Non‐invasive brain stimulation in modulation of mental rotation ability: A systematic review and meta‐analysis

Mental rotation, the ability to manipulate mental images, is an important function in human cognition. This systematic review and meta‐analysis investigates the potential of non‐invasive brain stimulation in modulation of this component of visuo‐spatial perception. The PubMed database was reviewed prior to 31 September 2020 on randomized controlled trials investigating the effects of repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), and transcranial alternating current stimulation (tACS) on the mental rotation ability in healthy persons. A total of 17 studies (including 485 subjects) matched our inclusion criteria. Within their scope, overall, 46 sham‐controlled experiments were performed. Methodology and results of each experiment are presented in a meta‐analysis. The data show a large variety of methods and effects. The influence of (1) stimulation‐technique (tDCS, tACS, and rTMS), (2) stimulation protocol (anodal, cathodal, bilateral tDCS, tACS, high‐frequency rTMS, low‐frequency rTMS, paired pulse rTMS, and theta burst stimulation), (3) stimulation timing (preconditioning and simultaneous), (4) stimulation location (left, right hemisphere, frontal, and parietal area), and (5) stimulus type (bodily and non‐bodily) is discussed. The data indicate a beneficial effect of anodal tDCS and of tACS and no effect of cathodal tDCS on the mental rotation ability. Bilateral tDCS protocols both improved and worsened the parameters assessed. The small effect sizes obtained in mostly rTMS experiments require cautious interpretation.

assessed. The small effect sizes obtained in mostly rTMS experiments require cautious interpretation.
K E Y W O R D S mental rotation, non-invasive brain stimulation 1 | INTRODUCTION

| Mental rotation
Mental rotation is psychological operation in which a mental image is rotated around some axis in threedimensional space (Zacks, 2008). Since first presented by Shepard and Metzler (1971), numerous concepts were developed to assess this component of visuo-spatial perception ( Figure 1). Current data repeatedly demonstrates correlations between superior mental rotation abilities and "success" in daily life. For example, optimal neurodevelopment during the first few months of life is associated, among others, with good mental rotation ability at age 6-10 years (Serdarevic et al., 2016). A higher mental rotation performance is significantly correlated to better mathematical achievement in boys between 7 and 9 years old and thus has implications for school practice (van Tetering, van der Donk, de Groot, & Jolles, 2019). Athletes and artists may present above average mental rotation performance (Sluming, Brooks, Howard, Downes, & Roberts, 2007;Voyer & Jansen, 2017). For example, professional musicians are 25% faster at mental rotation tasks than academically educated peers (Sluming, Brooks, Howard, Downes, & Roberts, 2007). Amateur gymnasts and orienteers demonstrate 50% better performance in mental rotation of cubes than contemporaries who take no regular sports (Voyer & Jansen, 2017).
On the other hand, relationships were also found between the reduction of mental rotation performance and ageing-or illness-related decline. Elderly (63-80 years) show significantly poorer mental rotation ability than young adults (17-29 years) (Techentin, Voyer, & Voyer, 2014). People who suffered from a major depressive disorder have a worse mental rotation performance than healthy peers (Chen et al., 2014), and the severity of depression symptoms correlates with the reduction of mental rotation ability (Oshiyama et al., 2018). Stroke victims show significantly poorer mental rotation performance than healthy controls (Braun et al., 2017;Daprati, Nico, Duval, & Lacquaniti, 2010), and the amount of sensitivity (Braun et al., 2017) and motor (Daprati, Nico, Duval, & Lacquaniti, 2010) deficits correlates with the disruption of mental rotation performance. Pain-related diseases are also associated with impaired mental rotation performance (Baarbé, Holmes, Murphy, Haavik, & Murphy, 2016;Kohler et al., 2019).
Some data explain the neurological background of illness-related disruption of mental rotation (Kohler et al., 2019;Yan et al., 2012). A magnet resonance imaging study revealed that patients who suffer from complex regional pain syndrome have a reduced activation in certain areas including the subthalamic nucleus, nucleus accumbens, and putamen, during mental rotation tasks (Kohler et al., 2019). An electroencephalography study in stroke patients demonstrated a hypoactivity in frontal and central areas of the ipsilesional hemisphere, as well as hypo-activity of the frontal cortex bilateral during a mental rotation task (Yan et al., 2012). Thus, illness-induced reduction of the mental rotation ability seems to be associated with suppressed neural processing. In contrast, a superior mental rotation performance appears to be linked to enhanced neural processes. A fMRI study found that orchestral musicians have a significantly increased activation in Broca's area during a mental rotation task, in addition to the visuospatial network, which was activated in both musicians and non-musicians (Sluming, Brooks, Howard, Downes, & Roberts, 2007). 1.3 | Non-invasive brain stimulation in modulation of neural and cognitive processing Repetitive transcranial magnetic stimulation (rTMS) and transcranial direct current/alternating current stimulation (tDCS/tACS) are innovative methods that can modulate neural processing within the cortex (Giordano F I G U R E 1 Common strategies to assess mental rotation ability: (1) congruence determination (same/mirrored) of two stimuli, which are rotated relative to each other (e.g., Shepard-Metzler objects, Shepard & Metzler, 1971;hands, Bode, Koeneke, & Jäncke, 2007), (2) congruence determination (congruent/incongruent) of rotated items in relation to "standard item" (e.g., animals, Snodgrass & Vanderwart, 1980;letters, Corballis & Sergent, 1989;cubes, Gittler & Glück, 1998), (3) laterality determination (right/left) of foots, hands (Sekiyama, 1982), identifying which hand (right/left) was marked depending on the stimulus dot presented in the lower portion of the figure (Ratcliff, 1979), and (4) paper folding and cutting task (determination of the item that represents what the folded and cut piece of paper will look like when it is unfolded) (Oldrati, Colombo, & Antonietti, 2018Herrmann, Rach, Neuling, & Strüber, 2013;Siebner & Rothwell, 2003) and have therefore the potential to influence the mental rotation ability. TDCS/tACS consists of the application of a low-intensity direct current/alternating current that flows between two electrodes (anode and cathode) (Herrmann, Rach, Neuling, & Strüber, 2013;Nasseri, Nitsche, & Ekhtiari, 2015). One of the electrodes is positioned over the target area (active electrode), the other (reference electrode) over another cranial or extracranial position. TMS is produced by passing short high-current pulses through an insulated coil of wire held over the scalp. The electric pulse induces a rapidly changing magnetic field with lines of flux running perpendicular to the coil (Lang & Siebner, 2007;Siebner & Rothwell, 2003). Acute physiological effects of non-invasive brain stimulation techniques distinguish tDCS/tACS from TMS. TMS produces high intensities of short-lasting electromagnetic currents that lead to a supra-threshold activation of the neurons. In contrast, tDCS/tACS does not generate action potentials in neurons, but bi-directionally modifies their spontaneous firing activity via subthreshold activation (Yavari, Jamil, Mosayebi Samani, Vidor, & Nitsche, 2018). Potentially oversimplifying, the basic idea is (i) anodal tDCS (anode over the target area) and facilitatory rTMS (high-frequency rTMS [≥5 Hz], intermittent theta burst stimulation [iTBS], and paired-pulse rTMS [pprTMS] [inter-stimulus interval >5 ms]) increase neuronal excitability and may consequently enhance cognitive performance; (ii) cathodal tDCS (cathode over the target area), inhibitory rTMS (low-frequency rTMS [1 Hz], continuous theta burst stimulation [cTBS], and pp rTMS [inter-stimulus interval ≤ 5 ms]) decreases neuronal excitability and subsequently worsen cognitive performance (Fertonani & Miniussi, 2017;Lang & Siebner, 2007). Indeed, the factors influencing the interindividual variability of the effect of brain stimulation on neural and cognitive processing are not completely understood. Recent studies show, for example, high interindividual variability regarding the neural responses to "up" and "down" regulating brain stimulation protocols (Hamada, Murase, Hasan, Balaratnam, & Rothwell, 2013;Wiethoff, Hamada, & Rothwell, 2014). It was also demonstrated that stimulation-timing (preconditioning/preceding versus simultaneously application) plays a critical role on stimulation effects on cognition (Hill, Fitzgerald, & Hoy, 2016;Yeh & Rose, 2019). In the past few years, there has been a rapid increase in the application of non-invasive brain stimulation techniques in modulation of mental rotation in healthy persons. This systematic review aims to investigate if non-invasive brain stimulation is effective in supporting this relevant visuo-spatial ability.

| Study selection
The PubMed database was reviewed up to 31 September 2020 for papers reporting the use of non-invasive brain stimulation in modulation of mental rotation ability in healthy subjects. The screening was performed according to the PRISMA guidelines. Search term (1) "transcranial direct current stimulation" and "mental rotation," (2) "transcranial alternation current stimulation" and "mental rotation," and (3) "repetitive transcranial magnetic stimulation" and "mental rotation" were used. Studies matching the following criteria were included: (1) human-studies, (2) prospective studies, (3) healthy subjects, (4) modulation of mental rotation ability by non-invasive brain stimulation, and (5) placebocontrolled study or study with at least two experimental groups/treatments. The screening was performed by two independent reviewers (JV and AE). Disagreements were resolved by consensus.

| Data extraction and risk of bias
Outcomes reported were (1) participants characteristics (age and gender), (2) methodological approach (number of participants, crossover/parallel groups design, assessments, number and scheduling of evaluations, and stimulation positioning techniques), (3) intervention characteristics (stimulation protocol, stimulation duration, stimulated area, and number of sessions), and (4) outcomes (assessments and between-group differences). The Oxford quality scoring system was applied to evaluate the methodological quality of trials included, such as random allocation, subjects and assessor blinding and description of dropouts and withdrawal (Jadad et al., 1996). Its overall score ranges between 0 and 5. The higher the score, the better the methodological quality of the study. Oxford quality score for studies enrolled in our meta-analysis is presented in Table 1 (tDCS/tACS) and

| Data synthesis and statistical analysis
Effect size and 95% confidence interval were calculated for each study and each assessment. On this basis, means were calculated for each stimulation technique (tDCS/ tACS and rTMS), stimulated hemisphere (left, right, bilateral, and central), stimulated area (frontal, parietal, frontal +parietal, and cerebellum), stimulation timing T A B L E 1 Studies investigating tDCS/tACS in modulation of mental rotation, included in our meta-analysis (preconditioning = offline, simultaneous = online), and stimuli (bodily and non-bodily). The effects on mental rotation ability were classified using Cohen's effect size definition (d < 0.2 "no effect," d = 0.2-0.49 "small effect," d = 0.5-0.79 "medium effect," and d ≥ 0.8 "large effect") (Campbell, Machin, & Walters, 2007). The homogeneity of effects across studies was evaluated using the inconsistency test (I 2 ), where values above 50% were considered indicative of high heterogeneity (Higgins, Thompson, Deeks, & Altman, 2003).

| RESULTS
In total, 56 articles were identified for the above search procedure. Thirty-nine of them were excluded because of inappropriate article type, outcome, intervention, and/or population. The remaining 17 manuscripts corresponded with our inclusion criteria and were selected for this systematic review and meta-analysis. The detailed summary of the literature search is depicted in Figure 2. Overall, 485 subjects were enrolled. There are no reports of serious adverse events. The trials show a high variability regarding stimulation parameters, participant's characteristics, methodological approach, and effects of noninvasive brain stimulation on mental rotation performance ( Figure 3). Table 3 demonstrates relationships between stimulation-induced effects on mental rotation ability and stimulation technique used, stimulation location and timing and stimuli used. Greater effects of mental rotation ability are associated with (1) tDCS and tACS, (2) stimulation of right hemisphere, (3) combined stimulation of frontal and parietal areas, and (4) using bodied stimuli. For more clarity, we split our presentation into studies investigating tDCS/tACS and studies investigating rTMS below.

| Repetitive transcranial magnetic stimulation
Ten studies tested rTMS on influencing the mental rotation ability. In total, 34 placebo-controlled experiments were applied (Table 2).

| DISCUSSION
The aim of this systematic review and meta-analysis was to investigate whether non-invasive brain stimulation can support the mental rotation ability in healthy subjects. The available data demonstrate great inhomogeneity of effects across the experiments and indicate that non-invasive brain stimulation can both support and deteriorate this important visuo-spatial ability. In fact, the effect size calculation detected a relevant intervention-induced improvement of mental rotation in nine protocols. In contrast, eight protocols evoked a relevant worsening of this ability. The remaining 23 protocols did not have a clear effect (Figure 3, Table 3). The reasons for these inconsistencies may be the high heterogeneity of experiments regarding stimulation techniques (tDSC/tACS and rTMS), stimulation parameters (different protocols, durations, and intensities), stimulation location (different brain areas and hemispheres), stimulation timing (online and offline), and mental rotation objects (bodily and non-bodily). We will discuss how these factors may influence the effects of non-invasive brain stimulation on the mental rotation ability.

| Stimulation technique-dependent effects
The effect size calculation demonstrated that tDCS/ tACS induced greater effects on mental rotation ability than rTMS. In fact, only two rTMS protocols but 10 tDCS/tACS protocols induced relevant effects (Figure 3, Table 3). It is not clear whether and to what extent the differences reflect the differential impact of the various techniques on the human brain and cognition, and which role other factors (such as patient characteristics, study design, etc.) play. The included tDCS/ tACS studies differ considerably from rTMS studies regarding the applied methodological approaches. A major part of the rTMS studies performed stimulation during the cognitive task (online). In contrast, tDCS/ tACS studies performed simultaneous and preconditioning stimulation equally frequently (Figure 3). This is an important point because previous reviews and meta-analyses indicate that stimulationtiming may relevantly impact stimulation effects of cognitive performance (Hill, Fitzgerald, & Hoy, 2016;Yeh & Rose, 2019). Moreover, most rTMS studies did not perform a conventional repetitive application of TMS pulses with a constant frequency. Instead, short sequencies of two to five pulses (with a given frequency) were time locked to the stimulus presentation during the cognitive task. A repetitive application of 38-200 sequencies were performed in each subject within a few minutes (Table 1, Figure 3). The question is whether those protocols can be clearly considered as repetitive application of TMS? In contrast, a conventional uninterrupted stimulation was performed in all tDCS/tACS studies. Given these inconsistencies, we will discuss the rTMS and the tDCS/tACS studies separately.

| TDCS/tACS for modulation of the mental rotation ability
The effect size calculation indicates that tDCS/tACS is effective in the modulation of mental rotation (Figure 3, Table 3). However, the effects are highly inhomogeneous. We will discuss how stimulation-timing, electrodes-polarity, stimulation location and type of stimuli may influence the effectiveness.

| Stimulation-timing dependent effects
Overall 12 placebo-controlled experiments investigated tDCS in the modulation of the mental rotation ability. Only two protocols did not induce any relevant effect on the observed parameters (Foroughi, Blumberg, & Parasuraman, 2015;van Elk, Duizer, Sligte, & van Schie, 2017). Interestingly, both ineffective protocols performed simultaneous stimulation (online). Similarly, an earlier review and meta-analysis demonstrated that only offline (but not online) anodal tDCS significantly supports working memory in healthy populations (Hill, Fitzgerald, & Hoy, 2016). Interestingly, an opposite effect (supportive influence of online, but not offline anodal tDCS) was observed in neuropsychiatric populations (Hill, Fitzgerald, & Hoy, 2016). In any case, the available data indicate that tDCS/tACS timing may significantly influence the stimulation effect on human cognition.

| Electrodes-polarity dependent effects
Our data show polarity-dependent effects in accordance with the oversimplified theory that indicates that anodal tDCS/tACD improves and cathodal tDCS/tACS deteriorates cognitive performance (Fertonani & Miniussi, 2017). Similar results are presented in an earlier systematic review and meta-analysis that investigates the effect of tDCS over the dorsolateral prefrontal cortex on human cognition (Dedoncker, Brunoni, Baeken, & Vanderhasselt, 2016). Only anodal stimulation supported cognitive processing in both healthy and neuropsychiatric samples. Cathodal protocols did not induce significant effects (Dedoncker, Brunoni, Baeken, & Vanderhasselt, 2016). The data also show that a reverse positioning of electrodes during simultaneous unilateral stimulation evokes reverse cognitive changes (Kikuchi et al., 2017). While anodal stimulation over the right parietal cortex and cathodal stimulation over the right frontal cortex improves the mental rotation ability, the reverse electrode placement leads to its deterioration (Kikuchi et al., 2017). This result is supported by a previous systematic review and metaanalysis that detected between-hemispheric asymmetries during mental rotation tasks (Zacks, 2008). Parietal cortex activity is somewhat more consistently observed in the right hemisphere, whereas frontal cortex activity is more consistently observed in the left hemisphere (Zacks, 2008). This explains the supportive effect of anodal (and not of cathodal) tDCS over the right parietal regions.

| Stimulation-location dependent effects
Our data indicate that the positioning of the active electrode over the right hemisphere is more effective than both left hemispheric and simultaneous bilateral stimulation (with the anode over one hemisphere and the cathode over the other) ( Figure 3, Table 3). Finally, simultaneous application over the right hemisphere (with electrodes over parietal and frontal cortical areas) induces greater effects than all another protocols (Kikuchi et al., 2017). In accordance, a previous meta-analysis emphasizes the superior role of the right hemisphere during some mental rotation tasks (Tomasino & Gremese, 2016). Whereas mental rotation of bodily stimuli (e.g., hands and feet) leads to the activation of the bilateral sensorimotor network, non-bodily stimuli (e.g., Shepard-Metzler objects) induce right lateralized activation (Tomasino & Gremese, 2016).

| Stimulus-type dependent effects
The effect size calculation indicates that tDCS/tACS is more effective in modulation of the mental rotation of bodily stimuli than in the modulation of non-bodily objects. A previous review refers to stimulus-dependent differences of neural-processing (Searle & Hamm, 2017). BOLD activation within the ventral stream and the premotor cortex increases linearly with stimulus discrepancies during mirror/normal discriminations, but only during mental rotation of bodily stimuli. Similarly, slow negativity over centroparietal regions as recorded by EEG increases for greater rotations only during mental rotation of bodily objects. Higher difficulty during mental rotation of non-bodily stimuli did not induce higher activation of the neural network (Searle & Hamm, 2017). It is conceivable, that stronger activation of the neural network during mental rotation tasks provides more opportunity for modulation with non-invasive brain stimulation and has thus greater effects on cognitive performance.

| rTMS for modulation of the mental rotation ability
The effect size calculation demonstrates that rTMS has only limited effectiveness in the modulation of the mental rotation ability (Figure 3, Table 3). From a total of 28 experiments, only two protocols induce relevant effects. However, Table 2 shows that any part of mental rotation tasks could be significantly influenced by this method. We will discuss how different factors influence the effectiveness of rTMS.

| Stimulation-timing dependent effects
The effect size calculation demonstrates relevant changes of mental rotation ability in one-half of the experiments that applied preconditioning stimulation and only in onefifth of experiments that performed a simultaneous stimulation ( Figure 3, Table 3). Thus, offline application seems to have more potential to influence the visuospatial ability. Similarly, a meta-analysis detected timingdependent effects of rTMS for modulation of working memory (Yeh & Rose, 2019). Offline 20-Hz rTMS induces an improvement and online 20-Hz rTMS a deterioration of the observed parameters (Yeh & Rose, 2019). Moreover, in the framework of online stimulation, the timing of rTMS pulses in relation to cognitive tasks significantly impacted their effects. A study demonstrated that 20-Hz rTMS that was applied over the right lobus parietalis superior 400-600 ms after stimulus onset affected the reaction time during mental rotation of alphanumeric characters (Harris & Miniussi, 2003). The same protocol applied 200-400 ms and 600-800 ms after stimulus onset did not induced any effects (Harris & Miniussi, 2003

| Stimulation-frequency dependent effects
The data did not demonstrate frequency-dependent effects in accordance with the assumed supportive effects of high-frequency (≥5 Hz) and deteriorating effects of low-frequency (1-Hz) rTMS (Siebner & Rothwell, 2003). The studies show both supportive and disruptive effects of high-frequency protocols ( Table 2). Ten-hertz rTMS induced worsening (Cona, Panozzo, & Semenza, 2017;Pelgrims, Andres, & Olivier, 2009) as well as improvement of the mental rotation ability (Cona, Marino, & Semenza, 2016). Twenty-hertz rTMS evoked deterioration (Feredoes & Sachdev, 2006;Harris & Miniussi, 2003) or no changes of the assessed parameters (Klimesch, Sauseng, & Gerloff, 2003). Individual alpha frequency + 1-Hz rTMS improved the mental rotation ability (Klimesch, Sauseng, & Gerloff, 2003). Individual alpha frequency-3-Hz rTMS did not affect the visuo-spatial performance (Klimesch, Sauseng, & Gerloff, 2003). Only one study tested a low-frequency stimulation protocol up to now. Its results show supporting effects of 1-Hz rTMS on the mental rotation ability (Zeugin, Notter, Knebel, & Ionta, 2020). One trial applied 50-Hz cTBS and demonstrated worsening of the mental rotation ability (Picazio, Oliveri, Koch, Caltagirone, & Petrosini, 2013). This is in accordance with the simplified concept that indicates supporting effects of iTBS and impeding effects of cTBS on neural and cognitive processing (Siebner & Rothwell, 2003); 0.2-Hz pp rTMS did not modulate the performance during the visuo-spatial task (Wang, Callaghan, Gooding-Williams, McAllister, & Kessler, 2016) and contradicts the positive role of paired-pulse rTMS with inter-stimulus interval >5 ms on neural and cognitive processes (Siebner & Rothwell, 2003). In accordance with our findings, the available systematic reviews and metaanalyses demonstrate that effects of diverse rTMS protocols for modulation of human cognition and behaviour are often contrary to the conventional frequency-dependent view (Siebner & Rothwell, 2003). A recent meta-analysis indicates a disruptive effect of 10-Hz and 20-Hz rTMS and no effect of 1 Hz or 5 Hz on attention, executive, language, memory, motor, and perception (Beynel et al., 2019). Another meta-analysis shows both improving and deteriorating effects of 20-Hz rTMS on working memory depending on the stimulation-timing (Yeh & Rose, 2019).
In contrast, only positive effects were found for 1-Hz rTMS (Yeh & Rose, 2019

| Stimulated hemisphere-dependent effects
The effect size calculation did not detect relevant effects regarding the stimulated hemisphere (right versus left) on rTMS effectiveness (Figure 3, Table 3). However, a closer look at our studies detected some hemispheredependent effects (Table 2). On the one hand, 10-Hz rTMS over both the right and the left dorsal premotor cortex impaired accuracy for the same stimuli during mental rotation of Shepard-Metzler objects (Cona, Panozzo, & Semenza, 2017). Only right hemispheric stimulation delayed reaction time for the same stimuli during mental rotation of both Shepard-Metzler objects and hands (Cona, Panozzo, & Semenza, 2017). This result contradicts the findings of an earlier meta-analysis that detected a more consistent parietal cortex activity in the right hemisphere, and frontal cortex activity in the left hemisphere during mental rotation tasks (Zacks, 2008). On the other hand, 20-Hz rTMS applied over the right lobus parietalis superior 400-600 ms after stimulus onset affected reaction time for alphanumeric characters (Harris & Miniussi, 2003). The same stimulation over the homologous area did not evoke any changes (Harris & Miniussi, 2003). This finding confirms the prominent role of right parietal areas (in comparison to the left hemisphere) during neural control of visuo-spatial tasks (Zacks, 2008). Another study indicates the prominent role of the left lateral cerebellum (in comparison to the homologous area) during mental rotation (Picazio, Oliveri, Koch, Caltagirone, & Petrosini, 2013). Fifty-hertz cTBS over left but not over the right hemisphere delayed reaction time during embodied and abstract mental rotation tasks (Picazio, Oliveri, Koch, Caltagirone, & Petrosini, 2013). One study demonstrates a differential involvement of both hemispheres depending on the stimuli angular disparity (Feredoes & Sachdev, 2006). Twenty-hertz rTMS over both the right and the left posterior parietal cortex impaired mental rotation of Shepard-Metzler objects (Feredoes & Sachdev, 2006). While the right hemispheric stimulation influenced stimuli with angular disparity of 120 , the left hemispheric stimulation affected stimuli with angular disparity of 180 . Stimuli with angular disparity of 0 and 60 were not affected (Feredoes & Sachdev, 2006). The differential involvement of the hemispheres depending on the stimuli angular disparity is a new finding, which has not been described in previous reviews and meta-analyses (Beynel et al., 2019; Yeh & Rose, 2019).

| Stimulated area-dependent effects
The effect size calculation shows no relevant impact of the stimulated region (frontal versus parietal) on rTMS effects ( Figure 3, Table 3). However, some studies found significant stimulated area-dependent effects ( Table 2). The data indicate that a simple subdivision of brain network in either "frontal" or "parietal" is too coarse to detect the effective involvement of specific brain regions during mental rotation tasks (Cona, Marino, & Semenza, 2016;Pelgrims, Andres, & Olivier, 2009). Tenhertz rTMS over both the right and the left lobus parietalis superior delayed the reaction time during mental rotation of letters but not hands (Pelgrims, Andres, & Olivier, 2009). The same protocol applied over both the right and the left supramarginal gyrus slowed reaction time during mental rotation of hands but not letters (Pelgrims, Andres, & Olivier, 2009). Similarly, 10-Hz rTMS over the left supplementary motor area improved reaction time and accuracy during mental rotation of Shepard-Metzler objects and of hands (Cona, Marino, & Semenza, 2016). Ten-hertz rTMS over the left primary motor cortex did not induce any changes (Cona, Marino, & Semenza, 2016).

| Stimulus-type dependent effects
The effect size calculation shows relevant effects in one third of the experiments that used bodied stimuli and only in one fifth of experiments that used non-bodied stimuli ( Figure 3, Table 3). Similar effects were also detected for tDCS/tACS studies. The reason for this phenomenon may be stimulus-dependent differences of neural processing as described on a previous review (Searle & Hamm, 2017). Increased difficulty during mental rotation of bodily stimuli induces an increased activation of the neural network. Higher difficulty during mental rotation of non-bodily stimuli did not induce any effects (Searle & Hamm, 2017).

| CONCLUSIONS
The systematic review and meta-analysis presented here shows that non-invasive brain stimulation is effective in modulating the mental rotation ability. The available data indicates a supportive effect of anodal tDCS and of tACS and no effect of cathodal tDCS. Bilateral tDCS/tACS both supported and deteriorated the visuospatial performance. Only small effects were obtained in most rTMS experiments. Stimulation timing, stimulation location and stimulus type are factors that impact the effects of both tDCS/tACS and rTMS. Future research may apply these methods in cohorts with impaired mental rotation ability, such as major depressive disorder, stroke, and pain-related diseases.