The myelin story, and what actually survives scrutiny

If you have watched more than three guitar practice videos you have heard this one. Every repetition fires a neural pathway. Repeated firing causes your body to wrap that pathway in myelin, a fatty insulating layer. Thicker myelin means a faster, cleaner signal. Therefore: perfect repetitions build myelin, myelin is skill, and sloppy repetitions wrap insulation around the wrong wire.

I believed this for years. I built practice habits around it. Then I went looking for the studies, and what I found was a real phenomenon supporting a claim it cannot actually carry.

What is true

Myelin is real biology, not a metaphor invented for guitar YouTube. Oligodendrocytes really do wrap axons, myelination really does increase conduction velocity, and activity-dependent myelination is a genuine and active area of neuroscience. Learning does change the brain, structurally, and myelin is part of how brains change.

What does not follow

The claim that gets made is much more specific than the biology supports: that repetition at the edge of your ability builds skill via measurable myelination of that exact circuit, and that this is the mechanism you should practise toward.

There is no evidence tying a specific act of practice to measurable myelination of specific cells in a way that would let you make practice decisions from it. The chain from "I played this lick fifty times cleanly" to "myelin thickened on the relevant axons" to "that thickening is why I improved" is assumed, not demonstrated. It is a plausible-sounding mechanism that has been promoted to a load-bearing one.

This matters practically, because if myelin is your model, certain conclusions feel obvious and are wrong. That one sloppy rep "wraps insulation around a bad pathway" and undoes your work. That volume of perfect reps is the primary variable. That you should never practise anything you cannot execute cleanly, which rules out speed bursts entirely.

What actually predicts learning

Here is the study that changed how I practise, and it does not mention myelin.

Duke, Simmons and Cash (2009) had advanced pianists learn a difficult passage, then tested retention the next day. They found no relationship between next-day retention and practice time, number of trials, or number of complete correct trials. Reps did not predict learning.

What did predict it was error handling. The players who retained the material were the ones who located the precise source of an error, immediately rehearsed the corrected version, and repeated until the error was gone. Not "more perfect reps." Better response to imperfect ones.

That reframes the whole activity. Practice is not accumulating clean repetitions to trigger a biological process. It is a debugging loop: find the exact failure, fix that specific thing, verify it is gone.

Turned into something you can actually do at the guitar, it looks like this:

1. Catch the moment the error happens. Not "that run was messy," but the specific note or transition where it went wrong. If you cannot say where, you are not yet practising, you are just playing.
2. Go back a step. Drop the tempo, and drop the amount of material to the smallest fragment that still contains the failure. Usually two or three notes either side of it.
3. Work out what caused it, slowly and deliberately. Is it a fretting finger arriving late, the hands drifting out of sync, a string change the picking hand is not set up for, tension creeping into the wrist? Name it. This step is the whole point, and it is the one people skip in favour of another rep.
4. Speed it back up until the error disappears, and keep going only as long as it stays gone. If it returns, you are back at step one with better information than before.

The reason this beats repetition is that a repetition contains no diagnosis. You can play a passage fifty times and, if you never isolate what fails, install the fault fifty times more securely. That is the grain of truth inside the myelin story, arrived at without needing myelin: what you repeat is what you get, so make sure the thing you are repeating is the corrected version.

The other stories in the same bucket

Once you start checking, several familiar claims turn out to be weaker than their popularity suggests.

Ten thousand hours. The Macnamara meta-analysis found deliberate practice explained about 21% of performance variance in music. Practice is the largest factor you control, and it is necessary. It is nowhere near sufficient, and the popular framing badly oversells it.

Micro-offline gains. The idea that ten-second rests between reps are themselves consolidating the skill was substantially undercut by 2025 replications: the effect shows up even with unlearnable random sequences, and is largely explained by pre-planning. Keep the short breaks, but justify them as fatigue and injury management, not a magic consolidation window.

Fast-twitch fibres for picking speed. Picking speed is a motor-control and movement-efficiency problem, not a muscle fibre type problem. The sprinter physiology story does not transfer to a small, learned hand motion.

Why bother demoting a harmless metaphor

Because the metaphor generates bad advice. "Myelin wraps whatever you repeat" implies that practising anything less than perfectly is actively damaging, which pushes you toward endless slow repetition and away from at-tempo speed bursts. Bursts are where you find the fast motion, and they are supposed to be messy. If your model tells you messy is harmful, you will skip the one thing most likely to move your ceiling.

Keep myelin as a loose way of saying "practice changes the brain." Do not practise toward it. Practise toward finding errors precisely and killing them, which is the part that has evidence behind it.

If you want the practical framework built on that evidence, the three zones is where it turns into a session plan.