FMRI Decoding of Auditory Information Based on Multi-voxel Pattern Analysis

Abstract

Auditory information decoding is very important for human-machine interaction and brain-inspired study. Whether the auditory content and location information can be both decoded from the human brain is still unclear. In this study, we employed support vector machine (SVM) classification method to decoding the auditory category and direction based on brain signals in two auditory core regions (STG and BA 41, 42 regions). Results showed that the proposed method successfully decoded the categories and directions with high accuracies than chance level (p<0.05). The results also suggested the totally blind subjects are better at discriminating the auditory categories and directions in STG. The findings in this study demonstrate the effectiveness of SVM in brain auditory information decoding and also promote our understanding of the auditory cognitive mechanism.