Beyond Linear Neural Envelope Tracking: A Mutual Information Approach

Abstract

The human brain tracks the temporal envelope of speech, which contains essential cues for speech understanding. Linear models are the most common tool to study neural envelope tracking. However, information on how speech is processed can be lost since nonlinear relations are precluded. As an alternative, mutual information (MI) analysis can detect both linear and nonlinear relations. Yet, several different approaches to calculating MI are applied without consensus on which approach to use. Furthermore, the added value of nonlinear techniques remains a subject of debate in the field. To resolve this, we applied linear and MI analyses to electroencephalography (EEG) data of participants listening to continuous speech. Comparing the different MI approaches, we conclude that results are most reliable and robust using the Gaussian copula approach, which first transforms the data to standard Gaussians. With this approach, the MI analysis is a valid technique for studying neural envelope tracking. Like linear models, it allows spatial and temporal interpretations of speech processing, peak latency analyses, and applications to multiple EEG channels combined. Finally, we demonstrate that the MI analysis can detect nonlinear components on the single-subject level, beyond the limits of linear models. We conclude that the MI analysis is a more informative tool for studying neural envelope tracking.

Neural substrates and behavioral relevance of speech envelope tracking: evidence from post-stroke aphasia

“…Envelope processing and auditory abilities in general have received limited attention in aphasia research over the past decades. Nevertheless, both the current paper and recent literature convincingly demonstrate the impairment of such fundamental auditory processes in aphasia at the behavioral (Kries et al, 2023b) and neural (De Clercq et al, 2023a;Kries et al, 2023a) level. Efficient rise time processing is, however, critical for higher-level speech comprehension as well, as these carry critical cues for detecting and segmenting phonemes, syllables and words (Goswami et al, 2011;Hämäläinen et al, 2012;Oganian and Chang, 2019).…”

“…Among various disorders, neural tracking has found application in the study of aphasia, a language disorder commonly caused by a stroke in the language-dominant left hemisphere of the brain (Papathanasiou and Coppens, 2017). In response to narratives, individuals with post-stroke aphasia (IWA) display decreased neural tracking of the envelope and linguistic speech representations (Kries et al, 2023a;De Clercq et al, 2023a, 2024b. Neural tracking measures further predicted aphasia at the individual level (De Clercq et al, 2023b), with prediction accuracy almost entirely driven by envelope tracking rather than linguistic tracking (De Clercq et al, 2024b).…”

Exploring neural tracking of acoustic and linguistic speech representations in individuals with post-stroke aphasia

“…Once neural tracking in aphasia is better understood, the application potential of this method could be manifold. A study that is currently in preparation found that IWA can be distinguished from healthy controls with 83% accuracy based on neural envelope tracking with mutual information of only 5 to 7 minutes of EEG data (De Clercq et al, 2023). Looking a step further into the future, an aphasia diagnosis based on processing levels of different speech representations could complement behavioral diagnosis and inform speech-language pathologists further which functions should be trained in therapy.…”

Exploring neural tracking of acoustic and linguistic speech representations in individuals with post‐stroke aphasia

“…Once neural tracking in aphasia is better understood, the application potential of this method could be manifold. A study that is currently in preparation found that IWA can be distinguished from healthy controls with 83% accuracy based on neural envelope tracking with mutual information of only 5–7 min of EEG data (De Clercq et al, 2023). Looking a step further into the future, an aphasia diagnosis based on processing levels of different speech representations could complement behavioral diagnosis and inform speech‐language pathologists further which functions should be trained in therapy.…”

“…healthy controls with 83% accuracy based on neural envelope tracking with mutual information of only 5-7 min of EEG data(De Clercq et al, 2023). Looking a step further into the future, an aphasia diagnosis based on processing levels of different speech representations could complement behavioral diagnosis and inform speech-language pathologists further which functions should be trained in therapy.Test-retest practice effects (i.e., improved performance on repeated tests due to remembering items or training test-specific skills) could be avoided during therapy follow-up.…”