Difficulties, Desirable and Not
Desirable difficulties are conditions that slow visible progress during practice while deepening long-term retention — but a difficulty stops being desirable the moment the learner cannot overcome it. · 10 min
A pattern has been building across this unit, and it is time to name it. Spacing your study feels worse than massing it, yet lasts longer. Testing yourself feels worse than rereading, yet holds better. Shuffling problem types feels worse than blocking them, yet transfers further. In each case the arrangement that struggles during practice wins at the test. Robert Bjork gave this family of arrangements a single name — desirable difficulties — and it comes with a theory precise enough to say when a difficulty helps and when it merely gets in the way. That boundary is the most useful thing in the folio.
Guess before you learn
A tutor gives two students the same material. Student A's practice is arranged so she is right about 90% of the time and it feels smooth. Student B's practice is arranged so he is right about 60% of the time and it feels like a struggle. Both have equal ability. Three weeks later, on a hard test, who tends to score higher?
Student B tends to win. A struggle you can overcome — the right level of hard — builds memory that smooth practice does not. If you chose Student A, you trusted practice accuracy as a gauge of learning; the two come apart, and closing that gap is what this folio is about.
9–12
3–5
Here is a strange truth: the study that feels easy often teaches the least, and the study that feels hard often teaches the most. Reading your notes again feels smooth — and slips away fast. Closing them and testing yourself feels rough — and stays. Scientists call the good kind of hard a desirable difficulty.
But there is a limit. A difficulty helps only if you can get through it. A problem with no clue how to start is not desirable — it is just too hard, and you need an easier step first.
6–8
Desirable difficulties are conditions that make practice harder now but learning stronger later. Robert Bjork named four you have already met: spacing (folio 7), interleaving (folio 9), testing yourself (folio 5), and generating an answer instead of reading it (folio 6). Each slows you down during practice — more errors, less fluency — and each pays off on a delayed test.
The catch is in the word desirable. A difficulty helps only when you have the background to overcome it. Face a problem you have no way to begin, and the difficulty stops being desirable: it just blocks you. The skill is telling the two apart.
9–12
The mechanism runs on a distinction from Bjork and Bjork's theory of memory: retrieval strength — how easily you can call something up right now — and storage strength — how deeply it is learned. Conditions that lower retrieval strength during practice, forcing effortful reconstruction, produce larger gains in storage strength. Easy practice raises retrieval strength briefly; difficult practice builds the durable kind.
This is why performance during practice misleads. Soderstrom and Bjork (2015) draw the line sharply: performance is what you can do while training; learning is what remains later, and the two routinely diverge. High practice performance can signal a difficulty too low to build storage strength — smooth today, gone next month.
K–2
Some hard things are the good kind of hard. When you close the book and try to remember, it feels harder than reading again. But that harder way is the one that helps you keep it.
Not every hard thing is good. If a book is far too hard to read at all, that is just hard. The good kind of hard is the kind you can still do.
Undergrad
The New Theory of Disuse (Bjork and Bjork, 1992) makes the prediction precise: retrieval strength and storage strength are separable, and current retrieval strength is a poor index of storage strength. A condition that depresses momentary performance while increasing the storage-strength gain per successful retrieval is, by definition, a desirable difficulty. Spacing, interleaving, testing, and generation all fit the description, which is why they recur throughout this course.
The boundary is not decorative. A difficulty is desirable only within the range where the learner can still respond successfully; below the knowledge required to succeed, added difficulty subtracts. This is why the same manipulation — reduced feedback, higher variability, interleaving — can help a competent student and harm a novice, and why instruction must fit difficulty to current ability rather than simply maximize it.
Postgrad
Bjork's framework reframes a century of training research: the conditions that most improve long-term retention and transfer frequently depress acquisition-phase performance, so training curves systematically mispredict retention (Soderstrom and Bjork, 2015). The generation effect (Slamecka and Graf, 1978), contextual interference, spacing, and testing are unified as manipulations that trade fluency now for durability later.
The prescription resists maximization. Because desirability is bounded by the ability to overcome the difficulty, optimal instruction is adaptive: introduce difficulties as competence permits, scaffold where success is not yet attainable, and read a dip in practice performance as possible evidence of learning rather than its absence. The mature stance treats present performance and latent learning as distinct dependent variables, not as one reading of the same gauge.
desirable difficulty
A study condition that slows visible progress during practice while improving long-term retention and transfer — desirable only when the learner can overcome it.
Why is this true?
Why can high accuracy during practice be a warning sign rather than good news?
Because accuracy during practice measures retrieval strength — how accessible the material is right now — which fades. Smooth, error-free practice often means the difficulty was too low to raise storage strength, so the memory is not being made durable, only briefly fluent.
Read the crossover carefully, because it is the whole idea. During practice, the easy arrangement is ahead — higher accuracy, less effort, better feelings. At the delayed test, the order reverses. The two numbers measure different things: practice accuracy is present performance, and the delayed score is learning. Bjork's language for it is retrieval strength versus storage strength. Easy practice pumps up retrieval strength, which decays. A difficulty you can overcome converts effort into storage strength, which stays. The uncomfortable consequence is that how well practice is going, right now, is a poor guide to how much you are learning.
Now the boundary, which matters as much as the effect. A difficulty is desirable only if you can overcome it. Interleaving a problem type you cannot yet solve is not a challenge — it is noise, because with no method in hand there is nothing to choose between. Reducing feedback helps a competent student and strands a beginner. The test is simple: after honest effort, can you succeed at the harder arrangement often enough to learn from it? If yes, the difficulty is working. If you are only failing, you need an easier step first — a worked example, a short blocked run — and the difficulty becomes desirable later, once the groundwork is in.
That closes the unit on when to make studying harder. The remaining chapters turn from arrangement to understanding: not just how to schedule practice, but how to encode an idea so richly that it holds. The next folio starts there, with the simplest deep technique of all — asking how and why until a new fact connects to what you already know.
Practice — new ink and old, interleaved
1.From memory: what is interleaved practice, and what happened in the 54-classroom trial?
Interleaving mixes related problem types in one session so each problem must be classified before it is solved. In Rohrer's 2020 randomized trial, interleaved classes scored 61% to the blocked classes' 38% one month later — d ≈ 0.83.
How close were you? Grade yourself honestly — it sets your review date.
2.From folio 3: where is the forgetting curve steepest, and what does that imply for timing your first review?
3.You feel your practice going badly — slow, error-prone. From this folio, what should you conclude?
4.From folio 10: which desirable difficulties does successive relearning combine?
5.Which change to a study session adds a desirable difficulty?
6.From folio 5: a classmate says testing just feels harder than rereading, so rereading must be better. In one sentence, correct him using this folio's distinction.
7.A meta-analysis reports retrieval practice at g ≈ 0.61. What does that number mean?
8.Which two techniques does successive relearning combine?
9.From folio 9: why does interleaving count as a desirable difficulty?
10.From folio 7: spacing is a desirable difficulty. For material you need to hold for 60 days, a review gap of about 15% of that interval is roughly how many days?
11.From memory: state the retrieval-strength versus storage-strength distinction, and why it explains desirable difficulties.
Retrieval strength is how accessible material is right now; storage strength is how durably it is learned. Desirable difficulties lower retrieval strength during practice while raising the storage-strength gain per successful retrieval, so they feel worse now and last longer.
How close were you? Grade yourself honestly — it sets your review date.
12.Match each activity to what it actually practices.