Abstract
Visual indicators can be used by occupants to determine if there is a fire emergency in the building. Past research indicates that, when encountered, the visual characteristics of flames are used by occupants to determine posed risk. However, judgments made by observers contain biases such as overestimating how long it takes for fires to reach a specific size displayed via images and, in turn, how much time is available for someone to take protective actions. This raises questions about how well individuals can detect the growth of developing fires. The present study investigated both the sensitivity of human observers to fire growth and whether rendering fires as three- versus two-dimensional objects affected sensitivity. Participants (N=40) judged if a video (8s duration) contained a fire that was or was not growing in intensity. Videos were generated from simulations via software used in fire safety research and design (Fire Dynamics Simulator). The growth rates for fires were based on the prescribed values in fire safety handbooks that estimated, using non-linear curves, slow to ultra-fast developing fires. Two types of virtual renderings of fires were created (2D and 3D) from the same viewpoint to display the fire width and height, with 3D fires also including depth. Using d-prime as an index of sensitivity, participants were significantly above chance at detecting growth (Cohen’s ds > .5) with greater sensitivity the larger the growth rate (Cohen’s f=.69) and no significant impact of rendering type. This provides initial evidence that visual indicators of fires can be used to judge growth even at the slowest growth rates. Furthermore, for the type of fires simulated, the use of depth of the fire did not affect performance. Subsequent research can examine whether occupants under duress display similar performance when determining whether a fire is increasing in intensity.