Predictive engineering models based on the EPIC architecture for a multimodal high-performance human-computer interaction task

Abstract

Engineering models of human performance permit some aspects of usability of interface designs to be predicted from an analysis of the task, and thus they can replace to some extent expensive user-testing data. We successfully predicted human performance in telephone operator tasks with engineering models constructed in the EPIC ( E xecutive P rocess- I nteractive C ontrol) architecture for human information processing, which is especially suited for modeling multimodal, complex tasks, and has demonstrated success in other task domains. Several models were constructed on an a priori basis to represent different hypotheses about how operators coordinate their activities to produce rapid task performance. The models predicted the total time with useful accuracy and clarified some important properties of the task. The best model was based directly on the GOMS analysis of the task and made simple assumptions about the operator's task strategy, suggesting that EPIC models are a feasible approach to predicting performance in multimodal high-performance tasks.