Unthinkable: Who Survives When Disaster Strikes - and Why

by Amanda Ripley

  At the time of the fire, Norris Johnson and William Feinberg were sociology professors nearby at the University of Cincinnati. They were riveted by the news of the fire, like everyone else in the area. As sociologists, they were particularly curious to know how a group of strangers in tuxes and ball gowns had behaved during a sudden and merciless fire. Eventually, they managed to get access to the police interviews with hundreds of survivors—a rare and valuable database. “We were just overwhelmed with what was there,” remembers Feinberg, now retired. People were remarkably loyal to their identities. An estimated 60 percent of the employees tried to help in some way—either by directing guests to safety or fighting the fire. By comparison, only 17 percent of the guests helped. But even among the guests, identity influenced behavior. The doctors who had been dining at the club acted as doctors, administering CPR and dressing wounds on the grounds of the club like battlefield medics. Nurses did the same thing. There was even one hospital administrator there who—naturally—began to organize the doctors and the nurses.

  The sociologists expected to see evidence of pushing or selfish behavior. But that’s not what they found. “People kept talking about the orderliness of it all. It was really striking,” says Feinberg. “People used what they had learned in grade school fire drills. ‘Stay in line, don’t push, we’ll all get out.’ People were queuing up! It was just absolutely incredible.”

  Safety in Numbers

  Life is simpler when lived alone. That may be why far more creatures roam the planet alone than in groups. Male elephants, which have no real predators, do not bother with other elephants except to mate. They stomp across the land unburdened by others.

  Humans are not so self-sufficient. We mingle in groups our whole lives, and we cling to one another in disasters. After the terrorist bombings on the London transit system on July 7,2005, which killed fifty-two and wounded hundreds, some victims actually resisted leaving the tube station. “I needed the [others] for comfort,” one victim explained to U.K. psychologist Drury. “I felt better knowing that I was surrounded by people.”

  When children are involved, the reason for solidarity is plain. A species’ survival depends upon protecting its young, and human babies are more vulnerable longer than any other animal offspring. But we see the same behavior among adult groups. One study of mining disasters found that miners tended to follow their groups even if they disagreed with the group’s decisions. Grown men trapped underground would rather make a potentially fatal decision than be left alone.

  Why is that? Morality aside, is there any practical reason to value camaraderie as much as we do? Why don’t we behave like animals in these situations? Or do we? I called primatologist Frans de Waal to see if other mammals behave the same way in life-or-death situations. Maybe our own behavior is just a by-product of civilization—a charming but unnatural bit of gallantry.

  De Waal has written eight books on chimpanzees and spent decades observing them in captivity. Chimps and humans share about 99.5 percent of their evolutionary history, so the comparison can be instructive. When chimpanzees see a possible enemy, de Waal says, they gather close and start to touch each other. They might even embrace. In other words, they act a lot like humans. “A common threat has a unifying effect,” he says.

  Why do chimps band together? For one thing, de Waal says, the show of affection might intimidate the enemy. If the predator wants one of them, he will have to deal with all of them. Companionship also calms the chimps, making them better able to handle the stress of the threat. The same is true in humans. In laboratory experiments, people who are asked to complete tasks with a friend at their side exhibit lower heart and blood pressure rates than when they go through the tasks alone.

  Even before a challenge materializes, camaraderie has clear benefits. In the early 1980s, primatologist Carel van Schaik went to Indonesia to try to understand monkey groups. Van Schaik studied two communities of long-tailed macaques—one on the island of Simeulue, a blissful macaque paradise with no cat predators, and one in Sumatra, a far scarier place populated by tigers, golden cats, and clouded leopards. After watching both communities, van Schaik found a significant difference. On Sumatra, where there were more predators, the macaques traveled in much larger groups. The bigger the threat, in other words, the bigger the posse—and the more eyes, ears, and nostrils with which to detect a predator. On Simeulue, the monkeys traveled in very small groups—the smallest groups of virtually any macaque community ever studied. They simply didn’t need one another the way the Sumatra macaques did.

  Even lower-order animals band together in times of danger. Fish cluster in tight schools, and birds call to each other to warn of an approaching hawk. But they don’t do it to be nice. As evolutionary biologist Richard Dawkins has noted, “Anything that has evolved by natural selection should be selfish.” We help one another because we get benefits from doing so, if not immediately and directly, then eventually or indirectly. Evolutionary biologists call this “reciprocity.” And, in evolutionary terms, it means that an animal does something to improve its odds of passing on its genes—either by reproducing or by protecting its relatives.

  If I carry Lou Lesce’s briefcase down the stairs of the World Trade Center, I might not get anything tangible in exchange, but I might still get something. “Evolutionary biology has a hypothesis to account for small conspicuous acts of kindness,” says animal behavior expert John Alcock. “My guess is that the people who assisted others in the World Trade Center knew that others were observing. They were not calculating, but their desire to commit some nice act—commiserating, directing, guiding others to safety, might have had a substantial payoff in terms of improved reputation.” There may have also been something calming about helping others that day; it lent a sense of normalcy and orderliness to the abnormal and disorderly.

  Until about a fraction of a second ago, in terms of human history, we lived in small, extended-family bands. We all knew one another, from birth to death. So by aiding one another, we could build a reputation for being generous and helpful, which would encourage others to cooperate with us—which would in turn boost our chances of reproductive success. “I know this sounds cold and analytical to the average layperson,” says Alcock. “I’m not saying that people who do this are not motivated by impulses that are moral, ethical, and desirable. I’m saying that it’s precisely because these impulses are adaptive that we admire them as moral and ethical.”

  Sometimes what looks like altruism is actually quite the opposite. In an elegant paper published in 1971, biologist W.D. Hamilton described what he called the “geometry for the selfish herd.” When a dog runs after a herd of sheep, the sheep at the back of the pack will butt or jump his way into the ranks ahead of him—leading the herd to become more and more tightly packed together. From afar, it may appear that the sheep are banding together in a grand show of unity. In fact, each is just trying to avoid being eaten, by reducing his “domain of danger,” as Hamilton puts it. The animals on the edge of the herd are the easiest to pick off. So no one wants to be on the edge. Genes that help us avoid the edge are genes that are likely to endure through natural selection. So a human who feels compelled to stick with a group of strangers in a sinking ship may be doing it for all kinds of reasons—including, perhaps, a long-ago, deeply subconscious desire to avoid being eviscerated by a passing hyena.

  It’s difficult to know for sure whether humans had enough predators to need to form selfish herds. After all, humans’ greatest predators have always been other humans. But we make formidable predators, and our best chances of survival are usually improved by sticking together. Groupthink, then, is the adaptive strategy of prioritizing group harmony. Dissent is uncomfortable for the group because it can be dangerous to the individual. Sometimes, when we appear to value the group ahead of our own skin, we are actually doing something else altogether.

  The Wedding Party

  At the Beverly Hills, the waitress told McCollister, the young bride, that her
party had to leave. There was a small fire, but they could come back as soon as it was extinguished, she said. Hearing the news, McCollister turned to find her mother. As she turned, she noticed another waitress opening a folding wall in the center of the room. The opened partition revealed a wall behind it, entirely in flames. Then McCollister looked down the hallway and saw smoke flooding in like water. She felt a clarity of purpose, all of a sudden. “‘Everybody out, now,’” she said, ushering the guests out the French doors to the garden. “I put my arms out and was pushing people out the door, kind of like cattle, to show them where to go.” People were obedient and calm, she says, remarkably so. She felt filled with responsibility for her guests. “This is my party. They were there because of me.”

  A woman in an adjoining room appeared to be trapped. So one of her guests lifted up a chair to throw it against the glass and rescue the woman. McCollister’s brother-in-law stopped the man. “Don’t throw the chair. They will sue you,” he said. But then another man threw the chair anyway, and the woman ran out.

  McCollister was one of the last to leave the room. It seemed like a quarter of an hour had passed since they had learned of the fire, but it was probably just a couple of minutes. She and her guests got about seven feet into the garden when the room exploded behind them. The back of her dress was blackened. She turned around and stared at the inferno. “Most of us just sat there in shock,” she says. “I think this is where the groupthink happened. It looked like a movie. We just stood in disbelief.”

  But not everyone stayed with the herd. The band members reached into the room and began throwing instruments out of the fire. McCollister’s cousin ran back in and rescued a wedding gift, a Baccarat crystal jelly jar that McCollister still has. One bridesmaid, Kathy, sprang into action. She was training to be a nurse, so she started helping the medics who arrived. Then she jumped in the ambulance and kept helping, shuttling back and forth between the fire and the hospital. “I remember seeing her carry IV bags back and forth. I was so amazed. She just picked up and went at it,” McCollister says. “Her dress got torn, she lost her shoes, and finally she just took her dress off and wore a hospital smock. She was awesome, absolutely awesome.”

  McCollister herself picked up the train of her dress and wrapped it around her waist. She put her aunt and uncle in a jeep that was taking people away from the scene. Out of the corner of her eye, she saw one of her other friends leave with her husband. They had medical training, but they left. She was appalled. “You just don’t leave.” The women had been close friends, and McCollister was supposed to be in the woman’s wedding shortly after her own. But she never spoke to her again. The woman had abandoned the group.

  Along with her new husband, McCollister joined a human chain that had formed to pull people out of the Cabaret Room, just off the garden. This room in particular seemed to have no end of victims. Most of them were dead or dying when they came out. One stranger died in her husband’s arms. Finally a firefighter decided to close the doors to contain the fire.

  The Science of Evacuation

  For a long time, engineers assumed people would move out of a building like water. They would fill the space they had, coursing down the staircases and flowing out to safety like a river of humanity. Buildings were constructed accordingly. The problem is, people don’t move like water.

  Ed Galea has spent his career trying to understand crowd behavior in fires. He manages a team of mathematical modelers, behavioral psychologists, and engineers at the University of Greenwich, Old Royal Navy College Campus, in London. In his office in an ancient building alongside the river Thames, he has three framed photos of the burning World Trade Center. Another wall is decorated with eight pictures of train wrecks, plane crashes, and other assorted tragedies. If another man worked here, such a display might seem odd, even callous. But when I meet Galea here on a summer morning, it quickly becomes clear that he takes every disaster personally. I sit down amid the piles of videotapes and books, and we speak without interruption for four hours. Galea’s intensity is contagious. He knows more about human behavior in fires than almost anyone in the world, and he agonizes over the needless loss of lives.

  Galea is from Australia, where he trained as an astrophysicist. His specialty was using computer models to describe how magnetic stars are born. But not many people would pay him for this service. So he took a rather dull job as an industrial mathematician in the steel industry. He moved to the United Kingdom in the mid-1980s, and, not long afterward, an airplane caught fire during takeoff in Manchester, England. It was a bizarre accident. The Boeing 737 never became airborne, and firefighters quickly doused the plane in foam. There was no crash, just a fire. The last survivor was pulled out five minutes after the plane had come to a stop. And yet fifty-five people still died. Why hadn’t they gotten out? In standard evacuation tests, this type of plane had been evacuated in just seventy-five seconds. What had gone wrong? “I couldn’t understand how fifty-five people could have died. That caught my attention.”

  Fire modeling had only just begun, and the United Kingdom led the world. After London burned down twice—in 1212 and 1666—the country became a world model for fire safety. The first seminar on human behavior in fire was held at the University of Surrey in 1977. Galea managed to talk his way onto a project studying the crash at the University of Greenwich in London. But most of the early models were based on how fire moves in relatively small square rooms. And none could explain what happened on the Manchester flight. When Galea and his colleagues modeled the crash, they were amazed at how fast the fire could spread from one of the engines to the fuselage and then inside the cabin, filling the plane with black, toxic smoke. But that still didn’t explain the casualty rate on the flight. “We understood the fire, but why did so many people die? Why couldn’t they get out?”

  So Galea decided to create a model to explain not the fire but the people. It was a revolutionary idea. He took the idea to the U.K. Civil Aviation Authority, which had funded his original model. They turned him down. “They said, ‘This is impossible. You might be able to model fire but you can’t model people.’ To me, that was like waving a red flag in front of a bull.”

  Galea is a confident man. His website features nine photos of himself, including several of him receiving a prize for his work from Queen Elizabeth II in 2003. There are also pictures of his desk, from multiple vantage points. This is not a man to be easily put off. So Galea did what desperate professors do: he got some graduate students to work for free. For one year, they constructed a very crude model based on the scant research available into human behavior. It was called EXODUS. When Galea showed U.K. Aviation officials what he’d come up with, they agreed to give him funding—and have been doing so ever since.

  The problem with treating people like water is that water molecules do not experience pain or fear. Water molecules don’t make decisions, and they don’t stumble or fall. Human beings, on the other hand, fill a space unevenly, in clusters. They take shortcuts and pause to rest when they can. Once committed to a path, they don’t easily change course. Groupthink has a momentum of its own.

  EXODUS tries to treat people like people. Each evacuee receives a specific age, name, sex, breathing rate, and running speed, among other characteristics. Then EXODUS gives individuals behavioral capacities “so they can make decisions.” For instance, until EXODUS, models assumed that people would begin to evacuate as soon as an alarm went off. Of course, anyone who has ever heard a fire alarm knows this is not what happens. With EXODUS, evacuees have lives and brains. Before leaving, each performs certain tasks—like grabbing a briefcase or searching for a child. And they have the ability to see an exit sign—and follow it—or not. (In experiments with real human beings, Galea has found that many people simply fail to see exit signs, even when they are in plain view. It remains unclear why.)

  Most important, the newest version of EXODUS recognizes that people move in groups. That is a difficult behavior to model, which is partly why s
o few models have ever tried to do it. But it is essential. Galea and his colleagues have analyzed a database of 1,295 survivor accounts from plane accidents. About half of the survivors said they were traveling with someone else at the time of the accident.

  EXODUS helped Galea understand that passengers on the Manchester flight had not reacted like synchronized swimmers. Some remained frozen in their seats. Others climbed frantically over seat backs, while still others piled up at an overly narrow exit row, slowing the evacuation to a standstill. One passenger tried to open the exit door beside her, not realizing that she was actually yanking on her armrest. Human behavior, combined with the noxious mix of smoke, heat, and gases, meant that the passengers had very poor odds of getting out.

  Today, Galea’s software is used in thirty-five countries. Galea would prefer that it be used before a fire—before a structure is even built. But it is often used as a forensic tool during investigations. The country with the most licenses—and a history shot through with disasters—is Korea. The United States has been “very backward in adopting this kind of technology in the design stage,” he says. “It’s very dangerous.” Before 9/11, most U.S. buildings were constructed without the help of any evacuation modeling at all, he says. Now, models are in vogue, but they vary dramatically in quality. Many of them still treat people like water.

  I ask Galea when most architects, engineers, and regulators started taking human behavior seriously. “They’re still not,” he says. “We’re sometimes still told that EXODUS is too complex and has too much human behavior. They want to know, ‘If I have someone here, how long will it take to get out?’ They don’t want to know how they move or if they move in groups. These guys who build buildings don’t want to know about this.”