This paper discusses several aspects of performance measure selection, which is the most difficult problem of designing studies in drugs and driving. This difficulty arises from the present limited understanding of the behavioural demands of driving, for which there is not yet any classification. The appropriate approach depends on the researcher's goal. One possibility is the empirical classification of a drug as potentially harmful. Another is to view research on the effects of drugs on driving as a scientific enterprise at the intersection of engineering psychology and psychopharmacology. Here, the aim would be to classify active drugs so that their behavioural side effects would be more predictable, or so that drugs that avoid side effects could be designed. A fundamental problem is lack of knowledge of what important elements of the world of driving should be reporduced by a driving simulator. Simulator tests are at present very different from tests of real driving situations, as the author found out in two simulator studies that he conducted in 1971. Even if simulator studies report significant positive findings, they do not make clear what behavioural functions a drug really affects. Although simulators have some advantages, they omit or attenuate many important elements of a typical driving situation.
Abstract