Hierarchy of speech-driven spectrotemporal receptive fields in human auditory cortex

Area of Science:

• Neuroscience
• Auditory Processing
• Speech Perception

Background:

• Cortical speech processing exhibits hierarchical organization, with early areas handling acoustic details and later areas abstract patterns.
• The precise speech information encoded at different hierarchical levels remains incompletely understood.
• Spectrotemporal receptive fields (STRFs) offer a method to investigate cortical speech processing.

Purpose of the Study:

• To estimate speech-driven STRFs from fMRI data of cortical responses to continuous speech.
• To characterize the spectrotemporal modulation (STM) patterns processed by different brain regions.
• To elucidate the hierarchical encoding of acoustic and linguistic information in speech perception.

Main Methods:

• fMRI was used to measure cortical responses to aurally presented sentences.
• A novel method involving filtering spectrotemporal modulations (STMs) was employed to estimate STRFs.
• A data-driven clustering algorithm categorized estimated STRFs, and 'behavioral STRFs' were derived from intelligibility judgments.

Main Results:

• STRFs in the supratemporal plane broadly represented STMs, while those in the lateral temporal lobe captured intelligibility-relevant patterns.
• A bilateral organization was observed, with posterior-lateral regions processing phonological information and anterior-lateral regions processing word/phrase information.
• Lateral Heschl's gyrus preferentially processed vocalic (pitch) information.

Conclusions:

• Cortical speech processing is hierarchically organized, with distinct regions specializing in different levels of speech information.
• STRF analysis effectively maps spectrotemporal patterns to specific brain regions involved in speech perception.
• This hierarchical encoding, from acoustic to linguistic features, is crucial for successful speech recognition.