Carl Westin

1-6 Research scope 9 contrast, real-time simulations were used to better capture the reactive elements and time pressures of the real world that influence decision-making. Additionally, this neutralizes the impact of hindsight biases on memory retrieval known to influence subjective methods. 41 Finally, measure of acceptance was based on the degree to which resolution advisories were accepted. This is perhaps the first time that re- search has tried to empirically define and quantify trust and acceptance of decision advisories that perfectly fits a person’s preferred way of solving problems (in fact, because it is a “replay” of their own solution). This thesis combines current and complementary work into both optimized technology-centered and heuristic human-centered approaches to ATC display and automation technology, by systematically evaluating algorithmic and heuristic ap- proaches to CD&R decision support systems. Findings extend current state-of-the- art with respect to automation design principles and personalized decision support applications. 1-6 Research scope This scope of this thesis is narrowed to semi-automated decision support systems providing support to short-term strategic and tactical task in demanding and time- critical complex systems, specifically ATC. This involves more advanced types of information analysis automation and decision aids that provide specific advisories for solving a particular control problem. More strategic decision support systems, working more on a planning-basis, are not considered, although it can be expected that this type of support system will become increasingly important in line with the desire for high performance in complex environments. Fully automated systems have not been addressed, except for learning from relevant robotics research and autonomous agents considered in artificial intelligence. The ambitious aim and novel research approach were not without challenges and risks. An iterative design protocol had to be developed for the creation of conflicts, scenarios, and test simulations. For example, conflict and scenario design had to ensure that task load was neither trivially low nor excessively high, and set to a point at which the decision to use automation is a meaningful one. Furthermore, an experimental protocol for controlling strategic conformance was needed, allowing for a scale up from preliminary simulations with novice university students and small groups of retired controllers, to large simulations with active controllers. It was essential that the same controllers participated in both the prequel and experiment simulations. Experiments hinged on the ability to convince controllers that they were not merely observing replays of their own previous performance (or of their colleagues) since this could influence their solutions. As such, it was