Overlearning: When Extra Practice Stops Paying Off

What is overlearning?

Overlearning means continuing to practise something after you've already reached the point of being able to do it correctly. You've solved the equation type, you understand it, and you do fifteen more of the same anyway — on the assumption that extra reps mean extra durability.

It's one of the most intuitive ideas in studying, and it's the implicit design of most textbook problem sets, which give you twenty near-identical questions in a row.

The intuition behind it isn't unreasonable, either. More practice does produce more fluency within the session, and that fluency is visible and satisfying. The question the research asks is narrower and more useful: not whether extra practice does anything, but whether it's the best available use of the same minutes.

Does overlearning actually work?

Doug Rohrer and Kelli Taylor tested this directly with mathematics practice, comparing extra within-session practice against distributing the same amount of practice over time. The overlearning produced little benefit for long-term retention, while distributing practice produced a substantial one.

The finding is not that the extra problems do nothing at all — it's that they are a poor use of the minutes. If you have thirty extra minutes, the evidence says putting them on a different day beats putting them on the end of today's session.

The comparison is what makes the result actionable. Both groups did the same total amount of practice — the only variable was whether the extra work happened immediately or later. That controls for effort and time and isolates the scheduling decision, which is the one thing a student can change for free.

Why do the extra repetitions buy so little?

Because by the time you're overlearning, the material is already fully available. You're retrieving from a memory that is sitting at the front of your mind, so each repetition is easy — and easy retrievals don't strengthen memory much.

Difficulty at the right level is what drives durable learning. A retrieval that takes effort, after a delay, when you weren't sure you'd manage it, does far more work than the fifteenth easy repetition of something you did four minutes ago.

This is the same principle that makes a delay useful rather than merely tolerable. When retrieval is effortful because some forgetting has occurred, the successful retrieval does real work. When it's effortless because you finished the last identical problem ninety seconds ago, there's very little for it to strengthen.

What should you do with that time instead?

Move it. Nicholas Cepeda and colleagues, synthesising a large body of spacing research, found that the same total study time distributed across separate sessions produces markedly better long-term retention than the same time massed together.

This is the single highest-leverage swap in most study plans, because it costs nothing. You are not being asked to study more — only to cut the tail off today's session and start tomorrow's a little earlier.

The practical shape of this is smaller sessions, more of them. Most study plans are built the other way round — long blocks per topic, each topic visited once — because that's how timetables and textbooks are organised. Restructuring toward shorter, repeated visits costs nothing but planning.

• Practise until you get it right two or three times in a row, then stop
• Put the remaining reps on a later day rather than at the end of the same session
• When you come back, start cold — no worked example in front of you
• If it's now too easy after the delay, stretch the gap further rather than dropping it

How do you know when you've practised enough?

The useful threshold isn't a number of repetitions, it's a quality of performance: can you produce it correctly, unaided, without a worked example visible? Once that's true a couple of times running, the session has got what it came for.

Be careful about judging this while the example is still on the page — that's precisely the condition under which people overestimate what they know. Cover everything before you decide you've got it.

Watch for the specific failure of judging readiness against a visible worked example. It's the same foresight problem that makes highlighting feel effective: with the method in front of you, every problem looks tractable. Cover the example, do one cold, and the answer is unambiguous.

Does this apply to flashcards too?

Yes, and it's why drilling a single deck repeatedly in one sitting feels productive and delivers less than it should. After the second or third pass in a session, you're recalling from short-term availability rather than genuinely retrieving.

A scheduling system handles this for you by design: it removes a card from today once you've got it and brings it back at a gap tuned to when you're likely to be on the edge of forgetting. The value is as much in what it stops you doing as in what it makes you do.

This also reframes what a scheduling algorithm is for. It isn't primarily a memory aid — it's a governor on the impulse to keep drilling what already feels shaky-but-known. Left to our own judgment most of us over-practise the material we've just done and under-practise the material we did last week, which is precisely backwards.

When is extra practice still worth it?

There are real exceptions. If a skill has to survive under pressure or without any warm-up — a procedure you'll perform under time limits, something safety-critical, or a motor skill — the extra automaticity has value beyond simple recall.

There's also a difference between repeating an identical problem and practising varied cases. Mixing problem types rather than repeating one is a different intervention from overlearning, and it tends to help. The thing the evidence discourages is specifically the twelfth near-identical repetition in a single sitting.

Worth separating two things that often get confused: repetition and retrieval strength. The exceptions above are cases where you need a response to be fast and automatic under load, which repetition does deliver. For material you need to recall accurately in an exam weeks away, spacing is the better buy, and for most academic study that's the case you're in.