Dissociated Vertical Divergence: Perceptual Correlates of the Human Dorsal Light Reflex

Background  Dissociated vertical divergence (DVD) has been attributed to a human dorsal light reflex that emerges when single binocular vision is precluded in infancy. If this is the case, then DVD should be associated with a subjective sensation of tilt.

Methods  Prospective examination of 9 patients with DVD and 9 control subjects to determine whether monocular occlusion and alternate occlusion induces a subjective sensation of visual tilt or body tilt.

Results  Alternate occlusion disclosed a tilt in the subjective visual vertical in 8 of the 9 patients with DVD and in none of 9 control subjects. On occlusion of the fixating eye, a vertical pencil positioned in the sagittal plane was perceived as instantaneously tilted, with its upper pole tipped toward the side of the covered eye. This visual tilt was quickly followed by a perceived rotation back to vertical, which coincided with the dorsally directed drift of the covered eye.

Conclusion  In patients with DVD, monocular occlusion is associated with a subjective visual tilt that is annulled by a cyclovertical divergence movement of the eyes. This observation supports the notion that DVD is a human dorsal light reflex, which functions to restore vertical visual orientation when unequal binocular visual input evokes a subjective sensation of visual tilt.

IN LIVING ORGANISMS, light from the sky above and gravity from the earth below have led to the evolution of sensory organs for vision and balance. The bright sky serves as a hemispheric light source that provides a stable visual reference for which way is up. In lower animals, the central vestibular system integrates visual input from the 2 eyes and graviceptive output from the 2 labyrinths to modulate postural and extraocular muscle tonus and maintain vertical orientation.^1-3 Although visual input is usually subordinate to vestibular input in establishing postural orientation, some lateral-eyed animals also maintain vertical orientation by equalizing visual input to the 2 eyes.⁴ Many fish and insects exhibit a dorsal light reflex in which illumination from one side evokes a reflex body tilt toward the light.^4-7 When a light is shined down from the right side, for example, the right eye receives greater visual input than the left eye (Figure 1). This binocular disparity would only exist in nature if the animal were tilted with its right side toward the sky. This visual imbalance causes the central vestibular system to register a leftward body tilt relative to the body position that would be necessary for the 2 eyes to receive equal binocular visual input, and to reflexively alter postural tonus to correct the tilt^1-4(Figure 1). In a vertically stabilized fish, the same stimulus evokes a vertical divergence of the eyes to reorient the interpupillary axis relative to the new light source, causing the eye with lesser visual input to shift dorsally, and the eye with greater visual input to shift ventrally³ (Figure 1).

The term dorsal pertains to the back or upper aspect of an animal. The dorsal aspect of the head in fish, quadrupeds, and bipeds corresponds to the top of the head. In bipeds such as humans, the back retains its phylogenetic dorsal orientation, although it is no longer the upper aspect in the upright position. Humans with congenital strabismus who never develop single binocular vision exhibit an atavistic resurgence of the dorsal light reflex in the form of dissociated vertical divergence (DVD).⁸ When these strabismic humans fixate monocularly, the nonfixating eye exhibits a slow dorsal rotation termed dissociated vertical deviation. This dorsal rotation can be elicited by optically or mechanically reducing visual input to either eye. It can also occur spontaneously when there is a fluctuating degree of sensory suppression in one or both eyes.⁸ Eye movement recordings have confirmed that DVD comprises a cyclovertical divergence in which the fixating eye depresses and intorts while the nonfixating eye elevates and extorts.^9-11 Thus, the alternative term, dissociated vertical divergence, provides a more accurate mechanistic description of this movement.⁸

In fish, abrupt fluctuations in binocular visual input cause the central vestibular system to register a tilt, and the ensuing dorsal light reflex serves to annul this tilt by realigning the eyes and body to new a orientation that the brain interprets as vertical. If DVD is a human dorsal light reflex, then humans with DVD might be expected to experience a perceived visual tilt(ie, a sensation that the visual environment is tilted) when the gradient of visual input to the 2 eyes is abruptly altered. In this case, the cyclovertical divergence associated with DVD should serve to reduce or eliminate the perception of tilt. Alternatively, if DVD is not a human dorsal light reflex, then the torsional component of DVD might be expected to induce a perception of visual tilt. In this case, patients should perceive no visual tilt prior to the onset of DVD, but a perceived visual tilt should be present after the DVD has occurred. To elucidate the perceptual correlates of DVD, I examined visual tilt perception in 9 patients with DVD and 9 normal control subjects.

Nine patients with bilateral DVD were examined prospectively and consecutively. Children who had been treated with previous cyclovertical muscle surgery were excluded from the study since the resulting fundus torsion could influence the perception of tilt. Dissociated vertical divergence was diagnosed when occlusion of either eye evoked a dorsal rotation of the occluded eye. The amplitude of the DVD was measured in both eyes using the vertical prism under cover test. Visual acuity was measured in each eye, versions were examined, field measurements were obtained, and a dilated retinal examination was performed to look for static torsion.

To determine whether patients experienced a perceived visual tilt (ie, a perceived tilt of the visual environment without any sensation of body tilt) or subjective tilt (ie, a perceived tilt of the body) under monocular conditions, each patient was instructed to view a pencil held vertically in the sagittal plane midway between the 2 eyes during occlusion of each eye and during alternate occlusion of the eyes. The patient's head was maintained in the upright position before and during testing. Care was taken to assure that the pencil was not slanted forward or backward in the sagittal plane, as a slant would optically induce a monocular image tilt. Each patient was instructed to hold up his or her index finger and to move this finger to demonstrate any perceived movement of the pencil as each eye was occluded. The cover test and alternate cover test were repeated 3 to 4 times in each patient to assure that perceptual responses were consistent from one trial to the next. When a perceived motion was noted, the patient was specifically asked whether the pencil appeared to move sideways or to tilt. The patient was then asked, "does it look like the pencil is moving, or does it feel like you are moving?" A labyrinthine imbalance such as that brought on by spinning or by disease of the semicircular canals, is generally perceived as a sensation of body movement in space. Thus, a perceived sensation of body movement in space would suggest that unequal visual input is inducing a labyrinthine imbalance, while perceived visual tilt with no sensation of body movement would suggest that the neural pathways activated by unequal visual input do not directly alter the relative output of the 2 labyrinths. This test was also performed in 9 control subjects who had normal stereopsis and no history of strabismus.

Patient ages ranged from 4½ to 51 years. All but 1 patient was younger than 15 years. Seven patients had a history of congenital strabismus that had been treated with only horizontal muscle surgery. One child with prominent DVD had a history of perinatal bruising of both eyelids, but had never developed congenital esotropia, latent nystagmus, or nasotemporal asymmetry.(The term nasotemporal asymmetry refers to a disparity in monocular nasally directed and temporally directed optokinetic responses, with normal nasal optokinetic responses, and impaired temporal optokinetic responses. This asymmetry is normal within the first 6 months of life and is retained throughout life in patients with congenital esotropia.) One adult with no history of horizontal strabismus also had bilateral DVD with no latent nystagmus or nasotemporal asymmetry. All patients were neurologically normal except for one prematurely born child with congenital esotropia and delayed walking who probably had periventricular leukomalacia.

The amplitude of the DVD in the 2 eyes was symmetrical in 5 patients and asymmetrical in 4 patients. Six patients had grossly visible latent nystagmus. Three patients had bilateral inferior oblique muscle overaction with a "V" pattern and bilateral static extorsion of the globes. The one patient with a history of prematurity had bilateral superior oblique muscle overaction with an "A" pattern and intorsion of the globes.

On monocular occlusion, 8 patients reported an instantaneous tilt of the pencil with its top tipped toward the side of the covered eye (Figure 2). This immediate perception of a tilt seemed to precede any motor movement of the uncovered eye and was probably caused by a change in perception only. This visual tilt was followed quickly by a perceived rotation of the pencil back to vertical, which coincided with a dorsally directed drift of the covered eye. (Note that if the fixating eye is making a torsional movement, the subject will perceive the pencil as moving in the opposite direction that the eye is rotating. Thus, if the subject sees the rotation of the pencil such that the top goes from being nasally tipped to straight up, this means that the 12-o'clock meridian of the cornea is rotating nasally. So an apparent extorsional movement of the pencil would correspond to an intorsional movement of the globe. This intorsional movement in the fixating eye begins the phenomenon known as DVD.^9-11)The one patient who did not report a perceived tilt was the adult who had idiopathic DVD and no other history of strabismus.

In patients who had markedly asymmetrical DVD, the visual tilt was usually observed only when the dominant eye was covered. All 8 patients who reported a sensation of visual tilt denied any sensation of head or body tilt. No spontaneous head tilting was observed during alternate cover testing. Three patients also perceived a sideways movement of the pencil during alternate occlusion. One adult with DVD perceived no visual tilt or sideways movement on occlusion of either eye. All 9 control subjects reported a horizontal movement of the pencil without tilt on alternate occlusion.

In DVD, monocular occlusion evokes a subjective tilt of the visual environment, which is followed by a cyclovertical divergence of the eyes and a perceived rotation of the tilted visual environment back to the vertical. This sequence of perceptual changes suggests that the cycloversional component of the DVD functions to correct a perceived tilt and to thereby reestablish vertical orientation under conditions of monocular fixation. In this discussion, I adhere to the convention in vestibular research of describing subjective visual tilt from the point of view of the subject rather than the examiner, although figures are shown from the perspective of the examiner to facilitate clinical application. According to this convention, when a patient looks "to the right," it is to the patient's right rather than the examiner's right, and so a clockwise rotation of visual environment or a torsional rotation of the patient's eyes must also be defined as clockwise from the patient's perspective.

True vertical corresponds to the gravitational vertical. Our perception of true vertical is influenced by graviceptive input to the 2 labyrinths and visual input to the 2 eyes. Subjective vertical applies to an individual's internal vertical orientation relative to true earth coordinates. When the subjective vertical is altered by neurologic disease or abnormal binocular vision input, the patient will experience a subjective visual tilt, which is a percieved tilt of the visual environment relative to the subjective vertical.⁸ As shown in Figure 1, increased luminance input to the right eye of a fish tilts the subjective vertical clockwise(as viewed by the animal), so that a vertical visual stimulus would then appear to be tilted counterclockwise relative to the subjective vertical. The direction of perceived visual tilt in DVD corresponds to the postural responses of lateral-eyed animals that exhibit a dorsal light reflex. The same perceptual shift is reported by strabismic humans with DVD when a binocular visual imbalance is induced by occlusion of one eye (Figure 2).Thus, with occlusion of the left eye, a right eye predominance would shift the subjective vertical clockwise (as seen by the patient), so that true vertical then appears to be rotated counterclockwise relative to the patient's altered subjective vertical (Figure 3). This counterclockwise tilt of the visual world, which is seen monocularly with the uncovered right eye, evokes a cyclovertical divergence movement of the eyes to erase the perceived tilt (Figure 3).

I have proposed that DVD is an atavistic resurgence of the dorsal light reflex that is evoked by a binocular visual disparity in humans with early-onset strabismus.⁸ Since the eyes retain some of their primitive function as balance organs in humans, unequal visual input induces a central vestibular imbalance in which the internal sense of vertical no longer corresponds to the gravitational vertical (Figure 3).⁸ Some patients with DVD also have a head tilt away from the side of the hyperdeviated eye,^12,13 which corresponds to the postural component of the dorsal light reflex in fish.¹⁴ This head tilt, which is not compensatory for binocular vision,^12,13 may serve to align the head to the tilted internal vertical representation(ie, the subjective vertical).¹⁴

Under monocular viewing conditions, patients with DVD may experience a "schizophrenic" perceptual situation in which the eyes tell the brain that the external world and the body are tilted relative to the altered internal representation of vertical, while the otoliths tell the brain that the head is upright.⁸ A similar sensory conflict is induced when humans view a tilted visual world under experimental conditions.¹⁵ Under the latter circumstances, the brain strikes a compromise, and the subjective visual vertical is tilted to an intermediate position between what the eyes and labyrinths are telling the brain.¹⁵

In DVD, the dorsal rotation of the visually deprived eye corresponds to the vertical divergence induced by the primitive dorsal light reflex. Although the deviating eye is said to drift "upward" in humans with DVD, the direction of rotation is not necessarily upward in space but always dorsal relative to the head, regardless of whether the patient is positioned in the upright, supine, or head-hanging position.¹⁶ Humans with DVD also display a phylogenetically newer cycloversional movement that corrects the perceived tilt of the visual environment.¹⁷ This ipsidirectional cycloversional movement can also be evoked by viewing torsional optokinetic stimuli or by inducing a static visual tilt of the visual environment.^18-22 All of these visual stimuli evoke a reflex cycloversional movement, which serves to align the tilted visual environment with the tilted internal representation of vertical.²²

Thus, the human dorsal light reflex comprises a twofold movement—a vertical divergence to realign the interpupillary axis of the eyes relative to the altered internal representation of the vertical, and a cycloversional movement that torsionally rotates the eyes in the direction of the tilted visual world to correct the perceived visual tilt (Figure 3). The vertical component of the human dorsal light reflex is a primitive adaptation that corresponds to the purely vertical divergence in fish. The phylogenetically newer cycloversional component conforms to Stephen Jay Gould's definition of an exaptation, which is a feature that did not arise as a primary adaptation, but one that was subsequently co-opted or grafted on to meet the newer demands of evolution (in this case, frontally placed eyes).²³ According to Gould, exaptations are features that were not originally built by natural selection for their current role, but that now enhance fitness (in this case, by restoring vertical orientation under monocular viewing conditions in the frontal-eyed human).²³

The human dorsal light reflex demonstrates how the central vestibular system can uncouple the vertical and torsional components of a cycloversional movement to allow each component to subserve its corrective function relative to specific conditions of a perceived tilt. During head or body tilt in humans, altered otolithic tone (ie, unbalanced input from the 2 labyrinths rather than from the 2 eyes) stimulates elevation and intorsion of the lower eye in space, and depression and extorsion of the higher eye in space.²⁴ This stimulus forms the basis of the Bielschowsky Head Tilt Test.²⁵ In DVD, however, the elevating eye extorts, and the depressing eye intorts, producing a cyclovertical divergence in which the torsional eye movements are opposite to those evoked by head tilt.⁸ This dissociation could occur only if the central vestibular system separately processes visual disparity input and graviceptive input from the trunk and otoliths, then integrates them to establish subjective vertical orientation as it does in fish.^1,8 The central ocular motor command centers must be at liberty to implement separate commands for vertical divergence and cycloversion in humans, although the 2 signals produce a single integrated extraocular movement.

This study needs to be viewed in light of its inherent limitations. First, it is a qualitative and subjective study that was performed in a clinic setting. The advantage of this simple method is that it enables the practicing ophthalmologist to confirm or refute these results without the need for special instrumentation. This disadvantage is that, because eye movement recordings were not obtained, a precise and quantitative temporal relationship between the appearance of DVD and the resolution of the subjective visual tilt could not be confirmed. Second, it is well recognized that both eyes normally extort in convergence.²⁶ One could therefore question whether the perceived image intorsion when the dominant eye is uncovered might result from each eye being in an extorted convergent position relative to the vertical object. If this were the case, however, one would expect the control group to have experienced the same perceptual changes since their eyes would also extort during convergence. However, this did not occur. Third, von Helmholtz²⁷ determined that the subjective vertical retinal meridian is tilted approximately 1° in each eye, with its top tipped temporally.²⁷ This subjective tilt could cause the top of a vertical line to appear to be tipped nasally(ie, tilted toward the side of the covered eye, as shown in Figure 2).²⁸ As such, it could be argued that this small degree of extorsional tilt in the subjective vertical retinal meridians of each eye could potentially contribute to the perception that the monocular image appears instanteously intorted when either eye is uncovered. If this were the case; however, one would again expect the control group to have perceived a similar visual tilt. Yet, only patients with DVD perceived a visual tilt under monocular conditions, suggesting that this effect did not influence the results of this study. Furthermore, the original von Helmholtz measurements of the subjective retinal vertical, like standard horopter measurements, were obtained using isolated visual stimuli in the absence of any surrounding contextual cues. In this study, normal background contextual cues were present, so one would not expect patients in either group to detect this small vertical bias. Fourth, it is not clear why the group of patients with DVD were less likely than the control group to report a perceived lateral movement of the pencil when the cover was switched from one eye to the other. It could be that the perception of tilt overrides the perception of lateral movement in these patients; that surgically corrected congenital esotropia is associated with an altered perception of visual space under monocular conditions; or that I was more attuned to elucidating the direction and sequential changes of the perceived tilt in patients with DVD, and that verbal or nonverbal cues could have influenced patient responses. Last, one could argue that if DVD is not a human dorsal light reflex, this cyclovertical divergence could be the cause rather than the result of the observed perceptual changes. In other words, extorsion of the covered eye in DVD could cause the vertical pencil to appear momentarily intorted when the either eye is uncovered. As the eye intorts to fixate, the image would extort, producing a similar sequence of perceptions as observed. If this were the case, one would expect patients with DVD to report that the image of the pencil shifted from an initial position of intorsion to a final position of extorsion. However, all patients with DVD stated that the tilted image of the pencil appeared vertical following its rotation, which would only occur if this reflex cycloversional movement served to correct the monocular visual tilt. This perceptual response suggests that this complex cyclovertical movement must function to reestablish vertical visual orientation when binocular vision is preempted.

In conclusion, DVD is associated with a subjective tilt of the visual environment and a reflex cyclovertical divergence of the eyes. This subjective visual tilt appears to drive both components of the resulting cyclovertical divergence. These perceptual correlates add to the accumulating body of evidence that DVD is a human dorsal light reflex, which serves to restore vertical visual orientation under monocular conditions. In DVD, a subjective sensation of visual tilt under monocular viewing conditions evokes 2 compensatory eye movements—a phylogenetically older vertical divergence movement (ie, a primitive adaptation) to realign the eyes relative to the altered internal representation of vertical, and an exaptive cycloversion movement that torsionally rotates the eyes in the direction of the tilted visual environment to restore vertical visual orientation by neutralizing the perceived visual tilt. The vertical component of this movement corresponds to the ancestral dorsal light reflex in fish, while the cycloversional component of the human dorsal light reflex appears to be an exaptation that functions to annul the subjective visual tilt under monocular conditions when the eyes are frontally placed. This twofold reflex movement corresponds both in theory and in actuality to the ocular motor response that would result from a tilted internal representation of the visual vertical. In the human dorsal light reflex, the direction of the cycloversion movement relative to the vertical divergence is opposite to that observed during a head tilt in space, demonstrating that the vertical divergence and the cycloversional component of visual tilt are independently programmed, and that these 2 extraocular movements can be dissociated by unequal visual input to the 2 eyes in humans with congenital strabismus.

Submitted for publication October 12, 2001; final revision received April 11, 2002; accepted May 23, 2002.

Supported in part by a grant from Research to Prevent Blindness Inc, New York, NY.

Reprints: Michael C. Brodsky, MD, Arkansas Children's Hospital, 800 Marshall St, Little Rock, AR 72202.