EEG Source Localization by Memory Network Analysis of Subjects Engaged in Perceiving Emotions from Facial Expressions Reshma Kar \ Amit Konar l , Aruna Chakrabor, Basabdatta Sen Bhattacharya 4 , Atulya K. Nagar 3 l Department of Electronics & Telecommunication Engineering, Jadavpur University, Kolkata, India 2 Department of Computer Science & Engineering, St. Thomas' College of Engineering & Technology, Kolkata, India 3 Department of Mathematics & Computer Science, Liverpool Hope University, Liverpool, UK 4 School of Engineering, University of Lincoln, United Kingdom rkar317@gmail.com, konaramit@yahoo.co.in, arunachakraborty.stcet@gmail.com, bbhattacharya@lincoln.ac.uk, nagara@hope.ac.uk Abstract-The memory network is a result of current dipoles created in the brain. Localizing the source of these current fows is known as source localization, and it could potentially reveal which parts of the brain are actually responsible for a particular brain activity. It would also increase the spatial resolution of an EEG recording by identifying the true source of multiple correlated readings. In our experiments, we employed memory networks to classify perception of emotional instances conveyed in facial expressions as well as to localize sources. These networks were created by selectively evaluating EEG channel signals pairwise for Granger causality. Channel selection was based on clustering of EEG features by Self Organizing Feature Map (SOFM). Principal Component Analysis (PCA) was employed for dimension reduction and noise elimination of EEG features. Finally a new metric based on Fischer's discriminant was used to compare different source localization techniques, where real source locations are unknown. The perception of the stimuli was classifed as belonging to one the following classes i) Happy ii) Sad iii) Fear iv) Relaxed. The created memory networks could classify perception of emotional content in 90.64% of cases. Comparison by the proposed Fischer Discriminant based metric revealed that the proposed network identifcation technique performs better at source localization as compared to independent component based source localization. Keywords-ource localisation; emotion perception; memor network; Granger causalit; Se/Organising Feature Maps; I. INTRODUCTION Emotions play a vital role in our daily activities. Experiencing emotion is a complex process involving multiple lobes of the human brain [15]. The exact mechanism of co ordination among brain modules until this date remains a mystery in cognitive neuroscience. A little thinking reveals that we recognize emotions revealed in a portrait fom our encounter with similar instances of emotion. Such experience usually is encoded in our memory network involving a number of neurons located over our brain network [23]. Understanding and 978-1-4799-1959-8/15/$31.00 @2015 IEEE recogllzmg emotion expressed in facial expression of human subjects is a frst step to express one's emotion. The process of perceiving emotion expressed in portraits of subjects thus calls for recalling memory to match the perceived instances with the stored reference emotional instances. Analysis of brain signals is instrumental in healthcare, rehabilitation and human computer interactions. Naturally, proximity to the source of these signals ensures precision in signal measurements. Unfortunately, collecting brain signals in vivo is not only expensive but bears a risk as well to the subject's health. EEG acquisition by placing electrodes on the scalp began in the 1930s [1]. EEG simplifes brain signal acquisition at low spatial resolution, because of the bluring effect caused by volume conductivity. The problem of volume conductivity arises due to the anisotropic property of the tissues surrounding the brain. This results in dispersion of electric potential across the scalp. This in tum is responsible for reduced spatial resolution. This is the one of the chief reasons why researchers are interested in locating the source of electric potential being measured across different electrodes. In fact, source localization can increase spatial resolution fom centimeter to millimeter scale [2]. Several clinical evidences on sensory working memory have led to the empirically and theoretically well-accepted hypothesis that memory is distibuted in almost all parts of brain including cortical stuctures [3], [4]. These [mdings [5], [6], [3] localize the role of the prefontal, temporal, parietal, fontal, motor and occipital areas in sensory and motor memory. We identifed connectivity among the areas associated with working memory while perceiving emotional instances fom facial expressions. Apart fom source localization, the proposed work will help us in establishing how memory facilitates co-ordination of sensory perception of emotional content. This will be helpfl in the