EEG-based characterization of auditory attention and meditation: an ERP and machine learning approach

Attar, Eyad Talal · 2025 · Crossref

DOI: 10.3389/fnhum.2025.1616456

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates how meditation modulates neural responses to auditory stimuli and whether electroencephalography (EEG) biomarkers can effectively distinguish between meditative states and cognitive tasks. Motivated by the need to understand the neurophysiological mechanisms of meditation and to develop real-time brain-computer interfaces, the research combines event-related potential (ERP) analysis, spectral power assessment, and machine learning classification. The authors aim to determine if meditation induces distinct neural signatures detectable through standard EEG metrics and if these signatures can be leveraged for automated cognitive state monitoring. The researchers utilized data from 13 participants (aged 24–58) sourced from the OpenNeuro dataset, categorized by meditation experience into novice, intermediate, and experienced groups. Participants performed an auditory oddball task involving standard tones, infrequent oddball tones, and distracting noise bursts. EEG data were preprocessed using independent component analysis to remove artifacts, followed by time-frequency analysis using continuous Morlet wavelets. The study analyzed ERP components, specifically the P300, and spectral power in theta, alpha, and beta bands. A Random Forest classifier was trained using features including ERP amplitude, spectral power, spectral entropy, and functional connectivity metrics to differentiate between meditation and cognitive task states. Model performance was evaluated using fivefold cross-validation and leave-one-subject-out (LOSO) validation to ensure generalizability. The results demonstrated that oddball stimuli elicited significantly larger P300 amplitudes compared to standard stimuli, indicating increased attentional allocation. Spectral analysis revealed that meditation was associated with increased frontal alpha and beta power and decreased central theta power, suggesting enhanced internal focus and reduced cognitive load. A positive correlation was found between meditation experience and frontal alpha power ($r = 0.45, p < 0.03$). The Random Forest classifier achieved an average accuracy of 86.7% and an ROC-AUC of 0.89 in distinguishing meditation from cognitive tasks. Under the more conservative LOSO validation, the model maintained an accuracy of 81.5% and an ROC-AUC of 0.85. Key discriminative features included P300 amplitude, frontal alpha power, and beta coherence. The findings provide strong evidence that meditation induces distinct, measurable neural modifications detectable via ERP and spectral analysis. The high classification accuracy demonstrates the viability of integrating EEG biomarkers with machine learning for real-time cognitive and emotional state monitoring. These results support the development of neurofeedback systems and brain-computer interfaces for mental health applications, offering a pathway for tailored interventions that enhance cognitive function and emotional regulation in both clinical and everyday settings.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• neuro workload indices