The Hierarchical Cortical Organization of Human Speech Processing

Area of Science:

• Neuroscience
• Cognitive Science
• Linguistics

Background:

• Speech comprehension involves transforming auditory spectral features into semantic meaning.
• Previous research often uses simplified speech stimuli and block designs, potentially limiting ecological validity.

Purpose of the Study:

• To investigate the neural representations of semantic meaning during natural speech comprehension.
• To map the hierarchical processing of spectral, articulatory, and semantic speech features in the brain.
• To compare findings using natural speech with traditional methods.

Main Methods:

• Functional magnetic resonance imaging (fMRI) experiment with passive listening to natural narrative speech.
• Voxelwise modeling to predict blood-oxygen-level-dependent (BOLD) responses using spectral, articulatory, and semantic speech features.
• Variance partitioning analysis to determine unique contributions of each feature space.

Main Results:

• Hierarchical representation of speech features along the temporal lobe, from spectral in primary auditory cortex (A1) to semantic in the superior temporal sulcus (STS).
• Both cerebral hemispheres show equal and active involvement in speech perception and interpretation.
• Early auditory responses in the STS correlate more strongly with semantic than spectral features.

Conclusions:

• Natural speech processing involves hierarchical feature extraction occurring early in the auditory cortex.
• Both hemispheres play a significant and balanced role in understanding spoken language.
• The methodology offers an efficient alternative for studying speech representations without artificial speech stimuli or block designs.