Retrieval practice and spaced retrieval improve achievement and retention by medium amounts, while interleaving helps category learning only for some materials and blocking beats it for words¶

Claim · All claims
Evidence · 3 studies · 3 quant-synthesis r? · q4 · i2 medium
Currency · newest study 2021 · 1 of 3 entries checked against their sources (automated) · not reviewed by a person

Introducing certain effortful conditions during learning — such as retrieval practice, spacing, and interleaving — can slow visible performance during instruction while improving long-term retention and transfer. The claim is about desirable difficulties specifically: conditions that are effortful but surmountable, not obstacles that exceed the learner's capacity.

Subclaims¶

q4 i2 Across 222 classroom studies with 48,478 students, practice testing (quizzing) raised academic achievement by a medium amount compared with restudying and other strategies, g = 0.499. → Yang et al. 2021

q4 i2 Spacing retrieval practice out over time produced much better final retention than massing it (g = 0.74, 29 studies), but expanding intervals were no better than uniform ones (g = 0.034). → Latimier et al. 2020

q4 i2 Interleaving produced a moderate overall benefit for inductive (category) learning (g = 0.42, 59 studies), but it depended on the material: blocking beat interleaving for words (g = -0.39), and results for expository texts were ambiguous. → Brunmair & Richter 2019

Evidence¶

Yang et al. 2021¶

Yang, C., Luo, L., Vadillo, M. A., Yu, R., & Shanks, D. R. (2021). Testing (quizzing) boosts classroom learning: A systematic and meta-analytic review. Psychological Bulletin, 147(4), 399–435. doi:10.1037/bul0000309

q4 · systematic review and meta-analysis · i2 · medium effect, g=0.499 · n=222 studies (48,478 students) · quant-synthesis · r?

A systematic review and meta-analysis of the testing effect in real classrooms rather than the laboratory, pooling 222 independent studies with 48,478 students. Taking quizzes or practice tests on studied material (retrieval practice) raised academic achievement by a medium amount compared with restudying and other learning strategies. The size of the benefit varied with what the control group did, whether the practice and final tests had the same format, whether corrective feedback was given, how many times learners were tested, and the study design.

Latimier et al. 2020¶

Latimier, A., Peyre, H., & Ramus, F. (2021). A Meta-Analytic Review of the Benefit of Spacing out Retrieval Practice Episodes on Retention. Educational Psychology Review, 33(3), 959–987. doi:10.1007/s10648-020-09572-8

q4 · meta-analysis (robust variance estimation) · i2 · medium-to-large effect, g=0.74 · n=29 studies (39 + 54 effect sizes) · quant-synthesis · r?

A meta-analysis of 29 studies on spaced retrieval practice, which the authors describe as combining two "desirable difficulties": retrieval practice and spacing. Spreading repeated retrieval of the same content over time gave a strong retention advantage over massing it (g = 0.74, 39 effect sizes). Expanding the gaps between retrievals was no better than keeping them even (g = 0.034, 54 effect sizes), although expanding schedules helped more when items were tested more often.

Brunmair & Richter 2019¶

Brunmair, M., & Richter, T. (2019). Similarity matters: A meta-analysis of interleaved learning and its moderators. Psychological Bulletin, 145(11), 1029–1052. doi:10.1037/bul0000209

q4 · multilevel meta-analysis · i2 · medium effect, g=0.42 · n=59 studies (158 samples, 238 effect sizes) · quant-synthesis · r?

A multilevel meta-analysis comparing interleaved with blocked presentation of items for inductive (category) learning, across 59 studies. Interleaving helped by a moderate amount overall, most for paintings and other visual materials (g = 0.67) and less for mathematical tasks (g = 0.34). It qualifies the claim: for word materials blocking beat interleaving (g = -0.39), and effects for expository texts and tastes were not significant. Interleaving helped more when categories were similar to each other and items within a category were less similar, so the difficulty is desirable only for some materials.

Studies bearing on this claim¶

When the outcome is retention at a delay, should effortful practice be reintroduced once the material is understood, and for which material? This table gathers the studies behind this claim and the 3 narrower claims that bear on it: 5 distinct studies (3 quant-synthesis, 1 theoretical, 1 review), q2–q4, 3 of 5 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

The difficulty must be surmountable. Desirable difficulties work because effortful retrieval and discrimination trigger deeper processing and strengthen memory traces — but only when learners can actually overcome the difficulty. If the challenge exceeds available working-memory resources, it becomes an undesirable difficulty and performance degrades, consistent with Cognitive overload degrades learning. and Chunking reduces working memory load. The designer's task is to keep difficulty within the learner's zone of capability, which is the same constraint that governs Cognitive load management and underpins Cognitive load reduction improves learning. — reducing extraneous load is precisely what frees capacity so that the intrinsic difficulty of retrieval and discrimination can be productive rather than overwhelming.

Expertise reversal. The same difficulty that helps novices can hinder advanced learners. As expertise grows, conditions that force extra effort on already-mastered material become redundant or counterproductive — see Expertise reversal effect and Cognitive load theory. Difficulty should be faded or re-targeted as competence develops, not held constant.

Transfer appropriateness. A difficulty is most desirable when it mirrors the conditions of the eventual test or use: practicing under varied, delayed, or interleaved conditions prepares learners for retrieval situations that blocked, immediate practice does not. Feedback must still be available so that errors made under difficulty are corrected rather than reinforced.

Fluency misjudgment. Learners judge their learning by momentary fluency, so they often prefer easier strategies that produce worse long-term outcomes — and may resist conditions that feel harder. Designers should expect short-term performance dips, communicate the rationale for effortful practice, and use assessments that reward delayed retention rather than immediate performance; productive struggle connects to Cognitive disequilibrium motivates conceptual change.

Evidence status. No specific studies are catalogued in the Evidence section yet; the framework originates with Robert Bjork's work on desirable difficulties, and key experimental and meta-analytic support (particularly for retrieval practice, spacing, and interleaving) still needs to be added before this claim can be rated.

Open questions. How much difficulty remains desirable as expertise grows, and how difficulty should be calibrated dynamically over the course of instruction, remain open design questions. Adaptive approaches that adjust challenge to current learner state — see Adaptive learning improves outcomes. — are one candidate mechanism, but their fit with deliberately under-scaffolding practice has not been established here.

Merged from “Desirable Difficulties Improve Retention” (desirable-difficulties-improve-retention): The core trade-off. The claim hinges on a dissociation between performance during learning and retention after it. A condition that produces fluent, error-free practice often feels effective and can look effective on an immediate test, yet yields poorer delayed retention than a condition that slows learners down [+S]. This is why "desirable" must be defined relative to a delayed retention or transfer measure, not to learner satisfaction or immediate performance. The distinction maps onto the broader contrast between conditions of learning and conditions of performance described in cognitive load theory.

What makes a difficulty desirable. The difficulties in question are those that force additional encoding operations — retrieval from memory, generation of an answer, comparison across contrasting cases, or re-engagement with material after partial forgetting. Difficulty that merely consumes working memory without triggering those operations (e.g., unnecessarily awkward formatting, split attention, or search that yields no answer) is not covered by this claim and is better treated through cognitive load reduction [-S]. In this sense the claim is the productive counterpart of active learning: generation and production during instruction are common sources of desirable difficulty.

Boundary conditions. Three moderators are likely to matter. First, the difficulty must be surmountable: if learners fail to reconstruct or generate the target content, the extra effort produces incomplete encoding rather than stronger memory [~M]. Second, learner expertise moderates the effect — for novices, high intrinsic load plus an added difficulty can push total load past working-memory capacity, degrading learning (see cognitive overload degrades learning and the expertise reversal effect) [~S]; for more advanced learners, the same difficulty may be beneficial or even necessary. Third, learners' preferences are an unreliable guide: conditions that feel harder are often the ones that produce more retention, so perceived difficulty and actual learning can diverge in either direction [~M] — a direct challenge to designs that optimize for learner satisfaction or fluency alone.

Open questions. The dose–response relationship (how much difficulty is optimal), the interaction between different difficulty types (e.g., spacing combined with interleaving), and the durability of effects across domains and age groups all remain to be established with evidence on this page. Until evidence entries are added, this page should be treated as a framing claim whose subclaims (e.g., for retrieval practice, spacing, and interleaving individually) will each carry their own evidence.

Learner Variables¶

• Long-Term Memory — outcome: instruction changes it

Related Claims¶

• Cognitive overload degrades learning. — difficulty beyond working-memory capacity is undesirable, not desirable
• Chunking reduces working memory load. — managing load keeps difficulty surmountable
• Cognitive load reduction improves learning. — reducing extraneous load makes room for productive intrinsic difficulty
• Cognitive disequilibrium motivates conceptual change. — productive struggle can drive conceptual change
• Adaptive learning improves outcomes. — dynamic calibration of difficulty is one way to keep challenges desirable
• Cognitive load management — load management is the complement to adding desirable difficulty
• Expertise reversal effect — the benefit of difficulty reverses as expertise grows
• Active learning improves exam performance. — generation and production during instruction are a common source of desirable difficulty
• Feedback Enhances Retrieval Practice — related
• Retrieval practice improves long-term retention — possibly the same claim (merge candidate)
• Retrieval Practice Improves Transfer — related
• Retrieval Failure Reduces Benefit — related
• Interleaving category examples improves inductive category learning for visual and mathematical materials, but not for expository texts or word categories — related
• Faster rate of learning may be negatively related to long-term retention (efficiency-effectiveness trade-off) — related
• Spaced Retrieval Improves Retention — related
• Spacing out retrieval practice produces better long-term retention than massing it — related
• Desirable difficulty: harder-to-learn conditions yield enhanced long-term retention — related
• Spaced retrieval practice outperformed both massed retrieval practice and no-practice control on delayed financial knowledge, while massed and control did not differ — related
• Productive difficulty at the edge of mastery, with high expectations, fosters deeper learning — related
• Spaced and interleaved homework practice improves learning and retention — related
• Struggling to encode, retrieve, and reorganize information produces deeper comprehension and retention, and AI-supplied ready-made solutions short-circuit these processes — a narrower finding that bears on this claim
• Classroom testing (quizzing) significantly boosts student academic achievement to a medium extent (g = 0.499). — possibly the same claim (merge candidate)
• The meta-analysis supports retrieval effort as a contributor to the testing effect — a narrower finding that bears on this claim
• Retrieval practice enhances learning across domains and the lifespan in applied educational settings — a broader claim this one bears on
• Testing benefits retention over a conservative restudy control, establishing the testing effect as robust — possibly the same claim (merge candidate)
• Interleaved practice produces higher mathematics test scores than blocked practice in a classroom experiment — related
• Active control over the learning experience enhances recognition memory from age 7 into adulthood — related
• Interleaved practice improves delayed posttest performance compared with blocked practice — related
• Retrieval practice produces better retention than restudying, with meta-analytic reviews reporting people 2.5 times as likely to remember tested information and medium-to-large classroom effects — related
• Group testing shows advantages in some studies but mixed results overall for long-term retention and transfer — a narrower finding that bears on this claim
• A systematic review of 50 classroom experiments found retrieval practice yields medium-to-large benefits in most cases, but only 6% were conducted in non-WEIRD countries — related
• Spacing content over time and interweaving different content strengthens learning and retention — related
• Interleaving benefits are small for mathematical tasks and absent or reversed for expository texts and words — related
• Spaced recall quizzing significantly improves overall post-test performance on fundamental mathematics without harming learning of current material — a narrower finding that bears on this claim
• Interleaving during concept learning benefits memory of studied items, with effect sizes (Hedges' g) of up to 0.65 — related
• Some participants used external memory aids such as notes or web sources during practice, and Experiment 2 analyses were checked excluding these participants. — related
• Deeper processing of target language improves lexical retention and long-term learning — related
• Interleaved practice outperformed blocked practice across eight STEM classrooms — related
• Strategies that reduce fluency and increase perceived difficulty (e.g., self-testing, spaced practice) often yield superior long-term learning, contrary to student beliefs — related
• Interleaved practice yields better category learning than blocked practice, and students who choose interleaved practice outperform those who choose blocked practice on novel materials — related
• Feedback likely supported both consolidation and rule induction in the observed retrieval practice benefits — a narrower finding that bears on this claim
• In applied settings, the beneficial effect of random practice on retention is almost negligible, as the competing meta-analysis itself reports — related
• Retrieval practice improves delayed but not immediate performance, consistent with a desirable-difficulties explanation — related
• Images during retrieval practice enhanced feelings of competence but not enjoyment — related