The correlation between Subjective Well-Being and EEG Frontal Asymmetry in different frequency bands

1629 

Abstract Submission 

Betty Wutzl¹, Kenji Leibnitz², Masayuki Murata³

¹Osaka University, Suita,Japan, ²National Institute of Information and Communication Technology, Suita, Japan, ³Osaka University, Suita, Japan

Betty Wutzl  
Osaka University
Suita,Japan

Kenji Leibnitz  
National Institute of Information and Communication Technology
Suita, Japan

Masayuki Murata  
Osaka University
Suita, Japan

A correlation between frontal alpha asymmetry, measured via electroencephalography (EEG), and Subjective-Well Being (SWB) has been reported in several papers, e.g., (Urry et al. 2004; Xu et al. 2018). In our previous work (Wutzl et al. 2023), we showed that such a correlation even holds when SWB is not changed by long-term psychological or psychiatric interventions, but also when it is changed on short time scales (60 or 30 seconds). We focused on the asymmetry of the most frontal EEG sensors (AF3 and AF4) and analyzed the influence of the specific time intervals on the calculation of frontal alpha asymmetry. Then, we focused on the asymmetry between different frontal sensors, as these sensor locations were reported to influence the asymmetry scores (Metzen et al. 2022). Here, we expand this research to include the asymmetries between all frontal sensors, as well as different EEG frequency bands.

We performed this experiment in 2022. In order to measure SWB on small time intervals, we changed the experimental room's temperature and humidity, and we recorded EEG with an Emotiv EPOC X headset (EMOTIV, San Francisco, USA) for up to nine minutes for six sets of different temperature-humidity settings. During each EEG recording, the participants were asked to orally report their SWB every 30 s on a scale from 1 (worst) to 10 (best). The EEG dataset for each subject was preprocessed following HAPPE (Gabard-Durnam et al. 2018), using EEGLAB (Delorme and Makeig 2004) and MARA (Winkler, Haufe, and Tangermann 2011; Winkler et al. 2014). Then, the data was filtered into one of the frequency bands: delta (0.5–3 Hz), theta (4–7 Hz), alpha (8–13 Hz), beta (14-30 Hz), or gamma (31-100 Hz). We also used the unfiltered signal as "non" (0.5-100 Hz). We calculated the power spectrum of two channels ch_1 and ch_2 and determined their asymmetry Asym by subtracting the natural logarithm of the power densities of ch_1and ch_2. Asym was then combined with the reported SWB value as tuple (Asym_band (ch_1,ch_2), SWB) for each subject and band. Participants tend to report mid-ranged SWB values SWB (6–8) more frequently than very low or very high SWB values (1–3, 10). Hence, we balanced the data set for each participant using SMOTE (Chawla et al. 2002) to have an equal number of samples per SWB, and we performed a linear regression with Asym as the independent and SWB as the dependent variable. We did this for each subject and then used a one-sided t-test to determine the statistical significance that the mean of the slopes of the linear regression from each subject is greater than zero. Figure 1 shows a graphical representation of the workflow and EEG sensor layout.

We acquired EEG and SWB data from 30 students (28 right-handed, 2 left-handed, 16 males, 14 females, ages 22.3 ± 4.2 years). Results with p-values of less than 0.01 are shown in Table 1. As expected from reports in the literature and our previous results, the alpha frequency band shows statistically significant results. However, filtering into the delta or theta bands, or not filtering at all (non), also yields a positive linear correlation between frontal sensors from contralateral brain hemispheres and SWB.

In our previous work, we focused on the alpha frequency band and the relationship between FAA and short-term SWB changes. Here, we present that also other frequency bands, i.e., delta or theta, or not filtering at all into a specific frequency band show similar results. Thus, we conclude that alpha is not the only EEG frequency band that should be investigated when focusing on short-term SWB changes.

EEG/MEG Modeling and Analysis ¹

Task-Independent and Resting-State Analysis ²

Data analysis

Electroencephaolography (EEG)

Emotions

Modeling

Other - Subjective Well Being, Brain Asymmetry