Stuck at the Same Level for Months? How to Get Past a Learning Plateau

Is it a real plateau, or just a bad week?

A real plateau is a flat line across many days, measured the same way each time. One rough evening, a cold or a busy month at work doesn’t count.

Measure at the right moment, too. Your last rep of the night shows what your hands can do while the move is still fresh. Researchers call that performance and keep it apart from learning, which is what remains later.

How you do during practice is often a poor guide to what you have learned (Soderstrom and Bjork, Perspectives on Psychological Science, 2015).

People also misread it the same way each time. In Simon and Bjork’s 2001 keypad experiment, repetitive practice looked better in the session, tested worse a day later, and still got learners’ vote.

Then ask whether you have been practicing the skill or just using it. Reps done the same way, at the same level, give the skill no reason to change.

Which drill fits the way you’re stuck?

Pick the row that matches your stall and start there. SOLID means repeated support from tests on people. PROMISING means real but thin or indirect evidence. STORY ONLY means common advice that nobody has properly tested.

Notice that the book’s only two untested drills are both plateau drills. Plateau tips are everywhere; trials of them are scarce.

Why does a skill stop improving when you use it every day?

Daily use is not practice. Once a skill is good enough for ordinary life, it runs on autopilot and nothing pushes it further.

The book’s example is an office worker whose typing speed has not changed in ten years, despite typing every working day. Writer Joshua Foer called this “the OK plateau” in a 2011 popular book. That label is an account, not a lab result, but the research lands nearby.

Detailed records of skilled performance show plateaus, then dips while people try new methods, then leaps when one works (Gray and Lindstedt, Cognitive Science, 2017).

A 2021 study of Tetris players by Gray and Banerjee pointed the same way: technique, not raw button speed, set the ceiling. So a flat line usually means a method has run out. Getting off it means making the skill demanding again and putting up with a worse week first.

For the typist, the new difficulty is ten minutes without the backspace key, copying text he has never seen. His mistakes stay visible, and within days a few problem letter pairs stand out.

Does slowing down actually help you break through?

It helps you get clean reps on a trouble spot. No experiment shows it raises your top speed.

The only tempo experiment the book found, Henley’s 2001 study of 60 high school wind players, saw no effect from the tempo pattern, though working with a model helped. It was a short, one-session test. A 2009 study of 17 advanced pianists by Duke, Simmons and Cash adds a clue: the next day, players with a higher share of correct practice attempts did better. That is a correlation, not proof.

So the book’s slow-practice drill aims at accuracy. Mark the spot that keeps breaking, lower the speed until it comes out right several times running, and count clean reps out of ten. Raise the speed a notch when most are clean, finish with a few reps near normal speed, and recount the next day.

Won’t 10,000 hours of practice get you there anyway?

Hours help, but the number was never a finish line, and how you practice predicts more than how long.

It comes from a 1993 study of 30 Berlin violinists by Ericsson, Krampe and Tesch-Römer. At 18, the top group averaged 7,410 hours, well short of 10,000. Ericsson later said the figure had no special power.

Pooled studies found deliberate practice accounted for 26 percent of differences between people in games, 21 percent in music and 18 percent in sports (Macnamara, Hambrick and Oswald, Psychological Science, 2014).

Ericsson and Harwell argued in 2019 that a stricter definition of practice raises those shares to 29 and 61 percent. Either way, the quality of each session is yours to change. That is also why counting clean attempts beats logging minutes.

How do you break a stuck skill into pieces?

Cut it into three or four chunks, smooth out each one, then practice the seams where they meet. Seams are where most sections fail.

People group movements into chunks on their own. In a 2003 sequence study, Sakai and colleagues found that shuffling a learned sequence hurt less when each person’s own groupings were kept, and the groupings varied between people. Cowan’s 2001 review put working memory at about four chunks, not the famous seven.

When a whole attempt fails and you can’t tell why, list the steps in order, drill the weakest one alone, and rebuild the full skill in the same session. A 2009 meta-analysis by Fontana and colleagues broadly backed this, with too few studies for some comparisons.

Should you mix up and vary your practice?

Yes, after a new move works at all. Learn it in a block, then shuffle it with related skills and vary one detail at a time.

Expect modest gains. The 2024 meta-analyses by Czyż and colleagues found a large benefit from mixed practice in labs and close to none in real-world settings. Compare a mixed week with a blocked week using cold scores. To vary, keep your real equipment and change one thing that affects the result: distance, tempo, key or angle.

How often should you practice to keep improving?

Short sessions on most days beat one long weekly session. This idea has some of the strongest support in the book, and its Practice Week Map is graded SOLID.

In a 2006 randomized trial, Moulton and colleagues gave 38 junior surgeons equal training time, either in one day or once a week. A month on, the weekly group kept the skill better on most measures.

Of 19 direct comparisons of spread-out and massed surgical skills training, 15 favored spreading and none favored massing (Jørgensen and colleagues, Surgical Endoscopy, 2025).

Session length is less certain. In Baddeley and Longman’s 1978 typing study, one hour a day taught the most per hour and two two-hour sessions the least; nothing under an hour was tested. The book’s 20-minute starting block is its own suggestion. Split your real weekly time into short sessions on separate days, and move a missed one to a spare day rather than doubling up.

What should you do with your misses?

Put each miss next to a make from the same session. The difference between them usually points at the cause.

Avoiding errors is no cure: a 2026 review of 29 papers by Wong and colleagues found errorless practice gave healthy children and young adults no significant edge. Reviewing helps. In Ellis and Davidi’s 2005 study, soldiers who went over both failures and successes improved more than those who reviewed failures alone. So after each session, write one target, one miss, one make and one change, and read it before the next session.

A weekly phone clip of five to ten reps, watched with a single question in mind, shows what feel hides. A 2026 review by Wang and colleagues found the gains from watching your own best reps had faded a week later, so use video as a check.

Are you too old to get past a plateau?

No. Motor performance tends to dip with age, but the capacity to learn holds up.

Older adults’ capacity to learn motor skills stays intact, though gains on complex fine-motor tasks may be smaller (Voelcker-Rehage, European Review of Aging and Physical Activity, 2008).

In a small 2016 study by Bugos and colleagues, 17 adults aged 60 to 85 with little musical background completed 30 hours of piano training. A 2016 sports review by Macnamara and colleagues found top athletes had not started earlier as children. And if anything hurts, stop and ask a professional.