Aviation Human Factors Division

We describe three applications of an attention-situation awareness model to predicting pilot performance in high fidelity flight simulations. The model incorporates the SEEV model of selective attention, in which eye movements are driven by salience, expectancy and value, and are inhibited by the effort required to move the eyes over longer distance and of concurrent tasks. In application 1, we predict the taxi-way errors made in a simulation in which commercial airline pilots navigate the airport surface at Chicago O'Hare airport. In application 2, we predict the scanning behavior of three commercial pilots flying a simulated approach to Santa Barbara Airport, using a synthetic vision system display, and responding to occasional unexpected events. In application 3, we predict scanning behavior and performance of 8 general aviation pilots flying a simulated landing approach to a terrain challenged airport. Various forms of the computational model are evaluated to predict scanning performance, off-normal event detection and, in application 3, multi-task performance of flight path control and traffic detection. The model is shown to be quite valid, accounting for around 85% of the variance in scanning across different areas of interest in the pilots' visual field. The model fit is not improved by including the effort-inhibition parameter. The degree to which pilots conform to the expected value (optimal) version of the model is shown to positively predict the indices of multi-task performance. The role of redundancy is also addressed.