If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Institute of Developmental Psychology, Faculty of Psychology, Beijing Normal University, Beijing, 100875, ChinaState Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, 100875, China
State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, 100875, China
State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, 100875, China
Key Research Base of Humanities and Social Sciences of the Ministry of Education, Academy of Psychology and Behavior, Tianjin Normal University, Tianjin, 300387, China
Corresponding author. State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, No. 19 Xinjiekouwai Street, Beijing, 100875, China.
State Key Laboratory of Cognitive Neuroscience and Learning & IDG/McGovern Institute for Brain Research, Beijing Normal University, Beijing, 100875, China
Anodal tDCS on one individual changed neural synchronization between individuals.
•
Anodal tDCS on one individual changed empathy shared by two individuals.
•
Anodal tDCS on one individual didn't change interaction behaviors.
•
Nonverbal but not verbal behaviors mediated the change in brain and empathy.
•
Our findings support the hierarchical architecture of the dual-brain function.
Abstract
Background
Previous studies have shown that interpersonal neural synchronization (INS) is a ubiquitous phenomenon between individuals, and recent studies have further demonstrated close associations between INS and shared external sensorimotor input and/or internal mental processes within a dyad. However, most previous studies have employed an observational approach to describe the behavior-INS correlation, leading to difficulties in causally disentangling the relationship among INS, external sensorimotor input and the internal mental process.
Objective/hypothesis
The present study aimed to directly change the level of INS through anodal transcranial direct current stimulation (tDCS) to test whether the change in INS would directly impact the internal mental process (Hypothesis 1) or indirectly through external sensorimotor input; the interaction behaviors were also changed (Hypothesis 2) or not (Hypothesis 3).
Methods
Thirty pairs of romantically involved heterosexual couples were recruited for a within-subjects design. Three conditions were assessed: a true stimulation condition with 20-min anodal high-definition tDCS to the right anterior temporal lobe (rATL) of women before they communicated with their partners, a sham stimulation condition and a control brain region stimulation condition. The comparison between the true and sham or control brain region conditions allows us to detect the true effect of brain stimulation on INS. Functional near-infrared spectroscopy (fNIRS) hyperscanning was used to simultaneously collect dyadic participants' hemodynamic signals during communication. INS, empathy, and interaction behaviors were examined and compared among different stimulation conditions.
Results
True brain stimulation significantly decreased INS between the rATL of the women and sensorimotor cortex (SMC) of the men compared to the sham stimulation condition (t(27.8) = −2.821, P = 0.009, d = 0.714) and control brain region stimulation condition (t(27.2) = −2.606, P = 0.015, d = 0.664) during communication. It also significantly decreased the level of emotional empathy (F(2,145) = 6.893, P = 0.001) but did not change sensorimotor processes, such as verbal or nonverbal interaction behaviors. However, nonverbal behaviors mediated the relationship between the changes in INS and emotional empathy (lower limit confidence interval = 0.01, upper limit confidence interval = 2.66).
Conclusion(s)
These findings support the third hypothesis, suggesting that INS is associated with the shared internal mental process indirectly via the sensorimotor process, but the sensorimotor process itself does not covary with the INS and the associated internal mental process. These results provide new insight into the hierarchical architecture of dual-brain function from a bottom-up perspective.
Maternal and paternal plasma, salivary, and urinary oxytocin and parent-infant synchrony: considering stress and affiliation components of human bonding.
] between children and their mothers. This synchrony may further extend into and underlie social interaction between adults, such as romantic couples, close friends, and even strangers [
]. Recently, with the emergence of the hyperscanning technique, the synchrony of neural activity (i.e., interpersonal neural synchronization [INS]) has been extensively investigated among adults [
]. These studies showed that INS is associated not only with shared sensorimotor input but also with shared mental processes, such as the representation of semantic and social concepts [
]. Most previous studies, however, have employed an observational approach to describe the behavior-INS correlation, leading to difficulties in causally disentangling the relationship among INS, external sensorimotor input and the internal mental process.
Here, we aimed to address this issue by directly changing the level of INS through transcranial direct current stimulation (tDCS) to test whether and how the change in INS would directly impact the internal mental process or indirectly through external sensorimotor input. Previous studies have shown that empathy is a key component of social interaction when an individual tries to understand the other's mental process [
] in the ATL of patients. Most importantly, a recent study on social interaction between romantic couples found that touch induced a significant increase in INS between the right ATL (rATL) in women and the right temporoparietal junction (TPJ) in men, and empathy modulated the relationship between INS and the strength of affiliative pair bonding [
]. The INS peaked when the rATL's activity preceded that of the TPJ by 4 s, which corresponded to verbal interaction behaviors that were initiated by women and had a duration of approximately 4 s [
]. Additionally, a similar study also showed a directional information flow from women to men using Granger Causality Analysis of brain signals during social interaction [
]. Thus, it seems that the rATL in women plays a leading role, whereas the TPJ in men plays a following role during communication. Finally, behavioral studies have also shown that men are more inclined to follow and be affected by women's behaviors during interaction [
]. Therefore, in this study, we directly stimulated the rATL in women prior to social interaction between romantic couples to test the relationship among INS, interaction behaviors, and empathy.
Based on previous studies, we proposed three hypotheses. The first hypothesizes that INS is directly associated with empathy (Hypothesis 1). Previous studies have shown that INS is associated with the shared representation of concepts [
]. Moreover, during social interaction, INS associated with shared linguistic components can be dissociated from that with shared sensorimotor input in both anatomical locations and temporal patterns [
]. Thus, brain stimulation of the rATL in women was predicted to significantly decrease INS between the rATL of women and brain regions of men and subsequently impact empathy but not interaction behaviors. The second hypothesis is that the change in INS is not directly associated with empathy (Hypothesis 2). Previous studies have shown that episodes with verbal turn-taking [
] during communication are associated with higher INS than those without such interaction behaviors. Moreover, INS associated with nonverbal behaviors is also associated with empathy [
]. Thus, it was predicted that brain stimulation of the rATL of the women would significantly decrease INS and then impact both interaction behaviors and empathy. There is also a third possibility that brain stimulation of the rATL would decrease INS and subsequently impact empathy through interaction behavior (i.e., a mediation effect); the interaction behaviors themselves, however, would not necessarily be changed (Hypothesis 3).
To date, only three studies have used transcranial alternating current stimulation (tACS) to examine the relationship between INS and interaction behaviors [
]. However, these studies have several limitations. First, they only assumed but never actually tested whether tACS changed INS. Second, none of these studies included a control brain region as a comparison, leaving an unresolved question of whether the stimulation effect on behavior was specific to the target brain region. Third, these studies only investigated the change in the sensorimotor process but not that of internal mental processes.
In this study, we applied tDCS to the rATL of romantically coupled women before they were involved in naturalistic communication with their partners (Fig. 1). Here, tDCS rather than tACS was used because previous research has shown that tACS may work by modulating the frequency of brain signal oscillations, such as the alpha band of EEG signals [
]. Thus, it would be better to know the specific frequency that we are interested in to ensure the effectiveness of tACS. However, in our study, we did not have such an a priori hypothesis. In contrast to tACS, tDCS does not have such an a priori requirement. When tDCS is applied, the resulting current flow in the brain induces a subthreshold alteration of neuronal resting membrane potentials. Both anodal and cathodal tDCS have been found to have significant impacts on the target brain region, particularly those associated with high-order cortical processes such as decision-making, memory, language and sensory perception [
]. Thus, anodal tDCS was employed in the present study.
Fig. 1The experimental setup. (a) Experimental procedure. (b) The optode probe set was placed on the bilateral frontal, temporal, and parietal cortices. CHs 11 and 25 were placed at T3 and T4, respectively, in accordance with the international 10–20 system. The Montreal Neurological Institute (MNI) coordinates of the probes are provided in Table S3. S, source; D, detector. (c) tDCS protocol. For each condition, the red circle sticker indicates the position of the anode, whereas the green circle stickers indicate the positions of the cathodes. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
We predicted that true tDCS would impact the INS between the rATL of women and men. Additionally, we also predicted that brain regions associated with social cognition, such as the sensorimotor cortex (SMC), might also be involved because SMC is an important component of the perception-action system in empathy [
]. Specifically, according to the perception-action model, the perception of an object's state will activate the subject's corresponding representations, which subsequently activate somatic and autonomic responses. This mechanism supports social behaviors relating to empathy, such as social facilitation, vicariousness of emotions, responsiveness, etc. [
]. Thus, stimulation of the rATL in women would decrease INS associated with the rATL or SMC in men.
2. Materials and methods
2.1 Participants
Thirty pairs of romantically involved heterosexual couples (mean age of the men = 22.43 ± 2.3 years, mean age of the women = 21.38 ± 2.40 years) were recruited through advertising in Beijing (20 pairs) and Tianjin (10 pairs). According to their self-report, the couples had known each other for an average of 20.60 months (S.D. = 12.45 months) and had been in romantic relationships for an average of 14.52 months (S.D. = 8.56 months).
The study protocol was approved by the Institutional Review Board of the State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University. Written informed consent was obtained from all participants.
2.2 Tasks and design
Each pair of participants experienced one experimental condition and two control conditions on three different days. In the experimental condition, true tDCS was delivered to the rATL of the women. In the sham control condition, sham tDCS was administered to the rATL of the women. In the control brain region condition, true tDCS was delivered to the occipital lobe of the women. The sequence of stimulation conditions was counterbalanced across participant pairs. The intervals between two stimulation conditions were longer than 24 h to avoid potential confounding between conditions.
On each day (Fig. 1a), the participants were first asked to complete behavioral assessments on the level of empathy (Table S1) and the strength of romantic love (Table S2). Second, participants were asked to complete a 5-min resting-state session (pre-stimulation rest). During this session, no tDCS was delivered, and the participants were asked to remain still with their eyes closed, relax their minds, and remain as motionless as possible. Third, immediately after the resting-state session, participants were requested to complete a 20-min brain stimulation session. During this session, the instruction to the participants was the same as that provided for the resting-state session. Fourth, a 5-min post-stimulation resting-state session followed the brain stimulation session. Fifth, participants performed a naturalistic communication task in which pairs of participants freely discussed a supportive topic [
] for 15 min to arouse empathy for their partners. During this process, the experimenter left the room to provide a private situation for the participants. Finally, they were asked to complete the same behavioral assessments as those in the first step at the end of the experiment. The entire experimental procedure was recorded on video with permission from participants.
2.3 Behavioral assessment of states of empathy and romantic love
Each day, participants were asked to complete behavioral assessments on the level of empathy (Table S1) and the strength of romantic love (Table S2) immediately before and after the experiment. The scales were modified from the Interpersonal Reactivity Index (IRI) [
] to measure the current state of empathy and the strength of romantic love, respectively. The interitem reliability of the modified IRI was satisfactory (Cronbach's α: perspective taking, 0.781; empathic concern, 0.728). The interitem reliability of the modified TLS was high (Cronbach's α: intimacy, 0.967; passion, 0.956; commitment, 0.967).
2.4 tDCS procedure
We used a high-definition anodal tDCS (HD-tDCS) system with a NeuroConn DC stimulator. The stimulator has five round rubber electrodes with a diameter of approximately 20 mm. For the true and sham conditions, the anode was centered over the rATL of the women (measurement channel [CH] 26, see below) with four cathodes spaced ∼3 cm away to establish a ring-like orientation (Fig. 1c). For the control brain region stimulation condition, the anode was centered over the women's occipital lobe. Both true and control brain region stimulation used an active stimulation protocol. During this protocol, the electrical current was initially ramped up to 1 mA over 10 s (fade-in), maintained for 20 min and then ramped down to 0 mA over 5 s (fade-out). The same fade-in and fade-out parameters were used in the sham condition, but the current was only maintained for 30 s. These manipulations were established based on previous tDCS research indicating that most individuals only have the itching sensation at the initial stage [
]. We followed this approach in our sham stimulation condition. Additionally, previous evidence has indicated that under a similar protocol, participants could not distinguish true stimulation from sham stimulation [
]. The participants in our study did not report any differences they had felt between the three days.
2.5 fNIRS data acquisition
Two LABNIRS systems (Shimadzu Corporation) were used to collect the fNIRS data. Four sets of customized optode probes (5 emitters and 5 detectors, 13 CHs) were used. Two sets were used for men, and the other two sets were used for women. For each participant, two sets of optode probes covered the frontal, temporal, and parietal cortices of the left and right hemispheres. The international 10–20 system was used to roughly localize the anatomical structures below the measurement CHs. Specifically, T3 and T4 in the left and right temporal cortices corresponded to the positions of CH11 and CH25, respectively [
]. The probe sets were examined to ensure that the positions were consistent among participants (Fig. 1b). To confirm the anatomical locations of the optode probes, we obtained MRI data from two female and two male participants who wore plastic caps on which the probes' true positions had been marked using Vitamin E balls (see the supplementary materials, SM). The Montreal Neurological Institute (MNI) coordinates of the probes are provided in Table S3.
The absorption of near-infrared light at three wavelengths (780, 805, and 830 nm) was measured at a sampling rate of 55.6 Hz. The changes in the oxyhemoglobin (HbO) and deoxyhemoglobin concentrations (HbR) were obtained for each CH based on the modified Beer‒Lambert law. Because previous studies showed that HbO was the most sensitive indicator of changes in regional cerebral blood flow and had the highest signal-to-noise ratio in fNIRS measurements [
], this study focused on changes in the HbO concentration.
2.6 Coding the interaction behaviors
Due to technical issues, we failed to capture a video of communication between one pair of participants; thus, complete data were obtained for 29 pairs. To test whether true brain stimulation changed the interaction behaviors, both verbal and nonverbal behaviors were coded according to the video recordings (Hypothesis 2). For verbal behaviors, the speech of each participant during communication was transcribed into text by six coders. The transcription was carefully inspected character-by-character by an experimenter (Y. L.) to ensure accuracy. Based on the text, two indices for verbal behaviors were calculated. One index was the number of turn-takings and responses, and the other index was the duration of speaking. For nonverbal behaviors, head movements, such as nodding or head shaking; gaze; facial expressions, such as smiling or laughing; hand touching; and manual gestures were coded by additional four coders who were unaware of the purpose of the experiment. The intercoder consistencies were acceptable (Cronbach's α > 0.7). To further exclude the potential influence of coders, we dummy-coded the identity of the coders and regressed them out from the percentage of verbal and nonverbal behaviors (i.e., the total duration of the task period was divided by the duration of verbal or nonverbal behaviors, and then the quotient was multiplied by 100). Then, we averaged the standardized residuals of verbal and nonverbal behaviors separately as features of verbal and nonverbal behaviors.
2.7 Behavioral data analyses
To test whether true brain stimulation had changed the process of empathy among dyads of participants (Hypotheses 1–3), scores of cognitive (i.e., perspective taking) or emotional (i.e., empathic concern) empathy obtained pre-stimulation were regressed out from those obtained post-stimulation using a linear regression method. The residuals of the regression were taken to index the empathy level of participants post-stimulation (i.e., the baseline level was controlled). Thus, the residuals were components that we were interested in. Next, given that no significant difference was found between the sham and control brain region stimulation conditions in the residuals, the residuals were averaged between sham and control brain region stimulation conditions. Then, the averaged residuals were subtracted from the residuals of the true stimulation condition. The difference was used to indicate the pure change in empathy due to true tDCS. Finally, a linear mixed model (LME) analysis was performed on the pure change in empathy. Here, gender, stimulation condition, and their interaction were fixed effects, while pair identity was a random effect. To test whether the interaction behaviors differed among conditions, we performed a similar LME on the features of verbal and nonverbal behaviors. The effect size of the LME was measured by d [
During preprocessing, the first and last 15 s of data in each task were removed to obtain data within the steady state period. Then, the data were downsampled to 11 Hz to reduce the computing times. The functions in Homer3 [
]. During this analysis, the components that accounted for 80% of the variance were removed from the raw HbO signals. Here, no filtering procedure was performed because, as shown below, INS was calculated using wavelet transform coherence (WTC), which can effectively reject physiological noise since each wavelet acts as a bandpass filter with very high attenuation in its stopband [
]. After calculating INS, the results associated with the heart rate (>0.8 Hz), breathing rate (0.15–0.3 Hz), Mayer waves (∼0.1 Hz), and very low frequency (<0.01 Hz) oscillations [
Current status and issues regarding pre-processing of fNIRS neuroimaging data: an investigation of diverse signal filtering methods within a general linear model framework.
To assess INS between the two fNIRS time series (one for each participant) within a pair, the MATLAB function “wcoherence” was used to perform WTC as a function of frequency and time [
] are the two most frequently used methods for calculating INS. PLV is performed on the frequency domain and is more widely used in EEG hyperscanning, whereas WTC is conducted on both time and frequency domains and is more widely used in fNIRS hyperscanning [
]. Additionally, compared to other methods in the time domain, WTC is capable of uncovering locally phase-locked response patterns at specific time points that are related to different social interaction behaviors [
The WTC result was a 2D matrix of the coherence values, where the column and row corresponded to a specific frequency and time point, respectively. All possible CH combinations between the two participants of a pair were examined (i.e., 26 × 26 = 676 in total). In addition, previous studies have indicated that INS usually involves a time lag probably due to the social prediction of upcoming information or a delayed shared representation [
]. To incorporate this effect, we calculated the coherence value by shifting the time course of the men forward or backward relative to that of the women from 2 to 12 s (step = 2 s) as well as when no time lag was added (i.e., the two series were temporally aligned). Finally, the coherence values were converted into Fisher's Z values and time-averaged across the task period. These procedures were conducted for all tasks.
2.8.3 Statistical analyses of INS during naturalist communication
To test whether tDCS changed INS during communication, the following analyses were conducted (Hypotheses 1–3).
2.8.3.1 Pattern of INS differences between conditions
First, INS was averaged across time. Next, for each participant pair in each stimulation condition at each frequency point, the INS for all CHs was concatenated into a 676-dimension vector (Fig. 2a, Step 1). Second, a LME analysis was performed on the concatenated INS vector for each pair of participants using the stimulation condition as a fixed effect, the CH identity as a random effect, and INS of pre-stimulation rest as a covariate (Fig. 2a, Step 2). This procedure produced beta values for each comparison (i.e., true vs. sham, true vs. control, and sham vs. control), indicating the difference between conditions. These analyses were conducted on each frequency point and time lag (i.e., when the brain signal of one individual lagged behind that of the other by 0–12 s; Fig. 2a, Step 3).
Fig. 2The overall analytic procedures. (a) Analytic procedures for communication. Step 1. Create an INS vector for each participant pair. Step 2. Obtain the beta values of each participant pair through LME, which indicates the difference between conditions. Step 3. Test whether the beta values of each comparison are different from 0 through one-sample t tests at the group level. Step 4. Select frequency clusters that had shared frequency ranges between the two comparisons of conditions (i.e., true vs. sham and true vs. control). INS in the selected frequency clusters were averaged, and t values between conditions were computed again. Then, the t values of the original dyads of participants were compared with the null distributions. Step 5. Specifically examine which CH combinations showed significant changes in INS through LME at the selected frequency band. (b) The overall analytic procedures for the resting state. Step 1. Create an INS vector of pre- and post-stimulation rest for each participant pair. Step 2. The r values of each participant pair were obtained through Pearson's correlation coefficient, which indicated similarity between pre- and post-stimulation rest. The r values were then converted into Fisher's Z values. Step 3. Fisher's Z values between conditions were compared through paired two-sample t tests at the group level. Step 4. Select frequency clusters that had shared frequency ranges between the two comparisons of conditions (i.e., true vs. sham and true vs. control). INS in the selected frequency clusters were averaged, and t values between conditions were computed again. Then, the t values of the original dyads of participants were compared with the null distributions. Step 5. Test whether the change in INS during rest contributed to the change in INS during naturalistic communication through a stepwise linear regression.
One-sample t tests (two-tailed) were applied to the beta values (Fig. 2a, Step 3). A cluster-based permutation method was used to correct for multiple comparisons across frequency points within each time lag (Fig. 2a, Step 4). First, male and female participants were randomly paired to form new pairs of participants who had been in the same stimulation condition but had not interacted with each other. INS was calculated within the new pairs. Second, the same aforementioned procedures (i.e., the LME and t-test) were conducted between true and sham stimulation conditions and between true and control brain region stimulation conditions. Frequency clusters showing significant differences between conditions at the P < 0.05 level were identified. The frequency clusters that had shared frequency points between the two comparisons (i.e., true vs. sham and true vs. control) were selected. INS in the selected frequency clusters were averaged. T values between conditions were computed again for that specific frequency cluster to indicate the overall difference between conditions of that frequency cluster. The aforementioned first and second procedures were repeated 1000 times to generate a null distribution of cluster-based t values for each pair of comparisons. Finally, the t values of the original dyads of participants were compared with the null distributions that were obtained from the new dyads. It was assumed that true brain stimulation should induce significant INS changes compared to both sham and control brain region conditions, but no significant differences were expected between sham and control brain region conditions. Based on this criterion, frequency clusters of interest were determined in the original participant pairs.
2.8.3.3 Spatial localization
After determining frequency clusters, we were interested in which CH combinations in the selected frequency clusters showed significant differences between conditions (Fig. 2a, Step 5). We compared INS at each CH combination between each pair of stimulation conditions using a LME analysis across all dyads of participants after averaging INS in the selected frequency cluster. Here, the effect of the stimulation condition served as a fixed effect, the pair identity of participants served as a random effect and INS of pre-stimulation rest served as a covariate. This analysis was only conducted on INS between the rATL of the women (CH26) and all CHs of the men because tDCS was only delivered to the rATL of the women. A permutation approach was used to correct multiple comparisons across CH combinations. Specifically, male and female participants were first randomly paired to form new pairs of participants who had been in the same stimulation condition but had not interacted with each other. Second, INS was computed within the new pairs of participants between the ATL of the women (CH26) and all CHs of the men and then averaged across time lags and frequencies. Third, a LME analysis was performed on INS. Fourth, the above procedures were repeated 1000 times to generate a null distribution of t values. Finally, the t values of the original pairs were compared with the null distribution (two-tailed threshold, P < 0.05). In this article, the uncorrected P values that survived corrections were reported.
2.8.3.4 Validation of the INS changes
To further validate the effect of true brain stimulation on INS, dyads of participants were split into three subgroups according to the experimental order: applying true stimulation on the first day (8 groups), the second day (11 groups), and the third day (11 groups). Then, a LME analysis was conducted on INS using stimulation condition, stimulation order, and their interaction as fixed factors. Additionally, we tested the INS difference across three stimulation conditions in each subgroup (Wilcoxon signed-rank test, Monte Carlo simulation with 10000 samples, one-sided, 95% confidence level).
2.8.4 Statistical analyses of INS during post-stimulation rest
To examine whether true brain stimulation also changed the INS of spontaneous brain activity even during the post-stimulation resting state, we compared INS during the resting state between stimulation conditions. It was expected that true brain stimulation would decrease the similarity of INS patterns between pre- and post-stimulation rests, whereas sham or control brain region stimulation would not. To this end, we performed the following analyses.
2.8.4.1 Pattern of INS differences between conditions
For each dyad of participants, the similarity of the overall INS pattern (Fig. 2b, Step 1) between pre- and post-stimulation rests across all CH combinations was calculated by Pearson's correlation coefficients. The correlation coefficient of the r-value was transformed into Fisher's Z (Fig. 2b, Step 2).
2.8.4.2 Frequency selection
A paired two-sample t-test (one-tailed, P < 0.05) was conducted between each pair of stimulation conditions on the Fisher's Z values across dyads of participants (Fig. 2b, Step 3). These analyses were conducted at each frequency point and time lag (Fig. 2b, Step 3). A similar cluster-based permutation method (2.8.3) as mentioned above was used to correct for multiple comparisons across frequency points within each time lag (P < 0.05, Fig. 2b, Step 4). After the frequency cluster of interest was selected, INS was averaged within the frequency cluster and used for subsequent analyses.
2.8.4.3 Spatial localization: the relationship between INS changes during rest and naturalistic communication
To test whether the INS change during the post-stimulation resting state contributed to the INS change during communication (Fig. 2b, Step 5), the following procedures were applied. Specifically, for each stimulation condition, INS during pre-stimulation rest was regressed out from that during post-stimulation rest, and the residuals were obtained. Second, the averaged residual of INS between sham and control brain region stimulation conditions was subtracted from the residual of INS during the true stimulation condition and used to indicate the pure INS changes in post-stimulation rest. The same procedure was performed on the INS during communication to obtain the pure INS changes in communication. Third, a linear stepwise regression modeling analysis was conducted using the pure INS change of communication as the dependent variable and the pure INS changes of the post-stimulation resting state from the ATL of the women to all CHs of the men as independent variables.
2.9 The association among INS, interaction behaviors and empathy
To test whether true stimulation-induced changes in INS impacted the level of emotional empathy through a mediation effect of interaction behaviors (Hypothesis 3) or not (Hypothesis 1), we conducted a chain mediation analysis (95% confidence intervals, 5000 bootstrap samples). We first obtained a pure change in emotional empathy and interaction behaviors using a procedure similar to that used to obtain the pure change in INS, except that no pre-stimulation interaction behaviors needed to be regressed out. Second, in the mediation model mentioned above, the pure INS change in communication was used as the independent variable of X, the pure change in emotional empathy was used as the dependent variable of Y, and the pure change in verbal or nonverbal behaviors was used as the mediator.
3. Results
3.1 True brain stimulation induced a significant change in INS during naturalistic communication
The results revealed only one frequency cluster of 0.04–0.05 Hz that showed significant changes in INS in the true stimulation condition compared with both the sham and control brain region stimulation conditions when the brain signals of the men lagged behind those of the women by 2–6 s (Fig. 3a and b). After averaging INS across time lags (i.e., 2–6 s) and frequency points (i.e., 0.04–0.05 Hz), INS was also significantly lower in the true stimulation condition than in the sham (t(29) = −2.775, P = 0.010, d = 0.510) or control brain region stimulation condition (t(29) = −3.645, P = 0.001, d = 0.659); however, no significant difference was found between the control brain region and sham stimulation conditions (t(29) = 0.198, P = 0.844, d = 0.034). The averaged INS was used for subsequent analyses.
Fig. 3The results of the comparison between stimulation conditions during communication. (a) The difference between true and sham (top), between true and control brain region (middle), and between control brain region and sham (bottom). The black rectangles indicate the time lags and frequency bands that showed significant differences both between true and sham conditions and between true and control brain region conditions but not between sham and control brain region conditions. (b) The null distribution of the t values generated by permutation when the signal of the men lagged behind that of the women by 6 s. Left, the comparison between true and sham conditions; right, the comparison between true and control brain region conditions. A t value of zero indicates that no clusters have been found in a given random sample. The red lines indicate t values within 0.04–0.05 Hz (averaged) in the original pairs, and gray areas indicate lower and upper 2.5% areas. (c-e) True stimulation decreased INS at ATLwomen→SMCmen during communication. (c) The anatomical positions of the ATLwomen→SMCmen. (d) The null distribution of the t value on INS between CH 26 of women and all CHs of men. Top, true vs. sham; bottom, true vs. control. The t value of ATLwomen→SMCmen (red line) in the original pairs was significant at the 5% level (two-tailed, gray color). (e) The residuals of INS at ATLwomen→SMCmen after regressing out the INS of pre-stimulation rest. The X-axis is the experimental order. Here, the uncorrected p values that survived the correction are reported. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Next, when comparing the averaged INS between each pair of stimulation conditions across all dyads of participants for each CH combination (Fig. 2a, Step 5), the results showed that INS between the rATL of the women (CH26) and the SMC of the men (CH20, Fig. 3c) was significantly lower in the true stimulation condition compared with the sham (t(27.8) = −2.821, P = 0.009, d = 0.714, Fig. 3d, top) or control brain region stimulation conditions (t(27.2) = −2.606, P = 0.015, d = 0.664, Fig. 3d, bottom). No significant difference was found between the control brain region and sham stimulation conditions (t(28.7) = −0.457, P = 0.651, d = 0.118). Together, these findings indicated that true stimulation prior to communication significantly decreased the level of INS during subsequent communication.
3.2 Validation of the INS change due to true brain stimulation
Again, the results showed that INS between the ATLwomen and SMCmen was significantly lower in the true stimulation condition compared with the sham (t(25.6) = −2.635, P = 0.014, d = 0.676) or control brain region stimulation condition (t(26.8) = −2.553, P = 0.015, d = 0.661). No significant difference was found between the control brain region and sham stimulation conditions (t(26.6) = −0.328, P = 0.754, d = 0.086). Most importantly, neither the main effect of stimulation order nor the interaction between stimulation order and stimulation condition reached significance (Ps > 0.05).
Tests on the INS difference across three stimulation conditions in each subgroup revealed that true stimulation on the third day induced significantly lower INS compared with the control brain region (Z = −1.867, P = 0.031) or sham stimulation (Z = −1.778, P = 0.038) on the other two days, but no significant difference was found between the control brain region and sham stimulations (Z = 0.533, P = 0.313, Fig. 3e, bottom). A similar pattern was found when true stimulation was applied on the second day (true vs. sham, Z = −1.511, P = 0.068; true vs. control, Z = −1.689, P = 0.046; Fig. 3e, middle). Although the differences between true and sham conditions and between true and control brain region conditions were not significant when applying true stimulation on the first day (true vs. sham, Z = −0.700, P = 0.270; true vs. control, Z = −0.980, P = 0.189), a similar tendency as mentioned before was also found (Fig. 3e, top).
3.3 The change in INS during communication could not be fully explained by the change in INS of spontaneous brain activity at rest
The results showed that true stimulation condition significantly decreased the similarity of INS between pre- and post-stimulation rests compared with the sham or control brain region stimulation condition within 0.03–0.04 Hz when the brain signal of the men lagged behind that of the women by 6–8 s. However, no significant difference was found between sham and control brain region stimulation conditions within this frequency range at these time lags (Fig. 4a and b). Statistical tests on the averaged INS across time lags (6–8 s) and frequency bands showed the same results (true vs. sham: t(29) = −2.961, P = 0.003, d = 0.541; true vs. control: t(29) = −2.520, P = 0.009, d = 0.460; control vs. sham, t(29) = 0.436, P = 0.667, d = 0.080).
Fig. 4Changes in INS during the resting state. (a) The INS difference between true and sham (top), between true and control brain region (middle), and between control brain region and sham (bottom) conditions. The black rectangle indicates the time lags and frequency bands showing significant differences between true and sham and true and control brain region stimulations but not between sham and control brain region stimulations. (b) The null distribution of the t values generated by permutation when the signal of the men lagged behind that of the women by 6 s. Left, the comparison between true and sham conditions; right, the comparison between true and control brain region conditions. A t value of zero indicates that no clusters have been found in a given random sample. The red lines indicate t values within 0.03–0.04 Hz (averaged) in the original pairs, and gray areas indicate lower 5% areas. (c) True stimulation-induced INS change at ATLwomen→TPJmen during post-stimulation rest could predict a minor part of the INS change at ATLwomen→SMCmen during communication. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
Next, the linear regression modeling analysis (Fig. 2b, Step 5) revealed that 12% of the variance in the pure change in INS during communication could be explained by the pure change in INS during post-stimulation rest with the ATLwomen→TPJmen making a significant positive contribution (F(1, 28) = 5.020, P = 0.033, Fig. 4c). Thus, a large proportion of variances were unexplained in the pure INS change during communication, suggesting that the pure INS changes during communication were mainly specific to interaction behaviors and mental processes rather than spontaneous brain activity.
3.4 True brain stimulation induced a significant change in the emotional empathy process but not in interaction behaviors
First, no significant difference was found among stimulation conditions for either verbal or nonverbal behaviors (Ps > 0.05).
Second, the analysis on empathy showed a significant effect of stimulation condition (F(2, 145) = 6.893, P = 0.001) for emotional empathy. Further pairwise comparisons showed that true stimulation significantly decreased the level of emotional empathy compared with sham (mean difference [MD] = −0.611, standard error [SE] = 0.172, P = 0.001, d = 0.628) and control brain region stimulations (MD = −0.467, SE = 0.172, P = 0.007, d = 0.480). No significant difference was found between the control brain region and sham stimulations (MD = −0.144, SE = 0.172, P = 0.403, d = 0.148) (Fig. 5a). No stimulation-related effects were found for cognitive empathy (Ps > 0.05).
Fig. 5Behavioral and cognitive changes. (a) True stimulation significantly decreased emotional empathy. (b) True stimulation-induced INS change between ATLwomen→SMCmen influenced emotional empathy through nonverbal behaviors.
Finally, to assess whether true stimulation changed the strength of romantic love, the same procedures were applied to the TLS scores. However, no significant results were found (Ps > 0.05).
3.5 The relationship among INS, interaction behaviors and emotional empathy
The results of mediation analyses showed a complete mediation effect for nonverbal behaviors, i.e., a change in INS significantly decreased the level of emotional empathy through nonverbal behaviors (lower limit confidence interval = 0.01, upper limit confidence interval = 2.66, Fig. 5b) but not through verbal behaviors.
4. Discussion
The major purpose of this study was to assess the causal relationship between INS and the shared sensorimotor input and mental processes within a dyad by applying tDCS to women's rATL prior to naturalistic communication. Based on previous studies, we hypothesize that INS is either directly associated with internal mental processes (Hypothesis 1) or indirectly associated through sensorimotor inputs (Hypotheses 2 and 3). Specifically, previous studies have revealed that INS associated with mental processes is dissociable from that associated with sensorimotor input regarding both temporal processes and spatial location [
]. For example, INS of the primary auditory cortex appeared immediately, but that of the TPJ appeared 4–6 s later, after the sound reached the ears of the listener [
]. These findings, however, cannot exclude the possibility that INS of the TPJ relates to semantic access through sensorimotor processes, such as speech perception. Our present results showed that true stimulation of the rATL in women significantly decreased the INS of ATLwomen→SMCmen as well as the level of emotional empathy but not interaction behaviors. Moreover, the change in INS at the ATLwomen→SMCmen significantly decreased the level of emotional empathy through nonverbal behaviors, although the nonverbal behaviors themselves were not changed. These findings supported the third hypothesis, suggesting that INS is associated with the internal mental process in an indirect manner, i.e., through sensorimotor input.
Specifically, first, we found that applying tDCS to the rATL in women decreased but did not increase the INS of the ATLwomen→SMCmen. To date, only a few studies have applied non-invasive brain stimulation to specific brain regions of interacting individuals in a “dual-brain stimulation” protocol to examine the relationship between INS and interaction behaviors. However, two of these studies showed that in-phase stimulation facilitated behavior synchrony [
], whereas the third study showed that both the same-phase-same-frequency and the different-phase-different-frequency stimulations decreased the synchrony of dyadic drumming relative to sham stimulation [
]. Thus, it is difficult to draw a robust conclusion from these studies. According to a recent perspective, INS is associated with shared representation of either external physical stimuli or the internal mental process [
]. Therefore, stimulation of one part of the interaction would disrupt the shared representation and result in a decrease in INS; on the contrary, delivering the same pattern of stimulation to both parts of the interaction might increase INS as well as behavioral synchrony. This conclusion is consistent with both the prior and present findings. Furthermore, we found that the change in INS during post-stimulation rest could only explain a minor portion of the variance in the change in INS during naturalistic communication, indicating that the tDCS-induced INS change was more related to the neural response to shared representation rather than to spontaneous brain activity. Therefore, these findings support the shared representation theory [
Second, we found that tDCS-induced INS change at the ATLwomen→SMCmen was associated with the change in empathy. To the best of our knowledge, only two studies have examined the relationship between INS and empathy. One study used EEG and found that when men in romantic couples were holding the hands of their female partner, INS increased during pain administration, and the increase in INS was correlated with the empathic accuracy of men [
]. However, one limitation of these two studies is that they cannot causally confirm the relationship between the changes in INS and empathy. Here, we directly showed that, compared with sham stimulation and stimulation of other brain regions, stimulation of the rATL in women within romantic couples specifically decreased the INS of the ATLwomen→SMCmen and empathy, which reliably confirmed the association between the shared representation of empathy and INS of the ATLwomen→SMCmen. Specifically, the ATL plays a key role in representing and retrieving social knowledge [
], and the SMC plays an important role in perceiving the state of others and subsequently activating the subject's corresponding representations during empathy [
]. Thus, the decline in empathy and INS might reflect a destruction of shared representations of empathy between partners due to brain stimulation.
Third, INS that was changed by tDCS showed a time-lag pattern, i.e., it appeared only when the brain signal of the men lagged behind that of the women by 2–6 s. According to predictive coding theory [
], in a hierarchical structure of the human brain, the higher level structure constantly produces predictions about the input of the lower level structure. Moreover, the lower level will calculate the prediction error (PE), i.e., the difference between the prediction and upcoming input, and back-project the PE to the higher level. Moreover, the prediction of the higher level may not be updated instantaneously when the PE is projected; in contrast, it may be updated only when sufficient information has been accumulated [
]. The sparse updating process might result in a time lag in INS between the brain signal of the higher level and that of the lower level. The present findings seem to support this hypothesis, suggesting that stimulating the rATL will affect its hierarchical relationship with the sensorimotor cortices in a time lag pattern, and the back-projection from the sensorimotor cortices will further impact the function of the rATL, showing a mediation effect of interaction behaviors.
Fourth, the influence of INS on empathy is mediated through nonverbal but not verbal behaviors. Basically, we express empathy in two different modes, i.e., verbal and nonverbal. A recent study showed that verbal and nonverbal languages are two different systems for understanding others. That is, while explicit verbal theory of mind (ToM) reasoning is supported by the precuneus and TPJ, nonverbal implicit ToM reasoning is additionally supported by the supramarginal gyrus [
] induced significantly higher INS than verbal communication for romantic couples but not strangers or friends. Thus, it seems that verbal behaviors are more associated with “cold” cognitive processes, whereas nonverbal behaviors are more associated with “hot” emotional states. Our results provide additional support for this speculation.
Finally, several limitations in the present study should be noted. First, as only the rATL of women was stimulated, it remains unclear whether and how INS and mental processes would change when stimulation is applied to the TPJ of men. Second, in this study, we only tested the effect of tDCS among romantic couples. Thus, future studies are needed to replicate or extend these findings to other social relationships, such as strangers. Third, although we employed several methods to remove systemic physiological noise in fNIRS signals, a better method is to use probes with different source-detector-separation distances [
In summary, this study showed that tDCS on the rATL of romantically involved women decreased INS of the ATLwomen→SMCmen and the level of dyadic emotional empathy. Moreover, the impact of INS on emotional empathy was mediated through nonverbal but not verbal behaviors. These findings suggest that INS relates to the cognitive process indirectly via the sensorimotor process, confirming the hierarchical organization of the human cerebral cortex. They also provide additional evidence for the idea of shared representation and interpersonal predictive coding during social interaction.
Funding
This work was supported by the National Natural Science Foundation of China (62293551, 61977008), China Postdoctoral Science Foundation (2021TQ0047) and Open Research Fund of the State Key Laboratory of Cognitive Neuroscience and Learning (CNLYB2101).
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
Maternal and paternal plasma, salivary, and urinary oxytocin and parent-infant synchrony: considering stress and affiliation components of human bonding.
Current status and issues regarding pre-processing of fNIRS neuroimaging data: an investigation of diverse signal filtering methods within a general linear model framework.
00257-1&id=gr1.jpg){kind=link}
00257-1&id=gr2.jpg){kind=link}
00257-1&id=gr3.jpg){kind=link}
00257-1&id=gr4.jpg){kind=link}
00257-1&id=gr5.jpg){kind=link}