Yili Liu, University of Michigan

Yili Liu, University of Michigan



In this talk I will describe our five (of the 7+/-2 planned) past and current steps toward developing a queueing network (QN) theory of mental architecture. First, a QN theory of reaction time (RT) was developed that integrates the influential architectural RT models as special cases, including the serial discrete-stages, the serial continuous-flow, and the discrete network models (such as the critical path network model). Further, the QN models cover a broader range of mental architectures and can be subjected to well-defined empirical tests. Second, the architectural RT models and the sequential sampling RT/accuracy models are unified through QN-RMD (Reflected Multidimensional Diffusions). Third, QN-MHP (Model-Human-Processor) was developed to bridge the mathematical and the symbolic models of mental architecture and to support mathematical modeling and real-time generation of task performance and mental workload. Fourth, QN-ACES is being developed to integrate QN and ACT-R, CAPS, EPIC, and Soar architectures for further theoretical unification and engineering application. Fifth, QN mental architecture is being extended to the neural and brain physiological domain. Our QN modeling work has been published in a dozen journal articles including 2 in Psychological Review, 2 in ACM Transactions on Computer Human Interaction, and several in IEEE Transactions.


Link to two papers:


Queueing Network Modeling of Elementary Mental Processes

AN-ACES:  Integrating Queueing Network and ACT-R.....