Abstrakt

Cognitive Neuroscience 2020: Machine learning and validity of binaural beat protocols: Trainability and interpretability

Muhammad Abul Hasan

Statement of the Problem: Binaural Beat (BB) is a form of sound wave therapy in which both ears received sounds of slightly different frequencies, yet auditory cortex perceived as a single signal [1]. BB therapy is provided in frequency ranges corresponding to electroencephalogram (EEG) bands (theta, alpha, beta, and gamma). Studies have shown benefits of different types of BB therapy for treatment of anxiety, depression, mood and memory [1-4]. Studies used different cognitive and EEG tests for studying psychological and neurological changes following BB stimulation [3, 4]. However, the unknown mechanism of BB therapy is a challenge for end users to implement BB in clinics [5]. The unknown mechanism might be due to lack in validation processes. In this study, machine learning and regression analysis was applied to study the neurological and psychological changes. Methodology & Theoretical Orientation: 21 participants in this study were divided into three groups for receiving different BB stimulation (alpha; 400-410 Hz low beta; 400-414 Hz, and high beta; 400-429 Hz stimulation). The stimulation was provided in three consecutive 5 minute sessions (15 min total length). Digit span (4 to 8 words) task was performed on each subject to study effects of BB on memory. EEG with 128 Hz sample rate in closed-eyes state was recorded with Emotive epoch 14 channels device for studying neurological changes.