ABSTRACT
In naturalistic situations and, more precisely, in dynamic risk environments, one of the major competences expected from operators is situativeness: "the tendency to take the particularity of the situation into account" (Hukki & Norros, 1998, p. 317; Hutchins, 1995; Suchman, 1987). However, training people to acquire such competence in their natural environment is difficult for obvious risk and economic reasons. The use of simulators is becoming more and more widespread. Up to now, the emphasis has mainly been put on (a) acquisition by the trainees of procedural knowledge and problem solving skills to control rare and potentially serious incidents, (b) acquisition of crisis management skills (Gabba, Fish, & Horward, 1994), and (c) methodological support for instructors (Hukki & Norros, 1998). Less research has been done in the United States on the temporal features of naturalistic situations that should be preserved in simulators, for they are part of the situativeness difficulty. In Europe, a strong trend of field and experimental studies, mainly in process control, has pinpointed this problem and stressed the "temporal errors" linked to these situational characteristics (De Keyser, 1995). Regarding time and situation, we have to fight two common views. The first restricts the problem to temporal pressure and dynamic decision making under stress. Based on field studies and microworld experiments in dynamic environments, Hoc (1989), Brehmer (1990a, 1990b), and Van Daele (1993) clearly demonstrated that many temporal characteristics other than temporal pressure shape the operator's behavior: time scales of actions, delay of feedback, slow/rapid tempo of process responses, and so forth. In fact, these characteristics determine, among other factors, the mode of control of the situation: from an anticipatory mode, where people plan their actions in advance and can predict the evolution of the process to a reactive one, where people react immediately to any change in the situation. Sometimes "scrambled" modes of control (Hollnagel, 1998) or cognitive breakdowns (Amalberti, 1996) emerge when people do not succeed in managing the situation. The other reductionist view is to think that the problem will be solved by increasing technical and functional validity of simulators. To a certain extent this is true. If we compare full-scale and screen-based simulators, the former preserves the temporal characteristics of the situation better than the latter, but it does not preserve all of them. Screen-based simulators could be useful training for some aspects of situativeness. In fact, what we need to assess is their psychological validity. This is the extent to which simulated situations generate similar psychological conditions of action to naturalistic situations (Baker & Marshall, 1989; Grau, Doireau, & Poisson, 1998; Leplat, 1989). Psychological validity stems from similarity between the conditions of action rather than similarity between material or technical characteristics per se. But conditions of action cannot be exclusively derived from formal task constraints; they depend strongly on the context in which the activity takes place. What has to be kept in simulators directly depends on the analysis of the activity in naturalistic situations. This activity analysis aims to identify the relationships between the contextual task constraints and the performance demands.
