A Computational Model of Attention/situation Awareness

A computational model of attention and situation awareness (SA) was developed and used to predict pilot errors in the task of taxiing from runway to terminal. The model incorporates a lowlevel perception/attention module and a higher-level belief-updating module. Attentional scanning is controlled by bottom-up and top-down processes, with the effectiveness of top-down guidance varying as a function of SA. Information sampled by the low-level module is fed forward to the higher-level module for consolidation within a working memory representation of the pilot’s situation, with the quality of this representation reflecting the pilot’s level of SA. The model was validated by comparing its predictions to the behavior of pilots performing a taxiway simulation. Results indicate that the model successfully predicts the improved performance associated with display augmentations, and provides construct validity regarding the effects of visibility, distraction, and degraded information quality.