Abstract
Purpose The simultaneous performance of tasks relying on the same cognitive processes might lead to gait disturbances. The study aims to explore the distinct involvement of working memory and inhibition during gait within a dual-task using 3D motion analysis and wireless electroencephalography (EEG), addressing limitations of previous research on the removal of gait-related artifacts from EEG signals and the EEG results’ generalizability. Methods 30 healthy participants performed two cognitive tasks designed to engage inhibition (Go-NoGo) and working memory (N-Back). Data acquisition was performed using a 19-channel wireless EEG device and a 3D optoelectronic system. 51 EEG features, seven spatial-temporal and nine kinematic parameters (seven spatiotemporal metrics and Gait Variable Scores for nine lower-limb joint ranges of motion) were analyzed. Riemannian geometry-based methods were employed to enhance artifact removal from EEG signals and to improve the EEG results’ generalizability across subjects. Results Gait analysis highlighted a predominant involvement of inhibition (p $$\le$$ ≤ 0.048 for the most of spatial-temporal parameters and p $$\le$$ ≤ 0.005 for GPS and GVS of Hip Flex-Extension among the kinematic parameters), whereas working memory engagement emerged mainly in response to higher cognitive demands ( p = 0.037 for mostly spatial-temporal parameters and p $$\le$$ ≤ 0.003 for GPS and GVS of Hip Flex-Extension among the kinetic parameters). Relative alpha desynchronization at F4 and at F3 emerged as candidate sensor-level EEG biomarkers of working memory- and inhibitory-related processes (p $$\le$$ ≤ 5 $$\times 10^{-5}$$ × 10 - 5 ) in the earlobe-reference configuration. A cross-domain mixed-effects analysis further showed that, among the tested EEG–gait associations, relative alpha power at F3 during the Go-NoGo task was significantly associated with stride length in the earlobe-reference configuration, suggesting that this sensor-level EEG feature may represent a candidate marker for monitoring the interaction between inhibitory control and gait adaptation during dual-task walking. Conclusions This study supports EEG-based monitoring of cognitive processes during walking and aids the design of tailored cognitive rehabilitation for motor-impaired individuals.