High-Frequency Power Reflects Dual Intentions of Time and Movement for Active Brain–Computer Interface

Abstract

Active brain-computer interface (BCI) provides a natural way for direct communications between the brain and devices. However, its detectable intention is very limited, let alone of detecting dual intentions from a single electroencephalography (EEG) feature. This study aims to develop time-based active BCI, and further investigate the feasibility of detecting time-movement dual intentions using a single EEG feature. A time-movement synchronization experiment was designed, which contained the intentions of both time (500 ms vs. 1000 ms) and movement (left vs. right). Behavioural and EEG data of 22 healthy participants were recorded and analyzed in both the before (BT) and after (AT) timing prediction training sessions. Consequently, compared to the BT sessions, AT sessions led to substantially smaller absolute deviation time behaviourally, along with larger high-frequency event-related desynchronization (ERD) in frontal-motor areas, and significantly improved decoding accuracy of time. Moreover, AT sessions achieved enhanced motor-related contralateral dominance of event-related potentials (ERP) and ERDs than the BT, which illustrated a synergistic relationship between the two intentions. The feature of 20–60 Hz power can simultaneously reflect the time and movement intentions, achieving a 73.27% averaged four-classification accuracy (500 ms-left vs. 500 ms-right vs. 1000 ms-left vs.1000 ms-right), with the highest up to 93.81%. The results initiatively verified the dual role of high-frequency (20–60 Hz) power in representing both the time and movement intentions. It not only broadens the detectable intentions of active BCI, but also enables it to read user’s mind concurrently from two information dimensions.

Keywords

• Active brain-computer interface
• high-frequency power
• single-interval timing prediction
• movement
• dual intention