Abstract
A study is described that examines the effect of providing systemic information on the acquisition and transfer of fault-finding skills. Systemic training information covers several structural, functional, physical (e.g., causal links), and supervisory features (e.g., system control and instrumentation). Rather than being constrained to a single fault-diagnostic strategy, the provision of systemic information could enable trainees to exploit a variety of strategies, including structural heuristics, pattern recognition, criticality analysis based on affected functions, and identification of unreliable cues hidden by supervisory features. An experimental study tested the hypothesis that training systemic information would enable trainees to develop skills in constructing their own diagnostic rules and modify or enrich them in the context of new situations. The performance of a group trained in systemic information was compared to the performance of 3 other groups trained in heuristics, fault-finding algorithms, and qualitative plant modeling. Learning curves on the original task showed that all groups achieved high accuracy scores in a thermo-mixer plant. A near-transfer study, in which the same plant was used but in the automatics mode, showed that the group who received systemic information was superior to the other groups trained in the heuristics or the fault-finding algorithm. On another far-transfer study, in which a new plant was used, the systemic group achieved the highest accuracy score both on their first encounter with this task and on subsequent practice sessions. Systemic information appeared to play a beneficial role in the transfer of fault-finding skills, which probably justifies the greater mental demands put on trainees to assimilate and translate instructions into an effective diagnostic strategy.
