Skip to content

← Psychology

Crowds and bystanders

Chapter 25 · Other people

In this chapter 5 sections
  1. Thirty-eight witnesses
  2. What happens when you aren’t alone
  3. Nine times out of ten
  4. Working together
  5. When it spreads through a school

There are two stories people like to tell about crowds. In the first, a crowd is a kind of beast. In 1895 a French writer called Gustave Le Bon put it in one sentence: a man alone “may be a cultivated individual; in a crowd, he is a barbarian”, a creature acting by instinct. In the second story the crowd does the opposite. It doesn’t riot; it stands and stares. Someone is in trouble, dozens of people see it, and everyone waits for someone else.

The two stories look like opposites, but they agree on one thing: among many people, you become less of a person. Less thoughtful in the first, less caring in the second. Both turn out to be too simple, and the truth is more hopeful than either.

Here it is in plain terms. When other people are around, each one of us is slower to step in and help, because the job feels shared. But a crowd is many people, and only one has to move. So in real emergencies, someone nearly always does. Most of this chapter is how psychologists found out both halves of that, and then what else happens to people in groups: why we pull less hard on a shared rope, how a committee of clever people can talk itself into a bad plan, and how a strange fear can spread through a school.

Thirty-eight witnesses

The second story has a name and a date. Early on the morning of 13 March 1964, at about a quarter past three, a young woman called Kitty Genovese was driving home from her job as a bar manager in Queens, in New York. She was 28. After she parked, near her apartment building, a man attacked her with a knife. He left, and came back, and found her again in a hallway at the back of the building. She died on the way to the hospital.

The New York Times first gave the murder four short paragraphs. It was one of 636 in the city that year. Two weeks later the paper put it on the front page, and this time the story was different. For more than half an hour, it said, 38 respectable, law-abiding citizens had watched a killer stalk and stab a woman, and not one of them had telephoned the police while it went on. One witness was quoted: “I didn’t want to get involved.”

That number, 38, stuck. Preachers, professors and newspaper writers searched for the reason, and the story went into the textbooks as a parable about modern city life: people packed together, and nobody caring.

It wasn’t true, or not like that. Researchers who went back to the court records found no evidence of 38 witnesses, no evidence that they had watched the murder, and no evidence that the witnesses had all done nothing. In 2016 the Times itself said its famous story had grossly exaggerated. Perhaps a dozen people heard something, though the number was never settled. Almost none could have seen the attack from beginning to end. Only a few glimpsed parts of it or understood the cries as a call for help; many thought they were hearing a drunk or a couple quarreling. There were two attacks, not three. Afterward, two people did call the police, and some neighbors later said a call was made even after the first attack. And a neighbor went out into the hallway and held Kitty Genovese in her arms as she was dying.

Some neighbors did ignore what they heard, and that is bad enough. But the story of 38 people coolly watching was the newspaper’s, not the night’s. Still, it set two psychologists in New York wondering, and what they found in their laboratory was real.

What happens when you aren’t alone

John Darley and Bibb Latané didn’t believe the easy explanation, that city people had simply stopped caring. They suspected something about the number of people itself. So in 1968 they built a test.

A student arrived for what she was told was a discussion about the personal problems of college life. She was put alone in a small room with headphones and a microphone. To spare everyone embarrassment, she was told, the students would talk over an intercom, each in a separate room, each in turn for about two minutes, with only one microphone on at a time. The experimenter wouldn’t listen.

In fact the other voices were tape recordings. The first speaker mentioned, hesitantly and with obvious embarrassment, that he was prone to seizures, especially when he was studying hard or had exams. When his turn came round again, his speech grew loud and confused. He asked for help. He said he was going to die. He choked, and went silent.

The student could do nothing for him through the headphones. To help, she had to leave her room and tell the experimenter. What changed from one student to the next was how many people she believed were listening. Some thought they were the only one. Some thought one other student could hear. Some thought four others could.

Thought they were the only listener85%average time to report: 52 secondsThought one other person was listening62%average time to report: 93 secondsThought four others were listening31%average time to report: 166 seconds0%25%50%75%100%
Figure 25.1 The share of students who went for help before the seizure ended, by how many other listeners they thought there were, and how long they took on average. From the table in Darley and Latané’s paper.

The difference was large. Of the students who thought they alone could hear, 85 out of 100 went for help before the seizure ended, and they went quickly. Of those who thought four others could hear, only 31 out of 100 went in that time, and they took about three times as long. Every student who believed she was alone reported it in the end. Of those who believed four others were there, only six in ten ever did.

What mattered was only the number. It made no difference whether the other listener was a man or a woman, or a medical student who might know what to do. Tests of the students’ personalities didn’t predict who would help. And afterward, the students in the larger groups said the other people hadn’t affected them at all.

The most telling part is how the students who stayed in their rooms behaved. They weren’t calm or uncaring. When the experimenter came into the room to end it, they asked, “Is he being taken care of?” and “He’s all right, isn’t he?” Their hands were trembling and their palms sweating; if anything, they seemed more upset than the students who had acted. They were stuck. They were afraid of making fools of themselves by overreacting, or of spoiling the experiment by leaving their room, and they felt they would be ashamed if they didn’t help, and while they hesitated, the time ran out.

Darley and Latané called it diffusion of responsibility. Alone, the whole job of helping is yours. With others there, it is split, and so is the blame for not doing it. And there is always the comfortable thought that somebody else must be doing something already. This became one of the best known findings in psychology, the bystander effect, and it has been tested again and again. Put the results of more than a hundred such tests together, with over seven thousand people, and the effect holds. It also has limits. It gets weaker when the situation is plainly dangerous, when there is a person doing harm in front of you, or when helping means real physical risk. And it gets weaker when the bystanders aren’t strangers to each other. Friends don’t wait for each other the way strangers do.

Nine times out of ten

There is a part of the 1968 paper that the famous version leaves out. Darley and Latané did a sum. Suppose each of five people is less likely to help than one person alone. What is the chance that at least one of the five helps? In their own experiment, the victim was about as likely to get help from two listeners as from one, and in the first minute more likely to get it from one or two than from five. But after the first minute, the chance that somebody came was high in every group. Each person in a crowd is less likely to act; a crowd still has many people who might.

For decades that second half was hard to test outside a laboratory. Nobody can stage a real fight in the street to see who steps in. But cities now have cameras, and in 2020 a team of researchers watched what they had recorded.

They took 219 arguments and fights caught on the cameras that are watched over the centers of three cities: Amsterdam in the Netherlands, Cape Town in South Africa, and Lancaster in England. They counted everyone in view as a bystander, on average about sixteen people. Then they noted who stepped in: anyone who made a calming gesture, put a hand on someone’s arm, got between the two, pulled one away, comforted the person being hurt, or helped in some other practical way.

In nine fights out of ten, at least one person stepped in, and usually several did. It was the same in all three cities, though people in Cape Town say they feel much less safe in public than people in Amsterdam do. And the more people there were around, the more likely help became, not less.

Is that the opposite of Darley and Latané? No. The cameras show what happens to the victim, not what goes on in each bystander’s head. Each person in a big crowd may well feel less pull to act, just as the students did. But one person in a crowd only has to move, and in a big crowd there are many who might. The two findings fit together, and the researchers say so: more people may make each one less likely to step in, and still make a bigger pool of people who might.

The cameras have their limits too. They have no sound, so a shout of “Stop!” wasn’t counted. They were in city centers full of shops, restaurants and bars, so they can’t say how often people step in elsewhere. And a camera shows what happened, not why.

The other old story, the crowd as beast, has also worn thin. Real panic, people trampling each other in a mindless rush, turns out to be rare. When bombs went off in London in July 2005, the survivors and witnesses described something else: commuters who were nearly all strangers, helping each other. The ones who helped often felt that the shared danger had made them, for a while, one group. People in crowds don’t lose themselves, as Le Bon thought. They often find a “we”.

Working together

Crowds aren’t the only groups that change how people behave. So do teams.

In the 1880s a French farm engineer named Ringelmann had men pull on a rope, alone and then together, and measured the force. A group pulled with less than the sum of its members pulling alone, and the bigger the group, the bigger the shortfall. Ringelmann himself put it down mostly to timing: people pulling together don’t all heave at the same instant, though he knew their effort might flag too. Much later, in the 1970s, researchers blindfolded students and told them others were pulling with them, when in fact each was pulling alone. They still pulled less hard than when they knew they were on their own. No timing problem could explain that. People simply put in less when their effort disappears into a group’s.

This came to be called social loafing. Like the bystander effect, it has been tested many times: in shouting, clapping, swimming, thinking up ideas and even judging poems. Put 78 of those studies together and it turns up almost everywhere, in men and women, and in studies from the West and from East Asia. It was smaller among women, and smaller in studies from China, Japan and Taiwan; the reviewers’ guess was that women, and people in those countries, may put more weight on the group’s goals. But it was there in all of them.

It also has a cure, or several. People stopped loafing when their own share could be seen and judged, when the task mattered to them, or when they cared about the group itself. Anyone who has done a group project at school knows the shape of this. The work goes better when everyone has a named part.

Groups can go wrong in a bigger way too. In April 1961, a force of Cuban exiles, trained and paid for by the American government’s intelligence agency, the CIA, landed at the Bay of Pigs in Cuba, planning to overthrow the new government there. The plan had been drawn up under one president and approved by the next, John F. Kennedy, after he talked it over with his advisers. Within two days the invaders were defeated.

How had a room of very able people agreed to such a plan? In 1971 a psychologist called Irving Janis offered an answer, groupthink. A close-knit group, he argued, can want agreement so much that people keep their doubts to themselves. Each silence makes the plan look more agreed, and so the next person keeps quiet too. Janis suggested ways out: one member given the job of arguing against the plan, a leader who doesn’t say at the start what he hopes for, and experts from outside invited to challenge the group.

The idea caught on, far beyond psychology. But it hasn’t held up as firmly as the bystander effect. A quarter of a century on, a review of the research found only a small body of real tests of it. And when one scholar went back to the documents released since, and to the memoirs, on the Bay of Pigs and on the Vietnam War, he argued that the presidents’ choices were shaped more by politics, by what each thought his choices would cost him politically, than by a cozy group silencing its doubts. Janis’s advice is still sensible. Whether his explanation is right, for the cases he chose, is argued.

When it spreads through a school

Sometimes what passes through a group is not a choice but a feeling, and in Nepal it has a name. In a school, one student, usually a girl, falls to the ground in the classroom or the yard, writhing, moaning, shouting and crying. Over the next days others follow. Some say they saw a dark figure trying to strangle them or drag them away, or the spirit of a woman who had died by suicide or some other unnatural death. Afterward they often remember little of it. People call it छोप्ने or छोपुवा, from छोप्नु, which can mean to seize or pounce, and also to faint or to be met by a spirit. Episodes can go on for weeks or months. They end by themselves, or after a traditional healer’s ritual, or after counselors come to talk with the students.

Outbreaks like this have been recorded in many places, in schools, factories and close communities, and doctors call them mass psychogenic illness: real symptoms, felt by real people, that spread through a group without a germ or poison to cause them. It is a description of how the illness travels, not a judgment of those who suffer it, and certainly not a claim that anyone is pretending. The fear is real, the falling is real, and so is the distress. For the families, a spirit is a serious explanation, held by the community, and a healer’s ritual is a way the community cares for its children.

Researchers from McGill University in Canada counted more than 130 schools in Nepal with outbreaks in about ten years, in at least 40 of the 77 districts. In 2015, three or four months after the great earthquakes, they went to twelve rural public schools in Sindhuli, Sindhupalchowk, Dang, Dolakha and Ramechhap. They talked with 194 students who had been affected, aged 11 to 18, and with 190 who hadn’t, mostly close friends the affected students named themselves. Nearly all of both groups were girls. A student counted as affected if teachers, families, healers or the student herself said so, and epilepsy was ruled out only by asking.

They looked for what set the affected students apart. Some things went with being affected: having been hit or neglected at home, living in a smaller family rather than a large joint one, depression, the after-effects of frightening events. The strongest link was with how readily a student responds to suggestion. In the test they used, a person listens to a set of eleven suggestions meant to bring on a hypnotic trance, and then says which ones she found herself carrying out. It measures a way of experiencing things that some people have more than others.

But when the researchers put everything together, the pattern didn’t hold up. Their full set of measures could not reliably say which students would be affected and which wouldn’t. And among the friends who hadn’t been affected, many were struggling too: about four in ten scored high for post-traumatic stress, and about a quarter for depression.

So the researchers could not say why one girl fell and her friend didn’t. Their ideas are offered as ideas. Hitting children to discipline them is common in Nepal, and the researchers wondered whether an outbreak may be one way children show a suffering they can’t put into words, at school, a place safer than home, where they can ask for help. They also wondered about families getting smaller as parents go abroad to work, leaving children with less support. And they pointed out that such outbreaks are often seen in times of fear and rapid change. Their own conclusion was that such outbreaks may involve more than something in a few individuals: things in the group, and in what it was going through, may be at work too.

All of this points the same way. A crowd is not a beast, and it is not a herd of people who have stopped caring. People in groups still think and feel. What a group changes is who feels the job is theirs. Alone, the job of helping is obviously yours; with others there, it is easy to wait, or to pull a little less, or to keep a doubt to yourself. Knowing this is the best way to beat it. Darley and Latané ended their paper with that hope: people who understand what makes them hesitate may overcome it. If you are ever one of many watching someone in trouble, the others may be waiting just as you are. Usually, someone moves. It might as well be you.

Found a mistake? Please tell us.