The most replicated finding in learning research is also the least followed: testing yourself on material beats reviewing it, by a wide margin, at equal time spent.
Rereading is the default study method almost everywhere, and it has one enormous advantage — it feels good. Text you have read before is easy to process the second time, and the brain interprets that ease as understanding. It is not. It is familiarity, and familiarity is what fails you in an exam room, where the question is phrased differently and nothing on the page is familiar at all.
Retrieving information from memory does something rereading cannot: it strengthens the path back to that information. The act of struggling to recall is not a side effect of studying, it is the mechanism. This is why a quiz you find difficult teaches more than a page you find easy, and why the difficulty is not something to engineer away.
The practical consequence is that a flashcard you nearly failed to answer was more valuable than three you answered instantly. Students often reshuffle their decks to avoid that discomfort, which quietly removes the benefit.
The second finding is that when you review matters as much as whether you do. Material reviewed just before you would have forgotten it sticks far better than material reviewed while still fresh. Reviewing too early is nearly wasted effort — you recall it easily, and easy recall strengthens little.
This is what a spaced-repetition scheduler automates. Each item carries its own interval, which stretches when you get it right and collapses when you get it wrong. Done properly, the total time spent falls while retention rises, because effort concentrates on the small set of things you are actually about to lose.
Left to choose freely, most people review their strongest material. It is more pleasant, it produces a higher percentage correct, and it feels like progress. It is also close to useless — you cannot improve on questions you already answer correctly.
Any honest system will therefore feel worse to use than an unstructured one. If your practice sessions are comfortable, that is evidence something is wrong with them.
Most students review by reading the explanation and moving on. That produces recognition of the explanation, not the ability to answer a different question on the same idea. A more useful review distinguishes between four very different failures:
| Answered quickly | Answered slowly | |
|---|---|---|
| Correct | Solid — retrievable knowledge | Fragile — reconstructed, will fail under time |
| Wrong | Misconception — a confident false belief | Gap — unfamiliar material |
The two wrong cases need opposite responses. A misconception is a belief you hold and will keep applying; re-reading the topic will not dislodge it, because you did not fail to learn it, you learned something incorrect. What corrects it is confronting the specific contrast — why the thing you believed applies to a neighbouring concept and not this one.
A gap is simply absence. Practice questions are an inefficient way to fill it; instruction is faster. Grinding questions on material you have never learned mostly generates frustration.
The fragile case is the one nobody tracks and the one that most often explains a plateau. You know it, you can reconstruct it, and under exam pacing the reconstruction does not finish in time. The fix is not more learning — it is speed work on material you already understand.
Studying one topic in a long block produces better practice-session performance and worse retention than mixing topics. Blocked practice lets you apply the same approach repeatedly without deciding which approach the problem calls for — and choosing the approach is most of the difficulty on test day, where questions arrive in no particular order.
Mixed practice feels worse and scores worse in the moment. That, again, is the point.