From Visual Panorama to Frontal Binocular Vision
Although we retain our primitive adaptations, the function of the human
oblique muscles has evolved to meet the needs of single binocular vision.
In the course of evolution, primitive adaptations give way to exaptations.
An adaptation is something fit (aptus) by construction for (ad) its usage.30 Exaptation is a relatively
new evolutionary concept advanced by Gould30
to describe a feature, now useful to an organism, that did not arise as an
adaptation for its present role, but that was subsequently co-opted for its
current function. Such structures are fit (aptus)
not by explicit molding for (ad) current use, but
as a consequence of (ex) properties built for other
reasons.30 According to this definition, a
mechanism must have a function and must enhance the fitness of its bearer
to qualify as an exaptation.30-31
For example, the feathers of birds may have originally evolved for thermal
insulation (an adaptation), only to be subsequently co-opted for flight (an
exaptation).31-32
Cyclovergence, Stereoscopic Perception, and the Pitch Plane
According to Blakemore et al,32 binocular
animals have abandoned the enormous biologic advantage of panoramic vision
in order to have their eyes pointing forward, the most obvious advantage of
which is stereopsis. Frontal repositioning of the eyes seems to have exapted
the oblique muscles to subserve stereopsis. Evolution has grafted a new torsional
control system that is subordinate to binocular vision on top of the "primitive"
dynamic torsional programming of the oblique muscles. Although the brain programs
eye torsional position by regulating the tonus of all extraocular muscles,
the oblique muscles have the predominant effect on ocular torsion. It is therefore
instructive to examine torsional eye position as a function of oblique muscle
innervation.
How do the human oblique muscles subserve stereopsis? Under conditions
of binocular fixation, an object closer in space than the fixation point will
produce an image on the temporal retinas, while an object farther in space
than the fixation point will produce an image on the nasal retinas.6 This horizontal disparity forms the basis for stereoscopic
perception. If one examines the circles that appear elevated on a Titmus stereoacuity
test under binocular conditions, examination with each eye will show a nasal
displacement of the circle in space, indicating that the image falls on the
temporal retina in each eye when the circle is viewed binocularly. When the
Titmus test is turned upside down so that the monocular image falls on the
nasal retinas of each eye, the circles appear to lie behind the plane of the
page.
Now consider a binocular individual with normal stereopsis who is fixating
on the center of a vertical object that is slanted so that its inferior aspect
is closer than the superior aspect (Figure
1). As the individual fixates the center of the slanted object,
the visual image of the upper pole is postfixational, which means that it
falls onto the nasal retina of each eye, whereas the visual image of the lower
pole is prefixational, falling onto the temporal retina of each eye. The reader
can appreciate this slant illusion by holding a pencil in the midsagittal
plane with the upper pole slanted away from the body and the lower pole tilted
toward the body. On occlusion of either eye, the upper pole of the pencil
will appear to be tilted, with the upper pole leaning toward the side of the
uncovered eye (Figure 1). 33 So under monocular conditions, the person perceives
a disconjugate image torsion that is analogous to how the image would be seen
if there were intorsion of each eye.
|
|
|
|
Figure 1. Tilt is the monocular correlate
of stereoscopic slant. A, An individual binocularly viewing a vertical object
that is slanted in the pitch plane. B, The monocular images corresponding
to the object are extorted when viewed with each eye.
|
|
|
If a vertical cyclodisparity in the 2 eyes is translated by the visual
cortex into a binocular sensation of depth in the pitch plane (ie, slant),
can retinal image torsion cause a vertical binocular image to be perceived
as slanted in the pitch plane? The answer is yes. The reader can appreciate
this phenomenon by placing a white Maddox Rods over each of the 2 eyes in
a trial frame, and looking toward a bright focal light source with the grids
oriented horizontally to produce a vertical line (Figure 2). Now counterrotate the lenses so that their upper pole
of each line moves nasally and their lower pole moves temporally until the
image of the binocular vertical line breaks into 2 tilted lines (Figure 2). If the rotation is stopped at
the break point, the torsional diplopia can be overcome and the 2 lines can
be fused. When cyclofusion occurs, which is almost purely on a sensory (as
opposed to a motor) basis, the single line will suddenly appear to be stereoscopically
slanted in the pitch plane, with the upper pole inclined toward the observer
and the lower pole inclined away from the observer. The opposite inclination
is seen when the upper poles of the Maddox Rods are rotated temporally. The
original treatise by Wheatstone34 describing
his invention of the stereoscope in 1838 provided the first example of pitch
stereopsis produced by dichoptic lines that are tilted in different directions.
|
|
|
|
Figure 2. Binocular cyclodisparity of vertical
lines is perceived stereoscopically as slant. A, Rotation of horizontal double
Maddox Rods to produce binocular image intorsion (as would be seen if both
eyes were extorted). B, Sensory cyclofusion causes the patient to stereoscopically
perceive a vertical line (solid line) as slanted in the pitch (ie, sagittal)
plane (dashed line).
|
|
|
This stereoscopic effect shows us something remarkable about the normal
binocular visual system. It tells us that when an isolated vertical image
cyclodisparity falls within the physiologic range of sensory fusion, it is
misregistered stereoscopically as a slant of the vertical
object in the pitch plane.7, 21-22,32-38
In a real world setting, however, perceived pitch is not solely a function
of retinal cyclodisparity, but it depends both on the brain's computation
of registered eye rotation and on retinal cyclodisparity. So to create an
accurate stereoscopic representation of vertical objects in the pitch plane,
cyclovergence should not occur and the eyes must be locked into a well-defined
static orientation relative to a given gaze position (ie, conforming to Donder's
law).1, 7, 21-22
The innervational patterns of oblique muscle recruitment, which counteract
the torsional actions of the rectus muscles in different positions of gaze,
must also be subordinate to this goal.
That binocular torsional control represents an active function of the
human oblique muscles rather than an evolutionary loss of contractile function
is seen in the kinematics of human convergence.39
It has long been recognized that both eyes extort during convergence and that
this extorsion increases in downgaze and decreases in upgaze.40-41
(Extorsion is even considered by some to be a component of the synkinetic
near reflex.42) While the existence of these
cyclovergence movements were once thought to constitute a violation of Listing's
law, they can be reconciled with Listing's law if it is assumed that convergence
is associated with a temporal rotation of Listing's plane in each eye (Figure 3).43
Vertical rotation of the eyes around these temporally rotated axes produces
incyclovergence of the eyes in upgaze and excyclovergence in downgaze.44-46 Several lines of
evidence suggest a neural and biomechanical basis for these cyclovergence
movements. In monkeys, Mays et al47 measured
single cell recordings within the trochlear nucleus and found decreased unit
activity during convergence. This decrease in firing rate was greater when
the monkey converged in downgaze than in upgaze, a finding that corresponds
to the observed convergence-associated torsional movements in humans. More
recently, dynamic magnetic resonance imaging by Demer et al48-49
have found that downward rotation of the lateral rectus muscle pulley and
medial rotation of the inferior rectus pulley during convergence, indicating
that the inferior oblique muscle may also play a role in convergence-associated
torsion, presumably via its collagenous attachments to the lateral rectus
and inferior rectus muscles.
|
|
|
|
Figure 3. View of both eyes from above demonstrating
orientation of Listing's plane (LP) during distance fixation (A) and convergence
(B). Curved arrows denote cyclovergence movements of the eyes associated with
vertical rotation about horizontal visual axes in Listing's plane. A "saloon
door" rotation of Listing's plane, which is opposite in direction to the ocular
rotation, can be used to reconcile the convergence-associated extorsion of
the eyes in downgaze (D) and intorsion of the eyes in upgaze (U) with Listing's
law.
|
|
|
From an evolutionary perspective, it is worth examining whether these
torsional movements during convergence simply represent primitive adaptations
that have been phylogenetically retained, like the small ocular counterroll
which has no known function in humans.17 In
the lateral-eyed animal, upgaze corresponds to an intorsional movement of
both eyes when the rotation is viewed from the frontal perspective, while
downgaze corresponds to an extorsional rotation of both eyes. One could argue
that convergence in humans may simply stress the system to a degree that prevents
binocular suppression of these primitive rotations. However, recent studies
suggest that the frontal binocular visual system has latched on to this primitive
torsional bias and exapted it to subserve stereoscopic perception in the pitch
plane.50-51 To subserve stereopsis,
the oblique muscles have been exapted to torsionally align the eyes with their
corresponding visual images in a way that preserves the binocular horizontal
disparities that produce stereoscopic perception in different positions of
vertical gaze.51 This exaptation serves to
minimize the brain's computational load for stereoscopic perception.51
Human cyclovergence is most robust at near fixation, where it plays
an active role in stereoscopic vision. If convergence were not linked to cyclovergence,
symmetrical convergence on a frontoparallel plane would induce incyclodisparity
of the horizontal images in upgaze and excyclodisparity of the horizontal
images in downgaze for each eye, solely on the basis of the geometric angle
from which each eye views a planar surface. (The opposite cyclodisparity bias
occurs for vertical visual landmarks due to projection geometry; however,
it is reduced or reversed by the horizontal retinal shear that was described
by Helmholtz.40) In convergence, the increased
intorsion of the eyes in upgaze and extorsion in downgaze helps to torsionally
align the horizontal meridians of the eyes with their respective horizontal
visual landmarks, thereby facilitating stereopsis. Since convergence is generally
used for downgaze, where near objects are situated, the innervational link
between convergence and extorsion presumably serves to set the operational
position for stereopsis as slightly in downgaze, where near objects can be
held by the arms and illuminated by overhead light.40-41,52
Although the torsional movements associated with convergence are preprogrammed,17 they exhibit remarkable plasticity53
and are enhanced by the depth perception of stereograms,54
demonstrating that they also rely on visual input to more accurately subserve
the needs of binocular vision and depth perception. Without these torsional
movements, convergence during vertical gaze would limit optimal stereoscopic
perception to 1 gaze elevation, requiring repositioning of the head to optimize
depth perception of targets at different earth elevations. Thus, this oblique
muscle exaptation provides the luxury-optimizing stereopsis for targets at
different eye elevations without the necessity of head movements in the pitch
plane.
Almost 30 years ago, Blakemore et al32
recorded action potentials from binocular neurons in the cat's visual cortex
and measured orientation selectivity during simultaneous binocular stimulation.
Certain binocular cells responded specifically to objects tilted in 3-dimensional
space toward the cat or away from it.32 Such
binocular cells may form at least part of the substrate for sensory cyclofusion
in humans. The survival value of stereoscopic spatial orientation explains
why cyclofusional movements are so limited in primary gaze and why sensory
cyclofusion is so well developed.3-7,55-56
Sensory cyclofusion without motor cyclofusion is a prerequisite for pitch
stereopsis (ie, slant perception). Motor cyclofusion of these torsionally
disparate vertical landmarks would induce a misperception of stereoscopic
slant for vertical lines. For any position of gaze, the oblique muscles must
torsionally anchor the eyes to produce a stable motor substrate for slant
perception of vertical objects.
Psychophysical experiments have shown that horizontal visual landmarks
are selectively used in humans to lock in torsional eye position, although
the neural feedback loops for this process are unknown. As early as 1861,
Nagel35 observed that the rotation of horizontal
fusion contours produced cyclovergence, while the rotation of vertical contours
produced only a stereoscopic effect.28 Ogle
and Ellerbrock37 noted that cyclofusion of
torsionally disparate horizontal lines that were presented dichoptically with
no visual background caused a previously fused vertical line to pitch in the
sagittal plane. According to Bradshaw and Rogers,57
cyclovergence is not well driven by disparities along vertical meridians even
when these are created by a real inclined surface. By inducing cyclodisparity
of horizontal lines to the 2 eyes dichoptically, small cyclofusional eye movements
can be elicited in humans under experimental conditions.7, 20, 23-24,37-38,56
The small size of these cyclofusional movements suggests that the human cyclovergence
system is equipped to provide a fine-motor modulation to a system that is
designed primarily for stability rather than movement.7
The greater stereoscopic value of vertically compared than horizontally tilted
images explains why vertically oriented gratings evoke smaller cyclovergence
movements than horizontally oriented gratings.23
Humans inhabit a terrestrial environment composed of primarily vertical
and horizontal landmarks that serve as reference points for vertical orientation.21-22 In a terrestrial setting, the most
prominent horizontal contour is the horizon, which may be the main visual
reference for stabilizing the eyes relative to the outside world.21 While cyclodisparities of vertical contours may be
caused by slant of the observed objects, cyclodisparities of horizontal contours
indicate cyclovergence errors that need correction.20-21
As summarized by Howard and Rogers33:
An orientation disparity between the images of lines in the horizontal
plane of regard can be due only to eye misalignment, whereas an orientation
disparity from a vertical line may be due to inclination of the line in depth.
It would therefore be adaptive if cyclovergence were evoked only by disparities
in horizontal elements, leaving residual disparities in vertical elements
intact as clues for inclination.
If cyclodisparities in horizontal visual landmarks of the visual scene
that occur at low stimulus frequencies and low amplitudes serve as the physiologic
stimulus for cyclovergence as proposed by Howard and Zacher,58
then horizontal cyclofusion must serve to torsionally anchor the eyes, allowing
the visual cortex to extract and construct a stable and reproducible representation
for pitch stereopsis.7 By using cyclodisparity
of horizontally oriented lines as the feedback signal for torsional misalignment
of the eyes and allowing cyclodisparity of vertically oriented lines to signal
depth (ie, slant), the brain can recruit the oblique muscles to control cyclovergence
and thereby assure accurate stereoscopic perception of vertical objects in
the pitch plane. While a cyclodisparity of horizontal lines are the driving
force for this cyclofusional reflex, stereopsis and vertical visual orientation
are its emergent functions.
If a binocular vertical cyclodisparity produces a stereoscopic sensation
of pitch, then ocular torsion produced by cyclovertical muscle palsy should
produce an abnormal sensation of pitch stereopsis. Not surprisingly, abnormal
pitch stereopsis is a common symptom of acquired superior oblique palsy. Lindblom
et al59 found that adults with acquired unilateral
or bilateral superior oblique palsy perceived the upper pole of a vertical
rod as being tilted toward them in the sagittal plane under binocular conditions.
This stereoscopic illusion corresponded to the associated extorsion of the
paretic eye (Figure 4). Subjects
also perceived the unfused image corresponding to the eye with the superior
oblique palsy as intorted (ie, tilted in the roll plane) relative to the other
image.
|
|
|
|
Figure 4. Right superior oblique palsy.
A, Examiner's view of patient's retinas showing extorsion of the right eye.
B, Under binocular conditions, the patient perceives a vertical object (solid
line) as stereoscopically slanted in the pitch plane (dashed line).
|
|
|
As seen in superior oblique palsy, the concept of Panum's space can
be extrapolated to torsional eye position. Ocular torsion within the realm
of fusion induces an illusory pitch stereopsis of isolated vertical lines,
while torsion outside the realm of fusion induces torsional diplopia in the
roll plane.5, 60-61
Ocular torsion of 5°, as generally occurs with unilateral superior oblique
palsy, is not an impediment to fusion, whereas ocular torsion of more than
10°, which accompanies bilateral superior oblique palsy, precludes fusion.6 In unilateral superior oblique palsy, it is often
stated that strabismus surgery or prismatic correction to vertically realign
the eyes is sufficient to restore cyclofusion, even when the extorsion persists
in the palsied eye. However, the persistence of extorsion in one eye is not
without perceptual consequence, and it should be remembered that sensory fusion
of torsional images can cause vertical objects to be perceived as slanted.
Psychophysical experiments by Howard and Kaneko62
have shown that an isolated shear disparity of vertical lines will induce
a stereoscopic slant, whereas a cyclodisparity that twists both vertical and
horizontal lines will not induce a perceived slant of the visual environment.
These experiments would predict that patients with unilateral superior oblique
palsy and extorsion of 1 eye would stereoscopically perceive isolated visual
landmarks in the sagittal plane as slanted toward them.
Cycloversion, Nonstereoscopic Perception, and the Roll Plane
Unlike cyclovergence, which is remarkably stable and seems to depend
primarily on where the eyes are looking, human cycloversion shows both intrasubject
and intersubject variation.39 These findings
implicate different neural control strategies for cycloversion and cyclovergence.39 One explanation for this disparity is that cycloversion
is probably not as important to stereoscopic vision as cyclovergence, which
determines stereoscopic volume at any given pitch plane, alters slant perception
of vertical objects, and is necessary for stereo constancy.39
Nevertheless, large cycloversional movements of the eyes create a problem
for stereoscopic perception. Misslisch et al17
have argued that the superimposition of a primitive cycloversion movement
of the eyes (such as an ocular counterroll evoked by a head tilt) on convergence
would induce a cyclodisparity and disrupt stereoacuity. The brain strikes
a balance between gyroscopic and stereoptic mechanisms by damping the ocular
counterroll by approximately 70% in convergence.18
In this way, exaptations of the neural circuitry that steers our cyclovergence
and cycloversion movements seem to override our primitive adaptations to promote
stereopsis.39, 50
While the human oblique muscles function under static conditions to
constrain torsional rotation of the eyes, it is not the physiologic role of
any muscle to simply constrain movement (check ligaments and muscle pulleys
are better suited to this function). To the surprise of many, Tweed et al63 have recently found that the human oblique muscles
execute large cycloversional saccades immediately preceding head movements
in the roll plane. The 3-dimensional scleral search coil recordings were performed
as normal study participants observed a laser spot while their eyes were directed
20° downward. The subjects then made combined eye-head movements to refixate
the laser spot as it jumped 20° (from right head tilt and right gaze to
left head tilt and left gaze). Eye movement recordings showed that these subjects
generated ipsiversive torsional saccades that ranged in size from 11°
to 17° and averaged 14.5°. These cycloversional eye movements preceded
the head movements by 20 to 60 milliseconds, indicating that these movements
were not vestibular in origin. The eyes arrived at the target first and locked
on, hanging in space as the head rotated around them (Figure 5). When the head came to a halt, the ocular torsion relative
to the head had stabilized near zero, and the eyes were poised for the swiftest
possible response to further movement of the target. These torsional eye movements,
which occur at the initiation of a head tilt and are not visible on gross
inspection, may reveal another exaptation of our torsional control system.
The human oblique muscles may have been exapted to generate saccadic torsional
eye movements to reestablish roll plane orientation in anticipation of a postural
rotation in the roll plane. These anticipatory saccades instantaneously recalibrate
torsional eye position to provide a stable visual representation of tilt in
the roll (frontal) plane. The evolution of frontally positioned eyes for stereopsis
may have created a survival advantage for the grafting of this new torsional
control system on top of the ancestral control system that produces the ocular
counterroll (Table 1). Similar
torsional saccades have not been observed in lateral-eyed animals (although
studies involving eye-head or eye-body coordination in animals are extremely
difficult to perform).
|
|
|
|
Figure 5. Depiction of torsional eye position
during head tilt from side to side. (1) Initial torsional position during
right head tilt. (2) During head tilt to the left, the eyes lead the head
and quickly assume their final torsional position corresponding to the left
head tilt. (3) The eyes "hang in space" until the head catches up. (4) Head
tilt to the right produces the reverse sequence of torsional eye movements.
(Reprinted with permission from The American Association for the Advancement
of Science, copyright 1999.63)
|
|
|
|
|
|
|
Oblique Muscle Adaptations and Exaptations
|
|
|
Reversion From Exaptation to Adaptation
Since ocular torsion within the realm of fusion can produce a stereoscopic
tilt in the pitch plane, it seems reasonable to ask whether strabismus or
other neurologic disease, which can be associated with a pathologic tilt in
the internal representation of the gravitational vertical, could recalibrate
prenuclear innervation to the extraocular muscles and produce a torsional
deviation of the eyes that conforms to this internal shift. Again, the answer
is yes. One of the primitive functions of the human oblique muscles is to
rotate the eyes toward the subjective visual vertical; when this internal
representation is altered under pathologic conditions, ocular torsion is the
inevitable result. In humans, as in lower animals, the central vestibular
system uses weighted input from the 2 labyrinths and weighted visual input
from the 2 eyes to establish subjective vertical orientation in pitch and
roll.64-65 In humans, cycloversion is evoked by visual or graviceptive imbalance in the roll plane,64 whereas cyclovergence
is evoked by a visual or graviceptive imbalance in the pitch plane.65 In the roll plane, unilateral loss of otolithic tone
secondary to brainstem, cerebellar, or utricular injury causes skew deviation,
whereas asymmetrical visual input in humans with congenital strabismus evokes
dissociated vertical divergence (Table 1).65 In skew deviation, unequal graviceptive
tone from the otoliths produces a pathologic tilt in the internal representation
of the visual vertical in the roll plane, which is associated with a corresponding
torsional repositioning of both eyes and a vertical divergence of the eyes
(Table 1).66
In dissociated vertical divergence, a dorsal light reflex in the roll plane
is also associated with visually induced tilt in the subjective visual vertical,
and the cycloversional component is ipsidirectional to the patient's perceived
visual tilt.67 Since the cycloversional component
of dissociated vertical divergence does not accompany the dorsal light reflex
in lateral-eyed animals, and it cannot be accounted for by anatomical repositioning
of the extraocular muscles, this component of the human dorsal light reflex
seems to represent an exaptation to restore vertical orientation during monocular
viewing.67 The associated cycloversion movement
that occurs when humans fuse vertically disparate images68-69
may indicate that binocular vertical disparity in humans is similarly misregistered
by the brain as tilt.70
In the pitch plane, these same primitive adaptations are operative.
Donahue and I65 have proposed that primary
oblique muscle overaction is associated with a slant of the internal representation
of the visual vertical in the pitch plane. A subjective inclination of the
superior portion of the visual environment toward the individual would produce
a corresponding extorsion of both globes and lead to inferior oblique muscle
overaction.65 Conversely, structural neurologic
disease within the brain stem or cerebellum would produce the intorsion and
superior oblique muscle overaction so commonly seen in children with Chiari
malformations, meningomyelocele, or hydrocephalus.65
The alternating skew deviation on lateral gaze with bilateral abducting hypertropia
that is associated with craniocervical disease may represent another central
vestibular disturbance in the pitch plane.71-72
Under pathological conditions, the oblique muscles still function to
keep the eyes in binocular torsional register with the perceived visual environment,
and an altered torsional position of the eyes constitutes an ocular motor
recalibration to the tilted or slanted internal representation of the visual
world that characterizes central vestibular disease. By recognizing that a
subjective tilt of the visual environment evokes a corrective torsional repositioning
of both eyes, we can begin to place the horse before the cart in understanding
congenital strabismus.
CONCLUSIONS
