Desirable Difficulties: Why Making Learning Harder Makes It Stick

The Paradox of Effective Learning

One of the more counterintuitive findings of cognitive psychology is that the conditions that produce the best learning outcomes are often subjectively unpleasant, while the conditions that feel most productive frequently produce shallow, impermanent learning. This paradox was crystallised in Robert Bjork's concept of "desirable difficulties" — conditions that introduce manageable challenges into the learning process in ways that slow apparent progress during practice while producing superior long-term retention and transfer.

The concept challenges a widely held assumption: that learning should be made as easy and fluent as possible. Reducing friction, removing obstacles, providing immediate feedback, presenting information in clear and accessible formats — these interventions often reduce the cognitive effort required during learning. And it is that effort, research suggests, that drives durable encoding.

What Makes a Difficulty Desirable

Not all difficulty is desirable. A difficulty is desirable if it slows initial acquisition while producing better long-term retention and transfer. It is undesirable if it impedes performance without producing this compensating benefit. The distinction matters because it prevents the concept from being used to justify arbitrary challenge or poor instruction.

Desirable difficulties that have been identified by research include spacing (distributing practice over time rather than concentrating it), interleaving (mixing different types of material within a session), retrieval practice (testing recall rather than reviewing), varying conditions of practice (changing the context, format, or conditions under which material is practised), and reducing feedback (providing less immediate correction than learners prefer).

Reduced Feedback as a Desirable Difficulty

One of the less intuitive desirable difficulties involves reducing the frequency of corrective feedback. In many training contexts, frequent feedback is assumed to be beneficial — it provides correction, builds confidence, and maintains engagement. Research on motor learning in particular has found that while frequent feedback does improve performance during training, it can create a dependency on external correction that undermines performance when feedback is removed. Learners who receive less feedback during training often perform worse during practice but better on delayed tests without feedback.

The explanation appears to involve the guidance hypothesis: frequent feedback guides performance rather than teaching the underlying skill. Learners who receive less feedback must develop more robust internal models of correct performance, and these internal models transfer better to novel conditions.

Generating Information vs Receiving It

A related phenomenon is the generation effect: information that is generated by the learner — through effort, inference, or partial completion — is retained better than information that is simply received. This has been demonstrated in studies where participants who completed a word fragment (e.g. "soc_al" for "social") later recalled the word better than participants who simply read the complete word. The generation process appears to deepen encoding in ways that passive reception does not.

Instructional approaches that exploit the generation effect include asking learners to attempt problems before being shown solutions, having learners predict outcomes before observing them, and prompting learners to generate examples of concepts rather than having examples provided. Each of these introduces an effortful generation step that improves retention.

Implications for Self-Directed Learning

The desirable difficulties framework suggests that learners who make their own study choices may systematically select suboptimal conditions. The strategies that feel most effective — re-reading, blocked practice, frequent review, immediate feedback — are often those that produce the smoothest performance during study. Strategies that produce better long-term outcomes — retrieval practice, interleaving, spacing, reduced feedback — are those that produce difficulty and apparent struggle. Awareness of this inversion does not guarantee that learners will adopt the more challenging approaches, but research suggests it does increase the likelihood.