A list of words is shown, the screen goes blank, and you write down whatever you can. This is free recall, and its appeal is that the pattern of what people remember is not flat. Plot recall against the position a word occupied and you get a curve with a raised start and a raised end — the serial-position curve, one of the most dependable findings in memory research. Words from the start receive more rehearsal, which carries them toward longer-term storage: the primacy effect. Words at the end were seen most recently and may still be in immediate memory: the recency effect. The middle enjoys neither advantage, which is why it is most often lost.
Twelve words appear one at a time, each for two seconds, in a freshly shuffled order every run. When the last has gone, a text box appears and you type as many as you can remember, in any order. Scoring is mechanical and strict: your entry is split on commas, semicolons and line breaks, each piece is lower-cased, and it counts only if it exactly matches a word that was presented. Order of typing is irrelevant, and the score is the number of presented words produced, out of twelve. Alongside it the tool draws a bar for every presentation position — first word on the left, twelfth on the right, recalled positions marked green — and groups them into primacy for positions 1–4, middle for 5–8 and recency for 9–12.
Suppose the shuffle produced an order in which you recalled the items that had appeared in positions 1, 2, 3, 7, 10, 11 and 12. Typed back in whatever order you liked, seven matched, so the score is 7 / 12. The blocks then split as follows: primacy covers positions 1–4 and finds three, so 3/4; middle covers 5–8 and finds only position 7, so 1/4; recency covers 9–12 and finds three, so 3/4.
The strictness matters more than it sounds. If the list contained river and you typed rivers, no credit — the comparison is on the exact lower-cased string. If you remembered candle and garden but typed them as candlegarden with no separator, that whole token matches nothing and both words are lost.
This is an educational demonstration of a classic paradigm, not a memory-clinic assessment, and no result from it should be read as evidence about anyone's memory or health.
They came too late for the extra rehearsal that helps the opening items and too early to still be in immediate memory, so they are squeezed from both sides.
No difference — the scorer checks membership, not order.
Almost always spelling or separators. Matching is exact, so a plural where the list had a singular, or two words typed with no comma between them, both fail.
There are no norms behind this page, and the task is easier than a laboratory free-recall test because recall begins immediately, so compare your own runs rather than a scale.