Tilt Suppression of the Vestibulo-ocular reflex

Timothy C. Hain, MD • Page last modified: March 7, 2021

Hain TC, Zee DS, Maria B: Tilt-suppression of the vestibulo-ocular reflex in patients with cerebellar lesions. Acta Otolaryngology. (Stockh), 105:13-20, 1988.

In this page, we reprise the results of our 1988 paper, include in a model that was not published in the original paper, and have added some results of subsequent workers on tilt suppression.


The effect of tilt on the time constant of postrotatory nystagmus was determined in a group of normal subjects and patients with cerebellar lesions. The normal subjects showed a vestibular response that decayed with a time constant of 19.6 seconds when upright and 7.2 seconds after tilt prone. Patients with midline cerebellar lesions near the uvula and nodulus had time constants that were normal in the upright position but were unaffected by head tilt. Patients with cerebellar lesions due to the Arnold Chiari malformation showed behavior that was intermediate between that of normal subjects and the patients with midline lesions. These results provide evidence that in humans, the midline cerebellum regulates a neural network in the brainstem that perseverates peripheral vestibular input.


In normal human subjects, rotation in darkness at a constant velocity induces a vestibular nystagmus that decays approximately exponentially. When the head is upright and the rotation is about earth vertical, the dominant time constant of the decay is about 20 seconds. When the head stops, a postrotatory nystagmus ensues with the same time constant. If the head is tilted prone just after the head is brought to rest, that is, at the onset of postrotatory nystagmus, the time constant of the vestibuloocular reflex (VOR) is diminished from approximately 23 to about 7 sec (Benson and Bodin, 1966).


Tilt suppression from Benson and Bodin 1966
Tilt suppression from Benson and Bodin, 1966

This phenomenon has been termed "tilt suppression" or "dumping" (Raphan et al, 1981). The vestibular response is obtained from activity in peripheral vestibular afferents modified by central circuits. To create the observed behavioral time constant of 20 seconds, which is based on the nystagmus response, the central mechanism must "store" the labyrinthine signal due to deflection of the cupula, which has a time constant of only about 6 seconds. For this reason the central processing has been likened to a "storage mechanism" that increases the ability of the vestibular system to respond to the low frequency components of head rotation (Raphan et al, 1979). Tilt suppression must derive from changes in velocity storage as afferent activity in vestibular nerve fibers that innervate the lateral canals is mainly determined by the mechanical characteristics of the canals and not appreciably affected by linear acceleration, i.e., gravity ( Goldberg and Fernandez, 1975). Furthermore, tilt suppression must be based upon a signal that reflects reorientation of the head with respect to gravity. Most evidence suggests that the otoliths are the source of the signal (Guedry 1965; Hain 1986).

Progress has been made in determining the anatomical and physiological substrate of tilt suppression. In monkeys, Waespe and coworkers (1985) have demonstrated that ablation of the cerebellar nodulus and uvula abolishes tilt suppression. Furthermore, electrical stimulation of this area causes a decrease in the vestibular time constant (Fernandez and Fredrickson, 1974). These findings suggest that activity in the nodulus and/or uvula impairs velocity storage. They also suggest that human beings with cerebellar lesions might show abnormal tilt suppression and we tested this prediction in patients.



MRI of one of the patients studied here with a midline cerebellar lesion. She had a medulloblastoma removed.

Sixteen patients with cerebellar lesions were studied. These included ten patients with midline cerebellar lesions due to tumors (mainly medulloblastomas) and six patients with the Arnold Chiari malformation (type I). The location of the lesion was determined by computed tomographic or magnetic resonance scans and from operative reports.? No patients had ocular motor palsies. ?? All but 2 of the patients with cerebellar lesions due to tumors had been operated upon and completed radiotherapy.? Seven normal subjects (laboratory personnel) were the control population. ? The figure above shows one of the medulloblastoma patients, where a cavity is seen in the midline cerebellum after treatment.

In all subjects the time constant of the horizontal VOR during postrotatory nystagmus was measured in two head positions? upright and face down (prone). This is similar to the methology of Benson and Bodin (1966). In order to induce the vestibular response the subject was seated upright in a motorized chair with the head stabilized from the back at two points. Subjects were instructed to remain alert and to attempt to view objects in the room as they rotated, even though it was completely dark. The chair was accelerated (100 deg/sec**2) in the dark to 60 deg/sec and maintained at that velocity while the perrotatory response was recorded. After three minutes or after the reversal phase had died out, whichever was longer, the chair was stopped and the postrotatory response recorded. In order to measure the effects of tilt prone, this procedure was repeated, but, just after the chair had stopped, the subject bent forward using a combination of a head and a trunk movement until the coronal plane of his head was parallel to the floor. The time at which the tilt occurred was determined from the transient change in vertical eye position that was induced by the tilt. One trial was obtained for each final head orientation and direction of rotation.

The time constants were obtained from eye movement data. The combination of all perrotatory and postrotatory responses to rotation while upright afforded six measurements of the upright time constant, which were averaged. The time constant following tilt prone was determined once for each direction and these values were also averaged.


Normal subjects?

The figure above shows the results from the seven normal subjects. Their average time constant with the head upright was 19.6 + 5.4 sec (mean + standard deviation). It was decreased to 7.2 + 1.8 sec by tilt prone (p < .001, t-test). Considering the subjects individually and calculating the change in time constant by dividing the difference between the upright and prone values by the upright value, the average percentage decrease in time constant was 62.0 + 7.4%.

Tilt suppression in Midline Cerebellar Lesions??

The figure above shows the results in 10 patients with midline cerebellar lesions (mainly medulloblastomas). For this group, the time constant upright (mean 19.5 + 3.0 sec) was not different from normal. Tilt reduced their time constants to an average of 17.4 + 4.2 sec, not significantly different from their upright values (p > .2). The values after tilt were, however, significantly different from the time constant after tilt of normal subjects? (p < .001). The average percent decrease in time constant evoked by tilt in the group with midline lesions was 9.4 + 21%. Only patient 4 had a percent decrease in time constant that fell within three standard deviations of the mean decrease found in normal subjects. Again, only patient 4 had a time constant in the prone position that was within three standard deviations of the mean time constant in the prone position found in normal subjects.

????Tilt suppression in Chiari Malformation

The figure above shows the time constants found in 6 patients with the Arnold-Chiari malformation. The mean time constant obtained in the upright position, 16.7 + 2.6 sec, was not significantly different from the value in normals. The time constant after tilt was 11.3 + 3.4 sec which is significantly less than the analogous response in normal subjects (p < .05). It corresponds to an average decrease in time constant of 32.4 + 15.7%. Two of these patients, 11 and 12, had tilt suppression by percent to within 3 standard deviations of the normal mean. All of these patients except number 15 had a mean time constant in the prone position that was within three standard deviations of the mean found in normals.


Our results in 1988 showed that tilt from upright to prone reduces the time constant of the vestibuloocular reflex (VOR) by about 2/3 in normal subjects, about 1/3 in patients with Chiari malformations, and essentially not at all in patients with large midline cerebellar lesions.? The dramatic effect of tilt??to prone upon the VOR of normal subjects is consistent with previous reports of tilt suppression (Benson and Bodin, 1975).? ? The finding that patients with midline cerebellar lesions do not show tilt suppression is consistent with the compelling evidence in experimental animals that suggests that the midline cerebellum and especially the nodulus and uvula regulate the velocity-storage mechanism and consequently the low-frequency response of the VOR.? In particular, lesions of the nodulus abolish tilt suppression of the VOR, increase the duration of rotatory and caloric responses and abolish the decline in duration of vestibular responses that occurs after repeated stimulation (habituation) (Waespe et al, 1985; Fernandez and Fredrickson, 1974; Singleton, 1967).? Stimulation of the nodulus/uvula complex in monkeys alters the low-frequency response of the VOR by decreasing its time constant (Solomon et al, 1985).???? Subsequent work to this has supported the conclusion that uvulo-nodular lesions abolish tilt suppresion (Wiest et al, 1999).

Subsequent work has documented that static tilts in pitch also suppress human optokinetic velocity storage. (Lafortune et al, 1990). Lee et al (2017) mapped out the structures that affected tilt suppression in 73 patients with cerebellar lesions and reported "Probabilistic lesion-mapping analysis showed that the nodulus and uvula are responsible for tilt suppression. "

In monkeys, Cohen et al (2002) reported concerning their work in monkeys that "after removal of the nodulus and rostral-ventral uvula, the spatial orientation of eye velocity to the GIA is lost and that eye velocity is then purely driven by the semicircular canals in a body frame of reference."

The nodulus/uvula complex has the appropriate connections to allow vestibular responses to be modulated by otolith inputs as there is both anatomical and physiological evidence that the nodulus and uvula receive input from the otoliths as well as project to the vestibular nuclei.? Most of the primary vestibular fibers that project to the cerebellum terminate in the "vestibulocerebellum" which consists of the flocculus, nodulus and uvula (Brodal, 1974). The primary projections of the otoliths to the cerebellum follow similar pathways (Carpenter et al, 1972). Secondary vestibular fibers are distributed more widely but again project mainly to the vestibulocerebellum.? Most secondary fibers take origin from caudal regions of the descending vestibular nucleus. In the cat, the uvula and nodulus project back to the caudal portions of the vestibular nuclei as well as to the nucleus prepositus hypoglossi (NPH) (Brodal, 1974).? Physiological studies suggest that vestibular stimulation affects neuronal activity in a large cerebellar cortical distribution including the entire vermis, flocculus, lingula, lobus simplex, and nodulus (Gilman et al, 1982).?


Model of dumping
Hypothetical model of how tilt suppression occurs in the brainstem and cerebellum

How does the vestibulocerebellum mediate tilt suppression?? Our concept of this is shown in the block diagram above. The low frequency response of the VOR is provided by the velocity storage mechanism (Raphan et al, 1979) and therefore tilt suppression is likely caused by influence of the vestibulocerebellum upon this central circuit.? However, the precise site in which velocity storage takes place has not yet been identified. The velocity storage mechanism must be located outside of the vestibulocerebellum as lesions therein do not abolish velocity storage. Two sites in the brainstem appear likely. The nucleus prepositus hypoglossi (NPH) is a possible site as it exhibits neural activity with appropriate frequency characteristics (Marini et al, 1975). Furthermore, NPH has reciprocal connections with the descending and medial vestibularnuclei (McCrea et al, 1979). Against this hypothesis is the observation that the nodulus and uvula send relatively few fibers to NPH (Marini et al, 1975).

Another possible site of velocity storage is the descending vestibular nucleus (DVN). Although DVN has abundant connections to other vestibular nuclei and to the cerebellum, it is not directly connected to a motor system such as the oculomotor nuclei or the spinal cord (Brodal, 1974). However, DVN may be a secondary processor of activity? a role which the velocity storage mechanism plays in most current models of the vestibuloocular reflex (Raphan et al, 1979; Hain et al, 1986).? This hypothesis would explain the observation that although stimulation of the nodulus inhibits DVN strongly (Precht et al, 1976) such stimulation has only a weak direct effect on the discharge of oculomotor neurons (Precht et al, 1977).

Finally, we must ask why the patients with Arnold-Chiari malformation exhibit a decrease in tilt suppression.? The ocular motor abnormalities that occur in this malformation are usually attributed to the abnormalities of the flocculus and paraflocculus and experimental lesions of these structures do not impair tilt suppression of the VOR (Waespe et al, 1983).? Both secondary compression of the medulla and downward herniation of the uvula and nodulus may be possible explanations.

In summary, these results provided early evidence evidence that in humans, the midline cerebellum, including the uvula and nodulus, regulates a neural network in the brainstem that perseverates peripheral vestibular input (i.e. the velocity storage system).



This page is an edited version of a paper published in 1988 (Hain, Zee, Maria). The figures were taken from a draft of that manuscript, and some simplification has been made in the text as well as updates to more contemporary references. We intend to update it as time permits and more evidence is provided in the literature.