Retrieval practice: memory is not application.

What remains supported

A 2021 systematic review of applied retrieval-practice research in schools and classrooms included 50 experiments and found that retrieval practice generally benefited student learning across a wide range of classroom conditions. Brali therefore keeps active recall as a legitimate method for retention rather than demoting it because newer papers found boundaries.

Boundary 1: the testing effect is not one clean phenomenon

Gereau and Mulligan published a 2026 meta-analysis focused on free-recall testing effects. They separated direct effects, better memory for tested material, from designs that can also contain forward effects on material learned after a test. In their included effect set, 66% came from mixed designs and 34% from designs capable of isolating the direct effect. Mixed designs also produced larger effects. Their conclusion is not that retrieval practice does nothing; it is that a broad “testing effect” label often collapses distinct outcomes and mechanisms.

Boundary 2: remembering a rule is not applying the rule

Ophuis-Cox, Catrysse, and Camp tested this distinction in a 2026 classroom study with 105 fourth-grade pupils learning Dutch verb-spelling rules. Adding retrieval practice to worked examples improved long-term retention of the rules after one week. But the combined group did not outperform the worked-example group on correct rule application. That is one specific study, not a universal verdict on procedural learning, but it gives Brali a useful product rule: if application is the goal, application needs its own practice and measurement.

What this changes in Brali

• The old Active Recall page is replaced with a sourced protocol that separates retention from application.
• The protocol is classified under Memory and Skill Learning, with Retrieval Practice as a Method.
• Unsupported percentage gains and arbitrary session lengths are removed.
• Two 2026 research candidates are moved to challenge-existing because they narrow how Brali should describe retrieval-practice evidence.
• AI integrations should name the target outcome: retention of tested material, later learning, or application. They should not borrow one broad “testing effect” claim for all three.

A practical rule for recommendations

If a user wants to remember vocabulary, definitions, concepts, or rules, retrieval practice is a reasonable candidate. If the user wants to solve equations, speak a language, write code, diagnose a fault, or apply a rule under realistic conditions, retrieval can still help with accessible knowledge, but the recommendation should also include practice that requires the actual performance.

Sources

• Agarwal PK, Nunes LD, Blunt JR. Retrieval Practice Consistently Benefits Student Learning: a Systematic Review of Applied Research in Schools and Classrooms. Educational Psychology Review. 2021. Supports retrieval practice as a useful classroom learning strategy across the reviewed applied experiments; it does not imply identical effects for every task or transfer outcome.
• Gereau AD, Mulligan NW. Research on the testing effect routinely conflates direct and forward testing effects: A meta-analysis of testing effects with free recall. Journal of Experimental Psychology: Learning, Memory, and Cognition. 2026. Supports separating direct and forward testing effects rather than treating the literature as one coherent testing-effect estimate.
• Ophuis-Cox FHA, Catrysse L, Camp G. Mixed effects of retrieval practice on the acquisition of procedural knowledge in verb spelling education. Learning and Instruction. 2026;104:102377. Supports separating retention from application in the studied classroom context.