Unveiling Brainwave Signatures of Devotional Practices Using OBiGSEN: An EEG-Based Study on Hare Krishna Mahamantra Meditation and Kirtan Kriya

A new EEG-based emotion recognition framework achieved 98.18% classification accuracy in distinguishing neural activity during Hare Krishna mantra meditation, Kirtan (musical rendition), and a resting control condition. The study used a 32-channel headset and applied Enhanced Synthetic Minority Oversampling Technique, Savitzky-Golay filtering, and Independent Component Analysis for signal processing. A Hybrid Burg-Hilbert Transform extracted spectral-temporal features, classified by an Optimized Bidirectional Gated Squeeze-Excited Network. Statistical tests confirmed significant differences in alpha and theta brain activity across conditions. Right-hemisphere coherence characterized meditation, while left-frontal engagement marked Kirtan. The findings suggest distinct neural profiles for devotional practices, with potential therapeutic applications for stress, anxiety, and depression.

Study at a glance

Abstract

Electroencephalography-based emotion recognition remains an active research area. Unlike many existing pipelines that rely on large, redundant feature sets, we propose a streamlined yet robust framework to examine the influence of chanting the Hare Krishna mantra (HKM) and its musical rendition (Kirtan) on neuronal dynamics-a relatively unexplored direction. EEG signals were acquired using a 32-channel headset under three conditions: HKM meditation, Kirtan listening, and resting control. To address class imbalance, Enhanced Synthetic Minority Oversampling Technique (ESMOTE) was applied. Signals were denoised using Savitzky-Golay filtering and Independent Component Analysis (ICA), followed by spectral-temporal feature extraction using a Hybrid Burg-Hilbert Transform (HBHT). We introduce the Optimized Bidirectional Gated Squeeze-Excited Network (OBiGSEN), with parameters fine-tuned via the Chaotic Honey Badger Algorithm. OBiGSEN achieved 98.18% classification accuracy and a mean squared error of 0.0485%, outperforming contemporary baselines. It effectively distinguishes meditationinduced neuronal coherence, Kirtan's balanced engagement, and the control state's disorganized patterns. Statistical validation using Kruskal-Wallis and Mann-Whitney U tests confirmed significant differences in alpha and theta activity across conditions. Topographic EEG analysis revealed strong right-hemisphere coherence during meditation and left-frontal engagement during Kirtan. These findings highlight the distinct neural profiles associated with devotional practices. The study underscores the therapeutic potential of mantra meditation and devotional music for managing stress, anxiety, and depression. The proposed lightweight, accurate EEG pipeline holds substantial promise for real-world clinical and wellness applications.