The Curse of Knowledge Is Why Experts Explain Things Badly
By Mira Lindqvist · Voice & Pace
In 1989, the economists Colin Camerer, George Loewenstein and Martin Weber ran an experiment in which some participants were given information others did not have: the actual earnings of real companies. They were then paid to predict what the uninformed participants would estimate those earnings to be.1 They could not set their own knowledge aside. Even with money on the line, their predictions drifted towards the answer they knew. The researchers called the pattern the curse of knowledge: once you know something, it is hard to accurately imagine not knowing it.
They also let participants trade with each other in a small market. Trading reduced the bias. It did not remove it.
The tapping study
The most famous demonstration came the following year, in a Stanford dissertation by Elizabeth Newton, and was later popularised by Chip and Dan Heath in Made to Stick.2 One group of participants, the "tappers," tapped out the rhythm of a well-known song, such as "Happy Birthday," on a table. A second group, the "listeners," tried to name the song.
Before the listeners guessed, the tappers predicted they would get it right about half the time. Out of 120 songs, the listeners identified three.
The tappers were not careless. While they tapped, they could hear the melody in their heads, with the words and the instruments. The listeners heard knocking. The tappers could not stop hearing what they knew, so they could not judge how little of it was coming through.
Experts are worse at this, not better
You might expect the people who know a subject best to be the best judges of how hard it is for a beginner. A 1999 study by Pamela Hinds found the opposite.3 Experts, intermediates and novices were asked to predict how long beginners would take to learn a task on a mobile phone. The experts were the most wrong, underestimating the difficulty more than people with less experience did.
Hinds also tried to fix it. She asked experts to think back to their own early experience, or to consider what makes the task hard for a beginner. Neither approach meaningfully reduced the bias.
It reaches past expertise
The curse is not limited to specialists. In a 2007 study, Susan Birch and Paul Bloom showed adults a version of a classic children's test: a girl puts her violin in one container and leaves, and someone moves it while she is away.4 Participants who knew where the violin had been moved were more likely to predict she would look there, even though she had no way of knowing. Knowing the answer bled into their judgement of what someone else would think.
A neighbouring finding is hindsight bias. In 1975 Baruch Fischhoff showed that once people learn how an event turned out, they overestimate how predictable it was, and how likely they would have been to predict it.5 Knowledge changes not just what you think, but what you believe you would have thought without it.
The same experiment, run on a room
An expert explaining their field to a general audience is running the tapping study on a room. They assume background the audience lacks, skip the connective steps that feel too obvious to state, and use field terms without registering that they are jargon. Arrogance and carelessness have little to do with it. They can hear the whole song, and the audience can hear the knocking.
The signs are recognisable once you know them:
- Undefined terms. A word you use every day at work is a foreign word to the room.
- Missing steps. "So obviously the next thing is..." where nothing about it is obvious.
- Answers before questions. A solution explained before the audience understands the problem it solves.
- Speed. Material you know well feels slow to you, so you rush the parts the audience most needs time for.
Why trying harder to imagine not knowing doesn't work
The consistent finding is that the bias is stubborn. Better-informed people in these studies did not fully correct for it when asked, when paid, or when prompted to recall their own beginner experience. Effort to imagine not-knowing has limited power against a bias this deep. Telling yourself "I'll keep the audience in mind" is not a plan.
What does work
The more reliable fixes stop relying on your intuition, because your intuition is the thing that has been compromised.
Test on someone who does not know. Give the talk, or just the explanation of the hardest idea, to a person outside your field. Watch where they look lost, and ask them to explain it back. Their version will show you exactly which steps you skipped.
Lead with a concrete example. An abstraction is easy for you to follow and hard for anyone else. A specific case gives the listener something to hold before the principle arrives.
Define as you go. If a term is necessary, give its plain meaning the first time, in a short clause, and keep using that meaning.
Build in checkpoints. In a talk, recap before you move on. In a meeting, ask what part needs more detail, rather than whether everything is clear. "Any questions?" invites silence. A specific question gets a real answer.
Slow down on the parts that feel easiest. The sections that feel too basic to dwell on are usually the ones the audience needs most.
References
- Camerer, C., Loewenstein, G., & Weber, M. (1989). The curse of knowledge in economic settings: an experimental analysis. Journal of Political Economy, 97(5), 1232–1254. https://doi.org/10.1086/261651
- Heath, C., & Heath, D. (2007). Made to Stick: Why Some Ideas Survive and Others Die. Random House. https://openlibrary.org/works/OL8423528W
- Hinds, P. J. (1999). The curse of expertise: The effects of expertise and debiasing methods on prediction of novice performance. Journal of Experimental Psychology: Applied, 5(2), 205–221. https://doi.org/10.1037/1076-898X.5.2.205
- Birch, S. A., & Bloom, P. (2007). The Curse of Knowledge in Reasoning About False Beliefs. Psychological Science, 18(5), 382–386. https://doi.org/10.1111/j.1467-9280.2007.01909.x
- Fischhoff, B. (1975). Hindsight is not equal to foresight: The effect of outcome knowledge on judgment under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, 1(3), 288–299. https://doi.org/10.1037/0096-1523.1.3.288