Beyond Cognitive Load: AI-Based Estimation of Cognitive Effort Using Brain Signals During Digital Tasks

Area of Science:

• Neuroscience
• Cognitive Science
• Artificial Intelligence

Background:

• Cognitive effort, the balance between cognitive load and task performance, is vital for understanding mental resource allocation.
• Unmanaged cognitive overload in public health and clinical training can lead to medical errors and burnout.

Purpose of the Study:

• To investigate if cognitive effort varies across different task segments.
• To determine if cognitive effort can be estimated at an individual level using brain signal data and machine learning.

Main Methods:

• Functional near-infrared spectroscopy (fNIRS) data were collected from 16 participants during a digital cognitive task.
• Cognitive effort was defined using relative neural efficiency and involvement, combining prefrontal hemodynamic activity with task performance.
• A two-stage analysis involved group-level assessment of cognitive effort across segments and participant-independent machine learning to predict performance.

Main Results:

• Significant differences in cognitive effort were observed across the four task segments, indicating task structure impacts cognitive efficiency.
• Machine learning models using fNIRS data accurately predicted individual task performance.
• Cognitive effort estimated from predicted performance scores closely matched that from actual scores, validating the brain signal basis of the metric.

Conclusions:

• The study demonstrates the feasibility of estimating cognitive effort from brain signals using artificial intelligence at both group and individual levels.
• This approach offers a scalable method for monitoring cognitive workload and mental fatigue in technology-mediated settings.