Testing a dynamic model of performance in aircraft conflict detection

The analysis of human decision making performance is vital for the development of effective training programs for operators. Across the human factors literature, Signal Detection Theory has been used to do this. Despite its broad utility, a limitation exists in that Signal Detection Theory does not account for operator response times and speedaccuracy trade-offs. Dynamic Signal Detection Theory has therefore been proposed, as it takes response times into account when analysing decision making performance. However, Dynamic Signal Detection Theory has not yet been validated in dynamic, complex environments. As such, this project aimed to test the validity of a dynamic model of signal detection in an industrial task. Twenty-one University of Queensland students completed a simulated air traffic control conflict detection task in which they were required to detect impending conflicts between aircrafts and intervene where necessary to prevent these conflicts from occurring. A one-way repeated measures design was used, with the amount of time available to detect conflicts and intervene varying across four experimental conditions. Results indicated that the time pressure manipulation did not yield speed-accuracy trade-offs, and the pattern regarding the relative speeds of correct versus error responses was not as predicted by the diffusion framework. The Dynamic Signal Detection Theory parameter of threshold was found to vary across the time pressure conditions in the expected fashion; however the discriminability and bias parameters did not. Explanations as to why inconsistent results were found are discussed along with future directions for research