Eye activity correlates of workload during a visuospatial memory task

Area of Science:

• Human-Computer Interaction
• Cognitive Psychology
• Ophthalmology

Background:

• Assessing operator workload is crucial for system safety and efficiency.
• Eye activity provides objective measures of cognitive and physical load.
• Visuospatial tasks, like target identification, demand significant attentional resources.

Purpose of the Study:

• To investigate how workload impacts various eye activity measures.
• To identify which eye activity metrics are most sensitive to workload changes.
• To develop a model for real-time workload estimation using eye-tracking data.

Main Methods:

• Eleven participants performed a simulated anti-air-warfare task with varying target densities to manipulate workload.
• Six eye activity measures (blink frequency/duration, fixation frequency/dwell time, saccadic extent, pupil diameter) were recorded.
• Participant-specific artificial neural network models were trained and tested to predict target density from eye activity.

Main Results:

• Blink frequency, fixation frequency, and pupil diameter showed systematic changes with increasing target density.
• Nonlinear regression identified these three measures as most predictive of workload.
• Artificial neural network models achieved a mean correlation of 0.66 in predicting target density.

Conclusions:

• Combining multiple eye activity measures with artificial neural networks provides reliable near-real-time workload indicators.
• This approach can be applied to monitor operator visual activity and scanning efficiency.
• Potential applications exist in optimizing human-system performance in demanding operational environments.