Desirable Difficulties: Why Making Learning Harder Actually Makes It Stick

Robert Bjork coined the term "desirable difficulties" in 1994, and it remains one of the most counterintuitive — and most actionable — ideas in all of learning science. The core claim is almost offensive to common sense: conditions that make learning feel harder in the short term tend to produce far stronger retention and transfer over time.

Photo by MART PRODUCTION on Pexels.

Not all difficulties qualify. Struggling because your textbook is poorly written, or because you're exhausted, or because the instructions are ambiguous — that's just noise. Desirable difficulties are specific kinds of friction that force your brain to process material more deeply. The struggle itself is doing the work.

So what makes a difficulty "desirable"?

The Four Heavy Hitters

Spacing. Instead of studying a topic in one long session, spread your practice across days or weeks. This feels less efficient because you forget things between sessions and have to re-learn them. That re-learning is not wasted effort — it's the point. Each retrieval strengthens the memory trace in ways that massed practice simply cannot.

Variation. Practice the same skill across different contexts, formats, and problem types rather than drilling one version repeatedly. A math student who only solves equations presented in identical format will be lost when the format shifts slightly. Varying the conditions forces your brain to extract the underlying principle rather than memorize surface features.

Interleaving — which we've covered separately — fits here too. Mixing problem types during practice is harder than blocking them by category, but it builds the ability to identify which approach a problem calls for, not just execute an approach you were just told to use.

Generation. Try to answer a question or solve a problem before you've been taught the solution. Get it wrong. Then study the material. Research consistently shows that attempting to generate an answer — even incorrectly — makes the subsequent instruction far more memorable. Your brain has primed the slot, so to speak.

Why Our Instincts Mislead Us

Here's the uncomfortable truth: the conditions that feel most productive are often least effective. Re-reading your notes feels good because it's fluent — everything seems familiar. Highlighting is soothing. Studying the same topic for three hours straight creates a satisfying sense of mastery.

Bjork calls this the difference between performance and learning. Performance is what you can do right now. Learning is what you can do later, under different conditions, without the cues that were present during study. These two things come apart constantly, and students — and instructors — almost universally optimize for the wrong one.

A learner who spaces their practice will perform worse during study sessions than someone who masses it. They forget more between sessions. They feel less confident. But test them two weeks later, and the spaced learner wins decisively.

A Practical Model

Here's how these difficulties interact across a typical learning cycle:

graph TD
    A[Attempt Generation] --> B{Get It Wrong?}
    B -->|Yes| C[Study the Material]
    B -->|No| C
    C --> D[Spaced Review Session]
    D --> E[Varied Practice]
    E --> F((Durable Learning))

Notice that getting it wrong in the generation phase doesn't branch you off to a failure path — it feeds directly into the same learning process as getting it right. The attempt is what matters, not the outcome of the attempt.

The Transfer Problem

If retention were the only goal, spacing and retrieval practice would be enough. But most of us need to use what we learn in new situations — that's transfer, and it's harder to achieve than simple recall.

Desirable difficulties, especially variation and interleaving, are specifically suited to building transfer. When you've only ever seen a concept in one context, you don't really understand the concept — you understand that context. Vary the contexts enough, and you start to see what stays constant. That invariant is the actual idea.

This is why worked examples that vary surface features while preserving deep structure are so powerful for technical subjects. You're teaching your brain to look past the clothing and recognize the shape underneath.

What to Do Monday

None of this requires special software or a new system. Close your notes and write down everything you remember from your last study session — that's generation and retrieval practice combined. Schedule your next review for three days from now, not tonight. When you practice problems, shuffle them instead of doing all of type A before moving to type B.

These are small changes. The discomfort they produce is the evidence they're working. Trust the friction.