That's not a flaw — it's how memory works. NeuroVocab is engineered around it: stories that encode words deeply, an algorithm that catches each word the moment it starts to fade, and seven practice modes that forge it into long-term memory.
The old watchmaker held the gear to the light, his face inscrutable — revealing nothing of whether he'd found the flaw that had silenced the clock for a decade.
Impossible to understand or interpret; enigmatic.
দুর্বোধ্য — যা বোঝা যায় না
In 1885, Hermann Ebbinghaus measured how fast memory decays — exponentially, losing most of what you learn within days. But each well-timed review makes the memory more stable, so it decays slower. The FSRS algorithm computes that exact moment for every word you learn, and schedules the review right before you'd forget.
Ebbinghaus, H. (1885). Über das Gedächtnis — replicated by Murre & Dros (2015), PLOS ONE.
A century of memory research reduces to three steps. Here’s what happens to a single word — say, ephemeral — inside NeuroVocab.
“The fireworks were dazzling but ephemeral — gone before she could even point.”
A word tied to a scene, an emotion, or a narrative forms far richer neural connections than a word on a flashcard. So NeuroVocab never shows you a bare definition first — every word arrives inside a story episode.
each successful review stretches the next interval
Memory decays on a predictable curve — and each well-timed review slows that decay. The FSRS algorithm computes the exact day each word would slip away, and schedules your review just before it does.
pulled from memory — not reread
Testing yourself builds dramatically stronger memory than rereading — the testing effect. So reviews are never passive: seven practice modes force you to pull the word out of memory, from different angles.
Each recall strengthens the memory, so the next review comes later — until the word no longer needs reviewing at all. That’s the whole system.
The testing effect: retrieving a memory strengthens it far more than re-reading it (Karpicke & Blunt, 2011, Science). Each mode sits at a different point on the recognition → recall spectrum, so every word is rebuilt from multiple angles — never just memorized.
Recognition memory — the gentlest cue. Ideal for words still in early-stage encoding.
Associative memory — binds word and meaning through relational pairing.
Orthographic encoding — producing the word form strengthens the phonological store.
Free recall with no scaffolding — the classic, high-effort retrieval cue.
Partial-cue retrieval in context — forces reconstruction, not recognition.
Time pressure removes the feel-of-knowing illusion — only true retrieval survives.
Error-weighted repetition — your weakest words automatically get the most reps.
Sessions blend modes automatically — easy retrieval for fresh words, deep retrieval for fading ones. The harder the recall, the stronger the trace it leaves.
Email or Google — takes about 30 seconds.
Barron's 333 · Business English · Academic Words
Your first words begin encoding into long-term memory. FSRS starts the clock.
Email or Google — takes about 30 seconds.
Barron's 333 · Business English · Academic Words
Your first words begin encoding into long-term memory. FSRS starts the clock.
Every day you wait, the curve resets. Your first episode takes three minutes — and your brain does the rest on schedule.
Free · No credit card · English ↔ বাংলা