eeg-cortical-speech-tracking-for-subjective-cognitive-decline

name	EEG Cortical Speech Tracking for Subjective Cognitive Decline
description	Research methodology and findings on using EEG cortical tracking strength (CTS) as a neural marker for early-stage cognitive decline. Combines speech encoding models with linguistic feature analysis to detect subjective cognitive decline (SCD). Use when: studying EEG speech processing, cognitive decline biomarkers, neural tracking of naturalistic speech, or linguistic feature encoding in aging populations.
metadata	{"arxiv_id":"2509.21277","published":"2025-09-25","authors":"Matthew King-Hang Ma, Yun Feng, Cloris Pui-Hang Li, Manson Cheuk-Man Fong","keywords":["EEG","cortical speech tracking","subjective cognitive decline","dementia risk","speech encoding model","linguistic features"],"category":"q-bio.NC"}

EEG Cortical Speech Tracking for Subjective Cognitive Decline (SCD)

Overview

This skill documents the methodology and key findings from arXiv:2509.21277, which investigates how self-perceived cognitive worsening shapes neural dynamics during naturalistic speech perception, and identifies cortical tracking strength (CTS) as a potential neural marker for early-stage cognitive decline.

Background

Subjective Cognitive Decline (SCD) doubles dementia risk. This study explores how self-perceived cognitive decline affects neural processing of naturalistic speech with varying prosodic contexts.

Methodology

Experimental Design

• Participants: 60 cognitively normal older adults
• Stimuli: Speech samples with four expressive styles:
  • Scrambled (low-level acoustic)
  • Descriptive (prosodically flat)
  • Dialogue (natural conversational)
  • Exciting (high prosodic variation)

Speech Encoding Models

Three speech representation layers mapped to EEG:

1. Acoustic features (low-level)
2. Subsyllabic segmentation (linguistic unit boundaries)
3. Phonotactic features (probability of sound sequences)

Analysis

• Cortical Tracking Strength (CTS): Correlation between speech features and EEG signals
• Comparison: Linguistic vs. acoustic feature tracking
• Correlation: CTS with SCD severity scores

Key Findings

1. Linguistic Features Outperform Acoustic

Subsyllabic linguistic feature models showed stronger CTS than acoustic models across all participants.

2. SCD-Related Neural Markers

Greater SCD severity associated with weaker CTS for:

• Subsyllabic linguistic features (but NOT acoustic features)
• Prosodically flat speech (scrambled and descriptive styles)

3. Specificity of Impairment

• Linguistic processing impaired in SCD
• Basic acoustic processing preserved
• Prosodic variation modulates the effect

Implications

Clinical Biomarker

CTS of higher-level linguistic features during prosodically flat speech may serve as an early neural marker for cognitive decline before clinical symptoms emerge.

Theoretical Insights

• Linguistic (not acoustic) processing vulnerable in early cognitive decline
• Prosodic context can compensate or exacerbate processing deficits
• Natural speech paradigms reveal subtle neural changes

Methodology Applications

When to Apply

• Studying early dementia biomarkers
• Investigating speech processing in aging
• Developing neural markers for cognitive screening
• Analyzing naturalistic speech perception

Key Technical Components

1. Encoding models: Map speech features → EEG
2. Linguistic feature extraction: Subsyllabic segmentation
3. Prosodic manipulation: Control speech expressiveness
4. CTS computation: Cross-correlation with lag optimization

Experimental Considerations

• Use naturalistic speech (not isolated syllables)
• Include multiple prosodic conditions
• Control for hearing ability and attention
• Account for linguistic background

Related Research Areas

• EEG speech tracking
• Cognitive decline biomarkers
• Naturalistic neuroscience
• Linguistic processing in aging
• Neural encoding models

Limitations and Future Directions

Current Limitations

• Cross-sectional design (causality unclear)
• Cognitive normal participants (clinical validation needed)
• Single language (Mandarin Chinese)

Future Work

• Longitudinal studies tracking CTS changes
• Clinical populations (MCI, dementia)
• Multi-language validation
• Integration with other biomarkers (MRI, CSF)

Citation

Ma, M. K.-H., Feng, Y., Li, C. P.-H., & Fong, M. C.-M. (2025). More than a feeling: 
Expressive style influences cortical speech tracking in subjective cognitive decline. 
arXiv preprint arXiv:2509.21277.

Keywords for Activation

EEG, cortical speech tracking, subjective cognitive decline, SCD, dementia biomarker, speech encoding model, linguistic features, prosody, naturalistic speech, neural tracking, cognitive aging