EEG Processing

You are here

EEG Processing

This research activity focuses on brain dynamics investigation during affective elicitation involving different sensory elicitation.

EEG is a signal with outstanding characteristics in terms of time resolution that allows studying brain dynamics until 50 Hz (and even more in same cases). Our EEG analysis is usually performed following a pipeline composed by pre-processing, power spectral density estimation, functional connectivity, statistical comparisons and machine learning methods. This processing chain aims at extracting features related to the brain activity and not corrupted by noise, which are used in order to assess central nervous system response to affective stimulation.

Many studies have been conducted in the field of affective computing, involving haptic, visual and olfactory senses. 


In this framework, we investigate how brain dynamics is modified by quantifiable, affective haptic stimuli. Prior arts have shown that a significant role in hedonic touch perception is played by the orbitofrontal cortex, the contralateral primary and bilateral secondary somatosensory areas, the contralateral middle and posterior insula cortex and the contralateral posterior parietal cortex.

We studied six combinations of force and velocity values in terms of loco-regional EEG power spectra in order to evaluate the subjects’ emotion elicited by the haptic stimulation. Results revealed a suppression of the oscillations over the contralateral somatosensory cortex, during caresses performed with the lowest force and the highest velocity. This occurred in all of the frequency bands considered alpha, beta and gamma. Concerning caresses administered at high force, there was a significant decrease of EEG oscillatory activity focused on mid-frontal electrodes, in all of the considered frequency bands, when the velocity of the caresses was the lowest. Significant sparse decrease of EEG power spectra, in all of the considered frequency bands, occurred at higher strength and velocity of the caress. 


Figure. Topographic maps of p-values obtained through wilcoxon nonparametric tests on PSD, considering EEG dynamics in response to each combination of force and velocity of caressing.


Another affective elicitation that has been studied is the hedonic olfactory stimulation. The emotional properties of odorants are well demonstrated by neuroanatomical studies. Specifically, a strict connection between olfactory efferent and neural substrates of emotional processing were reported. Moreover, several previous studies have also confirmed that human EEG activity changes during exposure to odorants, although with contradictory results. Some of the reported differences may be explained by considering the poor reliability of old EEG systems in terms of weak signals, confused specificity in electrode placement and no adequate experimental control. Starting from the above-mentioned ambiguity of results in EEG response we aims at implementing a robust system with respect to inter-subject variability. Our study rely on a pattern recognition approach based on Linear Discriminant Analysis to automatically classify the human EEG response as belonging to pleasant or unpleasant olfactory stimulation. 


Figure. Confusion matrix of pleasant smell vs. unpleasant smell.



Concerning central perception of visual affective stimuli, pre-frontal and the visual cortex play a relevant role in processing emotional reactions to different classes of stimuli. When this neural circuitry is impaired, then several pathological effects arise, such as anhedonia, i.e., the loss of pleasure or interest in previously rewarding stimuli, which is a core feature of major depression and other serious mood disorders.

An emotional visual stimulus can be conveyed through different modalities. One of the most widely used is displaying pictures taken from the International affective picture system (IAPS) database. This database includes hundreds of pictures, each of which reports a standardized value of arousal and valence. IAPS pictures with different levels of arousal and valence are often used in EEG studies. This modality of elicitation stimulates several cortical areas also allowing the prefrontal cortex to modulate the generation of cognitive perceptions. Previous studies have shown the relationship between induced emotions and brain power spectral density (PSD) in EEG activity over the frontal cortex. Of note, PSD as well as connectivity in the gamma band (over 30 Hz) have been found to be sensitive to valence changes. More specifically, in the literature the majority of studies have shown an increase of gamma rhythms during pleasant/unpleasant stimuli as well as an increase of the gamma band connectivity.

Our research group has investigated the neural correlates of pleasant/unpleasant as a function of arousing emotions through a spectral and functional connectivity analysis of EEG signals. Data were gathered from healthy volunteers while watching pictures gathered from the IAPS database. Considering 4 arousing levels, each of which including two valence levels (pleasant and unpleasant), we investigated EEG power spectra and functional connectivity. For example, focusing on gamma oscillations (over 32 Hz), because of their known sensitivity to valence changes, results revealed no significant changes between pleasant/unpleasant elicitation when lower and higher arousing level occurred. Conversely, valence changes in the intermediate arousing sessions were associated with changes in the prefrontal and occipital regions. Additionally, different arousing levels of pleasant elicitations affected short-range connectivity over the right hemisphere, whereas different arousing levels of unpleasant elicitations affected medium-range connectivity over the left hemisphere.


Figure. Statistical comparison (p-value maps) between unpleasant (i.e., neg) and pleasant (i.e., pos) sub-sessions of each arousal (i.e., ar) session, performed on EEG-PSD estimates calculated in the gamma band.


Computational Physiology & Biomedical Instruments Group, University of Pisa


Feel-ING is a new start-up specialized in developing and implementing innovative Information and Communication Technology solutions for the optimization of the psychophysical performance and the promotion of wellbeing and mental health