Introduction
Most students judge a study session by how it felt. A smooth session, where the notes made sense and nothing tripped you up, feels productive. A rough session, where you kept blanking and had to check the source, feels wasted. The research points the other way: many of the conditions that make learning feel harder are exactly the conditions that make it last.
Robert Bjork gave this idea a name — desirable difficulties — in a 1994 chapter, and he and Elizabeth Bjork developed it further in a widely cited 2011 chapter on making learning conditions productive. The core claim is that performance during practice and learning that survives to the next week are different things, and that the manipulations which depress the first often raise the second.
Performance Is Not Learning
This is the distinction the whole idea rests on. Performance is what you can do right now, with the material in front of you and the context fresh. Learning is the durable change — what you can still retrieve tomorrow, next month, or in an exam hall where the cues are different.
Rereading a chapter raises performance immediately. The sentences become familiar, you follow the argument without effort, and that ease gets misread as mastery. But familiarity with a text is not the same skill as producing its content from an empty page. When the page is empty, the fluency you built disappears, because it was never yours — it belonged to the text.
The Four Classic Desirable Difficulties
1. Spacing
Spreading study of the same material across separated sessions beats massing it into one block. Between sessions you forget a little, and the effort of recovering what slipped is what strengthens the memory. Cramming produces a high score on tonight's self-check and very little a week later. Read more in spaced repetition.
2. Interleaving
Instead of doing twenty problems of one type, mix types within a session. Blocked practice lets you apply the same method twenty times without deciding which method applies — and deciding is the part the exam actually tests. Interleaving feels messier and produces more mistakes during practice, then transfers better. Read more in interleaving practice.
3. Retrieval and Testing
Pulling an answer out of memory changes the memory. Reviewing does not. A quiz you take before you feel ready is more useful than one you take when you feel ready, because the effortful pull is what does the work. Read more in retrieval practice.
4. Generation
Producing an answer before you see it beats reading the answer, even when your attempt is wrong. Guessing at a definition, sketching an explanation, or predicting the next step in a derivation primes you to encode the correction. The related move is explaining an idea in your own words rather than the author's, which is the mechanism behind the Feynman technique.
Why Fluency Misleads
Every one of these techniques costs you something visible and pays you something invisible. That asymmetry is why students abandon them. You can see that you got six of ten flashcards wrong. You cannot see that you would have got nine of ten wrong next week had you reread instead.
Highlighting is the clearest case. It produces a page that looks studied, takes almost no effort, and leaves you with a strong sense of coverage. What it does not produce is a single occasion on which you retrieved anything. Rereading is the same problem with more time spent.
Useful self-check: if it feels easy, you are probably not learning. Not always — some material really is easy, and early exposure to a new topic should be easy. But when a session feels effortless on material you have not mastered, the ease is almost always coming from the source being in front of you.
The "Desirable" Caveat
Difficulty is not the goal; it is a side effect of the conditions that produce durable memory. A difficulty is desirable only if the learner can still succeed at it, at least some of the time. Push a beginner into interleaved problems before they can execute any single method and they do not build flexible knowledge — they build confusion and quit.
The practical rule: difficulty should be something you can overcome with effort, not something you can only fail at. If you are getting nearly everything wrong, the fix is not more grit. It is a smaller step — go back to worked examples, reduce the chunk size, or shorten the gap between reviews until you are succeeding often enough for the effort to have something to strengthen.
How This Maps to LearnCoachAssist
The four difficulties are not abstractions here; each corresponds to a tool you already have.
- Spacing is the SRS. The scheduler exists precisely so that cards come back at the point where recall takes effort rather than at the point where it is comfortable.
- Interleaving is Learning Paths and packs. A path mixes topics rather than running one subject to exhaustion, so you keep having to identify what kind of question you are facing. Browse them at Learning Paths.
- Testing is the quiz. Take it early and often, not as a final check once you feel confident. A quiz used as a measuring device wastes most of its value.
- Generation is explain-it-back mode. You type your own explanation before the reveal, which forces production rather than recognition. That single sequencing choice — your words first, model answer second — is what makes it a generation task instead of a reading task. Try it in explain-it-back study mode.
Calibrating: How Hard Is Hard Enough
The signal to watch is your Again rate — how often you fail a card and it comes back. As a rough rule of thumb, somewhere around 20 to 40 percent Again on review material suggests the difficulty is in a useful band. This is a heuristic for self-calibration, not a validated threshold, and the right number varies with subject, deck maturity, and how you write your cards.
What the extremes tell you is more reliable than the number itself. An Again rate near zero usually means intervals are too short or the cards are too easy — you are reviewing things you already know and paying for it in time. An Again rate above roughly half usually means the cards are overloaded, the material was never understood in the first place, or you are reviewing too late. In that case, fix the cards or add a comprehension pass before touching the schedule.
What to Change This Week
Pick one substitution rather than rebuilding your routine. Replace a rereading session with a closed-book retrieval attempt. Mix two subjects in one sitting instead of blocking them. Write your own explanation before opening the model answer. Each of these will make the session feel worse than the one it replaced. That is the point, and it is also the reason you will be tempted to switch back.
Frequently Asked Questions
Isn't struggling just wasted time if I get the answer wrong?
No — a failed generation attempt still improves how well you encode the correction afterwards, provided you actually see the correct answer and process it. The waste case is different: struggling with no feedback, or struggling with material so far beyond your level that you cannot make a meaningful attempt at all.
How do I tell a desirable difficulty from a bad study habit?
Ask what the difficulty is made of. Effortful recall, deciding which method applies, and producing your own words are desirable. Illegible notes, a noisy room, a badly written card with four facts crammed into it, and unclear instructions are just friction — they cost effort without generating any retrieval.
Should I always avoid rereading?
Not entirely. A first pass has to be a reading pass, and rereading is reasonable for a quick orientation before you start retrieving. The mistake is using rereading as the main study method, or as the last thing you do before an exam, where it produces confidence far out of proportion to what it produces in memory.
Conclusion
Desirable difficulties are an uncomfortable finding because they ask you to trust an outcome you cannot feel while you are producing it. The session that goes badly, where you blank on cards and get problems wrong in a mixed set, is frequently the session doing the most for you. The session that goes smoothly often just confirms that the book is still open.
Set the difficulty high enough that you notice it and low enough that you keep succeeding, then judge your study by what you can retrieve next week rather than by how the last hour felt.