The study cohort comprised 30 young adults, 15 men and 15 women, aged 21.37 ± 1.22 years (mean age ± standard deviation [SD]), who were recruited from Kio University where they were enrolled as students. All participants were right-handed according to the Edinburgh Handedness Inventory (Oldfield, 1971), and none had a previous diagnosis of a developmental disorder, or physical or mental disability. All participants were familiar with operating a personal computer.

The 30 participants were divided into three groups: the sham group (n = 10), or experimental groups, where tDCS was applied to the TPJ (n = 10) or IFC (n = 10). Each group was matched in terms of gender (with five men and five women in each group; χ² = 0.000, χ(0.95)2 = 5.991, p = 1.000) and age (TPJ: 21.20 ± 0.75 years; IFC: 21.50 ± 1.36 years; sham: 21.40 ± 1.43 years; p = 0.970).

The Ethics committee of the Graduate School and Faculty of Health Sciences at Kio University approved (approval number: H27-33) the experimental procedures. Participants provided background information and gave written informed consent. All of the participants read a tDCS information sheet and verified that they did not display any contraindications to tDCS. The procedures complied with the ethical standards of the 1964 Declaration of Helsinki regarding the treatment of human participants in research. There were no foreseeable risks to the participants, and no personally identifying information was collected.

After receiving tDCS, each participant performed two behavioral tasks (an imitation-inhibition task and visual perspective-taking task) and completed the AQ questionnaire. Each of the three tasks were performed in random order.

The tDCS was performed as previously described (Hogeveen et al., 2015). Briefly, tDCS was delivered through a pair of 35 cm² sponge electrodes, which were soaked in saline and connected to a neuroConn DC-stimulator Plus (neuroConn, Ilmenau, Germany). Stimulation sites for the tDCS protocol were identified using an EasyCap landmark cap (EasyCap, Herrsching, Germany), which were modified according to standard 10% landmarks. Previous studies primarily implicated a role for right-lateralized TPJ activity (Brass et al., 2005, 2009) and bilateral IFC activity (Brass et al., 2005; Cross et al., 2013) in the control of imitation. Further, Hogeveen et al. (2015) performed tDCS of the TPJ and IFC of the right hemisphere. Thus, anodal tDCS was applied to the right TPJ or IFC in the current study. The stimulation sites for the IFC and TPJ were FC6 (Holland et al., 2011; Hogeveen et al., 2015) and CP6 (Santiesteban et al., 2012a, 2015; Hogeveen et al., 2015), respectively. The reference electrode was placed horizontally over the vertex, individually measured, and then the vertex at 50% of the distance between the preauricular points, crossing a point 50% of the distance between the inion and nasion, was marked. For the TPJ stimulation, the anodal electrode was placed at CP6 with the cathodal electrode at the vertex. For the IFC stimulation, the anodal electrode was placed at FC6 with the cathodal electrode at the vertex. Both patterns of electrode placement were equally used for sham stimulation. The placement of these electrodes was the same as in previous studies (Santiesteban et al., 2012a, 2015; Hogeveen et al., 2015). For active tDCS, stimulation began with a 15-s ramp-up to 1 mA, proceeded to stimulation at 1 mA for 20 min, and ended with a 15-s ramp-down period. For sham stimulation, the same ramping procedure was accompanied by a 30 s stimulation period, yet participants were left in the room for the same total duration to mimic the experience of real stimulation without any neuromodulatory effect (Gandiga et al., 2006; Nitsche et al., 2008). During the stimulation period, participants were instructed to sit quietly with their eyes close and to think of nothing in particular, in order to minimize any attention to environmental stimuli (Damoiseaux et al., 2006; Tambini et al., 2010). Following tDCS stimulation, patients completed the imitation-inhibition task within 15 min, the perspective-taking task within 15 min, and the AQ within 10 min. The total time from the start of tDCS to the completion of all three tasks were less than 1 h. Previous studies using measures of corticospinal excitability have suggested that the neuromodulatory effects of 13 min of active tDCS are robust for 90 min post-stimulation (Nitsche and Paulus, 2001), suggesting that the current procedures were completed within the critical window.

The accuracy rate during congruent trials and incongruent trials, and the interference effect of ER were not significantly different between the groups (congruent trials, p = 0.126; incongruent trials, p = 0.961; ER, p = 0.961, Kruskal-Wallis test). In comparison to the accuracy rate for congruent trial, the accuracy of incongruent trials significantly decreased for all three groups (TPJ group, p = 0.017; IFC group, p = 0.017; sham group, p = 0.016; Figure 3A).

Based on a split-plot factorial design, an ANOVA analysis of the effect of between-group factors and between-trial factors on the RT revealed a main effect of between-trial factors (F_(1,27) = 111.936, p = 0.000, ηp2 = 0.806) and interaction effect (F_(2,27) = 7.019, p = 0.004, ηp2 = 0.342). A main effect of between-group factors was not noted (F_(2,27) = 1.416, p = 0.260, ηp2 = 0.095). A simple main effect analysis (with Bonferroni adjustment) revealed a significant increase in the RT for all three groups (p = 0.000 for all) during incongruent trials, in comparison to the RT during congruent trials. In addition, a simple main effect analysis (with Bonferroni adjustment) revealed no differences in the RT among groups during individual trials for congruent trials (TPJ vs. IFC, p = 0.548; TPJ vs. sham, p = 0.067; and IFC vs. sham, p = 0.902) and incongruent trials (TPJ vs. IFC, p = 0.332; TPJ vs. sham, p = 1.000; and IFC vs. sham, p = 0.190). A one-way ANOVA revealed significant differences in the interference effect of RT among groups (F_(2,29) = 7.019, p = 0.004). A multiple comparison test using Tukey’s method revealed that the interference effect of RT decreased significantly for the TPJ group and the IFC group, in comparison to the sham group (TPJ vs. sham, p = 0.027 and IFC vs. sham, p = 0.004). Significant differences in the interference effect of RT were not noted for the TPJ group and the IFC group (p = 0.693; Figure 3B).

The accuracy rate under control conditions and experimental conditions and the cost of perspective-taking (ER) were compared between groups using a Kruskal-Wallis test. There was no significant difference in the accuracy rate under control conditions and experimental conditions and the cost of perspective-taking (control conditions, p = 1.000; experimental conditions, p = 0.865; and the cost of perspective-taking, p = 0.865). The accuracy rate under control conditions and experimental conditions, which was compared (within groups) using a Wilcoxon signed-rank test, was not significantly different in the TPJ group or IFC group (p = 0.083 for both). However, a significant decrease in the accuracy rate under experimental conditions was noted in the sham group, when compared to the accuracy rate under control conditions (p = 0.046; Figure 4A).

The RT under control conditions and experimental conditions and the cost of perspective-taking (RT), which were compared between groups using the Kruskal-Wallis test, were not significantly different (control conditions, p = 0.966; experimental conditions, p = 0.282). However, significant differences in the cost of perspective-taking were noted among the three groups (p = 0.001). The Mann-Whitney U test with Bonferroni p-correction (p < 0.016) was performed as a post hoc test. Results revealed a significant decrease in the cost of perspective-taking in the TPJ group in comparison to the sham group (p = 0.000). A significant decrease in the cost of perspective-taking was also noted in the IFC group in comparison to the sham group (p = 0.005). Significant differences in the cost of perspective-taking were not noted for the TPJ group or IFC group (p = 0.579). The RT under control conditions and experimental conditions was compared (within groups) using the Wilcoxon signed-rank test, and revealed no significant differences in the TPJ and IFC groups (TPJ group, p = 0.114; IFC group, p = 0.093). However, a significant increase in the RT was noted in the sham group under experimental conditions, in comparison to control conditions (p = 0.005; Figure 4B).

The total AQ score and scores on the five AQ subscales, which were compared between groups using the Kruskal-Wallis test, was not significantly different (total AQ score, p = 0.527; social skills, p = 0.711; attention switching, p = 0.154; attention to detail, p = 0.633; communication, p = 0.963; imagination, p = 0.441; Figure 5).

The interference effect of RT decreased significantly in the TPJ group in comparison to the sham group, in line with several previous studies (Santiesteban et al., 2012a, 2015; Hogeveen et al., 2015; Sowden and Catmur, 2015). The current results support a role for the right TPJ in controlling self and other representations, i.e., it inhibits imitation of others and it enhances one’s ability to control oneself.

Further, the interference effect of RT decreased significantly in the IFC group in comparison to the sham group, which corroborates with a previous study (Hogeveen et al., 2015). Hogeveen et al. (2015) indicated that anodal tDCS of the right TPJ improved the inhibition of imitation, but it did not facilitate imitation. Further, anodal tDCS of the right IFC improved both inhibition of imitation and imitation. These finding indicate that the right TPJ is indirectly involved in controlling imitation, while the right IFC is directly involved in controlling imitation (Hogeveen et al., 2015). The results from the current study support a role for the right IFC in directly inhibiting imitation.

The cost of RT for perspective-taking decreased significantly in the TPJ group in comparison to the sham group. For the sham group, the accuracy rate significantly decreased under experimental conditions, while the RT significantly increased. However, there were no significant differences in the accuracy rate and the RT under experimental conditions in the TPJ group, which is similar to previous studies (Santiesteban et al., 2012a, 2015). There is ample evidence that the TPJ plays a role in social cognition, i.e., social moral judgment, hostile intention attribution, out-group punishment, false-belief and mentalizing (Donaldson et al., 2015). The results from the current study support the role of the right TPJ in inhibiting self-representation and enhancing representations of the other during perspective-taking.

A principal aim of the current study was to examine the effects of anodal tDCS of the right IFC on perspective-taking. The cost of RT for perspective-taking significantly decreased in the IFC group, in comparison to the sham group. For the sham group, the accuracy rate under experimental conditions significantly decreased, compared to that under control conditions. Further, the RT under experimental conditions significantly increased, compared to that under control conditions. However, there were no significant differences in the accuracy rate and the RT under experimental and control conditions in the IFC group.

Previous studies have noted increased brain activity in the right IFC during the visual perspective-taking task (Kaiser et al., 2008) and mental rotation of an object (Yassa et al., 2008; Hattemer et al., 2011; Semrud-Clikeman et al., 2012). Thus, the IFC group should have improved their ability to rotate an object mentally, so the group may be better able to perform a mental activity like taking the perspective of the director in the perspective-taking task.

Using fMRI, Mazzarella et al. (2013) demonstrated that the dorsomedialPFC was sensitive to the orientation of the actor in the altercentric task (i.e., object identification from an actor’s perspective), while the right IFC was sensitive to the orientation of the actor in the egocentric task (i.e., target object identification from their own perspective). Further, the dorsomedialPFC and the right IFC may play distinct but complementary roles in visual perspective-taking (Mazzarella et al., 2013). These findings suggested that anodal tDCS of the right IFC might facilitate awareness of the orientation of the director in the current study as well.

Hogeveen et al. (2015) indicated that anodal tDCS of the right IFC improves both imitation and inhibition of imitation. This finding indicates that the right IFC plays a role in inhibiting self-representation and enhancing representations of the other, and vice versa. Thus, the right IFC may facilitate inhibition of imitation, which requires enhancement of self-representation and inhibition of representations of the other, as well as perspective-taking, which requires inhibition of self-representations and enhancement of representations of the other. Although there may be other factors involved, the current study highlighted the role the right IFC plays in perspective-taking.

Further, both the TPJ and IFC are brain regions that play a role in being attentive to others in a social context (Kuang, 2016a). Perspective-taking requires an individual to direct attention away from oneself and towards others. Thus, stimulation of the TPJ and IFC could have improved attention to others.

The current study indicated that both the TPJ and IFC equally help to control different representations of self and others, i.e., inhibiting imitation of and taking the perspective of another. Kuang (2016b) states that instead of competing during action recognition during social situations, the mirror neuron (IFC) and mentalizing systems (TPJ) work in a synergistic and complementary manner to ensure appropriate social interactions in a given behavioral context. In other words, the dichotomy between mirroring and mentalizing processes suggests that the distinction between self and other operates at both the mental and physical level. Mirroring processes play a role in self-awareness and empathy, while mentalizing processes support bodily self-consciousness, i.e., a sense of body ownership and a sense of agency. The results from the current study support the theory that the self-other distinction is both mental (mirroring processes) and physical (mentalizing processes). The current results also indicate that either dysfunction of the IFC (the mirroring neuron system) or the TPJ (the mentalizing system) may lead to social-cognitive abnormalities, such as those seen in ASD.

The AQ reflects autism spectrum traits. Since dysfunction of the TPJ or IFC can cause ASD, we hypothesized that anodal tDCS of these brain regions would reduce an individual’s AQ score. However, tDCS did not affect the total AQ scores or any of the scores on its five subscales.

Previous studies, using MRI/fMRI, suggested that the activity of TPJ and IFC is related to the AQ score. A study on pervasive developmental disorders in adults reported that the AQ score and gray matter volume of the IFC are inversely correlated (Kosaka et al., 2010). In typically developing male adults, the resting-state functional connectivity of the default mode network (which includes the TPJ) and the AQ score is inversely correlated (Jung et al., 2014, 2015). Thus, results from the current study conflict with these previous studies. However, this discrepancy could be attributed to the difference in the methodology (i.e., tDCS vs. fMRI).

The AQ score and performance on a ToM task are significantly correlated (Carroll and Chiew, 2006). However, a study using fMRI indicated that the TPJ is part of the core network for ToM (Schurz et al., 2014), while Santiesteban et al. (2012a), found that anodal tDCS of the right TPJ produced no changes in the RT on a self-referential task, which included components of a ToM task. Further, anodal tDCS of the right and left TPJ produced no changes in accuracy on a ToM task (Santiesteban et al., 2015). Santiesteban et al. (2015) posited that a ToM task is insensitive to performance variation induced by stimulation in typical development adults, but that clinical populations would exhibit marked deficits in ToM. The AQ score is similarly less susceptible to variability due to transcranial stimulation. Although the imitation-inhibition task and perspective-taking task measure on-line social-cognitive processing, the AQ score reflects metacognition and memory based on previous social interactions. Therefore, the AQ is a “self” assessment of one’s social cognitive or social behavioral traits. Thus, the results in the current study suggested that neuromodulation is only effective on the functioning of single brain regions, with no effect on metacognition or memory about one’s mental state. However, this is purely speculative, and the limitations outlined in the next section must also be taken into account. The current study was unable to ascertain the relationship between the TPJ and IFC, and the AQ score, and thus, further work is required to ascertain these associations.

The current study had a number of limitations. The placement of tDCS electrodes could not be adjusted enough. In addition, control tasks, such as those not involving inhibition of imitation, were not performed to assess the duration of the modulatory effect of tDCS, which may explain why it did not appear to affect the AQ score. Although each participant performed the three tasks after tDCS in a random order, the performance on the imitation-inhibition task and the perspective-taking task significantly improved in the TPJ and IFC stimulation groups. Thus, although these technical limitations presumably had little effect, these issues need to be resolved in future studies.

In addition, the current study did not use an intra-subject design. In other words, the three types of stimulation, i.e., TPJ, IFC and sham stimulations, were not counterbalanced across subjects, which might have had better explanatory power and less noise. Nevertheless, the current study was designed as a comparison of unpaired groups in order to eliminate the effects of learning due to repetition of tasks. In fact, previous studies have used a similar design for the same reason (Santiesteban et al., 2012a, 2015; Hogeveen et al., 2015). Eliminating the effects of learning isolated the simple modulatory effect of tDCS.

The current study involved an extremely small sample of 10 subjects per group (5 men and 5 women in each). Thus, the modulatory effect of tDCS may not have been sufficient to affect the AQ score. ASD is more common in men than women (Kirkovski et al., 2013; Ruigrok et al., 2014; Van Wijngaarden-Cremers et al., 2014), and men generally have higher AQ scores in the general population and the highest individual ASD scores of all participants (Baron-Cohen et al., 2001). Further, although men showed inverse correlation between the resting-state functional connectivity of the default mode network and AQ score, this correlation was not present in women (Jung et al., 2015). Thus, in future studies, the sample size needs to be substantially increased, so that the effects tDCS of the TPJ and IFC have on the AQ score can be separately examined in men and women.

Since the participants in the current study were typically developing adults without ASD, these results are not reproducible in individuals with ASD or in different age ranges. Thus, future studies need to involve participants who are typically developing, have ASD and belong to different age groups.

In addition, the current study merely verified the immediate effects of tDCS. Studies in the area of rehabilitation have indicated that continuous tDCS improved motor function (Costa-Ribeiro et al., 2016; Ilić et al., 2016; Yozbatiran et al., 2016; Figlewski et al., 2017). Thus, continuous tDCS is likely to affect an individual’s AQ score.

Finally, the AQ measure has five subscales. Currently, the relationship between individual scores of AQ and brain region/neural network is unknown. Thus, brain regions and neural networks that correlate with scores on the subscales of the AQ need to be examined using neuroimaging techniques such as fMRI and electroencephalography. Those findings would presumably lead to the development of treatment strategies for ASD as a whole, as well as strategies targeting individual symptoms of ASD.