Abstract
Motor imagery (MI) is an endogenous mental process and is commonly used as an electroencephalogram (EEG)-based brain–computer interface (BCI) strategy. Previous studies of P300 and MI-based (without online feedback) BCI have shown that mental states like fatigue can negatively affect participants’ EEG signatures. However, exogenous stimuli cause visual fatigue, which might have a different mechanism than endogenous tasks do. Furthermore, subjects could adjust themselves if online feedback is provided. In this sense, it is still unclear how fatigue affects online MI-based BCI performance. With this question, 12 healthy subjects are recruited to investigate this issue, and an MI-based online BCI experiment is performed for four sessions on different days. The first session is for training, and the other three sessions differ in rest condition and duration—no rest, 16-min eyes-open rest, and 16-min eyes-closed rest—arranged in a pseudo-random order. Multidimensional fatigue inventory (MFI) and short stress state questionnaire (SSSQ) reveal that general fatigue, mental fatigue, and distress have increased, while engagement has decreased significantly within certain sessions. However, the BCI performances, including percent valid correct (PVC) and information transfer rate (ITR), show no significant change across 400 trials. The results suggest that although the repetitive MI task has affected subjects’ mental states, their BCI performances and feature separability within a session are not affected by the task significantly. Further electrophysiological analysis reveals that the alpha-band power in the sensorimotor area has an increasing tendency, while event-related desynchronization (ERD) modulation level has a decreasing trend. During the rest time, no physiological difference has been found in the eyes-open rest condition; on the contrary, the alpha-band power increase and subsequent decrease appear in the eyes-closed rest condition. In summary, this experiment shows evidence that mental states can change dramatically in the intensive MI-BCI practice, but BCI performances could be maintained.
Highlights
Motor imagery (MI)-based brain–computer interface (BCI) provides a novel communication method by decoding a human’s motor intention from brain signals such as electroencephalogram (EEG) (He et al, 2020)
sensorimotor rhythms (SMRs)-based MI-BCI enables the control of a robotic arm (Meng et al, 2016), virtual helicopters (Royer et al, 2010), communication (Perdikis et al, 2014), and video games (Bonnet et al, 2013)
We found that general fatigue, mental fatigue, and distress have significantly increased while engagement has decreased after a session of intensive MI-BCI operation
Summary
Motor imagery (MI)-based brain–computer interface (BCI) provides a novel communication method by decoding a human’s motor intention from brain signals such as electroencephalogram (EEG) (He et al, 2020). Fatigue effects during prolonged BCI operation are usually recognized to be negative factors (e.g., in P300-BCI) (Käthner et al, 2014). Another non-BCI study has shown that one session of MI training does not induce neuromuscular fatigue (Rozand et al, 2014). This could be generally considered as follows: EEG is a highly nonstationary signal, and a dramatic variation of features would be detrimental to the performance of BCI during prolonged operation. The fatigue effects on the MI-BCI with feedback require further investigations
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
