Queueing Network Modeling of Human Performance in Complex Cognitive Multi-task Scenarios.

Abstract

As the complexity of human-machine systems grows rapidly, there is an increasing need for human factors theories and computational methods that can quantitatively model and simulate human performance and mental workload in complex multi-task scenarios. In response to this need, I have developed and evaluated an integrated cognitive architecture named QN-ACTR, which integrates two previously isolated but complementary cognitive architectures – Queueing Network (QN) and Adaptive Control of Thought-Rational (ACT-R). Combining their advantages and overcoming the limitations of each method, QN-ACTR possesses the benefits of modeling a wider range of tasks including multi-tasks with complex cognitive activities that existing methods have difficulty to model. These benefits have been evaluated and demonstrated by comparing model results with human results in the simulation of multi-task scenarios including skilled transcription typing and reading comprehension (human-computer interaction), medical decision making with concurrent tasks (healthcare), and driving with a secondary speech comprehension task (transportation), all of which contain important and practical human factors issues. QN-ACTR models produced performance and mental workload results similar to the human results. To support industrial applications of QN-ACTR, I have also developed the usability features of QN-ACTR to facilitate the use of this cognitive engineering tool by industrial and human factors engineers. Future research can apply QN-ACTR – which is a generic computational modeling theory and method – to other domains with important human factors issues.