Modeling human perceptual decision-making using an artificial neural network

Computer-based graphical displays are increasingly used as aids to human decision-makers. However, effective use of this new technology requires an understanding of the perceptual components of decision skill. As a step toward meeting this need, a model of skilled human decision-making using dynamic, graphically displayed information is presented. The model uses the backpropagation architecture to describe how skilled decision-makers might become increasingly attuned to highly diagnostic features of a graphically displayed decision task. The proposed model is consistent with recent psychological research suggesting that decision-makers develop a `trained eye´ through experience. Results comparing the performance of the model with the performance of subjects in a laboratory experiment suggest that the model provides a promising approach for the study of decision skill