Sparse EEG Neurofeedback for Training Simulated Prosthetic Vision in Virtual Reality

Retinal prostheses can restore limited vision, but interpreting phosphene-based images is cognitively demanding and often requires months of training. This project develops an EEG-guided neurofeedback approach to support learning in a virtual reality (VR) object-localization task rendered with simulated prosthetic vision (SPV). Using a Galea dry-EEG headset, I compute a fronto-parietal neural index of engagement/workload and provide brief, sparse visual feedback after each trial. Participants complete baseline, training, and post-training blocks and are randomized to contingent neurofeedback or sham feedback. Primary outcomes include localization accuracy, completion time, and perceived workload (NASA-TLX). This work tests whether sparse EEG-guided feedback can improve SPV task performance and reduce cognitive workload, informing rehabilitation strategies for future vision-restoration devices.