Comparison of Amplitude-Time rewards and Sham-Control in EEG-neurofeedback.

The search for non-invasive and interactive therapeutic approaches has increased in the last decades, and the Neurofeedback (NF) has been explored in the context of brain disorders or enhancement of cognitive performance, both in research and clinical settings. The NF technique aims to train self-regulation of brain activity in real-time (RT), and it is primarily utilized in research with functional magnetic resonance imaging (fMRI) or electroencephalography (EEG). The training undergone with NF is based on associative learning, which offers a non-pharmacological and non-invasive treatment. Different levels of reinforcement can be used in the approach, and even though the underlying NF mechanisms remain to be fully elucidated, studies have shown effectiveness in improving the relief of symptoms. Despite promising results, several limitations exist, including emotional and cognitive confounds, and that about one-third of the population is not able to learn self-regulation of brain activity. The indexes utilized for learning outcome are not fully defined nor standardized and focus on the amplitude of the brain activity of interest, in detriment of time-bound information. Least but not last, only recently have sham and control groups been included regularly in study protocols. This project aims at studying the efficacy of an EEG-based NF-training protocol that includes reward for both amplitude (continuous feedback) and duration (discrete feedback) of the brain activity of interest. The protocol is embedded in a game-based learning approach, and the NF training group is compared to a control-sham group, simultaneously, attention, motivation, memory, and mood are assessed. The brain activity of interest was defined as the power of the upper alpha (UA) band, with the aim to increase its amplitude and reward when it’s done for longer periods. This choice binds to the existing know-how at the laboratory where the project was developed. The final goal would be to have a protocol design with proven efficacy that could be adapted to clinical settings in migraine patients. Twenty-two female participants were enrolled in the study, meeting inclusion criteria of no known neurological disorders and no use of psychotropic medication, and were divided in two groups, the real NFT and the control group. The protocol consisted in 4 training sessions to be performed in consecutive days at approximately the same time of the day, each session lasted about 2 hours and comprehend questionnaires and cognitive tests, a pre-baseline recording for calibration, the real training, and the post-training baseline calculation. Subsequently, the neurofeedback signal was calculated from channel Cz, and the EEG data were pre-processed using Matlab toolbox EEGLAB, and then divided into 30 epochs. Time-frequency (TF) decomposition was performed using wavelet transforms and timing information was tagged, to see when each participant achieved the goal of maintaining the NF signal above a predefined threshold. The information was subsequently analysed for the identification of training effects, both within and between sessions, for each of type of reinforcement feedback, and for each group. Finally, we aim at classifying between groups (NF and control) and identifying the features that better disentangle NF learners from non-learners.