Loading
Paivio isn't a hunch about studying. Every scheduling decision traces to findings cognitive scientists have replicated for more than a century, plus a randomized comparison of three fixed study formats to measure what they do in practice.
In 1885 Hermann Ebbinghaus memorized nonsense syllables and tested himself over days, and found that forgetting isn't random. It follows a curve: steep at first, then flattening. Recall drops fastest right after learning, and each time you let a memory decay and then retrieve it, the next drop is slower.
That single insight is the whole opportunity. If forgetting is predictable, then the best moment to review is knowable too: not while a memory is still fresh (wasted effort), and not long after it's gone (relearning from scratch), but at the edge, where recall is just starting to fail.
Paivio estimates that edge for every concept you're learning and schedules the review to land there. The dashed line is what happens to an untouched memory. The blue line is the same memory, caught at the edge three times.
These aren't fringe results. Each has held up across decades of studies, classrooms, and subjects, including dense medical material.
The same total study time produces far more durable memory when it's spread across expanding intervals than when it's massed in one sitting. Cramming feels productive because recall is high during the session; it collapses days later. Spacing trades a little short-term ease for a lot of long-term retention.
Retrieving an answer from memory strengthens it more than re-reading the answer does, even when re-reading feels more thorough. The effort of pulling a fact out is what consolidates it. This is why Paivio makes you recall rather than review, and why every card is a question, not a summary.
Desirable difficulty describes practice that demands effort without making success impossible. Paivio does not fit retrieval difficulty or format to an individual learner; which study mechanisms are delivered depends on the fixed study format assigned to the account.
Fixed-interval systems treat every card the same: right answer, multiply the interval by a constant. That works, but it's blunt. Paivio's scheduler models each concept's own memory stability and difficulty, in the spirit of the modern FSRS approach to spaced repetition, and predicts the probability you'll recall it on any future day.
It then schedules the review for the day your recall probability crosses your target, so you see hard concepts often and easy ones rarely, without you ever tuning a setting.
Reviewing a concept while it's still solid wastes the review; a retrieval only strengthens memory when it takes effort. Fixed intervals get this right on average and wrong on the concepts that matter most, the ones that are unusually hard or easy for you.
The science above holds on average. To check what it does in practice, Paivio runs a randomized comparison: accounts are assigned once, in equal proportions, to one of three fixed study formats, and Paivio collects outcomes to compare them on the outcome that matters: durable recall a week or more later.
Your review timing, your ratings and your due dates are identical in every format; only which study mechanisms are delivered on top of them differs, and FSRS still advances only from your explicit rating. Nothing is fitted to you as an individual, and the comparison is not adaptive: your format is fixed when your account is assigned.
This runs only with your consent, and how your learning is measured is shown in your learning profile in plain English. You can change how your data may be used at any time. Opting out of the comparison is permanent: an account that opts out stays out, its later study is not counted, and your cards and your schedule do not change either way.
Landmark work behind the scheduling. We stand on it; we don't overclaim beyond it.
You shouldn't have to design your own spacing schedule. Paivio does it, at the edge of forgetting, every day.
