Difficulty that helps

Treat difficulty as useful only when it makes the learner perform relevant work. Performance during practice is not the same as durable learning. A review of this distinction found that some conditions improve immediate performance while harming later learning, while others make practice performance worse but improve later retention or transfer.¹

Retrieval illustrates the difference. In the prose experiment, restudy felt easier and led after five minutes, while testing led after two days and one week. Across a meta-analysis, testing benefits were also larger after delays of at least one day than after shorter delays.² Struggling to retrieve can therefore coexist with better delayed retention.

That does not make struggle valuable by itself. In two retrieval-practice experiments, longer gaps made successful practice retrieval harder while improving final-test performance.³ Yet a separate long-term spacing experiment found that increasing the gap helped only to a point, after which longer gaps reduced retention.⁴ Difficulty can accompany effective practice, but continually increasing it is not a dependable rule.

A failed attempt can still be useful when the correct answer follows. Across six experiments using fictional general-knowledge questions and weakly associated word pairs, attempting retrieval before seeing the answer improved later learning even when the attempt failed. Those experiments always presented the correct answer afterward, so they do not show that unsupported errors improve memory.⁵

With a deck and twenty minutes, use prompts that require an answer rather than recognition alone. Pause long enough to make a genuine attempt. If retrieval fails, reveal and inspect the correct answer instead of continuing unsupported guessing. A difficult moment is acceptable when it belongs to this process; it is not a score to maximise.

Boundary. No evidence here supports difficulty independent of successful processing, correction, the material, the delay and the learner’s readiness. Researchers have proposed effortful processing as part of the testing effect, but no single mechanism fully explains it.² The retainable judgement is not that harder is better. It is that easy practice can mislead, while effortful retrieval with access to the answer can support later memory.