The nystagmus Ménière memory model (NMMM) a dynamic central memory theory integrating peripheral vestibular fluctuations and vestibular recalibration

1 Introduction—the clinician's dilemma

Spontaneous nystagmus during Ménière's attacks may change direction over hours, sometimes persisting or fluctuating for days (Hallpike and Cairns, 1938; Phillips et al., 2021; Lee et al., 2024; Nadol, 1998). After an attack, caloric testing may demonstrate a deficit (Lee et al., 2024; Nadol, 1998) while HIMP-vHIT can remain normal (McGarvie et al., 2016). This apparent paradox—alternating “irritative” and “paretic” phases and discordant vestibular test outcomes—remains difficult to explain using static peripheral models.

*Several pathophysiological models have been proposed, including mechanical and ionic theories associated with endolymphatic hydrops (Nadol, 1998; Pender and Rajan, 1996; Agrawal and Minor, 2010). However, models that implicitly depend on increased hydrostatic pressure (Baloh, 1998; Vibert et al., 1999) as the primary driver are not supported by experimental evidence* (Long and Morizono, 1987). A time-dependent explanation that integrates peripheral fluctuation with central compensation processes may better account for the observed attack-phase dynamics.

Hypothesis. *We hypothesize that nystagmus reversals and test dissociation in Ménière's disease can be explained by interaction between (i) a fluctuating peripheral vestibular output and (ii) a centrally stored vestibular set-point (“central tonic memory”) that updates with delay*.

2 Vestibular signaling substrates: the two-cell framework and afferent dynamics

Within labyrinth, two populations of hair cells share the task of signaling movement:

Type I hair cell (calyx synapse). Flask-shaped cell ensheathed by a calyceal afferent. High resting K⁺ conductance (gK, L) → smaller but faster receptor potentials; with maturation, non-quantal transmission appears in addition to quantal transmission. The non-quantal mode combines K⁺ accumulation in the synaptic cleft (ms timescale) and ephaptic/resistive coupling (even faster), providing more faithful tracking of high frequencies. These specializations support “irregular” afferent responses to transients >5Hz, though they are not in a 1:1 exclusive correspondence with all irregular fibers (Vibert et al., 1999; Manzari, 2026).

Type II hair cell (bouton synapses). More cylindrical cell with multiple bouton afferent endings. CaV1.3-dependent glutamatergic quantal transmission via ribbon synapses, sustained and linear, suitable for rate coding and low frequencies (< 5 Hz); in adult mammals, quantal exocytosis remains dominant relative to Type I (Manzari, 2026).

2.1 Frequency-domain framing

Vestibular afferent firing can be described by functional response dynamics (Vibert et al., 1999; Manzari, 2026):

• Regular afferents predominantly operate as a rate code and are most sensitive to low-frequency signals (typically < 5 Hz).
• Irregular afferents provide a temporal (timing) code optimized for high-frequency transients and rapid changes (typically >5 Hz).

These dynamics are consistent with the synaptic specializations above, but do not imply strict receptor exclusivity. Many afferents are dimorphic, and hair-cell/afferent coupling is not one-to-one.

2.2 Crista organization and zonation

The crista is not functionally homogeneous. A central–peripheral zonation introduces an organizational bias: irregular afferents (high-frequency, timing-coded) are relatively enriched more centrally with proportionally greater Type I (calyx) input, while regular afferents (low-frequency, rate-coded) are relatively enriched toward more peripheral regions with proportionally greater Type II (bouton) input. This is a bias, not an absolute segregation, but it provides an anatomical substrate for frequency-weighted signaling and may contribute to differential sensitivity of vestibular measures during fluctuating peripheral states (Manzari, 2026).

In Ménière's disease, endolymphatic hydrops has historically been described as exerting pressure and altering ionic homeostasis (Pender and Rajan, 1996; Agrawal and Minor, 2010). Animal models demonstrate preserved pressure despite hydrops (Long and Morizono, 1987), and direct pressure measurements in humans are lacking. Current evidence instead favors altered volume regulation, epithelial remodeling, and ionic/metabolic dysregulation as the primary drivers of reversible cochleo-vestibular dysfunction (Böhmer, 1993; Merchant et al., 2005; Rauch, 2010; Bryton et al., 2025). These perturbations likely do not affect the two cell types equally; however, to avoid overreaching, we keep a cautious formulation and interpret the caloric–HIMP/SHIMP discordance (van Dooren et al., 2022; Roh et al., 2019; Manzari et al., 2022) as reflecting differences in afferent channels and central tasks, rather than an obligatory, selectively targeted lesion.

This dissociation explains the discordant tests: caloric and HIMP—SHIMP paradigms measured by vHIT, which interrogate slow, sustained—tonic mechanisms, fluctuate wildly; vHIT, which depends on the high-frequency transient—phasic channel, often remains deceptively normal. Thus, the two-cell story already hints at a dual physiology—the sustained phasic world of precision and the transient—tonic world of balance, each following different rules during hydropic stress.

*Note:* In the manuscript we use the “phasic stream” for signals rich in high frequencies/transients (often carried by irregular afferents) and the “tonic stream” for slow/integrative signals (predominant in regular afferents). “Phasic” denotes responses rich in high-frequency content/transients (often weighted by irregular afferent dynamics and Type I–weighted input). “Tonic” denotes slow/integrative components and sustained low-frequency processing (often weighted by regular afferent dynamics and Type II–weighted input).

3 From vestibular compensation to central vestibular memory

When one labyrinth falls silent (e.g., vestibular neuritis), vestibular compensation recalibrates baseline firing in vestibular nuclei under cerebellar influence. In sequential bilateral vestibular loss, a characteristic nystagmus may occur beating toward the side of the first lesion; this classical observation—Bechterew's phenomenon (Kim et al., 2022)—reflects unmasking of a previously compensated asymmetry.

We began to wonder whether Ménière's disease, with its fluctuating peripheral output, might constantly awaken and re-test this memory. *Each hydropic fluctuation could act like a miniature “second lesion,” prompting the central system to compare the new signal against its remembered set-point and to recalibrate again. Nystagmus reversals during Ménière attacks are better conceptualized as time-dependent mismatch between fluctuating peripheral output and a delayed central recalibration state, with similarities to recovery nystagmus dynamics (mismatch between recovering peripheral input and stored compensation) (Vibert et al., 1999; Kim et al., 2022; Lacour et al., 2016), without implying a Bechterew-like mechanism. Accordingly, Bechterew's phenomenon is referenced here strictly as a historical analogy illustrating central storage of compensation, and not as a mechanistic explanation of Ménière-related nystagmus reversals*.

4 The nystagmus Ménière memory model (NMMM)

We therefore propose the Nystagmus Ménière Memory Model (NMMM), in which a central tonic memory (set-point) interacts dynamically with peripheral vestibular fluctuations.

Sequence of events

• Initial irritative phase (relative peripheral overshoot). Peripheral vestibular output transiently exceeds the central set-point, producing nystagmus toward the affected ear. Mechanistically, this phase is hypothesized to be driven by ionic/metabolic fluctuation (Griffith and Wangemann, 2011; Singh and Wangemann, 2008; Wangemann, 2011) that increases excitability or alters baseline firing rather than pressure-mediated distension.
• Collapse phase (relative peripheral undershoot). As the overstressed type II cells fatigue and ionic gradients fail, peripheral output drops below the stored set-point (e.g., through transient fatigue of ionic gradients and/or altered transduction). Nystagmus reverses, now beating away from the affected ear.
• Post-crisis adjustment (delayed recalibration phase). Central recalibration follows with delay. Residual drifts or asymmetries may persist until the set-point updates to a new equilibrium. Across repeated attacks, delayed updating can yield hysteresis: system behavior depends on recent history and does not necessarily return to a single baseline after each episode (Figure 1).

*To facilitate comparison with existing concepts, Table 1 provides an overview of previously proposed mechanisms for direction-changing nystagmus in Ménière's disease, together with their strengths and limitations relative to the Nystagmus Ménière Memory Model (NMMM)*.

Table 1. Comparative overview of vestibular pathophysiological models and explanatory scope in relation to the nystagmus Ménière memory model (NMMM).

5 Clinical testing evidence—what the eyes reveal

If NMMM is correct, the dissociation between vHIT and SHIMP should mirror the dialogue between the phasic and tonic worlds. The vHIT, driven by type I phasic afferents, measures the raw mechanical fidelity of the semicircular canals. It answers the question: Can the system react fast? The SHIMP, in contrast, challenges the oculomotor system to suppress that reflex for a moving target, then generate an anti-compensatory saccade. It asks: Can the system predict, inhibit, and redirect? Those tasks rely on tonic control and cerebellar supervision—the very domains where memory and adaptation reside.

In clinical practice, we and others have often found SHIMP to be the more sensitive barometer of instability. Even when vHIT gains remain normal, SHIMP may show reduced VOR gain, altered anti-compensatory saccade amplitudes, or bilateral irregularities (van Dooren et al., 2022; Roh et al., 2019; Manzari et al., 2022; Yacovino and Finlay, 2016; Chen et al., 2021). This pattern supports the view that SHIMP does not merely probe the periphery but the central history of its calibration. Their operational difference is central: SHIMP requires VOR suppression and generates anti-compensatory saccades. Therefore, SHIMP can be more sensitive to central calibration instability even when HIMP gain is preserved. A particularly telling observation comes from Roh et al. (2019), who compared SHIMP and conventional HIMP protocols in various peripheral vestibulopathies. Their cohort included Ménière's patients, although the results were not separated by etiology. They found consistently lower SHIMP gains and a reduction of covert saccades, suggesting that SHIMP exposes residual function and central adaptation in ways that HIMP cannot. The absence of Ménière-specific analysis in that study leaves a crucial gap that our model helps fill.

Recent prospective data make the picture clearer (see Table 2). Kirbaç et al. (2024) studied 20 patients with definite unilateral Ménière's disease, confirmed by caloric paresis and sensorineural hearing loss. The SHIMP-VOR gain on the affected side averaged 0.69, compared with 0.77 on the contralateral ear, while vHIT gains were significantly higher on both sides (p < 0.001). Abnormal SHIMP results occurred in 50 % of Ménière ears vs. 35 % on vHIT; even the “healthy” ears showed 35 % SHIMP abnormalities compared with 10 % on vHIT. The authors concluded that SHIMP and vHIT parameters alone could not clearly distinguish symptomatic from asymptomatic ears—a finding that we interpret as evidence of central cross-compensation and shared memory effects.

Table 2. Numerical data summary—SHIMP vs. vHIT in definite Ménière's disease.

Complementary observations from van Dooren et al. (2022) and Manzari et al. (2022) confirm that SHIMP yields slightly lower gains than HIMP/vHIT and reduces covert saccades, reflecting its dependence on central predictive mechanisms. Together, these studies support the notion that SHIMP is not redundant but reveals the living memory of vestibular balance.

6 Discussion—vestibular adaptation as a central memory process

Established: Ménière's attacks can show direction-changing nystagmus and dissociation between calorics and vHIT (Hallpike and Cairns, 1938; Phillips et al., 2021; Lee et al., 2024; Nadol, 1998; McGarvie et al., 2016; Baloh, 1998; McCaslin et al., 2015). Vestibular compensation encodes a central calibration state influenced by cerebellar circuits (Vibert et al., 1999; Lacour et al., 2016). Regular and irregular afferents exhibit distinct dynamics with frequency dependence (Manzari, 2026).

Plausible interpretation: If peripheral vestibular output fluctuates over minutes to hours while central set-point recalibration evolves over hours to days, *then direction-changing nystagmus can reflect the time-dependent sign of mismatch between instantaneous peripheral output and the stored set-point*.

Hypothesis (NMMM): The observed sequence—initial nystagmus toward the affected side followed by reversal—can be explained by relative overshoot then undershoot of peripheral output relative to a lagging central set-point, with hysteresis across repeated episodes (Figure 1). Zonation within the crista provides anatomical plausibility for differential sensitivity of low-frequency vs. high-frequency readouts.

7 Clinical implications

For clinicians, the NMMM redefines how we interpret vestibular tests and nystagmus direction:

• – Nystagmus Direction is time-dependent. A reversal often reflects lag between changing peripheral drive and a slowly updating central set-point. Direction should be interpreted alongside timing relative to symptom onset (hours from onset, interictal, post-attack).
• – HIMP and SHIMP are complementary. Both share the same fast peripheral drive; SHIMP adds central VOR suppression and anti-compensatory saccade generation. Normal HIMP with abnormal SHIMP is compatible with central set-point instability rather than occult canal failure.
• – Calorics probe low-frequency dynamics. Disproportionate caloric variability with preserved HIMP is compatible with frequency-dependent instability rather than fixed loss.
• – Consider zonation as a plausibility framework. The peripheral (regular/Type-II-weighted) (Manzari, 2026) stream may be destabilized earlier than high-frequency transient drive.
• – Contralateral SHIMP findings require caution. They can reflect cross-compensation within shared central circuits and do not alone prove bilateral disease. Serial correlation with audiology and clinical course is required.
• – Therapeutic monitoring should be longitudinal. Serial HIMP/SHIMP tracking may be more informative than isolated measurements.

8 Future directions

The present work is a hypothesis framework and does not present a dedicated animal/human experimental validation. Prospective longitudinal studies in confirmed Ménière's disease, with symptom-timed serial HIMP/SHIMP and caloric testing, are needed to test NMMM predictions and estimate central adaptation time constants. Computational modeling of a bilateral integrator with delayed adaptation may provide quantitative predictions for reversal latencies and hysteresis patterns. Where feasible, coupling longitudinal physiology with imaging/histopathology may help relate structural and microenvironmental changes to function.

9 Conclusion

We propose a time-resolved view of Ménière's disease in which direction-changing nystagmus and vestibular test discordance reflect interaction between fluctuating peripheral vestibular output and a centrally stored set-point that recalibrates with delay. The model avoids pressure-based assumptions and instead emphasizes ionic/metabolic mechanisms as plausible drivers of reversible dysfunction. Crista zonation provides anatomical plausibility for differential sensitivity of frequency-weighted readouts. NMMM generates testable predictions and supports longitudinal, symptom-timed interpretation of vestibular tests.

Hydropic-state fluctuation is conceptualized as modulating Type II–weighted tonic components and interacting with a central tonic set-point. Relative overshoot produces nystagmus toward the affected ear; relative undershoot produces nystagmus away from the affected ear. Central recalibration follows with delay, creating hysteresis across repeated episodes. Regular (rate) and irregular (timing) afferent dynamics weight the instantaneous error vs. its change over time differently; non-quantal transmission at the Type I calyx supports transient tracking, whereas quantal Type II transmission supports sustained integration.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. The author(s) used an AI-based language model (ChatGPT, OpenAI) solely for linguistic editing, structural organization, and clarity improvement during manuscript preparation. All scientific concepts, hypotheses, interpretations, and final editorial decisions were conceived, evaluated, and approved exclusively by the author(s).

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Alexander A. Tarnutzer, University of Zurich, Switzerland

Reviewed by: David Bächinger, University Hospital Zürich, Switzerland

Xiaobang Huang, Tianjin First Central Hospital, China