Updated on
September 1, 2026
Desirable difficulties are spacing, retrieval, interleaving, variation and reduced feedback used to improve later recall or transfer. This teacher guide shows when effort helps and when it becomes an access barrier.
What are desirable difficulties?
Desirable difficulties are learning conditions that can reduce fluency during practice but improve later recall or transfer. They include spacing, retrieval, interleaving, varied practice and reduced feedback. The challenge is useful only when learners have the needed prior knowledge, can make a successful attempt and become more independent on a delayed or changed task.
The term desirable difficulties refers to learning conditions that make practice feel harder now but improve later recall or transfer. Examples include spacing practice, retrieving information, interleaving related problems, varying examples and reducing feedback once a learner is ready. The difficulty is not valuable by itself. It must lead to more durable or flexible learning.
Soderstrom and Bjork (2015) explain why learning and current performance must be judged apart. A smooth lesson can produce strong immediate performance without secure learning. An effortful task can produce more errors during practice yet lead to better performance after a delay (Soderstrom and Bjork, 2015).
For teachers, the key question is not “How can I make this harder?” It is “Will this effort help these learners remember, choose or apply the idea later?”
For example, a Year 8 science teacher can revisit particle models after two days. Learners draw the model from memory before they see the diagram again. The brief delay makes recall harder, and the later explanation shows whether the idea has lasted.
The Bjorks use the term for conditions that can impair performance during learning while improving later retention or transfer. Their account names five recurring manipulations: spacing rather than massing, interleaving rather than blocking, varying the conditions of practice, using tests rather than repeated presentations and making feedback intermittent rather than continuous (Bjork and Bjork, 2011).
A Year 5 teacher can model equivalent fractions, guide two examples, then revisit the idea two days later among non-equivalent fractions. Learners must retrieve the rule and decide which comparison applies. The delay and mixed set are useful only if the earlier teaching gave them a sound starting point.
Performance is what a learner can do during a lesson. Learning is a relatively lasting change that remains after time has passed or support has changed. The two measures can move apart. Re-reading may make an answer feel fluent, while retrieval exposes gaps and strengthens access to the answer.
In one experiment, learners who repeatedly studied a passage performed well during practice. Learners who retrieved it performed less well at first but remembered more after a delay (Roediger and Karpicke, 2006). The experiment used prose that learners had already studied and did not test every form of classroom quizzing. In teaching, provide correction when an error would otherwise persist.
A Key Stage 3 history teacher can see the difference by asking for causes of the English Civil War at the end of the lesson and again the following week. A strong same-day answer shows performance with recent cues. The delayed answer provides better evidence about what can still be retrieved and connected.
A task is desirable when the learner can engage with the intended thinking and later becomes more independent. The same task can be desirable for one learner and unhelpful for another. Prior knowledge, task clarity, working-memory demands, feedback and anxiety all affect the result.
Use three tests before adding challenge:
If a Year 7 learner cannot identify numerator and denominator, mixing several fraction operations is premature. The teacher first explains and models the component knowledge. Interleaving becomes useful when the learner can perform each operation and now needs practice choosing among them.
Start with a small change. Add a short delay before recall, mix two known problem types or remove one prompt. Keep the rest of the task stable. This lets the teacher see what the added demand changes.
If accuracy falls but the later check improves, the effort may be useful. If learners cannot start, restore support and teach the missing part.
Feelings are useful evidence but not the final test. A learner may say that rereading feels easy because the text is familiar. Another may say that retrieval feels hard because the answer is not in view.
Ask both learners to try a short delayed check. The result helps the teacher judge whether the effort led to learning.
A Year 8 science class has learned particle models for solids, liquids and gases. The teacher wants learners to explain unfamiliar changes of state rather than repeat one rehearsed diagram.
Errors during the mixed set are information, not proof that the sequence has failed. The teacher looks for method selection and reasoning. If most learners are guessing, the next step is renewed modelling, not more confusion.
Spacing spreads encounters with an idea across time. A large research review found a robust benefit for distributed over massed practice, while also showing that the useful gap depends on the final retention interval (Cepeda et al., 2006). There is no universal spacing calendar.
A primary teacher can use spaced practice to revisit multiplication facts in brief sessions across a fortnight. The class retrieves previous facts before new examples. The teacher increases the interval only when accuracy remains high enough for learners to think rather than guess.
Retrieval asks learners to reconstruct an answer from memory. It can reveal what is missing and make later access more reliable. It is not a cold test of untaught material, and it should not become a public comparison of learners.
An English teacher uses retrieval practice by asking learners to list three features of a persuasive argument before reopening their notes. Learners then check, correct and add to the list. The attempt creates evidence for the teacher and a reason to attend to the missing feature.
Interleaving mixes related categories or problem types. Its main value is discrimination: learners must notice which method or concept fits. A meta-analysis found that effects vary substantially by material. Visual categories and similar problem types often benefit, while expository texts show much less consistent benefit (Brunmair and Richter, 2019).
A mathematics teacher first teaches area of triangles and parallelograms clearly. A later interleaved practice set removes the method labels. Learners identify the shape, select the formula and explain the choice before calculating.
Variation can help learners see which features matter and which are incidental. Change one useful dimension at a time. If the representation, vocabulary, method and context all change together, the task may measure adaptation to the format rather than the intended concept.
A geography teacher varies maps of river features by scale and orientation while keeping the identification criteria explicit. Learners annotate the evidence used for each decision. A short formative assessment shows when models can be reduced. The goal is flexible recognition, not novelty for its own sake.
Continuous immediate feedback can support early success but also make performance dependent on the prompt. Once the learner can attempt the task accurately, delay or reduce feedback so they must monitor the process. Restore support when errors show a missing concept rather than a useful retrieval gap.
A music teacher first gives immediate feedback on a short rhythm. On later attempts, the learner records the phrase, evaluates it against two criteria and receives teacher feedback after the whole phrase. The delay tests metacognitive monitoring without withholding essential teaching.
Use the mini app to check that the planned challenge targets learning rather than adding avoidable strain. Name the knowledge learners already have, the effort you want and the support that stays in place. Then define a delayed or changed check. The tool creates a printable plan for one lesson sequence.
Define the learning, check the foundation and name the later evidence before adding challenge.
Desirable difficulties is a framework for designing and interpreting practice, not a promise that every harder task will improve learning. Evidence is strongest for specific manipulations under defined conditions. It becomes weaker when several practices are bundled together and measured only through immediate classroom performance.
A large field experiment in mathematics found that interleaved assignments improved an unannounced short-term test by 0.28 standard deviations. It did not improve cumulative end-of-year assessments (Kremer et al., 2023). The result is useful because it shows both promise and a boundary. A short-term advantage should not be reported as a permanent or universal effect.
First, record what learners can do before the change. A short response, worked problem or oral account is enough. Second, note what happens during practice. Count the errors that show a wrong choice, not every slip.
Third, repeat a close task after a useful delay. Keep the check brief so it can guide the next lesson.
A Year 9 teacher who mixes equation types can compare three pieces of evidence. The first blocked set shows that learners know each method. The mixed set shows whether they can choose.
A short quiz one week later shows what remains. If the mixed set feels hard but the delayed choice improves, the difficulty has earned its place.
Do not rely on class average alone. Check whether some learners gain while others lose access to the task. A simple tally can show who still needs a prompt, who can work alone and who has misunderstood the core idea. This turns “productive struggle” from a slogan into a testable plan.
Maintain the cognitive goal while removing extraneous difficulty. A learner can retrieve an idea orally, select from accessible symbols or use assistive technology without making the thinking easier. The target is independent access to knowledge, not endurance of a particular format.
For a learner with weak reading fluency, a science retrieval task can be read aloud while the learner explains the particle model. For a learner who needs more processing time, reduce the number of mixed examples without labelling the method. In both cases, the learner still retrieves or selects the concept.
Use cognitive load theory and guidance on working memory to separate necessary thinking from avoidable demands. Use scaffolding when prerequisite knowledge or task access is insecure. Return to explicit instruction when the foundation is missing. Fading support should follow evidence, not a fixed timetable.
Most poor uses of desirable difficulties add strain without a clear learning gain. They mix untaught content, remove all support, treat errors as proof of effort or check only the end of the lesson. A sound plan names the target, keeps access in place and uses later evidence to decide whether the challenge should stay.
A department reviewing a new mixed-practice worksheet should begin with one question: what decision must learners make that they did not have to make in the blocked version?
The ideas overlap, but desirable difficulty has a specific learning test. The effort must improve later retention or transfer. A struggle that produces confusion without later improvement is not desirable.
Retrieval should require recall without making success remote. If most learners cannot begin, restore cues, reteach the foundation or shorten the delay.
Delay or reduce feedback after learners can attempt the process accurately enough to monitor it. Give timely correction when an error is likely to persist or when learners lack the knowledge to evaluate their answer.
Blocked examples can help during initial teaching. Interleaving becomes useful when learners know the component methods and need to choose between them.
Kremer, M., Gray-Lobe, G., de Laat, J. and van der Haar, L. (2023). A year of desirable difficulties: The impact of interleaving math practice in Nigeria. National Bureau of Economic Research Working Paper.
Bjork, E. L. and Bjork, R. A. (2011). Making things hard on yourself, but in a good way: Creating desirable difficulties to enhance learning. In M. A. Gernsbacher et al. (eds), Psychology and the Real World.
Brunmair, M. and Richter, T. (2019). Similarity matters: A meta-analysis of interleaved learning and its moderators. Psychological Bulletin, 145(11), 1029-1052.
Cepeda, N. J., Pashler, H., Vul, E., Wixted, J. T. and Rohrer, D. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354-380.
Roediger, H. L. and Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249-255.
Soderstrom, N. C. and Bjork, R. A. (2015). Learning versus performance: An integrative review. Perspectives on Psychological Science, 10(2), 176-199.
Next lesson, choose one important idea and add a delayed check before adding any other difficulty.
