A virtual reality study of behavioral sequences in residential fires

Abstract

Decision-making in residential fires can be difficult to study in an experimental setup. Virtual reality could be a way to expose participants to a residential fire scenario and collect data on their behavior. However, the validity of the data obtained needs to be assessed. To explore that, the data collected in two virtual reality residential fire scenarios in form of sequences of behavior was compared to the general model developed by Canter, Breaux and Sime. Two scenarios were developed, one with and one without a smoke alarm. Two samples of 20 participants each were exposed to a fire in a virtual house. The participants were residents of houses with the same layout as the virtual one, making them familiar with the building. The sequence of actions they performed were recorded and decomposition diagrams were drafted based on them, to then be compared to the general model. The results show that the participants’ sequences of behavior did not only fit those predicted by the general model, but also that the participants followed many different sequences, covering most of the possible patterns indicated in the general model, as it is expected in a real world fire.