The single most robust finding behind flashcards is that trying to remember something is not a way of checking whether you know it. It is a way of learning it — and it beats re-reading the same material, often substantially.
At a glance
| Restudying | Retrieval practice | |
|---|---|---|
| What you do | Read the material again | Try to produce it from memory |
| How it feels | Fluent, comfortable | Effortful, sometimes unpleasant |
| Short-term performance | Often looks better | Often looks worse |
| Retention after a week or more | Weaker | Stronger |
| What students predict | This will work better | This will work worse |
The experiment that anchors the field
The reference point most people cite is Roediger and Karpicke's 2006 work, Test-Enhanced Learning: Taking Memory Tests Improves Long-Term Retention, published in Psychological Science.
The design is simple enough to describe in a sentence. Students read a passage. One group then read it again; another group took a test on it, with no feedback. Later, everyone was tested for real.
Two results, and the second is the important one:
On a test very soon after studying, the re-readers did slightly better. Freshly re-read material is highly available.
On a test after a delay — days rather than minutes — the retrieval group did markedly better, and the gap widened as the delay grew.
That crossover is the whole story. Restudying buys short-term availability. Retrieval buys durable memory. A revision strategy optimised for how well you can perform this evening is optimised for the wrong thing, because the exam is not this evening.
It is not just about facts
A reasonable objection: perhaps retrieval only helps with rote recall, and understanding needs something richer.
Karpicke and Blunt addressed exactly this in Science in 2011, comparing retrieval practice against concept mapping — building an elaborate diagram of how ideas relate, which is about as far from rote as a study method gets. They tested not only verbatim recall but inference, the kind of question that requires connecting ideas.
Retrieval practice won on both. Even on the inference questions, the students who had simply practised recalling the material outperformed the students who had built concept maps.
And the students predicted the opposite. Asked which method would serve them better, they backed concept mapping.
How large, and how reliable?
Two meta-analyses are the standard citations for anyone asking whether this is a real effect or a handful of striking studies.
Rowland's 2014 review in Psychological Bulletin synthesised testing-versus-restudy comparisons and found a consistent advantage for testing, robust across materials and designs. Adesope, Trevisan and Sundararajan followed in 2017 in Review of Educational Research with a meta-analysis of practice testing specifically, reaching the same conclusion.
The effect also holds up in ways that matter practically: it works with feedback and (more weakly) without it, it works for young learners as well as undergraduates, and it works in real classrooms rather than only in laboratories.
This is about as settled as findings in education research get. The UK's Education Endowment Foundation reflects it in its guidance to schools, which is the register a teacher or head is most likely to recognise.
Why the discomfort is the mechanism
The uncomfortable part of retrieval — the pause, the groping for a word, the almost — is not a side effect to be minimised. It appears to be where the benefit comes from.
Robert and Elizabeth Bjork's framework of desirable difficulties is the usual way of explaining this: conditions that make learning feel harder in the moment often make it more durable. Retrieval is the clearest example. Reading is easy and does little; recalling is hard and does a lot.
Which means a study tool that removes the effort has removed the point. This is worth remembering when evaluating anything that promises easier revision.
What it means for a flashcard
Everything about the design of a good card follows from this research:
Your child must attempt the answer before revealing it. The attempt is the mechanism. A card flipped straight over is re-reading with extra steps, and our review flow requires the reveal to be a deliberate action for exactly this reason.
A failed attempt is still productive. Trying and failing, then seeing the answer, beats not trying. So a card graded Tricky is not a wasted card — which is why nothing is deducted for it.
One fact per card. Retrieval has to be a single, gradeable act. A card holding three facts produces a muddle rather than a retrieval attempt.
Feedback should be immediate. The answer appears the moment your child commits, which is the arrangement the research supports.
The one caveat worth stating
"Testing" in this literature means low-stakes retrieval practice: private, unmarked, no consequences. It does not mean more exams. Conflating the two is the most common misreading of this research, and it is how a genuinely humane finding gets used to justify something else entirely.
A child doing six minutes of flashcards is not being tested. They are practising the one activity that most reliably turns something they have met into something they know.
FAQ
What is the testing effect?
The finding that attempting to retrieve information from memory strengthens that memory more than restudying the same material. Retrieval is not just measurement — it changes what it measures.
Isn't testing just stressful for children?
High-stakes testing can be. Retrieval practice as a study method is low-stakes and private: no marks, no comparison, no report. Those are different things that unfortunately share a word.
Why do students prefer re-reading if recall works better?
Because re-reading feels more fluent. Familiarity is mistaken for knowledge, and the difficulty of retrieval is mistaken for failure. Students routinely predict re-reading will work better, and are wrong.