Retrieval-based JOL

Base your confidence rating on whether you can recall the answer, not on whether the material feels familiar.

Why it works

Fluency-based JOLs ask "does this feel familiar?" — a question working memory answers with misleading ease. Retrieval-based JOLs ask "can I produce this answer independently?" — which requires accessing long-term memory. By forcing recall before rating, the learner receives a direct sample of the retrieval process that the JOL is supposed to predict, rather than relying on a feeling that is only loosely coupled to recall.

How to do it

1. Before rating an item, close the material and attempt free recall.
2. Rate confidence based on the quality of recall (fluent? effortful? partial? absent?).
3. Never rate based on recognition or familiarity alone.
4. Score partial recall lower than complete recall even if the partial version felt confident.

Evidence

Retrieval-based JOL calibration is supported by research showing that testing improves both JOL accuracy and subsequent retention — the retrieval attempt provides both a better calibration signal and additional encoding. (observational)

Retrieval-based JOLs are more accurate but also more effortful; in time-pressured study, learners often revert to familiarity-based ratings. The practice requires deliberate habit.

Sources

• Roediger & Karpicke (2006), "Test-enhanced learning," Psychological Science

Common mistake

Rating "high" because the material looks familiar without testing whether it can be produced — then being surprised when it cannot be recalled at exam time despite confident study ratings.