Afferent arm of the pupil light reflex

The neural integration of the pupil light reflex begins with the afferent pathway in the retina, consisting of the photoreceptors, bipolar cells, and ganglion cells. For many years, it was disputed whether it was the rods or cones that contribute to the pupil light reflex and whether these were the same photoreceptors as those contributing to visual perception. Extensive experimental and psychophysical work has shown that the neuronal pathways mediating the pupil light reflex and visual perception share the same photoreceptors. Until recently, it was thought that all photoreceptor input was shared by both conscious perception of light and the pupil light reflex. In previous studies, almost all changes in light input producing a change in visual perception also produced a comparable change in pupil size. In fact, in almost every way measured, the pupil responses to light parallel those of visual perception. For example, the wavelength-sensitivity profile of pupil threshold for a small transient contraction as the color of light is changed from blue to red exactly parallels the same wavelength-sensitivity profile of visual perception. The shift in sensitivity is also the same as the eye is changed from a condition of light adaptation to dark adaptation (Purkinje shift), providing further evidence that the same photoreceptors are used for both pupil and vision. Patients with various abnormalities of rods and cones can be shown to have the same deficits in color vision or lack of appropriate sensitivity change during dark adaptation when the results of visual threshold are compared with pupil threshold for small contractions to light stimuli. Both rods and cones contribute to the pupil light reflex, but to a different extent depending on the lighting conditions.

Under conditions of dark adaptation and in response to low-intensity lights, the pupil light reflex becomes a sensitive light meter and is mediated primarily by rods, giving rise to low-amplitude pupil contractions. However, with brighter light stimuli and under conditions of greater light adaptation, the cones dominate most of the transient pupil light contractions. Therefore the rods are primarily responsible for the pupil’s ability to give rise to small contractions in response to low-intensity lights under conditions of dark adaptation – they provide a high sensitivity to the pupil light reflex at low light levels, just as they do for visual perception. The cones provide the input responsible for the larger transient pupil contractions that are easily observed under direct clinical observation, occurring at suprathreshold levels of light, mainly under photopic conditions. Loewenfeld has summarized the extensive literature on this subject and her text should be consulted by the reader desiring a more complete discussion of this topic. It is also believed that the bipolar cells, which receive input from the photoreceptors, are the same neurons providing input to the pupil light reflex as those mediating visual perception.

Although it appears that the pupil and visual systems share rod and cone photoreceptor input, a number of interesting observations in recent years has revealed a previously unrecognized and important new aspect of light transduction through the pupil pathway.

New evidence in the last few years has shown that the rod and cone input to the pupil light reflex is mediated by a special class of retinal ganglion cells containing the primitive visual pigment melanopsin found in the retina of lower animals. Besides being activated by rod and cone input causing a transient pupil response, the melanopsin retinal ganglion cell is also directly sensitive to light, providing a sustained steady-state pupil constriction to light. This intrinsic, direct activation pathway of the melanopsin-containing retinal ganglion cells causes the cell to discharge in a sustained way and is directly proportional to steady-state light input, similar to a DC light meter, which does not show classical light adaptation properties. In genetically altered mice that completely lack functional rods and cones, it was discovered that a rather robust pupil light reflex was still present. This unexpected finding was followed by a series of studies to identify what retinal element could be contributing to the pupil light reflex in the absence of rod and cone input. Through clever labeling experiments, a specific ganglion cell was identified containing melanopsin, which was itself photosensitive, with a broad spectral peak centering on about 490 nm, which is blue light. These melanopsin ganglion cells have been found to project both to the suprachiasmatic nucleus in the hypothalamus and also to the pretectal nucleus (the site of the first midbrain interneuron synapse for the pupil light reflex pathway ( Box 25.3 )).

Elegant electrophysiologic recordings coupled with the study of response properties of these ganglion cells have revealed that the melanopsin-containing retinal ganglion cells provide the midbrain pathway for the pupil light reflex and also provide light sensing information for the diurnal regulating areas of the hypothalamus that modulate the circadian rhythm. These melanopsin ganglion cells also receive rod and cone input to the pupil light reflex, but are also capable of transduction of light directly (under photopic conditions), without photoreceptor input, and may be responsible for providing more of steady-state light input to the brain. This helps to explain why some patients blind from photoreceptor loss still exhibit both a pupil light reaction to bright blue light and also maintain a circadian rhythm, while patients blind from optic nerve lesions (loss of melanopsin ganglion cell input) often lack a normal circadian rhythm.

As stated earlier, the melanopsin-containing retinal ganglion cells appear to mediate the classic midbrain pathway of the pupil light reflex. However, there is also evidence that ganglion cells conveying visual information to the occipital cortex may also play a role in modulation of pupil movement in response to different types of visual stimuli. For example, patients with isolated occipital infarcts have homonymous visual field defects that show corresponding pupil defects to small (2 degrees in diameter) focal lights presented to the same cortically blind visual field area. This phenomenon has been reported previously, but only recently has the correspondence between the shape characteristic of the homonymous pupil and visual field defect been fully appreciated. This correspondence provides compelling evidence for a role of cortical mediation of the pupil light reflex when small, focal light stimuli are used. In addition, the pupil has also been shown to respond to changes in complex stimuli such as spatial frequency, motion, and contrast, providing additional evidence for a higher-level cortical process that is capable of mediating pupil contractions to visual stimuli. Such evidence implies that other types of ganglion cells, in addition to the classic luminance responding cells that project to the midbrain, may also participate in the pupil reflex, perhaps the same ones that also mediate visual perception.

The interneuron arm of the pupil light reflex

The ganglion cell axons conveying light input to the classic pupil light reflex pathway segregate from the rest of the ganglion cell axons at the distal portion of the optic tract before the lateral geniculate nucleus. As in the visual input pathway, the pupil ganglion cell axons from the nasal retina (temporal field) decussate at the chiasm to the opposite side, and the axons from the temporal retina (nasal field) stay on the same side. Therefore pupil ganglion cell axons from homonymous areas of the visual field (temporal field from the contralateral eye and nasal field from the ipsilateral eye) distribute within the optic tract. From there, they travel in the brachium of the superior colliculus and synapse in the midbrain with the next neurons in the light reflex located at the olivary pretectal nucleus. These neurons represent interneurons because they serve to integrate the afferent input coming from the retina with the efferent output of the pupil light reflex exiting the midbrain.

The receptive field properties of the pretectal neurons have been elucidated in the awake primate. These interneurons are the way-station for the converging receptive field impulses of the retinal ganglion cells from the retina and are fewer in number, summating the ganglion cell input at this location. The receptive field of each pretectal neuron has been found to receive input from ganglion cells over a large area of retina (up to 20 degrees). Some of these neurons exhibit a “flat” response, firing equally well from input over its entire receptive field. However, another subset exhibits a “foveal-weighted” response, discharging at a higher frequency when a stimulus is placed near the center of the visual field (and receptive field of the neuron). This may partly explain why the pupil light reflex appears to be more sensitive to light coming from the center of the visual field.

Patients with a relatively small area of damage to their central visual field have also been found to show an obvious decrease in the pupil light reflex in the affected eye compared with the fellow eye, which may relate to the receptive field properties of the pretectal interneurons. The pretectal neurons discharge at a frequency that is linear to the log of intensity of the light stimulus given. However, not all of these neurons respond in the same range; it appears that some are more sensitive in different ranges of intensities, so together, there is an interneuron response covering at least a 4 log unit range of input. Neurons in the pretectum send crossed and uncrossed fibers, through the posterior commissure, to the small population of neurons comprising the paired Edinger–Westphal nuclei. This allows afferent input from the pretectal nucleus on each side to be distributed almost equally to the pupil efferent pathway originating in the Edinger–Westphal nucleus ( Box 25.4 ).

The uncrossed pathway appears to have evolved during development of binocularity and stereovision. Animals with eyes located more to the side of the head (e.g. birds, rabbits) have almost completely crossed pathways with no significant uncrossed component. That is why shining a light in one eye of these animals produces an almost totally crossed input to the pretectum, which then sends an almost completely crossed output to the Edinger–Westphal nucleus. The result is that only the pupil of the eye being stimulated will contract. Cats are between birds and primates in this evolutionary respect, with approximately 70 percent of their pupil pathway crossed. Placing a pet cat so that one of its eyes points more toward a light produces a greater reaction of the pupil in the eye facing the light, causing an anisocoria. In humans, in whom the crossed and uncrossed pathways are almost equal, the direct and consensual pupil light reflex is equal. This is why illuminating one eye normally does not result in pupil inequality (anisocoria). Similarly, input deficits to one eye caused by damage to the retina or optic nerve should not normally produce an anisocoria in humans.

In some individuals the crossed pathway slightly exceeds the uncrossed pathway in both the retina and midbrain, leading to a slightly greater pupil response in the eye stimulated compared with the pupil contraction of the fellow eye, similar to cats, but not to the same extent. This consensual deficit, termed contraction anisocoria , is small and can usually be recognized only with the aid of pupillographic recordings.

The efferent arm of the pupil light reflex

The efferent arm of the pupil light reflex is diagrammatically summarized in Figure 25.4 , which also shows the site of common lesions along this pathway that may be encountered in clinical practice. The neurons in the Edinger–Westphal nucleus send preganglionic axons into the right and left fascicle of the oculomotor nerve (third nerve) to join the motor axons destined for the eye muscles, as well as the preganglionic accommodative fibers that originate in nearby nuclei. The right and left fascicles of the third nerve exit the midbrain through the subarachnoid space, where each continues as the third nerve toward the orbital apex ( Box 25.5 ).

Innervation of the iris sphincter, from the Edinger–Westphal nucleus by way of oculomotor nerve, ciliary ganglion, and short ciliary nerves, with some of the causes of a fixed dilated pupil.

After passing through the cavernous sinus to the orbital apex, the preganglionic pupillary fibers and accommodative fibers synapse in the ciliary ganglion (parasympathetic ganglion). A lesion at this site may produce an Adie’s pupil. From here, the last neurons in the chain, the postganglionic neurons, pass into the eye by way of the short ciliary nerves, located in the suprachoroidal space, where they distribute to the anterior segment of the eye to innervate the iris sphincter muscle. The postganglionic accommodative fibers, which outnumber the pupil fibers 30 : 1, supply the ciliary muscle within the ciliary body of the eye. The postganglionic pupillary neurons appear to innervate the iris sphincter muscle in a segmental distribution over approximately 20 clock-hour sections. This is why lesions of the ciliary ganglion, such as in Adie syndrome, usually cause a number of sectors of the iris sphincter to become acutely denervated, with loss of the pupil contraction only in these segments ( Box 25.6 ).

The pupil near reflex and accommodation

When fixation of the eye is shifted from a far to a near object of interest, the eyes converge, the intraocular lenses accommodate, and both pupils constrict. This triad of ocular vergence, accommodation, and pupil contraction is called the near response . Despite many contentions in the literature claiming that this pupil constriction is exclusively dependent on either convergence or accommodation, clinical and experimental data indicate that any one of the three functions can be selectively abolished or elicited without affecting the others. These experimental and clinical observations have resulted in general agreement that the impulses that cause accommodation, convergence, and pupil constriction must arise from different cell groups within the oculomotor nucleus and travel by way of separate fibers to their effector muscles. Accommodation, convergence, and pupillary constriction are associated movements and are not tied to one another in the manner usually referred to by the term reflex. They are controlled, synchronized, and associated by supranuclear connections, but they are not caused by one another. The components of the near pupil reflex were recently summarized.

The miosis of the near response and the pupillary constriction to light have a single final common efferent pathway from the Edinger–Westphal nucleus to the iris sphincter by way of the ciliary ganglion. They primarily differ in the origin of the supranuclear pathways that are elicited by light and near that both converge on the Edinger–Westphal nucleus. With a near stimulus, such as accommodation, the pupil normally constricts even with no change in retinal luminance. It is important to realize that this near reflex of the pupil is mediated by the same efferent nerve pathway originating from the same neurons in the Edinger–Westphal nucleus that mediate the pupil light reflex. There does not appear to be a separate neuronal efferent pathway that mediates the near pupil constriction.

However, the supranuclear control over this response is different from the one mediating the light reflex. In the case of the light reflex, the supranuclear input comes from the pretectal nucleus, as described in the previous section. In the case of the near reflex involving accommodation, convergence, and miosis, the supranuclear input is thought to originate from cortical areas surrounding visual cortex and from cortical areas within the frontal eye fields. The cortical neurons providing input for the near reflex are thought to synapse at least once, before passing ventral toward the visceral neurons overlying the oculomotor complex in the midbrain. This is because a cortical lesion in this area does not produce atrophy within the oculomotor nuclear complex (it is at least one synapse removed). It is also important to realize that the near reflex consists of convergence of the eyes, accommodation, and pupil contraction, all of which should be thought of as co-movements and, as stated earlier, are not strictly dependent on one another. Any one of the three co-movements may occur in the absence of the others, as discussed by Loewenfeld. Because the supranuclear pathway for the near reflex passes ventral in the midbrain and the supranuclear pathway for the light reflex passes dorsal in the midbrain, the two systems may be differentially affected by disease processes ( Box 25.4 ).

The supranuclear neuronal input from a near visual task stimulates the pupil constrictor neurons located in the visceral part of the Edinger–Westphal nuclei. The same supranuclear neuronal input also stimulates the more numerous accommodative neurons, located nearby in the remaining visceral portion of the Edinger–Westphal nucleus. These preganglionic neurons give rise to accommodative axons that travel together with the pupil preganglionic light reflex neurons within the oculomotor nerve to synapse at the ciliary ganglion in the orbit (see previous section).

In summary, with a near stimulus, both the accommodative neurons (which mediate ciliary muscle contraction) and light reflex neurons (which mediate iris sphincter contraction) in the Edinger–Westphal visceral motor nuclei are stimulated from a supranuclear level. This gives rise to a separate neuronal output of accommodative and light reflex preganglionic neurons by way of the oculomotor nerve to the ciliary ganglion, which in turn gives off separate postganglionic innervation to the ciliary body and iris sphincter muscles. The preganglionic and postganglionic light reflex pathways make use of the same neurons to mediate pupil contraction to either near or light stimuli.

Pupil reflex dilation: central and peripheral nervous system integration

Normally, when the pupil dilates, two integrated processes take place: the iris sphincter relaxes, and the iris dilator contracts, helping actively pull the pupil open. Because the iris sphincter is stronger than the dilator muscle, pupil dilation does not readily occur until the sphincter muscle relaxes. Relaxation of the iris sphincter is accomplished by supranuclear inhibition of the Edinger–Westphal nucleus at a central nervous system level, most notably from the reticular activating formation in the brainstem. It appears from animal studies that this neuronal inhibitory pathway involves the central nervous system’s sympathetic class of neurons. These sympathetic neurons pass through the periaqueductal gray area and innervate the pupil efferent neurons at the Edinger–Westphal nucleus and at the synapse there is an α ₂ -adrenergic receptor activation. When this central inhibition is active, the preganglionic parasympathetic output from the Edinger–Westphal nucleus is suppressed, resulting in a relative relaxation of the iris sphincter and pupil dilation. When this inhibition is inactive, such as during sleep, with anesthesia, or with narcotics, the preganglionic neurons fire at a high rate, causing miosis. The neurons of the Edinger–Westphal nucleus are unique in this respect because their baseline discharge frequency, without any input, is high. If all input to these neurons is disconnected, they fire at a high rate, which results in a sustained pupil contraction and miosis. This is why deep sleep or anesthesia, which reduces almost all inhibitory supranuclear input to the Edinger–Westphal nucleus, results in small pupils.

Alternatively, during a state of wakefulness, the supranuclear inhibition is active and the neurons of the Edinger–Westphal nucleus are suppressed, causing the pupils to become larger again. If a light stimulus is given at this point, a train of neuronal impulses from the retina and then the pretectal interneuron will arrive at the Edinger–Westphal nucleus, which momentarily overcomes this inhibition, causing the pupil to constrict. If the light is turned off or the retina begins to become light adapted, the supranuclear inhibition again dominates, causing a reflex dilation of the pupil.

Almost all of the conditions mentioned previously cause changes in pupil diameter resulting from the modulation of the neuronal output from the Edinger–Westphal nucleus. In addition, the same factors causing a reflex dilation of the pupil also result in an increase in output to the peripheral sympathetic nervous system innervating the iris dilator muscle. The sympathetic nerve activity can be thought of as a “turbo-charge” for pupil dilation. Peripheral sympathetic nerve activation is not a requirement for pupil dilation to occur (parasympathetic inhibition alone can accomplish that to some extent), but it greatly enhances the dynamics of pupil dilation in terms of speed and maximal pupil diameter attained.

The sympathetic outflow to the iris dilator muscle can be thought of as a paired three-neuron chain ( Fig. 25.5 ) on both the right and left side of the central and peripheral nervous system without decussations. The first neuron originates in the hypothalamus and descends through the brainstem on each side into the lateral column of the spinal cord, where it synapses at the cervicothoracic level of C7–T2. The second preganglionic neuron leaves this level of the spinal cord and travels over the apical pleura of the lung and into the spinal rami to synapse at the superior cervical ganglion at the level of the carotid artery bifurcation on the right and left side of the neck. The third neuron, the postganglionic neuron, follows a long course along the internal carotid artery into the head and orbit. As these neurons pass through the cavernous sinus, they are associated with the abducens and then the trigeminal nerve before entering the orbit and distributing to the iris dilator muscle via the long ciliary nerves.

Sympathetic innervation to the eye, showing the three-neuron chain of central, preganglionic, and postganglionic fibers.

(Modified from Maloney WF et al: Am J Ophthalmol 1980; 90:394.)

In addition to the neuronal mechanisms involved in pupil dilation, humoral mechanisms also may contribute to pupil diameter. Circulating catecholamines in the blood (e.g. a bolus released from the adrenal glands) may act directly on the iris dilator muscle either through the bloodstream or, potentially, indirectly through the tears, resulting in mydriasis. Clinical conditions that influence the integration of the parasympathetic inhibition, sympathetic stimulation, and humoral release of neurotransmitters may take various forms and may affect the dynamics of reflex dilation in a characteristic manner that may be diagnostic of clinical conditions. This is revisited later in this chapter when pupil inequality and conditions that impede pupil dilation are discussed.

Other neuronal input to the iris

In addition to the autonomic nerves supplying the iris, sensory innervation to the iris is provided by the ophthalmic division of the trigeminal nerve. However, these sensory nerves may play an additional role in modulating pupil diameter. It is well known to cataract surgeons that mechanical and chemical irritation of the eye can cause a strong miotic response that is non-cholinergic and fails to reverse with autonomically acting drugs. In rabbits and cats the response seems to be caused by the release of substance P or closely related peptides from the sensory nerve endings, but in monkeys and humans, substance P has little or no miotic effect. Cholecystokinin (in nanomolar amounts) caused contraction of isolated iris sphincter from monkeys and humans. Intracameral injections in monkeys caused miosis that was not prevented by tetrodotoxin or indomethacin, indicating that the miosis was not caused by either stimulation of nerve endings or release of prostaglandins but by direct action on sphincter receptors. The cholecystokinin antagonist lorglumide caused competitive inhibition of the response.

Iris sphincter, iris dilator, and iris color

It is important to understand the structure of the iris and its histology to understand how the iris tissue accommodates changes in pupil diameter during contraction and dilation and how disorders of the iris tissue affect pupil movement. The iris can be divided into two main layers: the posterior leaf and the anterior leaf ( Fig. 25.6 ). The posterior iris leaf contains the dilator muscle, the sphincter muscle, and the posterior pigmented epithelium. From a front view of the iris, the dilator muscle is located circumferentially, in the midperiphery of the iris.

Histology of the iris in cross section. Upper arrow points to sphincter muscle drawn in higher magnification; lower arrow points to dilator muscle of bleached preparation drawn in higher magnification.

(From Saltzmann M: Anatomy and histology of the human eye-ball, Chicago: University of Chicago Press, 1912.)

The sphincter muscle is located just inside the pupillary border; its circumference is made up of approximately 20 motor segments connected together but innervated individually by postganglionic branches of the ciliary nerve. In the normal iris these segments receive nerve excitation in a roughly simultaneous fashion, and the entire circumference contracts in concert. Both the dilator and sphincter muscles are derived embryologically from the anterior layer of the two layers of posterior pigmented epithelium.

The more superficial, anterior iris leaf consists of connective tissue stroma with cells, blood vessels, and nerves supplying the sphincter and dilator, but there is no epithelial layer in primate species. The different components of the posterior and anterior iris undergo structural alterations to accommodate changes in pupil diameter during contraction and dilation. These alterations in structure confer mechanical non-linearities influencing how much the pupil can constrict or dilate in response to changes in light or pharmacologic stimulus and was extensively studied by Loewenfeld.

The mechanical non-linearities are important because they impose limitations on the range of pupil diameter over which the extent of pupil movement can be used for assessing neuronal reflexes to light stimuli or near stimuli or for pharmacologic testing of the pupil. For example, a person with small, 3-mm diameter pupils in dim light would obviously not show as large a pupil contraction to a standard light stimulus compared with a person with 5-mm diameter pupils, but the retina and optic nerves of both persons may be completely normal. A similar situation would occur if one attempts to quantify the response to a topical miotic or mydriatic agent. Therefore, the structure of the iris can pose physical constraints on pupil movement in response to sensory stimuli or pharmacologic agents, and this should be considered carefully when comparing the response in different eyes.

The color of the iris is determined by its mesodermal and ectodermal components. In Caucasians, the stroma is relatively free of pigment at birth. The stroma absorbs the long wavelengths of light, allowing the shorter (blue) wavelengths to pass through to the pigmented epithelium, where they are reflected back, causing the iris to appear blue. If pigmentation does not develop in the anterior stromal layers, the iris remains blue throughout life. If the stroma becomes denser and contains significant numbers of melanosomes, the blue color gives way to gray. The accumulation of pigment in the iris melanocytes of individuals destined to have non-blue irides occurs during the first year of life and is dependent on sympathetic innervation of the melanocytes (derived from neural crest cells). Interruption of the oculosympathetic nerve supply to one eye during this time period usually results in heterochromia, with the denervated iris remaining blue. In a heavily pigmented iris, the fine pattern of iris vessels is hidden by pigment and the surface of the iris looks brown and velvety.