EEG Power Analysis of Imagination, Perception, and REM Sleep: A Step Toward Understanding Dream Generation

The brain generates vivid sensory experiences both through externally driven perception and in the absence of external input, via imagery during wakefulness and dreaming during sleep. Understanding the neural mechanisms behind these forms of conscious experience remains a major challenge. This thesis compares the spectral characteristics of brain oscillations during internally generated imagery, externally driven perception, and REM sleep, a stage characterized by vivid dreams, using high density EEG (256-channel Geodesic Sensor Net). Twenty-one healthy participants completed a wakefulness experiment involving resting state, mental imagery (free and prompted), mental arithmetic, and video watching, and underwent overnight sleep recordings from which REM sleep segments were analyzed. We developed a comprehensive preprocessing pipeline, including bad channel detection, artifact removal via ICA, and quality controls. Spectral power was analyzed across six frequency bands and three scalp regions (frontal, centro-parietal, occipital) using cluster-based permutation test. Imagery showed higher alpha and beta power relative to video watching, particularly over posterior regions, while video watching exhibited stronger delta, posterior theta, and gamma activity. Wake-REM comparisons showed higher delta and theta power in REM sleep, while wakefulness exhibited higher alpha, sigma, beta, and gamma activity. Regional variations emerged: theta elevation in REM sleep was widespread relative to eyes-closed wakefulness but localized to frontocentral regions relative to eyes-open conditions. Eye state further modulated spectral dissimilarity with REM sleep, with eyes-closed conditions appearing more dissimilar. This work provides initial spectral characterization of sensory experiences but is limited to scalp-level power analysis. Future work using source reconstruction and directional connectivity analyses are essential to better address questions about dream generation.