Try a small experiment before reading further.
On a scale of one to seven, how well do you understand how a bicycle works? Most people answer five or six. It is a simple machine, visible in its entirety, and nearly everyone has ridden one.
Now draw one. Specifically: draw the frame, and place the chain, the pedals and both wheels correctly.
When psychologist Rebecca Lawson ran this study, a large proportion of adults produced bicycles that could not exist — chains looping around the front wheel, frames with bars running straight through where the wheel must turn, pedals attached to nothing. Many of the same people had rated their understanding highly moments before. Confidence dropped sharply after the attempt.
The original studies
The effect was named by Leonid Rozenblit and Frank Keil at Yale in 2002. Their procedure was simple and has been replicated many times.
1. Ask people to rate how well they understand how something works — a zip, a helicopter, a flush toilet, a combination lock.
2. Ask them to write a detailed, step-by-step causal explanation of it.
3. Ask them to rate their understanding again.
Ratings fell reliably between step one and step three. People discovered, in the act of trying to explain, that what they had was not an explanation but a familiarity.
Rozenblit and Keil called it the illusion of explanatory depth: the belief that you understand a mechanism in far more detail than you actually do.
Why the illusion exists
We confuse recognition with comprehension. You can identify a toilet, use one, and picture one clearly. That vivid representation feels like understanding, but knowing what something looks like is not knowing how it works.
Explanations are stored hierarchically, and we only check the top level. You know a toilet involves water, a handle, and a drain. That top layer is genuinely correct, so nothing signals that the layers below it are missing.
Mechanisms are rarely tested. You get instant feedback on whether you can ride a bicycle. You get almost no feedback on whether you can explain one. Skills get corrected; explanations do not.
Knowledge is distributed. We live inside a community of knowledge — someone knows how the water system works, someone knows how the vaccine works, and society functions. The trouble is that the mind does not clearly mark the boundary between what you know and what is merely known by someone reachable. Access feels like possession, and search engines have made this dramatically worse.
The illusion is specific to mechanisms. Rozenblit and Keil found it was much weaker for facts, procedures and narratives. Nobody thinks they know more capital cities than they do. It is causal, multi-part systems where the illusion concentrates — which is unfortunate, because those are exactly the systems policy and technology are made of.
Where it gets expensive
Political opinion. In a widely cited study, Philip Fernbach and colleagues asked people to state their position on contested policies, then to explain in mechanistic detail how each policy would actually produce its intended effects. After attempting the explanation, participants rated their understanding lower and, notably, held their positions less extremely. Asking people to justify their position with reasons did not have this effect — only asking them to explain the mechanism did. Reasons are easy to generate; mechanisms are not.
Technology. Users routinely believe they understand what an app does with their data, what a model is doing when it answers, or what a system will do at the edges. Confidence in these judgements is largely unrelated to accuracy.
Business. Strategies are approved on the basis of a plausible-sounding narrative that nobody has been asked to trace step by step. "We'll increase engagement and monetise later" is a story, not a mechanism.
Learning. Students who reread notes feel fluent and confident, then perform poorly. Fluency of reading is mistaken for depth of understanding — the same illusion, wearing a different hat.
Everyday life. Home repairs, car problems, medication, insurance, contracts. In each case people frequently discover the boundary of their understanding only when acting on it.
The good news: it is easy to detect
Unlike most cognitive biases, this one has a reliable self-test that takes ninety seconds.
Explain it out loud, end to end, with no gaps. Not what it does — how it does it. If you find yourself using a phrase that summarises a step rather than describing it ("and then it processes the data", "and the pressure does the rest"), that phrase is covering a hole.
Teach it to someone who will interrupt. The Feynman technique in its useful form: explain it to a curious twelve-year-old who is allowed to keep asking why. Every question you cannot answer marks a boundary.
Write it down. Writing is less forgiving than thinking. Ideas that feel complete in the head reveal their missing joints on paper.
Draw the causal chain. Boxes and arrows, each arrow labelled with the actual mechanism. Unlabelled arrows are unexamined assumptions.
Predict something. Real understanding of a mechanism lets you say what will happen if you change one part. If you cannot predict the effect of a change, you have a description, not a model.
What to do with the discovery
The point is not to feel bad about the gaps. Nobody can hold deep mechanistic models of more than a handful of systems, and demanding otherwise would be paralysing. Relying on a community of knowledge is not a failure — it is how civilisation works.
The point is calibration. What the illusion costs is not ignorance but unwarranted confidence: strong opinions, firm decisions and public claims resting on models that would collapse under a single "and then what happens?"
Three habits follow naturally.
- Before arguing strongly for something, try explaining its mechanism once, silently. Notice where you go vague.
- When someone else explains, listen for the summarising phrases that stand in for steps.
- Treat "I understand this well enough to act, and here is exactly where my understanding ends" as the goal, rather than fluent confidence.
The short version
Feeling that you understand how something works is generated by familiarity, vivid imagery and top-level knowledge — none of which require the mechanism to be present in your head at all. The illusion dissolves the instant you attempt a complete step-by-step explanation, which is why that attempt is one of the cheapest and most useful intellectual habits available.
Try it on the bicycle. Then try it on the thing you are most certain about.
Be the first to share a thought.