Carl Westin

2-2 Resolving automation acceptance issues 25 2-2-3 Individual-sensitive automation A logical next step, further reducing the compatibility gap between human and au- tomation problem-solving styles, is to consider individual-sensitive automation. de- cision aiding automation is typically designed to fit with a group rather than the individual. A limitation of the above heuristic forms of automation is that, although individual differences were acknowledged, the resulting decision aid treated opera- tors as a homogeneous group. As such, they have not been able to ensure complete harmony between the individual human and automation. On the basis of individual differences in personalities 36 and cognitive styles 37 that influence how problems are approached and solved, it can be expected that automation sensitive to individual preferences would be beneficial to automation acceptance. This notion is supported by several technology acceptance theories (see next section) that consider cognitive style to be an important factor influencing acceptance. 92–95 One example of a cognitive style dimension is that of impulsivity- reflectivity. On a continuum measuring response time in problem-solving, individ- uals are either fast (impulsive) or slow (reflective) in making decisions. 37 Recently, Liu et al. 96 argued that decision aiding automation that embraces in- dividual differences will become increasingly important for user attitudes and per- formance in situations with vaguely defined tasks and problem-solving processes, of which ATC is a prime example. While overlooking individual differences has been sufficient for today’s system, several researchers have argued for more indi- vidualized automation to foster successful human-automation teamwork in future ATC. 97–99 In theory, if controllers are to remain in the loop while working with de- cision aiding automation, they should evaluate the output of automation against an on-going appraisal. 100 Given what is known about naturalistic decision-making and biases in human decision-making, however, Kirwan and Flynn 29 were probably cor- rect in their observation that “if the resolution advisory is similar to that which the controller might have thought of, and if the controller can quickly infer why such a resolution was made, then such a rapid evaluation and decision will be possible...” (p. 4). Following this line of thought, we have investigated the potential benefits of in- dividual sensitive, strategic conformal decision aiding automation on acceptance in previous work. 101 Although it was technically not feasible to create such advanced automation, we simulated it by replaying controllers’ own solutions to pending sep- aration conflicts and disguised it as “automated advice.” In a first set of simulation trials, 16 controllers used a constraint-based interface (the Solution Space Diagram, SSD 102 ) to formulate and implement heading and/or speed solutions to traffic conflicts in the horizontal plane. The interface was novel in the sense that it revealed all possible heading and/or speed combinations that would