In 2011, Scott Young decided to teach himself the entire four-year MIT computer science curriculum in twelve months, using only the publicly available course materials and examinations. He passed the final exams in all 33 courses, including Circuits and Electronics, Theory of Computation, and Artificial Intelligence. He called the project the MIT Challenge. The methods he used — intensive self-testing, deliberate retrieval practice, project-based application, and compressed spaced repetition — were not intuitive. They were drawn directly from cognitive science. Young did not simply study more hours. He studied using techniques that the research on human memory shows are dramatically more effective than the default approaches most people use.

The gap between how most people study and how the brain actually learns is one of the most well-documented findings in cognitive psychology. Since Hermann Ebbinghaus mapped the forgetting curve in 1885 — showing that we lose the majority of newly learned information within days unless it is reviewed — researchers have accumulated decades of evidence about which study techniques work, which feel effective but do not, and why the methods that feel most productive are often the least productive. The research is unambiguous. Most people simply have not been told what it says.

Consider what a typical student does when preparing for an exam or a professional attempting to learn a new skill. They read and re-read the material. They highlight important passages. They watch explanatory videos. They review their notes. These activities feel productive. They produce a warm sense of recognition: "I know this." But recognition is not retrieval, and retrieval is what matters when you need to use the knowledge. The comfort of familiarity is a trap that the brain has laid for itself.

"Testing is not just a way of measuring knowledge. It is a way of creating it. The act of retrieving information from memory makes that information more retrievable in the future." -- Henry Roediger III and Jeffrey Karpicke, Psychological Science, 2006

Key Definitions

Retrieval practice (testing effect): The finding that attempting to recall information from memory produces greater long-term retention than equivalent time spent re-studying the same material. Roediger and Karpicke (2006) demonstrated that students who practiced retrieval retained 50 percent more material one week later than students who re-read. The mechanism is that retrieval actively strengthens the neural pathways associated with the memory, whereas re-reading activates those pathways passively.

Spaced repetition: The technique of distributing study sessions over increasing time intervals rather than concentrating them in a single session (massed practice). Based on Ebbinghaus's forgetting curve research (1885) and validated in Cepeda and colleagues' 2006 meta-analysis of 254 studies, spaced repetition consistently produces two to three times greater long-term retention than equivalent massed practice.

Desirable difficulties: A term coined by Robert Bjork at UCLA to describe learning conditions that slow initial acquisition but produce significantly better long-term retention. Interleaving, spacing, and retrieval practice are all desirable difficulties. They feel harder than passive review, but the difficulty is what drives deeper encoding.

Cognitive load: The amount of working memory resources required by a learning task, defined by John Sweller in his 1988 cognitive load theory. Sweller identified three types: intrinsic (inherent complexity of the material), extraneous (complexity from poor instructional design), and germane (cognitive effort that contributes to learning). Effective learning design maximizes germane load while minimizing extraneous load.

Chunking: The cognitive process of grouping individual units of information into larger meaningful patterns. Chase and Simon's 1973 study of chess masters found that expert players did not have better memory for individual pieces but organized board positions into recognizable chunks — meaningful patterns that could be stored and retrieved as single units. Experts in any domain have larger and more sophisticated chunk libraries in their field.

The Forgetting Curve and Why It Matters

Hermann Ebbinghaus, a German psychologist working in the 1880s, spent years memorizing thousands of nonsense syllables and testing himself at different intervals to map how memory decays over time. His forgetting curve — one of the most replicated findings in all of psychology — showed that newly learned information decays exponentially: roughly 50 percent is lost within an hour, 70 percent within 24 hours, and 90 percent within a week if no review occurs.

What Ebbinghaus also discovered was the solution to his own problem. Each time he reviewed the material, the decay rate slowed substantially. More importantly, information reviewed multiple times at spaced intervals became progressively more stable, eventually requiring very long intervals before meaningful decay. The mechanism is now understood: each retrieval episode reactivates and reinforces the neural representation of the memory, making subsequent retrieval easier and reducing the decay rate.

This finding has two direct implications for anyone trying to learn effectively. First, a single study session — no matter how concentrated or effortful — is insufficient to create durable long-term memories for most material. The brain needs multiple retrieval events spread over time to consolidate memories from fragile short-term storage into robust long-term storage. Second, the timing of review sessions matters enormously. Reviewing material too soon (when you still remember it clearly) produces much smaller benefits than reviewing it when you are on the verge of forgetting it — the "desirable difficulty" that Bjork identified as central to durable learning.

Evidence-Ranked Learning Techniques

Technique Effectiveness Key Research How to Apply
Retrieval practice (testing effect) Very high Roediger & Karpicke, 2006; meta-analysis effect size 0.50 Quiz yourself instead of re-reading; use flashcards or practice tests
Spaced repetition Very high Ebbinghaus (1885); Cepeda et al. (2006) Review material at increasing intervals before forgetting occurs
Interleaving High Taylor & Rohrer (2010) Mix different problem types within a study session
Elaborative interrogation High Pressley et al. (1987) Ask "why" questions about each new concept
Self-explanation High Chi et al. (1994) Explain the material as if teaching it (Feynman Technique)
Re-reading Low Roediger & Karpicke, 2006 Provides fluency illusion; poor long-term retention
Highlighting / underlining Low Dunlosky et al. (2013) Passive; does not require active processing

Retrieval Practice: The Most Powerful Learning Tool

The Testing Effect

The finding that testing yourself improves learning more than re-studying the same material has been replicated in hundreds of studies across different materials, populations, and time scales. The landmark modern demonstration was Roediger and Karpicke's 2006 study, published in Psychological Science, which tested students on a prose passage using three conditions: study only (re-reading four times), study-test (reading once, then testing once), and test-test (testing twice without additional reading). One week later, the test-test group retained 61 percent of the material; the study-test group retained 56 percent; and the study-only group retained only 40 percent.

The effect is large, replicable, and robust. A 2013 meta-analysis by Rowland covering 159 experiments confirmed the testing effect across virtually all conditions studied. The effect size for retrieval practice compared to re-reading averaged 0.50 — a large practical difference by educational research standards.

Why does retrieval practice work so much better than re-reading? The emerging consensus from cognitive neuroscience is that the act of retrieval itself — struggling to pull information from memory — strengthens the neural encoding in a way that passive exposure cannot. The difficulty is the mechanism. When information is retrieved with effort, the memory becomes more resistant to future forgetting. When it is passively read, no such strengthening occurs.

How to Use Retrieval Practice

The practical implementation is straightforward and requires no special tools.

Closed-book summarization: After reading a section, close the material and write down everything you can remember. Do not check the source until you have written everything you can recall. Then compare your summary to the original and note what was missing.

Flashcards used correctly: Flashcards are powerful when used as retrieval tools — looking at the question and forcing yourself to produce the answer before flipping the card — but weakly effective when used passively as a reading aid. The critical discipline is to actually generate the answer, even if you are uncertain, before revealing it.

The Feynman Technique: Named for physicist Richard Feynman, this technique involves explaining a concept from memory as if teaching it to someone with no background in the subject. If you cannot explain it clearly, you have identified a gap in your understanding. You return to the source material, close it again, and repeat. The technique combines retrieval practice with elaborative encoding — linking the concept to language and examples you generate yourself.

Practice testing: Working through practice problems, past examinations, or application exercises is more effective than equivalent time reviewing notes or readings, even when the specific questions are unfamiliar. The cognitive processes required — retrieval, application, and problem-solving — are closer to the processes required when you need to actually use the knowledge.

Spaced Repetition in Practice

Cepeda and colleagues' 2006 meta-analysis, which examined 254 studies and over 14,000 participants, established the optimal spacing intervals for different retention goals. For material you want to retain for one week, a gap of one day between the first and second study session maximizes retention. For material you want to retain for a year, gaps of weeks to months between sessions are optimal.

The practical challenge is that optimal spacing intervals depend on both the material and the individual's forgetting rate, making manual scheduling difficult. This is the problem that Anki and similar spaced repetition software solve. Anki uses an algorithm developed by Piotr Wozniak called SuperMemo, which tracks your performance on each card and schedules the next review at the interval that maximizes long-term retention per unit of study time. Research on Anki users suggests that students can retain material with 30 to 40 minutes of daily practice that would otherwise require several hours of traditional studying.

For learners without spaced repetition software, a simple manual schedule for new material works well in practice: review on the day you learned it, then three days later, then one week later, then two weeks later, then one month later. The exponential spacing produces most of the benefit of optimally scheduled review without requiring software.

Interleaving and Blocked Practice

One of the most counterintuitive findings in learning research concerns the structure of practice sessions. The intuitive approach is to practice one type of problem until you have mastered it before moving to the next type — what researchers call blocked practice. This feels effective because performance improves rapidly during the blocked session. But the improvement is deceptive: when tested later on the same material, blocked practice produces substantially worse retention than interleaved practice, in which different problem types are mixed randomly.

Kornell and Bjork (2008) demonstrated this with art appreciation: students who studied paintings by an artist in a blocked format (all paintings by each artist grouped together) performed worse on identifying the artist from a new painting than students who studied the same paintings in an interleaved, randomly ordered format. The interleaving forced students to actively compare and discriminate between artists rather than passively absorbing a block of examples.

The performance paradox — interleaving feels harder and produces lower performance during practice, but produces higher retention and transfer — is one of the clearest demonstrations of the desirable difficulty principle. The difficulty of interleaving is the mechanism: when you must determine which strategy or category applies before solving the problem, you build the discrimination capability that blocked practice bypasses.

For practical skill learning, interleaving means mixing different types of problems, techniques, or topics within a single study session rather than mastering one before moving to the next. A guitar student who alternates between scales, chord progressions, and sight-reading in a single session will initially progress more slowly than one who spends the entire session on scales — but will develop more robust and transferable musical capability over time.

Sleep, Memory Consolidation, and the Biology of Learning

Matthew Walker's research at UC Berkeley, summarized in Why We Sleep (2017), established sleep as not merely helpful but essential for memory formation and consolidation. The brain's processes during sleep are active, not passive: during non-REM slow-wave sleep, the hippocampus replays the day's experiences, and during REM sleep, it integrates new material with existing knowledge structures and forms remote associations between concepts.

The practical implications are several.

Pre-sleep encoding: Material studied in the hour before sleep benefits from proximity to the consolidation window. This does not mean cramming the night before — cramming without adequate sleep defeats the purpose — but it does mean that the pre-sleep period is a high-value encoding window that most people waste on passive entertainment.

Post-learning sleep: Walker's research found that sleeping within 24 hours of initial learning produces significantly better consolidation than sleeping later. The brain cannot permanently store new information until it has processed it during sleep.

The 40 percent impairment: Walker's studies on sleep deprivation and memory found that one night of poor sleep (less than six hours) reduces the hippocampus's capacity to form new memories by approximately 40 percent. This means that studying for an exam after a poor night's sleep is dramatically less effective than studying after adequate sleep, regardless of hours invested.

Naps: A 90-minute nap in the early afternoon, which includes both slow-wave and REM sleep, has been shown to restore hippocampal learning capacity and produce retention benefits comparable to a full night's sleep for material encoded that morning.

Deliberate Practice and the Expert Performance Framework

Anders Ericsson's 1993 paper "The Role of Deliberate Practice in the Acquisition of Expert Performance," published in Psychological Review with co-authors Krampe and Tesch-Romer, established what remains the most influential framework for understanding skill development. Studying violinists at the Berlin Academy of Music, Ericsson found that the difference between elite performers and good performers was not innate talent or total hours practicing — it was accumulated hours of deliberate practice, specifically defined.

Deliberate practice, as Ericsson defined it, has four essential features that distinguish it from ordinary practice. First, it is designed to target specific weaknesses rather than to perform the parts of the skill you are already competent at. Second, it requires concentration at the edge of your current ability — beyond your comfort zone but within the range of possible performance with effort. Third, it involves immediate, specific feedback on your performance so that you can adjust. Fourth, it requires mental engagement so intense that it cannot be sustained for more than one to four hours per day.

The last feature is important and often missed. Deliberate practice is not the same as practice time. Elite performers typically engage in two to four hours of concentrated deliberate practice per day — not eight hours of playing, performing, or practicing in the ordinary sense. The additional time is spent on related activities (performance, recovery, lower-intensity work), but the deliberate practice window is limited by the cognitive resources it demands.

For practical skill development, the deliberate practice framework prescribes a specific approach. Identify the specific sub-skill that is limiting your overall performance. Design a practice task that specifically targets that weakness. Get feedback — from a coach, from objective measurement, from recordings of your own performance — that tells you precisely how you are performing relative to the target. Practice at the edge of your current ability until you have mastered the sub-skill, then identify the next limiting constraint and repeat.

Scott Young and the Ultralearning Approach

Scott Young's MIT Challenge and subsequent research — documented in Ultralearning (2019) — synthesized the cognitive science literature into a practical learning methodology for autodidacts and rapid skill acquisition. Young identified nine principles that characterized the most successful self-directed intensive learning projects he studied.

Three of these are particularly relevant to the underlying science. The principle of directness holds that learning is most effective when conducted as close as possible to the context in which the skill will be used. If you want to speak Spanish conversationally, the most direct practice is conversation, not grammar exercises. If you want to code for data analysis, the most direct practice is analyzing real data sets, not reading about algorithms. This principle maps to the cognitive science of transfer: skills learned in one context transfer poorly to different contexts unless practice was conducted in conditions similar to the application context.

The principle of retrieval directly reflects the testing effect research. Young explicitly structured his MIT projects around practice exams and problem sets rather than lectures and readings. The lectures were necessary for exposure; the problem sets were where actual learning happened.

The principle of experimentation reflects Bjork's desirable difficulties framework. Young found that deliberately varying his practice methods — changing the tools, the problem types, the level of assistance — produced better skill consolidation than consistent practice using a single approach, even when the variation made individual practice sessions harder.

Chunking and the Architecture of Expertise

William Chase and Herbert Simon's 1973 study of chess players, published in Cognitive Psychology, produced a finding that reframed the understanding of expert performance. When chess masters and beginners were shown board positions from real games for five seconds, masters could recall virtually all the pieces; beginners recalled only four or five. This seemed to confirm the idea that experts have exceptional memory. But Chase and Simon added a critical control: when the same test was run with randomly arranged pieces — positions that could never occur in a real game — masters and beginners performed identically.

The finding was decisive. Chess masters did not have better memory for pieces; they had better memory for meaningful patterns. They perceived the board not as 32 individual pieces but as a smaller number of meaningful chunks — attack formations, defensive structures, common middle-game configurations — each of which could be stored and retrieved as a single unit. Through years of study and play, they had accumulated a library of roughly 50,000 such chunks, according to Simon's estimates.

This insight generalizes across all expert domains. Expert radiologists perceive X-ray images as patterns rather than pixel arrays. Expert programmers perceive code in functional blocks rather than individual statements. Expert writers perceive sentences in rhetorical moves rather than word sequences. In each case, expertise involves the accumulated library of domain-specific chunks that allows complex information to be processed within working memory's severe capacity limits.

The learning implication is that acquiring chunks should be an explicit goal of study, not a byproduct of accumulated exposure. When learning a new domain, identifying the recurring patterns — the standard moves, configurations, arguments, structures — and drilling those patterns until they are instantly recognizable accelerates learning faster than processing each new example as a unique event.

Practical Takeaways

Replace re-reading with retrieval practice. After reading any section of material you want to retain, close the source and write or speak everything you can recall. Do this consistently and the retention difference will be measurable within weeks.

Implement spaced repetition from day one. When learning new material, schedule reviews at day one, day three, day seven, day fourteen, and day thirty. Use Anki for vocabulary, facts, and definitions. Use practice problems and application exercises for conceptual and procedural knowledge.

Interleave your practice. Resist the pull toward blocked practice. Mix different problem types, topics, or sub-skills within a single study session. Expect it to feel harder. That difficulty is the mechanism.

Protect your sleep. Do not trade sleep for study time. One hour of study after adequate sleep is worth more than two hours of study followed by reduced sleep. The encoding and consolidation benefits of sleep are not recoverable through compensatory studying.

Use the Feynman Technique for anything you need to understand deeply. Attempt to explain it plainly, identify where the explanation breaks down, return to the source only to fix the gap, and repeat until the explanation is fluent.

Design deliberate practice, not just practice time. Identify the specific sub-skill limiting your performance. Target it directly. Get feedback. Work at the edge of your ability. Do this for two to four focused hours per day rather than eight hours of comfortable practice.

Build chunks explicitly. When learning a new domain, identify recurring patterns — standard examples, common structures, typical problem types — and practice recognizing them until the recognition is automatic.

References

  • Ebbinghaus, H. Memory: A Contribution to Experimental Psychology. Teachers College, Columbia University, 1885 (trans. 1913). https://psychclassics.yorku.ca/Ebbinghaus/
  • Roediger, H. L. & Karpicke, J. D. "Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention." Psychological Science, Vol. 17, No. 3, 2006.
  • Cepeda, N. J. et al. "Distributed Practice in Verbal Recall Tasks: A Review and Quantitative Synthesis." Psychological Bulletin, Vol. 132, No. 3, 2006.
  • Kornell, N. & Bjork, R. A. "Learning Concepts and Categories: Is Spacing the 'Enemy of Induction'?" Psychological Science, Vol. 19, No. 6, 2008.
  • Bjork, R. A. "Memory and Metamemory Considerations in the Training of Human Beings." In Metcalfe, J. & Shimamura, A. (eds.), Metacognition: Knowing About Knowing. MIT Press, 1994.
  • Ericsson, K. A., Krampe, R. T. & Tesch-Romer, C. "The Role of Deliberate Practice in the Acquisition of Expert Performance." Psychological Review, Vol. 100, No. 3, 1993.
  • Sweller, J. "Cognitive Load During Problem Solving: Effects on Learning." Cognitive Science, Vol. 12, No. 2, 1988.
  • Chase, W. G. & Simon, H. A. "Perception in Chess." Cognitive Psychology, Vol. 4, No. 1, 1973.
  • Walker, M. Why We Sleep: Unlocking the Power of Sleep and Dreams. Scribner, 2017. https://www.simonandschuster.com/books/Why-We-Sleep/Matthew-Walker/9781501144325
  • Young, S. Ultralearning: Master Hard Skills, Outsmart the Competition, and Accelerate Your Career. Harper Business, 2019. https://www.scotthyoung.com/blog/ultralearning/
  • Rowland, C. A. "The Effect of Testing Versus Restudy on Retention: A Meta-Analytic Review of the Testing Effect." Psychological Bulletin, Vol. 140, No. 6, 2014.
  • Csikszentmihalyi, M. Flow: The Psychology of Optimal Experience. Harper & Row, 1990.

Tags

Learning Memory Study Techniques Cognitive Science Deliberate Practice

Frequently Asked Questions

What is the most effective study technique according to research?

The research consensus from cognitive psychology and educational science identifies retrieval practice — actively recalling information from memory rather than passively re-reading it — as the single most effective technique for durable learning. Henry Roediger and Jeffrey Karpicke's 2006 study in Psychological Science found that students who practiced retrieval retained 50 percent more material one week later than students who spent the same time re-studying. The superiority of retrieval practice holds across subject domains, age groups, and time scales. Spaced repetition — distributing study sessions over time rather than massing them together — is a close second, and the two techniques compound each other powerfully when combined: retrieving information at spaced intervals produces substantially better long-term retention than either technique alone.

Why does re-reading feel productive but isn't?

Re-reading produces a strong feeling of familiarity with the material, which the brain mistakes for understanding and retention. This is what cognitive scientists call a 'fluency illusion': the text feels familiar because you have seen it before, but familiarity is not the same as retrievability. When you try to recall the information without the text in front of you — in an exam, a work meeting, or a practical application — you find that the familiarity has not translated into accessible memory. Research by Roediger and Karpicke, confirmed in subsequent meta-analyses, consistently shows that re-reading produces minimal long-term retention compared to retrieval practice conducted at equivalent time investment. The problem is structural: re-reading keeps the information in front of you, so the retrieval circuits in your brain are never required to work. Without being exercised, those retrieval pathways remain weak.

How does sleep help you learn?

Sleep plays two distinct roles in learning and memory that Matthew Walker summarized in his 2017 book Why We Sleep. Before learning, sleep prepares the brain's hippocampus to receive and encode new information — sleep-deprived individuals show up to 40 percent reduced capacity for forming new memories. After learning, sleep consolidates memories through a process of neural replay during non-REM slow-wave sleep (for factual memories) and REM sleep (for procedural and associative memories). During this consolidation, memories are transferred from short-term hippocampal storage to long-term cortical storage and are integrated with existing knowledge structures. The practical implication is that the period immediately before sleep is among the most effective for encoding new information, and sleep within 24 hours of study is essential for consolidation. Pulling an all-nighter before an exam dramatically impairs the brain's ability to access what was studied.

What is spaced repetition and how do you use it?

Spaced repetition is a technique in which study sessions on the same material are distributed over increasing time intervals rather than concentrated in a single session. The technique is grounded in Ebbinghaus's forgetting curve (1885), which showed that information is forgotten at a predictable exponential rate unless reviewed, and that each review resets the decay rate at a higher baseline. Cepeda and colleagues' 2006 meta-analysis of 254 studies confirmed that distributed practice produces retention two to three times greater than massed practice ('cramming') at equivalent total study time. In practice, spaced repetition works by reviewing material just before you are about to forget it — when the retrieval requires genuine effort but is still possible. Anki and similar software automate this scheduling using algorithms that adjust review intervals based on your performance. Without software, a simple implementation is reviewing new material at intervals of one day, three days, one week, two weeks, and one month.

How is deliberate practice different from regular practice?

Anders Ericsson's 1993 research on expert performance, published in Psychological Review, identified a specific form of practice he called deliberate practice that is qualitatively different from ordinary practice. Regular practice involves doing something you can already do, which builds fluency and confidence but does not expand capability. Deliberate practice requires working at the edge of your current ability — on specific, identified weaknesses — with immediate feedback and the explicit goal of improving performance. Ericsson's research on musicians, chess players, athletes, and medical professionals found that the amount of deliberate practice, not total years of experience, predicted expert performance. Key features of deliberate practice include: a clearly defined skill target at or just beyond your current ability; immediate, specific feedback on your performance; concentrated focus (deliberate practice is mentally exhausting and cannot be sustained for long periods); and repetition with reflection, not just repetition.

How long does it take to become proficient in a new skill?

The commonly cited '10,000 hours to mastery' figure, derived from Ericsson's research, refers to world-class expert performance in highly competitive domains — classical music, chess, elite sport — not general proficiency. For most professional and practical skills, the learning curve is much more favorable. Josh Kaufman's analysis in The First 20 Hours (2013) found that twenty hours of deliberate practice — approximately forty minutes per day for a month — is sufficient to move from complete incompetence to reasonable competence in most skills. This matches the empirical shape of the learning curve: the most rapid gains occur in the first hours of practice as you move from no knowledge to basic functional understanding. Subsequent gains require progressively more practice time per unit of improvement. The practical implication is that you can become usefully competent in a new skill much faster than you might expect, but achieving expert-level performance requires a substantially larger and more disciplined investment.

What learning methods work best for adults?

Adult learners differ from children in several ways that affect learning strategy. Adults have larger existing knowledge structures that can serve as frameworks for integrating new information — this is an advantage, since research by Ausubel (1960) and others shows that anchoring new material to existing knowledge dramatically improves retention. Adults also have stronger metacognitive awareness, meaning they can more reliably identify what they know and do not know, which enables better calibration of study effort. The most effective adult learning strategies combine retrieval practice and spaced repetition with active connection-building — explicitly linking new concepts to existing knowledge. Adults also benefit from learning in context and for application: when the learning is tied to a real use case, retention and transfer are significantly higher than when it is abstract. The main vulnerability for adult learners is time scarcity, which creates pressure toward passive methods like re-reading and video watching that feel efficient but produce poor retention.

Share this article