Influence of User Tasks on EEG-based Classification Performance in a Hazard Detection Paradigm.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Attention-based brain-computer interfaces (BCIs) require user engagement for effective control and environmental perception.
• Auxiliary tasks like mental counting or button presses are often used to enhance user motivation and BCI performance.
• The optimal auxiliary task for BCI paradigms, particularly in complex scenarios like hazard detection, remains an area of investigation.

Purpose of the Study:

• To compare the efficacy of mental counting versus button-press tasks within a hazard detection BCI paradigm.
• To evaluate the impact of these tasks on electroencephalogram (EEG) classification performance.
• To assess user preference between the two auxiliary tasks in the context of BCI operation.

Main Methods:

• Implemented a hazard detection paradigm using driving videos.
• Recorded electroencephalogram (EEG) data from participants performing either mental counting or button-press tasks.
• Analyzed BCI classification performance and user preference data for both tasks.
• Utilized data preprocessing techniques, including projection of motor-related potentials.

Main Results:

• Button-press tasks resulted in significantly higher binary classification performance based on EEG data compared to mental counting.
• Users expressed a higher preference for the button-press task over mental counting.
• Evoked response amplitudes were greater for the mental counting task, even after accounting for motor potentials.

Conclusions:

• Button-press tasks appear to be a more effective auxiliary task for attention-based BCIs, enhancing both prediction performance and user satisfaction.
• The choice of auxiliary task significantly influences BCI outcomes and user experience.
• Further research may explore the neural underpinnings of task-specific BCI performance.