Speech categorization is better described by induced rather than evoked neural activity.

Area of Science:

• Neuroscience
• Cognitive Science
• Speech Processing

Background:

• Categorical perception (CP) explains how the brain processes variable speech acoustics into discrete categories.
• Understanding the neural mechanisms of CP is crucial for deciphering speech perception.
• Electroencephalogram (EEG) offers insights into brain activity during cognitive tasks.

Purpose of the Study:

• To investigate the neural correlates of categorical perception (CP) in speech.
• To compare the effectiveness of evoked versus induced EEG activity in decoding speech categorization.
• To identify specific brain regions and frequency bands associated with CP.

Main Methods:

• Listeners performed a speech categorization task on a vowel gradient while EEG was recorded.
• Source-reconstructed EEG data were analyzed for band-specific evoked and induced neural activity.
• Support vector machine models were used to decode speech categorization from neural data.

Main Results:

• Induced gamma-band (γ-band) oscillations showed superior decoding accuracy (~95%) for speech categories compared to evoked beta-band (β-band) activity (~70%).
• Induced γ-band activity was dominant for CP decoding in the left hemisphere, while θ-band activity was dominant in the right hemisphere.
• Induced γ-band modulations were most strongly linked to behavioral CP, with specific brain regions identified for both induced and evoked activity.

Conclusions:

• Induced, high-frequency brain rhythms, particularly in the gamma band, play a dominant role in the category-level organization of speech.
• The findings highlight the distinct contributions of evoked and induced neural activity to speech categorization.
• Hemispheric differences in neural processing (gamma vs. theta bands) suggest specialized roles in speech perception.