Don’t put your phone down. Just thirty seconds.
Explain how a fridge works. Not roughly, not “something to do with gas” or “the motor makes it cold”. The mechanism. How does the cold get in, what happens to the heat, why does the thing hum?
I’ll wait.
First comes the confidence. Of course you know how a fridge works; you open one every day. Then the attempt. There’s a compressor, and it somehow cools the air, and then –
And then nothing.
You’re not alone. In 2002 the psychologists Leonid Rozenblit and Frank Keil studied exactly this pattern at Yale University. They gave students a list of everyday things – zips, flush toilets, helicopters, sewing machines – and asked them to rate on a scale from one to seven how well they understood how each one works. The participants gave themselves middling scores, on average just under four. Then they were asked to write down the mechanism step by step.
After the attempt to explain, the ratings dropped by almost a point. Then came a detailed question – how do you open a cylinder lock without a key? – and the score ended up around one and a half points below where it started. “I have an idea” became “I have a hunch, nothing more”. The two researchers called the phenomenon the illusion of explanatory depth. The fridge is my example.
The finding has held up. Rozenblit and Keil also showed it for natural phenomena, and later work has confirmed it repeatedly. Only for political beliefs is the picture murky: early studies found the same effect, several replication attempts did not. For everyday mechanics it is undisputed.
Gaps in knowledge are universal. What stands out about this effect is the surprise.
Until the moment you start explaining, you feel competent, with a calm certainty that says: of course I know this. Only then does the gap become visible.
I once tried the experiment at a dinner. Eight adults, all with university degrees. I asked the table whether anyone could explain how a toilet flush works. First laughter, then silence, then twenty minutes of discussion that ended with someone pulling out a phone. Nobody could explain it. Everyone had believed they could.
Rozenblit and Keil looked into the overestimation. They had forty devices rated on five scales, among them these three: how familiar they are, how many parts they have, how many of those parts you can see in operation. Then they checked which of these measures predicted the overestimation. How familiar a device was did not predict it. Nor did the number of parts. It was the ratio of visible to hidden parts – a single value that explained almost half of the differences.
Whatever you can read off an object, you book as knowledge in your head. The zip lies open, teeth visible, slider too; as long as you’re looking, you don’t have to remember anything. Take it away and you notice how little of it was in you. In a fridge, most of it is hidden. The overestimation is therefore likely to be smaller than with the zip; the exercise still shows you the gap.
The illusion isn’t everywhere, either. The researchers tested four other kinds of knowledge. For devices and natural phenomena, self-assessment collapsed. For capital cities, film plots and recipes it stayed largely stable – people asked to describe how to bake chocolate cookies rated themselves accurately from the start.
That comes down to the structure of the knowledge itself. Factual knowledge has hardly any intermediate stages: you either know the capital of Portugal or you don’t. Explanatory knowledge is graded, with levels, layers, depths. That is exactly where the trap lies. We mistake “understanding a bit” for “understanding”, sense an outline and take that feeling for competence.
You have no direct access to how much you know. You only have access to what it feels like to know something.
That feeling lies.
Do the test again, this time with numbers. How does a toilet flush work? Before you answer, rate on the scale from one to seven how well you know. Write the number down. Then explain it, out loud or in writing. What happens when you press the lever? Where does the water come from, where does it go? What stops the cistern from overflowing? Now rate yourself again and write the second number next to the first.
If you were honest, it’s lower. For most people by about a point. You have just watched the illusion at work in yourself.
The fridge does nothing to help you with your illusion. It just stands there. The systems that explain things to you today come with the explanation built in.
Ask yourself the same question after the next explanation one of them gives you. Ask a language model, for example, how it arrived at its answer: you’ll get a neat derivation, step by step. What those steps have to do with the calculation behind them is the subject of the chapter on the eloquent machine in my book (in German).
The full essay on this topic:
