Fast mental states decoding in mixed reality.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Virtual Reality

Background:

• Brain-Computer Interface (BCI) technology enables real-time monitoring and decoding of brain activity.
• Combining BCI with virtual and mixed reality (MR) offers potential for adaptive learning in ecological scenarios.
• Discriminating brain states during MR experiences is crucial for tailoring data presentation to user activity.

Purpose of the Study:

• To assess continuous electroencephalographic (EEG)-based discrimination of cognitive tasks within MR environments.
• To evaluate the performance of Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) classifiers for brain state detection.
• To explore the potential of BCI-controlled MR for adaptive information processing.

Main Methods:

• Recorded EEG data from participants engaged in MR scenarios using the eXperience Induction Machine (XIM).
• Applied LDA and SVM classifiers for continuous, single-trial classification of cognitive states (spatial navigation, reading, calculation).
• Utilized a 5-second time-window, shifting every 200 ms, for dynamic classification and mental workload differentiation.

Main Results:

• Achieved high accuracy in discriminating multiple brain states and differentiating between high and low mental workload using both LDA and SVM.
• Demonstrated superior performance of the LDA classifier compared to SVM.
• Showcased the feasibility of real-time EEG-based brain state prediction within an MR setting.

Conclusions:

• The proposed BCI-MR approach enables accurate and dynamic discrimination of cognitive states.
• LDA classifier shows promising results for BCI-controlled MR applications.
• Predicting brain states can drive adaptive data representation to enhance information processing in MR environments.