It is difficult to study panic in emergencies. In the immediate aftermath of a terrorist attack or natural disaster, the focus is on minimizing victims and damage — not collecting data on crowd dynamics. And re-creating these events in a laboratory raises a number of ethical questions. So how do researchers investigate crowd behavior in emergencies?
“It is impossible to study such situations experimentally.”
A team of scientists from Max Planck Institute for Human Development, Disney Research Zurich, ETH Zurich, and Rutgers University have created emergencies in virtual environments to investigate mass panic in the real world.
“In the study of crowd behavior, emergency situations are the least understood … because it is very difficult to collect data,” Mehdi Moussaïd, a researcher at the Max Planck Institute for Human Development, told Digital Trends. “In the past, we have analyzed a few camera recordings of places where an accident happened, but they are always difficult to evaluate.
“Furthermore, it is impossible to study such situations experimentally,” he continued. “It’s unethical and would be dangerous for our participants. Understanding precisely crowd behavior during emergencies with real-life data is a dead-end — at least with the existing methods.”
So Moussaïd and his team created a virtual environment with doors as bottlenecks, red blinking lights as stressors, and fires that block certain exits. They gave 36 participants control of their own avatar and let them observe the movement of avatars around them, but restricted them from communicating with each other.
“Everyone rushed to the exit, creating a panic-like movement.”
With the virtual fires lit, the researchers ordered the participants to evacuate the building. “Everyone rushed to the exit, creating a panic-like movement,” Moussaïd said. “Some participants did not make it out the building … but of course, in real-life, everyone was just sitting in front of a computer.”
Interestingly, participants in the virtual reality behaved a lot like people in response to real world emergencies. “We didn’t observe any significant difference between real and virtual settings,” Moussaïd said.
For example, most people in the virtual world pass on the right-hand side, just like people in real life. Participants did, however, move more slowly with their avatars then people tend to in reality — perhaps because the flight response is less intense when the danger isn’t virtual.
Moussaïd and his team hope that the data they collected will help other researchers understand crowd behavior but they are not done placing people in virtual danger. “We need to collect a large body of data about people’s behavior under a variety of emergency situations,” he said. “These will help us understand precisely what is going on. From there, we can start working on practical solutions to make stressful evacuations less dangerous.”
Urban planners can use the simulation to design their own virtual environments and test the safety of a building before it is built. By simulating the layout and analyzing a crowd’s response to an emergency, architects can locate bottlenecks and areas of hazard, potentially making buildings safer in worst-case scenarios.