Chapter 7: The Visual System

As a species, humans depend more on vision than any other sensory modality. If strength in numbers is an indication of the importance of vision, then a simple axon count and an inventory of brain areas devoted to vision are very telling. The optic nerves, which connect the eyes with the visual processing centers of the brain, each contain about 1 million axons. The cerebral cortex, where visual messages from our eyes are analyzed and perceptions are formed, contains a bewilderingly large number of distinct areas—more than two dozen by some estimates—devoted to one or another aspect of visual processing.

In many ways, the visual system is organized like the somatic sensory systems, considered in Chapters 4 and 5. For instance, the topography of connections in the visual system is determined largely by how the receptive sheet is organized. In fact, these connections are so systematized and predictable that clinicians can use a visual sensory defect to pinpoint with remarkable precision the location of central nervous system damage. Another similarity is that both systems have a hierarchical and parallel organization. In a hierarchically organized system, distinct functional levels can be discerned with respect to one another, each with clear anatomical substrates. In vision, as in somatic sensation, multiple hierarchically organized pathways carry information from receptors to structures in the central nervous system. Each pathway processes visual information for a different purpose, such as perception of the shape, motion, and color of objects.

Visual perception, like perception for the other senses, is not a passive process; our eyes do not simply receive visual stimulation. Rather, the position of the eyes is precisely controlled to scan the environment and to attend selectively and orient to specific visual stimuli. In addition to the pathways from the retina to the cortex for perception, there is a separate pathway to the brain stem for controlling eye movements. This chapter begins with an overview of the visual pathways for perception and eye movement control. It then considers the structure and anatomical connections of the components of these pathways. Finally, the chapter examines how the clinician can, with remarkable precision, localize disturbances of brain function by using knowledge of the organization of the visual system. The visual path for eye movement control will be revisited when we consider the circuits for eye movement control (see Chapter 12).

Anatomically Separate Visual Pathways Mediate Perception and Ocular Reflex Function

The visual pathway that mediates perception and the pathway that controls eye movement originate in the retina, a thin sheet of neurons and glial cells that adheres to the posterior inner surface of the eyeball (Figure 7–2A). Also located here are the photoreceptors (Figure 7–2A, inset), which synapse on retinal interneurons that, in turn, synapse on ganglion cells. Ganglion cells are the retinal projection neurons, which synapse in the thalamus and brain stem. The axons of ganglion cells travel in the optic nerve (cranial nerve II), and ganglion cells from each eye contribute axons to the optic nerve on the same side. Some ganglion cell axons decussate in the optic chiasm (Figure 7–2A, B) en route to the thalamus and brain stem, whereas other axons remain uncrossed. Together the crossed and uncrossed ganglion cell axons, reordered according to a precise plan (see below), course in the optic tract (Figure 7–2B). Thus, each optic tract contains axons from both eyes.

The Pathway to the Primary Visual Cortex Is Important for Perception of the Form, Color, and Motion of Visual Stimuli

The principal thalamic target for ganglion cells is the lateral geniculate nucleus. This thalamic relay nucleus is analogous to the ventral posterior nucleus, the main somatic sensory relay nucleus. The lateral geniculate nucleus projects to the primary visual cortex via a pathway called the optic radiations. This projection is important for perception. It comprises multiple functional pathways, whose axons intermingle: one that is more important for perceiving the form of visual stimuli, another for color, and a third for the location and speed of motion of stimuli.

The primary visual cortex, or V1, is located in the occipital lobe, along the banks and within the depths of the calcarine fissure (Figure 7–2A). The primary visual cortex is also referred to as the striate cortex because myelinated axons (termed the stripe of Gennari) form a prominent striation. Efferent projections from the primary visual cortex follow one of three principal pathways. One pathway projects to the secondary and higher-order visual cortical areas in the occipital lobe (see Figure 7–16). The higher-order visual areas partially encircle the primary visual cortex. Each of these visual cortical areas is retinotopically organized. Whereas the primary visual cortex is important in visual signal processing that is fundamental to all aspects of visual perception, many of the higher-order cortical areas are each important in different aspects of vision. For example, the primary visual cortex processes information for stimulus form, color, and motion; one of the higher-order areas (V4) is important for color vision; and a different area is important for discriminating the direction and speed of a moving stimulus (V5; see Figure 7–15). The axons of the second path from the primary visual cortex decussate in the corpus callosum and terminate in the contralateral primary visual cortex. This projection is also for visual perception. It helps to unify images from the two eyes into a perception of a single visual world. Finally, the third path from the primary visual cortex descends to the visuomotor centers of the midbrain for focusing the images of interest on the retina and for moving the eyes to salient objects.

The Pathway to the Midbrain Is Important in Voluntary and Reflexive Control of the Eyes

Certain retinal ganglion cells project to the midbrain directly (Figure 7–2A, C), principally to two structures: the superior colliculus and the pretectal nuclei. The ganglion cell axons traveling to the midbrain skirt the lateral geniculate nucleus and course in a tract named the brachium of superior colliculus (Figure 7–2B; see also Figure 7–9). The superior colliculus, found in the tectum of the midbrain (see Figure 2–9), is dorsal to the cerebral aqueduct (Figure 7–2C). In lower vertebrates, such as amphibians and birds, the superior colliculus is termed the optic tectum and is the principal brain structure for vision, in lieu of a visual cortex. In mammals and especially humans, the superior colliculus has a minimal role in perception but an important role in the control of saccades, rapid eye movements that quickly shift visual gaze from one object to another. The pretectal nuclei are rostral to the tectum, at the midbrain-diencephalic junction (see Figure 7–9). The pretectal nuclei participate in pupillary reflexes, which regulate the amount of light reaching the retina, as well as other visual reflexes (see Chapter 12).

Additional brain stem and diencephalic projections of the optic tract serve other functions. For example, one projection reaches midbrain nuclei that are important in the reflexive control of eye position to stabilize images on the retina when the head is moving. These nuclei comprise the accessory optic system. As another example, a retinal projection to the hypothalamus is important for circadian rhythms (see Chapter 15).

The Visual Field of Each Eye Partially Overlaps

When the eyes are fixed straight ahead, the total area seen is called the visual field, the combined visual fields of each eye (Figure 7–3A). Although the visual field can be divided into right and left hemifields, the visual field of each eye is not simply one hemifield. Similar to when you look through binoculars, the field of view of each eye overlaps extensively. As a result, the visual field includes a central binocular zone (Figure 7–3A, dark shading), where there is stereoscopic vision, and two monocular zones (Figure 7–3A, lighter shading). Each hemifield is therefore seen by parts of the retina of each eye.

Optical Properties of the Eye Transform Visual Stimuli

After light enters the eye through the cornea, the transparent avascular portion of the sclera, it is focused onto the retinal surface by the lens (Figure 7–4A). The lens inverts and reverses the visual image projected on the retina. When you look at an object, you move your eyes so that the object's image falls upon the fovea, a specialized high-resolution portion of the retina. The fovea is centered within a morphologically distinct region of the retina called the macula lutea (Figure 7–4B). The brain precisely controls the position of the eyes to ensure that the key portion of an image falls on the fovea of each eye. A vertical line passing through the fovea divides the retina into two halves, a nasal hemiretina and a temporal hemiretina. Each hemiretina includes half of the fovea and the remaining perifoveal and peripheral portions of the retina. The anterior portions of the nasal hemiretinae correspond to the temporal crescents, which are monocular zones receiving visual information from the temporal parts of the visual fields (Figure 7–5).

Consider the relationship between an object being viewed and where its image falls on the retina (Figure 7–5). When you look at someone's face and, for example, fixate on the person's nose, the left side of the face is within the left visual hemifield. The image of the left side falls on the nasal hemiretina of the left eye and the temporal hemiretina of the right eye. Although each eye views the entire face, visual information from the visual hemifield on one side is processed by the visual cortex on the opposite side (see below).

Figure 7–4B also shows the optic disk, where retinal axons leave the eyeball and the blood vessels serving a part of the retina enter and leave the eye. This corresponds to the blind spot (Figure 7–3B) because the optic disk has no photoreceptors. Interestingly, an individual is not aware of his or her own visual blind spot until it is demonstrated. The fovea and optic disk can be examined clinically using an ophthalmoscope to peer into the back of the eye.

The Retina Contains Three Major Cell Layers

The retina is a laminated structure, as revealed in a section oriented at a right angle to its surface (Figure 7–6). Other components of the visual system also have a laminar organization. Lamination is one way the nervous system packs together neurons with similar functions and patterns of connections. The spatial reference point for describing the location of the different layers is the three-dimensional center of the eye. The inner, or proximal, retinal layers are close to the center of the eye; the outer, or distal, layers are farther from the center (Figure 7–6, inset).

Although the retina has many anatomically distinct layers (Figure 7–6), the cell bodies of most retinal neurons are located within three layers. This can be best observed in a micrograph of the retina of the mouse (Figure 7–7), in which the various cell types can be identified genetically or immunohistochemically.

1. The cell bodies of the two classes of photoreceptors—rods, for night vision, and cones, for high-acuity daylight vision—are located in the outer nuclear layer.

2. The cell bodies and many dendritic processes of retinal interneurons—bipolar, horizontal, and amacrine cells—are located in the inner nuclear layer.

3. Ganglion cells, the retinal projection neuron, are located in the innermost retinal cell layer, the ganglion cell layer.

Cones contain the photopigments for color vision and come in three different classes according to their absorption spectra: red, green, or blue. Cone density is highest at the fovea and decreases continuously to the peripheral retina. This is why visual acuity is greatest at the foveal and decreases to the peripheral retina. Rods contain the photopigment rhodopsin and are optimally suited for detecting low levels of illumination, such as at dusk or at night. In fact, a single photon can activate a rod cell. Rods are absent in the fovea and are densest along an elliptical ring in the perifoveal region, which is the location of maximal light sensitivity. This helps to explain why, when discerning a faint object at night, we do best by looking off to one side, rather than directly at it.

Bipolar cells connect photoreceptors directly with the ganglion cells (Figure 7–7). Of the two principal classes of bipolar cells, cone bipolar cells and rod bipolar cells, the former receive synaptic input from a small number of cone cells to give high visual acuity and color vision. By contrast, rod bipolar cells receive convergent input from many rods for less visual acuity but increased sensitivity to low levels of illumination. The actions of horizontal cells and amacrine cells enhance visual contrast through interactions between laterally located photoreceptors and bipolar cells. Horizontal cells are located in the outer part of the inner nuclear layer, whereas amacrine cells are found in the inner portion. Many amacrine cells contain dopamine, which plays a role in adapting retinal synaptic activity to the dark.

There are two major classes of retinal ganglion cells—M and P cells. The M (or magnocellular) cell has a large dendritic arbor, enabling it to integrate visual information from a wide portion of the retina. M cells are thought to play a key role in the analysis of stimulus motion as well as gross spatial features of a stimulus. The P (or parvocellular) cell, with its small dendritic arbor, processes visual information from a small portion of the retina. These cells are color sensitive and are important for discriminative aspects of vision, such as distinguishing form and color. Ganglion cell axons collect along the inner retinal surface (Figures 7–4B, 7–6, and 7–7) and leave the eye at the optic disk (Figure 7–4B), where they form the optic nerve.

Connections between many retinal neurons are also made within specific laminae (Figure 7–7). Connections between photoreceptors and retinal interneurons are in the outer synaptic (or plexiform) layer. Bipolar cells synapse on ganglion cells in the inner synaptic (or plexiform) layer.

The cellular organization of the retina might seem unexpected because light must travel through retinal layers that contain axons, projection neurons, and interneurons to reach the photoreceptors. The consequences of this organization on visual acuity are minimized by an anatomical specialization at the fovea. Here the retinal interneurons and ganglion cells are displaced, exposing the photoreceptors directly to visual stimuli and optimizing the optical quality of the image (Figure 7–7, inset). Moreover, ganglion cell axons are unmyelinated while they are in the retina, which increases the transparency of the retina and facilitates light transmission to the photoreceptor layer. Ganglion cell axons become myelinated once they enter the optic nerve. Since the retina develops from the central nervous system, the myelin sheath surrounding ganglion cell axons is formed by oligodendrocytes (see next section).

There are important nonneural cells in the retina. Müller cells, a kind of astrocyte, have important structural and metabolic functions. Their nuclei are located in the inner nuclear layer, and their processes stretch vertically across most of the retina (Figure 7–6). The other nonneural element associated with the retina, the pigment epithelium, is external to the photoreceptor layer (Figure 7–5A) and serves metabolic and phagocytic roles. For example, cells in the pigment epithelium help remove rod outer segment disks that are discarded as part of a normal renewal process. Because the retina does not tightly adhere to the pigment epithelium, it can become detached following a blow to the head or eye. This results in a partially detached retina and loss of vision in the detached portion.

The circulation of the retina has a dual organization. The arterial supply of the inner retina is provided by branches of the ophthalmic artery, which is a branch of the internal carotid. The outer retina is devoid of blood vessels. Its nourishment derives from arteries in the choroid, the layer of ocular tissue between the inner retina and the outer sclera. This may be why the photoreceptors are in the outer retina.

Each Optic Nerve Contains All of the Axons of Ganglion Cells in the Ipsilateral Retina

The optic nerve is cranial nerve II, but it is actually a central nervous system pathway rather than a peripheral nerve. This is because the retina develops from a displaced portion of the diencephalon, rather than from neural crest cell, as primary somatic sensory neurons. The optic nerves from both eyes converge at the optic chiasm (Figure 7–8). The axons of ganglion cells of each nasal hemiretina decussate in the optic chiasm and enter the contralateral optic tract, whereas those of each temporal hemiretina remain on the same side and enter the ipsilateral optic tract (Figure 7–8). Thus, each optic tract contains axons from the contralateral nasal hemiretina and the ipsilateral temporal hemiretina (Figure 7–8). Despite the incomplete decussation of the optic nerves in the chiasm, there is a complete crossover of visual information: Visual stimuli from one half of the visual field are processed within the contralateral thalamus, cerebral cortex, and midbrain.

The Superior Colliculus Is Important in Ocular Motor Control and Orientation

The optic tract splits on the ventral diencephalic surface. The major contingent of axons terminates in the lateral geniculate nucleus and gives rise to the pathway for visual perception (see next section). A smaller contingent skirts the lateral geniculate nucleus and passes over the surface of the medial geniculate nucleus, which is the thalamic auditory nucleus (see Chapter 8). These axons collectively are termed the brachium of superior colliculus (Figure 7–9C) because their major site of termination is the superior colliculus.

The superior colliculus is laminated on microscopic appearance: Incoming visual information is processed by the dorsal layers (Figure 7–9C), whereas somatic sensory, auditory, and other information is processed by neurons in the ventral layers. The ventral layers of the superior colliculus contain part of the neural apparatus for eye and neck muscle control (see Chapter 10). A function of the superior colliculus is to combine visual and other sensory information to generate motor control signals to help orient the eyes and head to salient stimuli in the environment.

The neural systems for visuomotor function and visual perception appear to converge in the cerebral cortex. Certain superior colliculus neurons have an axon that ascends to two thalamic nuclei that serve more integrative functions than sensory relay alone, the lateral posterior and pulvinar nuclei of the thalamus (see Table 2–1; Figure 2–13). These thalamic nuclei project primarily to higher-order visual areas and to the parietal-temporal-occipital association areas. One function of this ascending projection from the superior colliculus may be to inform cortical areas important for visual perception about the speed and direction of eye movements. This information is important for distinguishing between movement of a stimulus and movement of the eyes.

The Lateral Geniculate Nucleus Transmits Retinotopic Information to the Primary Visual Cortex

The major retinal projection is to the lateral geniculate nucleus of the thalamus. This nucleus forms a surface landmark on the ventral diencephalon that is sometimes called the lateral geniculate body (Figures 7–10 and 7–11). It is located just lateral to the medial geniculate nucleus, the thalamic auditory nucleus (see Chapter 8). The lateral geniculate nucleus is retinotopically organized. The fovea is represented posteriorly in the lateral geniculate nucleus, with progressively more peripheral parts of the retina represented anteriorly. The medial superior part of the lateral geniculate nucleus represents the inferior visual field, and the lateral inferior part, the superior visual field.

The lateral geniculate nucleus sends its axons to the primary visual cortex via the optic radiations (Figures 7–10 and also 7–9A, B). The optic radiations take an indirect course around the lateral ventricle to reach their cortical targets. A portion of the optic radiations transmitting visual information from the superior visual field courses rostrally within the temporal lobe (termed Meyer's loop), before heading caudally to the primary visual cortex.

The primary visual cortex, which is located mostly on the medial brain surface, corresponds to Brodmann's cytoarchitectonic area 17 (see Table 2–2; Figure 2–19). The retina and, in consequence, visual space are precisely represented in the primary visual cortex (Figure 7–10) because of the orderly projection from the thalamus to the cortex. The foveal representation, corresponding to central vision, is caudal to the perifoveal and peripheral portions. The inferior retina, receiving information from the upper visual field, is represented in the inferior bank of the calcarine fissure. The superior retina, receiving visual input from the lower visual field, is represented in the superior bank. Although the fovea region is a small portion of the retina, the area of primary visual cortex devoted to it is greatly expanded with respect to the rest of the retina. This organization is similar to the large representation of the fingertips in the primary somatic sensory cortex (see Figure 4-9B).

The Primary Visual Cortex Has a Columnar Organization

Different areas of the cerebral cortex have a similar anatomical organization: They each have six principal cell layers, often with multiple sublaminae, and the thalamic relay nucleus makes most of its synapses within layer IV. Different cortical areas also share a similar functional organization: Neurons located above and below one another—yet in different layers—have similar properties. This is the columnar organization of the cerebral cortex. In the primary somatic sensory cortex, neurons in a cortical column all process sensory information from the same peripheral location and the same somatic submodality (see Chapter 4; Figure 4–11).

The primary visual cortex also has a columnar organization (Figure 7–11). Neurons in a cortical column have similar properties and functions because local connections primarily distribute the thalamic input vertically, from layer IV to superficial and deeper layers, rather than horizontally within the same layer. Horizontal connections do exist; however, they mediate other kinds of functions, such as enhancing contrast and helping to associate visual information from different parts of a scene to form perceptions. These horizontal connections run in the stria (or stripe) of Gennari.

The primary visual cortex has at least three types of columns (Figure 7–11): (1) Ocular dominance columns contain neurons that receive visual input primarily from the ipsilateral or the contralateral eye; (2) orientation columns contain neurons that are maximally sensitive to visual stimuli with similar spatial orientations (see Figure 7–13); and (3) color columns, termed blobs, are vertically oriented aggregates of neurons in layers II and III that are sensitive to the color of visual stimuli.

Ocular dominance columns segregate input from the two eyes

The lateral geniculate nucleus contains six principal cell layers, stacked on top of one another. Each layer receives information exclusively from ganglion cells from the ipsilateral or the contralateral retina. Neurons in the dorsal four layers have different functions than those in the ventral two layers (see below). In layer IV of primary visual cortex, the axon terminals of lateral geniculate neurons that receive input from the ipsilateral retina remain segregated from the terminals of neurons that receive their input from the contralateral retina (Figures 7–11 and 7–12).

Ocular dominance columns can be shown in human primary visual cortex at autopsy in a person who had one eye removed before death, for example, because of an ocular tumor. When stained for the mitochondrial enzyme cytochrome oxidase, tissue sections show alternating stripes of reduced and normal staining (Figure 7–12A). The stripes with reduced staining correspond to the ocular dominance columns of the removed eye, which were inactive following enucleation. Normal staining corresponds to the columns of the intact eye, active until death. The ocular dominance columns can be analyzed on histological sections and the three-dimensional configuration of the columns drawn on the surface of the primary visual cortex (Figure 7–12B, C).

Mixing of information from both eyes, giving rise to binocular inputs, occurs in neurons located above and below layer IV. These binocular interactions are mediated largely by cortical interneurons. The binocular neurons receive a stronger synaptic input from the same eye that projected information to the monocular neurons in layer IV, and a weaker input from the other eye. This pattern of lateral geniculate axon terminations in layer IV (Figure 7–11) and blending of connections above and below layer IV forms the anatomical basis of the ocular dominance columns. A given retinal location in each eye is represented in the cortex by a pair of adjacent ocular dominance columns. Horizontal connections between neurons in adjacent ocular dominance columns are thought to be important for depth perception.

Orientation columns are revealed by mapping cortical functional organization

Physiological studies have shown that most neurons in the primary visual cortex respond to simple bar-shaped stimuli with a particular orientation. However, unlike ocular dominance, which is an attribute based on anatomical connections from one eye or the other, orientation specificity of neurons in a column in the primary visual cortex is a property produced by synaptic connections between local cortical neurons. Orientation columns can be revealed in experimental animals using methods that provide an image of neuronal function, such as neuronal activity or local blood flow, which correlates with neural activity. Figure 7–13 is an image of a small portion of the surface of the primary visual cortex of a monkey as it viewed contours of different orientations. The figure shows the pattern of activation of cortical neurons in response to stimuli of different orientations. Neurons sensitive to particular stimulus orientations are located within territories of one or another color. Neurons sensitive to all orientations are present within a local area, but they are distributed in a radial pattern resembling a pinwheel. Cells selective for stimulus orientation (and therefore the orientation columns themselves) are located from layer II to layer VI, and spare a portion of layer IV, which contains neurons that are insensitive to stimulus orientation.

Clusters of color-sensitive neurons in layers II and III are distinguished by high levels of cytochrome oxidase activity

Neurons sensitive to the wavelength of the visual stimulus are clustered within the ocular dominance columns in layers II and III. The locations of these color-sensitive cells correspond to regions of primary visual cortex that have high levels of activity of the mitochondrial metabolic enzyme cytochrome oxidase (Figure 7–14). The regions of increased enzyme activity, which correspond to the clusters of color-sensitive neurons, are termed blobs (Figure 7–14, small dots). The adjoining secondary visual cortex (area 18; V2) has no blobs, but shows alternating stripes of increased (thick and thin stripes) or decreased (pale interstripe) cytochrome oxidase activity (Figure 7–14). The section on higher-order areas addresses how neurons in the thick stripe, thin stripe, and pale interstripe are part of distinct visual processing channels.

The Magnocellular and Parvocellular Systems Have Differential Laminar Projections in the Primary Visual Cortex

The M (magnocellular) and P (parvocellular) ganglion cells give rise to visual information channels that process distinctive stimulus features. M ganglion cells synapse in the ventral laminae of the lateral geniculate nucleus, whereas P cells synapse in the dorsal laminae; this is in addition to receiving input from ganglion cells from either the ipsilateral or the contralateral halves of the retina (Figure 7–11). Because the thalamocortical neurons located in the ventral layers are larger than those in the dorsal layers, they are also called magnocellular and parvocellular layers, respectively. M cells are the major input to a visual circuit for the analysis of stimulus motion and generalized aspects of form, whereas P cells provide the input for analyzing the color and fine details of stimuli.

Neurons in the magnocellular and parvocellular layers of the lateral geniculate nucleus project to different sublaminae in layer IV of the primary visual cortex. The magnocellular system projects primarily to layer IVCα, whereas the parvocellular system projects primarily to layers IVA and IVCβ. Interneurons in the layer IV sublaminae connect with neurons in superficial and deeper cortical layers, which distribute visual information to other cortical and subcortical regions (Figure 7–11). The differential laminar projections of the magnocellular and parvocellular systems set the stage for distinct visual processing channels that distribute information about different aspects of a stimulus to the secondary and higher-order visual cortical areas.

Higher-Order Visual Cortical Areas Analyze Distinct Aspects of Visual Stimuli

The higher-order visual areas, located in Brodmann's areas 18 and 19, are located in the occipital lobe; they partially encircle area 17 (Figure 7–15). They receive visual information directly or indirectly from the primary visual area as well as from the integrative thalamic nuclei, the pulvinar and lateral posterior nuclei. Each is also retinotopically organized. (Higher-order visual areas are collectively termed the extrastriate cortex because they lie outside the primary area, or striate cortex, which contains the stripe of Gennari.) Intracortical connections between the visual areas are bewilderingly complex; they have both a hierarchical and a parallel component. For example, the primary visual cortex projects to the secondary visual cortex (V2), which in turn projects to V5. This is a hierarchically organized projection to V5. The parallel projection to V5 is a direct one, skipping V2. Although it can be deduced that less information processing occurs in the parallel projection, it is not yet clear how the parallel and hierarchical paths differ functionally.

Research analyzing connections of the monkey visual system suggests that out of the myriad cortico-cortical projections among the primary and higher-order visual areas, different pathways are involved in perceiving stimulus motion, color, and form (Figure 7–16A). The secondary visual cortex (V2) plays a key role in all three pathways.

1. The motion pathway derives from the M-type ganglion cells. Information passes through the magnocellular layers of the lateral geniculate nucleus, to neurons in layer IVCα of the primary visual cortex (Figure 7–11), and from there to neurons in layer IVB (Figure 7–11). Neurons in layer IVB project directly to V5 and indirectly via neurons in the thick cytochrome oxidase stripes of V2 (Figure 7–16). In the rhesus monkey, V5 corresponds to a region named MT, for middle temporal area. This region is important not only for motion detection but also for regulating slow eye movements (see Chapter 12). The pathway from VI (and V2) to V3 may be important for analyzing aspects of visual form in motion. A region thought to be analogous to V5 in the human can be imaged while the subject views moving visual stimuli (Box 7–1).

2. The color pathway derives from the P-type ganglion cells, which terminate in the parvocellular layers of the lateral geniculate nucleus. From there the thalamocortical projection is, via neurons in layer IVCβ (Figure 7–11), to neurons in the color blobs (layers II and III), then to the thin stripes in V2 (Figure 7–16), and next to V4. A region that may be equivalent to V4 in the human cortex has been described using functional imaging (Box 7–1). The color blobs also receive a direct projection from the interlaminar neurons in the lateral geniculate neurons, which are located in the region between the principal magnocellular and parvocellular cell layers.

3. The form pathway also derives primarily from the P-type ganglion cells and the parvocellular layers of the lateral geniculate nucleus. In V1, neurons in layer IVCβ (Figure 7–11) project to the interblob regions of layers II and III, and from there to the pale interstripe portion of V2 (Figure 7–16). Next, V2 neurons project to V4. Whereas the motion and form systems are thought to contribute to depth perception, the color system does not.

Object Recognition Is Transmitted by the Ventral Stream and Spatial Localization and Action, by the Dorsal Stream

The notion of functionally distinct pathways for different attributes of a visual stimulus helps to explain the remarkable perceptual defects that occur in humans following damage to the temporal and parietal lobes. Damage to the inferior temporal lobe produces a selective defect in object recognition. By contrast, damage to the posterior parietal lobe impairs the patient's capacity for object localization in the environment but spares the patient's ability to recognize objects. These findings suggest that there are two streams of visual processing in the cortex (Figures 7–15 and 7–16): The ventral stream to the temporal lobe carries information about specific features of objects and scenes, and the dorsal stream to the parietal lobe carries spatial information. Thus, the ventral stream is concerned with seeing what, as opposed to where, which is the function of the dorsal stream. Although extensive interconnections exist, the ventral stream for object recognition may receive a preferential input from the parvocellular, or form and color, system. In contrast, the dorsal stream for localization receives input primarily from the magnocellular system. The dorsal stream for stimulus localization is also important using visual information to guide movement. Through connections to the frontal lobe, the where stream is also an action, or how, system. The dorsal-ventral pathway distinction is also present in the mechanosensory (Chapter 4) and auditory (Chapter 8) systems.

The Visual Field Changes in Characteristic Ways After Damage to the Visual System

The pattern of projection of retinal ganglion cells to the lateral geniculate nucleus and then to the cerebral cortex is remarkably precise, defined by the retinotopic organization. Damage at specific locations in the visual pathway produces characteristic changes in visual perception. This section examines how clinicians can apply knowledge of the topography of retinal projections to localize central nervous system damage.

Functional connections in the visual system can be understood by delineating the visual field. Recall that the visual field corresponds to the total field of view of both eyes when their position remains fixed (Figure 7–3). The visual fields of the two eyes overlap extensively. A change in the size and shape of the visual field—a visual field defect—often points to specific pathological processes in the central nervous system (Table 7–1). Such defects may reflect damage to any of six key visual system components (Figure 7–18).

• Optic Nerve: Complete destruction of the optic nerve produces blindness in one eye (Figure 7–18A; Table 7–1); partial damage often produces a scotoma, a small blind spot. When a scotoma occurs in the central field of vision, for example, in the fovea, the patient notices reduced visual acuity. Remarkably, a peripheral scotoma is often unnoticed. This emphasizes the importance of foveal vision in our day-to-day activities (see below). Optic nerve damage also produces characteristic changes in the appearance of the optic disk (Figure 7–4B) because the damaged ganglion cell axons degenerate. Tumors and vascular disease commonly cause optic nerve damage.

• Optic Chiasm: Ganglion cell axons from the nasal halves of the retina decussate in the optic chiasm (Figure 7–8). These fibers transmit visual information from the temporal visual fields. A common cause for chiasmal damage is a pituitary tumor. The pituitary gland is located ventral to the optic chiasm. As the tumor grows it expands dorsally, because the bony floor of the cavity in which the pituitary gland is located (the sella turcica) is ventral to the pituitary gland. The mass encroaches on the optic chiasm from its ventral surface. This results in preferential damage of the decussating fibers and produces a bilateral temporal visual field defect (bitemporal heteronymous hemianopia) (Figure 7–18B; Table 7–1). Patients may not notice such a defect because it occurs in their peripheral vision. They commonly come to an emergency room following an accident caused by peripheral visual loss, for example, a traumatic injury incurred from the side, such as being hit by an automobile.

• Optic Tract or the Lateral Geniculate Nucleus: Damage to the optic tract or the lateral geniculate nucleus, also due to tumors or a vascular accident, produces a defect in the contralateral visual field (homonymous hemianopia) (Figure 7–18C; Table 7–1). If a lesion is due to compression, such as produced by a tumor, the basis pedunculi can become affected (Figure 7–9C), resulting in contralateral limb motor control impairments.

• Optic Radiations: Axons of lateral geniculate neurons course around the rostral and lateral surfaces of the lateral ventricle en route to the primary visual cortex at the occipital pole (Figure 7–9B2). Neurons in the lateral geniculate nucleus that mediate vision from the superior visual fields have axons that course rostrally into the temporal lobe (Meyer's loop) before they course caudally to the primary visual cortex. Temporal lobe lesions can produce a visual field defect limited to the contralateral upper quadrant of each visual field (quadrantanopia) (Figure 7–18D; Table 7–1). This is sometimes referred to as a "pie in the sky" defect because it is often wedge shaped. Neurons in the lateral geniculate nucleus that serve the macular region and the lower visual field project their axons laterally around the ventricle and caudally through the white matter underlying the parietal cortex. A lesion of the white matter within the parietal lobe can affect the optic radiations and produce visual field defects (homonymous hemianopia) (Figure 7–18E; Table 7–1).

• Primary Visual Cortex: Damage to the primary visual cortex, which commonly occurs after an infarction of the posterior cerebral artery, produces a contralateral visual field defect that can sometimes spare the macular region of the visual field (homonymous hemianopia with macular sparing) when the lesion affects visual cortex gray matter not subcortical white matter (Figure 7–18F; Table 7–1). Two factors contribute to macular sparing. First, in the case of infarctions, the arterial supply to the cortical area that serves the macular region is provided primarily by the posterior cerebral artery, with a collateral supply coming from the middle cerebral artery (see Figure 3–4B). After occlusion of the posterior cerebral artery, the middle cerebral artery can rescue the macular representation. Second, the area of cortex that mediates central vision is so large that a single infarction, or other pathological process, rarely destroys it entirely. Although rare, a traumatic injury to the occipital pole can produce a defect involving only the macular region (Figure 7–18G; Table 7–1).

Retina

The retina is the peripheral portion of the visual system (Figures 7–6 and 7–7). Retinal neurons are located in three cell layers. The cell bodies of photoreceptors are located in the outer nuclear layer (1): Cones are the photoreceptors for color vision and high-acuity vision; rods are for night vision. The cell bodies of retinal interneurons—bipolar cells, amacrine cells, and horizontal cells—are located in the inner nuclear layer (2). Ganglion cells are located in the ganglion cell layer (3) (Figures 7–6 and 7–7). Connections between many retinal neurons are also made within specific laminae (Figure 7–7). Connections between photoreceptors and retinal interneurons are in the outer synaptic layer. Bipolar cells synapse on ganglion cells in the inner synaptic layer. Light must pass through the ganglion cells and interneurons before reaching the photoreceptors. Müller cells are the principal retinal neuroglia.

Visual Field and Optic Nerves

The retina receives a visual image that is transformed by the optical elements of the eye (Figure 7–5): The image becomes inverted and reversed. Images from one half of the visual field (Figure 7–3) are projected on the ipsilateral nasal hemiretina and the contralateral temporal hemiretina (Figure 7–5). The axons from ganglion cells exit from the eye at the optic disk (Figure 7–8B). Axons from ganglion cells in the temporal hemiretina project into the ipsilateral optic nerve and ipsilateral optic tract (Figure 7–8B). Ganglion cell axons from the nasal hemiretina project into the ipsilateral optic nerve, decussate in the optic chiasm, and course through the contralateral optic tract (Figure 7–8).

Midbrain Projections for Eye Movement Control

Ganglion cell axons destined for the midbrain leave the optic tract and course in the brachium of superior colliculus (Figures 7–2C and 7–9). A key midbrain site for the ganglion cell axon terminals is the superior colliculus, a laminated structure (Figure 7–9). The superficial layers of the superior colliculus mediate visuomotor and visual reflex function, and the deeper layers subserve orientation of the eyes and head to salient stimuli. The pretectal nuclei, where interneurons for the pupillary light reflex are located, also receive retinal input (Figure 7–9B; see Chapter 12).

Thalamic and Cortical Projections for Perception

The lateral geniculate nucleus (dorsal division) is the thalamic nucleus that receives the principal projection from the retina (Figures 7–9 and 7–10). Like other structures in the visual system, the lateral geniculate nucleus is laminated, and each of the six layers receives input from either the ipsilateral or contralateral retina. Visual information comes from the contralateral visual hemifield.

Visual Cortical Areas

The lateral geniculate nucleus projects to the primary visual cortex via the optic radiations (Figures 7–9 and 7–10), which course through the white matter of the temporal, parietal, and occipital lobes. Thalamic input terminates principally in layer IV, in sublaminae A and C, of the primary visual cortex (Figure 7–12). Input from the ipsilateral and contralateral eyes remains segregated in this layer. This is the anatomical substrate of the ocular dominance columns (Figures 7–11 and 7–12). Another type of column is the orientation column (Figures 7–11 and 7–13). Vertically oriented aggregates of neurons in layers II and III, centered in the ocular dominance columns, are color-sensitive columns (or color blobs) (Figures 7–11 and 7–14), the third column type.

The primary visual cortex is retinotopically organized (Figure 7–10B). The primary area projects to the higher-order visual areas of the occipital, parietal, and temporal lobes (Figures 7–15 and 7–17). There are at least three functional pathways from the primary visual cortex to higher-order visual areas: (1) for perception of stimulus form, (2) for perception of stimulus color, and (3) for perception of stimulus motion. The ventral stream comprises the pathway into the temporal lobe for object recognition. The dorsal stream is the path for stimulus location and action.

Visual Field Defects

Damage to the visual pathway produces characteristic changes in visual perception (Figure 7–18; Table 7–1): (1) complete transection of the optic nerve, total blindness in the ipsilateral eye, (2) optic chiasm, bitemporal heteronymous hemianopia, (3) optic tract and lateral geniculate nucleus, contralateral homonymous hemianopia, (4) optic radiation in the temporal lobe (Meyer's loop), contralateral upper quadrant homonymous hemianopia, (5) optic radiations in parietal and occipital lobes, contralateral homonymous hemianopia, and (6) primary visual cortex, contralateral homonymous hemianopia with macular sparing.