What the Research Actually Says About Knowledge Decay in Aviation

I trained as a surgeon before I ever held a pilot certificate, and the two disciplines taught me the same uncomfortable lesson from opposite directions: skill is not a thing you acquire. It's a thing you maintain. The certificate in your wallet and the board in my office are both snapshots of a competence that was real on a particular afternoon — and that, from that afternoon forward, begins to fade.

We don't like to talk about fading. In aviation we talk about currency, which sounds like proficiency and is measured like a parking meter. But the science of how humans forget is large, old, and surprisingly consistent, and almost none of it supports the idea that legal currency and retained skill are the same thing. I want to walk through what that research actually says — carefully, because this is the kind of subject where a confident overstatement can get someone killed — and then say plainly what I think it means for the rest of us who fly.

The forgetting curve is real, and it is steep

The starting point is the oldest result in experimental psychology. In the 1880s Hermann Ebbinghaus memorized lists of nonsense syllables and measured how much effort it took to relearn them after a delay. The "savings" — how much faster relearning went the second time — dropped sharply within the first day and kept declining for weeks. That shape, the forgetting curve, became one of the most cited ideas in the study of memory.

For a long time it was also one of the least re-tested. So it matters that in 2015 Jaap Murre and Joeri Dros published a careful replication in PLOS ONE and got a curve very close to the original (Murre & Dros, 2015, doi:10.1371/journal.pone.0120644). I want to be honest about its limits, because honesty is the whole point of an article like this: their replication, like Ebbinghaus's own work, rests on a single dedicated subject relearning nonsense material. It is a genuine confirmation of an old and well-corroborated effect, not a thousand-pilot field study.

What survives that caveat is the shape, not a precise number. Retention falls fast at first and then more slowly. The curve isn't even perfectly smooth — there's a small overnight recovery, the fingerprint of sleep doing its consolidation work. The lesson is not "you forget X percent by Tuesday." The lesson is that forgetting is the default, it starts immediately, and it is fastest right after you learn — which is exactly the window in which we tend to assume the knowledge is safely ours.

Skills decay too — and the ones you'd bet on aren't the ones that go first

A reasonable objection is that nonsense syllables aren't flying. Flying is a skill, not a vocabulary list. Does skill decay the same way?

The best single answer is a 1998 meta-analysis by Winfred Arthur and colleagues, Factors That Influence Skill Decay and Retention, which pooled 189 data points across 53 studies (Arthur et al., 1998, doi:10.1207/s15327043hup1101_3). Two findings hold up. First, trained skills decay during periods of disuse. Second, the loss grows with the length of the layoff — the longer you go without practicing, the more you've lost. I'll resist quoting their largest effect size as if it were a universal constant, because it isn't: the worst decay showed up after a year or more of no practice, and it varied a great deal by the type of task. That variation is the interesting part.

Here is where aviation gets its own, specific, and frankly counterintuitive result. In a 1986 study in Perceptual and Motor Skills, Childs and Spears examined flight-skill decay and found that cognitive and procedural skills decay faster during disuse than control-oriented, "stick-and-rudder" skills (Childs & Spears, 1986, doi:10.2466/pms.1986.62.1.235). Read that twice, because it inverts the usual worry. We tend to fret about whether our hands still remember how to land. But the hands hold up comparatively well. What degrades first is the procedural and decision-making layer — the flow that tells you which checklist, the mental model of a failing system, the sequence under load, the judgment about when to divert.

That should reframe how a pilot thinks about getting "rusty." The thing most likely to bite you after a layoff is not that you'll grease the landing a little worse. It's that the engine-failure-after-takeoff brief, the unfamiliar-airport decision, the partial-panel scan — the cognitive scaffolding — has quietly thinned.

Currency is a recency rule, not a measure of competence

Now hold the science next to the regulation. Legal currency in the United States is defined by recent experience. To carry passengers you need three takeoffs and three landings in the preceding 90 days; instrument currency requires six approaches, holding, and course intercepting and tracking within the preceding six calendar months (14 CFR §61.57).

Notice what those rules measure: that you did some things recently. They don't measure how well you did them, whether you'd handle the abnormal, or how much of your procedural knowledge has eroded since your last real workout. They are a proxy — a reasonable, administrable proxy, but a proxy.

This isn't a renegade reading of the rules. It's the FAA's own position. The agency deliberately stopped calling the flight review "biennial" precisely because the word implied that a check once every twenty-four months was enough to stay proficient, and it encourages currency training as often as a pilot's individual needs require (FAA AC 61-98E). The regulator is telling us, in writing, that the legal minimum is a floor, not a finish line.

So we have two things that the public — and, if we're candid, plenty of certificate holders — quietly treat as equivalent: current and competent. The research on forgetting says they come apart almost immediately after training, and the aviation-specific research says the part that comes apart first is exactly the part recency rules measure least well.

Does the gap actually hurt anyone? Here I have to be careful.

It would be easy, and wrong, to tell you that lapsed currency causes accidents. I can't show you a clean study that takes a population of pilots, measures days since their last meaningful practice, and plots accident probability against it. As far as I can find, that longitudinal aviation study doesn't exist yet.

What I can show you is a chain of reasonable inferences, each link sourced. General skill decays with disuse (Arthur et al.). In flying, the procedural and decision layer decays first (Childs & Spears). And when researchers look at why general aviation accidents happen, training deficiencies — especially a lack of recurrent practice in emergency procedures — show up as a leading contributing-factor theme. A 2025 systematic review of general-aviation accident factors ranked training deficiencies the second most prominent of five themes across the studies it pooled (Sheffield, Lee & Zhang, 2025, doi:10.1016/j.jairtraman.2025.102859). That's a review of contributing factors across GA accidents broadly, not proof that rust kills — but it points the same direction the decay literature does, and it points at the emergency and procedural skills the decay literature says fade first.

I'd rather give you an honest inference than a false certainty. The honest inference is this: the skills most likely to have quietly degraded since you were last sharp are the skills most implicated when things go wrong, and the rule that's supposed to keep you sharp doesn't actually test them.

What the science says to do about it

The good news is that the same research tradition that documents forgetting has spent fifty years documenting how to fight it, and the answer is robust enough that I'm comfortable stating it plainly.

Two effects do the heavy lifting. The first is spacing: practice distributed over time produces far more durable retention than the same practice crammed together. The canonical synthesis, Cepeda and colleagues' 2006 meta-analysis in Psychological Bulletin, pooled 839 assessments across 317 experiments and found that distributed practice reliably beats massed practice — and, importantly, that the best gap between sessions grows as the interval you need to remember over grows (Cepeda et al., 2006, doi:10.1037/0033-2909.132.3.354).

The second is the testing effect: the act of retrieving something — making yourself recall it, not just reread it — strengthens memory more than additional study does. Roediger and Karpicke showed that in 2006: when the final test came five minutes later, rereading won; but on delayed tests days later, the students who had been made to retrieve the material remembered substantially more (Roediger & Karpicke, 2006, doi:10.1111/j.1467-9280.2006.01693.x). It isn't a lab curiosity, either — the same advantage shows up for real curriculum remembered over months, as in a classroom study of eighth-graders retaining U.S. history facts on a test nine months later (Carpenter, Pashler & Cepeda, 2009, doi:10.1002/acp.1507). Put spacing and retrieval together and you have what learning scientists now simply call spaced repetition, an efficient and effective way to make knowledge stick (Kang, 2016, doi:10.1177/2372732215624708).

Translate that out of the journals and into the hangar and it says something concrete. Re-reading the AIM the night before a flight review is the cramming the research warns against. Pulling up an emergency memory item and forcing yourself to produce it from nothing, in short sessions spread across weeks, is the protocol the research endorses. The procedural knowledge that fades first is precisely the knowledge that responds best to spaced retrieval — which is a hopeful symmetry, if we're willing to act on it.

Why I built what I built

I'll declare my interest plainly: I founded FlyCurrent, and our practice engine is built on spaced retrieval — the spacing and testing effects above, scheduled deliberately rather than left to chance. I'm not going to dress that up as a guarantee of anything; the research doesn't license guarantees, and a physician who oversells a treatment loses the right to be believed. What I claim is narrower and I believe defensible: the recency rules define a legal floor, the science says proficiency erodes above that floor in predictable ways, and the most evidence-based countermeasure we have is spaced, effortful retrieval of the procedural knowledge that fades first.

You can do this on your own — and a few disciplined pilots do. But doing it well is harder than it sounds: knowing which items are about to fade, surfacing each one at the right interval, making yourself retrieve rather than reread, and sustaining that across the hundreds of things a competent pilot is supposed to hold. That scheduling-and-retrieval problem is exactly what we built FlyCurrent to solve — the spacing and testing effects, applied deliberately to aviation knowledge, so the procedural material that fades first is the material you keep practicing. I won't promise you an outcome the research can't underwrite. What I'll say plainly is that this is the evidence-based way to fight the curve, and we built the tool to run it for you instead of leaving it to willpower and the night before a flight review.

There's a second thing I'm after, and I'd rather be open about it. Remember the gap I admitted earlier — that no one has actually measured how a pilot's knowledge decays across the months between training, because the longitudinal data has never existed. Every time a pilot practices on FlyCurrent, with their consent and in de-identified form, we learn a little more about how aviation knowledge is genuinely retained and lost. Done responsibly, that becomes one of the most valuable datasets in aviation safety — the very curve this article had to reason around instead of cite — and it lets the platform measurably sharpen how it keeps each pilot current over time.

So here is the whole of it. The forgetting curve runs whether or not we acknowledge it. My hope is straightforward: that pilots who train this way gain knowledge instead of quietly shedding it, that they carry more of it into the cockpit on the day it counts, and that they are safer for it — which, in the end, makes the skies safer for all of us. If that is the kind of pilot you want to be, that is the reason FlyCurrent exists.

Fly current. From now on, FlyCurrent.

---

Stephen J. Ronan, MD, is a board-certified plastic surgeon, a pilot, and the founder of FlyCurrent. This article was reviewed for accuracy by Mike Kloch, CFI. Sources are linked above and listed in full at the foot of the page.