Skip to main content

Selective EEG analysis for emotion recognition using multidimensional directed information criteria


Brain waves captured by multiple electrodes during an electroencephalogram (EEG) recording, are derived as time series signals that represent the brain activity of various scalp sites. One important feature of these multiple time series is the information flow from one to another that under specific processing can reveal useful brain functionality.

Materials and methods

In this work, the Multidimensional Directed Information (MDI) [1] concept is adopted in order to examine the flow of information between EEG recordings from three different scalp sites with the objective to bring out and define the interconnections of different scalp sites of frontal and prefrontal cortex during an emotionally charged situation. In the line of these objectives, EEG signals were recorded from 16 healthy right-handed subjects during a specifically designed emotion elicitation experiment. Two monopole and one dipole EEG channels were placed at Fp1, Fp2 and F3/F4 positions respectively, according to the international 10/20 system [2].


The EEG signals were analyzed using MDI as a parameter that could identify the EEG sections that contribute the most to the emotion recognition. In this way, an efficient emotion categorization could be achieved by keeping the effective parts of the EEG signal and further analyze it feeding a classifier towards the development of a robust and effective emotion recognition system from EEG recordings.


The encouraging preliminary results justify the feasibility of the proposed approach, stressing the importance of the targeted selection of the information source within the EEG recordings before any further categorization analysis.


  1. 1.

    Sakata O, Shiina T, Saito Y: Multidimensional directed information and its application. Electronics and Communication in Japan. 2002, 85: 45-55. 10.1002/ecjc.1086.

    Article  Google Scholar 

  2. 2.

    Jasper H: The ten-twenty electrode system of the International Federation 1958. Electroencephalogr Clin Neurophysiol. 10: 367-380.

Download references


The authors would like to thank all the 16 subjects participated in the experiment for their patience during the tedious EEG recording phase.

Author information



Corresponding author

Correspondence to Panagiotis Petrantonakis.

Rights and permissions

Open Access This article is published under license to BioMed Central Ltd. This is an Open Access article is distributed under the terms of the Creative Commons Attribution 2.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and Permissions

About this article

Cite this article

Petrantonakis, P., Kaltsa, V. & Hadjileontiadis, L. Selective EEG analysis for emotion recognition using multidimensional directed information criteria. Ann Gen Psychiatry 9, S149 (2010).

Download citation


  • Emotion Recognition
  • Series Signal
  • Emotion Categorization
  • Multiple Electrode
  • Multiple Time Series