Comparative analysis of cognitive load across VR locomotion tasks using EEG and NASA-TLX

Virtual reality (VR) is increasingly recognized as a transformative medium across various domains, including education, healthcare, and rehabilitation. One of the primary challenges in immersive environments is understanding how different locomotion techniques influence users’ cognitive load. This study addresses this gap by systematically evaluating cognitive load across ten distinct VR locomotion tasks using both objective (electroencephalography, EEG) and subjective (NASA-TLX) measures.The analysis revealed substantial task-dependent variations in cognitive load, particularly within the Physical Demand, Effort, and Frustration dimensions. Notably, tasks such as Climb and Climb Hanging elicited the highest cognitive load levels, while less demanding tasks like Blink and Shift showed significantly lower scores. For instance, Climb and Climb Hanging yielded the highest weighted composite NASA-TLX scores (999.06 and 965.00, respectively), in stark contrast to Blink (682.19) and Shift (679.69), suggesting that vertical locomotion imposes considerably greater cognitive strain.In addition to behavioral measures, machine learning classifiers were employed to evaluate the discriminative potential of EEG data in identifying cognitive states induced by locomotion. Among the models tested – RF, SVM, and kNN – the RF classifier achieved the strongest performance (weighted F1-score = 0.893), demonstrating strong potential for real-time cognitive load monitoring in VR.Overall, these findings underscore the importance of task-specific locomotion design and highlight the value of combining physiological and subjective assessments to comprehensively evaluate cognitive load in immersive virtual environments.