Mindfulness-based neurofeedback: A systematic review of EEG and fMRI studies

Abstract

Neurofeedback concurrent with mindfulness meditation may reveal meditation effects on the brain and facilitate improved mental health outcomes. Here, we systematically reviewed electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) studies of mindfulness meditation with neurofeedback (mbNF) and followed PRISMA guidelines. We identified 9 fMRI reports, consisting of 177 unique participants, and 9 EEG reports, consisting of 242 participants. Studies of fMRI focused primarily on downregulating the default-mode network (DMN). Although studies found decreases in DMN activations during neurofeedback, there is a lack of evidence for transfer effects, and the majority of studies did not employ adequate controls, for example, sham neurofeedback. Accordingly, DMN decreases may have been confounded by general task-related deactivation. EEG studies typically examined alpha, gamma, and theta frequency bands, with the most robust evidence supporting the modulation of theta band activity. Both EEG and fMRI mbNF have been implemented with high fidelity in clinical populations. However, the mental health benefits of mbNF have not been established. In general, mbNF studies would benefit from sham-controlled RCTs, as well as clear reporting (e.g., CRED-NF).

Keywords: DMN; EEG; fMRI; mindfulness; neurofeedback; theta.

Fig. 1.

Brain networks involved in mindfulness meditation. Central executive network, in red; default-mode network, in blue; salience network, in green. DLPFC: dorsolateral prefrontal cortex; PPC: posterior parietal cortex; PCC: posterior cingulate cortex; MPFC: medial prefrontal cortex; ACC: anterior cingulate cortex; insula: insular cortex. Adapted with permission fromTreves et al. (in press).

Fig. 2.

EEG frequency bands. This visualization demonstrates the differing frequencies of the various EEG bands. Created in Google slides.

Fig. 3.

PRISMA flow diagram depicting number of identified and evaluated articles for concurrent mindfulness and fMRI or EEG neurofeedback procedures.Note. Studies refer to unique samples, while reports refer to publications on said samples. Our review identified four samples which corresponded to more than one published report, as indicated in this flowchart and in the study summary tables (Tables 1and2).

Fig. 4.

fMRI targets for neurofeedback. In blue, default-mode network (DMN) regions. In red, the central executive network (CEN). Networks are taken from the Yeo atlas (Thomas Yeo et al., 2011). *Kim used DMN-SN slope, **Pamplona used sustained attention network.

Fig. 5.

This figure displays the variety in methods of calculating EEG-neurofeedback. In (A), the spatial layouts of the neurofeedback targets are displayed.Hinterberger and Fürnrohr (2016)calculated alpha, theta, beta, gamma, USP, SCP, wide, delta1, and delta2 from CPz.Kosunen et al. (2016)calculated theta and alpha using an average from F3, F4, C3, C4, P3, and P4.Salminen et al. (2023)calculated theta power using an average F3 and F4, while alpha power was calculated as an average of F3, F4, C3, C4, P3, and P4.Brandmeyer and Delorme (2020)used Fz to calculate frontal midline theta.Chen et al. (2021)used FP1, F3, F7 and FP2, F4, F8 to calculate alpha of the left and right frontal lobes, respectively.Dunham et al. (2018)placed the BIS sensor on the left forehead and the temporal fossa.Dunham et al. (2019)placed the BIS sensor on the left or right forehead and the temporal fossa.Prestel et al. (2019)used independent component analysis (ICA) with a 32-channel setup to calculate frontal midline theta.van Lutterveld et al. (2017)used source estimation with a 128-channel setup to determine gamma band PCC activity. In (B), the frequency bands used by each report are displayed. Theta was used the most by five reports, closely followed by alpha, which was used in four reports.