This paper presents an exploratory study of mechanisms by which exposure influences accident occurrence. The paper is based on a previous paper, inwhich exposure was defined as the occurrence of specific events that represent an opportunity for an accident to happen. Four such events were considered: encounters, simultaneous arrivals in intersections or at pedestrian crossing facilities, lane changes and braking. This paper reviews some studies that permit the shape of the relationship between the number of events and the probability of accident occurrence to be determined. It is found that, in general, the rate of accidents per event falls sharply as the number of events increases. This can be interpreted as a learning curve: the more often road users encounter a specific event, the more skillful they become in recognizing and controlling the risks involved in that event. A measure of learning efficiency is proposed and it is shown that learningefficiency varies between different types of events. It is suggested thatthe complexity, controllability and predictability of an event influenceslearning efficiency. More complex, less controllable and less predictableevents reduce learning efficiency, because these events are associated with cognitive overload and violation of road user expectations. Examples are given of events of varying degrees of complexity, controllability and predictability. Another mechanism by which events involving conflict or physical contact between road users become recorded as accidents is incompatibility, which refers to differences in the kinetic energy produced by different types of vehicles and the degree to which they protect road users from impact.
Abstract