Prior to the current use of non-invasive neurostimulation methods to target the cerebellum, studies in the 1970–1980s used surgical methods to implant a cerebellar pacemaker in patients with schizophrenia (17–20). This approach was motivated by animal research showing that the deep cerebellar nuclei are connected to the limbic system and play an important role in affective processing (21). As part of this approach, the pacemaker was implanted into the left side of a patient’s chest and connected to electrodes on the cerebellar surface. A battery-operated stimulator worn by the patient then delivered the stimulus through an antenna taped to the skin. During stimulation, electroencephalography (EEG)-based auditory and somatosensory evoked potentials were reduced in amplitude (18). While some participants did benefit, the cerebellar pacemaker was not always well-tolerated by patients. These studies were fraught with high rates of non-compliance (17–19), with many patients refusing to wear the pacemaker and multiple incidents of device and antenna breakage. Long-term use also led to frontal headaches and vertigo in a subset of patients (19). Critically, these invasive surgical procedures were also inherently associated with serious surgical risks, including air embolisms, formation of cerebrospinal fluid fistula, shifting of implanted electrodes, acute inflammation, and/or seizures (18, 20).

Recent technological advances led to more effective and tolerable, non-invasive brain stimulation methods that can be safely applied to the cerebellum (22, 23). Consequently, there has been exponential growth in studies using the methods depicted in Figure 1 such as repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), and transcranial pulsed current stimulation (tPCS). TMS consists of the generation of a brief, high-intensity magnetic field by passing a brief electric current through a magnetic coil (24, 25). This magnetic field will either excite or inhibit a targeted region underneath the coil. Intermittent theta burst stimulation (iTBS) is a newer form of rTMS that provides a two-second train of bursts (30 pulses) every 10 s, and is the most commonly used method of cerebellar stimulation (26). Relative advantages of iTBS vs. traditional rTMS is that stimulation sessions are shorter, utilize a lower threshold intensity, and have more long-term excitatory meta-neuroplastic effects (27, 28). Another form of TBS that has been used is continuous TBS (often referred to as cTBS, though we note that this acronym has also been used to refer to cerebellar TBS in some studies). Continuous TBS provides a burst of 3 pulses at 50 Hz for either 20 or 40 s (29). While iTBS is considered facilitatory, continuous TBS is thought to suppress cortical excitability (29, 30). Although even continuous TBS, which is considered an inhibitory protocol, can lead to downstream increases in functional connectivity between brain areas (31).

In addition to magnetic stimulation approaches, electrical cerebellar stimulation methods have also been used, albeit not as commonly. Through the use of anodal or cathodal stimulation, tDCS can modulate cortical excitability of targeted neural circuits by either increasing or decreasing intrinsic neural firing (32). tDCS entails delivery of a weak direct current through a surface scalp electrode over the cerebellum (33). It is thought that tDCS modulates neural activity in a polarity-specific manner (32, 34). Although, tDCS has poorer focality than TMS (35). tPCS, another method of transcranial electrical stimulation that is used to directly modulate neuro-oscillations, delivers a pulsating current of a specific frequency over a targeted area (36). Advantages of tDCS and tPCS are that these devices are typically inexpensive, battery-operated, and portable (35). Thus, tDCS and tPCS can be used for in-home treatment (32, 37). Initial research across six clinical trials showed the feasibility and tolerability of implementing repeated sessions of at-home tDCS with remote supervision (23). These different approaches highlight the complex and far-reaching capabilities of neurostimulation; capabilities of which we are just beginning to understand.

Previous brain stimulation studies in schizophrenia have typically targeted cortical regions, such as the frontal cortex and the motor cortex, in an effort to improve positive symptoms (38, 39), negative symptoms (40–42), and cognitive deficits (42, 43). The rationale for the cerebellum as a brain stimulation target in schizophrenia is at least twofold: first, there is increased knowledge documenting relationships between cerebellum abnormalities and clinical features of schizophrenia (8–10), and second, the cerebellum has several unique attributes that make it an attractive stimulation site, such as its immense and distributed connections throughout the cortex, impressive processing capabilities, and inherent plasticity (33, 44–46). The cerebellum contains over 50–80% of the brain’s neurons while only accounting for 10% of the brain’s volume (47, 48). To accommodate all of these neurons within a small volume, the large number of cells is packed in a columnar array with modules that are perpendicular to the cortical surface and parallel to each other (3–5). This organizational structure is conducive to massive parallel processing (3–5) and has been likened to a biological equivalent of a modern microprocessor chip (5). The cerebellum is also located immediately below the skull making it a convenient site for electrode placement (44). Moreover, the cerebellar cortex has been found to be highly responsive to electrical and magnetic stimulation (44).

One of the critical advantages of cerebellar stimulation lies in the potential for modulating cerebello-cerebral circuits, and in turn, impacting cognitive and behavioral functions that depend on these distributed circuits (12, 14). The cerebellum is structurally and functionally connected to numerous cortical and subcortical regions (6, 49–51), with closed parallel loops that link the cerebellum to distant cortical regions (6, 12, 52). Consequently, the cerebellum has been described “as a window to the whole brain” (15). By stimulating the cerebellum, researchers can indirectly modulate dysfunctional cortical circuitry via cerebello-cerebral circuits (14, 15, 44). Additionally, it has been posited that cerebellar stimulation could lead to long-lasting modulatory effects in schizophrenia through the induction of cerebellar plasticity (53). It is thought that the cerebellum has both long-term synaptic and non-synaptic plasticity (45, 46, 54), both of which drive new learning (46, 54). This notion is supported by evidence of induced plasticity in cerebellum-involved pathways (e.g., cerebello-premotor-motor and cerebello-frontal pathways) following rTMS (55, 56).

The precise mechanisms of non-invasive stimulation of the human cerebellum are unknown. One theory is that at least some forms of neurostimulation, like TMS and tDCS, modulate the excitability of Purkinje cells (PCs), a class of GABAergic inhibitory neurons found in the superior cerebellum (53). PCs are large cerebellar output neurons that play a central role in the cerebellar cortical circuit by modulating activity in the deep cerebellar nuclei outflow. Pre-clinical findings have shown that rTMS using a low-intensity current in mice can alter dendritic and spine morphology of Purkinje cells (25). Similarly, it was recently shown that while tDCS-induced electrical field changes can reach deep cerebellar nuclei, PCs were the most sensitive cell type to tDCS (57). More specifically, tDCS anodal stimulation has an excitatory effect that increases output of PCs, and consequently, leads to greater inhibition of cerebello-cerebral pathways; cathodal stimulation has the opposite effect, and is thought to be inhibitory to PCs, leading to disinhibition of the cerebral cortex (14, 58).

As noted above, cerebellar neurostimulation has the potential to induce cerebellar plasticity as seen in healthy individuals (55) and in stroke patients (59). TMS has been found to effect such change through the induction of cerebellar long-term plasticity (LTP) (60), and it is thought that tDCS effects change via a comparable system (15). The most common form of LTP, and its inverse long-term depression (LTD), depend on activation of N-methyl-D-aspartate receptors (NMDAR) (28, 61). The relationship between LTP and NMDAR is evident by the fact that plasticity-inducing effects of neurostimulation effects can be blocked by the administration of NMDAR antagonists, like memantine and dextromethorphan (27, 62). Both LTP and NMDAR abnormalities are also implicated in the pathophysiology of schizophrenia (63–68).

Additionally, the effects of neurostimulation can extend throughout the brain, beyond the initial stimulation target. For instance, tDCS effects may influence activity in both the specific target region and multiple network systems by way of increasing/decreasing release of monoamine neurotransmitters, like dopamine, onto circuits that do not even involve the anodal stimulation site (69, 70). Studies of rTMS (including iTBS) also show downstream effects of stimulation on broader networks (64). These downstream effects on cerebello-cerebral networks are thought to be beneficial in ameliorating clinical symptoms and cognitive deficits.

Over two decades ago, Andreasen et al. (71) called attention to the cerebellum through their cognitive dysmetria hypothesis, which posits a deficit in the underlying neural system responsible for coordinating the processing, prioritization, and expression of information among people with schizophrenia (71). Since then, a number of other mechanistic hypotheses involving the cerebellum have been proposed to explain clinical phenomena in psychosis (72). These studies have not only led to new discoveries regarding the cerebellum’s role in the pathophysiology of schizophrenia, but also have implications for treating schizophrenia.

An initial systematic review reported on 10 studies (3 randomized controlled trials [RCTs], 3 open-label studies, and 2 case reports) of cerebellar stimulation in schizophrenia (26). These studies found that cerebellum stimulation produced clinical changes in negative and depressive symptoms, as well as cognitive functioning domains. Critically, cerebellum modulation showed potential for alleviating schizophrenia symptoms that are less responsive to antipsychotic medications, i.e., negative symptoms (73). These promising findings garnered further enthusiasm for cerebellum stimulation as a treatment for schizophrenia (11, 13), as evidenced by multiple published studies following Escelsior’s initial review and several ongoing clinical trials.

This manuscript provides a systematic update regarding the effects of cerebellar stimulation in schizophrenia. We discuss the effects on clinical symptoms, cognition and behavior, functional brain networks and underlying neuro-oscillations, movement, and physiology. We also review the tolerability of this intervention method for individuals with schizophrenia. We close by discussing issues and technical considerations regarding implementation of cerebellar stimulation as well as recommendations and future directions.

This systematic review was pre-registered on PROSPERO (CRD42022346667) and adheres to PRISMA guidelines (74). JPYH searched research databases (PubMed, Scopus, and PsycInfo) to identify published articles from inception until October 11, 2022. All empirical studies (i.e., excluding reviews and meta-analyses) that reported on the effects of cerebellar stimulation, obtained by physical or pharmacological means (e.g., electric or magnetic stimulation or in situ injection), among animal models of schizophrenia, patients with schizophrenia-spectrum disorders, and individuals at risk of psychosis were included in the review. Articles were included if they included just an active stimulation arm comparing baseline to post-stimulation, or if they included a comparison of active cerebellar stimulation to a sham or active control condition. Unpublished papers and clinical trials were excluded from the systematic review. We systematically searched titles and abstracts using the following Boolean search terms: schizoph* AND cerebell* AND (modulation OR intervention OR stimulation OR transcranial OR TMS OR tDCS OR TBS OR tACS OR injection). References from all included papers as well as a previous systematic review (26) were also evaluated. This screening process was followed by independent full-text screening of all potentially relevant articles and data extraction by JPYH and SVA. Extracted study data included author name and year, description and size of study sample, type of research design (i.e., RCT, open-label uncontrolled study, pilot study, case report, or preclinical study), names and types of measures, assessment timepoints, cerebellar stimulation and sham parameters, and study outcomes (i.e., clinical, cognitive, behavioral, connectivity and oscillatory, movement, physiological, and tolerability/side effects).

Quality of included studies was classified based on Nathan and Gorman’s criteria (75) for rating the methodological rigor of study designs. According to this classification system, there are six levels of studies from Type 1 (most rigorous) to Type 6 (least rigorous). Type 1 studies are double-blind, randomized, prospective, controlled clinical trials. These studies involve comparison of randomized groups, state-of-the-art diagnostic and assessment methods, appropriate analytic methods, clear exclusion and inclusion criteria, and adequate sample size. Type 2 studies are clinical trials that lack some of the rigorous criteria of a Type 1 study, such as small sample sizes, lack of clearly defined inclusion and exclusion criteria, and problems with the randomization protocol. Type 3 studies are open treatment studies and include designs such as pilot and case-control studies. These studies are often methodologically limited by observer bias, retrospective recall error, and uncontrolled data collection. Type 4 studies entail sophisticated analysis of secondary data analyses (e.g., meta-analysis). Type 5 studies are review studies that do not include data analysis. Type 6 studies are case studies, opinion pieces, and essays. Based on the article inclusion criteria for the current study, Type 4 and 5 studies were not included.

The systematic literature search yielded a total of 1,510 published studies (see Figure 2 for PRISMA diagram). Of these, JPYH identified 31 published articles for full review. No additional articles were identified from review of study bibliographies. Based on full-article review, JPYH and SVA independently identified 20 published articles for this review (see Tables 1, 2).

Cerebellar pacemaker studies were the pioneer studies that tested the effectiveness of cerebellar stimulation on alleviating symptoms and improving functioning in schizophrenia. These treatment-resistant, small sample studies showed modest improvements in clinical symptoms and functioning (e.g., living at home, little to no medication, little to no psychotic symptoms) at different follow-up periods (17–20); however, results are difficult to interpret since assessments were not standardized and there were challenges arising from faulty equipment and low treatment tolerability. This approach is rather controversial and no longer recommended (76). Nonetheless, cerebellar pacemaker studies set the stage for the current non-invasive stimulation methods. Due to the increased scientific rigor of more modern studies, we weigh these studies more heavily in our results summary.

Of the studies utilizing non-invasive brain stimulation, 12 examined the effects of cerebellar stimulation on clinical symptoms. The most consistently examined clinical domain was psychotic symptoms, with all 12 studies including measures of psychotic symptoms; 10 studies specifically used the total and/or the positive, negative, and general psychopathology subscales from the Positive and Negative Syndrome Scale (PANSS) (77). Three other studies looked at psychotic symptoms using the Scale for the Assessment of Negative Symptoms (SANS) (78) and Scale for the Assessment of Positive Symptoms (SAPS) (79). Additionally, three studies used a measure of overall clinical impression and four used depression and mood inventories, including the Calgary Depression Scale for Schizophrenia (80) and Visual Analogue Scales.

Six studies examined the effects of cerebellar stimulation on cognition measured from tasks and paper-pencil tests, and three studies examined effects on behavior based on task performance. Three RCTs in individuals with schizophrenia, two of which had relatively larger sample sizes, found no significant effect of iTBS on cognition (82, 90, 93); and that both the active and sham groups similarly improved on multiple cognitive measures over the course of the study (82, 90). In contrast to the null findings for iTBS RCTs, a small open-label study found that patients with schizophenia had improved performance on a continuous performance test and a visuospatial test (87). Gupta et al. (94) also found that individuals with non-clinical psychosis (i.e., high schizotypy) performed better on a pursuit rotor task following active tDCS stimulation; more specifically, the non-clinical psychosis group exhibited a greater rate of improvement on the pursuit rotor task following active stimulation compared to sham, whereas this interaction was not significant for the unaffected comparison group. In fact, the non-clinical psychosis group performed at a comparable level to the control group after active stimulation (94). Additionally, one patient in a case study Laidi et al. (85) improved across a broad range of cognitive functions (i.e., verbal episodic, short term, and working memory, executive, and attention).

Three studies examined how cerebellar stimulation impacted behavior. In a preclinical study, researchers blocked medial prefrontal cortical dopamine receptors in rats as a model of prefrontal abnormalities characteristic of schizophrenia [as evidenced by performance on an interval timing task (95)]. Optogenetic stimulation at 2Hz delta (but not 4, 10, or 20 Hz) of lateral cerebellar projections in these rats rescued behavioral deficits (95). There was no effect of stimulation on other prefrontally-mitigated behaviors, like lever pressing or open-field activity. Comparatively, a clinical study of patients with schizophrenia did not show enhanced performance on the interval timing task following stimulation with iTBS, delta tPCS, or theta tPCS (93). During an eye blink conditioning task which captures associative learning via a simple reflex pathway independent of motivation, one individual with schizophrenia showed progressive conditioning after cerebellar tDCS (85).

Six studies examined the effects of cerebellar stimulation on underlying brain dynamics, with two studies utilizing resting-state functional connectivity and four studies utilizing electroencephalography (EEG) to derive outcome variables. There is a well-established literature documenting aberrant connectivity between the cerebellum and prefrontal cortex in schizophrenia (96). In a large RCT of individuals with schizophrenia, resting-state functional connectivity increased between the cerebellum and the right inferior frontal gyrus, right pallidum, and right frontal pole following iTBS stimulation relative to sham (90). A different RCT reported increased resting-state functional connectivity between the cerebellum and dorsal prefrontal cortex after active iTBS stimulation relative to sham in participants with schizophrenia (89); further, increased cerebellar-dorsolateral prefrontal cortex connectivity correlated with reductions in PANSS negative symptoms characterized by a large effect size (r = −0.81), though we note this sample was quite small (N = 11).

Underlying neuro-oscillations measured with EEG can be abnormal in schizophrenia, such as frequencies associated with perception, memory, and synaptic plasticity, including theta and gamma (97–99). Theta oscillatory power was significantly improved following theta tPCS, but not delta tPCS, as evidenced by greater power in the midfrontal region (93). Participants with schizophrenia also showed a more normal pattern of reduced gamma spectral power in the left frontal and temporal cortex after rTMS (86). Further, reduced gamma power in frontal and temporal cortices correlated with negative symptom reductions, while the left frontal cortex corresponded with less severe depressive symptoms (86). In contrast, a case study showed increased gamma spectral power in the left/right frontal and left occipital cortex as well as decreased gamma spectral power in the left temporal region following iTBS (88). In addition to being used as a treatment modality, cerebellar stimulation can be used to probe deficits and to better understand mechanisms underlying the pathophysiology of schizophrenia. In line with this work, Daskalakis et al. (100) used TMS to probe cerebellar inhibititon (i.e., an important measure of cerebellar activity and cerebello-thalamic-cortical pathway integrity) in individuals with schizophrenia. As predicted, individuals with schizophrenia showed significant deficits in cerebellar inhibition compared to unaffected comparison participants.

Three studies examined the effects of cerebellar stimulation on movement-related symptoms. Two RCTs found no significant effect of iTBS. In one study, individuals with schizophrenia showed decreased extrapyramidal symptoms and ataxia at 6-week follow-up irrespective of their treatment condition (90). In the other RCT, individuals with treatment-resistant schizophrenia showed no effect of condition or time on extrapyramidal physical symptoms (e.g., gait, rigidity, and tremor) (82). In contrast, a small pilot study showed numerical decreases in clinician-rated tardive dyskinesia (84).

Two studies examined the effects of cerebellar stimulation on physiology (i.e., blood pressure, heart rate/pulse). One RCT found no effect of iTBS; individuals with schizophrenia showed decreased diastolic blood pressure at 6-week follow-up irrespective of their treatment condition (90). An open-label uncontrolled study found that diastolic blood pressure increased immediately post-stimulation and 5 min after, but soon returned to baseline levels. There was no significant change for systolic blood pressure, or heart rate/pulse (87).

As described earlier, pioneer studies using invasive cerebellar stimulation methods (17–20) had poor tolerability and high rates of non-compliance. Of the 16 studies using non-invasive methods, 10 reported on adverse events and side-effects following stimulation. Of these 10 studies, two reported no side-effects and 6 reported mild side-effects including headaches that were relieved with analgesics (81, 82, 87), pain (83, 87, 90), dizziness and nausea (83), mild skin burn (84), and excessive sleepiness (81). For more serious side effects, cerebellar stimulation (i.e., rTMS) was terminated for one patient due to increased frequency of auditory hallucinations and associated distress (92). Additionally, two participants exhibited increased mania/hypomania after iTBS (90, 91).

When solely focusing on the 7 RCTs that examined cerebellar stimulation in schizophrenia, results remain largely the same as when all studies are included because the results from RCTs (Types 1 and 2) in this systematic review took precedence over less rigorous study designs (Types 3 and 6).

Over 80% of articles examined whether cerebellar stimulation could alleviate clinical symptoms of schizophrenia. There was some evidence that cerebellar stimulation reduced total psychotic symptoms (i.e., the sum of positive, negative, and general psychopathology symptoms). However, it is unclear whether this reduction was driven by a more specific reduction in negative symptoms. Negative symptoms are thought to account for much of the long-term morbidity, functional impairments, and poor quality of life in schizophrenia, and as such remain a critical unmet need in schizophrenia treatment (101). Cerebellar stimulation was most effective in treating negative symptoms (with some studies reporting reductions maintained up to 24-weeks follow-up), while the findings for positive and general psychopathology symptom reductions were weaker. Because antipsychotic medication is less effective in treating the negative symptoms of schizophrenia (73), the possibility of cerebellar stimulation reducing these symptoms is especially noteworthy. Though we note that existing studies examined overall negative symptoms, and future studies may consider evaluating changes in specific domains of negative symptoms (i.e., experiential/motivational vs. expressive/affective deficits), as they may map onto separate neurobiological systems (102).

Schizophrenia and depression are highly comorbid disorders (103), with both disorders sharing overlapping symptoms, such as anhedonia (104). Initial evidence also raises the possibility that cerebellar stimulation can reduce depressive symptoms in schizophrenia. This is consistent with views of the cerebellum as an “emotional pacemaker,” with the cerebellar vermis in particular believed to modulate emotional processing (105). It is also consistent with research showing that the cerebellum modulates reward processing and controls social behavior (106). Unfortunately, many of the cerebellar stimulation studies that included mood measures lacked a neurostimulation control condition; thus, it cannot be ascertained whether mood changes were the result of active stimulation or simply non-specific treatment effects. However, these studies are an important first step in testing the efficacy of cerebellar stimulation for treating depressive symptoms in schizophrenia.

Surprisingly, cerebellar stimulation did not improve cognition in people with schizophrenia. This contrasts with research in non-psychiatric groups as well as non-psychotic cerebellum-involved disorders, in which participants showed significant gains in learning post-cerebellar stimulation (107, 108). While the previous systematic review (26) concluded that cerebellar stimulation may improve cognitive functioning in schizophrenia, 5 of the 6 papers published after that review showed mixed findings, with comparisons across studies difficult due to study design differences (e.g., different stimulation methods, randomized studies vs. uncontrolled studies), and lack of standardization in cognitive measures and domains assessed. Nonetheless, our understanding of the effects of cerebellar stimulation on cognition in schizophrenia is still an emerging area that would benefit from more rigorous and standardized procedures.

Few studies have looked at whether cerebellar stimulation can impact specific behavioral changes. Interesting findings in rodents showed changes on an interval timing task that captures one’s ability to maintain various temporal intervals in working memory (95); this study raises that possibility that stimulating cerebellar projections to the thalamus may be able to boost cognitive control. Along these lines, another future direction is to examine whether augmentation of this cerebello-thalamic circuit using cerebellar stimulation could modulate sensory prediction deficits present in schizophrenia that depend on this circuit.

Part of the utility of cerebellar stimulation lies in its potential for having widespread impact on distributed cortical networks (6, 12, 49–52). Consistent with this theory, increased functional connectivity between the cerebellum with the frontal cortex (89, 90) and the right pallidum (90) was observed following cerebellar iTBS relative to sham. Gains in cerebellar-to-prefrontal cortex connectivity were also linked with negative symptom reductions (89), suggesting that modulation of cerebellar-cerebral networks via the cerebellum could be an approach to improving symptoms in schizophrenia. Studies examining EEG-related oscillations found that individuals with schizophrenia had a more normal pattern of increased theta oscillatory power in the midfrontal region (93), as well as a more normal pattern of reduced gamma oscillatory power in the left frontal and temporal cortex (86, 88), following tPCS or TMS, respectively. The gamma power reduction corresponded with reductions in negative and depressive symptoms among individuals with schizophrenia (86). Taken together, these studies illustrate how modulation of the cerebellum can impact cerebello-cerebral circuits and their underlying oscillatory dynamics. In turn, this modulation appears to be related to symptom reduction.

The cerebellum is heavily involved in movement and coordination, and movement abnormalities are present in schizophrenia (109, 110). Based on two RCTs, there was no effect of iTBS on movement-related symptoms (82, 90), although a small pilot study showed numerical decreased in tardive dyskinesia (84). More research in this area is needed to establish the effect of cerebellar stimulation for schizophrenia patients in this domain.

The brainstem might be inadvertently affected during cerebellar stimulation, and as such, it is recommended that studies systematically monitor physiological symptoms (14, 21). Of the included studies, two studies examined effects on physiology (i.e., blood pressure, heart rate/pulse) in individuals with schizophrenia. In an RCT, there was no significant effect of iTBS with both active and sham conditions showing decreased diastolic blood pressure at 6-week follow-up (90). An open-label uncontrolled study found increased diastolic blood pressure immediately post-stimulation and 5 min after, with no significant change for systolic either blood pressure or pulse (87). Neither study reported any clinically significant or concerning changes in participants’ physiological activity.

Although the side effect profiles for modern cerebellar stimulation methods are generally low (14, 15), stimulating the cerebellum entails additional risk compared to the rest of the cortex due to its potential to induce painful neck muscle contractions and twitching (14, 21). Across all studies that reported side effects in this systematic review, only two participants reported neck pain during stimulation (that was alleviated with analgesics). Other reported side effects included headaches, dizziness and nausea, mild skin burn, and excessive sleepiness; these side effects were reported in approximately 10% of participants and were mild, temporary, and alleviated by analgesics. Overall, non-invasive brain stimulation methods appear well-tolerated by individuals with schizophrenia and pose minimal safety risks.

To date, the optimal cerebellar stimulation parameters are unknown (15, 21, 44, 111). Efficacy of brain stimulation is determined by coil geometry (for TMS), stimulus intensity, duration and frequency of sessions, depth of the targeted tissue, and location of the cerebellar target. Research in this area is important for increasing efficacy ensuring patient tolerability and developing more personalized treatments.

Despite the cerebellum being a deep brain structure that requires cerebellar-specific stimulation parameters (e.g., coil types and stimulation intensity), most studies have followed standard parameters from cortical stimulation studies (112). Preliminary research on cerebellar-specific stimulation parameters has sought to identify the optimal TMS equipment for effective and tolerable stimulation. These studies compared different TMS coil shapes to find that double-cone (113, 114) and batwing (113) coils, which are designed to stimulate deeper tissue like the cerebellum (115), can effectively stimulate cerebellar targets, with the double-cone-coil being the most effective. Comparatively, one report concluded that the standard figure-of-eight coil produced unreliable results (113). Tolerability of the double-cone coil was significantly less than that of the figure-of-eight, and the authors therefore recommended a double-cone coil at 60% maximal stimulation output to balance reliability and tolerability. Of note, in our systematic review, one study used the angled double-cone coil while the remaining TMS studies used the standard figure-of-eight coil. As for location of the cerebellar target, most of the transcranial electric stimulation studies identified the cerebellar vermis as 1–2 cm below the inion, which is consistent with recommended practice and the majority of cerebellar stimulation studies (15, 33). However, some studies used different methods to identify the cerebellar target, such as MRI-guided neuronavigation (87, 89–91) and the 10–20 international EEG coordinate system (82, 86, 88, 92). This distinction is relevant as neuronavigation helps maximize the precision of the stimulation location for a given individual. Standardization is needed as electrode placement can impact the direction of the current flow direction and orientation of the electric field (15, 33). In line with this, there has been an effort to optimize and standardize procedures of transcranial electric stimulation for cerebellar targets (15, 116). These studies devise a protocol covering optimal electrode montages for cerebellar stimulation, for balancing optimal efficacy with minimal side effects.

Another technical issue to consider when conducting cerebellar stimulation RCTs is selection of the sham condition. There is great variability in sham methods employed in the field (e.g., similar sound and scalp contact but without stimulation, stimulation using the same pulse frequency but with the coil flipped, stimulation using a different frequency, etc.). Selection of the sham condition can lead to differential biological effects beyond the intended transient sensations, which in turn affect the results (117). For sham TMS, changing the position of the coil does not completely exclude residual brain stimulation, which is why one common method is to turn the coil upside down (118). Another recommended approach is to combine a purpose-built coil that mimics sound and scalp contact with surface skin electrodes that provide electrical stimulation time-locked to a TMS pulse (119). For sham tDCS, common approaches are to apply stimulation for a few seconds at the beginning of a session or to stimulate at a constant low intensity for the entire duration (118). It is important to ascertain the efficacy of blinding to condition for both participants and researchers (117), which the majority of the included RCTs did not do. While it has become more common to report on blinding success, this is not yet the standard in the field. Participants’ and researchers’ expectations regarding stimulation/treatment can produce placebo or nocebo effects that impact results (118).

A major advancement in this field (that is currently underway) is the implementation of RCTs to evaluate cerebellar stimulation in schizophrenia (11, 13). 57% studies included in this review were designed with an active arm only, meaning there was no control condition to determine the specificity of treatment effects. A strength of a recent study is the direct comparison of different cerebellar stimulation approaches [i.e., theta tPCS vs. delta tPCS (93)]. rTMS has been used the most frequently, especially iTBS which has relative advantages over traditional rTMS in that stimulation sessions are shorter, utilize a lower threshold intensity, and exhibit greater long-term excitatory meta-neuroplasticity (27, 28). It is unclear if using rTMS, particularly iTBS, is based on historical practice or if TMS is more effective than transcranial electric stimulation approaches (i.e., tDCS, tPCS, or even transcranial alternating current stimulation [tACS]) when targeting the cerebellum. These latter approaches could be advantageous as they are not known to induce contraction of neck muscles in patients (21). Moreover, these approaches are also less expensive, more portable, and have potential as in-home treatments.

It has also been argued that research linking clinical symptoms to neurobiological measures is hampered by research design obstacles, many of which were present across these studies (120). Notably, larger sample sizes with greater power are needed to establish the reliability of cerebellar stimulation effects. Most studies to date included fewer than 20 individuals with schizophrenia. Alternatively, standardization across sites and studies would allow for the pooling of data. This point is made not only for the stimulation methods/parameters, but also for the assessments, particularly cognitive batteries (as there is more consistency in the clinical symptom inventories used). Furthermore preclinical models of psychosis are needed to test mechanistic hypotheses of cerebellar stimulation.

Longitudinal designs that extend beyond 6 weeks can help clarify the longevity of effects and whether additional doses/boosters are needed. More nuanced longitudinal studies could also help to clarify whether there are individual plateaus in treatment effects, i.e., the subject-specific point after which there are diminishing returns. Available studies varied widely in terms of when they assessed treatment effects. While some studies assessed change throughout the stimulation period, others only compared pre- and post-completion timepoints.

Additional research is needed to understand who will most benefit from cerebellar stimulation (111). Many studies recruited individuals who were treatment-resistant (17–19, 82, 87, 92) or who had at least moderate symptoms (20, 90, 91); however, it is not clear whether these individuals were more likely to benefit from treatment than those with fewer symptoms. Evaluating individuals across the psychosis spectrum can help elucidate whether less symptomatic individuals or those earlier in the illness course can similarly benefit from cerebellar stimulation. For instance, Gupta et al. (94) provided preliminary evidence that cerebellar stimulation improved cognition in non-clinical high schizotypy individuals, whereas this effect was not present in other studies of chronic schizophrenia patients.

Another future direction is to combine cerebellar stimulation with multiple neuroimaging modalities (MRI, EEG) and behavioral tasks to drill down on the underlying circuits impacted by cerebellar stimulation (121). That is, single studies can benefit from the complementary spatial resolution of MRI and the temporal resolution of EEG to clarify how stimulation modulates specific cerebellar-mediated behaviors. One example is the prediction of self-generated stimuli that is feasibly measured using tasks where participants both vocalize brief sounds and listen to playback of themselves (122). The ability to anticipate self-produced auditory stimuli is notably impaired in schizophrenia, as evidenced by deficient suppression of auditory cortical signals measured with EEG and by failures to deactive auditory cortex (122–124). Importantly, this sensory prediction process is supported by an underlying cortico-cerebellar-thalamo-cortico circuit (10). Testing whether stimulation of the cerebellum can augment the underlying cortico-cerebellar-thalamo-cortical circuit and thus improve sensory prediction, is an important and novel future direction.

A caveat regarding the results of this systematic review is the potential for publication bias, especially since many of the older included studies were small open-label or case studies. Although a systematic review was conducted on multiple databases for published articles, studies finding null results might not have been published due to rejection based on small sample size or because the authors did not attempt to publish the results. As studies in the field shift to larger and more rigorous RCTs or longitudinal designs, which can provide more power for detecting effects and can reduce the probability of a Type II error that is more prevalent in small sample studies, the likelihood of publication of null results becomes greater.