Sources checked 31 October 2026. The study figures below are the published averages; the scenarios are worked examples, not advice or a diagnosis.
Maya stands up to present to her team. Inside, an alarm is going off. Her heart is pounding, her hands are a little shaky, her voice feels thin. She is sure everyone can see it. She imagines the room reading her like a billboard: she's nervous, she's not ready, she doesn't belong up here.
Out in the room, people see something else. A colleague with slides. Maybe a slightly fast first sentence. Then they start thinking about the chart, or lunch, or their own presentation next week.
The gap between those two pictures has a name. Psychologists call it the illusion of transparency, and it is one of the more useful biases to know about, because knowing about it seems to help.
What is the illusion of transparency?
The illusion of transparency is the tendency to overestimate how much of what we feel inside is visible to other people. Nervousness, disgust, guilt, alarm: we experience them loudly, so we assume they leak out just as loudly. Usually they leak out much less. [1]
The name comes from a 1998 paper by Thomas Gilovich, Kenneth Savitsky and Victoria Husted Medvec. They opened it with Edgar Allan Poe's The Tell-Tale Heart, in which a murderer, chatting calmly with three police officers, becomes convinced they can hear his victim's heart beating under the floor. It is his own heart he hears. He confesses. The officers had suspected nothing. [1]
The key word is overestimate. The research does not say that feelings are invisible. It says that people reliably think their feelings are more visible than they are.
The lying game: what the 1998 study found
In the first study, Cornell students played a lying game in small groups of five. Each round, every player drew a card and answered a question out loud in front of the others. One card per round told its holder to lie, for example to claim they had met a famous TV host. Everyone else tried to spot the liar, and a prize went to the best lie detector. [1]
After each round, the liar estimated how many of the other four players would pick them out. The answer, averaged over eight sessions: liars expected 48.8% of the others to catch them. In fact 25.6% did. With one liar among four suspects, pure guessing would find the liar 25% of the time. In other words, the observers did about as well as chance, while the liars felt nearly twice as exposed. [1]
Try it before you look:
Published averages from Gilovich, Savitsky & Medvec (1998), Study 1a. Chance was 25%.
The same pattern came back in two variations. In Study 1b, liars expected 50% of the group to catch them and 27% did. In Study 1c, they expected 44.3% and 32.4% did; observers who watched the same lies guessed that only 25.3% of players would spot them, much closer to the truth. [1]
It is not just lying. In Study 2a, people tasted drinks on camera, some of them deliberately foul, and tried to keep a straight face. Tasters guessed that about 4.9 of 10 observers would spot which drinks were nasty. On average 3.6 did, which was not reliably better than chance (3.3). For pleasant drinks, where there was little to hide, their guesses were about right. [1]
Why your nerves feel so visible
The authors' explanation is anchoring. When you try to imagine how you look from outside, you start from the only thing you have direct access to: what you feel. Then you adjust for the fact that other people can't feel it. The adjustment is real, but too small. You end up somewhere between "they can see everything" and the truth, and usually closer to the first. [1]
There is a simple reason the anchor is so heavy. You get the whole signal: the pounding heart, the tight throat, the thought "I'm going to forget the next slide." Your audience gets a thin slice of it: a face, a voice, a pair of hands. Much of what you feel never reaches the surface at all, and what does reach it is easy to read as something else. A quick sentence can be enthusiasm. A pause can be thought.
What speakers think and what audiences see
In 2003, Savitsky and Gilovich took the idea to public speaking. Forty Cornell students were paired up and each gave a short, unprepared speech while the other watched. Afterwards each rated, from 0 to 10, how nervous they had felt, how nervous they thought they had looked, and how nervous their partner had looked. [2]
On average, speakers felt moderately nervous (5.95). They thought they had looked even more nervous than that (6.55). Their partners rated how nervous they had looked at 5.25. And 27 of the 40, about two in three, said they had looked more nervous than their partner. In eight pairs both people said so. [2]
Published counts from Savitsky & Gilovich (2003), Study 1. The 'no bias' view is a worked example: in pairs, about half should say they looked more nervous.
Notice what this does not say. The partners did see some nervousness; 5.25 out of 10 is not zero. The gap is between how visible people believed their nerves were and how visible they turned out to be. The authors also repeated the study at Williams College with observers who did not have to give a speech themselves, to rule out the idea that partners were too nervous to notice anything. [2]
Does knowing about it help?
This is where the illusion becomes useful. In a second study, 77 Cornell students had five minutes to prepare a speech that would be videotaped. Before speaking, they were split into three groups. [2]
Control (20 speakers): no extra instructions.
Reassured (18 speakers): told it is normal to be nervous and not to worry about what other people think.
Informed (39 speakers): told about the illusion of transparency: that speakers feel their nervousness is obvious, but audiences pick up on it much less than speakers expect.
Observers who didn't know which group a speaker was in then rated the videos. The informed speakers were rated as more composed and as giving better speeches. Simple reassurance did not produce the same effect. [2]
Published means from Savitsky & Gilovich (2003), Study 2, on a 1–7 scale (higher = better).
The authors' reading is that speech anxiety can feed on itself. You get nervous, you worry that the nervousness shows, and that worry makes you more nervous. Knowing that it mostly doesn't show breaks one link in that loop. Reassurance tells you not to care what people think, which is hard to do on command. The informed instruction tells you something more specific and more believable: they can't see as much as you think. [2]
What the studies don't tell you
These are good, clever experiments, and they are also small and specific. Keep the limits in view.
The samples were small and similar. A few dozen to about a hundred American college students per study, mostly at one university, in the late 1990s and early 2000s. Treat the exact sizes of the gaps as what happened in these rooms, not as universal constants.
The situations were mild. A lying game for an ice-cream coupon, foul drinks, short speeches to strangers. Panic attacks, visible blushing or trembling, and very strong emotion may well be easier to see.
Partly visible is still visible. Observers in the speech study still noticed some nervousness. The lesson is not "nobody can tell"; it is "they can tell less than you feel."
Knowing is not a cure. The informed speakers did better on average, by about three-quarters of a point on a seven-point scale. That helps. It is not a replacement for practice, preparation, or professional help if anxiety is stopping you from doing things you need to do.
A related demonstration makes the same point in a different way. In a study described in the 1998 paper, people tapped the rhythm of a well-known song on a table and guessed how many listeners would name it. Tappers hear the melody in their heads while they tap; listeners hear only knocks. Tappers overestimated the listeners badly. [1] Your nerves are like that melody: vivid to you, mostly knocks to everyone else.
How to use it before you speak
The practical version is short. The next time you're about to present, ask for a raise, or walk into an interview, remind yourself that the alarm is mostly inside. You will feel more exposed than you are. Feeling exposed is not the same as being exposed.
Some ways to make that reminder concrete:
Name the illusion, not the fear. "This is the illusion of transparency" is easier to believe than "don't be nervous."
Ask what they can actually see. Not your heart rate, not your thoughts. A face, a voice, your hands. Steady the parts that are visible, like a slower first sentence, and let the rest be.
Don't announce it. Saying "sorry, I'm so nervous" hands the audience information they probably didn't have.
Check afterwards. Ask one person how you came across. Most people are surprised by the answer.
Example wording based on the studies above. It is not therapy and does not replace help for severe anxiety.
You can't switch the alarm off on demand. But you can stop believing that everyone else hears it too.
Sources
Gilovich, T., Savitsky, K., & Medvec, V. H. (1998). The illusion of transparency: Biased assessments of others' ability to read one's emotional states. Journal of Personality and Social Psychology, 75(2), 332–346. doi:10.1037/0022-3514.75.2.332. Lab studies with Cornell undergraduates (lying game in groups of five, foul-tasting drinks, a simulated emergency); small samples and mild, artificial situations.
Savitsky, K., & Gilovich, T. (2003). The illusion of transparency and the alleviation of speech anxiety. Journal of Experimental Social Psychology, 39(6), 618–625. doi:10.1016/S0022-1031(03)00056-800056-8). Study 1: 40 Cornell students in pairs (plus a replication with 17 Williams College speakers); Study 2: 77 speakers and 40 observers. Short, unprepared speeches to strangers; it did not test long-term effects or people with diagnosed social anxiety.
Watch the episode
One email when we publish. Research, product decisions, and what teams report back.
