HMN 2026: How choices made by crowds in a train station are guided by strangers

In crowds, most people are strangers to you, and everyone else for that matter. However, until now, the effect of stranger-to-stranger interactions on the choices people make in crowds has not been properly examined. Ziqi Wang and Federico Toschi from the TU/e Department of Applied Physics and Science Education, along with Alessandro Gabbana at the University of Ferrara in Italy, explored how strangers influence people’s choices in crowds at Eindhoven Centraal railway station. The research is published in the journal Proceedings of the National Academy of Sciences.

“Using a collection of special overhead sensors, we gathered data on how pedestrians move over a three-year period, from March 2021 to March 2024,” says Toschi. “This amounted to about 30,000,000 pedestrian trajectories and included people getting off trains and those waiting on the platform. We collaborated with ProRail on this project, as we have done in previous studies on how pedestrians move in Eindhoven Centraal station.”

Toschi has been studying pedestrian dynamics for some time and was jointly awarded the 2021 Ig Nobel Prize for physics for work on how pedestrians keep a certain distance from each other in crowds.

Pedestrian choices on track four

All 30,000,000 pedestrian movements were not included in the study. “We were only interested in a certain group of pedestrians for this study,” says Wang. “We focused on passengers who left trains on track four at specific locations close to a platform kiosk. Then we looked at the paths taken by passengers around the kiosk and how their choice is influenced by strangers in front of them.”

The researchers were left with about 100,000 passengers who exited from door zones close to the kiosk and who then made a choice about whether to follow one of two paths (Path A or Path B). Density of the pedestrians along the paths depended on the door zone.

“When passengers exit zone L3, there is a more symmetric use of both paths as the length of the paths are similar in length,” says Toschi. “But, of course, this is a real-world setting, and so many things can play a role, such as what passengers can see, if they know the layout of station and platform, or if they are in a rush. Thanks to the large number of passengers in our data and the focus on three door zones, these effects are minimized.”

Effect of the stranger

So, how did the researchers identify if passengers disembarking trains were strangers to each other and not part of the same group?

“We developed a mathematical detection method that looked at how close people were to each, and how much they matched each other’s speed and direction,” says Wang. “Detected groups were normally two people, one member and one leader.”

With groups identified, the researchers then looked for interactions between strangers, and the results startled the researchers. “We didn’t expect such a strong tendency for passengers to follow the person in front of them, particularly when they didn’t know them at all,” noted Toschi.

To validate the results from Eindhoven Centraal station, the researchers built a numerical model which confirmed that people imitate the paths of strangers in front of them.

For Wang, it was fascinating to see three years of human behavior summarized in visual patterns. “This was my first time looking at data on how people move. The ‘stranger-following effect’ ended up explaining that crowds in the train station don’t always act rationally or efficiently.”

Avalanches!

Motivated by the discovery that strangers follow strangers, the researchers checked if decisions to follow strangers were clustered together. Put simply, when one person follows a stranger in front of them, are they then followed by a stranger themselves? And how large are these choice “avalanches,” as they are known?

“We were curious to see if the choice of a person to take path B could display avalanches,” says Toschi. “We are studying similar clustering or avalanching in dense emulsions as models for (induced) seismicity. So, we would like to study the physics of avalanches in my research group.”

Analysis of the data revealed that once a pedestrian opted for Path B, other pedestrians were more likely to opt for Path B too. “Seeing these avalanches in the data was unexpected and exciting to say the least,” adds Wang.

Besides informing people about how they should interact with strangers in crowds, this work could be used in the future to assist with the design of safer transit infrastructure, as well as real-time crowd monitoring and management.

The content is provided for information purposes only.