This research aims to investigate human perception of architectural spaces through EEG signal recognition and develop an innovative interactive-based creative mode. Artificial intelligence algorithms are increasingly used in architectural design, but they are currently limited to imitating and replicating the neural mechanisms of the biological brain. Architectural space extends beyond mere form and style. This paper explores the dynamic connection between architectural space and human cognition by utilizing brain-computer interface (BCI) technology to obtain a direct brainwave signal source. The focus is specifically on the relationship between architectural space and emotional response. Emotion recognition is achieved through the implementation of advanced deep-learning algorithms. Simultaneously, we utilize VR technology to immerse testers in the architectural environment, studying neural responses to various spaces. Our experiments have preliminary revealed spatial features that are linked to human emotions, as well as differences between the designer’s expectations and the audience’s experience. The AIGC model converts emotional data into corresponding geometric spatial features and generates new spatial scenes that contain emotional, temporal, and personal attributes. This approach comprehensively understands the interplay and co-creative essence between architecture and human consciousness, as well as externalizes human consciousness as a method in architectural design.