translation; adult reminders and exercises are editorial adaptations. Not all examples apply to every context.
id
toolbox-2013
title
Strengthening the Student Toolbox: Study Strategies to Boost Learning
url
https://www.aft.org/ae/fall2013/dunlosky
kind
professional
accessed at
2026-10-01
supports
Dunlosky’s public article explains ten evaluated study methods, practice-testing and distributed-practice ratings, and the purposes of memory and comprehension.
limitations
Public translation of the strategy review, not a new independent trial or a fresh review of that review’s full text.
id
miller-memorize
title
Helping Students Memorize: Tips from Cognitive Science
url
https://www.facultyfocus.com/articles/teaching-and-learning/helping-students-memorize-tips-from-cognitive-science/
kind
professional
accessed at
2026-10-01
supports
Miller explains the purpose and context of facts and active flashcard answering. Published online in 2019 from a 2014 teaching article.
limitations
Teaching guidance, not app or AI validation. Its generic mixing language is narrowed using the interleaving review. Sort/filter illustrations are independently authored.
id
gurung-study
title
How Should Students Study? Tips, Advice, and Pitfalls
url
https://www.psychologicalscience.org/observer/how-should-students-study-tips-advice-and-pitfalls
kind
professional
accessed at
2026-10-01
supports
Gurung and McCann describe generating examples and verbalizing steps as part of effective study.
limitations
Teaching synthesis, not experimental validation of every checklist suggestion.
id
bain-interview
title
Patrice Bain: Teaching with 'Power Tools'
url
https://www.stlouisfed.org/education/teach-economics-podcast-series/teaching-with-power-tools-patrice-bain
kind
professional
accessed at
2026-10-01
supports
Bain’s public interview illustrates replacing a teacher’s recap with learner-generated recall and low-stakes classroom practice.
limitations
Historical practice account, not independent verification of classroom anecdotes or universal schedules.
id
sumeracki-examples
title
Episode 15 - How Students Can Use Interleaving, Elaboration, Dual Coding, and Concrete Examples
url
https://www.learningscientists.org/learning-scientists-podcast/2018/3/21/episode-15-how-students-can-use-interleaving-elaboration-dual-coding-and-concrete-examples
kind
professional
accessed at
2026-10-01
supports
Sumeracki explains how to connect concrete examples through their underlying principle.
limitations
Professional translation, not a new trial of these adult illustrations. No universal endorsement of every analogy or representation.
id
ai-math-2025
title
Generative AI without guardrails can harm learning: Evidence from high school mathematics
url
https://www.pnas.org/doi/10.1073/pnas.2422633122
kind
study
accessed at
2026-10-01
supports
Classroom-randomized study of nearly 1,000 Turkish high-school students compares GPT Base, GPT Tutor and controls: assisted gains differed from unaided exam outcomes.
limitations
One school, high-school mathematics and specific GPT-4 tools. Tutor used teacher inputs, not just a prompt slogan. No universal adult tool prescription or claim Tutor improved unaided exams.
id
memory-medlineplus
title
Memory
url
https://medlineplus.gov/memory.html
kind
official
accessed at
2026-10-01
supports
NIH/NLM summary attributed to NIA advises discussing worrying forgetfulness and difficulty with familiar activities with a health care provider.
limitations
Health-literacy guidance, not diagnosis or treatment.
id
professional-jeffrey-karpicke
title
Jeffrey Karpicke — official role record
url
https://learninglab.psych.purdue.edu/people/
kind
official
accessed at
2026-10-01
supports
Purdue Cognition and Learning Lab principal investigator; learning and memory researcher. Defines retrieval as a learning event and distinguishes direct benefits from correction-guided restudy. University-hosted research papers and the 2025 retrieval chapter explain learning for educators and readers.
limitations
Knowledge experiments and synthesis do not validate every adult skill or clinical memory program.
id
professional-pooja-agarwal
title
Pooja Agarwal — official role record
url
https://college.berklee.edu/faculty/pooja-k-agarwal
kind
official
accessed at
2026-10-01
supports
PhD in cognitive psychology; Berklee associate professor. Coauthored the 2021 classroom review; makes low-stakes retrieval intelligible to teachers. Official profile documents RetrievalPractice.org, public teaching, workshops, books and media explanations.
limitations
Primary research supports the learning claims; books, reach and consulting are not independent validation.
id
professional-janet-metcalfe
title
Janet Metcalfe — official role record
url
https://psychology.columbia.edu/content/janet-metcalfe
kind
official
accessed at
2026-10-01
supports
Columbia professor of psychology; PhD, University of Toronto. Explains attended correction, reasoning about errors and monitoring what is known. Public explanatory error-learning review and membership of the official IES practice-guide panel.
limitations
Educational feedback and metacognition are not individualized clinical rehabilitation guidance.
id
professional-mark-mcdaniel
title
Mark McDaniel — official role record
url
https://psych.washu.edu/people/mark-mcdaniel
kind
official
accessed at
2026-10-01
supports
WashU professor emeritus of psychological and brain sciences; research scientist and CIRCLE director. Connects learning strategies with instructional choices; differentiates memory and use. Official practice-guide panel work and university teaching research translate memory science to instruction.
limitations
Prospective memory, content recall and transfer have different outcomes; an institutional role does not settle every technique.
id
professional-doug-rohrer
title
Doug Rohrer — official role record
url
https://www.usf.edu/arts-sciences/departments/psychology/people/faculty/doug-rohrer.aspx
kind
official
accessed at
2026-10-01
supports
University of South Florida professor of psychology; PhD in psychology. Coauthored Cepeda's timing study; makes spacing and category-selection purposes distinct. Official profile links publicly available pedagogical research and explanatory materials.
limitations
No universal spacing timetable or endorsement of indiscriminate task mixing.
id
professional-elizabeth-ligon-bjork
title
Elizabeth Ligon Bjork — official role record
url
https://bjorklab.psych.ucla.edu/people/
kind
official
accessed at
2026-10-01
supports
UCLA psychologist with PhD in psychology; official lab documents former professor and distinguished research professor roles. Helps distinguish desirable challenge from a blanket celebration of difficulty. Official lab describes teaching awards and translation of memory research into instructional practices.
limitations
The lab verifies historical roles; no current administrative appointment inferred. Robert A. Bjork, not Elizabeth, coauthored the cited Soderstrom review.
id
professional-john-dunlosky
title
John Dunlosky — official role record
url
https://www.kent.edu/psychology/profile/john-dunlosky
kind
official
accessed at
2026-10-01
supports
Kent State professor emeritus of psychological sciences. Evaluates actual study activities by their learning purpose instead of popularity. Public American Educator strategy guide and Kent teaching-institute outreach.
limitations
Strategy ratings are not individualized prescriptions or a universal number of repetitions.
id
professional-katherine-rawson
title
Katherine Rawson — official role record
url
https://www.kent.edu/psychology/comprehension-memory-lab
kind
official
accessed at
2026-10-01
supports
Kent State professor of psychological sciences; Comprehension and Memory Lab researcher. Coauthored the learning-technique evaluation; keeps comprehension, practice and retention separate. Kent’s Building a Better Student account describes her public teaching-institute discussion.
limitations
The source does not establish an exact number of successive-relearning attempts.
id
professional-michelle-miller
title
Michelle Miller — official role record
url
https://experts.nau.edu/en/persons/michelle-miller-2/
kind
official
accessed at
2026-10-01
supports
Northern Arizona University professor of psychological sciences. Connects facts to purpose and distinguishes active flashcard answering from passive reviewing. Public authored memorization guidance addresses teachers' questions about facts and higher-order thinking.
limitations
Practical teaching article is not a trial validating apps, every mixing procedure or AI tutors.
id
professional-regan-a-r-gurung
title
Regan A. R. Gurung — official role record
url
https://ctl.oregonstate.edu/directory/regan-ar-gurung
kind
official
accessed at
2026-10-01
supports
Oregon State professor of psychological science; associate vice provost and executive director of the Center for Teaching and Learning; PhD. Makes reasoning visible by generating examples and verbalizing solution steps. Teaching workshops and an APS coauthored guide explain useful student practice.
limitations
Teaching synthesis includes nonexperimental evidence; not all-context causal proof for its recommendations.
id
professional-patrice-bain
title
Patrice Bain — official role record
url
https://ies.ed.gov/ncee/wwc/PracticeGuide/1
kind
official
accessed at
2026-10-01
supports
Classroom educator and author; named teacher member of the 2007 IES practice-guide panel. Shows how a teacher's recap can become a learner's remembered answer. Her St. Louis Fed educator interview gives examples of low-stakes retrieval.
limitations
Historical panel role, not an assumed current school appointment. Interview anecdotes and commercial books are not independent outcome evidence.
id
professional-megan-sumeracki
title
Megan Sumeracki — official role record
url
https://www.ric.edu/people-directory/megan-sumeracki
kind
official
accessed at
2026-10-01
supports
Rhode Island College professor of psychology; PhD in cognitive psychology. Links different examples through a shared principle rather than memorable surface features. The Learning Scientists podcast explains the use of examples and representations.
limitations
Public teaching translation is not direct validation of every analogy or individual implementation.
id
professional-andrew-butler
title
Andrew Butler — official role record
url
https://education.washu.edu/people/andrew-c-butler
kind
official
accessed at
2026-10-01
supports
WashU professor of education and psychological and brain sciences by courtesy; PhD. Tests learning transfer while exposing the feedback and hints used in specific designs. Official profile describes teaching, translational education work and public research explanations.
limitations
Cued new-domain use is not spontaneous relevance recognition or automatic real-world transfer.
id
professional-douglas-larsen
title
Douglas Larsen — official role record
url
https://profiles.wustl.edu/en/persons/douglas-larsen/
kind
official
accessed at
2026-10-01
supports
WashU professor of neurology and pediatrics; vice chair for educational affairs. Extends retention evidence to residents and distinguishes knowledge outcomes from clinical performance. Public institutional publications and medical-education research provide an adult professional learning example.
limitations
Retained answers do not establish patient benefit, safe independent practice or professional certification.
id
professional-denise-park
title
Denise Park — official role record
url
https://profiles.utdallas.edu/denise
kind
official
accessed at
2026-10-01
supports
Late UT Dallas cognitive-aging researcher; official memorial lists distinguished university chair, professor and Center for Vital Longevity research director. Keeps age and the particular memory domain visible when considering how evidence generalizes. Official research profile presents lifespan cognitive research and linked public center resources.
limitations
Memorial record, not a current practitioner. No dementia-prevention or memory-treatment promise inferred.
id
rawson-public-teaching
title
Building a Better Student
url
https://www.kent.edu/provost/news/building-better-student
kind
official
accessed at
2026-10-01
supports
Kent’s November 17, 2014 account describes Rawson’s spring teaching-institute discussion and Dunlosky’s public teaching work.
limitations
Historical outreach record, not current appointment verification or workshop impact evaluation.
id
unesco-genai
title
Guidance for generative AI in education and research
url
https://www.unesco.org/en/articles/guidance-generative-ai-education-and-research
kind
guideline
accessed at
2026-10-01
supports
Publication summary, updated January 2026, addresses human agency, pedagogical validation and privacy protection in educational AI.
limitations
Publication summary read, not full report. Guidance is not a causal learning trial, workplace-specific policy or a claim about current law.

professional panel

name
Jeffrey Karpicke
credential
Purdue Cognition and Learning Lab principal investigator; learning and memory researcher
official url
https://learninglab.psych.purdue.edu/people/
public usefulness
University-hosted research papers and the 2025 retrieval chapter explain learning for educators and readers.
contribution
Defines retrieval as a learning event and distinguishes direct benefits from correction-guided restudy.
boundary
Knowledge experiments and synthesis do not validate every adult skill or clinical memory program.
reviewed at
2026-10-01
name
Pooja Agarwal
credential
PhD in cognitive psychology; Berklee associate professor
official url
https://college.berklee.edu/faculty/pooja-k-agarwal
public usefulness
Official profile documents RetrievalPractice.org, public teaching, workshops, books and media explanations.
contribution
Coauthored the 2021 classroom review; makes low-stakes retrieval intelligible to teachers.
boundary
Primary research supports the learning claims; books, reach and consulting are not independent validation.
reviewed at
2026-10-01
name
Janet Metcalfe
credential
Columbia professor of psychology; PhD, University of Toronto
official url
https://psychology.columbia.edu/content/janet-metcalfe
public usefulness
Public explanatory error-learning review and membership of the official IES practice-guide panel.
contribution
Explains attended correction, reasoning about errors and monitoring what is known.
boundary
Educational feedback and metacognition are not individualized clinical rehabilitation guidance.
reviewed at
2026-10-01
name
Mark McDaniel
credential
WashU professor emeritus of psychological and brain sciences; research scientist and CIRCLE director
official url
https://psych.washu.edu/people/mark-mcdaniel
public usefulness
Official practice-guide panel work and university teaching research translate memory science to instruction.
contribution
Connects learning strategies with instructional choices; differentiates memory and use.
boundary
Prospective memory, content recall and transfer have different outcomes; an institutional role does not settle every technique.
reviewed at
2026-10-01
name
Doug Rohrer
credential
University of South Florida professor of psychology; PhD in psychology
official url
https://www.usf.edu/arts-sciences/departments/psychology/people/faculty/doug-rohrer.aspx
public usefulness
Official profile links publicly available pedagogical research and explanatory materials.
contribution
Coauthored Cepeda's timing study; makes spacing and category-selection purposes distinct.
boundary
No universal spacing timetable or endorsement of indiscriminate task mixing.
reviewed at
2026-10-01
name
Elizabeth Ligon Bjork
credential
UCLA psychologist with PhD in psychology; official lab documents former professor and distinguished research professor roles
official url
https://bjorklab.psych.ucla.edu/people/
public usefulness
Official lab describes teaching awards and translation of memory research into instructional practices.
contribution
Helps distinguish desirable challenge from a blanket celebration of difficulty.
boundary
The lab verifies historical roles; no current administrative appointment inferred. Robert A. Bjork, not Elizabeth, coauthored the cited Soderstrom review.
reviewed at
2026-10-01
name
John Dunlosky
credential
Kent State professor emeritus of psychological sciences
official url
https://www.kent.edu/psychology/profile/john-dunlosky
public usefulness
Public American Educator strategy guide and Kent teaching-institute outreach.
contribution
Evaluates actual study activities by their learning purpose instead of popularity.
boundary
Strategy ratings are not individualized prescriptions or a universal number of repetitions.
reviewed at
2026-10-01
name
Katherine Rawson
credential
Kent State professor of psychological sciences; Comprehension and Memory Lab researcher
official url
https://www.kent.edu/psychology/comprehension-memory-lab
public usefulness
Kent's opened Building a Better Student account documents her public teaching-institute discussion.
contribution
Coauthored the learning-technique evaluation; keeps comprehension, practice and retention separate.
boundary
The source does not establish an exact number of successive-relearning attempts.
reviewed at
2026-10-01
name
Michelle Miller
credential
Northern Arizona University professor of psychological sciences
official url
https://experts.nau.edu/en/persons/michelle-miller-2/
public usefulness
Public authored memorization guidance addresses teachers' questions about facts and higher-order thinking.
contribution
Connects facts to purpose and distinguishes active flashcard answering from passive reviewing.
boundary
Practical teaching article is not a trial validating apps, every mixing procedure or AI tutors.
reviewed at
2026-10-01
name
Regan A. R. Gurung
credential
Oregon State professor of psychological science; associate vice provost and executive director of the Center for Teaching and Learning; PhD
official url
https://ctl.oregonstate.edu/directory/regan-ar-gurung
public usefulness
Official biography documents teaching workshops; opened APS coauthored guide explains student practice.
contribution
Makes reasoning visible by generating examples and verbalizing solution steps.
boundary
Teaching synthesis includes nonexperimental evidence; not all-context causal proof for its recommendations.
reviewed at
2026-10-01
name
Patrice Bain
credential
Classroom educator and author; named teacher member of the 2007 IES practice-guide panel
official url
https://ies.ed.gov/ncee/wwc/PracticeGuide/1
public usefulness
Opened St. Louis Fed educator interview offers simple low-stakes retrieval illustrations.
contribution
Shows how a teacher's recap can become a learner's remembered answer.
boundary
Historical panel role, not an assumed current school appointment. Interview anecdotes and commercial books are not independent outcome evidence.
reviewed at
2026-10-01
name
Megan Sumeracki
credential
Rhode Island College professor of psychology; PhD in cognitive psychology
official url
https://www.ric.edu/people-directory/megan-sumeracki
public usefulness
Opened Learning Scientists podcast gives public practical explanations of examples and representations.
contribution
Links different examples through a shared principle rather than memorable surface features.
boundary
Public teaching translation is not direct validation of every analogy or individual implementation.
reviewed at
2026-10-01
name
Andrew Butler
credential
WashU professor of education and psychological and brain sciences by courtesy; PhD
official url
https://education.washu.edu/people/andrew-c-butler
public usefulness
Official profile describes teaching, translational education work and public research explanations.
contribution
Tests learning transfer while exposing the feedback and hints used in specific designs.
boundary
Cued new-domain use is not spontaneous relevance recognition or automatic real-world transfer.
reviewed at
2026-10-01
name
Douglas Larsen
credential
WashU professor of neurology and pediatrics; vice chair for educational affairs
official url
https://profiles.wustl.edu/en/persons/douglas-larsen/
public usefulness
Public institutional publications and medical-education research provide an adult professional learning example.
contribution
Extends retention evidence to residents and distinguishes knowledge outcomes from clinical performance.
boundary
Retained answers do not establish patient benefit, safe independent practice or professional certification.
reviewed at
2026-10-01
name
Denise Park
credential
Late UT Dallas cognitive-aging researcher; official memorial lists distinguished university chair, professor and Center for Vital Longevity research director
official url
https://profiles.utdallas.edu/denise
public usefulness
Official research profile presents lifespan cognitive research and linked public center resources.
contribution
Keeps age and the particular memory domain visible when considering how evidence generalizes.
boundary
Memorial record, not a current practitioner. No dementia-prevention or memory-treatment promise inferred.
reviewed at
2026-10-01

Download original public profile Download complete article

End of Learning and remembering