The vestibular nerve transmits self-motion and orientation information from the vestibular end organs (the semicircular canals, saccule, and utricle) to the brainstem, cerebellum, and cerebral cortices. This transmission is crucial for oculomotor and postural control, as well as spatial orientation. Additionally, branches of the vestibular nerve project to the mastoid region behind the ear. It is possible to stimulate the vestibular system by applying a low-intensity electric current to the vestibular nerve. This is typically accomplished by placing electrodes on the mastoids and delivering transcutaneous current to local vestibular afferents. This noninvasive form of stimulation is known as Galvanic Vestibular Stimulation (GVS), named after Luigi Galvani who used electric current to evoke frog muscle contractions (Galvani, 1953; Dlugaiczyk et al., 2019). It has been used for its various diagnostic and therapeutic capabilities as well as to investigate vestibular physiology. For example, it can induce a complex whole-body response of oculomotor and balance reflexes, which vary greatly depending on the waveform and other stimulation parameters.

While much of the research on GVS in humans has been restricted to non-invasive methods due to ethical and practical considerations, substantial invasive studies on animals have provided deeper insights into the mechanistic effects of GVS. Johann Purkinje first noted in 1820 that passing electric current through the human head led to balance and equilibrium disturbances (Purkinje, 1820). Later, Josef Breuer attributed these effects to the galvanic stimulation of the vestibular system, observing distinct head movements resulting from the stimulation of individual semicircular canals (SCCs) in birds (Breuer, 1874). Many years later, Chapman et al. (2019) applied targeted GVS to rat SCC nerves and found higher concentrations of c-Fos protein in the contralateral medial vestibular nucleus. These findings by Breuer and Chapman et al. suggest that GVS stimulates the SCC rather than the otoliths. On the other hand, Holstein et al. (2012) applied low-frequency sinusoidal GVS (0.02–0.04 Hz) to rats and found c-Fos protein in the nuclei of otolith-driven neurons, indicating that GVS serves as an otolithic stimulus. To resolve this inconsistency, Kwan et al. (2019) conducted a study in which they inserted recording electrodes into the vestibular nerves of rhesus macaques to measure the activity of individual afferents. Their findings demonstrated that both otolith and SCC afferents are activated by sinusoidal GVS. These results suggest that GVS does not replicate natural head motion as closely as may have been previously assumed, as regular head rotations typically do not stimulate both otoliths and SCC simultaneously.

Nguyen et al. (2021) investigated the effects of GVS on mice following unilateral labyrinthectomy. They found that 1 Hz sinusoidal GVS at 0.1 mA accelerated the recovery of spatial memory and locomotor function in these mice, indicating a potential therapeutic benefit of GVS in vestibular dysfunction. In another study, Sabzevar et al. (2023) applied GVS to rats in direct current (DC) pulses at the minimum intensity that induced eye movement and discovered electrophysiological signals in the tail of the striatum, a recently discovered sensory region, suggesting that vestibular projections also extend into this area. These studies highlight the potential of GVS to influence not only vestibular functions, but also broader neural circuits involved in spatial memory and sensory processing.

The complexity of vestibular signaling and the effects of GVS contribute to significant variability in observed outcomes, which depend on factors such as waveform, anatomical location of stimulation, and other selected parameters.

The mechanisms of suprathreshold and subthreshold are presumably different. Suprathreshold GVS, delivered in either DC or sine waveform, modulates the firing rates of vestibular afferent neurons in a manner that is directly proportional to the applied current, as demonstrated in neural recordings from non-human primates (Kwan et al., 2019; Forbes et al., 2023). This neural activation is subsequently transmitted to the vestibular nuclei in the brainstem, where it mediates a complex whole-body response involving balance reflexes. These reflexes are influenced by various factors, including the head and body’s orientation and position, the specific task being performed, and the sensory input received from other sensory modalities (Fitzpatrick and Day, 2004). In response to this perception, oculomotor and postural responses are triggered directed toward the anodal current (Fitzpatrick and Day, 2004).

Conversely, subthreshold GVS when delivered as a noisy waveform (nGVS) is thought to act by inciting stochastic resonance, increasing the discharge and resting state activity of vestibular afferents, thereby improving neural activation “information transfer” even at intensities that do not cause gross motor activity (Wuehr et al., 2016b). Behavioral responses have been observed that are consistent with this hypothesized mechanism, such as reduced vestibular perceptual thresholds with moderate amplitudes of nGVS (Galvan-Garza et al., 2018), and improvements in balance in both healthy (e.g., Keywan et al., 2018) and diseased populations (e.g., Fujimoto et al., 2018). However, to our knowledge, no studies have quantified vestibular afferent neuron firing rates during nGVS application in non-human primates, as has been done with suprathreshold DC or sine GVS (Kwan et al., 2019; Forbes et al., 2023).

Electrode positioning strongly impacts the induced current flow pattern and electric field at the vestibular organs of interest (Thomas et al., 2020; Truong et al., 2024). This consideration is critical as current flow is believed to modulate the firing rate of vestibular hair cells that is ultimately relayed to the vestibular nuclei in the brainstem (Wardman and Fitzpatrick, 2002). Functional MRI has revealed that bilateral GVS stimulates the temporoparietal junction, central sulcus, anterior interior intraparietal sulcus, and premotor regions of the frontal lobe (Lobel et al., 1999).

Consequently, it is crucial to establish stimulation parameters tailored to the desired application to ensure accuracy and reproducibility. To identify the optimal waveform parameters, we conducted a systematic review of clinical trials on GVS according to the PRISMA guidelines (Page et al., 2021) and extracted stimulation parameters relevant to each application. While GVS is employed both as a diagnostic tool (Dlugaiczyk et al., 2019) and as a method to enhance or modulate balance in healthy individuals (e.g., Wuehr et al., 2016a; Brunyé et al., 2024), this review specifically concentrates on its therapeutic applications. Additionally, although other forms of vestibular stimulation, such as thermal (e.g., caloric) and invasive techniques, are worthy of consideration, our focus is on conventional, non-invasive GVS.

We conducted a literature search of the PubMed and Embase databases for clinical trials with the keyword “galvanic vestibular stimulation” up to July 1, 2024. We included research articles, letters, and case studies, but excluded literature reviews and abstracts. A research team member screened each result and removed duplicates, studies that used GVS solely for diagnostic or screening purposes, and studies that applied GVS simply to enhance functions in healthy subjects rather than to treat a specific symptom or disorder. We segmented the remaining results into three groups: (1) GVS for treating impaired postural control, (2) GVS for post-stroke patients, and (3) GVS for other clinical applications. We extracted the stimulation parameters of each study as well as the study design and number of diseased patients in each arm. We only report that a study was double-blind when the authors mention that the investigators were blinded to the treatment conditions. Otherwise, we assume that blinded studies were only single-blind.

The stimulation parameters included electrode montage, waveform, frequency, intensity, sham procedure, stimulation length, number of sessions, and inter-session interval as follows:

Our search returned a total of 239 results. We then removed 92 duplicates, 88 articles that did not meet the inclusion criteria, and 2 papers that were not available (Figure 3). Among the remaining 57 publications, four of them were secondary analyses of other studies. This left us with 53 independent studies discussed below.

The studies discussed demonstrate that there is a notable divergence in the selection of stimulation parameters. Some parameters are widely accepted, while others remain poorly defined and require further investigation to identify the most effective settings. Studies on impaired balance usually employed noisy waveforms, driven by the beneficial effects of input noise in enhancing neural system sensitivity. This occurs through noise-enhanced responses of nonlinear systems to weak signals, a phenomenon known as stochastic resonance (Wiesenfeld and Moss, 1995). It is hypothesized that a central circuit involved in signaling movement initiation, with a pathologically elevated threshold, may benefit from noisy modulation of afferent firing rates (Pan et al., 2008). Conversely, for treating impaired verticality, studies often employed DC stimulation to counteract the misalignment (Oppenländer et al., 2015a).

With respect to current intensity, the choice has varied between choosing an individualized optimal intensity (i.e., maximizing an individual’s performance on a task), individualized subthreshold intensity, or a fixed intensity. This decision is guided by the exact outcome measure being tested and as applicable, highlights the need for personalization, as the same intensity can elicit different vestibular sensations and physiological effects in different subjects, including within the same patient population.

As expected, the choice of waveform has a bearing on the length of the stimulation session. As noisy GVS at imperceptible levels does not cause noticeable side effects in the form of oculomotor and postural responses, some studies have delivered stimulation spanning hours to an entire day. When using DC, studies have generally not exceeded the 30 min duration and inter-session duration of 1 day based on the safety criteria and protocols in place for the more thoroughly researched tDCS modality.

Repeated applications leading to cumulative and/or sustaining effects are well documented in other brain stimulation modalities (e.g., Razza et al., 2020; Black et al., 2023; Valter et al., 2024). However, most GVS studies have been limited to single session protocols. As such, there is a gap in our understanding of the long-term and cumulative effects of GVS. This highlights the need for comprehensive research to establish the appropriate treatment indication and protocol, and to examine the cumulative and long-term impacts of GVS.

In conclusion, certain GVS parameters are well-accepted for specific applications, such as the need to deliver a noisy stimulus waveform to improve postural control in Bilateral Vestibulopathy, whereas other parameters still lack consensus, such as the number of stimulation sessions necessary for improving spatial neglect in post-stroke patients. Also, a wide variety in noise frequency ranges have been used for improving postural control in subjects with vestibular disorders. While the wide range makes it difficult to determine the most effective frequency band for a given indication, it also demonstrates that a wide range of parameters are biologically active. Additional studies are therefore necessary to clarify the optimal frequency for each population and application.

GVS continues to hold significant potential for clinical utility. Furthermore, owing to low intensity current needs, GVS devices are portable and can be therefore administered across a wide variety of settings such as laboratory, clinical and home locations, mimicking protocols developed in other non-invasive brain stimulation modalities (Valter et al., 2021; Pilloni et al., 2022; Black et al., 2023). Research has also demonstrated that GVS application is even feasible in spaceflight due to its low mass and power requirements, low power usage, and minimal side effects (Putman et al., 2021; Sherman S. et al., 2023; Sherman S. O. et al., 2023). Given the safety profile, protocols can explore delivering multiple sessions over weeks and thereby expect developing clinical utility in myriad conditions. Our review serves as a suggestion for future GVS research protocols and clinical applications with respect to the choice of stimulation parameters and related considerations for specific patient populations and conditions.

Results of this review indicate that the field of GVS is still evolving, with no universally standard application guidelines. Instead, current evidence appears to be condition-dependent or patient population-specific, limiting making conclusions on an overall all-cause acceptable GVS parameter protocol, but does align more favorably with the changing clinical landscape of precision medicine practices. Nevertheless, elucidative information has surfaced from the current review that serves as the basis for early key recommendations, critical to moving the field of GVS forward. The following are those recommendations that scientists can use as reference when designing GVS research studies. First, it is essential to perform (at least one) sham condition, appropriate as a control condition for the study’s scientific objectives, and provide sufficient methodological details regarding the sham (e.g., was current applied away from the mastoids, was no current applied, were electrodes on, was the subject blinded to the GVS application, etc.). This is particularly critical for studies investigating after-effects or the longitudinal application of GVS, where there is otherwise a confound between the presence of GVS and time (which could impact learning, fatigue, boredom, etc.). We also encourage investigators, particularly for subthreshold GVS, to assess whether subjects are able to feel tingling sensations or other side effects from the GVS application, in a two-alternative forced choice test (where GVS is presented in one interval and not in the other, and the subject is required to guess which interval contained the GVS), and investigators ensure the recognition performance is no better than chance.

Second, while the literature suggests that GVS applications may require (or benefit) from personalization to each individual, this can incidentally provide investigators excessive opportunities for identifying a positive effect of GVS. For example, testing each subject with multiple amplitudes of GVS (to identify which is preferred for each individual) and then comparing each individual’s optimal amplitude to a sham condition can increase potential for false positive findings. Personalization procedures therefore need to be carefully pre-defined to ensure scientific rigor. Third, future research should aim to systematically compare multiple GVS applications (waveform, frequency, intensity, etc) in the same study for a given patient population and outcome. Nearly all prior research has only assessed one particular GVS application, requiring the field to compare across studies, performed in different laboratories, with nuanced differences in assessments, metrics, timing, and other procedures. Current, variegated approaches across studies preclude determining ‘best’ sets of GVS parameters given a specific use case. Finally, to aid in standardized GVS applications across studies, we encourage the publication of null result outcomes from GVS studies. By publishing these findings, more comparisons can be made between GVS applications that otherwise may not be noted in the literature. Further, this approach will provide advanced guidance on what GVS applications may not be useful for specific patient populations and outcomes.