Research on Immersive Exhibition Design of Central Plains Intangible Heritage Culture Based on Multimodal Perceptual Evaluation

The transformation of intangible cultural heritage exhibitions into immersive narratives is subject to the inherent limitations of the traditional user experience assessment paradigm, the over-reliance on subjective questionnaires, and the lack of a real-time, objective feedback mechanism for physiological dimensions. To overcome this evaluation bottleneck, this study proposes a multimodal perceptual evaluation framework that accurately quantifies visitors' immersion state using collaborative EEG and eye-tracking signals. Under this framework, we construct a multi-head cross-attention network to dynamically fuse the temporal dependence of EEG signals with the spatial characteristics of visual gaze, and we innovatively introduce the honest causal forest paradigm to tightly infer and isolate asymmetric risk factors (such as visual dizziness and interaction latency) that induce negative interactions. In the empirical evaluation of "Yinxu Digital Immersion Exhibition", the MHCAN model achieved an F1 score of 94.5% on the immersion-state recognition task, which is significantly better than the traditional single-modal baseline. Further quantitative physiological analysis shows that, compared with static graphic displays, the VR immersive interaction paradigm can significantly enhance users' mental input and positive emotional valence. Based on the above multidimensional physiological characterization, this study finally proposes a theoretical framework for an adaptive exhibition control strategy driven by real-time neural and visual feedback. This study not only provides a rigorous closed-loop algorithmic basis for the objective quantification and dynamic optimization of subjective cultural experience, but also demonstrates broad potential for clinical and practical applications in the fields of adaptive cultural space design, human-computer interaction intervention, and neuroaesthetics.