Free recall
Pull what you have learned out of memory with the source closed, then check it against the source.
Variants
- default
- Mark how sure you are of each item
- Answer numbered prompts (also a pretest before first study)
Data fields
The fields a spec's page can set for this component (see switchback show free-recall --json):
topicprompts
Grounding
Helps: when studied material has to last days or weeks: writing out what you remember with the book closed, then checking it against the source in a second colour, beats rereading for the same time on a later test, by a small to moderate margin in this free-recall format, though not on a test minutes away
Backfires: when so little comes back that there is nothing to check, or when the check pass is skipped or scored loosely: errors recalled without correction tend to persist, and people marking their own recall credit wrong answers unless they check it idea by idea
- ARetrieving studied material from memory (for example, writing down everything you can recall with the book closed) improves retention at a delay of days or more, compared with spending the same time restudying it.Rowland (2014), Adesope et al. (2017), Pan & Rickard (2018), Roediger & Karpicke (2006)
- BChecking recall against the source right after retrieving it increases the retention benefit and corrects errors that would otherwise persist. Without feedback, errors produced during recall tend to be retained.Pashler et al. (2005), Butler & Roediger (2008), Butler, Karpicke & Roediger (2008), Dunlosky et al. (2011)
- BErrors a person made with high confidence are more likely to be corrected on a later test, once feedback is seen, than errors made with low confidence.Metcalfe & Finn (2011), Butler, Fazio & Marsh (2011), Metcalfe (2017)
- BTrying to answer questions about material before studying it, even when most answers are wrong, improves later retention of that material compared with studying it without a pretest.Richland, Kornell & Kao (2009), Kornell, Hays & Bjork (2009), Pan & Carpenter (2023), Carpenter & Toftness (2017)
- CCoarse, three-level confidence ratings discriminate correct from incorrect judgements above chance (resolution), even though people are typically overconfident in absolute terms. Items marked low-confidence are therefore a useful flag for risk.Fleming & Lau (2014), Moore & Healy (2008), Koriat, Lichtenstein & Fischhoff (1980)