Follow the crowd

In a virtual world people Also behave like herd animals. (Photo: Thinkstock)

In a virtual world people Also behave like herd animals. (Photo: Thinkstock)

Earthquakes, tsunamis, rushing mobs – in the Transport and Planning department, test subjects are being immersed in a ‘disaster’, with the ultimate goal of improving the usability of traffic models for evacuation.

The day started out fine in the virtual world that the scientists of the Transport and Planning department (CEG) use to chart human behaviour in order to improve their traffic models.

‘You will be attending a concert on an island’, PhD student ir. Mignon van den Berg informs her dozens of ‘guinea pigs’, many of whom have been recruited through newspaper advertisements. We are all seated at computers. ‘You will go there by helicopter. Once you are there, each of you will take a car in order to drive to the concert. You can use these arrows to move forward, backward, left and right’, she explains, pointing to the keyboard.

The group that has gathered is quite diverse, including young as well as old people, from all ranks of society. A teenager wearing a baseball cap and holding a can of soda is well aware of how to navigate in a gaming environment. An older woman is reading the instructions carefully.

Although the situation in which the participants find themselves is fictitious, the sounds, the rushing avatars and the clock that continues to tick makes people nervous. ‘This is reflected in the responses to the questionnaire that we have the participants complete at the end of the session’, notes van den Berg. ‘After the session, people report that they have experienced an increased level of stress.’

The PhD student is part of the research team of Prof. Serge Hoogendoorn. In 2009, the NWO awarded a Vici grant to this professor of transport to conduct research into questions including how traffic models could be made more useful for evacuations.

Models assume that travellers are largely aware of what they can expect on the road, that they have clear destinations and that they make well-considered choices regarding their route and time of departure. But this would obviously not be the case if a dyke should break or if a tsunami should occur.

‘In such situations, people behave like herd animals’, asserts Hoogendoorn. ‘They will do what they are used to doing. This mental state is known as “bounded rationality”. People will not adjust their behaviour unless they become truly aware that their actions are not wise.’

Can virtual worlds be used to study this type of traveller behaviour during crisis situations? According to the TU Delft scientists, not much is known in this regard. Van den Berg: ‘I was therefore interested to see whether we could use this technology to develop a quantitative overview of this type of following behaviour during evacuations. To date, this has been impossible.’

Test subjects
In recent months, van den Berg has conducted ten experiments, each with about 30 subjects. She is now charting the behaviours of these test subjects. Who went in which direction, and when? How did people influence each other?
That last question is particularly interesting. ‘It seems that people also behave like herd animals in this virtual world. This provides an initial indication that the programme could be useful for improving evacuation models.’

Hoogendoorn has great expectations. ‘Our current models – like the one for predicting how quickly an area can be evacuated in case of a flood – have been developed with the understanding that we do not know much about how people react. This results in margins of uncertainty. We hope that this type of research, using avatars, will help to reduce these margins.’

The researchers are also able to manipulate the situation. Unbeknownst to the other participants, several of van den Berg’s colleagues also play along as avatars. ‘These “moles” have been instructed to run away from the concert at particular times’, the PhD student explains. If the moles run away as soon as they feel the earthquake, without waiting for the news bulletin, many of the participants will survive the disaster, provided that the moles run in the right direction.

‘This program is quite advanced’, observes Hoogendoorn. ‘We can analyse the avatars to determine exactly how they reacted after receiving particular information. Did they start to flee when they felt the earth shaking? Did they leave only after they heard the news bulletin? Or did they follow the rest of the group? This type of analysis is usually dependent upon surveys after the fact.’

The design of the experiment is largely van den Berg’s brainchild. The underlying programming is from the National Institute of Informatics in Tokyo. Hoogendoorn and his colleague, Prof. Hans van Lint, who was recently appointed as an Antoni van Leeuwenhoek professor, had been invited to deliver lectures there several years ago. The plan to use the computer program in Delft was concocted later in a Belgian bar in Japan’s capital city.

In the coming years, Hoogendoorn and van Lint would like to conduct more experiments with virtual worlds. ‘We are thinking about designing a football stadium from which all of the visitors must evacuate en masse’, relates Hoogendoorn. Such research will take place in a new laboratory that van Lint will be starting: the Delft Integral Traffic & Transport Laboratory, or DITT Lab.

It remains a remarkable fact that people can experience stress from a simulated earthquake and tsunami, even when safely seated at a PC. To van den Berg’s colleague, psychologist Erica Kinkel (Transport and Planning), this does not come as much of a surprise. ‘The situation was deliberately created to immerse people as deeply as possible in the virtual world. The sounds that the participants hear through their headphones help in this regard. The individual participants must do the best they can to make it to the helicopter by a certain time in order to evacuate. Even if the situation is not life-threatening, people experience stress due to time pressure.’ The psychologist admits that the comparability of created crisis situations with actual disasters remains a complicated issue. ‘It’s the same in real life, however, given that no two earthquakes or tsunamis are the same.’


Stay informed about the research

Receive the Delft Outlook newsletter 4 times a year