[Eeglablist] EEG-based Emotion Recognition

Tarik S Bel-Bahar tarikbelbahar at gmail.com
Thu Feb 23 11:16:33 PST 2017


Hi Matheus, some brief notes below, hope they help. cheers.


**********************************************************************************

If you don't have an actual psychological measure of emotion, you
might be able to consider the options below. If the participants were
not being emotional either in any way (e.g., eyes-closed resting) then
you can't easily assume there are emotion-related dynamics to begin
with in the sample. However, one can say that there is a continuous
emotional experience, however low on arousal they might be, most
phenomenological experiences have a tinge of valence. There are of
course other issues, such as individual differences on emotional
states and traits in the people in the study. If you had any measure
of their faces or psychophysiology that would help you a lot.

For your purposes, review the EEG-bci-emotion-IEEE literature and
build EEG metrics based on that, use the most predictive spectral
metrics at the most specific bands at the most predictive channel
locations. This is regardless of whether they used stimuli or not.
Preferably use something where they use a continuous signal like yours
and associated that with self-reports of trait or state emotion. You
can also look at the EEG asymmetry literature, which would allow to
supposedly index ongoing emotion via lateralized EEG metrics. See also
alpha burts. These are easy to compute, and assume everyone has
ongoing emotional or motivational biases that influence assymetry. So
as a start focus on alpha, and perhaps theta metrics, and go from
there.

*you may also use existing downloadable databases of EEG + emotion
(e.g., DEAP) to build your own predictive framework.

Here's some google scholar titles you might like below:

Toolbox for Emotional fEAture extraction from Physiological signals (TEAP)

Detection of human emotions using features based on the multiwavelet
transform of EEG signals

Fusion of EEG and Musical Features in Continuous Music-emotion Recognition

Prediction of subjective ratings of emotional pictures by EEG features

Real-time EEG-based emotion monitoring using stable features

Human emotion variation analysis based on EEG signal and POMS scale

Automated estimation of human emotion from EEG using statistical
features and SVM


Improving BCI-based emotion recognition by combining EEG feature
selection and kernel classifiers

Toward automatic detection of brain responses to emotional music
through analysis of EEG effective connectivity

Deap: A database for emotion analysis; using physiological signals

Stress, emotion regulation and cognitive performance: the predictive
contributions of trait and state relative frontal EEG alpha asymmetry

Frontal Theta Activity as an EEG Correlate of Mood-Related Emotional
Processing in Dysphoria

EEG Correlates of Ten Positive Emotions

Using Deep and Convolutional Neural Networks for Accurate Emotion
Classification on DEAP Dataset.

Study on an effective cross-stimulus emotion recognition model using
EEGs based on feature selection and support vector machine

Spontaneous EEG activity and spontaneous emotion regulation

Physiological sensing of emotion  J Healey - The Oxford handbook of
affective computing, 2014 - books.google.com

Application of Entropy-Based Metrics to Identify Emotional Distress
from Electroencephalographic Recordings

On Mon, Feb 20, 2017 at 6:06 PM, Matheus Segalotto
<matheus.segalotto at gmail.com> wrote:
> Dear eeglablist,
>
> I collected raw EEG, using 14 channels at 256Hz in my experiment about
> source code.
> I have a question about emotions and how to classify it.
>
> Is it possible to classify emotions even though my experiment is not
> inducing explicitly any emotions?
> Is there a method that can classify emotions without a training the subject?
>
> I searched many articles and so far all of them have an experiment with
> pictures or music with pre-defined emotions on these inputs.
>
> Thank you for your help.
>
> --
> Atenciosamente,
> Matheus Segalotto
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu


More information about the eeglablist mailing list