Neural Mechanisms of Semantic Processing in Japanese-Chinese Bilinguals: A Study Based on Deep Learning

Aiming at the problem that the different representation patterns generated by the brains of Japanese and Chinese bilinguals when they perform semantic processing tasks have poor decoding effect using traditional machine learning algorithms, this paper analyzes the task-state functional magnetic resonance imaging (tfMRI) data of 15 Japanese and Chinese bilinguals based on deep learning network model for the first time, and further explores the neural mechanism of the brains of Japanese and Chinese bilinguals. The experimental results showed that the model effectively decodes the neural information related to semantic processing, that is, the bilateral superior temporal gyrus (STG), the middle temporal gyrus (MTG), the angular gyrus (AG), the triangular part of the inferior frontal gyrus (PT), the right subfrontal gyrus (PO), and the supramarginal gyrus (SMG). And the left fusiform gyrus (FFG) showed significant activation differences during semantic processing. The results validate the potential of deep learning models in the study of complex brain neural mechanisms, and provide new theoretical support for the neural basis of bilingual semantic processing.