Analyzing the Potential Contribution of a Meta-Learning Approach to Robust and Effective Subject-Independent, Emotion-Related Time Series Analysis of Bio-signals

Studies in computational intelligence, Год журнала: 2024, Номер unknown, С. 139 - 187

Опубликована: Янв. 1, 2024

Язык: Английский

---

Frequency-specific cortico-subcortical interaction in continuous speaking and listening

eLife, Год журнала: 2024, Номер 13

Опубликована: Дек. 23, 2024

Speech production and perception involve complex neural dynamics in the human brain. Using magnetoencephalography, our study explores interaction between cortico-cortical cortico-subcortical connectivities during these processes. Our connectivity findings speaking revealed a significant connection from right cerebellum to left temporal areas low frequencies, which displayed an opposite trend high frequencies. Notably, high-frequency was absent listening condition. These underscore vital roles of connections within speech network. The results new enhance understanding brain processes, emphasizing distinct frequency-based interactions various regions.

Язык: Английский

---

Interpretability of statistical approaches in speech and language neuroscience

Опубликована: Ноя. 29, 2019

Traditional theoretical models conceive the neural system of speech and language as a set hierarchical modules that transform continuous acoustic stream into discrete concepts. This modular view arises from traditional neuropsychology has largely been backed up by statistical allow for controlled variation few experimental factors at time, thus allowing clear interpretations to be made. Recently, exploration large datasets led emergence more complex can capture patterns distributed across space time. However, interpretation these is challenging due increased correlations spatio-temporal dependencies between variables, which obscure links activations linguistic functions. To guide experimenter data analyst through complexity approaches in neuroscience, we have designed taxonomy delineates trade-off model interpretability.

Язык: Английский

---

Speech and music recruit frequency-specific distributed and overlapping cortical networks

Research Square (Research Square), Год журнала: 2022, Номер unknown

Опубликована: Окт. 11, 2022

To what extent do speech and music processing rely on domain-specific domain-general neural networks? Adopting a dynamical system framework, we investigate the presence of frequency-specific network-level selectivity combine it with statistical approach in which clear distinction is made between shared, preferred, category-selective responses. Using intracranial EEG recordings 18 epilepsy patients listening to natural continuous music, show that majority focal activity shared processing. Our data also reveal an absence regional selectivity. Instead, restricted dis- tributed coherent oscillations, typical spectral fingerprints. work addresses longstanding debate redefines epistemological posture how map cognitive brain functions.

Язык: Английский

---

Beyond Linear Neural Envelope Tracking: A Mutual Information Approach

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2022, Номер unknown

Опубликована: Авг. 15, 2022

Abstract The human brain tracks the temporal envelope of speech, which contains essential cues for speech understanding. Linear models are most common tool to study neural tracking. However, information on how is processed can be lost since nonlinear relations precluded. As an alternative, mutual (MI) analysis detect both linear and relations. Yet, several different approaches calculating MI applied without consensus approach use. Furthermore, added value techniques remains a subject debate in field. To resolve this, we analyses electroencephalography (EEG) data participants listening continuous speech. Comparing approaches, conclude that results reliable robust using Gaussian copula approach, first transforms standard Gaussians. With this valid technique studying Like models, it allows spatial interpretations processing, peak latency analyses, applications multiple EEG channels combined. Finally, demonstrate components single-subject level, beyond limits models. We more informative Significance statement In present study, addressed key methodological considerations applications. Traditional methodologies require estimation probability distribution at first. show step introduce bias and, consequently, severely impact interpretations. propose parametric method, demonstrated against biases. Second, analysis, there variance explain proving its statistically powerful tracking than addition, retains characteristics processing when complex deep networks.

Язык: Английский

---