Next-Generation Neural Mass Models Reproduce Features of Speech Processing

Area of Science:

• Neuroscience
• Computational Neuroscience
• Auditory Processing

Background:

• Syllable segregation in speech processing is vital but challenging due to continuous acoustic input.
• Neural speech tracking aligns brain activity with speech rhythm, but underlying cortical mechanisms are unclear.
• Competing hypotheses for neural speech tracking include phase-resetting and evoked responses.

Purpose of the Study:

• To investigate the cortical circuits underlying neural speech tracking.
• To evaluate if a biophysical neural mass model can replicate features of neural speech tracking.
• To compare the neural mass model against phenomenological models of phase-resetting and evoked responses.

Main Methods:

• Developed and tested a biophysical next-generation neural mass model.
• Used phenomenological models as algorithmic baselines.
• Evaluated model dynamics through in-silico EEG experiments, Phase Concentration Metric, syllabic rate variations, and Inter Event Phase Coherence analysis.

Main Results:

• All models reproduced sharpness-tuned rhythmic speech tracking.
• The evoked response model required pre-processed acoustic stimuli.
• The neural mass model demonstrated thresholded phase-resetting triggered by speech envelope onsets, producing cross-frequency nested oscillations matching experimental data.

Conclusions:

• The biophysical neural mass model provides a mechanistic link between cortical population dynamics and speech tracking.
• Nonlinear dynamics in the neural mass model explain peak-rate event representations in auditory cortex.
• This model offers insights into how the brain effortlessly segregates syllables from continuous speech.