Towards a Measure of Working Memory Capacity via Eye Gaze Measurements
Description/Abstract/Artist Statement
Working Memory Capacity (WMC) is a psychological construct used to study an individual’s attention level, capacity to multitask, and fatigue. While reliable methods of measuring WMC have been developed and utilized, none of them accommodate measuring a subject’s WMC in real time as they perform a task not specifically designed to measure WMC, such as driving. Such a measurement would be beneficial to understanding how various tasks affect WMC. This work explores the feasibility of the development of this measurement using eye gaze measurements to predict cognitive workload, which is inversely related to WMC. Three publicly available datasets containing eye gaze measurements collected from participants performing tasks designed to induce various levels of cognitive load were used to train and test machine learning models based on logistic regression, linear regression, support vector regression, ridge regression, and lasso regression for each trial in each dataset. The target variables were the NASA-TLX subscale scores, so each trial and machine learning algorithm had six models developed for it, one for each NASA-TLX subscale. Predictive importance of the gaze measurements was calculated for the best performing models revealing the measurements most predictive of cognitive workload scores. Saccade main sequence relationships were also graphed to determine if the relationships hold for individuals experiencing raised levels of cognitive load. Preliminary results suggest main sequence relationships hold for individuals experiencing cognitive load. The results provide insight into the feasibility of developing a WMC measurement utilizing a relationship between eye gaze measurements and WMC through cognitive load.
Faculty Advisor/Mentor
Sampath Jayarathna
Faculty Advisor/Mentor Department
Computer Science
College Affiliation
College of Sciences
Presentation Type
Oral Presentation
Disciplines
Cognitive Science | Data Science
Session Title
College of Sciences 2
Location
Learning Commons @Perry Library, Room 1311
Start Date
3-30-2024 9:30 AM
End Date
3-30-2024 10:30 AM
Towards a Measure of Working Memory Capacity via Eye Gaze Measurements
Learning Commons @Perry Library, Room 1311
Working Memory Capacity (WMC) is a psychological construct used to study an individual’s attention level, capacity to multitask, and fatigue. While reliable methods of measuring WMC have been developed and utilized, none of them accommodate measuring a subject’s WMC in real time as they perform a task not specifically designed to measure WMC, such as driving. Such a measurement would be beneficial to understanding how various tasks affect WMC. This work explores the feasibility of the development of this measurement using eye gaze measurements to predict cognitive workload, which is inversely related to WMC. Three publicly available datasets containing eye gaze measurements collected from participants performing tasks designed to induce various levels of cognitive load were used to train and test machine learning models based on logistic regression, linear regression, support vector regression, ridge regression, and lasso regression for each trial in each dataset. The target variables were the NASA-TLX subscale scores, so each trial and machine learning algorithm had six models developed for it, one for each NASA-TLX subscale. Predictive importance of the gaze measurements was calculated for the best performing models revealing the measurements most predictive of cognitive workload scores. Saccade main sequence relationships were also graphed to determine if the relationships hold for individuals experiencing raised levels of cognitive load. Preliminary results suggest main sequence relationships hold for individuals experiencing cognitive load. The results provide insight into the feasibility of developing a WMC measurement utilizing a relationship between eye gaze measurements and WMC through cognitive load.