Learning to Learn — Learning Library¶
A learning library about learning itself, built the same way as its siblings math-learning-library ↗ and rust-learning-library ↗: one idea per page, and every claim backed by a program that actually runs.
No page here hand-types what a program prints. Each lesson links a real .py file; a tool runs it, checks the output against a recorded answer key, and pastes that verified output into the page. CI fails if any of the three drift apart. So when a page says "a student who reuses the last formula scores 75% on the blocked sheet and 25% on the exam", that is not a promise; it is a test result.
Some claims about learning no program can check: what sleep does to a memory, what a walk does to attention, whether a supplement does anything. Those pages do the next best thing. Every claim on them carries a grade, well established, good evidence, weak or speculative, with the source it rests on, and no page gives a dose or recommends anything to take.
The examples are stdlib-only, on purpose. If you have python3, you can run every page in this repo.
📖 Read it as a site: https://masiarek.github.io/learning-to-learn-library/
Start here¶
01_Knowing_What_You_Know/ — How do you know that you know?
| Lesson | What it teaches |
|---|---|
| Metacognition: judging what you know | Flavell's four abilities, and the one a program can measure: confidence against results, the Brier score, and why honest confidence minimises it |
| Count the vowels | McGuire's exercise: 3 of 15 phrases remembered after counting vowels, 12 after knowing the goal and the principle |
| Studying vs learning | Bloom's six levels climbed on the Pythagorean theorem, up to a formula that re-creates the facts a student would memorise |
02_Making_It_Stay/ — How do you learn so that it lasts?
| Lesson | What it teaches |
|---|---|
| Spaced retrieval | The forgetting curve, why testing beats rereading, the study cycle, and why a geometric review schedule makes remembering cost a logarithm |
| Interleaving | A blocked sheet of volume problems asks you to choose a formula 4 times in 12, a mixed exam 7 in 8; a student who reuses the last formula scores 75% on the sheet and 25% on the exam |
| Cognitive load | Working memory holds about four chunks, and a chunk is whatever practice made automatic: (a+b)² = a² + 2ab + b² is 19 items to a beginner and 1 to an expert |
| Focused and diffuse thinking | Small uphill steps stop on the nearest hill: a thousand focused steps end where three did, a wide survey finds the right hill but not its top, and focus, step back, focus again reaches the answer in fifteen looks |
03_Believing_You_Can/ — What does a belief have to do with the evidence?
| Lesson | What it teaches |
|---|---|
| Learned helplessness | Honest counting, (s + 1)/(n + 2), leaves a learner who failed ten times never trying the lever that works; four small successes, or four watched ones, get them started |
| Day or night: perspective taking | "When does night begin?" has four exact answers in Starbuck, WA, 20:49 to 23:39, one per cutoff; the sorites, and fuzzy logic's third way out |
04_Planning_the_Work/ — Why does a good plan still fail, and what fixes it?
| Page | What it teaches |
|---|---|
| The buffer hour | Task times are skewed, so five honestly estimated one-hour tasks fit in five hours one day in thirteen; a reserved hour gives 47%, and a recorded overrun ratio corrects the estimates |
| Goals and schedules | Specific goals just below your best, process goals, stretch goals, a written week with an hour in reserve, to-do rules, and which of the circulating "protocols" are findings |
| Learning with AI | Answers are rereading made perfect; questions, hints, quizzes and a student to teach make every technique cheaper |
05_Body_and_Brain/ — Which claims about sleep, exercise, food and the brain should a learner believe?
| Page | What it grades |
|---|---|
| How memory works | The pipeline from attention to storage, the three processes, the forgetting curve, the big four techniques and desirable difficulties |
| Sleep | Sleep after learning, sleep loss before it, naps, quiet rest, and what is mostly from rats |
| Exercise | Long-term brain health, attention for the next hour, the small and inconsistent effect on a study session, and BDNF |
| Food and supplements | Caffeine, deficiencies, diet over decades, and why the nootropic evidence is thin; no doses |
| Brain chemistry | LTP, acetylcholine, dopamine, norepinephrine: good evidence in animals, and what does and does not follow |
| Mindset and motivation | Self-regulation, self-efficacy, growth mindset's real size, two biases, imposter syndrome, stress, intrinsic motivation |
Other ways in¶
- Start here, topic map, glossary, resources (books), references (papers, with what kind of study each is), roadmap.
- Every lesson ships an Anki deck and ends with a short summary in Polish.
Running the examples¶
python3 02_Making_It_Stay/interleaving/examples/interleaving.py # any one page
python3 tools/run_examples.py --check # all of them, as CI does
Contributing¶
House rules are in CONTRIBUTING.md: one idea per folder, output generated never typed, stdlib only, evidence grades in place of a program where no program can check the claim, and a place in the topic map for every page.