Emotional States Estimation from Multichannel EEG Maps

Abstract

The notion of emotional empathy, that is the estimation of responses from electrophysiological signals has recently gained attention among the designers of brain computer/machine interfaces (BCI/BMI). This is largely due to the relative simplicity and convenience of electroencephalograph (EEG) recordings. Whereas this problem is not entirely new, engineering approaches connecting the emotional states of humans and the observed EEG recordings are still in their infancy. In this paper we present an initial set of results on the estimation of basic emotional responses generated while watching short videos with dynamic emotional contents. A novel multichannel EEG analysis approach is proposed in order to discover representative components of the emotional responses. This is achieved based on distinct spatial patterns exhibited within the EEG data recorded over frontal lobes, and averaged over a number of trials and subjects. Simulation results support the proposed approach, and confirm the initial hypothesis.