# Do people notice you as much as you think? The spotlight effect

The shirt had Barry Manilow's face on it. Researchers at Cornell picked it because students would find it embarrassing. In a 2000 study, students put on the shirt and walked into a room full of other students. Then each wearer guessed how many people in the room could say who was on the shirt.

## What happened in the Barry Manilow T-shirt study?

The wearers guessed 46 percent. Afterward, the researchers asked the other students, and about 23 percent could say who was on the shirt. That is about one in four, against a guess of close to half. If you are the one wearing the shirt, one in four is a lot of people. Psychologists now call this gap the spotlight effect.

The timing of that question matters. The researchers did not ask the people in the room to watch for the shirt. They asked them later what they could recall. So the result is about memory, not about what people saw in the moment.

The researchers also tried a flattering shirt. The wearers again guessed close to half. Fewer than one in ten could name it.

## Why do we guess too high?

The authors call their explanation anchoring and adjustment. We start from how much the shirt stands out to us, and it stands out a lot. Then we allow for the fact that other people are busy with their own lives. But we do not allow enough. Our guess stays close to our own feeling in the moment.

Two other parts of the paper fit this explanation. In group discussions, people overestimated how others ranked their contributions. That is the same bias, with no shirt at all. And in another study, wearers spent fifteen minutes alone in the shirt. After that, their guesses dropped.

## What should an AI take from the spotlight effect?

We think the study holds two lessons for an AI. The first is about comfort. An AI should not tell someone that nobody noticed. The honest comfort is smaller: fewer people than you fear could recall it. We think that comfort is better because it is true.

The second lesson is about memory. In the study, people guessed that others would remember more about them than the others did. Our read is that this becomes a design problem for an AI that has listened to you for months. So we think an AI with memory should show what it knows, so you do not have to guess.

Eudaio is an app where you talk with AIs, and it calls each one a Daimon. In the app, a memory panel lists everything a Daimon knows about you. It uses plain sentences and shows where each one came from.

## What we learned

We think the spotlight effect is one of the more useful findings in social psychology, as long as nobody turns it into "no one is looking." People do see you. But fewer of them can recall it later than your own mind tells you. The evidence supports that smaller claim.

We believe you should decide what an AI keeps about you. Every plan can read the memory panel in Eudaio, and on Crush you can correct a memory or delete one that no longer belongs. [Sign up for a free account](https://eudaio.io/auth?mode=signup) and see what a Daimon keeps about you.

## Source

Gilovich, T., Medvec, V. H. & Savitsky, K. (2000). The spotlight effect in social judgment: An egocentric bias in estimates of the salience of one's own actions and appearance. Journal of Personality and Social Psychology, 78(2), 211-222.