Files
Abstract
Monitoring cognitive effort is challenging due to complex neural networks and variability in task engagement. This study presents an electroencephalogram (EEG) and machine learning-based approach to estimate cognitive effort during listening tasks. We developed an EEG processing pipeline incorporating a novel method for separating ocular artifacts from cortical signals and evaluated multiple classification strategies. While alternative classifiers performed similarly to prior research, they highlighted limitations of current approaches. Extending EEG-based cognitive load estimation to a new population, we applied the system to individuals with aphasia and controls during naturalistic listening tasks of varying complexity. The system distinguished EEG data from difficult versus easy passages in both groups, though classification accuracy was lower than in previous studies, likely due to subtle task manipulations and inconsistent effort levels. These findings inform future refinements for cognitive monitoring in clinical populations.