Current Issue
This Month's Print Issue

Follow Fast Company

We’ll come to you.

3 minute read

Virtual Worlds Are Helping Decode Stampedes And Mass Panic Behavior

Panic is hard to measure, but it turns out it can be simulated.

It’s hard to study mass panic in the real world. When people panic and stampede, there usually isn't anyone around to observe and record the proceedings. That’s why researchers at the Max Planck Institute decided to move their emergency situations into the virtual world.

In a series of experiments designed to observe how crowds behave in mass panic situations, 36 participants were asked to control an avatar and navigate it through a virtual building. The building was on fire, and the stress levels were increased to mimic real-world panic by introducing a time limit tied to financial rewards. If the participants took longer than 50 seconds to exit the building, they stood to lose points and, therefore, money. "Further stress-inducing elements were poor lighting, red blinking lights, and fires at the blocked exits," says the Max Planck Institute’s Kerstin Skork. In addition, other behaviors would lose points for the participants—bumping into other players, for example.

The behavior in the video looks instantly familiar. You see individuals milling around and blockages occurring in the most dangerous spots: places where bottlenecks are formed by the building’s design, dead ends, and points where participants were forced to make decisions.

The reality level was increased yet further by giving some group members directions. These lucky folks saw a guiding arrow at the the top of the screen. All of the participants knew about this, but they didn’t know who could see the arrow. This mimicked the presence in a crowd of people who know the building, and know the best escape routes.

"One crucial aspect of crowd dynamics lies in the social interactions that take place between individuals," say the authors. "These interactions create feedback loops and amplification effects and give rise to self-organized macroscopic patterns."

The simulation was also run without the stressors and this led to some interesting differences. For instance, "In the absence of stress, participants tended to keep reasonably safe distances from their neighbors in order to avoid the collision penalty," says the report. This means that there were almost zero body contacts. In the regular escape, the participants were prepared to sacrifice points in order to jostle their way to a faster exit.

Herding was also studied. The experiments found that people were more likely to follow their neighbors when stress levels were high. This wasn’t down to a psychological cause, though. It was just because the place was so crowded. In a stampede, you’re always following somebody, even if you try not to.

Other observations include our learned tendencies of how to behave in a space. Ninety five percent of participants favored moving to the right to avoid somebody. Previous studies have shown that Europeans "tend to intuitively walk on the right-hand side," writes Skork.

Despite some limitations with the current simulation—flow rates don’t always match up with known real-world data, for example—the researchers are happy that the simulations can provide essential insight into how we behave in emergencies.

One thing missing from the experiments, though, is real interaction between subjects. In a real fire, people familiar with the building might call out directions to others. "As past research has shown," says the paper, "social identification among individuals tends to promote inter-individual cooperation and enhance the efficiency of emergency evacuations."

By contrast, the design of this virtual experiment put people in cubicles, which "could have encouraged participants to behave in a more competitive manner." Future experiments, then, should allow people to communicate with each other.

Eventually, we could design buildings to better suit the panic instincts of the people that will use them, resulting in a better chance of escaping even during a panicked stampede. This research could also help in regular crowd-control situations, such as getting people out of sports stadiums after a game, or on and off planes faster.

Have something to say about this article? You can email us and let us know. If it's interesting and thoughtful, we may publish your response.

The Fast Company Innovation Festival