Automatic classification of eye activity for cognitive load measurement with emotion interference

Area of Science:

• Human-Computer Interaction
• Cognitive Science
• Biomedical Engineering

Background:

• Accurate cognitive load measurement (CLM) is crucial for patient treatment, reducing medical errors, and evaluating health information systems.
• Existing CLM methods may be susceptible to confounding factors like emotional interference, necessitating robust approaches.
• Eye activity presents a promising, non-invasive modality for real-time CLM.

Purpose of the Study:

• To propose and validate an automatic cognitive load measurement (CLM) system using eye activity.
• To investigate the influence of emotional arousal on eye-based CLM and develop strategies to mitigate it.
• To assess the feasibility of using eye features for near-real-time CLM.

Main Methods:

• Investigated pupillary response, blink patterns, and eye movements (fixation, saccade) as indicators of cognitive load.
• Conducted experiments combining arithmetic tasks with affective image stimuli to assess emotion interference.
• Examined feature selection strategies to minimize arousal effects and proposed a feature set for classifying cognitive load levels.

Main Results:

• Eye activity features demonstrated susceptibility to emotional interference, but cognitive load effects dominated during task execution.
• Specific segment selection for eye-based features effectively minimized arousal interference.
• The proposed feature set achieved cognitive load level prediction performance comparable to reaction time measures.

Conclusions:

• Eye activity features are feasible for near-real-time cognitive load measurement (CLM).
• The developed system shows potential for improving patient care, reducing clinical errors, and enhancing health system evaluations.
• Further research can refine CLM systems by optimizing feature selection and addressing confounding variables.