In the 1880s, German psychologist Hermann Ebbinghaus memorized lists of nonsense syllables and tested himself over time, producing what's now called the forgetting curve: in his data, more than half of new material was gone within an hour, about two-thirds within a day, and about three-quarters within a week if he never revisited it. Modern replications confirm the same basic shape. This isn't a sign your memory is bad — it's what every untrained memory does, because the brain has no way of knowing, in the moment, which of the hundreds of things you encounter today will actually matter next month. The entire premise of this course is that you can tell it, deliberately and cheaply, which things are worth keeping.
Three stages, simplified: encoding (getting information in, which happens in working memory — small, and it fades in seconds without action), storage (holding it, which for anything long-term means the hippocampus first, then gradual transfer to the cortex during sleep — more on this in 3.3), and retrieval (pulling it back out, which turns out to be far more than a passive readout — the act of retrieving something is itself what makes it stick better next time, the single most important idea in this course).
The testing effect is one of the most replicated findings in learning science: actively trying to retrieve information strengthens memory of it far more than simply re-exposing yourself to it again. In a landmark study by Roediger and Karpicke, students who took a practice recall test on a passage remembered more of it a week later (about 56%) than students who spent that time rereading it (about 42%), even though the rereaders felt more confident. Rereading feels productive because the material feels familiar — but familiarity with the page is not the same as being able to produce the information yourself, and the test is the thing that actually trains that skill.
The second major finding, distributed practice, is almost as well established: spreading review sessions out over time beats cramming them together, even when the total study time is identical. A large meta-analysis by Cepeda and colleagues confirmed this holds across a wide range of material and time scales, and some studies have found spaced review can improve long-term retention by up to 200% compared to massed practice. A practical, Ebbinghaus-informed starting schedule: review once within the first hour, again within 24 hours, again within a week, and again within a month — each successful review can stretch the next gap further, because remembering something at the edge of forgetting it is precisely what re-strengthens it.
A 2013 review by Dunlosky and colleagues ranked ten common study techniques by how much evidence actually supports them. Practice testing and distributed practice — exactly the two techniques in 1.3 and 1.4 — came out on top. Highlighting, underlining, rereading, and summarizing — the techniques most students actually default to — were rated low utility: not useless, but far weaker than what most people assume they're doing when they "study." Cramming the night before can produce a passable score on tomorrow's test while producing almost nothing that survives a month; it optimizes for the wrong curve entirely.
Say you're learning 30 words in a new language. The instinct is to write them all down and reread the list daily. Here's the same task run through this module instead: Day 1 — learn the words, then immediately close the list and try to recall all 30 from memory (1.3), checking which ones you missed. Later that same day — a second recall attempt, within the first-hour-to-24-hour window (1.4). Day 2 — recall again; words you got right move to a longer gap, words you missed stay on a short one. Day 8, then roughly Day 30 — recall attempts continue, spacing out further each time a word is remembered correctly. By day 30, you've spent less total time than daily rereading would have taken, and you can actually produce the words — not just recognize them on a list.
You don't need software to use spacing (a simple box of index cards sorted into "review today / review this week / review this month" piles works), but flashcard software automates the scheduling math. The classic algorithm, SM-2 (Piotr Wozniak, 1987), tracks an "ease factor" per item and multiplies the review interval by it each time you get something right, so intervals grow geometrically — days, then weeks, then months. Newer software increasingly uses FSRS, a model fit to hundreds of millions of real review sessions that targets a specific desired retention rate directly, and needs roughly 20–30% fewer total reviews than SM-2 for the same result. The algorithm matters far less than actually using one consistently — a slightly-less-optimal schedule you stick to beats a perfect one you abandon after a week.
The method of loci, used since antiquity and still the core technique behind competitive memory sports, works by anchoring pieces of information to specific locations along a familiar route — your childhood home, your commute — and picturing a vivid, often strange image of each item at each spot, then "walking" the route to retrieve them in order. Meta-analyses confirm it works: participants recall roughly 20% more information using it compared to standard memorization. It's most useful for ordered lists and discrete facts — a speech's talking points, a shopping list, the planets in order — less so for deep conceptual understanding, which the techniques in 2.4 handle better.
Working memory holds surprisingly little at once — the famous "seven, plus or minus two" figure from 1956 has been refined by more recent research (Nelson Cowan) to closer to four meaningful chunks at a time. Chunking is the trick that gets around this limit: grouping individual pieces into fewer, larger units. A phone number is easier to hold as three chunks (555-867-5309) than ten separate digits; a long password is easier to hold as a short memorable sentence than as random characters. The chunk itself doesn't have to be simple — an expert chess player "chunks" an entire board position into a handful of familiar patterns, which is a large part of why they can recall positions novices can't.
Elaborative interrogation — repeatedly asking yourself "why is this true?" or "how does this connect to what I already know?" — was rated moderate-to-high utility in the same Dunlosky review from 1.5, and it works by giving new information more connection points to existing memory, which makes it easier to retrieve later from more than one direction. Dual coding — pairing words with an image or a diagram, rather than relying on text alone — adds a second, independent retrieval path for the same idea. Neither technique replaces recall and spacing; both make each individual review session more effective.
Interleaving means mixing different topics or problem types within a single study session, rather than blocking all of one type together before moving to the next. It reliably feels harder and slower in the moment — you lose the "flow" of doing the same kind of problem repeatedly — but it forces you to actively identify which method or fact applies each time, which is itself a form of retrieval practice. This is one of the more counterintuitive, well-supported findings in the field: the version that feels worse while you're doing it is often the version that produces better long-term results.
A practical, everyday application of nearly everything above: when you meet someone, repeat their name back immediately ("Good to meet you, Maria" — retrieval, right away), make a deliberate association between the name and a distinctive feature or an image (elaboration and dual coding), and use the name again within the conversation and once more before you part (spaced review, compressed into minutes instead of days). None of this is a trick specific to names — it's the same system from this whole module, applied to a situation most people never think to apply it to.
Not everything should be scheduled the same way. Discrete facts (vocabulary, dates, definitions) fit the classic expanding-interval flashcard system in 2.1 cleanly. Skills (a language's grammar, a musical instrument, a sport) need spaced practice too, but of doing, not just recalling — the same expanding-gap logic applies, just applied to performance rather than retrieval. Deep conceptual understanding (why a historical event happened, how an economic system works) benefits most from the elaboration and interleaving in 2.4 and 2.5, with spaced review used to check whether you can still explain the concept in your own words, not just recite a definition.
Competitive mnemonists memorizing hundreds of random digits typically use the Major System: each digit maps to a consonant sound (0=s/z, 1=t/d, 2=n, and so on), which turns strings of numbers into pronounceable words that can then be placed into a memory palace (2.2) as vivid images. It's a genuinely learnable skill, not a talent some people are born with — but it takes real dedicated practice to build the digit-to-word mappings until they're automatic, and it's realistically worth the investment only for people who need to memorize numbers often (performers, specific competitive or professional contexts) rather than for everyday use.
Sleep isn't just rest between study sessions — it's an active part of the memory system itself. During sleep, newly formed memories are gradually transferred from the hippocampus, where they're first and fragile, to the cortex, where they become more stable and resistant to interference; slow-wave sleep and sleep spindles appear especially important for this process. Research has found measurable memory benefits from both a full night's sleep and shorter one-to-two-hour naps. Practically: a study session followed by a full night's sleep genuinely outperforms the same study session followed by staying up later to cram more in — the sleep is doing real consolidation work, not just recovering energy.
People who compete in memory championships — memorizing the order of a shuffled card deck in under a minute, or hundreds of random digits — aren't working with a different kind of brain. Research on top competitors has found their underlying memory capacity isn't unusual; what's different is the deliberate, structured technique (palaces, the Major System, extensive practice) layered on top of an ordinary memory. This is genuinely encouraging: "photographic memory," in the sense of a perfect, effortless total recall, doesn't hold up as a real, well-documented phenomenon — what looks like exceptional memory is, in essentially every studied case, trained technique plus practice.
Each successful recall doesn't just refresh a memory — it measurably flattens the forgetting curve for that item, so the next gap can be longer than the last one. Run this cycle enough times on something and it can become close to permanent — which is the whole logic behind why language learners who use spaced systems can still recall vocabulary years after their last formal lesson, while cramming for a final exam is famously gone within weeks. The system compounds: the same five minutes of review is worth more in month three of a spaced schedule than it is in week one, because the curve is already flatter by then.
Not everything is worth memorizing this deliberately, and treating every fact you encounter as flashcard material is its own kind of mistake — the same honest caution this library's Speed Reading course applies to reading speed applies here to memory technique: match the tool to the actual goal. Information you can look up in ten seconds when you need it (a rarely-used reference number, a fact you'll only ever need once) usually isn't worth the ongoing review cost; understanding a concept well enough to reason about it flexibly matters more than memorizing its exact phrasing; and for material you'll use constantly in your actual work or life, the practice of doing the work itself often builds the memory as a byproduct, without a separate system layered on top.
A realistic starting plan, not everything in this course at once.
Forgetting-curve figures reflect Hermann Ebbinghaus's original 19th-century research and subsequent replications. Distributed-practice findings reflect Cepeda et al.'s meta-analysis on spacing effects. The testing-effect comparison reflects Roediger & Karpicke's retrieval-practice study. Study-technique utility rankings reflect Dunlosky et al. (2013), "Improving Students' Learning With Effective Learning Techniques," Psychological Science in the Public Interest. Method-of-loci effectiveness reflects meta-analytic findings on the technique. SM-2 and FSRS details reflect published documentation of these spaced-repetition algorithms. Sleep-consolidation findings reflect current sleep and memory research. This is educational content, not a substitute for professional guidance on learning disabilities or memory-related medical conditions.
📘 Make It Stick: The Science of Successful Learning — Brown, Roediger & McDaniel (the most accessible full treatment of retrieval practice and spacing)
📘 Moonwalking with Einstein — Joshua Foer (a journalist's deep dive into competitive memory technique, including the method of loci and the Major System)
🌐 Any spaced-repetition flashcard app's own documentation on their scheduling algorithm — worth reading once you're ready to build a real system (2.1)
🌐 Related pieces in this library: Memory: How to Make Things Stick (category 17) and the Speed Reading course's lesson on long-term retention, which this course builds directly on