Date of Award
Campus Access Dissertation
Doctor of Philosophy (PhD)
Dan A. Simovici
Enhancing user performance is a constant goal for human computer interaction (HCI) researchers. To achieve this, we need to have an accurate measurement of a user's performance that reflects moment-to-moment of the user's cognitive state. By measuring users' cognitive state during a specific task, such as visual search, reading, or counting, we can improve a given interface based on this implicit data stream. The measurement of cognitive load per se can be improved using psychophysical and machine learning methods. In this dissertation, I study how both this measurement and the users' task performance can be enhanced using various methods, with a focus on eye movement data as an indicator of cognitive state. I used a safe, non-invasive eye tracking technology that can collect pupil dilation and eye fixation as indicators of cognitive state during certain searching and reading tasks. To enhance users' performance, I discuss the applicability of using two main interaction areas in HCI: interaction techniques using auditory feedback during search tasks, and the contextual design of reading interfaces. I examined the users' performance using cognitive load measurement and recommended the appropriate interface that can optimize performance. To obtain better measurement of cognitive load as an indicator of a user's task performance, I ran a series of controlled experiments. I studied the automatic recognition of the level of cognitive load in different reading and visual search tasks. I improved the measurement of cognitive load during a search task, focusing on filtering the eye data either by using machine learning classification methods or statistical analysis. Also, I was able to predict the user's efficiency by measuring eye movement data at the early stage of the task using a novel interface during two different search tasks. For the reading tasks, I presented an interface that improved speed reading by guiding gaze fixations toward the middle of a word and measured the correlation between pupil size and comprehension. The results were significant in measuring cognitive processing and can be used to determine the users' comprehension level.
Attar, Nada, "Enhancing Cognitive Load Measurement and User Performance in Human-Computer Interaction" (2016). Graduate Doctoral Dissertations. 292.