The word rusty is doing a lot of work
There is something you used to be able to do. You can still feel the shape of it. You sit down to do it now and what comes out is slower and thinner than what used to come out, and the word everybody reaches for is rusty.
Rust is corrosion. It is material loss. A rusted bolt has less metal in it than it had when it went in. The word carries a claim about what happened to the skill while you were not using it, and the claim is that some of it is no longer there.
That is probably not what happened. There is an account in memory research which says the thing you learned is still at full strength, that something else entirely is what fell away, and that the two are so different that confusing them changes what you should do next, how long it will take, and what you are entitled to say about yourself while it is happening.
One distinction, read slowly
In 1992 Robert Bjork and Elizabeth Bjork proposed that anything held in memory carries two separate strengths rather than one.
Storage strength is how thoroughly the thing is learned. How many times it was practiced, how deeply it was encoded, how tightly it is tied to everything else you know. It is a property of the learning.
Retrieval strength is how accessible the thing is at this moment. Whether it comes when called, and how fast.
Everyday language has one word for both of these and that is where the trouble starts. We say somebody knows something, and we mean either that they learned it well once or that they can produce it right now, and those are different facts that come apart all the time.
The theory holds that storage strength does not decrease. Whatever was learned thoroughly stays learned thoroughly. Retrieval strength is the one that falls with disuse, and it falls whether the underlying learning was shallow or deep.
Two consequences follow that are worth sitting with, because neither is obvious.
The first is that high storage strength and low retrieval strength is a specific state, and it is not the same state as never having learned the thing. Two people who cannot currently do something are in different positions if one of them used to do it well. They look identical from outside, and they perform identically on the day, and they will not perform identically six weeks from now.
The second is stranger. In this account, retrieving something when its retrieval strength is low does more for storage strength than retrieving something that was already accessible. Difficulty at the moment of recall is not a sign that the recall failed to do its work. It is closer to the opposite.
What Ebbinghaus measured, which was not forgetting
Ebbinghaus is remembered for a curve, and the curve is the least interesting thing he did.
Working alone on himself in the 1880s, he learned lists of nonsense syllables to a fixed standard, which was reciting the list correctly, and recorded how long that took. Then he waited, for 20 minutes or a day or a month, and learned the same list again to the same standard. The number he cared about was not how much he could recall. It was how much less work the second learning took than the first.
He called that difference savings, and it is a different instrument from a recall test in a way worth being precise about. A recall test asks whether the thing comes back. It returns a yes or a no, and after a long enough interval it returns no for almost everything. Savings asks what the first learning is still worth, and it keeps returning a number after the recall test has gone silent.
Material he could no longer produce at all still cost less to learn the second time than material he had never seen. Something survived the forgetting that the recall test could not detect, and the only way to see it was to measure the relearning rather than the remembering.
The limits are severe and he stated most of them himself. One subject, who was also the experimenter. Nonsense syllables chosen precisely because they carried no meaning, which is the opposite of a skill. Nothing about it settles what happens to a person picking up a guitar.
What it establishes is narrower and it is the thing this piece rests on. The failure of a recall test is not a measurement of what remains. If you want to know what is left of something you used to be able to do, the test is not whether you can do it now. It is what the next six weeks cost.
The limits of that, stated here rather than later
A skill is not a syllable list, and the honest version of this argument has to say where the theory stops.
Bjork and Bjork's account is a theory of memory. A great deal of what you mean by a skill is procedural and physical, involves timing, and lives in systems this framework was not built to describe. The bridge from one to the other is an argument, not a finding.
The skill decay literature is more directly about the thing you are asking about, and it is messier. Arthur and colleagues reviewed the quantitative work on decay and retention and found that how much is lost depends on how long the interval was, what kind of task it is, and how the skill was trained and tested in the first place. There is no single decay rate to quote at you, and anybody who quotes one is rounding off the part of the evidence that matters.
What survives both of those caveats is narrow and still useful. Not doing something for a long time reliably damages your access to it. It does not reliably delete what you learned. The felt experience of being rusty is a measurement of access, and access is the recoverable half.
Which skills hold, and which fall
Arthur and colleagues found more than the absence of a single decay rate, and the more specific finding is the one that answers the reader who suspects their own case is the hopeless one.
Decay was not uniform across kinds of task. Physical and speed based skills held up better over a period of disuse than cognitive and accuracy based ones. The things a body does, at pace, with the whole sequence running as one movement, came back more readily than the things a mind does, slowly, where being right matters more than being quick.
That cuts across the way most people rank their own skills. The instrument, the knife work, the language you spoke fluently, the sport: those are the ones people assume are gone, because they feel furthest away. They are the ones the evidence treats most kindly. The spreadsheet model you built over two years, the statute you knew cold, the codebase you could hold in your head: those feel closer, because you can still talk about them, and they are the ones more likely to have thinned.
There is a reason in the theory for it, and it is worth stating as a reason rather than as a finding. A physical skill run at speed is one long chain where each step cues the next, so retrieving any part of it drags the rest along. A cognitive skill built on accuracy is a set of separately retrievable facts and judgments, and they fall away one at a time without the others noticing.
Hold the caveat from the section above beside it. This is a pattern across studies rather than a rule about your case, the categories overlap in real work, and almost nothing you do is purely one or the other. Driving is physical and full of judgment. Writing is cognitive and has a motor rhythm anybody who has lost it will recognize.
What it is good for is a correction to the intuition. The skill you feel furthest from is not necessarily the one furthest gone, and the ranking you carry in your head was built on how strange it feels to try, which is a measurement of retrieval strength and nothing else.
What you can still judge
Here is the part that gives the distinction teeth, and it is the part people notice in themselves and then dismiss.
You can usually still tell good work from bad work in a skill you have not practiced in years.
Play something badly on an instrument you have not touched since your twenties and you will hear that it is bad, and you will hear specifically what is wrong with it, and you will know what it is supposed to sound like. Write in a language you have lost and you will not produce the right sentence, but you will know that the one you produced is not it. Watch somebody else do the thing and you can still see the difference between competent and excellent, and say what the difference consists of.
Judgment survives execution by a long way. That is not a finding anybody ran on lapsed performers, and it should not be dressed up as one. Polanyi's observation that we know more than we can tell is usually cited about the knowledge a person cannot put into words while performing, and the extension here is mine: the same gap opens in the other direction, in somebody who cannot perform at all and can still name every fault.
That asymmetry is worth something on its own terms, before any question of recovery. It is why a lapsed practitioner is useful in a room full of people who never learned the thing. Recognizing what good looks like is a separate capability from producing it, and it is the one that stays.
What practice will not return
There is a thing this argument cannot reach and it should be said plainly, because the reader who came here with a real loss will notice its absence.
When someone misses a skill, they are often not missing the skill. They are missing the conditions that surrounded it. The room, the hour of the day, the other people who were doing it too, the reason it mattered to do it at all, the version of themselves that had the time.
None of that is retrieval strength and none of it comes back by practicing. A person can recover the skill completely and find that the missing thing is still missing, because the missing thing was never the skill.
That is worth separating out before you spend six weeks on the wrong recovery. Some of what feels like a lost capability is a lost circumstance wearing its clothes.
Four objections
This is a theory of memory and a skill is not a memory. Correct, and the section above concedes it. The claim being made is narrower than the theory: that the distinction between what was learned and what is currently accessible applies to skills too, which is an argument from resemblance rather than a demonstration. Treat it as a better hypothesis than the rust one, not as a proof.
Some skills really are gone. Some are. Skills that depend on a body that has changed, on equipment nobody has any more, on a context that does not exist. The account here does not promise recovery of everything. It says that the learning is unlikely to be what you lost, which leaves open that something else was.
This sounds like permission not to practice. It is closer to the reverse. If relearning is cheaper than learning, then the case for going back to something is stronger than it looks from the outside, not weaker. The person who says it is too late to pick it up again is pricing the work as though they were starting from nothing, and they are not.
What if it is not worth recovering? Then say so and stop carrying it. A skill you dropped for good reasons and do not want back is not a failure, and the guilt attached to it is doing no work. The question at the foot of this piece asks what you miss, which is a narrower thing than what you lost.
An hour, if you want one
Take the skill and write three things down. Not a plan, three facts.
First, what you can still judge. Describe, in a sentence or two, what separates good work from adequate work in this skill. If you can write that, you have put on the record something that survived, and the record did not exist five minutes ago.
Second, the last time you did it. Where you were, who else was there, and what it was for. This is the test for whether you are missing a capability or a circumstance, and you will usually know which one it is by how the sentence comes out.
Third, what six weeks would buy. One hour a week, no more, and be specific about what you would expect to be able to do at the end. If the answer is that you cannot tell, that is the real finding, and it is more useful than an ambitious schedule you will not keep.
What no system recorded
Your employment history holds the roles. Your qualifications hold what you were certified in. Your performance reviews hold what you were doing during the years somebody was watching.
None of them hold the thing you could do before that started, or alongside it, or the year you got good at something and then stopped.
So begin with the cheapest of the three facts to check, which is the second one. Whether what you miss is the skill, or the room you used to do it in.