How Learning Actually Works — The Science of Getting Better at Anything

Decades of cognitive science research have transformed what we know about learning — but most education hasn't caught up. The most popular study methods (re-reading, highlighting) are among the least effective. The most powerful methods (retrieval practice, spaced repetition, interleaving) are widely understood but rarely used. Here's what actually works, based on research that has been replicated for decades, with the key studies named so you can check them yourself.

Educational content, not medical advice. Talk to your doctor before changing any medication, treatment, or routine. Information here reflects current research but every person's situation differs.

Key takeaways

• Memory is reconstructive, not a recording — every act of recall changes the memory.
• The most effective study technique is retrieval practice (testing yourself), not re-reading.
• Spacing learning over time beats massed practice (cramming), and the best gap grows with how long you need to remember.
• A 2013 review of ten common techniques rated only two — practice testing and distributed practice — as high utility.
• Sleep is essential for consolidating new memories.
• Most learners feel like they're learning during ineffective methods, and feel like they're failing during effective ones.

How memory actually forms

Learning involves three rough stages, each with distinct brain mechanisms. For the cellular detail — synapses, the hippocampus and long-term potentiation — see our explainer on how memory works in the brain; here the focus is on what those mechanisms mean for anyone trying to learn something.

Encoding

Initial perception and processing. New information becomes neural activity patterns in the cortex. What you pay attention to, and how deeply you process it, largely decides what gets encoded at all. Information you merely glance at leaves a weak trace; information you connect to things you already know leaves a richer one.

Consolidation

Over hours to days (especially during sleep), memories are stabilised through processes involving the hippocampus and gradually become more dependent on long-term cortical storage. At the cellular level this involves changes in the strength of connections between neurons — the same capacity for change described in our piece on neuroplasticity and brain rewiring.

Retrieval

Activating the stored pattern — recalling, recognising, applying. Crucially: each retrieval reshapes the memory.

The third point matters enormously: memory is reconstructive. Every time you remember something, you slightly modify it. Strong memories aren't the ones encoded once and stored intact — they're the ones retrieved many times.

Why we fool ourselves about learning

A central finding of learning science: our subjective sense of learning is unreliable.

What feels productive

• Re-reading a textbook chapter.
• Highlighting passages.
• Watching a video.
• Following along with examples.

What's actually productive

• Closing the book and trying to recall what you read.
• Generating examples from scratch.
• Testing yourself.
• Explaining concepts to someone else (or aloud).

The more effortful the recall, the better it sticks. The easier it feels in the moment, the less it transfers to long-term memory.

This mismatch — feeling unproductive while learning effectively, feeling productive while learning ineffectively — is why so many students develop poor study habits. Psychologists Robert and Elizabeth Bjork call the effective-but-uncomfortable conditions "desirable difficulties": spacing, interleaving and testing all slow down apparent progress during practice while improving what remains weeks later.

Part of the illusion is fluency. When you re-read a page, the words feel familiar, and the brain treats that familiarity as a sign of knowledge. But recognising something when it is in front of you is a different act from producing it when it isn't — and exams, conversations and jobs almost always demand the second.

The landmark studies

Most of what follows rests on a handful of well-known papers. Knowing what they actually found — rather than what popular summaries say they found — helps you apply them sensibly.

Roediger & Karpicke (2006): the testing effect

In Psychological Science, Henry Roediger and Jeffrey Karpicke of Washington University in St. Louis had 120 undergraduates read short prose passages ("The Sun" and "Sea Otters"). After the first reading, students either restudied a passage or took a free-recall test on it without feedback. They then took a final test 5 minutes, 2 days or 1 week later.

The results reversed with time:

Final test delay	Restudied	Tested	Winner
5 minutes	81%	75%	Restudy
2 days	54%	68%	Testing
1 week	42%	56%	Testing

Source: Roediger & Karpicke (2006), Experiment 1, proportion of idea units recalled.

The tested group remembered about as much after a week as the restudy group did after two days. A second experiment with 180 students compared reading a passage four times (SSSS), three times plus one test (SSST), and once plus three tests (STTT). After a week, the STTT group recalled 61% versus 40% for SSSS — even though they had read the passage far fewer times. Students who restudied also predicted they would remember more, which is the fluency illusion in a single dataset.

Cepeda et al. (2006): how much spacing?

Nicholas Cepeda, Harold Pashler, Edward Vul, John Wixted and Doug Rohrer published a meta-analysis in Psychological Bulletin covering 839 assessments from 317 experiments in 184 articles on verbal recall. Spaced presentations beat massed presentations reliably. Their most practical finding was an interaction: the gap between study sessions that produces the best retention grows as the retention interval grows. Material you need for next week benefits from short gaps; material you need next year benefits from longer ones.

Dunlosky et al. (2013): a report card for ten techniques

John Dunlosky, Katherine Rawson, Elizabeth Marsh, Mitchell Nathan and Daniel Willingham reviewed the evidence for ten study techniques in Psychological Science in the Public Interest (January 2013). They rated each on how well it generalises across learners, materials and test types.

Technique	Utility rating
Practice testing	High
Distributed practice	High
Elaborative interrogation	Moderate
Self-explanation	Moderate
Interleaved practice	Moderate
Summarisation	Low
Highlighting / underlining	Low
Keyword mnemonic	Low
Imagery for text	Low
Re-reading	Low

Source: Dunlosky et al. (2013); the APS press release confirms the high and low ratings for testing, spacing, summarisation, highlighting and re-reading.

Two caveats from the authors are often dropped. "Low utility" does not mean useless — highlighting can help as a first pass if it leads to something more active. And the techniques only help learners who actually use them consistently.

Rohrer & Taylor (2007): interleaving in maths

Doug Rohrer and Kelli Taylor taught college students to calculate the volumes of unfamiliar geometric solids. One group practised problems blocked by shape; the other practised them shuffled. During practice the blocked group did better. On a test a week later, the interleaved group scored 63% versus 20% for the blocked group (Experiment 2 of the paper, which had only 18 participants, so treat the exact figures as illustrative). The lesson: blocked practice tells you which formula to use; interleaved practice forces you to work it out, which is the skill a real test demands.

What actually works (and why)

Retrieval practice

Testing yourself, free-recalling material, working problems from scratch. Practice testing earned the highest rating in Dunlosky's review, and the testing effect has been replicated across ages, subjects and test formats.

Why: each retrieval strengthens access to the memory and reveals what you actually know versus what you think you know. Low-stakes quizzes count — the benefit comes from the act of retrieval, not from the grade.

Spaced repetition

Distribute study over multiple days or weeks rather than cramming.

Why: some forgetting between sessions makes the next retrieval harder, and effortful retrieval strengthens memory more than easy retrieval.

Practical tools: Anki, Quizlet, hand-built flashcard systems. The spacing effect is one of the most robust findings in cognitive psychology, documented since Hermann Ebbinghaus's experiments in the 1880s.

Interleaving

Mix different problem types in a study session rather than blocking ("all multiplication, then all division").

Why: forces your brain to identify which strategy applies, building stronger discrimination.

Counterintuitive: blocked practice feels easier and produces better immediate results. Interleaved practice feels harder and, in studies like Rohrer & Taylor's, produces much better performance on delayed tests. Interleaving works best when the mixed items are related enough to be confused — mixing algebra with French vocabulary gains little.

Elaboration

Explain concepts, connect to existing knowledge, ask "why" and "how." Dunlosky's review rated elaborative interrogation (asking "why is this true?") and self-explanation as moderate utility.

Why: encoding with rich connections to existing knowledge creates more retrieval paths.

Concrete examples

Generate or work through specific examples of abstract concepts.

Why: moving between abstract and concrete strengthens both, and examples you generate yourself double as retrieval practice.

Dual coding

Combine words and images. Diagrams, sketches, timelines alongside text.

Why: multiple representations of the same content create more retrieval cues. Note that this is a property of the material, not of the learner — it is not the same idea as "visual learners".

Sleep

Adequate, regular sleep, particularly the night after learning.

Why: memory consolidation continues during sleep. Our article on what sleep does and why you need it covers the stages involved.

What doesn't work as well as people think

Re-reading

Common but weak beyond the first read. Familiarity feels like understanding but doesn't translate to long-term memory. Dunlosky's review rated it low utility.

Highlighting

Adds little on its own. The act of highlighting feels productive but the marked text usually isn't engaged with deeply afterwards.

"Learning styles" (visual/auditory/kinaesthetic)

This popular framework doesn't hold up. In a 2008 review in Psychological Science in the Public Interest, Harold Pashler, Mark McDaniel, Doug Rohrer and Robert Bjork set out what a valid test would need: measure learners' styles, randomly assign them to matched or mismatched instruction, and give everyone the same test. Few studies used that design, and those that did found no support for matching teaching to claimed styles. People do have preferences — but preferences are not evidence of better learning.

Cramming

Works for short-term retention (the test next morning) but produces little long-term retention. Material crammed in one sitting is largely forgotten within days, as the Roediger & Karpicke and Cepeda data both imply.

Massed practice

Doing the same thing repeatedly in one session. Easier than spaced or interleaved practice, but weaker long-term outcomes.

Passive listening

Lectures, podcasts, audiobooks alone. Information passes through without strong encoding unless you engage with it actively — pausing to summarise from memory turns passive input into retrieval practice.

How long does learning actually take?

The "10,000 hours rule" (popularised by Malcolm Gladwell from K. Anders Ericsson's 1993 research on violinists) is widely misquoted. Ericsson himself stressed that the number was an average for one elite group, not a threshold, and that the type of practice mattered more than its quantity.

• Deliberate practice (focused, effortful, with feedback, targeting weaknesses) drives expertise more than mere repetition.
• The amount required varies enormously by domain.
• Naive practice (just doing the activity) can plateau at low levels indefinitely.

Practice is not the whole story, either. A 2014 meta-analysis by Brooke Macnamara, David Hambrick and Frederick Oswald in Psychological Science pooled 88 studies and found deliberate practice explained a varying share of performance differences:

Domain	Share of performance variance explained by deliberate practice
Games (e.g. chess)	26%
Music	21%
Sports	18%
Education	4%
Professions	under 1%

Source: Macnamara, Hambrick & Oswald (2014), via the Association for Psychological Science.

The takeaway isn't that practice is pointless — it's that well-structured practice is necessary but not sufficient, and that starting knowledge, age of starting, working memory and opportunity also matter. The professions figure rests on a small number of studies, so treat it cautiously.

The forgetting curve

Hermann Ebbinghaus (1885) memorised lists of nonsense syllables and measured how much effort he saved when relearning them after delays ranging from 20 minutes to 31 days. He found that forgetting is steepest soon after learning and then levels off: most of the loss happens within the first day, and what survives a few days tends to persist.

Popular versions of the curve often quote precise "percentages forgotten", but Ebbinghaus measured savings (how much faster relearning was), not the share of facts you can recall, so those numbers don't transfer neatly to studying a textbook. A 2015 replication by Jaap Murre and Joeri Dros in PLOS ONE, using a single subject who spent 70 hours learning and relearning lists, produced a curve similar to Ebbinghaus's original — with a small upward bump at the 24-hour point that the authors linked to sleep.

What matters practically: forgetting is fast at first, so the first review should come soon (within a day or so), and later reviews can be spread further apart. That is exactly what spaced-repetition software does — reviewing items at increasing intervals (for example 1 day, 3 days, 1 week, 2 weeks) and adjusting each card based on whether you remembered it. For a deeper look at why memories fade, see why we forget things.

Practical study workflow

What an evidence-based study session looks like:

1. Review what you remember (5-10 min)

Before opening notes, recall what you learned previously. Write down everything. Compare to notes.

2. Engage with new material (20-40 min)

Read or watch content. Take notes in your own words, not verbatim. Generate questions as you go.

3. Self-test (15-20 min)

Close the material. Free-recall: write everything you remember about the topic. Test yourself with practice questions.

4. Identify gaps

Compare your recall to source. Note what you missed or got wrong.

5. Active review

Re-engage specifically with weak spots. Generate examples. Explain concepts aloud.

6. Sleep on it

Don't underestimate this. Consolidation continues overnight.

7. Return tomorrow

Spaced repetition begins. Re-test on what you covered before tackling new material.

This approach feels harder than highlighting and re-reading but produces much more durable learning.

A worked example: one exam, three weeks

Suppose you have three weeks before a biology exam covering six chapters. A typical plan reads two chapters a week, then re-reads everything the weekend before. An evidence-based plan looks different:

Week	New material	Retrieval and spacing
1	Chapters 1–2	Daily 10-minute recall of everything covered so far; flashcards for key terms
2	Chapters 3–4	Mixed practice questions from chapters 1–4; one blank-page "brain dump" per chapter
3	Chapters 5–6	Full mixed practice tests every other day; re-study only the items you missed
Night before	Nothing new	Light review, then a full night's sleep

The total hours can be identical. The difference is that most of the second plan's time is spent retrieving rather than re-reading, every chapter is revisited several times at increasing gaps, and the final week mixes topics the way the exam will.

Tools that align with the science

Anki

Free flashcard app with algorithm-driven spaced repetition. Widely used by medical students, language learners and anyone needing durable memorisation. The software is only as good as the cards: short, single-fact cards you write yourself work better than downloaded decks you never understood.

Quizlet

More user-friendly than Anki, with less control over scheduling. Useful for quick self-testing.

Obsidian / Notion and similar

Note-taking apps that encourage linking and elaboration. Useful for building connections, but writing notes is encoding, not retrieval — pair them with self-testing.

Practice problems

Textbooks with solved problems, Khan Academy and similar. Attempting problems before checking solutions beats reading solutions.

AI tutors

Chatbots can generate quiz questions, explain a concept a second way, or check an explanation you wrote. Used to quiz you, they support retrieval practice; used to summarise everything for you, they replace the effort that makes learning stick. Our guide to using AI for studying covers how to keep the effort on your side.

Domain-specific notes

Languages

• Daily practice beats once-weekly cramming.
• Spaced repetition for vocabulary.
• Comprehensible input (slightly above your level) builds passive understanding.
• Speaking practice — even imperfectly — builds active fluency.

Mathematics and STEM

• Solving problems beats reading proofs.
• Work through derivations from scratch.
• Interleave problem types, as in Rohrer & Taylor's study.
• Mistakes are signals — don't avoid hard problems.

History and humanities

• Connect events and concepts (elaboration).
• Generate arguments rather than only memorising dates.
• Write and speak about content.

Skills (music, sports, programming)

• Deliberate practice with feedback.
• Identify specific weak points and drill them.
• Recovery and rest are part of skill consolidation.
• Plateaus are normal; pushing through usually requires changing the practice, not just repeating it.

Why motivation and attention matter

Encoding depends on attention, and attention depends partly on motivation. The brain's reward circuitry — discussed in our article on dopamine and motivation — helps tag information as worth keeping, which is one reason material you care about or are curious about tends to stick better.

That cuts both ways. Phones and notifications split attention, and switching between tasks carries a measurable cost each time you return. Studying in shorter, fully focused blocks with real breaks generally beats long sessions half-spent on something else.

Common myths

"Some people are just bad at math/languages/etc."

Innate differences exist, as the Macnamara meta-analysis implies, but effective methods produce real gains for nearly anyone willing to apply them consistently.

"I have to learn it perfectly first time"

Multiple exposures and partial forgetting are normal and even useful for long-term memory — retrieval after some forgetting is what strengthens it.

"Studying with music helps focus"

Evidence is mixed. Music with lyrics is more likely to interfere with reading and verbal learning; some people find quiet instrumental music neutral. If in doubt, test yourself both ways.

"Multitasking saves time"

Task-switching studies consistently show that switching back and forth costs time and accuracy compared with doing tasks one at a time.

"Some subjects don't require effort"

The subjects that feel easy are usually the ones with prior connections to existing knowledge. When learning feels effortless, often little new is being learned.

Sleep and learning specifically

The role of sleep is too important to bury:

• New learning continues to consolidate during sleep; both slow-wave sleep and REM sleep appear to contribute, though their exact roles are still debated.
• Sleep deprivation the night after learning reduces retention.
• Sleep deprivation the night before learning impairs encoding.
• Daytime naps can help retention of recently learned material in some studies.

Pulling all-nighters before exams is close to the worst strategy from a memory perspective: it cuts encoding for anything studied late and removes the consolidation window. Our article on REM sleep and the brain explains what happens during that window.

How to teach yourself anything

1. Pick a clear goal. "Learn Python" is too vague. "Build a working web scraper" is concrete.
2. Find quality material. Books, courses and primary sources rather than random clips.
3. Plan deliberate practice. Schedule specific sessions with retrieval practice and progressive difficulty.
4. Build feedback loops. Self-tests, projects with measurable outcomes, peer review.
5. Apply quickly. Use what you're learning in real contexts. Theory without application doesn't stick.
6. Track progress. Note what you've covered and what gaps remain.
7. Stay consistent. Short daily sessions spread practice out in a way one long weekly session cannot.

Open questions in learning science

The core findings are robust, but researchers are still working on several questions:

• Optimal spacing schedules. Cepeda's work shows the best gap depends on the retention interval, but exact schedules for real courses — with many topics competing for review — are still being refined.
• Transfer. Testing reliably improves memory for the tested material; how far it improves the ability to apply ideas in new situations varies by study.
• Classroom scale. Many landmark studies used university students and short materials. Larger classroom trials generally support the findings but show smaller effects.
• Individual differences. Working memory, prior knowledge and conditions such as ADHD affect how much a learner benefits from each technique.

Frequently asked questions

Q: What is the single most effective study technique? A: Retrieval practice — testing yourself without looking at the answers. Dunlosky et al.'s 2013 review rated practice testing as high utility, and in Roediger and Karpicke's 2006 study students who took a recall test remembered 56% of a passage after a week versus 42% for those who restudied it.

Q: How far apart should I space my study sessions? A: It depends on how long you need to remember the material. Cepeda et al.'s 2006 meta-analysis found that the best gap grows as the retention interval grows, so review soon after first learning and then at gradually longer intervals.

Q: Are learning styles real? A: People have preferences, but a 2008 review by Pashler and colleagues found no adequate evidence that matching instruction to a visual, auditory or kinaesthetic "style" improves learning.

Q: Why does re-reading feel so effective if it isn't? A: Re-reading makes material feel familiar, and the brain mistakes that fluency for knowledge. In Roediger and Karpicke's experiments, students who restudied predicted better recall than students who tested themselves, yet remembered less a week later.

Q: Is the 10,000-hour rule true? A: Not as a rule. It is a popular simplification of Ericsson's research. A 2014 meta-analysis found deliberate practice explained 26% of performance differences in games, 21% in music and 18% in sports, and far less in education and professions.

Q: Does sleep really affect learning? A: Yes. Memory consolidation continues during sleep, sleep loss before learning impairs encoding, and sleep loss after learning reduces retention.

Q: How do I know if I'm actually learning? A: Test yourself. If you can recall details and apply concepts from scratch, you've learned them. If you can only recognise material when it is in front of you, you haven't yet.

The bottom line

Learning is a well-studied process. The most effective methods — retrieval practice, spaced repetition, interleaving, elaboration, sleep — are widely known but rarely used. The most popular methods — re-reading, highlighting, cramming — feel productive but produce poor long-term outcomes. The mismatch between subjective ease and objective effectiveness is the central trap. Effective learning feels harder, slower and more uncertain than ineffective learning. Accepting that discomfort is what separates durable knowledge from surface familiarity.

Sources

• Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning. Psychological Science, 17(3), 249–255. PDF
• Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T., & Rohrer, D. (2006). Distributed practice in verbal recall tasks. Psychological Bulletin, 132(3), 354–380. Record
• Dunlosky, J., et al. (2013). Improving students' learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4–58. APS press release
• Rohrer, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35, 481–498. Record
• Pashler, H., McDaniel, M., Rohrer, D., & Bjork, R. (2008). Learning styles: Concepts and evidence. Psychological Science in the Public Interest, 9(3), 105–119. Record
• Macnamara, B. N., Hambrick, D. Z., & Oswald, F. L. (2014). Deliberate practice and performance. Psychological Science. APS press release
• Murre, J. M. J., & Dros, J. (2015). Replication and analysis of Ebbinghaus' forgetting curve. PLOS ONE, 10(7), e0120644. Record

Related Reading

• How the Human Brain Actually Works — Neurons, Networks, and Consciousness
• The Science of Exercise — Why It Works and How Much You Actually Need
• The Science of Mental Health — What We Actually Know
• The Science of Sleep — Why You Need It and What Happens Without It
• How Does Memory Actually Work?