Decoding auditory frequencies and directions based on brain functional features

Decoding auditory stimuli based on brain function data is of great significance to understand auditory functional mechanism. At present, there is still controversy about whether the brain processing mechanism of auditory stimuli is parallel hierarchical processing or distributed processing [1]. Different from previous studies that used univariate analysis to study auditory processing, this study intends to build a decoding model and study the auditory processing mechanism from the perspective of multivariate pattern analysis. In this paper, we analyzed functional MRI data from 27 subjects under perception of different auditory frequencies and directions, using brain activation and functional connectivity as features, and using support vector machine for decoding. The decoding accuracy of frequencies and directions was 70.7% and 71.6% with brain activation features. On the other hand, the accuracy rate reached 73.7% and 77.7% respectively with functional connectivity features. Then we analyzed the weights and found that the activation patterns in precuneus and the superior temporal gyrus (STG) contributed to sound frequency discrimination, and STG also represented differences in direction. The connectivity patterns between the bilateral precuneus showed obvious changes under different frequency conditions, while the bilateral middle occipital gyrus and STG showed significant changes under different directions of sound stimulation. The results support a distributed auditory processing model.