Introduction

The presence of others can have a positive effect, known as social buffering, a phenomenon that leads to better recovery from aversive experiences1. Social buffering is the phenomenon whereby the presence of others leads to a subjective perception of reduced task difficulty and pain intensity. Physiological changes associated with social buffering include lower heart rate and cortisol levels, reduced activity in the anterior cingulate cortex (ACC), and activation of reward circuits such as the amygdala2,3,4,5. It is important to note that these assessments are objective and supported by empirical evidence. Similarly, being in the company of familiar people increases feelings of safety compared to being alone. This increased sense of safety is associated with increased parasympathetic activity6. We have recently reported that social buffering effects occur in occupational therapy settings and that this effect is associated with insular cortex activity7. However, in this report, the analysis of autonomic activity included only the time immediately after the start and just before the end of the task. Therefore, changes throughout the task were unclear.

On the other hand, physiological synchrony (PS) can occur between individuals when they are in close proximity. This phenomenon has been observed in a variety of ways, including heart rate8, the electrocardiogram (ECG)9 and others10,11. This dyadic PS is associated with togetherness12 and cohesion11 and may be related to cortical activity associated with being awareness of others, such as the medial prefrontal cortex and the temporal-parietal junction13. This PS and interpersonal physiology has been reported by many researchers and can be considered an established fact, at least with respect to skin conductance and heart activity. However, it is not well established as a theory, and theoretical interpretations based on embodiment theory, systems models, alliance, and empathy have been attempted, and theoretical specificity is lacking14. Perhaps the theoretical framework with the most explanatory power is Relational Psychophysiology by Ham and Tronick15. There, the synchrony phenomenon in therapeutic interactions can be interpreted in the broader context of attachment theory and dyadic models.

Occupational therapy is a profession that prioritizes the client's needs and promotes health and well-being through meaningful activities16. Crafts have long been used in psychiatric occupational therapy17,18. The use of crafts is decreasing, which may be related to the difficulty in explaining the physiological evidence, suggesting the need for a clear explanation of how occupational therapy can help19. In this regard, Shiraiwa et al. found that individuals who exhibited a frontal midline theta rhythm (Fmθ) during craft activities had increased sympathetic and parasympathetic activity20. Fmθ is a theta wave with a frequency of 5–7 Hz as identified by Ishihara and Yoshii21. It is a current-source electroencephalogram (EEG) component of the bilateral medial prefrontal cortex, including the ACC, that occurs when mental concentration is required22. Occupational therapy also includes group work as one of its eight intervention modalities23,24. Mosey proposed a five-level theory of group interaction. The lowest level, the parallel group, is a group of individuals who work and play with minimal task sharing and collective stimulation in the current of others25,26. The use of groups has been reported to improve subjective well-being and self-efficacy27. Lloyd suggested that when providing occupational therapy services in psychiatric settings, there are some key areas to consider when planning the group program, namely the client's expectations of group therapy, the structure of the group, and the importance of client feedback28. In terms of client expectations of group therapy, the basic therapeutic factors defined by Yalom are evaluated by clients in psychiatric occupational therapy groups29. Yalom called these basic therapeutic factors "curative factors", and the curative factors he listed were cohesiveness, altruism, catharsis, universality, existential factors, self-understanding, family reenactment, guidance, identification, instilling hope, interpersonal learning input, and interpersonal learning output30. Of these, Yalom found that different client populations rated cohesion as the most helpful therapeutic factor31. In other words, the theoretical basis is that groups used in occupational therapy have an effect of cohesion, togetherness, and a relaxation effect due to universality and social buffering, but the physiological basis for this effect is not fully understood. Cohesion and togetherness are related to the physiological synchrony of the dyad, and the relaxation effect can be assessed by measuring the activity of the autonomic nervous system.

The aim of this study was to investigate the effects of parallel groups in an occupational therapy setting on dyadic PS, and on subjects’ autonomic activity and EEG. The hypothesis was formulated as follows: Crafting in parallel groups will increase dyadic PS and subjects’ parasympathetic activity, as well as cortical activity in the medial prefrontal cortex (MPFC) and the temporal-parietal junction associated with being aware of others, and the insular cortex associated with parasympathetic activity.

To test these hypotheses, the dyadic PS and the subjects’ autonomic and EEG activity during task performance were simultaneously measured and analyzed. Cross-correlation analysis has been used to study synchrony in a variety of contexts, including posture and movement between two individuals32, emotion between mother and child33, heartbeat intervals34, maternal and fetal R-R interval (RRI) variability35,36, and electrodermal activity37,38. This method is particularly useful for analyzing heart rate synchrony over time. The Lorenz plot analysis39 was also used to assess the autonomic activity in the subjects. The Lorenz plot analysis allows the independent calculation of the index of sympathetic activity (cardiac sympathetic index; CSI) and the index of parasympathetic activity (cardiac vagal index; CVI). The Lorenz plot analysis was used in this study because of its sensitivity40 and lack of influence from the respiratory component41, making it a valuable tool for analyzing autonomic activity during craft activity. And exact Low-Resolution Electromagnetic Tomography (eLORETA)42 was used to estimate the current source density (CSD) of the EEG. Without assuming a specific number of active sources, eLORETA measures the electrical activity of neurons using CSD. The solution space is restricted to cortical gray matter and contains 6239 voxels with 5 mm spatial resolution.

Results

Data were collected from a sample of 30 participants consisting of 18 males and 12 females with a mean age of 20.17 ± 0.91 years. The cross-correlation coefficient (CCC) and lag values, a measure of dyadic heart rate PS, were missing for 4 participants in the alone conditions, for a total of 26 participants with available data in the alone condition. The 26 participants included 16 males and 10 females with a mean age of 20.23 ± 0.86 years. All were in good health with no history of brain or head trauma or mental disorders.

Comparison between conditions

Comparison of the 3 conditions revealed significant differences in the subjects’ CVI change and dyadic CCC change. Both subjects’ CVI change and dyadic CCC change were higher in the parallel condition (CVI change: 0.07 ± 0.17, CCC change: 0.03 ± 0.11) compared to the alone condition (CVI change: − 0.07 ± 0.22, CCC change: − 0.06 ± 0.14) (CVI change: p = 0.009, CCC change: p = 0.030) (Table 1).

Table 1 Comparison between conditions.
Full size table

Multiple regression analysis

Only the parallel condition (alone-parallel: p = 0.006, β = − 0.72, nonparallel-parallel: p = 0.038, β = − 0.51) and Fz theta power (p = 0.001, β = 0.37) were significant explanatory variables for the subjects’ CVI change (F(6, 79) = 3.76, p = 0.002, adjusted R2 = 0.16) (Table 2). There were no significant in the CSI/ RRI/ CCC/ lag change when used as a dependent variables.

Table 2 Multiple regression analysis results.
Full size table

Comparison of CSD between conditions and correlation with other indices

Analysis of CSD between conditions using eLORETA showed no significant differences.

Correlation analysis of EEG data from all conditions showed that CSD of the left fusiform gyrus in the delta band showed a significant positive correlation with CVI change (p < 0.01). CSD of the left insular cortex in the theta band also showed a significant positive correlation with CVI change (p < 0.01). CSD of the medial frontal gyrus in the beta 1 band showed a significant negative correlation (p < 0.05) with dyadic lag change. No significant correlations were found between subjects' CSD and other variables (Fig. 1A).

Figure 1
figure 1

Correlation between CSD and other indicators for all conditions (A) and in parallel condition (B). The color scale indicates the strength of the correlation, with lighter shades of yellow or blue representing larger t-values and smaller p-values. (A-Upper panel) CSD of the left fusiform and its peripheral area gyrus in the delta band showed a significant positive correlation with CVI change (p < 0.01). (A-Middle panel) CSD of the left insular cortex and its peripheral area in the theta band showed a significant positive correlation with CVI change (p < 0.01). (A-Lower panel) CSD of the medial frontal gyrus and its peripheral area in the beta 1 band showed a significant negative correlation with dyadic lag change (p < 0.05). (B-Upper panel) CSD of the ACC and its peripheral area in the delta band showed a significant positive correlation with dyadic lag change (p < 0.05). (B-Lower panel) CSD of the right inferior frontal gyrus and its peripheral area in the high gamma band showed a significant positive correlation with dyadic lag change (p < 0.05).

Full size image

In the parallel condition, the CSD of the ACC in the delta band showed a significant positive correlation with dyadic lag change (p < 0.05), and the CSD of the right inferior frontal gyrus in the high gamma band showed a significant positive correlation with dyadic lag change (p < 0.05) (Fig. 1B). No significant correlations were found in the alone and nonparallel conditions.

Discussions

The aim of this study was to investigate the effect of parallel groups in an occupational therapy setting on dyadic PS, and on subjects’ autonomic activity and EEG. The increase in CCC and CVI in the parallel condition may be due to PS and social buffering. These were observed even though there were no differences in task performance between conditions. In real occupational therapy in mental health, psychological effects are more important than task performance in crafts43. In other words, this study shows that craft activities in parallel groups in occupational therapy produce physiological changes due to social buffering and physiological synchronization. The results of this study support the physiological evidence of group cohesiveness, universality, and other curative factors through the theoretical rationale cited by Yalom30. They also highlight the complex interplay between social dynamics and physiological responses and may shed light on the mechanisms underlying population-based occupational therapy treatment.

It has been reported that the higher the PS, the greater the sense of togetherness12 and cohesion 11. PS is also higher for couples in a new relationship and couples in a special relationship than for acquaintance couples, suggesting that PS has a non-linear relationship with the level of intimacy44. In this study, other people close to the subjects were recruited as experimenters, and the higher PS in the parallel condition can be interpreted as a reasonable result. On the other hand, factors such as sexual attraction may influence PS in the male–female dyad45. In other words, it is possible that PS increased in each condition due to the subjects' sexual attraction to the experimenter who composed the field. Despite these limitations, PS is thought to occur mainly during joint tasks9 and face-to-face interactions10, but it has also been reported to occur between individuals who are simply present without direct interaction46. In the present study, the parallel condition involved individuals engaged in a craft activity without direct interaction. It was observed that similar PS occurred in this situation. Previous studies have shown that heart rate synchrony is related to emotion34,47. In the nonparallel condition, the participants experienced stress from being observed, which may have led to a lack of increased PS due to the discrepancy between their emotional movements and those of the observed individuals. In addition, the present study suggests that social buffering can occur even in situations where there is no physical contact48 or social interaction49. It is important to note that there was no correlation between PS and parasympathetic activity. Research has shown that PS, specifically related to the sympathetic nervous system, is positively correlated with human relationships50. Therefore, it is reasonable to conclude that the increased dyadic PS and the increased parasympathetic activity in the parallel group are due to separate mechanisms. In addition, social buffering has been linked to oxytocin51, which promotes in-group trust and cooperation52. In craft in parallel groups, these common neural and endocrine mechanisms may be involved in these phenomena of social buffering and PS. Our findings are positioned as supportive of these theoretical frameworks. In particular, the physiological changes observed in the social buffering and intimate peer-to-peer crafting situations provide evidence that the environmental setting of treatment reduces individual stress and physiological modulation. In conclusion, the results of PS and autonomic activity and the physiological data of ECG in this study suggest that the use of parallel craft activities in clinical occupational therapy may enhance a sense of unity and cohesion and induce a state of relaxation through another mechanism.

The Fmθ may increase parasympathetic activity. Previous studies have shown that the current source of Fmθ, ACC22, is associated with autonomic activity53,54. The results of the present study also support these findings. These results suggest that inducing a strong state of attentional focus to direct craft activity may effectively increase parasympathetic activity, such as relaxation, during treatment sessions. It is important to adapt the method of teaching the subject how to engage in craft activities should be adjusted based on the treatment goals.The lag change, which is an index of dyadic PS of heart rate, was found to have a negative correlation with frontal lobe beta activity. Administration of benzodiazepines, such as lorazepam and diazepam, which increase GABA action, can increase beta activity55,56. In addition, positive feedback can increase medial beta activity in the frontal cortex57. This suggests that frontal beta activity may contribute to the mechanism that produces anxiolytic effects and indicates a positive mental state. In the parallel condition, the study found that the positive correlation between ACC delta power and dyadic lag changes. ACC delta power has been reported to be associated with the worsening of neuropathic pain58 and anhedonia59. This suggests that the more negative the lag change, the lower the ACC delta activity, resulting in a more inhibitory effect, such as relief of neuropathic pain and anhedonia. These correlated regions overlap with the MPFC, and since the MPFC has been reported to be associated with awareness of others13. The present results that higher activity in this region leads to lower lag values, i.e., higher synchrony, are quite plausible. High gamma activity is observed in the right inferior frontal gyrus during auditory speech comprehension60. In order to approximate the timing of dyadic PS and induce a negative lag change, it may be necessary to avoid exposure to auditory stimuli and environmental noise and to avoid interference with auditory comprehension. In the present study, the content of these dyadic lag value correlations may reflect mental states and cognitive processes. Although the importance of dyadic lag values in PS of heart rate has not been adequately discussed, further studies should focus on them.

The study found a correlation between increased individual parasympathetic activity and increased delta activity around the left fusiform gyrus, as well as increased theta activity around the left insula. Previous research has shown that stress can increase activity in the fusiform gyrus61, particularly on the left side, which is also associated with fear perception62. Interpreting the increased delta activity as a modulation of network activity that should be inactive for task performance63, it is consistent with the findings of previous studies that increased delta activity in this region is positively correlated with increased parasympathetic activity, which produces effects such as relaxation. In addition, a study found a negative correlation between theta power in the left insular cortex and measures of depression64. As described above, it is also natural that the theta-band activity in the left insular cortex would be positively correlated with parasympathetic activity, as shown in previous studies. In summary, the results of the EEG data suggest that while the present findings reinforce previous findings showing that autonomic and cortical activity are related, the use of parallel handicrafts in clinical occupational therapy with a low lag between dyads, so to speak, and a highly synchronous dyad may have an anxiolytic effect and positive psychological states.

This study has several limitations to this study. The sample size in this study is 30, which is comparable to previous studies7 using similar statistical methods. However, it is not a large or sufficient sample size, so caution should be exercised in generalizing the results. Also, since the effect may be affected by whether the others in the field are of the same or opposite sex, caution should be exercised in generalizing the results in this regard as well. The parallel condition was the only situation in which both the subject and the experimenter were engaged in the same activity. However, it cannot be assumed that the heart rate changed significantly between conditions, as the amount of individual RRI change did not differ significantly between conditions. Only the rate of change per beat, i.e., the rhythm, was affected. In the parallel condition, the CCC change was statistically significant, but the value of the CCC was not high. This study used cross-correlation analysis, which allows the window size to be adjusted. It is important to confirm the robustness of the results by changing the window size or increasing the sample size. Furthermore, the ecological validity of the study is limited as the controlled experimental conditions may not fully replicate the complexity of real-world clinical occupational therapy settings. Potential biases could also arise from the subjective selection of experimenters who were close to the participants.

In this study, we also recruited others who were close to the subjects as experimenters. In a clinical occupational therapy setting, it is important to test not only one-on-one craft activities in a short time frame, but also longer periods of time involving multiple people with different interrelationships, such as language exchange and relationships. Regarding the minimum number of RRI data required for Lorenz plot analysis, it has been reported that a minimum of 30, 50, or 100 beats of RRI data are required65, and another study also found that a minimum of 100 beats of RRI data are required66. Therefore, the approximately 200 RRI data obtained from the 3-min task performed in this study should not be problematic. However, longer time periods may be useful for more accurate measurement of autonomic activity, and longer task durations are needed to replicate clinical occupational therapy situations. Validation of this study in a group of patients with psychiatric or mental disorders, the target population for clinical occupational therapy, will demonstrate the physiological efficacy of occupational therapy as routinely provided and lead to more effective rehabilitation. Incorporating diverse craft types, extending the duration of tasks, and involving a broader range of participant demographics and backgrounds in future studies would enhance the generalizability of the findings. Additionally, incorporating qualitative measures to assess subjective experiences and perceptions during group activities could provide a more comprehensive understanding of therapeutic processes.

In conclusion, tasks performed in parallel conditions, such as those used in occupational therapy, increase an individual's parasympathetic activity and dyadic PS. In addition, dyadic PS, particularly in parallel conditions, is supported by associations with frontal delta and gamma activity. These findings suggest that there are multiple physiological effects of engagement in parallel conditions. In other words, strong attentional focus on craft activities in the parallel group may be effective when relaxation is the goal. On the other hand, craft activities in the parallel group may also be useful for improving the sense of group unity, but unity is not associated with attentional focus. Therefore, it is necessary to selectively manipulate the level of attentional focus in order to selectively elicit each effect. In conclusion, tasks performed in parallel conditions, such as those used in occupational therapy, increase an individual's parasympathetic activity and dyadic PS. In addition, dyadic PS, particularly in parallel conditions, is supported by associations with frontal delta and gamma activity. These findings suggest that there are multiple physiological effects of engagement in parallel conditions. In other words, strong attentional focus on craft activities in the parallel group may be effective when relaxation is the goal. On the other hand, craft activities in the parallel group may also be useful for improving the sense of group unity, but unity is not associated with attentional focus. Therefore, it is necessary to selectively manipulate the level of attentional focus in order to selectively elicit each effect.

Methods

Subjects

We recruited 30 rehabilitation students aged 18–22 years who were able to complete craft activities. Individuals with a history of trauma to the brain or head, or with mental disorders, were excluded. This study was approved by the Ethics Committee of the Graduate School of Rehabilitation Sciences of Osaka Metropolitan University (2022–212) and Osaka Kawasaki Rehabilitation University (OKRU-RA0032), and the experiments were conducted in accordance with the ethical standards of the Declaration of Helsinki. Written informed consent was obtained from the participants.

Procedure

Subjects participated in a craft activity called net crafts, similar to a previous study20. Net crafting is a process of making bookmarks by threading single-color metallic yarn through a soft polyethylene net (a rectangle approximately 40 mm * 97 mm with 3 mm * 3 mm holes) using a 5.5 cm needle specially designed for metallic yarn. Net crafting is commonly used in occupational therapy in Japan because it is a simple task with many repetitions and is easier to do than knitting. The activity lasted for 3 minutes and was performed in 3 different conditions: alone, nonparallel, and parallel (Figure 2). In the alone condition, only one subject was seated while performing the craft. In the nonparallel condition, the experimenter observed the subject, i.e., only the subject worked on the craft, and the experimenter did not work on the craft, but only paid attention to the subject's work. In the parallel condition, both the subject and the experimenter were engaged in their respective crafts, i.e., there was little direct interaction, and each was absorbed in the craft. In both conditions, the subject and experimenter were instructed not to engage in verbal interaction. To minimize artifacts introduced into the EEG and ECG, subjects and experimenters were instructed to minimize body movements during manual handling. Subjects were equipped with electroencephalographs and electrocardiographs and were measured after sufficient practice with the actual task. To avoid subject and experimenter fatigue, drowsiness, and increased scalp electrode impedance, each condition lasted 3 minutes, with 1 minute 30 seconds of rest before and after each condition. This ensured that the time from electrode application to removal was approximately 1 hour. The experimenter for each subject was a close relative of the subject and was recruited at the same time as the subject. This is based on the fact that physiological synchrony is more likely to occur with close others, as in previous studies9 that observed PS while controlling for intimacy, and that in clinical occupational therapy, group members are assumed to be homogeneous others.

Figure 2
figure 2

Experimental conditions and procedures. The image above shows the three conditions: alone, nonparallel, and parallel. In each condition, the subject rested for 1 min and 30 s before and after a 3-min task.

Full size image

Measurement

EEG

EEG activity during each craft activity was recorded using a Polymate Pro MP6100 electroencephalograph (Miyuki Giken Co., Ltd.) with 19-channel of electrodes placed on the subject's scalp according to the international 10–20 system. The sampling rate was 1000 Hz, and the electrode impedance was less than 10 KΩ. EEG activity was measured for the subject only.

ECG

The Portable Electrocardiogram Polyam (ECG) IIB (Nihon Santeku Co., Ltd.) was used to measure the ECG of both the subject and the experimenter during each craft activity at a sampling rate of 1000 Hz. In the alone condition, the subjects and the experimenter were not visible to each other, and their ECGs were measured simultaneously while only the subject engaged in the craft activity. In both the nonparallel and parallel conditions, both subjects' ECGs were measured simultaneously while both were visible to each other. Two identical electrocardiographs were prepared, one for the subject and one for the experimenter, so that the electrocardiograms of the two subjects were measured simultaneously. The electrocardiographs worn by the subject and experimenter, respectively, were externally connected to the "SENSOR" channel of the electroencephalograph used in this study. Therefore, the two ECG signals were perfectly synchronized in time with the electroencephalograph, and the ECG data between the two were set up in such a way that no temporal deviation of ECG data occurred between them. The electrodes were placed as follows: -(G1) on the upper sternum, E(BE) on the lower right intercostal space, and +(G2) on the lower left intercostal space. This is a specific setting that does not correspond to the general standard 12-lead ECG, and is a method of placement in which a waveform similar to induction II, in which the R wave is large and upward, is observed. MATLAB (R2022a) was used to extract the RRI every second from the measured ECG data. The smoothed RRI waveform was analyzed using a moving average over a 3-second interval.

To assess subjects’ autonomic activity, we calculated the CSI as a measure of sympathetic activity and the CVI as a measure of parasympathetic activity using Lorenz plot analysis39 of the subject's smoothed RRI. Lorenz plot analysis was performed using a computer program developed by Sato et al.67.

To assess the interpersonal PS of heart rate between subject and experimenter, cross-correlation analysis was performed on the second-by-second rate of change of the smoothed RRI of the subject and the experimenter for the first and last 90 seconds of data. The CCC, which was maximized within a 10-second lag value at that time, was also calculated using MATLAB (R2022a). The amount of change was determined by subtracting the first 90 seconds of data from the last 90 seconds of data (CSI/ CVI/ RRI/ CCC/ lag change). In other words, the first 90 seconds were used as a baseline to calculate how the data changed from the first half to the second half.

Task performance

The number of steps achieved by the completed net crafts was measured.

Statistical analysis

The EEG data were downsampled to 500 Hz using EEGLAB (v2021.1) in MATLAB and bandpass filtered between 0.5 and 100 Hz. Independent Component Analysis (ICA) was performed using the runica algorithm in EEGLAB68. After applying ICA, we removed the frontal components identified as eye movement and blinking, and the occipital component identified as primarily an electromyogram for neck retention. The artifact-influenced interval was then subtracted without removal. To study Fmθ quantitatively, we used the Darbeliai v2022.12.21.1 plug-in in EEGLAB to calculate power values in the theta band at the Fz electrode. The window length was set to 2 seconds and the spectral steps in 1 Hz were set to 10. The theta band was set to 4–8 Hz.

Subjects' autonomic data for CSI/ CVI/ RRI change and interpersonal PS data for CCC/ lag change were analyzed between conditions using one-way ANOVA, assuming normality and equality of variance. We checked the assumptions of normality using the Shapiro-Wilk test and the equality of variance using Levene's test before performing the ANOVA. The Tukey's HSD test was applied to correct for multiple comparisons. Because CSI change and Fz theta power were not normally distributed, a Kruskal-Wallis test was used instead of ANOVA. To analyze the variables affecting CSI/ CVI/ RRI/ CCC/ lag change, we performed a multiple regression analysis. We used CSI/ CVI/ RRI/ CCC/ lag change data as dependent variables and other variables as independent variables. That is, when CSI change was the dependent variable, condition and task performance, Fz theta power and CCC /lag change were the independent variables, and when CCC change was the dependent variable, condition and task performance, Fz theta power, CSI/CVI/RRI change were the independent variables. We checked the assumptions for multiple regression analysis, including linearity, independence of errors, homoscedasticity, and absence of multicollinearity. Variance Inflation Factor (VIF) was used to assess multicollinearity, ensuring that all VIF values were below 10. We performed these statistical procedures using jamovi 2.4.0.069,70,71,72 for these statistical procedures, and set the significance level for each test to 5%.

During the craft activity, EEG activity was evaluated using eLORETA42. In the present study, the CSD value was used to estimate the brain electrical source responsible for the generation of scalp recording potentials in each frequency band. EEG data preprocessed in EEGLAB at 0.5–100 Hz were used to calculate eLORETA cortical CSDs for eight frequency bands: delta (1–3.5 Hz), theta (4–8Hz), alpha 1 (8.5–10Hz), alpha 2 (10.5–13 Hz), beta 1 (13.5–20 Hz), beta 2 (20.5–30 Hz), low-gamma (30.5–60 Hz), and high-gamma (60.5–100 Hz). Statistical analysis of CSD between conditions was performed using nonparametric mapping (SnPM) method in eLORETA. Regression analysis using eLORETA software was employed to analyze the association between CSD and CSI/CVI/RRI/CCC/ lag data and task performance. Multiplicity was determined using SnPM with a critical threshold for setting the p-value at p = 0.05 and correction for multiple comparisons at all frequencies. In the resulting 3D statistical images, functional associations between ROIs with significant differences were identified using a nonparametric randomization/replacement procedure. Randomization and permutation tests were performed by estimating the empirical probability distribution of the "maximum statistic" under the null hypothesis. eLORETA used 5,000 data randomizations to determine the threshold for marginal detection and corrected multiple comparisons by total FC and total frequency, independent of Gaussianity73. In addition, eLORETA's nonparametric randomization procedure based on "maximum statistics" was used to correct for multiple comparisons. The omnibus null hypothesis was rejected if at least one t-value exceeded a critical threshold of p = 0.05 determined by 5,000 data randomizations.