Cavallo, V., INRETS and Dumont, E., LCPC, France
The hypothesis that speed is underestimated in foggy driving conditions (Brown, 1970) has recently given rise to a series of empirical studies using a simulator. Snowden et al. (1998) showed that participants instructed to adopt target speeds drove faster as the visual scene became foggier. Cavallo and Pinto (2001) were not able to confirm these results, despite the wide range of experimental methods used. In the present experiment, we attempt to determine whether these discrepant results can be explained by the way fog is simulated and/or by how the visual scenes are presented. A speed production task was used (target speeds of 50, 80, and 110 km/h) on a textured, tree-lined road with pavement markings. Although no effect of the visual presentation mode was found (monitor vs. large screen), the results pointed out the crucial role of the characteristics of the simulated fog. A speed increase was noted only when there was a uniform attenuation in contrast (Snowden et al., 1998). This simulation mode produces a veil-like effect that deteriorates the discriminability of details but barely reduces the preview distance. On the other hand, an exponential attenuation in contrast as a function of distance, which corresponds to the physical phenomenon, limits visibility beyond a certain distance (the "visibility distance"), but does not lead to a rise in speed (Cavallo & Pinto, 2001). It is concluded that speed perception is not altered in conditions where fog is simulated in a realistic way.