6 min read

Built on the Science of Memory

noruditmethodologylearning sciencememory

Everything on the Norudit homepage rests on a single line: built on the science of memory. This is that science, unfolded: five principles, each with a name, a paper, and a job.

Most study apps are built on a hunch about what feels productive: highlighting, re-reading, a tidy summary. Norudit is built on a small set of findings that have survived decades of replication, and it uses each one for exactly what it is good at. No more. The learning flywheel explains how these are arranged into a sequence. This piece is the other half: what each principle is, and why it earns its place in the machine.

Spaced repetition: reviews timed to forgetting

In 1885 Hermann Ebbinghaus measured his own forgetting and drew the curve every student lives on: memory decays fast at first, then levels off, and each time you successfully recall something, the next drop is slower. A century of work turned that curve into a rule: the same hour of study spread across days beats the same hour crammed into one (Cepeda et al., 2006). Review just before you would have forgotten, and each pass buys more time than the last.

Fixed-card tools like Anki proved the science works. Norudit schedules with FSRS-6, a modern algorithm fitted to your forgetting curve rather than a textbook average. And then it does one thing a flashcard can't: it rewrites the card each time, so you are tested on the concept, never on recognising a card you've seen forty times. The full story is The Deck That Rewrites Itself, and how it fits itself to you is The Academy That Remembers You.

Knowledge tracing: a live map of what you know

Spacing tells you when to review. It doesn't tell you what. For that you need an estimate, held per concept and updated every time you answer, of how well you actually know each idea. That model has a name: Bayesian Knowledge Tracing (Corbett and Anderson, 1994), first built to decide which skill a student should practise next.

Norudit keeps that estimate for every concept in your course. It is what lets the review deck surface your weakest ideas first instead of marching through the material in order, and what turns an exam's error analysis into a targeted plan rather than a grade. You are never studying the average topic; you are studying the one the model thinks is about to slip.

Retrieval practice: recall it, don't reread it

Here is the most counter-intuitive finding in the whole field: the act of pulling a fact out of memory strengthens it far more than putting it in again. In the canonical study, students who read a passage and then took a recall test remembered dramatically more a week later than students who simply re-read it four times (Roediger and Karpicke, 2006). Re-reading feels like progress. Retrieval is progress.

It also feels worse, and that is the point. The effort of dragging something up unaided is not the cost of the exercise; it is the mechanism (Bjork and Bjork, 2011). Norudit leans into this everywhere it can, and hardest at the blank page of free recall, where there are no prompts to lean on. Why the discomfort is the feature is its own essay: Why Difficulty Is the Point.

The Feynman technique: explain it simply, or you don't know it

Richard Feynman's study method survives in one blunt sentence: if you can't explain it simply, you don't understand it. The research calls it the self-explanation effect: students who explain material back to themselves, in their own plain words, understand it more deeply than those who read it twice (Chi et al., 1989). Plain language has nowhere to hide; the exact spot where the simple words run out is the exact spot you don't yet understand.

This is the one hard gate in Norudit. You don't pass it by receiving an explanation; you pass it by giving one, to someone who won't let you wave your hands. It is the one test you can't fake, and it sits precisely where the feeling of knowing gets mistaken for knowing.

Grounded in your material: the principle that makes the other four trustworthy

The fifth is not a memory principle at all. It is the answer to AI's biggest flaw: models make things up. A tutor that spaces, traces, and tests you flawlessly is worthless if what it teaches is confidently wrong.

So Norudit teaches, questions, and grades from your uploaded material (your textbook, your notes, your lecture PDFs), not a web search and not its training data. It isn't perfect, but it is far lower-risk, and it always hands you the source page to check. The architecture behind that is Grounded on the Page. It is what lets the other four principles be trusted: reviewing, tracing, and recalling the right thing only matters if the thing is right.

The whole point

These are not features bolted on for a marketing page. Each is a mechanism with a single job, chosen because the evidence says it works, and arranged so each one feeds the next.

A hunch about what feels productive resets every session. A principle with a paper behind it compounds.

None of this is hidden behind the sign-up wall as theory. The spacing, the tracing, the retrieval, the gate, and the grounding are all built and running on whatever material you bring. Start your first class →

If you would rather go to the sources than take our word for it, the books and papers that shaped the method are walked through in The Bookshelf Behind the Method.

Sources