| Autor: |
Coutinho E; Department of Music, University of Liverpool, Liverpool, United Kingdom.; Department of Computing, Imperial College London, London, United Kingdom., Gentsch K; Swiss Center for Affective Sciences, University of Geneva, Geneva, Switzerland., van Peer J; Behavioural Science Institute, Radboud University Nijmegen, Nijmegen, Netherlands., Scherer KR; Swiss Center for Affective Sciences, University of Geneva, Geneva, Switzerland., Schuller BW; Department of Computing, Imperial College London, London, United Kingdom.; Chair of Complex & Intelligent Systems, University of Passau, Passau, Germany. |
| Jazyk: |
angličtina |
| Zdroj: |
PloS one [PLoS One] 2018 Jan 02; Vol. 13 (1), pp. e0189367. Date of Electronic Publication: 2018 Jan 02 (Print Publication: 2018). |
| DOI: |
10.1371/journal.pone.0189367 |
| Abstrakt: |
In the present study, we applied Machine Learning (ML) methods to identify psychobiological markers of cognitive processes involved in the process of emotion elicitation as postulated by the Component Process Model (CPM). In particular, we focused on the automatic detection of five appraisal checks-novelty, intrinsic pleasantness, goal conduciveness, control, and power-in electroencephalography (EEG) and facial electromyography (EMG) signals. We also evaluated the effects on classification accuracy of averaging the raw physiological signals over different numbers of trials, and whether the use of minimal sets of EEG channels localized over specific scalp regions of interest are sufficient to discriminate between appraisal checks. We demonstrated the effectiveness of our approach on two data sets obtained from previous studies. Our results show that novelty and power appraisal checks can be consistently detected in EEG signals above chance level (binary tasks). For novelty, the best classification performance in terms of accuracy was achieved using features extracted from the whole scalp, and by averaging across 20 individual trials in the same experimental condition (UAR = 83.5 ± 4.2; N = 25). For power, the best performance was obtained by using the signals from four pre-selected EEG channels averaged across all trials available for each participant (UAR = 70.6 ± 5.3; N = 24). Together, our results indicate that accurate classification can be achieved with a relatively small number of trials and channels, but that averaging across a larger number of individual trials is beneficial for the classification for both appraisal checks. We were not able to detect any evidence of the appraisal checks under study in the EMG data. The proposed methodology is a promising tool for the study of the psychophysiological mechanisms underlying emotional episodes, and their application to the development of computerized tools (e.g., Brain-Computer Interface) for the study of cognitive processes involved in emotions. |
| Databáze: |
MEDLINE |
| Externí odkaz: |
|
|
Nepřihlášeným uživatelům se plný text nezobrazuje |
K zobrazení výsledku je třeba se přihlásit.
|
