Functional Organization of Sound Categories in Human Auditory Cortex

Abstract

How the human auditory cortex processes complex auditory stimuli, in addition to lower-level tonotopical organization, remains not well understood. In this experiment, 13 subjects underwent functional magnetic resonance imaging while being presented brief sounds from six categories: adult speech, kids speech, animals, instruments, objects, and Pokémon. There were regions along Heschl’s gyrus and superior temporal gyrus (STG) that were more responsive to instruments and speech respectively. Representational similarity analysis revealed a strong distinction between the neural responses to speech and non-speech categories, particularly in regions around STG. Spectral analysis and animacy ratings obtained for all sounds suggest that frequency, a lower-level sound feature, may underlie the functional representation of sound categories in both primary auditory cortex (PAC) and STG, while animacy may drive the organization of only STG. A classifier trained on the neural data was able to predict the sound category presented with slightly above chance accuracy. Specific binary decoding analyses for speech and instrument categories performed with higher accuracy. Overall, these findings indicate that the auditory system may exhibit both parallel and hierarchical processing of auditory information. The functional organization of the six sound categories tested in this experiment is not distinct enough for a classifier to decode categories with high accuracy.