Transcranial direct current stimulation of the occipitotemporal cortex enhances identification of ambiguous faces Authors: Seyed Javad Saghravanian, Hossein Esteky Published: 2025 ABSTRACT AbstractTranscranial direct current stimulation (tDCS) has emerged as a promising tool for investigating the neural mechanisms underlying various brain functions and for enhancing related behavioral performance. In this sham-controlled, between-subjects study, we examined the effects of tDCS applied to the right and left occipitotemporal cortex (OTC) on face identification performance using a delayed match-to-sample task with morphed face images of varying discrimination difficulty. Three active montages were tested: (M1) anode on the right OTC and cathode on the left OTC, (M2) anode on the left OTC and cathode on the right OTC, and (M3) anode on the right OTC with the cathode on the left shoulder. A total of 47 participants were randomly assigned to one of four groups: M1 (n = 13), M2 (n = 12), M3 (n = 13), and Sham (n = 9). The results showed a significant improvement in face discrimination performance with the M1 montage relative to the sham condition. Although performance also improved in M2 and M3, these effects were not significantly different from the sham group. Across all montages, the largest impact on non-match discrimination was observed under more ambiguous conditions, particularly at a 30% morph distance, which corresponded to the point of subjective equality. Furthermore, tDCS stimulation led to longer reaction times across all montages, with the significant delays noted in M1 and M3. Findings of this preliminary study suggest a hemispheric laterality bias in face identity discrimination following tDCS-induced neural modulation.Keywords: Face recognition, TDCS, Delayed match to sample, Occipito-temporal cortexSubject terms: Neuroscience, Cognitive neuroscience Introduction Face recognition is central to social cognition and interaction, engaging multiple regions within the occipito-temporal cortex (OTC)—notably the fusiform face area (FFA), the occipital face area (OFA), and the superior temporal sulcus (STS)1–3. Each area specializes in processing different facets of facial information: the OFA encodes low-level facial features4,5, the FFA represents social attributes such as gender5,6, and the STS processes more dynamic cues, including facial expressions7–9. Evidence from invasive electrical stimulation studies using electrocorticography (ECoG) in humans10–12 and depth electrode recordings in non-human primates13,14 has demonstrated a causal relationship between activity in these cortical face areas and face perception. Non-invasive transcranial current stimulation (tCS) over the OTC has likewise been applied to study both face15–20 and object16,17,19,21 recognition. One such neuromodulatory technique is transcranial direct current stimulation (tDCS), which typically delivers a low-intensity (1–2 mA) current between two scalp electrodes—a cathode and an anode—placed to target specific cortical regions. While some studies attribute subthreshold depolarization primarily to the anode and hyperpolarization to the cathode22–25, others suggest that both electrodes may induce depolarization26,27, making the precise excitatory or inhibitory nature of tDCS uncertain. The acute effects of tDCS likely stem from alterations in subthreshold membrane potentials, while longer-lasting after-effects are believed to involve calcium-dependent plasticity mediated by glutamatergic and GABAergic mechanisms28–31. Within this framework, multiple studies have employed tDCS over the OTC to investigate face19,20 and facial expression recognition (Willis et al., 2019), face memory16,32, and holistic face processing15. For instance, Yang and colleagues (2014) applied tDCS bilaterally to the right and left OTC, aiming to excite one hemisphere with anodal stimulation while inhibiting the other with cathodal stimulation. They observed a reduction in holistic face processing, as measured by the composite face effect, following stimulation to both hemispheres. In contrast, Barbieri et al. (2016) found that delivering anodal tDCS before (but not during) a cognitive test enhanced the perception and memory of both faces and objects. Willis and colleagues (2019), however, reported no significant effects on face perception when attempting to replicate Barbieri’s protocol. Additionally, anodal tDCS applied to the right STS has been shown to improve recognition of emotional facial expressions33,34, and focal stimulation of the FFA can facilitate facial feature recognition and discrimination18,35, whereas stimulation to the OFA does not appear to produce the same effect35 aside from impairing Mooney-face detection36. Although both hemispheres house important neural populations for face processing, numerous studies emphasize right-hemisphere superiority in face recognition37–41. This phenomenon has often been linked to a left-visual-field bias associated with contralateral face-selective areas, particularly in the right fusiform face area (FFA)42. Furthermore, Verosky & Turk-Browne (2012) provided fMRI evidence suggesting that while the left FFA plays a primary role in processing facial identity, the right FFA provides additional support43. Behavioral and electrophysiological studies using chimeric or composite face stimuli also typically demonstrate right-hemisphere dominance44,45. Nonetheless, not all individuals show this pattern; some fMRI findings point to bilateral or even left-hemisphere dominance46,47, and ECoG data indicate that the right hemisphere may be more susceptible to distortions in face perception when stimulated10,11,48. Similar variability in hemispheric specialization exists in lower primates, suggesting that right-hemisphere dominance in humans, although common, is not absolute49. Because tDCS outcomes depend heavily on the electrode montage, current direction, and intensity, electrode placement relative to the neuronal orientation is critical50,51. Current traveling from dendrites to soma typically increases excitability, whereas the reverse direction often reduces it, and perpendicular current flow seems to have negligible effects. Such considerations apply broadly to tDCS, whether it is targeting the motor cortex to improve motor learning25,52, frontal regions to influence decision-making or working memory53,54, or the OTC to modulate face recognition. The stimulation duration and intensity further shape both immediate and longer-term tDCS after-effects55, with certain protocols extending benefits for up to 90 min post-stimulation52,56. Notably, suboptimal choices in these parameters can invert or diminish the expected outcomes57–59. In light of these complexities, results from anodal (excitatory) and cathodal (inhibitory) tDCS in cognitive tasks remain inconsistent, likely reflecting differences in stimulation design, individual brain anatomy, and the specific neural circuits engaged60. This underscores the importance of refining tDCS protocols to maximize its research and therapeutic potential. Sensory detection and perceptual performance are closely linked to the baseline excitability of sensory neurons. When baseline neural activity is elevated, the sensory system becomes more responsive, facilitating improved detection of incoming stimuli61–64. Modulating cortical excitability by shifting the resting membrane potential of neurons toward depolarization with tDCS65 may enhance sensory responsiveness. This heightened neural sensitivity becomes particularly relevant when processing ambiguous or degraded stimuli, such as perceptually bistable images or signals embedded in noise62,63. Notably, baseline activity in sensory regions can be influenced by multiple factors, including slow intrinsic oscillations66,67, contextual cues64, and top-down attentional modulation68, suggesting a dynamic interplay between internal brain states and external sensory processing. In the present study, we evaluated how tDCS applied over the right or left OTC, a region central to sensory information processing, influences face identification performance in a Delayed Match-to-Sample (DMS) task using morphed face images with varying degrees of ambiguity. We hypothesized that the impact of tDCS would be shaped by both hemisphere and task ambiguity, with minimal effects expected in easy conditions with large shape differences that yield high performance. Based on our previous electrical stimulation studies in non-human primates13, we expected larger effects in ambiguous conditions, particularly around the point of subjective equality (PSE). Our findings showed that tDCS indeed enhanced performance in ambiguous conditions at and near PSE and was associated with increased response times. Materials and methods Participants Forty-seven healthy participants (mean age 27 ± 6 SD) were randomly assigned to one of four groups: three real stimulation groups and one sham group. All participants had normal or corrected-to-normal vision. Each participant underwent a single 1.5-h session and was randomly assigned to either the right-anode, left-anode, or sham stimulation groups. Recruitment and informed consent Before participation, individuals received a detailed explanation of the experiment and registered online if they consented to participate. On the day of the experiment, participants signed a consent form before starting the procedure. Ethics and dissemination This study followed protocols approved by the Ethics Committee of the Pasargad Institute for Advanced Innovative Solutions (code number: ETH-2023–12-1) and was in accordance with the Declaration of Helsinki. All experimental procedures were conducted at the National Brain Mapping Lab (NBML), under project number P-402–136. Transcranial direct current stimulation (tDCS) Transcranial direct current stimulation (tDCS) was administered using a battery-driven current stimulator (Neuroelectrics®, Barcelona, Spain) with a pair of saline-soaked circular sponge electrodes (5 cm in diameter). A constant current of 1.5 mA (current density: 0.060 mA/cm2) was applied for 20 min, with ramp-up and ramp-down phases of 30 s each. These parameters were selected based on previous studies demonstrating the efficacy and tolerability of tDCS in modulating activity in visual and face-selective regions of the OTC15–17. In the sham condition, the stimulation was delivered only during the ramp-up and ramp-down stages to mimic the sensation without causing neurophysiological changes. Three electrode montages were used in this study. In the first two montages, the anode electrode was placed on the right (M1) or left (M2) occipito-temporal cortex (OTC; P8 or P7 according to the EEG 10–20 system), with the cathode electrode positioned over the contralateral OTC. The third montage (M3) targeted the right OTC with the anode electrode, and the cathode electrode was placed over the left shoulder as an extracerebral site. During the 20-min offline stimulation period, participants were seated quietly, facing a black screen. They were instructed to keep their eyes open, remain calm and relaxed, and refrain from engaging in any task, conversation, or exposure to visual or auditory stimuli. The study used a single-blind design in which participants were unaware of their assigned stimulation group. Importantly, they were not informed about the presence of a sham group. Following each stimulation session (whether real or sham), participants were asked to report any sensations they experienced during the stimulation. Participants reported typical tDCS-related sensations, such as mild tingling, itching, or warmth, and none expressed any doubt about receiving active stimulation. While these reports were qualitatively similar across sham and active groups, we did not collect detailed quantitative data on the frequency or intensity of sensations. Therefore, we cannot statistically confirm successful blinding. Task and stimuli In each session, first, the subjects received a 20-min tDCS or sham stimulation. Then, they performed a delayed face identification match-to-sample task. We used a modified version of the face stimuli originally developed by our laboratory69 (Fig. 1a). The original seed stimuli consisted of 3 families, each with 3 individual members (9 faces). These original face images were created using FaceGen (Singular Inversions Inc.). Here, we selected all potential pairs of faces across the three face families, totaling 27 face stimulus pairs. Then, morph face images were generated from all of the selected face pairs using FantaMorph 5 Deluxe (Abrosoft Co.) software. For each pair of stimuli, multiple transitional morph images were created with 5% steps. This procedure generated 19 face images with various levels of contribution from each of the two seed images, ranging from 5 to 95% of the seed faces. Using this procedure, we created a stimulus set with 522 stimuli. Experimental Design. (a) The nine seed images used to create morphed face stimuli. Adapted from Salehi et al. (2017) 69. (b) In non-match trials, the test stimulus was a morphed image derived from two seed faces, presented with a 20–60% difference relative to the sample image. (c) Overview of a single experimental session. (d) Schematic of the Delayed Match-to-Sample (DMS) task. A fixation cross is displayed for 1000 ms to cue the upcoming stimulus. Next, the sample stimulus appears for 400 ms. Following a 2000 ms blank interval, the test stimulus is presented for another 400 ms. Participants indicate whether the test stimulus matches the sample by pressing designated keys on the keyboard. The task was controlled using a Lenovo laptop running MATLAB-based Psychophysics Toolbox (Psychtoolbox-3) and was presented on a 21-inch LED monitor (60 Hz, resolution 1708 × 960) positioned 60 cm in front of the participant. A total of 200 trials were presented, with half being match trials and the other half non-match trials. Non-match trials included five difficulty levels, with differences between sample and test stimuli ranging from 20 to 60% at 10% intervals (Fig. 1b). First, subjects performed 10 training trials to familiarize themselves with the task instructions before the main task. The delayed match-to-sample (DMS) task began with a 1000-ms cross-stimulus at the center of the monitor to indicate the onset of a new trial and cue the upcoming stimulus. Subjects were instructed to maintain fixation at the cross-stimulus and its position during the whole trial. The sample face stimulus, 4.5 degrees in size, was displayed at the fixation point for 400 ms, followed by a 2000 ms delay period with a blank screen. The test face stimulus, identical in size to the sample, was then presented for 400 ms. Participants reported whether the test stimulus matched (match trials) or differed (non-match trials) from the sample stimulus by pressing the right or left key on the keyboard (Fig. 1d). The response right and left keys were assigned randomly to the match/non-match response conditions in each session. Data analysis Data analysis was performed using custom MATLAB scripts (MathWorks Inc.). A two-way ANOVA with factors Group (e.g., M1 vs. Sham) and Condition (morph levels) was conducted to compare the non-match recognition percentages between stimulation groups. The ANOVA results reported in the text reflect the main (Group and Condition) and interaction (Group × Condition) effects. The non-match recognition percentage represents the rate at which subjects perceived differences between the sample and test stimuli under each task condition. Post hoc pairwise comparisons with Šidák correction were conducted to compare the mean non-match recognition percentages between groups at each morph level. To determine the point of subjective equality (PSE), a logistic function was fitted to the sham data using the Palamedes toolbox70. MATLAB’s built-in functions were used to compute the correlation coefficient (corr) and the slope of the regression line (polyfit). To compute the correlation between reaction time and recognition accuracy, each participant contributed six data points corresponding to the six task conditions. These individual data points were pooled within each group, resulting in 54, 78, 72, and 78 data points for the sham, M1, M2, and M3 groups, respectively. Subsequently, Z-scores were calculated to statistically compare the regression line slopes of each stimulation group against the sham group. We conducted a sensitivity power analysis using G*Power71 to determine the minimum detectable effect size given our actual sample sizes (M1: n = 13, M2: n = 12, M3: n = 13, Sham: n = 9). For independent-samples comparisons with α = 0.05 and power = 0.80, the smallest detectable standardized mean differences (Cohen’s d) ranged from 1.15 to 1.30 across the pairwise groups. Correspondingly, the achieved power with the current sample sizes was very low for small effects (d = 0.2; ~ 7–8%) and moderate effects (d = 0.5; ~ 19–23%), modest for large effects (d = 0.8; ~ 42–50%), and only adequate (~ 80%) for very large effects (d ≥ 1.2). These results indicate that the present study was sufficiently powered to detect only very large effects of tDCS, whereas small-to-moderate effects (d ≈ 0.3–0.6) would require substantially larger sample sizes. Results To investigate how stimulation of the occipitotemporal cortex (OTC) affects face identity recognition in different levels of ambiguous conditions, we assessed performance across three active transcranial direct current stimulation (tDCS) montages (M1–M3) and a sham condition. Forty-seven healthy volunteers performed a Delayed Match-to-Sample (DMS) task for identification of face images following 20 min of one of the tDCS or sham conditions (Fig. 1). To assess the psychometric function of subjects’ face identity recognition performance, we performed several preliminary experiments and asked subjects (n = 20) to perform the DMS task (data not shown). Using the results of these preliminary experiments, we selected morphed face distances that best covered the performance spectrum. Morphed face stimuli with 20, 30, 40, 50, and 60 percent distances yielded performances from about 35 to 90% with the 30% morphed face distance yielding about 50% correct responses in both the preliminary and main experiments that are reported here. Note that for each of the morphed face stimuli, distances between the sample and match pairs of faces were selected randomly from stimuli in various positions in the face morph continuum. tDCS effects on task performance In all of the tested conditions, subjects performed well in the conditions where the sample and match were the same stimulus, regardless of the tDCS condition (matched trials; 78.5 ± 1.8%). Also, faces with 50 and 60% morph distances were easily detectable by subjects, yielding high face recognition performances (50%; 85.7 ± 2.3%, 60%; 89.5 ± 2.3%). No significant differences were observed between the sham and stimulation groups at these easy levels (p > 0.05, Šidák-corrected). In fact, the point of subjective equality (PSE), which indicates chance performance, was calculated to be at 26% face morph distance in the pooled data of the sham trials (Fig. 2). Importantly, when present, tDCS effects on face recognition performance were observed only for the difficult face identity recognition conditions, particularly for the 30% and 40% face morph distances (Fig. 3). Non-Match Recognition Percentage at the Point of Subjective Equality (PSE). (a) Psychometric curves fitted to the sham group’s non-match recognition data using a logistic function. The black lines highlight the 50% level of non-match recognition (y-axis) and the corresponding sample-to-test difference (x-axis), identified by the psychometric curve (red dashed line) as 26%. (b) Non-match recognition percentages for individual participants in match (0%) and non-match (20–60%) conditions, shown in different colors. DMS Task Results for M1, M2, and M3 Montages. The average non-match recognition percentage for each montage group—(a) M1 (green line, n = 13), (b) M2 (cyan line, n = 12), and (c) M3 (magenta line, n = 13)—is plotted alongside the sham group (red dashed line, n = 9). (d–f) Non-match recognition percentages for individual subjects (match: 0%; non-match: 20–60%) are plotted in different colors, with the black dashed line indicating the mean non-match recognition in the sham condition for each montage. tDCS effects on subjects’ performance were observed for all of the tested montages with various degrees of effectiveness. The largest effects were observed when anodal electrodes were placed over the right OTC (M1 condition). Two-way ANOVA revealed a statistically significant main effect of tDCS compared to the sham condition in M1, but not M2 montages (M1: F(1,120) = 4.7, p < 0.05; M2: F(1,114) = 1.59, p = 0.2). The Group × Condition interaction for the M1 group compared to the sham shows only a trend toward significance (Table 1; M1: F(5,120) = 2.26, p = 0.053; M2: F(5,114) = 1.48, p = 0.2), suggesting that group differences may vary across task difficulty levels. Given our a priori hypothesis that neuromodulation effects could emerge preferentially at intermediate morph levels (near the point of subjective equality), we conducted post hoc pairwise comparisons for each morph level using the Šidák correction. These analyses revealed a significant improvement in recognition performance for the M1 group compared to Sham at the 30% morph distance (p < 0.01, Šidák-corrected; Fig. 4). At this morph level, the mean recognition performance was 50.55 ± 3.76% for the Sham group (red bar), 66.53 ± 4.17% for the M1 group (green bar), and 62.10 ± 3.66% for the M2 group (cyan bar). The difference between M2 and Sham did not reach significance (p = 0.09, Šidák-corrected). Table 1Non-match recognition percentage ANOVA results.MontagesF-ANOVA (main)P-ANOVA (main)F-ANOVA (interaction)P-ANOVA (interaction)M1/Sham4.74340.03142.260.053M2/Sham1.59380.20941.480.2016M3/Sham2.46070.11941.480.2024M1/M20.68560.40910.1550.9784M1/M30.15970.69001.650.1503M2/M30.14460.70430.9940.4240This table shows the main (Group and Condition) and interaction (Group × Condition) two-way ANOVA analysis results for each pair of montage arrangements. Significance is indicated in bold digits. Non-match recognition percentage ANOVA results. This table shows the main (Group and Condition) and interaction (Group × Condition) two-way ANOVA analysis results for each pair of montage arrangements. Significance is indicated in bold digits. Non-Match Recognition Percentage at PSE. A bar plot showing the non-match recognition percentage for each group at the sample-to-test difference closest to the sham group’s PSE (Fig. 2), determined to be 30%. Colors match those in Fig. 3. Asterisks (**) denote statistically significant differences (p < 0.01, Šidák-corrected) between the anodal stimulation groups and the sham group. Cathodal stimulation has been shown to suppress52 or excite26 the neural structure close to the stimulation area. So, it is plausible that the impact of M1 tDCS montage on face identity recognition is due to the rebalancing of the activity of both right and left OTC in the M1 condition, confounding the observed results depicted in Fig. 3a. To address this issue and investigate the impact of right OTC stimulation alone, we conducted the DMS experiment on another population using the M3 montage. In this montage, the anode electrode was placed over the right OTC, while the cathode electrode was positioned on the left shoulder as an extracerebral site. Anodal stimulation in the M3 montage did not produce a statistically significant enhancement in face identity recognition performance compared to the Sham group. Two-way ANOVA revealed no significant main effect of Group (F(1,120) = 2.46, p = 0.12) and no significant Group × Condition interaction (F(5, 120) = 1.48, p = 0.2). Moreover, post hoc comparisons indicated that the observed improvements of tDCS in the 30% and 40% morphed distances in the M3 montage relative to sham were not reached significant (Fig. 3c; Šidák-corrected: p = 0.35 and p = 0.31, respectively). Among the active stimulation groups, only M1 showed a statistically significant improvement in face recognition performance compared to the sham at the 30% morph distance (Fig. 4). Direct comparisons between M1, M2, and M3 did not reach significance (all p > 0.05, two-way ANOVA; Table 1), indicating that we cannot conclusively claim a montage-specific effect relative to other active groups. tDCS effects on reaction time An important parameter in studying the neuromodulatory effects of tDCS is the influence of stimulation on subjects’ reaction time (RT) as a measure of cognitive function72. As depicted in Fig. 5a and 5b, two-way ANOVAs on reaction time revealed significant main effects of stimulation for the M1 and M3 montages relative to Sham (Table 2; M1: F(1,120) = 4.67, p < 0.05; M3: F(1,120) = 13.67, p < 0.001). The interaction between Group and Morph Level did not reach significance for either montage (M1: F(5,120) = 0.3, p = 0.90; M3: F(5,120) = 0.17, p = 0.97), indicating that the stimulation effect was broadly distributed across difficulty levels rather than confined to a particular morph condition. Consistent with this, follow-up post hoc comparisons using the Šidák correction did not reveal significant differences between the M1 or M3 groups and Sham at any individual morph level. The M2 montage showed a non-significant trend toward increased reaction times relative to Sham (main effect of Group: F(1,114) = 3.19, p = 0.07; interaction: F(5,114) = 0.22, p = 0.94). The bilateral montages (M1 and M2) produced comparable overall RTs (all p > 0.05, two-way ANOVA; Table 2). Finally, although M1 and M2 tended to yield shorter RTs than the M3 montage, these pairwise differences did not reach statistical significance (all p > 0.05, two-way ANOVA; Table 2). DMS Task Reaction Time (RT) for Different Stimulation Montages. (a) RT plots for the M1 (green line, n = 13), M2 (cyan line, n = 12), and M3 (magenta line, n = 13) right-anode groups compared to the sham group (red dashed line, n = 9). (b) Mean RT differences (ms) for M1 (green), M2 (cyan), and M3 (magenta) montages relative to the Sham group across different sample–test morph levels. (c–f) Individual participant performance across different conditions as a function of RT. Regression lines show a negative correlation between performance and RT, with the sham group exhibiting a steeper slope compared to the three stimulation groups. Each condition is represented by a distinct color for clarity. Table 2Reaction time ANOVA results.MontagesF-ANOVA (main)P-ANOVA (main)F-ANOVA (interaction)P-ANOVA (interaction)M1/Sham4.67120.03270.3080.9073M2/Sham3.19030.07670.2290.9492M3/Sham13.67850.00030.1700.9733M1/M20.06050.80610.0990.9920M1/M32.78130.09750.15109,794M2/M33.30980.07100.1030.9913This table shows the main (Group and Condition) and interaction (Group × Condition) two-way ANOVA analysis results for each pair of montage arrangements. Significance is indicated in bold digits. Reaction time ANOVA results. This table shows the main (Group and Condition) and interaction (Group × Condition) two-way ANOVA analysis results for each pair of montage arrangements. Significance is indicated in bold digits. To explore the relationship between performance and RT across different task conditions, we plotted individual subjects’ performance as a function of their reaction time (Fig. 5c-f). In the sham group, a negative correlation was observed between subjects’ performance and RT (r = −0.52, p < 0.001). This negative correlation was attenuated across all active stimulation groups. Specifically, the M1 (r = −0.19, p = 0.09) and M3 (r = −0.14, p = 0.22) groups displayed non-significant correlations, whereas the M2 group maintained a significant negative correlation (r = −0.36, p < 0.01). Comparing the slope of the regression line between the sham (slope = −0.09%/ms) and stimulation groups revealed a significant attenuation in the M1 (slope = −0.02%/ms, z-value = − 2.7, p < 0.01) and M2 (slope = −0.01%/ms, z-value = −3.2, p < 0.01) groups. Although the M3 group also exhibited a lower slope relative to the sham (slope = −0.04%/ms), this difference did not reach statistical significance (z-value = −1.94, p = 0.051). Discussion In this study, we employed transcranial direct current stimulation (tDCS) to examine the roles of the right and left hemispheres in face identification during a delayed match-to-sample (DMS) task. Anodal stimulation was applied to the right or left occipito-temporal cortex (OTC), a well-established region for processing face-related information1–3. To investigate the effects of tDCS across a range of task difficulties, we designed the experiment with multiple morph levels. Because it was unclear whether neuromodulation would selectively affect performance at the point of subjective equality (PSE). We used morphed face images with varying levels of shape similarity to create different degrees of ambiguity in face recognition. Non-matching face stimuli ranged from 20 to 60% shape difference (in 10% intervals) relative to the sample. Our results revealed a significant enhancement in face recognition performance, primarily when anodal and cathodal electrodes were placed over the right and left OTC, respectively (M1 montage). The effects of tDCS depended on task difficulty, showing the largest improvement at 30% morph-distance between sample and test stimuli. This level was closest to the point of subjective equality (PSE) derived from the psychometric function in the sham condition. In the M3 montage (anode on the right OTC, cathode on the left shoulder), stimulation did not yield a statistically significant performance enhancement relative to sham. However, the overall improvement was also not significantly different from the M1 group (Table 1). Additionally, we observed that reaction times (RTs) were generally longer than in the sham condition across all montages, although this difference reached statistical significance only in the M1 and M3 groups (Table 2). Several factors influence the outcomes of tDCS studies, and task difficulty is one of the most critical. This work is the first to systematically employ a range of stimulus differences to precisely gauge the effects of tDCS on face recognition. Previous findings on tDCS over the OTC have been mixed. Some studies have shown moderate to negligible improvements in face identification16,17 or in recognizing own-race versus other-race faces19,20. In these reports, baseline performance in sham or pre-test conditions typically hovered around 70%, which lies outside the performance range most influenced by tDCS in our experiment. As demonstrated here, when face stimuli in the DMS task were either very easy or highly difficult, the impact of tDCS was minimal. Consistent with our findings, Brunyé et al. (2017) showed that tDCS effects were most pronounced under high memory loads (up to four faces simultaneously) in a working memory task, supporting the idea that tDCS efficacy depends on task difficulty and cognitive load. Another key factor in tDCS efficacy is the specific stimulation protocol. In this study, the cathode electrode in M1 and M2 was positioned over the contralateral OTC, enabling bilateral modulation of visual cortical regions. The M1 but not M2 or M3 montages improved face identity recognition around the PSE; however, the difference between M1, M2, and M3 did not reach statistical significance. According to the literature, the right OTC is predominantly associated with face recognition38,41 and holistic face processing73,74, whereas the left OTC is more specialized for feature-based processing73,75. With the M1 montage, anodal tDCS over the right OTC may enhance excitability and improve neuronal processing, leading to better DMS performance. Simultaneous cathodal stimulation of the left OTC may reduce competing activity from the left hemisphere, optimizing right OTC function. This interhemispheric balance aligns with the natural asymmetry in face processing, explaining why M1 led to greater performance gains. In contrast, M2 likely facilitates the left OTC and inhibits the right, prioritizing feature-based processing over holistic processing—thus disrupting the natural right-hemisphere dominance in face recognition. Previous research has also examined unilateral and bilateral transcranial electrical stimulation (tES) over the OTC for face perception. Similar to our data and contrary to the classic excitatory/inhibitory interpretation of anodal/cathodal stimulation, Yang et al. (2014) found that bilateral tDCS over the OTC could enhance face perception and holistic face processing; however, their findings showed improvement regardless of polarity placement. Costantino et al. (2017) likewise noted that cathodal stimulation over the OTC alone did not impair face perception or memory. While Barbieri et al. (2016) observed that unilateral right OTC stimulation improved both face perception and memory, Willis et al. (2019) did not replicate these results, although their findings were qualitatively similar. Bilateral transcranial random noise stimulation (tRNS) over the OTC has improved facial identity recognition for upright faces76 and face matching77, though other studies report minimal effects on face perception17. In our experiment, the M3 montage (anode on the right OTC, cathode on the left shoulder) was specifically designed to assess the effect of targeting the right OTC in isolation. The results indicated that the M3 montage, similar to the M2 montage, did not produce significant performance improvements compared to the sham group. In contrast, the M1 montage resulted in statistically significant enhancements at a 30% morph distance. Although direct comparisons between stimulation groups did not reach significance, the relatively weaker modulation observed with M3 highlights the potential importance of bilateral stimulation for enhancing face identity discrimination, which is observed in M1. A similar phenomenon was described by Fecteau et al. (2007), who reported that bifrontal stimulation of the dorsolateral prefrontal cortex influenced risk-taking behavior, whereas unilateral stimulation had no significant effect. The authors concluded that balancing activity between the two hemispheres is key to maximizing the impact of tDCS. With the M3 montage, placing the cathode extracephalically may alter current flow, possibly directing current away from the cortex and into deeper brain structures78,79. This shift, along with the lack of interhemispheric modulation, could explain why M3 was comparatively ineffective. Overall, these findings suggest that bilateral OTC stimulation is more beneficial for face identification, underscoring the need for coordinated activity across hemispheres. Several studies have found no significant reaction time (RT) differences between tDCS and control conditions for tasks such as own/other-race face recognition20,80, face detection36,81, face perception and memory16,17, and working memory32. Other research, however, has reported shorter RTs following tDCS, particularly for facial expression recognition34,82. In contrast, our study showed that both bilateral (M1) and unilateral stimulation (M3) led to longer RTs in the DMS task, yielding a flatter performance/RT slope compared to the sham condition (Fig. 5). Because tDCS over the right or left OTC can also modulate areas tuned to visual categories other than faces83, these broader neural interactions might delay response times. For instance, Barbieri et al. (2016) showed that OTC stimulation improved perception and memory for both faces and objects, suggesting that tDCS may introduce additional competition in neural networks during perceptual decision-making, thereby prolonging response times. Similar inconsistencies have been observed for tDCS over the prefrontal cortex. Some studies have shown increased RT after bifrontal stimulation84,85, while others found decreased RT following unilateral stimulation86 or differential RT outcomes depending on current polarity87. In summary, our results highlight the notable effects of tDCS over the OTC in improving face identification during a DMS task, particularly at specific difficulty levels. Bilateral OTC modulation appears more effective than unilateral right OTC stimulation, likely due to interhemispheric rebalancing. The consistently longer RTs, particularly in M1 and M3 groups, relative to sham, suggest that tDCS-induced neuromodulation can influence neurons selective for both face and non-face categories, increasing the computational demands of sensory discrimination and thus extending the time required to make a decision. These findings deepen our understanding of how tDCS alters cognitive processing and underscore the importance of both stimulation parameters and task characteristics in modulating face recognition performance. Both spontaneous and attention-driven increases in baseline activity of sensory neurons result in higher sensibility and detection performance of incoming sensory information61–64. tDCS increases neural membrane potential65 and may exert its effect on sensory information processing by promoting higher neural sensibility. The enhanced sensibility is particularly evident in the detection of ambiguous stimuli such as the noisy condition62 or the ambiguous vase-faces stimulus63. Changes in baseline neural activity and membrane potential can occur spontaneously by ongoing slow brain oscillations66,67, stimulus context64, and/or attentional influence68. In our study, the largest impact of tDCS was observed for the ambiguous condition, for which some, but not adequate, visual information was available (30% morphed distance). These results are consistent with findings related to the impact of attention as well as those showing spontaneous fluctuation of baseline activity and its impact on the elevation of neural sensibility and detection performance. Using our experimental paradigm and data, we cannot decipher which of these mechanisms is responsible for our findings. tDCS, particularly when applied over frontoparietal regions involved in selective attention, has been shown to modulate attentional processes88–90. While activation of the general attentional process in our study cannot be ruled out. It should be noted that the absence of an active control condition leaves open the possibility that general, nonspecific factors such as arousal or attentional modulation may have contributed to the observed effects, rather than specific neuromodulatory influences targeting the OTC face selective neurons. However, we consider this interpretation unlikely. Non-specific arousal effects, such as heightened alertness or attentional modulation, are typically associated with reduced RTs91,92, which is in contrast with our findings. As shown in Table 2, two-way ANOVAs comparing the M1 and M3 stimulation groups to sham revealed significantly longer RTs, suggesting that the observed performance enhancements were not driven by generalized arousal. This study suggests that tDCS effects are most pronounced under conditions of intermediate perceptual ambiguity. Future investigations could build on this observation by focusing exclusively on stimuli near the PSE, thereby increasing statistical power and experimental sensitivity. Conducting a targeted replication study using only the 30% morph condition would help verify the reliability of the current findings and refine stimulation protocols for subsequent research. Such an approach would also reduce the number of statistical comparisons required and minimize the likelihood of Type I errors. Moreover, future studies should include non-face control conditions, such as objects or scenes, to determine whether the observed effects of stimulation are specific to face identity processing or reflect broader visual or cognitive mechanisms. While the current study focused solely on face stimuli to reduce task complexity and isolate effects related to perceptual ambiguity, this design limits conclusions about face-specific tDCS impact. Given the modest sample size and the absence of a double-blind design, the present findings should be interpreted with caution. Nevertheless, they offer valuable preliminary evidence supporting the potential of tDCS-induced neural modulation in face identity discrimination. Future studies should employ larger sample sizes, implement double-blind protocols, and include systematic assessments of blinding efficacy to enhance methodological rigor and replicability. Acknowledgements The authors thank Fatemeh Mohammadi for her help with data collection. Author contributions H.S. and S.J.S. designed the experiment, formulated the analysis, and wrote the manuscript. S.J.S. collected the data. H.S. supervised the project. Data availability The data that support the findings of this study are available from the corresponding author upon reasonable request. Declaration Competing interests The authors declare no competing interests. Ethics approval and consent to participate Informed consent to participate was obtained from all subjects and/or their legal guardian(s); this study followed protocols approved by the Ethics Committee of the Pasargad Institute for Advanced Innovative Solutions (code number: ETH-2023–12-1). All experimental procedures were conducted at the National Brain Mapping Lab (NBML), under project number P-402–136. Consent for publication Informed consent to participate was obtained from all subjects and/or their legal guardian(s). Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data that support the findings of this study are available from the corresponding author upon reasonable request.