Carl Westin

7-1 Retrospective 151 7-1 Retrospective The focus of this thesis has been on the cognitive compatibility mismatch between operators and their automated decision aids. To bridge this compatibility gap, the automation’s problem-solving should match the human’s problem-solving style. It was hypothesized that the acceptance and understanding of a decision aid’s advice would benefit if the aid was perceived to “reason” similar to the human it was inter- acting with. This degree of match between the automation’s solution and apparent underlying operations, with that of the human, was conceptualized as strategic con- formance . As perceived by the human, the automation’s reasoning can vary between being conformal (a match) and being nonconformal (a mismatch). To start with, a broad literature search was conducted across a variety of so- ciotechnical work domains for research on automation acceptance in relation to individual-sensitive systems (Chapter 2). Specific focus was given to ATC related research since this was the target domain of both the MUFASA project and this thesis. The foundation for strategic conformance was found in two complementary domains. The cognitive engineering domain was relevant because of its research on automation use decision in dynamic, high-risk, and time critical contexts. The information systems domain was relevant because of its well-developed and well- accepted acceptance models. Of particular relevance was the compatibility con- struct, considered in both domains, which asserts that human and automation must be compatible in order to facilitate acceptance. A model of human-machine com- patibility was proposed, ranging from a basic level of response compatibility that considers simple handling qualities (i.e., consideration of physical aspects that re- inforce correct use) to the highest level of cognitive compatibility that considers the strategic conformance in problem-solving (i.e., consideration of psychological aspects in decision-making that support the operator’s problem-solving style). 7-1-1 Empirical findings on strategic conformance Strategic conformance was empirically evaluated in three real-time simulations with air traffic controllers. A conformal advisory was always based on the controller’s own implemented solution to the same conflict as recorded in a previous prequel simulation (in which conflict were solved manually with the support of the heading band SSD). A nonconformal advisory was always based on a contrasting solution made by another controller participating in the same real-time simulation. Overall, findings indicate a preference among controllers for conformal advisories. These were accepted more often than nonconformal advisories in two studies (First empir- ical and Automation transparency studies), and equally often in one study (Source bias study). This is noteworthy since no justification, rationale, or additional infor-