Eyes-Closed Resting EEG Predicts the Learning of Alpha Down-Regulation in Neurofeedback Training

Wenya Nan, Feng Wan, Qi Tang, Chi Man Wong, Boyu Wang, Alessandro Pasquale De RosaFrontiers in Psychology 2018 · DOI

TL;DR

This study reveals that a participant's resting brainwave patterns, specifically their eyes-closed resting EEG, can predict whether they will successfully learn to down-regulate their alpha brain activity during neurofeedback training. This finding offers a promising method for early identification of "non-learners," potentially saving time and resources in neurofeedback interventions.

What Problem Does This Paper Address?

Neurofeedback (NF) training is a powerful technique that allows individuals to learn self-regulation of their brain activity, leading to improvements in cognitive and behavioral performance. However, a significant challenge in NF is the phenomenon of "non-learners" – individuals who fail to achieve the desired changes in their brain activity despite training. This lack of understanding about why some people learn NF well while others don't, particularly in the context of alpha down-regulation (which is associated with increased cortical activation), hinders the optimization and application of NF. This paper aims to shed light on the learning process of alpha down-regulation and, crucially, to identify reliable predictors that can distinguish potential learners from non-learners before extensive training begins.

Methods

The researchers employed a two-day neurofeedback training protocol designed to help participants learn to reduce their alpha brainwave activity. The core of their approach involved:

  • Neurofeedback Training: Participants received real-time visual feedback (a sphere and cube changing size and color) based on their alpha amplitude at the Oz electrode. The training aimed to decrease this alpha amplitude.
  • EEG Recording: Eyes-closed and eyes-open resting EEG data were collected before and after training. During training, EEG was recorded continuously.
  • Alpha Learning Assessment: A "learning index" was calculated for each participant. This index quantified the average change in alpha amplitude from the first block of training to subsequent blocks within a session, averaged across both training days. A negative learning index indicated successful alpha reduction.
  • Predictive Modeling: To identify predictors of learning, a stepwise linear discriminant analysis (LDA) was used. This statistical technique systematically selected the most informative features from the resting EEG data (specifically, amplitudes in various frequency bands from eyes-closed resting baseline) to build a model that could differentiate between learners and non-learners.

Datasets

The study utilized data from 29 healthy volunteers (21 females, 8 males) aged 22-32 years. Resting EEG data (eyes-open and eyes-closed) were collected before and after the two-day neurofeedback training. No new benchmark datasets were introduced; the study used data collected from its participants for both training and evaluation of the predictive model.

Key Insights and Ideas

  • Resting EEG as a Crystal Ball: The most significant insight is that the brain's baseline activity, even before any training begins, holds crucial information about an individual's capacity to learn specific neurofeedback protocols. This challenges the notion that learning is solely a product of the training process itself.
  • Alpha Down-Regulation Learning is Variable: The study confirmed that a substantial portion of individuals (27.59% in this study) struggle to learn alpha down-regulation, aligning with previous reports of "non-learners" in various NF paradigms. This highlights the need for personalized approaches.
  • Specific Resting Frequencies Matter: The predictive power wasn't just about general brain activity but specifically linked to the amplitudes in theta, lower alpha, sigma, and beta 1 frequency bands during eyes-closed resting states. This suggests these brainwave patterns are intrinsically related to the neural mechanisms involved in learning to suppress alpha.
  • Inverse Relationship for Alpha Amplitude and Learning: Interestingly, a moderately positive correlation was found between the learning index and resting alpha activity. This implies that individuals with lower resting alpha might have a better learning index (i.e., learn to down-regulate more effectively), a finding that contrasts with alpha up-regulation studies.

Key Results

  • Predictive Model Accuracy: The stepwise LDA model, using eyes-closed resting EEG amplitudes in theta, lower alpha, sigma, and beta 1, achieved an impressive 86.2% leave-one-out cross-validation accuracy in predicting whether a participant would be a learner or a non-learner.
  • Learner/Non-Learner Split: Out of 29 participants, 21 were classified as learners (negative learning index), and 8 were classified as non-learners (positive learning index).
  • Alpha Amplitude Trends: While there was a general trend for alpha amplitude to decrease over training periods, with a significant main effect of session, individual differences were substantial. Resting alpha amplitudes did not significantly change between baseline measurements.

Limitations and Future Work

  • Limited Sample Size: The study involved 29 participants, which, while sufficient for initial findings, could be expanded for greater generalizability.
  • Short Training Duration: The two-day training protocol might be considered short. Future work could explore whether these predictive relationships hold for longer or more intensive training regimens.
  • Focus on Alpha Down-Regulation: The findings are specific to alpha down-regulation. Future research should investigate if similar predictive patterns exist for other NF protocols (e.g., alpha up-regulation, SMR training) and in different populations (e.g., clinical groups).
  • Sustained Learning: The study focused on within-session learning. Investigating whether these learned skills transfer to situations outside of the training environment (retention and transfer tests) is a crucial next step.
  • Functional Changes: Exploring the impact of this learned alpha down-regulation on actual functional changes in the brain and behavior would further validate the efficacy of NF.

Summary updated

Abstract

Neurofeedback training, which enables the trainee to learn self-control of the EEG activity of interest based on online feedback, has demonstrated benefits on cognitive and behavioral performance. Nevertheless, as a core mechanism of neurofeedback, learning of EEG regulation (i.e., EEG learning) has not been well understood. Moreover, a substantial number of non-learners who fail to achieve successful EEG learning have often been reported. This study investigated the EEG learning in alpha down-regulation neurofeedback, aiming to better understand the alpha learning and to early predict learner/non-learner. Twenty-nine participants received neurofeedback training to down-regulate alpha in two days, while eight of them were identified as non-learners who failed to reduce their alpha within sessions. Through a stepwise linear discriminant analysis, a prediction model was built based on participant's eyes-closed resting EEG activities in broad frequency bands including lower alpha, theta, sigma and beta 1 measured before training, which was validated in predicting learners/non-learners. The findings would assist in the early identification of the individuals who would not likely reduce their alpha during neurofeedback.
BibTeX
@inproceedings{nan2018eyes,
  author = {Wenya Nan and Feng Wan and Qi Tang and Chi Man Wong and Boyu Wang and Alessandro Pasquale De Rosa},
  title = {{Eyes-Closed Resting EEG Predicts the Learning of Alpha Down-Regulation in Neurofeedback Training}},
  booktitle = {Frontiers in Psychology},
  year = {2018},
  doi = {10.3389/fpsyg.2018.01607},
}
Figure