Ten healthy naïve volunteers (seven males and three females; mean age ± standard deviation (SD) = 24.9 ± 1.3 years) participated in the study. The number of participants was determined by statistical power analysis (GPower 3.1, Faul et al., 2007). The effect size was estimated according to the two previous studies that demonstrated the effects of dual-hemisphere tDCS over the PO of stroke patients (Fujimoto et al., 2016) and over the S1 of healthy participants (mean d_z = 1.43). All participants were right hand dominant according to the Edinburgh Handedness Inventory (Oldfield, 1971), and none had a history of psychiatric or neurological illness. This study was carried out in accordance with the recommendations of the local ethics committee of Tokyo Bay Rehabilitation Hospital. All participants gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the local ethics committee of Tokyo Bay Rehabilitation Hospital.

We delivered direct current tDCS using a DC Stimulator Plus (NeuroConn, Germany) with two sponge surface electrodes (each with a surface area of 25 cm²). The stimulation intensity was 2 mA based on our previous study (Fujimoto et al., 2016). Two milliampere stimulation has been reported to minimize the intra- and inter-individual variability of the tDCS effect (Wiethoff et al., 2014; Laakso et al., 2015) and to induce consistent intra- and inter-individual increases of M1 excitability (Ammann et al., 2017). In all conditions except for the sham condition, we applied a direct current for 15 min. During stimulation, we gradually increased the current from 0 mA to 2 mA for the first 15 s and gradually decreased the current from 2 mA to 0 mA for the last 15 s. The current density at the stimulation electrodes was 0.08 mA/cm². These parameters are in accordance with a safety criterion and are far below the threshold for tissue damage (Nitsche et al., 2003; Poreisz et al., 2007). For the sham condition, we used the same procedure but applied current for only 15 s (Gandiga et al., 2006).

To identify the PO, we obtained T1 anatomical images for all participants using magnetic resonance imaging (Intera 1.5 T, Philips, Netherlands) prior to the tDCS experiment. For each participant, the centers of the stimulation electrodes were placed over the PO, identified via the individual T1 anatomical images. The target area was localized using a frameless stereotaxic navigation system (Brainsight2, Rogue Research Inc., Montreal, Canada). Our previous study confirmed that this procedure affected the SEP of the PO, but not the hand area of S1 (Nakagawa et al., 2017).

We adopted a double-blind, crossover, sham-controlled experimental design (Hummel et al., 2005; Gandiga et al., 2006). The experiment involved four types of interventions: two types of dual hemisphere tDCS, uni-hemisphere tDCS and sham stimulation. In one type of dual tDCS, the anode and cathode electrodes were placed over the left and right PO, respectively (Dual-Anodal-Left condition). In the other type of dual tDCS, these electrodes were placed over the opposite hemispheres (Dual-Anodal-Right condition). In uni-hemisphere tDCS, we administered anodal tDCS over the left PO and cathodal tDCS over the forehead above the contralateral orbit (Uni-Anodal-Left condition). This uni-hemisphere stimulation was used for comparison with Dual-Anodal-Left condition to examine the contribution of the ipsilateral PO when the right finger was stimulated. Each participant underwent four sessions, each of which involved one type of intervention and was performed between 9:00–18:00 on the same day. The tDCS sessions were separated by at least 3 days and the order of the sessions was counterbalanced among participants. In all sessions, we placed three electrodes in the same positions (two electrodes over the PO and the one electrode on the orbit) on all participants. The experimenter who applied the tDCS then discreetly connected the two active electrodes to the DC Stimulator. Thus, neither the experimenter who measured task performance nor the participant knew which stimulation type was applied during each trial.

To examine the effect of tDCS over the PO, we used the GOT (Van Boven and Johnson, 1994). The GOT is widely used as a measure of tactile spatial acuity (Sathian et al., 1997; Goldreich and Kanics, 2003; Ragert et al., 2008). During this task, participants sat blindfolded on a chair in a comfortable position. The tactile stimuli were five hemispherical plastic domes with grooves of different widths (0.5, 0.75, 1.0, 1.2 and 1.5 mm) cut into their surfaces (Tactile Acuity Grating, Miyuki Giken). We used a custom-made device to control the up-down movements of the domes, such that one dome was presented to the participant in each trial. A single skilled investigator tested all participants to minimize possible variance of stimulation.

Each session involved three task blocks, each of which was presented before the tDCS (Pre), during the tDCS (During), and 10 min after the tDCS (Post 10 min). Each task block in the dual-hemisphere tDCS and sham conditions consisted of 200 trials: 100 trials each for the left and right index fingers. The participants completed all trials for one finger first, followed by all trials with the other finger. The order of the fingers was counterbalanced. Each task block of the uni-hemisphere tDCS contained 100 trials for the right finger. For each finger, one of the five domes was successively presented during 20 trials. We presented the domes in the following order: 1.5 mm, 1.2 mm, 1.0 mm, 0.75 mm and 0.5 mm. Each dome was presented in each of two orientations 10 times and the order of the orientations was pseudo-randomized (20 trials for two orientations × 5 domes = 100 trials for each finger). In each trial, the groove of the dome was randomly oriented in one of two directions: orthogonal or parallel to the axis of the index finger. The dome was applied with moderate force onto the fingertip for 2 s. After the dome left the finger, participants were asked to verbally report whether the orientation of the grating of the presented dome was parallel or orthogonal in a two-alternative forced choice paradigm. Before the first session of the experiment, the participants were familiarized with the task.

To assess the participants’ subjective state during tDCS, we asked them to complete a questionnaire that measured their levels of attention, fatigue, pain, sleepiness and discomfort at the end of each session. The questionnaire had a four-point scale (e.g., attention [1 = no distraction; 4 = highest level of distraction]).

We calculated the groove width for which 75% of the responses were correct in each block. We defined this width as the discrimination threshold and used it as a primary outcome measurement. We calculated the threshold using a linear interpolation method, as follows (Ragert et al., 2008). Threshold = Gbelow + (0.75 − Pbelow) × (Gabove − Gbelow)(Pabove − Pbelow)

We then statistically evaluated the patterns of the thresholds using SPSS software (version 18; IBM Corporation, Armonk, NY, USA). We conducted two analyses. The purpose of the first analysis was to examine whether dual-hemisphere tDCS over the PO affected thresholds of healthy participants in a similar way to the stroke patients in our previous study (Fujimoto et al., 2016). In this analysis, we compared the effects of dual hemisphere tDCS with that of the sham condition. In the second analysis, we tested the main hypothesis, that dual-hemisphere tDCS would produce a stronger effect than uni-hemisphere tDCS. In both analyses, we initially conducted an analysis of variance (ANOVA), then conducted post hoc pairwise comparisons using Dunnett’s correction (two-tailed).

We estimated the EFs that were produced by dual- and single-hemisphere electrodes in the brain, as in our previous study (Laakso et al., 2016). First, the cortical EFs were numerically calculated in 62 individual magnetic-resonance-imaging (MRI)-based anatomical models using the finite-element method. The sizes and locations of the electrodes were similar to those in the actual experiment (Figure 3A). Other parameters of the computer simulations were identical to our previous study (Laakso et al., 2016). The individual EFs were then registered with each other and mapped to the standard brain template (Fonov et al., 2009, 2011) using FreeSurfer image analysis software (Dale et al., 1999; Fischl et al., 1999; Fischl and Dale, 2000) and the spherical demons algorithm (Yeo et al., 2010). This allowed us to determine the population-average EFs and their variability in standard brain space.

To compare the EFs in the PO and S1, regions of interest (ROIs) were defined as follows. We initially defined S1 and PO from the SPM anatomy toolbox (Eickhoff et al., 2005). We then limited the S1 to the hand area, in which Montreal Neurological Institute (MNI) Z coordinates ranged from Z = 45 to Z = 65. This range of Z coordinates was determined according to anatomical landmarks (i.e., inverse omega shape on horizontal sections) and previous findings (Kitada et al., 2005, 2006; Yang et al., 2017). In each ROI, we reported the peak and mean EFs, calculated as the absolute value of the EF. The peak EF was defined as the EF at the point with the maximum average EF, and the mean EF was defined as the EF averaged over each ROI. Paired t-tests (two-tailed) with Bonferroni-correction were used for statistical testing.

Figure 1 left shows the effect of dual-hemisphere tDCS over the PO compared with that of the sham stimulation. The obtained data are available in Supplementary Tables S1, S2.

We performed a three-way repeated measures ANOVA on grating thresholds, with the factors of Finger (right and left finger), Intervention (Dual-Anodal-Right, Dual-Anodal-Left, and sham) and Time (Pre, During and Post 10 min). This analysis revealed a significant three-way interaction (F_(4,36) = 18.42, P < 0.001, ηp2 = 0.67); significant two-way interactions of Finger × Intervention (F_(2,18) = 19.84, P < 0.001, ηp2 = 0.69) and of Finger × Time (F_(2,18) = 6.32, P = 0.008, ηp2 = 0.41); and a significant main effect of Time (F_(2,18) = 6.19, P = 0.009, ηp2 = 0.41). As we found a significant three-way interaction, we examined the effects of tDCS for each finger separately.

To test our main hypothesis, we then compared the effects of dual-hemisphere vs. uni-hemisphere tDCS. Figure 2 shows the threshold patterns. Two-way ANOVA (three levels of Intervention × three levels of Time) showed a significant interaction (F_(4,36) = 5.91, p = 0.001, ηp2 = 0.40), as well as significant main effects of Intervention (F_(2,18) = 4.33, p = 0.029, ηp2 = 0.32) and Time (F_(2,18) = 11.01, P = 0.001, ηp2 = 0.55). We conducted post hoc pairwise comparisons with Dunnett’s correction (two-tailed) to compare the dual-hemisphere tDCS and sham with uni-hemisphere tDCS condition. This analysis revealed no significant differences between the interventions before tDCS was applied (P values > 0.8) and after the end of tDCS (P-values > 0.3). Conversely, we found a significantly lower grating threshold in the Dual-Anode-Left condition compared with the Uni-Anode-Left condition during tDCS (P = 0.017, d_z = 0.93). The comparison between the Uni-Anode-Left and Sham conditions showed a trend towards a significant difference (P = 0.054, d_z = 0.61). No other significant differences were observed.

None of the participants reported side effects. Table 1 shows ratings of attention, fatigue, pain and discomfort reported by participants at the end of each intervention. A one-way ANOVA (four levels of Intervention) for each of the post-experimental ratings showed no significant differences (P-values > 0.1).

EF modeling revealed that the highest group-average EFs were located in regions in and around the lateral sulcus (Figures 3B,C). EFs in the PO and S1 hand area are shown in Figure 3D. As expected, EFs in the bilateral PO were symmetric in dual-hemisphere tDCS (P-values > 0.9), whereas the left PO showed stronger EFs than the right PO in uni-hemisphere tDCS (t₍₆₁₎ = 17.56, P < 0.001, d_z = 2.23 for peak EF; t₍₆₁₎ = 38.22, P < 0.001, d_z = 4.85 for mean EF).

We compared EFs in the PO with EFs in the S1 hand area. Both the peak and mean EFs were significantly higher in the left PO than in the hand area of the left S1 in dual-hemisphere tDCS (t₍₆₁₎ = 11.42, P < 0.001, d_z = 1.45 for peak EF; t₍₆₁₎ = 17.63, P < 0.001, d_z = 2.24 for mean EF) and in uni-hemisphere tDCS (t₍₆₁₎ = 9.24, P < 0.001, d_z = 1.17 for peak EF; t₍₆₁₎ = 22.08, P < 0.001, d_z = 2.80 for mean EF).

EF modeling indicated that both dual-hemisphere and uni-hemisphere tDCS stimulated the region in and around the PO. In S1, for instance, the face area is the region most adjacent to the PO, and thus could be affected by tDCS. However, the S1 hand area is approximately 3 cm away from the PO and the effects of stimulation over this area was weaker than for the PO. Indeed, the same tDCS protocol used in the present study has been found to affect SEPs in the PO but not in the S1 (Nakagawa et al., 2017). Thus, although we cannot rule out the contribution of the S1 hand area, its role is minor compared to the PO. Similarly, the inferior parietal lobule is located postero-superior to the PO. This region is involved in orientation discrimination and shows right hemisphere laterality (Kitada et al., 2006). If the effect of the tDCS is mainly from activity in this region, such laterality would be expected to be reflected in task performance. However, such an effect was not clearly observed, suggesting that the influence of tDCS on the inferior parietal lobule is smaller than that on the PO. Nevertheless, it is necessary to test the relative contributions of these adjacent areas in future work.

One of the potential issues is the small number of the participants (10 participants). This number was determined according to our previous studies that demonstrated the effect of the dual-hemisphere tDCS over the somatosensory cortices on tactile orientation discrimination in stroke patients (Fujimoto et al., 2016) and healthy participants (Fujimoto et al., 2014). Indeed, although these previous studies involved similar sample sizes, they successfully demonstrated the effect of dual-hemisphere tDCS on their behavioral performance. As we used double-blinded design, the effect in the present study cannot be explained by any observer effect. More importantly, we found the tDCS effect in the majority of the individual data (Figures 1C, 2B), as well as the group result. Thus, although it is important to replicate our findings in the future, it is unlikely that our result is biased by a small number of the participants.

Several important limitations should be considered when interpreting the current results. First, task performance was not significantly different in the left finger between sham and dual-hemisphere tDCS (with the anode on the left scalp). Thus, future work should investigate the possible asymmetric effects of dual-hemisphere tDCS. Second, we only compared the effects of dual-hemisphere tDCS with those of uni-hemisphere tDCS on the right finger, because of hand laterality. Indeed, most previous neuroimaging studies examining ipsilateral PO activation stimulated only the right hand (Roland et al., 1998; Karhu and Tesche, 1999; Kitada et al., 2005; Stilla and Sathian, 2008; Yang et al., 2017). However, confirming this result in future studies by testing the effects of uni-hemisphere tDCS on the left finger will be critical for clarifying this issue. Finally, the numbers of male and female participants in the experiment were not matched. Given the previously reported sex differences in tactile spatial acuity (Goldreich and Kanics, 2003), it may be useful to examine sex differences in the effects of tDCS over the PO.

Despite these shortcomings, the current findings may contribute to the development of therapeutic approaches for recovering somatosensory function following somatosensory cortex damage. Indeed, we demonstrated that the effects of tDCS can last for 10 min after the end of the stimulation, in accord with previous findings (Fujimoto et al., 2016). Thus, the results of the present study constitute an important step toward the clinical application of tDCS over the PO. In conclusion, we compared the effects of dual-hemisphere tDCS with those of conventional uni-hemisphere tDCS over the PO. We observed a greater improvement in the threshold for tactile grating orientation discrimination with dual-hemisphere vs. uni-hemisphere tDCS. This result indicates that inhibition of the ipsilateral PO enhances tactile orientation perception, supporting the hypothesis that the bilateral PO may function in competition, rather than in harmony.