The aim of this experiment was to determine the kind of visual information used in anticipating rear-end collisions according to the visual information available in the road environment and the perceptual style of the driver. Nine field independent and seven field dependent subjects had to judge time-to-collision under various visual and spatio-temporal conditions. Results showed that only under poor visual conditions (presence of the obstacle alone) were temporal estimates affected by approach speed, obstacle distance and actual time-to-collision. Under these conditions the estimates appeared to be mainly based on static depth cues (angular size). The enrichment of the visual scene (texture) allowed the processing of motion-based information (optical flow) and made the estimates independent of spatio-temporal conditions; variability, overshoots and subjective difficulty were reduced. Whereas both field-dependent and independent observers took global visual changes into account, only field-independent subjects were able to use local motion as well. The findings suggest that multiple sources of information play a part in time-to-collision judgements.
Abstract