Carl Westin

6-1 Introduction 115 A BSTRACT Although people are recognized to vary in their decision-making and problem-solving pref- erences, automated decision aids have generally not been designed to consider or accommo- date individual cognitive differences. However, researchers have argued that personalized automation will become increasingly important for facilitating safe and efficient working relationships between humans and automated agents, particularly in complex, dynamic, and time-critical automation-dependent environments such as air traffic control. An issue in the development of such individual-sensitive automation is that people may be inconsistent in their decisions and actions. Previous research provides an inconclusive picture in regards to consistency and agreement in conflict resolution. This paper presents two studies investigat- ing decision-making consistency (intra-rater reliability) and agreement (inter-rater reliabil- ity) in conflict resolution among air traffic controllers. Conflict solutions were collected in real-time simulations, in which participants unknowingly encountered the same conflict four times. Consistency and agreement in conflict solutions were determined against a solution framework building on and extending conflict resolution strategies identified in previous research: the solution parameters hierarchy classification, the control problem classifica- tion, and solution geometry classification. Controllers were consistent but did not agree. As such, controllers cannot be considered homogeneous and their individual differences in conflict resolution decision-making need to be acknowledged in human factors models and automation design. 6-1 Introduction Similar to other control-room environments, ATC automation design has generally followed a “one-size-fits-all” approach suitable to an average, generic, stereotype controller. Although controllers currently can customize basic system properties, such as interface appearance (e.g., zoom, brightness, information layers), there are no current applications sensitive and adaptive to the controller’s cognition. The generalization is further exacerbated by a widespread assumption that controllers are homogeneous which can partly be attributed to rigorous screening and selection procedures and training that converge towards creating a uniform work force. 235, 236 Human factors researchers have argued for more individual-sensitive, idio- graphic , 237 automation design approaches, as opposed to current generic ones, suited to the individual’s abilities, needs, and preferences. 126, 154, 155, 197, 198 Simi- larly, individual differences has been outlined as pivotal in future ATC automation design. 97–99 In support of this, extensive research on cognitive styles and personal- ities has demonstrated that people search and process information, make decisions, and solve problems differently. 37, 238–240 Despite this knowledge, however, individ- ual differences have traditionally not been accommodated for in automation design, although exceptions can be found in research on automated cars (e.g., personalized