Multimodal and hemispheric graph-theoretical brain network predictors of learning efficacy for frontal alpha asymmetry neurofeedback

Abstract

BackgroundEEG neurofeedback using frontal alpha asymmetry (FAA) has been widely used for emotion regulation training, but its effectiveness is controversial because some participants failed to learn self-regulation of FAA signals. It is desired to reveal neuroimaging predictors of learning ability in FAA neurofeedback, which cannot only improve our knowledge of the mechanisms of neurofeedback but can also enable personalized neurofeedback scheme. Previous studies have indicated that individual differences in neurofeedback training can be traced to neuroanatomical and neurofunctional baseline features. However, these studies only focused on regional measures of brain structure or function, and overlooked possible neural correlates of brain network measures. Besides, no neuroimaging predictors for FAA neurofeedback protocol have been reported so far. MethodsThe present study designed a single-blind pseudo-controlled FAA neurofeedback experiment and collected multimodal neuroimaging data (structural magnetic resonance imaging and diffusion tensor imaging, and resting-state functional MRI) from healthy participants before neurofeedback training. We assessed the learning performance for evoked EEG modulations during training (performance index L1) and at rest (performance index L2), and investigated performance related predictors based on a combined analysis of multimodal brain networks and graph-theoretical features. ResultsParticipants from both real and sham groups could increase their FAA during training blocks by maintaining a positive emotional state, but only the real group showed a signi cant increase in FAA at rest. The predictors of learning performance were different for evoked EEG modulations during training blocks and at rests: L1 was correlated with the graph-theoretical metrics (normalized clustering coe cient and local e ciency) of the right hemispheric gray matter network and functional brain network, while L2 was correlated with the graph-theoretical metrics (local and global e ciency) of the whole-brain and left hemispheric functional brain network. ConclusionsThe individual differences in FAA neurofeedback learning could be explained with individual variation in structural/functional architecture, and the correlated global topological metrics of different learning performance indices showed different laterality of hemispheric brain networks. These results provided an insight into the neural correlates of inter-individual differences in neurofeedback learning.

Decoded EEG Neurofeedback-Guided Cognitive Reappraisal Training for Emotion Regulation

“…Compared with fMRI neurofeedback, EEG neurofeedback has the advantage of being more widely available, cost-effective, and including ambulatory settings. Existing EEG studies for emotion regulation used single-frequency band EEG activity in specific brain regions for neurofeedback signal calculation, such as alpha symmetry (Quaedflieg et al, 2016;Mennella et al, 2017;Li et al, 2023), and high-beta down-training (Paquette et al, 2009;Cheon et al, 2016;Wang et al, 2019). However, the generation, perception, and regulation of emotions depend on the synergistic work of multiple brain regions or brain networks (Morawetz et al, 2020).…”

“…Particularly in light of its relatively low cost, portability, and easy of use compared to fMRI, the use of EEG neurofeedback to improve emotion regulation hold promise for improving results via personalized, targeted protocols. For example, frontal alpha asymmetry EEG neurofeedback has been applied in both healthy (Quaedflieg et al, 2016;Li et al, 2023) and depressed participants for emotion regulation training (Wang et al, 2019). One recently published study designed a real-time EEG-based brain-computer interface system for neurofeedback training (Huang et al, 2021).…”