IMMERSIVE NEUROLEARNING: A CNN–LSTM-DRIVEN BCI–XR HYBRID MODEL FOR ADAPTIVE NEUROCOGNITIVE EDUCATION

Abstract

The convergence of Brain–Computer Interfaces (BCIs) and Extended Reality (XR) is bringing about promising new paradigms for implementing real-time, data-driven, and learner-specific adaptive educational systems. Immersive NeuroLearning enables advanced deep learning capabilities based on Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) network for real-time analysis of electroencephalography (EEG) signals. As a result of the integration of spatial patterns of neural activity and temporal dynamics of cognitive states, the model achieves high prediction accuracy for engagement and mental workload, and allows the XR learning environment to dynamically and continuously adapt. The simulation results show that the proposed model can reach 92.8% accuracy, which is higher than traditional single CNN (85.4%) and LSTM (87.2%) models. This enhanced result further shows the representativeness of spatial and temporal neural features. In addition, the prediction-based adaptive XR interface enables users to better enjoy the immersive experience, maintain attention stability and keep overall cognitive retention by adjusting the complexity of stimulus and interactions with the learner’s online mental state. Taken together the findings emphasize the transformative potential of neuroadaptive learning systems. By combining high-fidelity neural decoding with interactive XR experiences, Immersive NeuroLearning represents a fundamental advance for intelligent, personalized, and cognition-aware educational systems that can enable lifelong, data-driven human–AI co-learning, and thus holds the promise for the development of next-generation educational technologies.