How do we learn?
We are trying to answer that question. We write beliefs about how people learn that could turn out to be wrong, and we attack them with real studies.
What survives stays. What fails is narrowed or dropped. We do not score beliefs — we only count the attacks.
What we know so far
A short story of the beliefs that have been tested. Each line says no more than that belief says tonight.
1. What “learned” means, and how memory holds
On this site, something is learned when you can recall it later, without looking, from prompts that match what you practiced. After that first step, recalling it repeatedly with feedback beats rereading on tests a day or more later. Spreading practice out and mixing types also beats cramming on tests weeks later, though a gain that lasts a whole school year has not been shown. And unmarked extras that wander off-topic can get in the way of applying what you learned, though that harm is not consistent.
More in Memory and practice · Screens, devices and attention.
2. What you already know changes what you can take in
Knowing a topic helps you understand texts about it, though teaching that knowledge does not always raise general reading scores. For children just cracking the reading code, teaching letter–sound links step by step helps those who struggle to decode; better comprehension from that teaching has not been shown.
More in Reading and comprehension.
3. The body and feelings
Getting enough sleep in the nights before you learn protects what was learned. Feelings matter at test time too: higher test anxiety goes with lower scores and grades, a small to moderate link. Acute stress just before a memory test lowers adults’ recall by about a fifth of a standard deviation, but only in studies that left out women taking hormonal contraceptives. Whether starting the school day later helps is a separate policy question: it can add sleep for teenagers, but better results are not guaranteed.
More in Sleep and the body · Emotion, stress and arousal.
4. How people get better
Teaching pupils to plan, monitor and check their own learning can raise later school scores. Practice matters, but it explains only part of why some people perform better than others. Promising a prize simply for doing work people already find interesting can make them choose it less once the prize stops. And explaining what you just learned can help you later, if you studied it expecting to teach.
More in Metacognition and self-regulation · Expertise and practice · Motivation and curiosity · Learning from other people. Growth-mindset lessons sit in motivation: a short growth-mindset lesson does little on average; extra help for struggling teens is unproven.
5. How this shows up with tools and in school
When practice is handed to a tool, letting AI give you the finished answers often makes you look better now and know less later. The medium matters for reading: past early primary school, paper often edges screens when deep reading is timed. In the classroom, phone bans have raised results in some places, with no clear average gain, and on open-ended work, comments on how to improve beat a grade alone.
Further areas: AI and thinking · Screens, devices and attention · Teaching and feedback · Learning outside school · Special education and inclusion · Language learning · Music and arts training · Adult / workplace learning · Bilingualism and dual literacy · Early childhood and language.
What we do not know yet
Questions the attacks have opened and not answered, and ideas that failed fair tests.
Suggest a belief
Write one sentence about how people learn that could turn out to be wrong. It will not appear on this site unless it is approved. If it is, it will first be checked against what we already know, and only then — if it still earns a place — enter the nightly attacks.
Prefer a quiet page? Open the suggestion form on its own.
Last night: 2 trials on C10 (Tutoring software in school use) and C19 (Metacognitive strategies); 1 held, 1 narrowed · 26 newly added, still under test · How we work