← All postsStudy techniques

Desirable difficulties: When harder studying actually helps

Learn what desirable difficulties are, why some harder study conditions improve later performance, and how to separate productive effort from confusion.

The Bananote teamOctober 5, 20269 min read

A page of practice problems can feel reassuring when every question uses the same method. The heading names the topic, the previous answer suggests the next step, and the work gets faster as the pattern repeats. A mixed set removes those clues. Progress slows because each problem now begins with a decision.

That extra work can be useful, but the uncomfortable feeling does not prove anything by itself. A confusing explanation, a missing prerequisite, and a well-designed retrieval question can all feel difficult for completely different reasons.

The practical idea behind desirable difficulties is narrower: some learning conditions reduce performance during practice while improving later retention or transfer. The difficulty earns the word desirable only when it prompts useful encoding or retrieval that the learner is equipped to complete. (Bjork, 2013)

A hard session and a useful session are not the same thing

Performance is what can be observed during practice: the percentage correct, the speed of a response, or the amount completed. Learning has to be inferred from what remains available later and whether it can be used in a relevant situation.

Those measures can move together, but they do not always. A review by Nicholas Soderstrom and Robert Bjork documents conditions in which rapid improvement during practice did not produce the strongest long-term retention, as well as conditions that slowed acquisition but improved later performance. The same review also warns that no difficulty is inherently desirable; its effect depends on the learner, the task, and the criterion used to measure learning. (Soderstrom and Bjork, 2015)

This distinction explains why a smooth session can be misleading. It does not mean that smooth work is bad or that struggle is good. The useful question is whether the study condition prepares the student to produce the kind of response required later. The guide to the illusion of competence covers how to check learning when familiarity is supplying false confidence.

Which study methods can create desirable difficulties?

Several well-studied methods can make practice less fluent. Their benefits come from different processes, so they should not be treated as interchangeable ways to “make studying harder.”

Spacing requires returning after time has passed

Spacing distributes encounters with material instead of placing them in one continuous block. The later encounter may require more reconstruction because the answer is no longer as accessible as it was moments after study.

A 2022 review of spacing and retrieval research found broad benefits across ages, subjects, and educational settings, while emphasizing that useful intervals depend on the material and how long the learning needs to last. A gap is part of the method; there is no universal point at which forgetting becomes optimally difficult. (Carpenter, Pan, and Butler, 2022)

The forgetting-curve guide explains how to choose a return point without pretending one schedule fits every course.

Retrieval practice requires producing an answer

Retrieval practice asks for an answer before the source supplies it. That can mean explaining a process, solving a problem, writing what is remembered, or answering a question.

Retrieval practice has strong support compared with restudying, including in classroom research. A 2025 meta-analysis found a much smaller average advantage over established elaborative strategies, and that advantage depended on corrective feedback; without feedback, elaborative encoding performed better. The practical rule is to attempt an answer, then check and correct it, rather than treating retrieval as automatically superior to every other active strategy. (McDermott, 2021; Gonçalves, Muniz, and Jaeger, 2025)

Interleaving mixes related examples or problem types so the method is not announced in advance. The added work is discrimination: noticing which features matter and selecting an approach.

A meta-analysis found a positive average interleaving effect with substantial variation by material. Visual categories showed larger benefits, mathematics showed a smaller positive average, and word learning favored blocking. The interleaving guide covers when that comparison is useful and when a mixed set becomes random noise. (Brunmair and Richter, 2019)

Variation should match the flexibility the course requires

Changing examples, problem formats, or application contexts can prepare knowledge for use beyond one familiar presentation. Random changes are not automatically productive. The variation should require a relevant decision or application rather than adding unrelated inconvenience.

An uncomfortable chair or a poor scan adds friction without engaging a learning-relevant process. Changing locations is different: environmental context can affect recall, and varying context has sometimes helped, but the effects are conditional rather than a universal reason to move rooms. (Smith and Vela, 2001)

How can you tell whether a difficulty is useful?

Judge the design of the task before judging the feeling it creates.

Name the later performance. Decide what the course will require: recall a fact, distinguish two cases, solve an unfamiliar problem, explain a mechanism, or apply a rule.
Identify the added work. A useful difficulty should rehearse something connected to that performance, such as retrieving an explanation or choosing among similar procedures.
Check the prerequisites. The learner needs enough knowledge and support to make a serious attempt. If every response is a blind guess, reduce the scope or study an example first.
Provide a correction path. Use the course source, an answer key, instructor feedback, or another trustworthy reference to identify why the attempt succeeded or failed.
Measure the result later. Test the same knowledge after a delay or in a new but relevant example. Immediate frustration and immediate fluency are both weak verdicts on their own.

This test also catches false positives. Copying a blurry handout is difficult, but the effort does not practice a useful course response. Solving an unlabeled statistics problem can be difficult for a productive reason because selecting the method is part of the task.

What does the adjustment look like in one course?

Suppose a statistics class has introduced independent-samples t-tests, paired-samples t-tests, chi-square tests, and one-way analysis of variance. A student can follow each procedure when the worksheet heading names it, but an exam will present an unlabeled study and ask which analysis fits.

A useful progression starts with enough worked examples to understand the variables and assumptions behind each method. The next set removes the labels and mixes a small number of related cases. Before calculating anything, the student identifies the method and names the evidence: the type of outcome, the number of groups, and whether observations are paired or independent.

If every choice is a guess, the set has arrived too early or includes too many distinctions at once. Return to two methods, compare their decision rules, and try another mixed pair. If the method can be chosen but the calculation fails, the difficulty has exposed a different skill that needs practice.

That diagnosis is the value. “This felt hard” does not tell the student what to do next; “I confuse paired and independent samples when the design is described in prose” does.

Build a useful difficulty into Bananote practice

Start with one course note created from a supported source: a live recording, uploaded audio, PDF, pasted text, a YouTube lecture, an existing note, or a printed page captured with Scan Text on iPhone or iPad. Verify technical terms, rules, formulas, and examples against the original source.

Then design the practice around the later task:

Use flashcards for facts or relationships that need retrieval.
Use a scored quiz for a broader sample of the verified note.
Use AI chat when the task requires an explanation, comparison, or a sequence of questions.
Answer before opening the note, then check both the response and the generated feedback against the source.
Save the exact error and return to it in a later session rather than repeating the same question immediately until it feels familiar.

For the statistics example, a source-bounded prompt might be:

> Using only this verified note, give me one unlabeled study scenario at a time involving the statistical methods described here. Wait for me to choose a method and explain why. Then quote the parts of the note that support or contradict my choice, and mark anything the note cannot resolve.

Bananote reduces the setup required to keep verified material, questions, and later review together. The student still chooses the comparison, attempts the answer, checks the reasoning, and decides whether the challenge needs to be narrowed or increased.

Frequently asked questions

What are desirable difficulties?

They are learning conditions that can reduce performance during practice while improving later retention or transfer. Examples include spacing, retrieval practice, interleaving related material, and useful variation, but each works only under suitable conditions.

Why do effective study methods feel worse?

Some methods remove cues or require choices that make the answer slower to produce. That lower practice performance can coexist with stronger later learning, but the feeling of difficulty alone does not show that the method is effective.

Can a difficulty be undesirable?

Yes. Missing prerequisites, confusing instructions, excessive scope, and absent feedback can block learning. A useful difficulty requires the learner to perform a relevant process and still provides a path to an accurate correction.

How do I know if my studying is working?

Choose a task that resembles the course requirement, attempt it without the original cues, and check it against a trustworthy source. Return after a delay or use a new relevant example; no single delay fits every subject and deadline.

How does Bananote help?

Bananote can create flashcards, quizzes, and single-note chat practice from verified course material. Those tools make it easier to attempt an answer before looking and to return to errors later, while the student remains responsible for checking the source and judging the reasoning.

A useful study challenge has a job: it prepares a response the course will require and produces evidence that can be checked. If the only result is confusion, change the task rather than assuming more struggle will fix it.

Try Bananote with one verified course note, attempt a question before opening the source, and use the checked result to plan a later review.

Sources

Study less, remember more

Record a lecture, get flashcards, and let the schedule do the remembering.

Try it free Download