This master's thesis sought to investigate the efficacy of neurofeedback in meditation training by comparing an embodied social robot guide (Epi) against a disembodied voice-only guide and a control condition. The thesis aimed to determine if physical embodiment improves the attainment of meditative states compared to standard audio feedback or no feedback at all. By analyzing electroencephalogram (EEG) measures, specifically the alpha/theta ratio and its stability, alongside self-reports, the work evaluated the impact of these different feedback modalities on focused attention meditation.
The technical challenge of this project resided in creating a closed-loop system capable of offering real-time preprocessed data on meditative states. This involved developing experimental software to process live EEG signals, specifically calculating the alpha/theta ratio and alpha/theta variance semi-instantaneously, and translating these biological metrics into dynamic feedback cues. A key component of this engineering was designing and sequencing distinct humanoid movement patterns for the robot (Epi) to create an embodied effect that differentiated it from the audio-only condition. The system required precise synchronization on multiple levels to ensure the readings accurately reflected the user's current neural state during the meditation sessions. The complete repository can be found below.
View Code on GitHubYou can view the complete academic work "Brainwaves, Breath and Robots: Rethinking Feedback in Meditation Training" below.
Open Thesis (PDF)