Desirable difficulties are learning conditions that make study feel harder and slower in the moment but improve long-term retention and the ability to apply knowledge in new situations.
How it works
Researchers distinguish between performance, how well you do during study, and learning, what you can still do later. These often move in opposite directions. Rereading, practising one problem type at a time and studying a topic in one long block all make performance smooth, but much of that apparent learning fades. Conditions that slow you down by requiring effortful retrieval or decisions tend to produce more durable learning.
The main desirable difficulties are:
- Spacing study over time instead of massing it; see spaced repetition.
- Interleaving different topics or problem types; see interleaving.
- Testing yourself instead of restudying; see the testing effect.
- Generation, producing an answer before seeing it, which Slamecka and Graf (1978) showed improves memory.
- Varying conditions of practice, such as different problem formats.
How to use desirable difficulties (steps)
- Expect it to feel harder. Struggling during practice is a sign the method is working, not that it is failing.
- Test before you review. Start sessions with retrieval practice.
- Spread sessions out. Study a topic for shorter periods on several days.
- Mix problem types once you can do each on its own.
- Generate first. Try to answer before reading the explanation.
- Judge by results. Use delayed tests, not how study feels, to decide what works. That is good metacognition.
- Keep it achievable. If you are getting almost everything wrong, step back to easier material.
Evidence
Robert Bjork introduced the term in a 1994 chapter on memory and metamemory in training, and Bjork and Bjork (2011) later summarised the research for a general audience in a chapter titled “Making things hard on yourself, but in a good way”. The individual difficulties are well supported: Cepeda and colleagues’ 2006 meta-analysis for spacing, Rohrer and Taylor (2007) and Kornell and Bjork (2008) for interleaving, Roediger and Karpicke (2006) for testing, and Slamecka and Graf (1978) for generation. A consistent finding across these studies is that learners misjudge which conditions help them most.
Using desirable difficulties with Duetoday
Duetoday makes desirable difficulties easy to build into each session. Its flashcards use spaced repetition and require you to recall before revealing. Practice tests generated from several lectures mix topics. And when you are stuck, Galileo can give you a hint instead of the full answer, so you still do the work of retrieval.
