Julie Onton, PhD, Swartz Center, UCSD 

Emotions are a fundamental part of being human, yet recording brain activity during genuine emotional states is difficult to achieve in a laboratory setting. Many studies have attempted to recreate natural emotional reactions by showing subjects pictures of intensely emotional or disturbing facial expressions or scenes. While such pictures can trigger a cascade of subjective and neural events leading to sustained emotional experiences, the responses of viewers depend in large part on the attitudes they bring to the viewing. As well, such methods ignore the broad natural range of emotion and feeling states that pervade our everyday experience. In particular, a wide variety of positive feelings--contentment, happiness, love, compassion, awe, etc.--experienced in milder forms during daily life are not  normally addressed in such studies. This study attempted to discover the brain dynamic correlates of imagined and embodied emotional feeling. Subjects were asked, via a voice recording, to recall and/or imagine a series of scenarios in which they had felt or would feel a series of suggested emotions, in each case allowing the imagery and somatic feeling sensations to become as vivid as possible. A series of fifteen suggested positive and negative emotions were separated by brief relaxation periods. During these sessions, we recorded 256 channels of EEG data from the scalp, neck and face. After decomposing the EEG data using Independent Component Analysis (ICA) into maximally independent time courses and associated spatial maps, we used a novel approach to identify independent power spectral modes active during the experiment. From this analysis, 8 major spectral patterns were identified: 3 classes of alpha (8-12 Hz), 2 classes of beta (15-30 Hz), and 3 classes of gamma (>30 Hz) modulations. Different emotional experiences were differentially associated with these major spectral modulations and, while results differed considerably across subjects, some consistent associations between brain patterns and emotion were identified. The results suggest that mood/emotion states bias EEG dynamics towards identifiable patterns of activity that could possibly be used in the future for mood, response and/or therapeutic monitoring.