Amy Sprague
May 13, 2025
By tracking eye movements and neural activity, Jiaxin Li, a Ph.D. student in the Department of Industrial and Systems Engineering (ISE), is developing tools to anticipate when we're about to make critical errors while juggling multiple tasks—potentially transforming high-stakes professions from air traffic control to emergency response.
Jiaxin Li, ISE Ph.D. student
In today's complex work environments, professionals often manage multiple demanding tasks simultaneously. Whether it's air traffic controllers monitoring dozens of aircraft, emergency responders coordinating rescue efforts, or industrial operators overseeing critical systems, the ability to multitask effectively is crucial—and the consequences of errors can be severe.
This challenge drives the research of ISE's Human and Systems (HAS) Lab doctoral researcher Jiaxin Li, whose innovative work is transforming our understanding of human performance during multitasking.
A OpenMATB visualization of eye-tracking on a task-focused screen which Ph.D. student Jiaxin Li can analyze to predict multitasking performance. Reference: Cegarra, J., Valéry, B., Avril, E., Calmettes, C., & Navarro, J. (2020). OpenMATB: A Multi-Attribute Task Battery promoting task customization, software extensibility and experiment replicability. Behavior research methods, 52, 1980-1990.
The science of predicting multitasking performance
Li's research tackles a fundamental challenge in safety-critical professions: how to predict and maintain peak performance when juggling multiple tasks where mistakes can have serious consequences.
"What makes Jiaxin's work so innovative is her integration of physiological sensing into predictive models," explains Associate Professor Ji-Eun Kim, Li's faculty adviser and director of the HAS Lab. "Instead of simply measuring errors after they happen, her approach allows us to anticipate performance decrements in real-time, creating opportunities for intervention before critical mistakes occur."
The technical backbone of Li's research involves dynamic Bayesian networks (DBNs)—sophisticated models that combine real-time data from eye movements and brain activity (EEG) to continuously predict multitasking performance. Her findings demonstrate that these integrated models significantly outperform those using single indicators, and even surpass other advanced machine learning techniques in accuracy.
"Traditional methods of measuring multitasking ability often look backward at error rates, which doesn't capture the moment-to-moment changes in our capacity to handle multiple demands," says Professor Kim. "Jiaxin's models offer a revolutionary approach—they can actually anticipate when someone is likely to make a mistake, potentially allowing for intervention before an error occurs."
Rethinking automation in complex systems
Another significant dimension of Li's research focuses on how automation can best support human operators. Rather than pursuing full automation, Li investigates which specific tasks should be automated for optimal human-system performance.
"The question isn't just whether to automate, but what to automate," notes Li. "The research shows that the benefits of partial automation depend significantly on factors like sensory modality. This nuanced understanding is critical for designing systems that genuinely enhance human capability rather than creating new vulnerabilities."
Using flight simulator software known as the Multi-Attribute Task Battery II (MATB-II), Li has demonstrated that partial automation provides greater benefits when the tasks involve different senses—a finding with important implications for cockpit design, control rooms, and industrial operations.
Li's work explores how factors like the sensory modalities of tasks (visual vs. auditory) and task priorities influence the effectiveness of partial automation. Her research suggests that automation leads to greater improvements in multitasking performance when the tasks involve different senses (cross-modality). These insights provide valuable guidance for designing automation in complex systems to better support human operators.
Transforming how we design critical systems
As workplace demands grow increasingly complex across many fields, Li's research at the HAS Lab represents a shift toward human-centered system design—creating environments that acknowledge and accommodate human cognitive limitations rather than treating them as flaws to overcome.
"What makes this research so exciting is its potential to transform how we approach high-stakes work environments," says Professor Kim. "Instead of designing systems and expecting humans to adapt, Jiaxin's work points toward a future where we design with human capacity and wellbeing as the foundation."
The implications of Li's research extend beyond theoretical understanding—they offer practical approaches to enhancing safety and performance in critical environments. Her predictive models could potentially be integrated into future systems that monitor operator state and provide assistance precisely when needed, before errors occur.
By advancing our ability to predict multitasking performance and optimize automation strategies, Li's research promises to enhance safety and efficiency across numerous high-stakes professions. From aviation to industrial control rooms, these insights could revolutionize how we design systems where multitasking is unavoidable and errors can have serious consequences.
"Ultimately, this research is about working with human nature rather than against it," Li explains. "By understanding the intricate relationship between physiological signals and multitasking performance, we can create systems that complement human abilities rather than demanding superhuman performance."
For a field historically focused on training humans to adapt to technological systems, Li's research represents a profound shift—one that places human capabilities at the center of system design, potentially transforming how we approach some of society's most critical and demanding jobs.