Attention Deficit/Hyperactivity Disorder (ADHD) is a common neurodevelopmental disorder that affects many individual’s ability to maintain and shift attention. Little is known about the connection between executive function and eye movement in individuals diagnosed with ADHD during multitasking. The objective of this research is to examine the relationship between patterns of eye movement and multitasking performance between students with and without ADHD. During the experiment, students with and without ADHD completed NASA’s Multi-Attribute Task Battery at both low and high task complexity. We found that students without ADHD showed higher error counts, higher workload ratings, and greater total number of fixations in the high task complexity than in the low task complexity. By understanding the relationship among eye movement, executive functioning, and mental workload, it is possible to gain insight in attention allocation in online learning environments and apply that knowledge to create tools for equal access to learning.

Although multitasking has been studied in the past few decades, there has been a lack of investigation on how individuals’ multitasking performance can be predicted using eye movement data. To address this gap, this study proposed an exploratory approach to understand the manifestation of eye movement patterns that could provide diagnostic and predictive information of multitasking performance. Nineteen participants completed Multi-Attribute Task Battery (MATB-II) experiments under both low and high workloads and their eye movement and MATB-II task performance were collected. We applied a hierarchical clustering method that classified the participants into three clusters – clusters with small, medium, and large number of fixations. Then, we compared the MATB-II performance of the three clusters. The results s howed significant differences in average reaction time to stimuli and average error count among the three clusters. Our study showed that hierarchical clustering of eye fixations can effectively predict multitasking performance.

This study examines the utilities of a dynamic Bayesian network (DBN) to predict multitasking performance. Multitasking is the practice of conducting more than one task simultaneously. Compared with BN (Bayesian network), the DBN has the advantage of encoding both spatial and temporal relationships of the multiple variables under uncertain information. We established the DBN model based on contextual and observable variables from 19 participants to predict multitasking performance over time. The proposed DBN outperformed the BN model with smaller prediction errors and showed great potential to be used as a tool in error management and operator screening in diverse work environments.