Pick up a language you abandoned a decade ago and something strange happens. The first week feels like grief — all those verbs, gone. Then, around week two, words start arriving before you ask for them. Conjugations you never consciously reviewed simply work. Within a month you're somewhere it originally took you a year to reach.
If that knowledge was truly gone, this shouldn't be possible. You'd be starting from zero, and relearning would cost what learning cost. It never does. The most useful explanation for why comes from cognitive psychologists Robert Bjork and Elizabeth Bjork, whose new theory of disuse proposes that every memory has not one strength but two — and that almost everything confusing about studying, from the false confidence of cramming to the quiet power of spacing, falls out of the difference between them.
One Memory, Two Strengths
In the Bjorks' framework, first laid out in 1992, any item in memory — a word, a formula, a face — carries two independent values.
Storage strength is how deeply the memory is entrenched: how thoroughly it's been woven into everything else you know. It grows with every meaningful encounter and, in the theory's boldest assumption, it never decreases. Once something is well learned, that learning is not undone by time.
Retrieval strength is how accessible the memory is right now — how easily you can summon it on demand. This is the volatile one. It rises sharply when you study or recall something, and it decays with disuse, which is exactly what it feels like when a name sits on the tip of your tongue.
The crucial insight is that these two can come apart completely. Your childhood phone number may have enormous storage strength and, this morning, almost no retrieval strength. The hotel room number you memorized an hour ago has the opposite profile: instantly accessible, barely stored. It will be unreachable by Thursday — not because it decayed more, but because it was never entrenched to begin with.
Forgetting Is Losing the Address, Not the House
This distinction echoes an older finding. In a classic 1966 experiment, Endel Tulving and Zena Pearlstone had people study lists of categorized words, then tried two kinds of tests. Asked to recall freely, participants came up short. Given category cues — a type of bird… — they suddenly produced words they had appeared to forget entirely. Tulving's conclusion: memories can be available in storage while being inaccessible at that moment. The failure wasn't erasure. It was retrieval.
Hermann Ebbinghaus saw the same shadow from another angle in the 1880s. He couldn't directly measure what remained of his memorized syllable lists, so he measured savings: how much faster he could relearn a list compared to learning it fresh. Even for material he could no longer recall at all, relearning was reliably cheaper. Something persisted below the waterline of conscious recall.
This is why your abandoned language comes back in weeks. The storage strength you built years ago is still there; what collapsed was only access. Relearning isn't rebuilding the house — it's repainting the address on the curb.
The Rule That Changes How You Study
Here is where the theory turns from consoling to practical. The Bjorks propose a counterintuitive relationship between the two strengths: the gain in storage strength from studying or retrieving something is greater when its current retrieval strength is lower.
Read that again, because it inverts how most people study. When a fact is fresh and easy to recall — high retrieval strength — reviewing it again teaches you almost nothing durable. The retrieval is effortless, and effortless retrievals barely move storage strength. But when a fact has faded, when you have to grope for it and drag it back, that difficult retrieval produces a large, lasting gain. The fading itself is what makes the review valuable.
Forgetting, in other words, isn't the enemy of learning. Partial forgetting is the precondition for the most profitable kind of practice. This is the engine underneath the spacing effect: waiting until a memory has grown less accessible before you review it doesn't just save time — it makes each review worth more.
Why Cramming Feels Like Learning
The two-strengths model also explains the most reliable illusion in studying. The night before an exam, you drill the same material over and over. Each pass gets smoother. By 1 a.m. everything comes instantly, and instant feels like known.
But what you can feel — what any self-quiz measures — is retrieval strength only. Storage strength is invisible. Cramming pumps retrieval strength to its ceiling while adding little entrenchment, because every repetition happens while the material is maximally accessible, exactly when the storage gains are smallest. You walk into the exam with a tall, hollow structure. It may survive the morning. It will not survive the month.
The cruel symmetry: the studying that feels best is the studying that stores least, and the studying that stores most — halting, effortful recall of half-faded material — feels like failure while it's happening. Your sense of fluency is measuring the wrong variable.
Studying With Two Strengths in Mind
Once you separate the strengths, a few practical rules follow directly.
Let things fade before you review them. The wobble you feel when recalling something days later isn't evidence the first session failed. It's the setup for a high-value retrieval. Schedule reviews at the edge of forgetting, not inside the glow of familiarity.
Test, don't re-expose. Rereading raises retrieval strength a little and storage strength almost none. Retrieving from memory — even slowly, even wrongly at first — is what converts access into entrenchment.
Stop treating relearning as defeat. Coming back to a neglected subject, the honest response to "I've forgotten everything" is: no, you've lost access to everything, and access is cheap to restore. Savings are real. The second climb is always faster, and the retrieval practice involved in relearning builds storage strength the first pass never did.
Distrust fluency at the moment of study. How easily something comes to mind tonight tells you almost nothing about whether it will come to mind in March. Judge your learning by what you can retrieve after a delay, because delay is the only test that isolates storage.
The Model Under the Hood
It's worth noticing how directly this fifty-year arc of theory has become engineering. Modern spaced-repetition algorithms like FSRS model each fact with variables that map almost exactly onto the Bjorks' two strengths: a stability value capturing how slowly the memory decays — storage strength, operationalized — and a retrievability value estimating the probability you can recall it right now. The scheduler's whole job is to watch retrievability fall and intervene precisely when a retrieval will buy the most stability. What the theory predicted, the algorithm executes: reviews timed to low retrieval strength, because that's where durable memory is made.
Doing that arithmetic by hand — tracking hundreds of facts, each fading at its own rate, each due at its own moment — is exactly the kind of bookkeeping human attention is bad at and software is built for. That's what Recall does: fast, beautiful flashcards scheduled by FSRS, so every card resurfaces right when its retrieval strength has sagged enough to make the review count. It works fully offline and imports your existing Anki and Quizlet decks, so the storage strength you've already built comes with you. You supply the effortful recall; it supplies the timing. The forgetting, it turns out, was never the problem — reaching things at the right moment was, and that part can be remembered for you.