Queueing Network Modeling of Human Performance and Mental Workload in Perceptual-Motor Tasks.

Abstract

Integrated with the mathematical modeling approaches, this thesis uses Queuing Network-Model Human Processors (QN-MHP) as a simulation platform to quantify human performance and mental workload in four representative perceptual-motor tasks with both theoretical and practical importance: discrete perceptual-motor tasks (transcription typing and psychological refractory period) and continuous perceptual-motor tasks (visual-manual tracking and vehicle steering with secondary tasks). The properties of queuing networks (queuing/waiting in processing information, serial and parallel information processing capability, overall mathematical structure, and entity-based network arrangement) allow QN-MHP to quantify several important aspects of the perceptual-motor tasks and unify them into one cognitive architecture. In modeling the discrete perceptual-motor task in a single task situation (transcription typing), QN-MHP quantifies and unifies 32 transcription typing phenomena involving many aspects of human performance--interkey time, typing units and spans, typing errors, concurrent task performance, eye movements, and skill effects, providing an alternative way to model this basic and common activities in human-machine interaction. In quantifying the discrete perceptual-motor task in a dual-task situation (psychological refractory period), the queuing network model is able to account for various experimental findings in PRP including all of these major counterexamples of existing models with less or equal number of free parameters and no need to use task-specific lock/unlock assumptions, demonstrating its unique advantages in modeling discrete dual-task performance. In modeling the human performance and mental workload in the continuous perceptual-motor tasks (visual-manual tracking and vehicle steering), QN-MHP is used as a simulation platform and a set of equations is developed to establish the quantitative relationships between queuing networks (e.g., subnetwork s utilization and arrival rate) and P300 amplitude measured by ERP techniques and subjective mental workload measured by NASA-TLX, predicting and visualizing mental workload in real-time. Moreover, this thesis also applies QN-MHP into the design of an adaptive workload management system in vehicles and integrates QN-MHP with scheduling methods to devise multimodal in-vehicle systems. Further development of the cognitive architecture in theory and practice is also discussed.