Cross-validated Adaboost Classification of Emotion Regulation Strategies Identified by Spectral Coherence in Resting-State.

Creative Commons License

Aydın S.

Neuroinformatics, vol.20, no.3, pp.627-639, 2022 (SCI-Expanded) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 20 Issue: 3
  • Publication Date: 2022
  • Doi Number: 10.1007/s12021-021-09542-7
  • Journal Name: Neuroinformatics
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED)
  • Page Numbers: pp.627-639
  • Hacettepe University Affiliated: Yes


In the present study, quantitative relations between Cognitive Emotion Regulation strategies (CERs) and EEG synchronization levels have been investigated for the first time. For this purpose, spectral coherence (COH), phase locking value and mutual information have been applied to short segments of 62-channel resting state eyes-opened EEG data collected from healthy adults who use contrasting emotion regulation strategies (frequently and rarely use of rumination&distraction, frequently and rarely use of suppression&reappraisal). In tests, the individuals are grouped depending on their self-responses to both emotion regulation questionnaire (ERQ) and cognitive ERQ. Experimental data are downloaded from publicly available data-base, LEMON. Regarding EEG electrode pairs that placed on right and left cortical regions, inter-hemispheric dependency measures are computed for non-overlapped short segments of 2 sec at 2 min duration trials. In addition to full-band EEG analysis, dependency metrics are also obtained for both alpha and beta sub-bands. The contrasting groups are discriminated from each other with respect to the corresponding features using cross-validated adaboost classifiers. High classification accuracies (CA) of 99.44% and 98.33% have been obtained through instant classification driven by full-band COH estimations. Considering regional features that provide the high CA, CERs are found to be highly relevant with associative memory functions and cognition. The new findings may indicate the close relation between neuroplasticity and cognitive skills.

CONCLUSION AND DISCUSSION:  The functional connectivity approaches are as follow: 1. coherence based on power spectral density estimation, 2. PLV based on phase difference between neuro-electrical activities generated by neuronal populations at right and left hemispheres, 3. MI based on statistical alterations throughout time instants in those neuro-electrical activities, and 4. WCOH based on wavelet transformed coefficients of those neuro-electrical activities.

Two groups include the individuals who use frequently and rarely use of both rumination and distraction (GP1 and GP2), the other groups include the individuals who use frequently and rarely use of both expressive suppression and reappraisal (GP3 and GP4). Thus, the groups (GP1 vs GP2, GP3 vs GP4) were classified by examining Adaboost classifiers with respect to non-averaged inter-hemispheric connectivity estimations from short-segments of 2 sec as well as ensemble averaged of the estimations over a trial of 2 min for 27 electrode pairs excluding middle-line sites and reference points as shown in Fig. 3. The resulting Classification Accuracies reveal that full-band coherence estimations provide the most meaningful and discriminative indicators in quantifying inter-hemispheric neuro-electrical synchronicity in resting state. Besides, ensemble averaging connectivity estimations over a trial of 2 min did not provide clear difference between contrasting groups. Due to nature of ongoing EEG series originated from spontaneous and time-varying brain functions, it is superior to determine a collection of coherence estimations as the large number of features in association with consecutive small-windows of 2 sec in a recording interval of 2 min.

GP1 and GP2 were successfully classified with high CA of 99.44%, GP3 and GP4 were successfully classified with high CA of 98.33% by using Adaboost classifiers driven by non-averaged full-band (0.5−40.5 Hz0.5−40.5 Hz) EEG coherence estimations. Regarding the statistical test results and corresponding EEG recording sites, more cortical regions are affected by ruminative thoughts by means of inter-hemispheric EEG coherence in comparison to optimistic thoughts. However, the findings reveal that particular BAs of 2, 5, 6, 8, 9, 18, 20, 37, 39, 40, 41, 42, 47 were commonly found to be sensitive to cognitive emotion management strategies in healthy adults. The main functions of these BAs are as follow:

2:sensory perception, motor learning (primary somatosensory cortex)

5:working memory, language processing, visuo-motor attention, pain perception, tactile localization, motor execution, bimanual manipulation, (somatosensory association cortex)

6:memory-language-motor functions, planning and sensory guidance of both muscle movement and complex motor movements (premotor cortex and supplementary motor cortex)

8:spatial memory, memory-guided saccades (prefrontal cortex)

9:working memory, planning, organization, and regulation of motor functions, sustaining attention (dorsolateral prefrontal cortex)

18:visual depth perception through receiving input from primary visual cortex, (secondary visual cortex).

20:visual fixation, identify intention (Inferior temporal gyrus)

37:processing of color information, recognition of face/body/word/numbers through visual perception, (occipito-temporal cortex)

39:speech fluency, language comprehension, (Angular gyrus, part of Wernicke’s area)

40:speech fluency, language comprehension, (Supramarginal gyrus, part of Wernicke’s area)

41/2:auditory working memory, Visual speech perception, (primary and secondary auditory cortex)

47:working and episodic memory, management of reward and conflict, spoken language, language semantics, identifying semantics, processing of linguistic information, (orbital part of inferior frontal gyrus).


In conclusion, overall results reveal that CERs are highly correlated with main associative brain functions such as working memory, visual/sensory perception and cognition. As well, expressive suppression cause decrease in full-band EEG coherence at mostly fronto-central regions. The reason of this findings may originated from lack of experience of positive emotions, since frequently use of SE increases the felt intensity of negative emotions, while reduces the felt of positive emotions, such as happiness as discussed in reference (Gross & Jazaieri, 2014)