Researchers stopped people to ask for directions, swapped in a different man while a door passed between them — and fewer than half noticed the person they were talking to had changed

On a sidewalk at Cornell University, a man holding a campus map stopped passersby and asked for directions to a nearby building. Ten to fifteen seconds into the conversation, two other men rudely shouldered a large door between him and the pedestrian. For about a second the door blocked the view. When it passed, a different man was standing there, holding the map, continuing the same conversation. He wore different clothes and spoke in a clearly different voice, and the two men differed in height by about five centimeters.

Fewer than half of the people noticed. In the first run of the experiment, only 7 of the 15 pedestrians realized they were now talking to a different person. The rest gave the stranger their directions and walked away, and were startled when the researchers explained what had happened. Asked afterward whether they had noticed anything unusual, many of them complained only that the door-carriers had been rude.

What the study was actually testing

The psychologists Daniel Simons and Daniel Levin ran the door study in 1998 to settle a specific argument, not to prove that people are oblivious. Earlier work had shown that observers often fail to spot large changes in photographs and films when the change is hidden by a blink, a flicker, or a cut between shots. Skeptics had a ready objection: maybe that only happens because a screen is passive and artificial, and in real life, fully engaged with another person, we would catch the change at once.

The door was a way to test that objection in the world. The conversation was live, the pedestrian was an active participant making eye contact and giving directions, and the substitution happened in the middle of a natural interaction. If real-world engagement protected against the effect, the swap should have been obvious. It was not. The failure that showed up on flickering screens showed up just as readily on the sidewalk, which told the researchers the effect was not an artifact of screens at all.

The name for it: change blindness

The phenomenon is called change blindness, the failure to notice a substantial change to a scene when the moment of change is masked. Our intuition insists we hold a detailed, photograph-like picture of what is in front of us, updated continuously. The door study, and a stack of experiments before it, suggest we do not. We hold a sparse sketch built around whatever we judge matters, and check that sketch for consistency rather than re-examining the world frame by frame.

Crucially, this is not a story about a faulty memory storing the scene and losing it. It is about what gets encoded in the first place. The pedestrians were looking right at their conversation partner. Attention, by itself, did not guarantee that they registered the features that would have flagged the man as a different person.

Who noticed, and why that is the interesting part

The most telling result is not the headline number but who fell on each side of it. Everyone who noticed the change in the first experiment was a student of roughly the same age as the two experimenters, somewhere around twenty to thirty. The people who missed it were older, roughly thirty-five to sixty-five.

Simons and Levin offered an explanation drawn from social psychology. We pay closer attention to the individuating details of people we see as members of our own group, and tend to process people we file as “other” in broad strokes, by category rather than by face. Their hypothesis was that the younger pedestrians saw the experimenters as peers and encoded them as individuals, while the older pedestrians saw them as generic young men and encoded only the type.

To test that, the pair ran a second experiment. The same two experimenters dressed as construction workers and repeated the routine near a construction site, this time stopping only younger students, the very group that had always caught the swap before. Now that the men read as members of a different group, the students stopped catching it. Only 4 of the 12 noticed the change. One woman who missed it described the mechanism herself, saying she had simply seen “a construction worker” and never coded the individual standing in front of her.

Everyone is sure they would catch it

Part of what makes the result land is that almost no one believes it could happen to them. Simons and Levin checked this directly. They read a class of about 50 introductory psychology students a plain description of the door scene, a lost man asking for directions, the rude door, a different person standing there afterward in different clothes, and asked who would notice the switch. By a show of hands, the students claimed without exception that they would.

That gap between confident prediction and actual behavior is part of why the study became a teaching staple. The conviction that we would obviously catch such a thing is exactly the intuition the experiment punctures. It is also a useful test the authors proposed for taking change blindness seriously: if naive observers are certain a change is unmissable and a real fraction of them still miss it, the failure is worth paying attention to rather than dismissing as a trick with trivial details.

How sure can we be from a sidewalk experiment

The door study is vivid, and vividness invites overstatement, so it is worth marking what it does and does not establish. The samples were small, 15 people in the first experiment and 12 in the second, the numbers a field study on a real sidewalk allows rather than the hundreds a lab survey might gather. The headline figures are best read as a striking demonstration, not a precise measurement of how often people miss such changes.

There are obvious objections, and the authors took them on. Perhaps subjects noticed the swap but were too polite to say so. Simons and Levin argued the opposite bias was at work: a few people who plainly had not noticed nonetheless claimed they had when asked directly, suggesting social pressure pushed reports up, not down. Perhaps giving directions distracted people from their partner, though the pedestrians appeared focused on the conversation, taking turns and making eye contact. These remain reasonable caveats rather than settled points.

The deeper limit is built into the design. The two experimenters were similar to begin with, both men of roughly the same age and appearance, so the change crossed no obvious category line. The researchers were explicit that a switch between, in their example, a short woman of one race and a tall man of another would almost certainly be caught, because it would change not just the details but the kind of person. Change blindness is about the details we discard, not about some general inability to see. We are not blind to the world. We are economical with it, and the door study is a measure of just how economical.

A small unsettling afterimage

What lingers is not the trick but the ordinary thing it exposes. The pedestrians were not careless or unusually unobservant. They were doing what everyone does, holding just enough of a stranger in mind to finish the exchange and move on. The door study turned that quiet economy into something you can watch happen, and left an uncomfortable question hanging over every brief encounter since: how much of the person in front of us are we actually keeping.