Desirable difficulty
Making learning harder in the right ways makes it last longer.
What it means
A desirable difficulty is a learning condition that slows acquisition and feels harder in the moment but produces more durable, flexible, and transferable long-term retention. Spacing practice, interleaving topics, varying conditions, and testing rather than rereading are canonical examples that impose beneficial effort during encoding and retrieval. The difficulty must be 'desirable' — within the learner's capability — since unproductive struggle without the resources to overcome it simply impairs learning. The principle exposes a treacherous gap between performance during study and actual learning, because fluent, easy practice breeds overconfidence and weak retention. It unifies the spacing, testing, generation, and interleaving effects under one organizing idea.
Why the effort pays off
The gains trace to how memory is strengthened by use. Retrieving a fact rather than rereading it forces the brain to reconstruct the memory trace, and each successful reconstruction makes the route back to it more reliable. Spacing works because a memory partly fades between sessions, so each return demands more effortful reconstruction and re-encodes the material against a changed context, leaving more retrieval cues. Interleaving forces the learner to discriminate between problem types and choose a strategy, not merely execute a rehearsed one, which is exactly what a later test or the real world requires. In each case the momentary struggle is the mechanism, not a side effect: easy, fluent practice leaves little to reconstruct and so encodes little that lasts.
What the evidence shows
The component effects are among the better-replicated in the learning literature, but their sizes vary. Dunlosky and colleagues' 2013 review rated practice testing and distributed practice as the two highest-utility techniques of the ten they examined, well above rereading and highlighting. Interleaving is more conditional. Brunmair and Richter's 2019 meta-analysis of 59 studies found a moderate overall benefit, Hedges' g near 0.42, but it was large for visually similar categories such as paintings, small for mathematics, and reversed — blocking won — for learning word pairs. So 'make it harder' is not a universal law. The benefit depends on what has to be told apart: where items look alike and must be discriminated, mixing them helps most; where the added difficulty gives nothing to discriminate, it can simply cost.
When difficulty stops being desirable
The word 'desirable' is doing real work. A difficulty helps only if the learner has the resources to overcome it; past that point it becomes noise. Novices are the clearest case. A 2025 study by Hwang found that interleaving vocabulary harmed low-achieving adolescents, who needed an initial block of massed practice to form basic form-meaning links before mixing paid off. Cognitive-load research points the same way: when material already has many interacting parts, added difficulty can push total load past what working memory handles, and learning collapses rather than deepens. Stacking difficulties is also not additive — two together are not reliably better than one and can interfere. The working rule is to build a foundation blocked and easy, then introduce difficulty as competence grows.
The performance-learning trap
The most useful and least intuitive part of the idea is that how well you perform while studying poorly predicts how much you will retain. Soderstrom and Bjork's 2015 review catalogues manipulations that lift practice performance yet leave long-term learning flat or worse, and others that depress it while improving retention. Because effortful methods feel unproductive in the moment, learners and teachers systematically prefer the fluent ones that show quick gains and fade fast: rereading feels like mastery, self-testing feels like failing. This gap between the felt sense of learning and its reality is why desirable difficulties are so hard to adopt by choice — the method that works is the one that feels worse while you use it. Judging learning by in-session ease reliably picks the weaker strategy.
Examples
Quizzing yourself across several spaced sessions feels harder than rereading once, yet yields far better retention weeks later.
Trying to recall the word before the app reveals it feels like failing; letting it flash the answer feels smooth and expert. The fumbling is the part you will still know next month.
A guitarist who interleaves three songs in one practice session fumbles more than one looping a single riff for an hour — and plays all three better at the gig a week later.
A surgical trainee who practices several procedures in random order fumbles more than one repeating a single technique all afternoon, but performs each more reliably when live cases arrive unpredictably.
A pilot who rehearses emergency drills spaced weeks apart and in scrambled order feels rusty each session, yet reacts faster and more correctly when a real malfunction hits.
First described in Robert A. Bjork (1994).
Key references
- Hwang, H.-B. (2025). Undesirable difficulty of interleaved practice: The importance of initial blocked practice for declarative knowledge development in low-achieving adolescents. Language Learning, 75(1). doi.org/10.1111/lang.12659
- Brunmair, M., & Richter, T. (2019). Similarity matters: A meta-analysis of interleaved learning and its moderators. Psychological Bulletin, 145(11), 1029-1052. doi.org/10.1037/bul0000209
- Soderstrom, N. C., & Bjork, R. A. (2015). Learning versus performance: An integrative review. Perspectives on Psychological Science, 10(2), 176-199. doi.org/10.1177/1745691615569000
- Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4-58. doi.org/10.1177/1529100612453266
- Rohrer, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481-498. doi.org/10.1007/s11251-007-9015-8
- Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 185-205). MIT Press. direct.mit.edu/books/edited-volume/3931/chapter/164557/Memory-and-Metamemory-Considerations-in-the