Human error models as predictors of accident scenarios for designers in road transport systems

Models of human error in technological systems have so far been used mainly retrospectively to explain accidents which have happened. Unless they can be used by designers of transport systems as prospective predictors of potential accident scenarios, their value as tools for accident prevention is limited. This paper reviews the theoretical and practical difficulties involved in closing the gap between human error models and the needs of transport system designers. The main conclusions are that cognitive psychologists should: (1) produce a pool of production rules which give an adequate description of driver behaviour in practice, plus a confusability index for those rules; and (2) document cases of erosion of rules by road users and develop a diagnostic tool for ‘susceptible’ rules.