Effects of gratitude meditation on neural network functional connectivity and brain-heart coupling

A sense of gratitude is a powerful and positive experience that can promote a happier life, whereas resentment is associated with life dissatisfaction. To explore the effects of gratitude and resentment on mental well-being, we acquired functional magnetic resonance imaging and heart rate (HR) data before, during, and after the gratitude and resentment interventions. Functional connectivity (FC) analysis was conducted to identify the modulatory effects of gratitude on the default mode, emotion, and reward-motivation networks. The average HR was significantly lower during the gratitude intervention than during the resentment intervention. Temporostriatal FC showed a positive correlation with HR during the gratitude intervention, but not during the resentment intervention. Temporostriatal resting-state FC was significantly decreased after the gratitude intervention compared to the resentment intervention. After the gratitude intervention, resting-state FC of the amygdala with the right dorsomedial prefrontal cortex and left dorsal anterior cingulate cortex were positively correlated with anxiety scale and depression scale, respectively. Taken together, our findings shed light on the effect of gratitude meditation on an individual’s mental well-being, and indicate that it may be a means of improving both emotion regulation and self-motivation by modulating resting-state FC in emotion and motivation-related brain regions.


Supplementary Material S2: Time intervals between all successive fMRI scans
Supplementary Figure S1 shows the experimental procedure with time stamps to illustrate the time intervals between consecutive fMRI scans, which were present because of differences in the duration of audio instructions before each fMRI scanning. We delivered a simple instruction before starting the intervention fMRI scans and resting-state scans via the built-in microphone. The time interval between the start time of the first intervention fMRI scan (t2) and end time of the baseline fMRI scan (t1) can be computed by a subtraction between two time stamps (t2-t1). Similarly, we computed the time intervals between all successive fMRI scans.
Supplementary Figure S1. Experimental procedure. The order of experiments (set I and II) were counter balanced across participants. Abbreviations: RS-fMRI, resting-state functional magnetic resonance imaging.
The intermissions before fMRI scans were not significantly different between the first intervention (t2-t1) and the second intervention (t6-t5) and between follow-up resting-states (t4-t3 vs. t8-t7), but significantly different between the intervention and follow-up resting-state (t2-t1 vs. t4-t3 or t6-t5 vs. t8-t7) (Supplementary Figure S2). Meanwhile, the average time intervals between during-and after-intervention fMRI acquisition were 33.0 ± 10.1 and 29.8 ± 10.9 seconds for the gratitude and resentment interventions, respectively, and those intervals were not significantly different between the two interventions (t31 = 1.71, p = 0.1). Figure S2. The average time intervals for all successive fMRI scans across the participants. Paired sample t-test revealed significant differences among time intervals based on Bonferroni-corrected P < 0.05. We indicated the significant differences as asterisks (*). Please refer to Supplementary Figure S1 for the information about the time stamp, ti. Blue bars indicate the time intervals between the resting-state fMRI scan and intervention fMRI scans, and gray bars indicate time intervals between the intervention fMRI and follow-up resting state fMRI scans.

Supplementary
Supplementary Table S1 shows the summary of the time intervals between all successive fMRI scans. Two-sample t-test revealed that there were no significant differences in the time intervals for the successive fMRI scans between set I and II. Heart rate data at the baseline, during the interventions, and after the interventions were presented in Supplementary Figure S3. Although we acquired heart rate data for the entire experimental period, we did not compare these heart rate values with heart rate acquired at the baseline because of the following two reasons. First, baseline heart rates were unstable because they were measured immediately after the body-posture change of the participant from stand up to supine 1 . Second, it may take some time for an individual to stabilize levels of anxiety about closed space (such as MRI scanner) and levels of heart rate until the first (baseline) scanning because heart rate is sensitive to anxiety or emotional stress 2 . Considering these viewpoints, it was a limitation of the current study that participants performed the baseline MRI scanning before they were sufficiently stabilized in the supine position inside the MRI scanner. Therefore, including heart rate at the baseline in the statistical analysis can make it difficult to interpret the results.

Supplementary
Supplementary Figure S3. Mean and standard error of heart rate at the baseline, during the interventions, and after the interventions.
Meanwhile, we computed heart rate during gratitude session as the 1st intervention and during resentment session as the 1st intervention (the second sessions of Set I and Set II), which were is 68.3 ± 6.9 and 73.4 ± 8.6. Although these two mean values were not statistically different between two sessions (t = -1.76, df=27, p = 0.089), the mean heart rate during gratitude after the baseline is quite lower than mean heart rate during resentment after the baseline (mean difference in heart rate = -5.1 bpm, 95% confidence interval = [-11.0, 0.5] bpm).

Supplementary Material S3: Summary for six seed regions of interests and functional connectivity.
Six seed regions of interests for investigating the default mode, emotion-regulation, and reward-motivation networks are summarized in Supplementary Table S2. In this section, we provide the detailed summary statistics that were presented in the form of visual illustration in the main manuscript. The supplementary results include temporal synchronization between dynamic functional connectivity and heart rate during the interventions (Supplementary Table S3), changes in seed-based functional connectivity during gratitude and resentment interventions relative to the baseline (Supplementary Tables S4 and S5), and comparisons of resting-state functional connectivity among the baseline, after-gratitude, and after-resentment (Supplementary Table S6).

Supplementary
In addition, Supplementary Figure S4 represent the statistical parametric mapping of significant seed-based resting-state functional connectivity for each seed region. Zmax, maximum z-value within the cluster; RM-ANOVA, repeated-measure analysis of variance; Nvox, number of voxel within a cluster; SD, standard deviation. † A significance level of post-hoc analysis among three sessions was corrected for multiple-comparisons using Bonferroni method.

Supplementary Material S4: Inter-network functional connectivity
In order to investigate inter-network functional connectivity, we produced spatial independent component maps and the corresponding time-courses using independent component analysis and spatiotemporal dual-regression. Ten independent component maps (Supplementary Figure S5) were identified using the template-matching method and were matched with the previously reported resting-state networks 654433 (Smith et al., 2009) (Smith et al., 2009). 6 Among them, given our hypothesis, five networks colored in cyan such as the default mode, auditory, salience, and bilateral frontoparietal networks, were selected for the effects of gratitude and resentment interventions on inter-network functional connectivity.
In this section, inter-network functional connectivity was evaluated using the time-courses obtained from the dualregression analysis.
Supplementary Figure S5. Ten independent component maps for the resting-state networks.
Subsequently, temporal synchronization between dynamic inter-network functional connectivity and heart rate were counted during the gratitude and resentment interventions (Supplementary Table S7). Inter-network functional connectivity was compared between the interventions and baseline (Supplementary Table S8), between the gratitude and resentment interventions (Supplementary Table S9), and between resting-states at the baseline, after the gratitude intervention, and after the resentment intervention (Supplementary Table S10).  (SD) are provided together with the statistics obtained from paired-sample t-test. The statistical significances were controlled for multiple comparison using false discovery rate (FDR) corrected P-value (PFDR). Abbreviation: df, degree of freedom.