Section III: Systems Neuroscience: Sensory & Motor Systems

OBJECTIVES

IMAGING & CAPTURING LIGHT

Vision begins with light passing through the cornea and lens, the optical elements that refract, focus, and transmit light to the innermost layer of the eye, the retina, the region that transduces light information to electrical signals and transmits them to the brain. Because the cornea is at the interface between air (index of refraction close to 1.0) and the corneal tissue (index of refraction about 1.38), the cornea does most of the focusing. The lens, with a slightly higher index of refraction than the aqueous humor (variable, but about 1.4 in its center), fine tunes the focus. Ciliary muscles modulate refraction by the lens, controlling focus. This process is called accommodation. Because the cornea and lens are converging and positive lenses, they project an inverted image on the retina.

Anatomy of the Eye

The main structural elements of the eye are shown in Figure 11–1.

The eye is composed of 3 main layers that enclose 3 transparent structures. The outermost layer, composed of the cornea and sclera, is called the fibrous tunic. The middle layer, composed of the choroid, ciliary body, and iris, is called the vascular tunic (uvea). The inner layer is the retina, which obtains its blood supply from the choroid and retinal vessels. Within these layers are 3 transparent structures: (1) the aqueous humor, filling the anterior and posterior chambers between the lens and cornea; (2) the gelatinous vitreous humor (vitreous body), filling the bulk (80%) of the eyeball between the lens and the retina; and (3) the flexible lens. The overall structure of the eye is illustrated in the section view in Figure 11–1A.

Figure 11–1B shows a magnified section of the anterior portion of the eye. Approximately 70% of the focusing of the light is done at the cornea at the air–tissue interface. The lens fine-tunes this focus by changing its shape, illustrated in Figures 11–1C and 11–1D. The control of focus is called accommodation. The pupil is an adjustable hole in the iris that controls the amount of light entering the eye. The pupil size is controlled by the third cranial nerve, the oculomotor nerve. This nerve arises from 2 nuclei in the anterior mesencephalon (midbrain): the oculomotor nucleus and the Edinger-Westphal nucleus. The Edinger-Westphal nucleus projects to the eye via the ciliary ganglion as part of the parasympathetic system. It controls pupil constriction via the sphincter pupillae muscle and accommodation via the ciliary muscle. The oculomotor nucleus also controls eye movements (see later discussion).

The Formation of Images on the Retina

The optical elements of the eye are critical for normal vision, and thus, dysfunction, aging, or errors in development of these elements can require corrective lenses or surgery. Conditions affecting the eye include the following:

1. Opacities in the cornea or lens (cataracts) that block or scatter light, resulting in degradation of image quality.

2. Asphericity in the cornea reducing focusing quality, called astigmatism. There are 2 types of astigmatism: meridional and nonmeridional astigmatism. Meridional astigmatism is characterized by a corneal surface that is slightly cylindrical rather than spherical, making its focusing power in 1 axis different than off that axis. If, for example, the cornea/lens optical system is set for proper vertical but not horizontal focus, vertical lines can be resolved, but horizontal lines would be blurry. Nonmeridional astigmatism, in contrast, occurs when the corneal surface is irregular in a complex way such that there is not even a single axis of good focus.

3. Conditions that affect focus, such as myopia and hyperopia. The eye is said to be emmetropic if the match between the focusing power of the cornea and lens and the length of the eye is such that light from distant objects is correctly focused on the retina when the lens is in its flattest, lowest power state. During normal development, the growth of the eye matches changes in the eye’s optics, but errors in this process result in either myopia, where the length of the eye is too long for the front of the eye optics, or hyperopia, where it is too short (Figure 11–2). In myopia, the power of the cornea/lens system is too high for the length of the eyeball, so that myopes cannot focus on distant objects. However, myopes can focus on objects closer to the eyes than those with normal emmetropic eyes. A negative lens (Figure 11–2A, bottom) is used to reduce the cornea/lens power and achieve normal focus. In hyperopia, the cornea/lens system has insufficient power for the length of the eye. In mild cases, lens accommodation may allow distant objects to be focused, but then there is no remaining accommodative power to focus near objects. This is corrected with a positive lens (Figure 11–2B, bottom).

4. Presbyopia (literally “old eye”) arising during aging, in which the lens hardens and can no longer be made to bulge to increase its optical power. If one is emmetropic prior to the onset of presbyopia, then distance vision, where the lens is flat, is normally unaffected, but lens power cannot be increased to see closer objects, requiring, for example, reading glasses. Someone who is not emmetropic prior to presbyopia onset may require different compensation lenses for distant versus close vision, such as bifocals.

Eye Movements

Our eyes move continually. Eye movements can be voluntary or involuntary and are essential for acquiring, fixating, and tracking visual stimuli. About 3 to 4 times a second, the eyes make large movements, called saccades, that shift the fovea to a new area of attention. The fovea is the small specific pit-like region of the retina, located within a disk called the macula that is involved in sharp central vision (see Figure 11–1).

Figure 11–3 shows eye movements recorded during free fixation of a photograph of a girl’s face. In between saccades, during fixation, our eyes make several types of small movements, called microsaccades, drift, and tremor, which are necessary for vision. Eye movements are controlled by 3 cranial nerves. The oculomotor nerve controls the majority of eye muscles, with the exception of the superior oblique, which is controlled by the trochlear nerve, and the lateral rectus, controlled by the abducens nerve.

If images are artificially stabilized on the retina, after a few seconds, all vision disappears because the retina adapts to constant light and only responds to changes. The brain takes into account the eye movements it commands with the associated image movements so that when you are tracking a bird moving across the sky and the image on the retina is relatively stationary, your brain knows that the bird is moving because it knows your eyes are moving with it. When we move, the output of the vestibular system (the semicircular canals in the ear) is stimulated by self-movement. This signal is combined with visual information so that we perceive movement of ourselves rather than that of things around us.

THE RETINA

Images transformed by the cornea and lens are projected onto the retina, the neural tissue in the innermost lining of the eye. The retina and optic nerve are derived from the optic vesicle of the prosencephalon of the neural tube and are considered part of the central nervous system (CNS) and brain. The retina is a component of the CNS that is located outside the skull or vertebral column, for the obvious reason of allowing light to be imaged by the transparent anterior eye components.

The retina is nourished by 2 main sources of vasculature. The first of these is the central retinal artery and vein, which enter through the optic nerve. These blood vessels ramify a capillary bed across the retinal surface everywhere except in the macular area of highest visual acuity. The other source of blood supply is the choroidal vasculature, which is just distal to the retinal pigment epithelium (RPE). These vascular structures are illustrated in Figure 11–4A. This figure also shows the macula lutea and fovea centralis, structures at the center of the retina that enable the highest visual acuity. The macula is blood vessel free and exhibits a high density of cones and rods. The fovea, in the center of the macula, has an ultra-high density of cones and is rod free. The fovea receives input from the central one-half degree of visual angle, which is about the size of the moon. The center of the fovea has only red and green cones, but no blue cones.

A typical region of the retina outside the fovea is illustrated in Figure 11–4B. Because the retina more or less lines the inside of the spherical eyeball, vision scientists use the center of the eye as the main reference point and the terms proximal and distal to refer to cell and synaptic layers closer to and farther from, respectively, this center point. The most proximal illustrated structure is a capillary on the proximal surface of the retina (next to the vitreous) from the central retinal artery. Distal to this is the ganglion cell axon layer, then the retinal neurons and synaptic layers, the RPE, and finally, the choroid. Light passes through all the cell and synaptic layers before reaching the photoreceptor outer segments at the top of Figure 11–4. To prevent scattering, light not absorbed by the photoreceptor outer segments is absorbed by the dark RPE.

Two nonneural cell types are important in the function of the retina. Figure 11–5 shows, on the left, a photomicrographic cross-section of the retina stained for cell bodies rather than processes. This reveals the neural cell layers. On the right of the figure is a depiction that includes 2 important nonneural cells: the RPE cell, discussed earlier, and the Müller cell. Radial Müller cells are glia that extend throughout the entire radius of the retina. Their end feet form the inner limiting membrane at the vitreal surface.

The relationship between the photoreceptor outer segments and the RPE is shown in Figure 11–6. Photoreceptor health and function are dependent on the RPE, which supplies vitamin A for production of retinal, the photon-absorbing molecule associated with the photoreceptor opsin. Lysosomal enzymes within the RPE also digest the rod outer segment tips that are constantly shed as their light-absorbing pigment is used up. Dark melanosomes within the RPE prevent scattered light and image blurring. Blindness in diseases such as retinitis pigmentosa (RP) stem from RPE dysfunction.

Retinal Neurons: Organization & Function

There are 5 major classes of retinal neurons whose location and connectivity are illustrated in Figure 11–7. The initial process of seeing, as mediated by the retina, starts with the capture of light photons by photoreceptor cells, which are specific types of sensory neurons. Light capture controls the release of glutamate by the photoreceptors. This glutamate modulates the activity of bipolar and horizontal cells to which they are connected in the outer plexiform layer. Bipolar cells, in turn, activate amacrine and retinal ganglion cells in the inner plexiform layer. The result of this process is the conversion of the optical image to several neural images represented in the firing of different ganglion cell classes whose axons form the optic nerve and project to visual processing centers in the brain. This is discussed in more detail later.

1. Retinal layers and cells. The retina is an ordered, layered, hierarchical structure with distinct cell body and synaptic layers. Starting from the most proximal layer at the vitreal surface (where light enters) is the optic nerve fiber layer. This layer contains the axons of retinal ganglion cells that will gather at the optic nerve head and exit the retina on their way to the optic chiasm and retinal recipient zones in the brain. Distal to that is the ganglion cell layer that contains the cell bodies of the retinal ganglion cells (2 types are shown: midget and diffuse). Within the ganglion cell layer are also several types of amacrine cells, which are interneurons that do not project to the brain (and are not shown in the ganglion cell layer in Figure 11–7). These are called “displaced” amacrine cells. Distal to the ganglion cell layer is the inner plexiform (plexiform = synaptic) layer, which contain the output terminals of bipolar cells that release glutamate onto the dendrites of amacrine and ganglion cells.

   Distal to the inner plexiform layer is the inner nuclear layer that contains the cell bodies of bipolar (midget, rod, and flat bipolar), horizontal, and amacrine cells. Distal to that is the outer plexiform layer, which contains the photoreceptor terminals that synapse onto bipolar cell and horizontal cell dendrites. Beyond that is the outer nuclear layer containing the photoreceptor cell bodies (rod and cone), and beyond that the photoreceptor inner and outer segment layers.

2. Major neuronal cell classes. The 5 major neuronal cell classes in the retina are photoreceptors, bipolar cells, horizontal cells, amacrine cells, and ganglion cells. The main pathway through the retina is the sequence of photoreceptor to bipolar cell to ganglion cell. Horizontal cells modulate the signals from photoreceptors to bipolar cells, and amacrine cells modulate the output of bipolar cells to ganglion cells.

3. Müller cells. Müller cells are radially oriented glial cells that provide the initial scaffolding for retinal development and provide metabolic support for the neurons in the developed retina (see Figure 11–5).

4. Pigment epithelial cells. RPE cells not only reduce the scattering of unabsorbed photons, but are also important in nourishing the photoreceptor cells and in phagocytosis and recycling photopigment (see Figure 11–6).

PHOTORECEPTORS: ANATOMY, DISTRIBUTION, & FUNCTION

Anatomy of Rods and Cones

The first step in seeing is the capture of photons by photoreceptors. The 2 major types of photoreceptors, rods and cones, function in dim and bright light, respectively. Figure 11–8 shows a typical rod (left) and cone (right). A major structural difference between rods and cones is that, in rods, the photopigment is contained in disks that are completely internal to the outer segment of the receptors, whereas in cones, the cell membrane itself is folded into a comb-like membrane process that contains the photopigment. Rods and cones also differ in their distribution within the retina.

Photoreceptor Anatomy

Photoreceptors convert light into a modulation of the release of the neurotransmitter glutamate. This process is called phototransduction. Photoreceptors have 3 main structural parts, as illustrated in Figure 11–8:

1. The outer segment contains the photon-absorbing photo-pigment in an array of disk-like membrane structures. The plasma membrane of the outer segment has a high concentration of cyclic nucleotide–gated sodium/calcium channels.

2. The inner segment contains mitochondria and biochemical machinery for transporting the photopigment, such as rhodopsin, to the outer segment from its sites of synthesis in the cell body. Within the inner segment is the cell body that contains the nucleus and protein-manufacturing machinery.

3. The synaptic terminal is where glutamate is released to regulate second-order cells (bipolar and horizontal cells).

The outer segments of rods and cones interdigitate into the RPE, which provides essential metabolic support, such as resupplying new photopigment after it is bleached by light. The RPE also has a phagocytic function to degrade the distal tips of the rod and cone outer segments that are constantly being shed (see Figure 11–6). RPE dysfunction may cause photoreceptor death. This is a frequent cause of blindness, such as in RP.

Distribution of Rods & Cones

Unlike the imaging chip in a camera, which typically has a uniformly dense detector array, the retina has a much higher density of receptors in the center than in the periphery. Moreover, the mix of receptors also differs as a function of retinal location. Several aspects of the receptor density function are shown in Figure 11–9.

The horizontal axis of Figure 11–9 is eccentricity, in units of visual angle. What this refers to is the projection of the visual world onto the retina along a horizontal line that runs through the fovea and extends from the periphery of the retina closest to the nose (nasal retina) to the periphery closest to the cheek (temporal retina). The density of cones is strongly peaked within the central retina, which corresponds to just a few degrees of visual angle. This area is the fovea (see Figures 11–1 and 11–4), and the immediately surrounding region is the parafovea. Outside this region, the cone density is relatively constant out to the far periphery of the retina. Of the approximately 6.5 million cones in the retina, more than half are in the foveal region.

Not shown in this diagram is that fact that within the fovea center there are only green and red cones, but no blue cones. Blue cone density relative to green and red cones peaks slightly just outside the fovea in the parafovea, and then constitutes a relatively constant percentage of all cones throughout the rest of the retina, where the relative percentages are about 64% red, 32% green, and 2% blue. There are approximately 6.5 million cones in the human retina. Cones mediate daylight vision, including color.

The rod distribution is quite different from that of the cones. The fovea and parafoveal region are virtually rod free. Rod density peaks just outside this region, and then gradually declines out to the retinal periphery. Because rods mediate vision in dim light, it has long been known that in trying to locate a very dim object, such as a faint star, one should not look directly at the assumed location but just to the side of that location. Rods greatly outnumber cones across the retina except in the central foveal region. The retina contains approximately 120 million rods. The vertical density function follows a similar, but not identical, eccentricity profile for both rods and cones.

The blind spot is the retinal area where the ganglion cell axons become myelinated and exit the retina to form the optic nerve. There are no photoreceptors there. Generally, we are not aware of the blind spot for 2 reasons: (1) the blind spots in the 2 eyes are at different visual field locations; and (2) the brain “fills in” the lack of input based on surrounding neural activity. This also occurs during the slow onset of blindness; patients may lose considerable vision before they become consciously aware that there is any problem.

Rods & Night Vision

Rods mediate vision in dim light. The vast majority of mammals, including humans, have a single type of rod whose maximum sensitivity is to a wavelength of approximately 500 nm. Figure 11–9 shows that there are many more rods everywhere in the retina except the fovea, where they are completely absent. The human retina contains approximately 120 million rods. Rods are specialized to operate in very low light levels. Under ideal circumstances, a brief flash can usually be detected if approximately 5 to 7 rods each capture a single photon of light at about the same time. Individual rod cells are more sensitive to light because they express more photopigment and have higher amplification in the phototransduction cascade. Vision at low light levels mediated by rods is called scotopic vision.

Cones & Color Vision

Humans, like many primates, have 3 types of cones, called short-wavelength (blue), middle-wavelength (green), and long-wavelength (red) cones. When photoreceptors capture a photon of visible light, their response is the same regardless of its wavelength, because the molecular cascade that modulates glutamate release is the same regardless of the wavelength of the photon that is absorbed. This means that in dim light, where only rods are sensitive enough to generate visual signals, there is no color vision (vision is scotopic) because nothing about the wavelength of the absorbed photon is communicated to higher order retinal neurons. In bright light, however, retinal cells are sensitive to the ratio of activity of different cones, which varies with wavelength, and vision is photopic. Vision at high light levels mediated by cones is called photopic vision. Figure 11–10 shows the relative absorption efficiency for the short-, middle-, and long-wavelength cones (and rods) as a function of wavelength. Photons of any wavelength of light produce a unique ratio of activity of the 3 cones, allowing color to be identified independently of light intensity level. There is a regime at middle light levels in which both cones and rods operate to some extent. This is called mesopic vision.

PHOTOTRANSDUCTION

The Phototransduction Second Messenger Cascade

The absorption of a light photon by the photopigment in the photoreceptor outer segment (Figure 11–11A) initiates a cascade of intracellular events that leads to the modulation of glutamate release by the photoreceptors on to bipolar and horizontal cells. Key aspects of this cascade (described in the following text) are illustrated in Figure 11–11B:

1. The photopigment is formed by the protein opsin, which is bound to a molecule of retinal. Opsins are transmembrane proteins in the G-protein–coupled receptor family. When the photopigment molecule, such as rhodopsin in rods (or photopsins in cones), absorbs a photon, the retinal changes its stereoisomer form from the kinked structure called 11-cis retinal to a straighter form called all-trans retinal.

2. All-trans retinal separates from the protein opsin to which it was bound, allowing the opsin to be active.

3. The released opsin then activates transducin, a type of G protein. When activated, transducin dissociates its bound guanosine diphosphate (GDP) and binds to guanosine triphosphate (GTP).

4. The transducin-GTP complex activates phosphodiesterase, the enzyme that breaks down cyclic guanosine monophosphate (cGMP) into 5′-GMP. cGMP is a second messenger that promotes the opening of the cyclic nucleotide–gated (CNG) sodium/calcium channels in the outer segment of the photoreceptor. The reduction in the levels of cGMP reduces the number of open CNG channels (see Figure 11–11). Note that only sodium moving through this channel is depicted in Figure 11–11, but some other cations also transit this channel.

5. Closure of CNG channels causes hyperpolarization of the photoreceptor. Hyperpolarization is also contributed by the potassium current in the inner segment that pulls the membrane potential down toward the equilibrium potential for potassium (E_K). The sodium-potassium transporter pump works to counteract the effect of the dark current and maintain the normal cell low sodium–high potassium concentration gradients.

6. The hyperpolarization of the photoreceptor causes its synaptic terminal (see Figure 11–8; also called the pedicle) to release less glutamate, the photoreceptor neurotransmitter.

7. The modulation of glutamate release drives other cells in the retina. Specifically, the outputs of photoreceptors drive 2 main types of cells, called bipolar and horizontal cells.

Dark Currents & Synaptic Glutamate Release

cGMP concentrations in photoreceptor outer segments are high in the dark, and this ensures that the percentage of open CNG sodium/calcium channels is high. A large number of potassium channels are continuously open in the inner segment. Thus, in the dark, the simultaneous entry of sodium and calcium in the outer segment and exit of potassium in the inner segment produce the dark current. The dark current holds the membrane potential at about –40 mV, so glutamate release is high in the dark. Closure of the outer-segment CNG sodium/calcium channels by the light, from reduction of cGMP concentration, allows the excess potassium current in the inner segment to hyperpolarize the photoreceptor to approximately –60 mV. This hyperpolarization reduces the release of glutamate from the photoreceptor terminal. Hyperpolarization in response to light occurs in the photoreceptors of all vertebrates. However, many invertebrates operate in an opposite mode and have photoreceptors that depolarize to light via photochemical mechanisms.

Photoreceptor Adaptation

The visual system has evolved to function from light levels at which single rods absorb a few photons per second, to levels where cones absorb millions. This dynamic range is mediated by 3 types of adaptation: (1) the use of rods in dim light and cones in bright light; (2) light and dark adaptation by rods and cones themselves within their operating range; and (3) adaptation by other retinal neurons. Cones function in light levels from daylight shadows to reflection off snow, which constitutes a range of about 8 log units of light intensity. Rods adapt over a range from a few photons per second to hundreds, or about 2 log units of background intensity. Adaptation allows the photoreceptors to generate relatively large signals for small changes in light level or the mean illumination at that time of day. Adaptation involves processes within photoreceptors and within the retinal circuitry. Schematically, regardless of the locus, adaptation depends on detecting the mean light level and dividing or subtracting that level from the direct response to light intensity, as illustrated in Figure 11–12.

One important place this is done is at the photoreceptor to bipolar cell synapse, mediated by horizontal cells. Horizontal cells summate input from neighboring photoreceptors and, via inhibition, reduce the photoreceptor to bipolar cell drive in a proportional manner. The result is a shift of the operating curve of the bipolar cell, so that its response range is adjusted to be centered at the current overall light level (illustrated in Figure 11–12).

Photoreceptor Diseases

Most blindness in the developed world is caused by photoreceptor degeneration (the major exception is glaucoma, in which retinal ganglion cells degenerate). Photoreceptors are highly specialized cells with intense metabolic and protein trafficking demands placed on them by the processing of visual pigments. Photoreceptors also depend on metabolic support by the pigment epithelium, into which their outer segments interdigitate, so that pigment epithelium dysfunction can secondarily produce photoreceptor degeneration.

1. Macular degeneration. Macular degeneration refers to a loss of vision in the center of the visual field (the macula) because of damage to the photoreceptors in this central region of the retina. It can be differentiated into so-called “dry” and “wet” types. Dry macular degeneration (often called age-related macular degeneration) is associated with the accumulation of drusen (a kind of cellular debris) that accumulates between the choroid (blood vessel network distal to the pigment epithelium that nourishes the retina) and the retina (see Figures 11–4 and 11–5). Death of photoreceptors mediating central vision occurs. Dry age-related macular degeneration is more prevalent in the elderly and typically has a slow progression. Its progression can sometimes be slowed with dietary modification.

   Wet macular degeneration occurs when blood vessels sprout (neovascularization) from the choroid behind the retina. These neovascular vessels often leak, leading to retinal scarring and cell death. The retina can also become detached, with the detached portion also dying. It is sometimes treated with laser photocoagulation therapy to limit neovascularization.

2. Retinitis pigmentosa. Retinitis pigmentosa is a progressive degenerative, inherited pathology caused either by genetic defects in the rod photoreceptors or the RPE. It typically begins with loss of night vision due to rod death and progresses to tunnel vision from death of all peripheral photoreceptors. Ultimately, total blindness can occur. Mutations in >35 genes have been linked to RP, with mutation in the rhodopsin gene responsible for approximately 25% autosomal dominant RP.

3. Color blindness and anomalies. Color vision is possible when there are different receptor types with different spectral sensitivities, which is the probability of absorbing light as a function of the wavelength (see Figure 11–10). Humans with normal color vision are called trichromats because they have 3 distinct cone types. However, most mammals and many partially colorblind humans have only 2 cone types and are called dichromats. Dichromacy is usually caused by a pigment allele mutation in which either the green cone is switched to a red pigment (deuteranope) or the red cone is switched to green (protanope), so that the dichromat has about the normal number of cones but no red-green opponency. It is also possible to lose the blue cone system altogether, although that is rarer (tritanope). Calling humans who are missing a cone type “colorblind” is a misnomer because these people are color deficient in only part of the spectrum. Colorblindness is an X-linked recessive disorder. The genes for the red and green cones are located on the X chromosome. Because women have 2 X copies, they need 2 defective genes to have this colorblindness. But because men have only 1 X chromosome, 1 mutation gives the defect to a man. Red-green color blindness occurs in approximately 5% of men, but only 0.25% of women. Tritanopes who are missing a blue cone do not see well at all in the blue end of the spectrum because they lack the blue cone that is sensitive there. Some hereditary conditions and retinal degenerations, such as RP, result in retinas with no rods. These people are night blind. Anomalous (abnormal) color vision occurs when a person has cones with a different spectral sensitivity than a “normal” person. This produces poorer color vision in most cases, but some cases result in better color vision. For example, because a woman has 2 X chromosomes, if she has 2 different genes for the same cone but with different spectral sensitivities, she can be effectively a tetrachromat (4 colors).

RETINAL VISUAL PROCESSING: RETINAL CIRCUITS

The end effect of the absorption of photons by photoreceptors, which communicate via graded potentials, is the modulation of the release of glutamate at the photoreceptor synaptic terminal. This modulated glutamate release activates the network of neurons in the retina that culminates in the firing of action potentials by retinal ganglion cells, whose axons form the optic nerve and project to several retinal recipient zones in the brain, most notably the lateral geniculate nucleus of the thalamus and the superior colliculus.

Photoreceptor Synaptic Release

The glutamate release from photoreceptor synaptic terminals occurs at what is called a ribbon synapse. This synapse, which involves robust release of glutamate due to its morphology, is common in the retina but rare elsewhere in the CNS. The postsynaptic receptors for glutamate released by the photoreceptors are found on the dendrites of 2 major classes of cells—bipolar and horizontal cells—with the synapses located in the outer plexiform layer (see Figure 11–7). Bipolar cells transmit the photoreceptor signal from the outer plexiform layer to the inner plexiform layer, where they synapse onto the dendrites of amacrine and ganglion cells. Horizontal cell processes modify the photoreceptor to bipolar cell message in a lateral inhibitory network.

Bipolar & Horizontal Cells

Horizontal Cells & Lateral Inhibition

Horizontal cells receive inputs from multiple photoreceptor cells and synapse back onto those photoreceptors. Their role in retinal signal processing is to modulate the signal from photoreceptor to bipolar cells via lateral inhibition. Horizontal cells are activated by photoreceptors over a large region. This produces an estimate of the average local light intensity, a percentage of which is then subtracted from the output of the central photoreceptor that activates a particular bipolar cell. Thus, horizontal cells cause photoreceptors to communicate to bipolar cells the difference between the light they receive and the surrounding light. This is called local contrast, which is relatively independent of overall luminance level (see Figure 11–12). Horizontal cells contact only photoreceptors and other horizontal cells and do not send outputs to ganglion or amacrine cells.

Because photoreceptors are hyperpolarized by light and inhibited by horizontal cells activated by surrounding photoreceptors, more glutamate is released for areas of the image that are darker than the local average, and less glutamate is released for areas brighter than the local average. At all but the lowest light levels, objects are represented by a small modulation around the ambient light level that is either above or below that level. Object detection based on detecting this modulation is robust and independent of the ambient light level. Horizontal cell lateral inhibition reduces redundancy in retinal signals. Thus, the bipolar cell signals the difference between the activation it receives from photoreceptors connected directly to it and the surrounding average intensity.

Bipolar Cells

Different bipolar cell types receive synaptic input from either rods or cones and are distinguished as rod or cone bipolar cells. Cone photoreceptors can synapse onto bipolar cells having 2 different types of response, hyperpolarizing or depolarizing, thus splitting the photoreceptor signal at the first synapse in the retina between depolarizing (on) and hyperpolarizing (off) pathways. This is accomplished by the presence of 2 main types of bipolar cells, which are called depolarizing (on) and hyperpolarizing (off), depicted in Figure 11–13. Hyperpolarizing (off) bipolar cells express AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid)/kainite–type receptors for glutamate and depolarize when their photoreceptors do, in the dark, and hyperpolarize in response to light. Depolarizing (on) bipolar cells have an inhibitory receptor for glutamate (metabotropic MGluR6) that inverts the signal from photoreceptors so that these bipolar cells hyperpolarize to dark and depolarize to light. In summary, in the dark, a photoreceptor cell releases glutamate, which hyperpolarizes (and inhibits) “on” bipolar cells and depolarizes (excites) “off” bipolar cells. In the light, photoreceptor cells release less glutamate; this causes the on bipolar cell to lose its inhibition and become active (depolarized), whereas the off bipolar cell loses its excitation (becomes hyperpolarized) and becomes silent. Similar to the photoreceptors, bipolar cells also use glutamate and graded responses to communicate to their targets, amacrine cells and retinal ganglion cells. (In Figure 11–13, the bipolar cells are referred to as on-center and off-center.) Most mammalian retinas actually have about 10 distinct types of bipolar cells. These include a depolarizing and hyperpolarizing bipolar cell type for each of the 2 or 3 cone types and at least 1 bipolar cell type that receives inputs from rods. Bipolar cells also vary regarding whether they receive input from either a large or small number of photoreceptors.

The bipolar cell situation is different for rods compared to cones. Rods connect only to depolarizing rod bipolar cells. These bipolar cells make gap junction contacts with depolarizing cone bipolar cells and thus activate on-center ganglion cells (and amacrine cells) through the depolarizing cone bipolar cell terminals. The depolarizing rod bipolar cells also have inhibitory, glycine-mediated outputs to hyperpolarizing cone bipolar cells. By disinhibiting these cells for light increases, they activate them for light decrements.

Amacrine Cells

The 2 major bipolar cell types establish 2 parallel pathways in the retina that respond to stimuli lighter than the background (on bipolar cells) or darker than the background (off bipolar cells). Bipolar cell terminals connect to the dendrites of 2 kinds of postsynaptic cells: amacrine and ganglion cells.

Bipolar Cell Inputs to the Inner Plexiform Layer

Bipolar output synaptic terminals end in separate layers of the inner plexiform layer, with depolarizing on bipolar cells terminating in the proximal part of the inner plexiform layer and hyperpolarizing off bipolar cells terminating in the distal inner plexiform layer. These parallel on and off pathways are maintained through the connections to the retinal ganglion cells and to the thalamus.

Laterally Inhibitory Amacrine Cells

Amacrine cells are laterally interacting interneurons. Some have a similar function in the inner plexiform layer to that of horizontal cells in the outer plexiform layer. These amacrine cells conduct inhibitory signals from the surrounding bipolar cells, so that the ganglion cell responds to the difference between the light in its central area and the surrounding area. This action reduces the amount of redundant information that ganglion cells must transmit.

Amacrine Cells Generate Complex Receptive Field Properties

Some amacrine cells generate complex receptive field properties. Other amacrine cells sculpt the bipolar cell output to ganglion cells to produce a variety of ganglion cell classes that respond to particular features of the visual input, such as motion, edges, or colors. Most amacrine cells use the neurotransmitter γ-aminobutyric acid (GABA), but others use glycine, neuropeptides, and even acetylcholine to produce complex ganglion cell response properties. Thus, from a limited number of bipolar cell types, there emerge ganglion cell classes that respond to different colors, movement, or the presence of long edges or corners. Over 30 classes of amacrine cells help to create all these ganglion cell classes.

Ganglion Cells

Retinal ganglion cells are the sole output from the retina. Their axons form the optic nerve that conducts action potentials to approximately 15 retinal recipient zones in the brain. The most important of these for visual acuity and perception is the projection to the lateral geniculate nucleus (LGN) of the thalamus. Important projections are also made to the hypothalamus and midbrain areas such as the pretectal and accessory optic system nuclei.

Thalamic Layer-Projecting Ganglion Cell Classes

On and off bipolar cells are connected to similarly responding ganglion cells called on-center and off-center ganglion cells, respectively (Figure 11–14). The 2 major ganglion cell classes project to distinct layers within the dorsal LGN of the thalamus. These ganglion cell classes are called parvocellular (which means small) and magnocellular cells, although they may also be called midget and parasol cells, respectively. Both parvo- and magnocellular ganglion cells have so-called “concentric” receptive fields. Concentric means that there is an excitatory central region, usually coextensive with the ganglion cell dendritic tree, in which light-on stimulation produces an excitatory response in on-center ganglion cells, whereas light-off stimulation produces and excitatory response in off-center ganglion cells. Usually the best response occurs when the light increment or decrement exactly matches or fills the excitatory receptive field center region.

Concentric receptive field ganglion cells have regions surrounding the excitatory center that have the following 2 main characteristics. (1) Stimulation with spots larger than the receptive field center, that is, invading the inhibitory surround, produces a weaker response than stimulation confined to the center; hence the name, inhibitory surround. (2) The surround also has a property called “antagonistic.” An antagonistic surround is one in which surround stimulation in an on-center ganglion cell produces light-offset excitation, whereas light-onset stimulation produces an excitatory response in off-center ganglion cells (see bottom of Figure 11–14). Both parvo- and magnocellular ganglion cells that form the main projection to the LGN of the thalamus have antagonistic surrounds, but many other ganglion cells have only inhibitory surrounds.

Parvocellular/Midget Ganglion Cells

In the fovea, all ganglion cells are a particular type of parvocellular ganglion cell in a neural circuit called the midget system, in which a single photoreceptor (a cone) activates a single bipolar cell, which feeds a single ganglion cell. However, outside the fovea, there are generally many photoreceptors (rods or cones) that are connected to 1 bipolar cell and several bipolar cells are input to 1 parvocellular ganglion cell. The responses of parvocellular ganglion cells resemble their respective depolarizing or hyperpolarizing bipolar cell input, except that the ganglion cell excitatory postsynaptic potential produces action potentials that will transmit the ganglion cell message via its axons to the brain. Parvocellular ganglion cells have brisk-sustained responses, meaning that they have high action potential firing rates that are well modulated by small changes in light intensity. These are the most numerous ganglion cells in the human retina, constituting approximately 1 million of the approximately 1.2 million total ganglion cells. They form the main projection to the upper 4 layers of the LGN of the thalamus, called the parvocellular layers, which is a relay in the high-acuity, conscious visual pathway. Parvocellular ganglion cells code for color and fine detail. Their small size and small receptive fields mediate very high acuity in the human fovea.

Magnocellular/Parasol Ganglion Cells

Virtually all mammalian retinas also have the magnocellular ganglion cell class. Although absent from the fovea itself, these ganglion cells range from just outside the immediate fovea to the extreme periphery. Magnocellular ganglion cells respond more transiently than parvocellular cells, and they have larger integration areas (receptive fields) than parvocellular cells at any given eccentricity. Magnocellular ganglion cells have relatively more amacrine inputs than parvocellular ganglion cells and respond well to changes in the visual image, such as motion, even at low contrast. They project to the bottom 2 layers of the LGN of the thalamus, the magnocellular layers, in primates.

Other Ganglion Cell Classes

The network of amacrine cells also gives rise to ganglion cell classes that respond only to specific features of the visual input. Some of these ganglion cell classes project to the thalamus, but others project to different brain areas. Bistratified retinal ganglion cells project to the koniocellular layers of the LGN and are proposed to function in color vision. Directionally selective ganglion cells are not only sensitive to movement, but also to its direction. These cells project to a variety of brain nuclei that enable target tracking and using vision to maintain balance and orientation. A small percentage of ganglion cells are intrinsically light sensitive. This means that they express their own photopigment that causes them to respond directly to light, without photoreceptor input. These intrinsically photoreceptive ganglion cells are sensitive to the overall light level, and they project to the suprachiasmatic nucleus in the hypothalamus and control circadian rhythms. Other intrinsically photoreceptive ganglion cells project to the Edinger-Westphal nucleus and control the pupillary light reflex.

Chromatic Processing

The visual system is thought to interpret color in an antagonistic way, with antagonism occurring at the level of ganglion cells. A range of wavelengths stimulate each of the cone types to varying degrees. Color vision requires that the signals from different cone types be kept separate until color-opponent ganglion cells can compare the ratio of activity between cones. The activity of red cones is typically opposed to the activity of green cones in some color-opponent ganglion cells. The other main type of color-opponent ganglion cell has blue opposed to the combination of red and green. Color opponency is another parallel visual pathway, like on versus off in the bipolar and ganglion cells.

Glaucoma

The major disease of ganglion cells is glaucoma, one of the leading causes of blindness. Glaucoma is an optic neuropathy associated with elevated intraocular pressure in the aqueous humor, in which retinal ganglion cells die due to damage to their axons. However, glaucoma causing optic nerve damage can also occur in individuals with normal intraocular pressure. Retinal ganglion cells tend to be lost in a characteristic pattern with an associated expanding visual field loss that often progresses to blindness. There are 2 main categories of glaucoma: open-angle and closed-angle. Open-angle glaucoma is chronic, typically symptomless, and of slow progression until significant vision has been lost. It is associated with poor drainage of the scleral venous sinus that builds up pressure in the anterior chamber and then the vitreous.

Closed-angle glaucoma refers to the situation where the iris sticks to the lens, preventing fluid from escaping from the vitreous to the anterior chamber. It often has a sudden painful onset. Treatment can involve using a laser to make a hole in in the iris to allow drainage, but without treatment, vision can rapidly deteriorate.

In both types of glaucoma, ganglion cell axons in the periphery tend to die first so that visual field loss tends to progress from the periphery to the center, yielding a stage near the end of the disease characterized as “tunnel vision.”

What Does the Human Eye Tell the Human Brain?

Although the exact number of distinct ganglion cell classes in the human eye is unknown, the number is likely to be at least 20, as is the case with other mammals in which this issue has been investigated. How are these ganglion cell classes distributed across the retina? Like the photoreceptors, ganglion cells show an eccentricity factor: Ganglion cell somas and dendritic trees are smaller in and near the fovea than in the periphery. At each locus (eccentricity) in the retina, each ganglion cell class appears to tile the retina, with no gaps between dendritic trees and little overlap. However, the size of each ganglion cell class at each eccentricity varies as a function of class.

Figure 11–15 illustrates how, at 1 particular retinal location, 6 to 7 parvocellular ganglion cells might tile the interior of 1 magnocellular ganglion cell. The magnocellular ganglion cells themselves form a similar tiling on a larger scale. If there are 20 ganglion cell classes and each tiles the retina, then every point on the retina is within the receptive field of 1 of each of the 20 different ganglion cell classes, each of which extracts some particular information from the visual scene for projection to the brain (where there are at least 15 different retinal recipient zones). Some of these ganglion cells care about color, others about edges, and still others about movement. The retina does not transmit a picture of the visual scene for the brain to analyze; rather, it transmits 20 transforms of the visual scene that are used by the brain for perception and visually guided behavior in many different ways. This must be clear from the numbers: There are >120 million photoreceptors but only approximately 1.2 million ganglion cells. Only in the midget/parvocellular system within the fovea does the retina approach the idea of sending the output of a single cone via a single ganglion cell to the brain. Elsewhere, each parvocellular ganglion cell extracts some average chromatic signal from the photoreceptors that influence it, and other ganglion cells extract more complex information.

SUMMARY

• Vision is a hierarchical process.

• Mechanical elements of the eye, such as the lens and pupil, control the focus and intensity of light reaching the retina.

• Eye movements are an integral part of visual processing.

• Retinal photoreceptors convert captured photons to modulated glutamate release.

• Retinal bipolar cells communicate positive and negative excursions around the mean light level to the inner retina.

• Amacrine and ganglion cells in the inner retina sculpt and transmit extracted aspects of the changing light distribution for transmission to the brain.

• Retinal ganglion cells project to 15 different retinal recipient zones in the brain.

• The most important retinal recipient zone is the lateral geniculate nucleus of the thalamus.

• Blindness and visual deficits occur primarily in the retina, but also in higher visual processing centers.

OBJECTIVES

THE OUTPUT OF THE RETINA: WHAT THE EYE TELLS THE BRAIN

The common metaphor that the eye is like a camera works reasonably well for some of the optical components of the eye such as the lens and pupil but fails dramatically in describing the output of the retina. A camera is a recording device whose purpose is to register (and, in the video case, transmit) a true rendition of the light distribution focused on the film or detector. Virtually all cameras have a homogeneously dense pixel array, whereas the photoreceptor density in the retina varies markedly from center to periphery. More importantly, a camera records or transmits the light intensity at each pixel, but in the retina, most ganglion cells receive inputs from many photoreceptors, and many ganglion cells extract a highly processed signal from these photoreceptor inputs, such as motion occurring in a particular direction or the presence of an edge.

A Review of Ganglion Cell Classes

The mammalian retina, including that of primates such as humans, contains at least 20 to 25 classes of ganglion cells whose axons terminate in at least 15 different retinal recipient zones. The responses of these different ganglion cells mediate many different types of visual functions by virtue of their receptive field properties (what causes them to respond) and central projections (where in the brain the ganglion cell axons terminate). Many retinal ganglion cells terminate in multiple retinal recipient zones and thus participate in multiple visual processing pathways.

Functions of Visual Output

There are several visually mediated functions of which we are conscious, and many of which we are not conscious. Conscious visual functions include perception, identification, and visually controlled movement guidance. Unconscious visually mediated processes include the control of eye position and pupil size and circadian rhythms. These processes are mediated by different ganglion cell classes in different visual pathways.

Ganglion Cell Response Selectivity

Different ganglion cell classes differ considerably in their response selectivity, which is related to the function of the visual circuit to which they project. At one extreme are the midget or parvocellular ganglion cells of the fovea whose input arises primarily from a single bipolar cell, which in turn is driven mostly by a single red or green cone. The intracellular voltage response of these midget ganglion cells closely resembles that of the single bipolar cell driving them. In the case of off-center ganglion cells, the hyperpolarizing bipolar cell response itself resembles that of the cone to which it connects. Because the spike output of the midget ganglion cell follows its intracellular potential fairly faithfully, this spike output is a reliable index of the photoreceptor response as though passed through a voltage to frequency converter. The bipolar cell connected to the on-center midget ganglion cell has an inhibitory mGluR6 receptor for glutamate, so its response is the reverse of that of the hyperpolarizing bipolar cell. The on-center midget ganglion cell thus signals light increases fairly faithfully, whereas the off-center midget ganglion cell signals light decreases.

Many retinal ganglion cells, in contrast, have response characteristics that depend on complex spatiotemporal characteristics of the light distribution over the thousands of photoreceptors and hundreds of bipolar cells to which they are connected. Notable examples include selectivity for direction of motion or the presence of edges. Many of these ganglion cells exhibit memory dependence in their responses such that the pattern of light stimulation many seconds previously strongly influences current responses, which underlies phenomena such as contrast gain control.

Overview of Ganglion Cell Retinal Recipient Zones

Conscious vision is mediated almost exclusively via the ganglion cell projection to the thalamus, specifically, the lateral geniculate nucleus of the thalamus. The next most important projection is to the superior colliculus (SC) for visual orienting and eye movement control, especially voluntary eye movements called saccades. Retinal ganglion cells also project to several pretectal and accessory optic nuclei, to the hypothalamus, and to several other less well-understood retinal recipient zones.

Thalamus

All sensory systems except for olfaction involve a projection to a sensory modality–specific region of the thalamus that in turn projects to what is called a primary cortex for that sense. The thalamic pathway is always crucial to conscious awareness of sensory input for that sensory modality. In the visual system, the visual thalamus region is called the lateral genicular nucleus (LGN). Two ganglion cell classes, called parvocellular and magnocellular ganglion cells, form the major projection to the LGN. Several other ganglion cell classes, usually termed koniocellular due to their smaller soma sizes, also project there. Sparser projections from the retina are also made to the ventral geniculate nucleus and possibly the pulvinar, which is an integrating and visual attention control thalamic area.

Superior Colliculus

The SC receives axons from almost all ganglion cell classes except the parvocellular cells. The SC evolved from its homolog in nonmammalian vertebrates, the optic tectum.

Pretectum

The pretectum is a set of midbrain nuclei that receive direct retinal projections. The main pretectal nuclei are the nucleus of the optic tract (NOT), the olivary pretectal nucleus (ON), and the anterior, medial, and posterior pretectal (PP) nuclei. The pretectum is involved in mediating visual reflexes such as optokinetic nystagmus (the cycle of tracking and saccadic return elicited by a stimulus such as a moving grating) and pupillary reflexes.

Accessory Optic System

The accessory optic system (AOS) consists of several retinal recipient nuclei that are involved in attention, eye movements, and vestibular coordination with visual input for balance. The AOS includes the medial terminal nucleus, lateral terminal nucleus, and dorsal terminal nucleus. Many ganglion cells projecting to the AOS are directionally selective in their responses.

The Hypothalamus & Other Specialized Recipient Areas

The suprachiasmatic nucleus of the hypothalamus, Edinger-Westphal nucleus, and other specialized recipient areas receive inputs from specialized retinal ganglion cells such as intrinsically photoreceptive ganglion cells.

VISUAL PROCESSING IN THE BRAIN: THE THALAMUS

Approximately 80% of all retinal ganglion cells, including the important midget system that comprises the fovea, project to the LGN of the thalamus. Virtually all magnocellular ganglion cells also project there, as well as several additional cell classes called koniocellular cells. By virtue of the optics of the eye, there is a correspondence between visual field angular location and retinal location, a topology. This topology is not retained within the optic nerve, but it is reconstituted in a layer- and class-specific manner in the LGN. This involves a sorting of fibers in the changeover from the optic nerve to the optic tract at the optic chiasm.

The Optic Nerve & Optic Chiasm

The output of the retina, the optic nerve, consists of the axons of all ganglion cells that exit the eye there. A few centimeters after the axons leave the eye, they enter a structure called the optic chiasm, which means “optic crossing,” where some of the ganglion cell axons from each eye cross at the chiasm and go to the other side of the brain and some do not. Figure 12–1 shows that ganglion cells from the part of the right retina farthest from the nose (called the temporal retina) project ipsilaterally to the LGN on the right side. This portion of the retina receives input from the left visual field. The other half of the retina (closest to the nose, called the nasal retina) projects contralaterally to the left LGN. The nasal retina of the right eye receives input from the right visual field. The nasal left retina, which receives input from the left visual field, projects to the right thalamus with the temporal right retina. Thus, the right LGN receive inputs from both eyes from the left visual field, and similarly, the left LGN receives axons from both eyes, which receive visual input from the right visual field. The nerves leaving the optic chiasm are called the optic tracts. The major destination of the optic tracts is the LGN, but ganglion cells also project to >10 other retinal recipient zones in the brain.

The Thalamic “Relay”

In the LGN, each thalamic relay cell receives inputs from 1 or a few similar ganglion cells that are either parvo- or magnocellular, depending on the layer. For this reason, and because LGN relay cells respond very much like their parvocellular or magnocellular ganglion cell inputs, LGN cells have often been called relay cells. The ratio of retinal input to LGN output is approximately 1:1, consistent with the idea that each retinal ganglion cell mostly drives a single LGN cell. However, other inputs to the LGN from other parts of the brain (including visual cortex) allow gating functions that are associated with attention. These inputs modulate the strength of a neuron’s responses to any particular stimulus based on the contextual importance of that stimulus. There are actually more reciprocal inputs to the LGN from the visual cortex, which feeds information back and modifies it, than from the retina going up to the LGN.

Note that no LGN cells are binocular, but the LGN on each side receives inputs from the portions of both eyes viewing the opposite side visual hemifield. Damage to the LGN on 1 side or to 1 optic tract, but not 1 optic nerve, produces a visual field scotoma (blind area) under binocular viewing. A different result occurs if there is damage to the upper 4 versus lower 2 layers of the LGN. Damage to the upper 4 parvocellular layers results in a loss of color and fine detail perception. Damage to the bottom 2 magnocellular layers reduces the ability to respond to changes in the visual image, such as motion, particularly at low contrast.

The main output of all LGN layers is to visual cortex in the occipital lobe via a fiber pathway called the optic radiation. There are also LGN cells between the 6 main layers called interlaminar cells, which project to visual cortex. The overall projection scheme from the retina to visual cortex is shown in Figure 12–2, with the scotoma (visual field loss) associated with damage to various loci in the pathway.

The LGN is located at the lateral inferior edge of the thalamus, just lateral to the medial geniculate nucleus that similarly relays auditory information, as shown in Figure 12–3. The LGN is also close to the midbrain inferior and superior colliculi, which relay auditory and visual information, respectively.

An important principle of cortical organization with respect to its representation of the sensory periphery is shown in Figure 12–4. This is the so-called cortical magnification factor, in which the foveal representation of the retina occupies a much larger cortical area than other retinal areas. The explanation for this is simple and fundamental, however. The neocortex neural circuitry is fundamentally similar throughout its extent (with important exceptions to be discussed elsewhere). This implies that its processing capabilities are similar per unit area, which is reflected in the fact that the number of thalamic inputs to a given cortical area is relatively constant for a given sensory modality. Because the density of retinal ganglion cells and thalamic relay cells is much higher in the center than the periphery of the retina, as shown in the top of Figure 12–4, this relatively small retinal area projects to a much larger cortical area than the periphery, keeping the number of retinal ganglion cells and thalamic relay cells relatively constant per unit of cortical area.

THE SUPERIOR COLLICULUS & EYE MOVEMENTS

Most ganglion cell classes except the parvocellular cells project to the SC in the midbrain. SC layers exhibit a retinotopic map of the surrounding space. This structure is the mammalian analog of the optic tectum in nonmammalian vertebrates who generally lack any significant visual cortex. In such animals, the optic tectum is the main visual processing center for controlling visually guided behavior. In mammals, however, the SC controls orienting and eye movements, and visually guided behavior and perception are mediated by thalamic pathways to the cortex. The superficial layers project to deeper layers that encode saccade targets. Thus, the output of the deep layers of the colliculus controls eye movements to visual targets. Neural activation at a specific locus in the SC evokes a response directed toward the corresponding spatial location. Lesions to the SC reduce or eliminate orienting eye and head movements to peripheral objects.

The main class of eye movements controlled by the SC are called saccades. Saccades are large jumps in eye position from one fixation point to another. Voluntary saccades occur at a rate of about 3 to 4 times per second. Involuntary saccades also occur when something appears in the visual periphery and we direct our gaze to place the central fovea at the peripheral target. Our eyes also constantly make smaller movements called microsaccades, drift, and tremor. One function of these movements is to prevent image fading, because artificially stabilizing images on the retina invokes adaptation mechanisms that eliminate neural firing and cause vision to disappear.

THE PRETECTUM

The pretectum is a set of midbrain nuclei that receive direct retinal projections and mediate visual reflexes such as optokinetic nystagmus (the cycle of tracking and saccadic return elicited by a stimulus such as a moving grating) and pupillary reflexes. The main pretectal nuclei are the NOT, ON, and the anterior, medial, and PP nuclei. These nuclei are located anterior to the SC, posterior to the thalamus, and above the periaqueductal gray. The NOT and ON receive inputs from intrinsically photosensitive ganglion cells. These ganglion cells have their own photopigment and respond directly to overall light levels.

Pretectal nuclei send their outputs to the thalamus, SC, reticular formation, pons, and inferior olive. The ON and PP nucleus project to the Edinger-Westphal nucleus, which controls pupil diameter. There are also pretectal projections to the pulvinar, a thalamic integration and control area for visual attention. The NOT receives inputs from directionally selective ganglion cells that mediate its involvement in smooth pursuit. Some pretectal nuclei may have roles in nociception and sleep regulation.

ACCESSORY OPTIC SYSTEM

Other retinal ganglion cell projection areas include the AOS, whose nuclei include the medial terminal nucleus, lateral terminal nucleus, and dorsal terminal nucleus. These areas are concerned with attention, eye movements, and vestibular coordination with visual input for balance. AOS nuclei receive projections from directionally selective retinal ganglion cells to mediate these functions.

THE HYPOTHALAMUS & OTHER SPECIALIZED RECIPIENT AREAS

Intrinsically photoreceptive retinal ganglion cells respond directly to light via their own membrane photopigment. These cells respond slowly at high firing rates to overall changes in average luminance, which is quite different from the complex, center-surround structure of most retinal ganglion cells. The intrinsically photoreceptive ganglion cells send their outputs to nuclei such as the suprachiasmatic nucleus, which controls circadian rhythms, and the Edinger-Westphal nucleus, which controls pupil dilation. There are indications of other retinal recipient areas of the brain such as the pulvinar, ventral thalamus, and Raphe about which little is known except that these projections are likely to be involved in unconscious or autonomic processing of visual information.

THE OCCIPITAL LOBE

As in all sensory systems with the exception of part of olfaction, the projection to visual cortex is primarily from a specific region of the thalamus. In the case of vision, the thalamic relay cells are in the LGN. Cells there send their axons to the visual area of the neocortex at the pole of the occipital lobe via a fiber tract called the optic radiation (see Figure 12–2). The recipient area of the occipital lobe is called visual area 1 (V1), as well as area 17 (from the Brodmann map) and the striate cortex, because of a dense cell layer that appears in histologic stains. This pathway mediates almost all the vision of which a person is conscious (versus vision functions, such as pupil contraction and dilation, which a person is neither conscious of nor able to control voluntarily).

Representation of the Visual Field

The 1 million LGN relay cells project in a topographic manner to primary visual cortex (V1) at the pole of the occipital lobe and within the calcarine sulcus. The left visual field is represented on the right visual cortex, and the right visual field on the left visual cortex.

Retinotopy

The cortical cells driven from each LGN form a map of visual space distorted by the higher retinal ganglion cell representation in the center versus the periphery. Figure 12–5 shows how the visual field and LGN thalamic representation are mapped onto the inner surface of the right visual cortex within the calcarine sulcus. Central visual areas, including the large foveal representation, occupy the most posterior regions of the sulcus, with the peripheral representation extending anteriorly.

Processing Expansion

The projection from the LGN to the visual cortex consists of approximately 1 million axons, similar to the number of ganglion cell axons that project to each LGN from the retina. However, this million-cell input from the LGN drives approximately 200 million cortical cells in V1. However, cells in V1 are more specialized for stimulus parameters such as orientation and movement direction. This results in a sparser representation of the visual stimulus in cortex than retina or LGN. Whereas in retina and LGN, most visual stimuli cause some change in firing of most cells in that area of the visual field, in visual cortex, only a small percentage of more selective cells respond well to any stimulus, depending on specific aspects of the stimulus such as its orientation or motion direction. Thus, V1 neurons extract local stimulus features from a restricted region of the retina.

Columnar Organization of V1

Organization of Inputs

The cellular architecture of V1, or striate cortex, is similar to that of neocortex throughout the brain in that it has 6 layers with a characteristic input/output pattern. Figure 12–6 illustrates the layers, the types of cells (pyramidal and local) that occupy the layers, and the characteristic inputs and outputs. The general plan for visual cortex is that thalamic inputs terminate in layer IV, and in this regard, the visual cortex is similar to other sensory cortices, and, in fact, all cortices. However, the visual cortex differs from other cortical areas in the density of inputs. There are >2 million LGN inputs in the visual pathway, versus, for example, a few tens of thousands of auditory inputs to the auditory cortex. The result in the visual cortex is a thickening of layer IV compared to other cortical areas with a very high cell density. This high-density, thickened layer IV makes the cytoarchitecture of layer IV stand out compared to other cortical areas and leads to the name “striate” cortex for V1. Layer IV in the striate cortex is so thick that it has been subdivided into sublayers A to C, and layer C is further subdivided in the sub-sublayers α and β, as shown in Figure 12–7.

The input to visual cortex from the LGN is different for parvocellular versus magnocellular layers and still partially segregated. Figure 12–7 shows that the parvocellular layer (layers 3 to 6) axons from LGN terminate in visual cortical layer IVCβ. Cells in this layer project upward to layer III. Magnocellular inputs arrive in cortical layer IVCα, however. There is also a horizontal regional specialization in striate cortex, referred to as “blobs” and “interblobs,” which are areas of high and low cytochrome oxidase staining, respectively. The cytochrome oxidase staining differentiates areas of relatively high versus low metabolic activity. Input to the blob areas is almost exclusively parvocellular. Input to the interblob areas, however, comes directly from the magnocellular input and secondarily through a cortical projection in layer III from parvocellular cells.

Cell Physiology

As mentioned earlier, each visual LGN relay cell drives, on average, several hundred V1 cells. The cells in V1 have more specificity in their responses than retinal ganglion cells or LGN relay cells. As Hubel and Weisel famously showed, V1 cells are almost all sensitive to the orientation of an extended stimulus on the retina and do not fire action potentials unless there is an edge, represented by several collinear ganglion cells. Many of these cells are also sensitive to direction of motion, and others are sensitive to edge length. Cells that are purely orientationally selective are called simple cells. Cells that are both orientationally selective and direction of motion selective are called complex cells, which also differ from simple cells in several other response properties. An additional class, called hypercomplex or end-stopped cells, is orientationally selective, directionally selective, and selective for stimulus length.

Electrophysiologic recordings made in V1 show that the orientation selectivity in V1 has a specific horizontal organization in sync with the blob and interblob regions. At any location in visual cortex, all cells at various depths have the same orientation selectivity, and the preferred orientation changes systematically as one moves the electrode horizontally across the cortical surface. A region whose extent includes all orientations is termed a column and tends to extend from one cytochrome oxidase blob to the next.

At the same time, V1 cells are the first in the visual pathway to be binocularly driven. The binocularity of cells also systematically changes with cortical horizontal position, such that as one moves a microelectrode, one sees a systematic change from cells totally driven by the left eye, to equal eye contribution, to right eye dominance. At any given cortical location, ocular dominance and orientation selectivity change at relatively right angles, leading to the idea of a 2-dimensional column (Figure 12–8) in which ocular dominance changes from left to equal to right and orientation selectivity changes from horizontal to vertical.

HIGHER CORTICAL PROCESSING AREAS

V1, at the posterior pole of the occipital lobe, contains an entire representation of the retina, and therefore the visual world, onto which orientation, motion direction, edge length selectivity, and ocular dominance are grafted onto the position map. V1 sends its outputs to several other cortical areas and back to the thalamus. The thalamic return projection modulates the input V1 itself receives in ways that are not completely understood but appear to contribute to the selectivity of V1 cells and serve as a basis for visual attention.

V1 projects to adjacent cortical areas such as V2 and V3. Neurons in these areas have similar response properties to those in V1, and the additional processing done in V2 and V3 is not well understood. After these areas, however, the visual cortical output diverges into 2 anatomically differentiated streams whose functions are relatively well understood and that have important clinical ramifications. An outline of this dual projection system is shown in Figure 12–9.

VENTRAL STREAM FOR OBJECT PERCEPTION

The V2-V3 complex forms 1 projection via the inferior longitudinal fasciculus into the inferior portion of the temporal lobe (also called inferotemporal cortex). This cortical analysis system is dominated by parvocellular input. It specializes in object identification and related aspects of perception and is sometimes called the what pathway. Cortical areas in the what pathway include the TE (temporal; also called PIT [posterior inferior temporal]) and TEO (temporal occipital; also called anterior inferior temporal) areas. The beginning stage for this pathway is usually taken to be area V4, where most neurons are color selective. More anterior neurons in this pathway (TEO) respond only to very complex patterns such as the shape of a hand. In the fusiform gyrus (fusiform face area) near this area are many neurons that are face selective. The output of the ventral processing stream includes the hippocampus and anterior areas of the frontal lobe.

Dorsal Stream & Visually Guided Behavior

A second processing stream from the V2-V3 complex passes through area V5 into the parietal lobe via the superior longitudinal fasciculus. Neurons in V5 (also called MT [middle temporal]) are almost all directionally selective, and this pathway is dominated by magnocellular input. Major areas in the dorsal stream include the middle temporal and medial superior temporal areas. The dorsal pathway has been termed the where pathway because of its involvement in the use of vision for navigating in space. Some researchers believe, however, that a better term would be the how to pathway because of its necessity for visually guided behaviors, such as catching a ball or running through an obstacle course. Damage to this area causes apraxia, which is the inability to skillfully conduct tasks such as manipulating objects based on visual information.

Dorsal-Ventral Crosstalk

Despite the fact that the ventral stream is dominated by parvocellular input and the dorsal stream by magnocellular input, there is crosstalk between them. In the structure-from-motion phenomena, a few visible spots on a moving object are sufficient to give away its shape identity during movement. In this case, dorsal motion-detecting neurons must communicate with object-detecting neurons in the ventral pathway. Depth perception is another example of a holistic perception that may be derived from ventral pictorial cues or dorsal motion parallax.

Blindness, Agnosias, & Visual Field Deficits

Blindness

Blindness typically originates from damage or cell loss early in the visual pathway, including retina, thalamus, and V1. Because of the parallelized processing at higher levels, losses there tend to be selective and are called agnosias, which is the loss of a specific visual ability without the loss of other abilities, and are usually due to damage to a high order cortical visual area. The term agnosia was coined by Sigmund Freud.

Most blindness originates in the retina. Photoreceptor retinopathies, such as macular degeneration, retinitis pigmentosa, and diabetic retinopathy, are caused by photoreceptor death. The other major retinopathy, glaucoma, is caused by death of retinal ganglion cells, usually associated with excessive pressure within the eye.

The cause of retinitis pigmentosa is a hereditary degeneration of retinal rod photoreceptors, progressing to loss of all vision due to secondary death of cones. The retinitis pigmentosa sequence starts with night blindness from rod death; later, all peripheral vision is lost, leading to tunnel vision. Central vision is spared for some time because there are no rods in the fovea. In the last phase, the fovea also degenerates, and all vision is lost. There is no treatment for retinitis pigmentosa, and the resulting blindness is irreversible.

Photoreceptors die in macular degeneration and diabetic retinopathy because of metabolic disorders that lead to neovascularization or inadequate retinal nutrition. The sequence typically begins with photoreceptor death and then is followed by death of other retinal neurons. Retinal death can also occur if the retina detaches from the underlying pigment epithelium after a blow to the head.

Glaucoma is caused by the death of retinal ganglion cells and is typically associated with excessive pressure within the eye restricting axonal trafficking at the optic nerve head. There are 2 major subtypes. Closed-angle glaucoma occurs when the iris obstructs drainage canals from the vitreous in the middle of the eye to the aqueous chamber behind the cornea, allowing vitreal pressure to build up. This type of glaucoma is the rarer of the 2 types and is treatable using laser surgery to blast small holes in the iris. Open-angle glaucoma usually has a heritable component and is often associated with high blood pressure. It can often be controlled with medication. In either glaucoma subtype, however, once the ganglion cells have died, the blindness is irreversible.

Amblyopia is a dysfunction in which 1 eye appears to be blind, but the eye itself presents as being neutrally normal. Its origin is often a congenital optical problem in the eye such as severe astigmatism or a cataract that degrades the neural output of the eye during the critical developmental period from birth to about 6 years. Cortical synapses are taken over almost completely by the better eye, and after the plastic critical period, the damage is irreversible, even if at a later age the optical problem with the amblyopic eye is corrected.

Dorsal/Ventral Stream Damage

Damage from a tumor or stroke to areas in the dorsal stream causes apraxias or loss of ability to use vision to manipulate objects. A well-known agnosia called akinetopsia was experienced by a patient with bilateral damage to V5. This patient was unable to avoid cars crossing the street or pour a cup of tea because she could not estimate time to impact or time to full cup from the motion signal. Her visual experience is something like that in a strobe light. Damage to the parietal lobe, particularly on the right side of the brain, results in hemineglect, in which people are inattentive to objects in their left visual field.

Ventral stream damage is often associated with loss of pattern perception. One example is achromatopsia, occurring from damage to V4 in the ventral stream. This is the loss of the ability to identify colors and is distinct from color blindness. Color blindness occurs because of the loss of 1 or more cone types, so that the person cannot see any difference between some colors. In achromatopsia, however, the person sees a difference between color patches, but they do not seem to have any quality of color but, rather, appear as different dirty shades of gray.

Face & Complex Perception

Although the ventral visual processing stream has no final output or highest level area, the most complex neural selectivity known in that stream is undoubtedly the face-selective cells in the fusiform face area. Many cells recorded in this area in nonhuman primates respond well only to faces, and this area is activated in functional magnetic resonance imaging scans in humans when they view faces. Patients with damage to this area also may lose the ability to visually recognize another’s face while still being able to identify the person by the sound of the person’s voice or the way he or she walks.

SUMMARY

• Vision is a hierarchical process.

• Retinal ganglion cells project to 15 different retinal recipient zones in the brain.

• The most important retinal recipient zone is the LGN of the thalamus.

• The LGN projects to the visual cortex, which engages in complex analysis of the visual scene, and projects to other brain areas.

• Visual areas of the brain produce neural images that represent different aspects of the visual world based on factors such as color, the location of edges, and movement.

• Blindness and visual deficits occur primarily in the retina, but also occur in higher visual processing centers.

OBJECTIVES

OVERVIEW

The sense of hearing allows us to capture sound energy from the environment and informs us about the identity of the sound emitter and its location. Complex sounds are also an essential means of complex communication for biological organisms, reaching their pinnacle in human language. Sound consists of a sequence of air pressure pulses created by vibrating objects such as vocal cords. Vibrations cause compression and rarefaction of the air, which result in the characteristics of frequency, the number of cycles per second, amplitude, and sound intensity, which is usually measured in decibels, a log scale. The sensitivity of the auditory system is very close to the absolute threshold created by random movement of air molecules. Sound reception begins with mechanical modification and transduction in the outer and middle ear and neural coding in the inner ear at the cochlea. Relays in the brainstem and thalamus pass the encoded auditory input to the auditory cortex in the mid-superior temporal lobe, where higher auditory processing allows us to understand language and appreciate music.

Near the cochlea of the auditory system are the semicircular canals of the vestibular system. Hair cells in those canals respond not to external sound input but to motion along 3 axes of rotation. The neural output of the vestibular system is essential for maintaining balance and works with motion-detecting cells from the retina in central pathways that code our motion in space and allow the performance of complex, balanced motion.

PROPERTIES OF SOUND

As any object vibrates, it compresses air on the side it is moving toward and rarefies air on the other side. These pressure pulses are transmitted and spread by collisions of the molecules in the air as longitudinal waves. This is illustrated in Figures 13–1 and 13–2. Most objects have a natural frequency at which, when struck, they vibrate. Objects will also typically vibrate at various amplitudes at all the integral multiples (harmonics) of that frequency.

Intensity & Loudness

Sound waves have the physical properties of intensity, which we perceive as loudness, and frequency, which we perceive as pitch. Figure 13–1A shows a sinusoidal pressure oscillation at a particular frequency on an arbitrary time and amplitude scale. If the amplitude is increased, the vertical pressure excursions are larger (Figure 13–1B). If the frequency is increased, the wavelength is shorter (Figure 13–1C).

There is a million-fold range of sound amplitudes to which our ears respond. A compact notion used to describe this range is the decibel scale, named after Alexander Graham Bell. Figure 13–3 shows the how the decibel varies logarithmically with sound pressure. The decibel scale is usually referenced to a basis pressure p₀ of 20 µPa, close to the threshold of hearing, and is then referred to as the sound pressure level in decibels.

The human hearing threshold is not the same for all frequencies. Figure 13–4 shows the audibility curve for human hearing, which shows the threshold, and the perceived loudness above threshold, as a function of frequency. Our threshold is lowest for frequencies of a few thousand hertz (hertz is sinusoidal cycles per second, after Heinrich Hertz). This frequency range is approximately the same as that of human speech, the range at which telephones work. The bass and treble controls on music stereo amplifiers are there to boost low and high frequencies that are below threshold when music is played at low volume. At higher sound pressure levels, the audibility curve is flatter and the compensation needed is much less. The level of 140 dB is the threshold of pain or feeling, and rapid damage occurs to the auditory hair cells at such levels. Prolonged exposure to levels of 120 dB or even lower can also damage hearing.

Frequency, Pitch, & Harmonics

Most real-world sounds are not simple sine waves, but neither are they arbitrary combinations of frequencies. Objects can generally sustain vibrations whose wavelengths fit an integral number of times within the length of the object. Figure 13–5 illustrates this phenomenon for the 1-dimensional case of a vibrating string. A plucked string may initially be excited by a range of frequencies, as shown in Figure 13–5A. However, because the 2 ends of the string are clamped, only frequencies whose wavelength is such that the amplitude is 0 at those points will be sustained. Other frequencies whose amplitudes would not be 0 at the clamped ends are quickly damped out.

Figure 13–5B shows that the set of sustainable frequencies begins with f1, whose wavelength is half the distance between the fixation points, and includes all integral multiples of f1, the harmonics of f1. The summation of all these frequencies at various amplitudes may look like Figure 13–5C, which is not a pure sine wave. Natural 3-dimensional objects emit sounds with different frequencies corresponding to different resonant lengths of the object along 3 axes and may be quite complicated. Later we will see that the auditory cortex is sensitive to such complex frequency sets and can distinguish them from randomly assembled sets of frequencies.

ANATOMY OF THE EAR

The first stages of auditory processing are mechanical transformations of the sound pressure variations reaching the ear in the outer and middle ear. Then, in the inner ear, sound pressure is transformed into neural impulses sent to the brain.

The Outer Ear & Localization: Tuning & Directing

The outer ear consists of 3 main elements: the pinna, auditory canal, and eardrum. The pinna is what people colloquially refer to as the ear. This structure filters the sound waves as it directs them into the auditory canal. The filtering changes the sound composition as a function of the elevation of the sound emitter, allowing localization of sounds in elevation by decoding this frequency alteration. The pinna and the nomenclature of its structures are shown in Figure 13–6.

Sounds then traverse the auditory canal on the way to the eardrum. The overall structures of the outer, middle, and inner ear are shown in Figure 13–7. The auditory canal has a resonant frequency of approximately several kilohertz that enhances the frequency range of best human hearing. The auditory canal ends at the tympanic membrane or eardrum, which is set in motion by the sound pressure as modulated by the pinna and auditory canal.

The Middle Ear Bones & Amplification

Figure 13–8 shows a more detailed view of the middle ear, which is composed of 3 bones called the malleus, incus, and stapes, the smallest bones in the human body. The movement of the tympanic membrane drives the malleus, which moves the incus, which in turn moves the stapes that is attached to the cochlear entrance at the oval window. The arrangement of this system performs a force amplification to translate a larger, weaker movement of the tympanic membrane in air to a stronger, smaller movement of the oval window bounding the fluid-filled cochlea.

The force amplification by the middle ear system is accomplished in 2 ways: (1) the tympanic membrane has a larger area than the oval window, and therefore, force is concentrated from a large to a small area; and (2) the arrangement of the middle ear bones forms a lever, with the tympanic membrane on the long arm and oval window on the short arm. The total force multiplication is on the order of 100 times.

The Inner Ear: Cochlea & Organ of Corti

Sound is transduced into neural signals in the cochlea, whose relation to the rest of the structures of the ear can be seen in Figure 13–7 and Figure 13–9. Situated near the cochlea are the semicircular canals that transduce head orientation and movement into neural signals. The output of both of these inner ear structures travels to the brain via the vestibulocochlear nerve, which is cranial nerve VIII.

THE COCHLEA & SOUND TRANSDUCTION

Sound Wave Transduction & Place Coding

A cross-section of the Organ of Corti within the cochlea is shown in Figure 13–11B. The cochlea has 3 fluid-filled chambers: (1) the scala vestibuli above the organ of Corti in which pressure waves are induced by the oval window, (2) the scala tympani below the organ of Corti, and (3) the cochlear duct containing the organ of Corti with the tectorial and basilar membranes and auditory hair cells. The leverage changes in the middle ear allow the larger, low-impedance vibrations of the tympanic membrane to produce smaller more forceful vibrations of the oval window to transmit pressure pulses into the fluid-filled cavity called the scala vestibuli.

The way sound energy is localized by frequency along the cochlea is shown schematically in Figure 13–10, with the cochlea schematically “unrolled.” The selectivity for vibration frequency along the length of the cochlea is called a place code. High frequencies cause large displacements of the basilar membrane at the oval window entrance to the cochlea, but not farther along its extent. Medium frequencies cause the maximal vibration in the middle of the cochlea, and low frequencies at the end of the cochlea, called the helicotrema, which is actually in the center of the coiled structure (shown in Figure 13–11A).

The place selectivity along the basilar membrane is much greater for high than low frequencies, shown in the bottom panel of Figure 13–10. High-frequency activation of the basilar membrane is restricted to small regions near the oval window, but very low frequencies vibrate almost the entire membrane. This place restriction difference is related to the way the nervous system encodes frequency. Because neural action potential and refractory periods last on the order of 2 milliseconds, neurons cannot generally fire >500 action potentials per second. At sound frequencies <500 Hz, a large percentage of neurons in the cochlear can fire every sound cycle, encoding sound frequency by their firing frequency. However, at higher frequencies, neurons cannot fire at the sound frequency, but at separated sound pressure peaks. Frequency is then encoded by which neurons along the basilar membrane are firing—a place code.

A more detailed, less schematic illustration of the cochlea and the organ of Corti within it is shown in Figure 13–11. In Figure 13–11A, the oval window is at the extreme left, underneath the text “Scala vestibuli.” The cochlea is coiled such that it traverses about 2.5 turns between the oval window and helicotrema. A cross-section of the cochlea is shown in Figure 13–11B, showing the scala vestibuli, scala tympani, and organ of Corti. The basilar and tectorial membranes are in the organ of Corti, as are the auditory hair cells that convert sound pressure changes in the fluid-filled scala into neural action potentials. Figure 13–11B also shows the location of the spiral ganglion that contains the cell bodies of the neurons whose axons make up the auditory nerve and cochlear branch of cranial nerve VIII. The spiral ganglion cells are bipolar neurons whose dendrites contact auditory hair cells and whose axons project to the cochlear nucleus.

Auditory Hair Cells

A detailed cross-section of the organ of Corti showing the auditory hair cells is shown in Figure 13–11C (Figure 13–11D shows an actual photomicrograph of this structure). The pressure changes at the oval window are transmitted to the scala vestibuli and tympani to produce a shearing action between the basilar and tectorial membranes. This deflects the stereocilia of the hair cells whose movement is transduced into action potentials in the auditory nerve. This is accomplished by specialized channels at the hair cell base that open when the cilia bend. This takes place for hair cell deflections as small as the width of an atom.

At the base of the hair cells are the endings of neurons whose somas are located in the spiral ganglia. Depolarization of the hair cell produces action potentials in the spiral ganglia neurons, which in turn project via cranial nerve VIII to the cochlear nucleus in the brain.

There are 2 major classes of hair cells: inner and outer. The inner hair cells are responsible for the majority of auditory transmission to the brain. These are contacted by myelinated axon-like processes of several spiral ganglion cells. The major function of the more numerous outer hair cells is to control the stiffness of the membrane at their base, enhancing hair cell output at low volume, but decreasing it and protecting the hair cells at higher volumes. Each outer hair cell is contacted by a single, usually unmyelinated spiral ganglion cell ending.

The transduction mechanism of auditory hair cells is shown in Figure 13–12. Potassium (K⁺) channels in the stereocilia are opened by deflection in one direction and closed for the other. The net effect of opening the potassium channels at the tips of the stereocilia is to open nonselective cation channels permeable to sodium, potassium, and calcium in the cell body of the hair cell. This depolarizes the hair cell, causing the release of an excitatory neurotransmitter (probably glutamate) onto receptors on the end process of the spiral ganglion cell.

The Spiral Ganglion Cell

Spiral ganglion cells are bipolar neurons whose somas or cell bodies are located next to the scala tympani in the bony part of the cochlea. The proximal process of the spiral ganglion cell is a conventional myelinated axon that projects to the cochlear nucleus. The distal process, sometimes referred to as a dendrite, has axonal properties such as the ability to conduct action potentials and being myelinated. In this regard, there is at least a superficial resemblance to dorsal root ganglion cells outside the spinal cord, which are pseudo-unipolar cells with 1 “conventional” axonal segment projecting centrally and another projecting peripherally at whose tip mechanotransduction produces action potentials that travel retrogradely toward the soma and beyond it to the spinal cord. The distal end of the spiral ganglion cell at the base of the auditory hair cell has receptors for the neurotransmitter (probably glutamate) released by the auditory hair cell base that has performed the mechanical transduction via its stereocilia.

Spiral ganglion cell distal processes make 2 different types of connections with auditory hair cells, called type I and type II, illustrated in Figure 13–13. Type I hair cells receive an enveloping spiral ganglion afferent nerve terminal (nerve calyx), which itself has a connection to an efferent nerve terminal. Type II hair cells receive an afferent axon-like nerve terminal from a spiral ganglion cell with an adjacent efferent nerve terminal on the hair cell itself. The efferent nerve terminals are from the lateral olivocochlear region of the brainstem. Details of the stereocilia and their mechanical linkage are shown in Figure 13–13B.

Auditory Nerve Representation

There are approximately 30,000 spiral ganglion cells contacting approximately 3500 inner and 12,000 outer hair cells. The auditory nerve is composed of these 30,000 spiral ganglion cell axons whose inputs come mostly from inner hair cells. This nerve also contains axons from the vestibular system. Axons from the spiral ganglia project to the cochlear nucleus, located in the dorsolateral brainstem at the border between the pons and medulla. Individual auditory nerve fibers are sensitive to particular frequency ranges, especially for higher frequencies, reflecting the frequency map of their position in the basilar membrane.

Conduction Versus Neural Hearing Loss

Hearing loss is called conductive when sound is not mechanically transmitted from the eardrum to the cochlea, usually due to damage to the eardrum itself or the middle ear bones. The eardrum can be damaged by diving into deep water or by sharp objects entering the auditory canal. Inner ear infections that literally erode the middle ear bones are a common cause of conductive hearing loss. Infections that reach the middle ear sometimes cause otosclerosis, or remodeling of the middle bones that reduces their ability to transmit sound from the eardrum to the cochlea. German measles (rubella) contracted during pregnancy can be passed to the fetus, in many cases causing the infant to be born deaf.

Neural hearing loss is due to neural damage, usually loss of auditory hair cells in the cochlea. This can occur from exposure to loud noise and commonly occurs in aging at the high-frequency end of the sound spectrum. At higher levels in the auditory system along the projection to auditory cortex, tumors or strokes can damage cells.

Hearing loss also occurs during aging (called presbycusis), much of which is associated with exposure to loud environmental sound. The frequency range for young people ranges from 20 to 20,000 Hz, but the elderly are relatively insensitive to sounds above approximately 15,000 Hz. Continuous exposure to moderately loud sounds or single exposures to very loud sounds can damage hearing. Male hearing loss with aging tends to be worse than that for females because of exposure to high-level workplace sound, such as in manufacturing.

Hearing aids are external audio amplifiers that compensate for the mechanical signal attenuation caused by conductive hearing loss. They can also compensate for low to moderate neural loss of cochlear hair cells. Because hearing loss usually is much higher for some frequencies than others (typically high frequencies), hearing aids must be customized to selectively amplify the attenuated frequencies without overwhelming the user with high-volume amplification of frequencies the person can still hear.

Cochlear Implants

If large-scale hair cell loss is present, external hearing aids are ineffective. The current best solution for this situation is the cochlear implant, which consists of approximately 30 electrodes on a flexible polymer tube threaded into the cochlea. These electrodes are driven by an external microphone and electronics to electrically stimulate the cochlear nerve fibers. Over 100,000 cochlear implants have been implanted worldwide. Many users with these devices have experienced considerable hearing restoration. Damage at higher levels in the auditory pathway cannot be treated with cochlear implants. Current research focuses on using microelectrode arrays implanted in nuclei along the auditory pathway past the side of damage.

Ringing & Tinnitus

Tinnitus is the perception of ringing in the ears in the absence of actual sound. It may arise from exposure to loud noises, ear infections, and allergies. However, in many cases, its cause is unknown. Tinnitus is divided into objective and subjective types. Objective tinnitus is often associated with movement spasms within the cochlea that actually produce sound, which can be detected by diagnostic instruments. Subjective tinnitus can result from known and unknown causes and is difficult to treat. It is often associated with some hearing loss and can occur and improve on its own over a period of months or years. One treatment involves the use of noise-producing earphones that mask the ringing sound. Extreme cases of tinnitus can be debilitating and have been associated with suicide. In some of these cases, surgical destruction of the auditory nerve has been done to produce relief, but this also causes loss of normal hearing.

AUDITORY PROJECTIONS TO THE BRAINSTEM

The Cochlear Nucleus

The spiral ganglion axons and those from the semicircular canals of the vestibular system project via cranial nerve VIII (vestibulocochlear nerve) to the ipsilateral cochlear nucleus (Figure 13–14). This nerve passes through the skull via the internal acoustic (auditory) meatus in the temporal bone. The cochlear nucleus has distinct dorsal and ventral nuclei, shown in Figure 13–15. The ventrolateral dorsal cochlear nucleus and anteroventral ventral cochlear nucleus receive axons from lower frequency sensitive auditory fibers that synapse on auditory hair cells distant from the oval window, whereas the dorsomedial portion of the dorsal cochlear nucleus and dorsal ventral cochlear nucleus receive inputs from high-frequency fibers originating near the oval window. This tonotopic organization of the cochlear nucleus is maintained in higher projection areas all the way to the primary auditory cortex in the superior temporal lobe.

The cochlear nucleus projects both ipsilateraly and contralaterally rostrally to the superior olivary nuclei on both sides (Figure 13–14). The contralaterally projecting fibers cross by passing through a structure called the trapezoid body (shown in Figures 13–14 and 13–15). The ascending fibers on both sides then project via the lateral lemniscus on each side to the inferior colliculus, a major auditory processing center. A cochlear projection called the dorsal acoustic stria ascends to nuclei of the lateral lemniscus.

Cochlear axons convey frequency and amplitude information explicitly and sound localization implicitly in their firing, because there is no explicit representation of sound location in the cochlea. Auditory neurons tend to fire at the peak of sound pressure waves. Sound amplitude is encoded by the number of spikes that are generated at each peak and the number of fibers that fire (fibers sensitive to particular frequencies have somewhat different thresholds).

The inferior colliculi and all later projection areas receive bilateral auditory input from both ears, which is important for sound localization. The inferior colliculus on each side projects ipsilaterally to the medial geniculate in the thalamus, which in turn projects to auditory cortex in the middle of the superior temporal lobe (superior temporal gyrus).

The Superior Olive, Lateral Lemniscus, & Directional Localization

There is no explicit representation of sound location in the cochlea, but rather a tonotopic or frequency representation along the basilar membrane. The superior olive is an important structure for sound localization in azimuth (horizontally) because it receives inputs from both cochlear nuclei that can be used to compute azimuth via interaural time difference and interaural intensity difference. Time difference computations are thought to be computed in the medial superior olive, whereas intensity difference is registered in the lateral superior olive.

The method by which the brain accomplishes interaural time difference discrimination for azimuth sound localization is shown schematically in Figure 13–16. Neural coincidence detectors are arrayed along axonal delay lines carrying spikes from the left and right ears. For a sound source directly ahead, spikes will reach only the middle coincidence detector at the same time. Sounds located to the left or right of this middle position will arrive synchronously at different coincidence detectors.

The time sensitivity of this system is extraordinary. A “back of the envelope calculation” using the speed of sound as 1000 ft/s and the distance between the ears as 6 in (0.5 ft) yields a maximum delay for a sound source directly to the left or right of 0.5 milliseconds, or 500 microseconds. Because the neural action potentials last about 1 millisecond, this means that the interaural time difference neural system can discriminate arrival time differences much less than the width of 1 action potential. The basis for this capability is the phase locking of auditory nerve firing in each ear to the exact peak of the sound wave. Interaural time difference works best for low frequencies where phase differences yield the largest absolute time differences and where shadowing by the head does not reduce the amplitude as much in 1 ear compared to the other. The neural structure and mechanism for interaural time difference sound localization have been demonstrated in the barn owl, a bird that can locate and strike prey in complete darkness using this type of mechanism. In humans, similar circuits exist in the superior olive and may be the reason why the cochlear nucleus projects both ipsilaterally and contralaterally close to the initial sound transduction.

The other mechanism for azimuth sound localization is interaural intensity difference. Sound amplitude will be greater on the side of the head closer to the sound source because of shadowing by the skull. This effect is greater for higher than lower frequencies. Different neuronal types in the lateral superior olive are sensitive to the sum of excitation (EE) from the 2 ears or the difference between the firing from each ear (EI). Dividing the EI output by the EE output (neurally) will give a normalized representation of the distance from the middle position.

Localizing elevation is more complicated because the 2 ears are in the horizontal plane and no difference in arrival time or intensity occurs at the ears because of pure elevation changes. However, one of the functions of the complex shape of the pinna or external ear (see Figure 13–6) is that it alters the frequency spectrum of complex sounds reflected into the auditory canal as a function of elevation. Specifically, the middle-frequency spectrum of complex sounds reflected by the pinna into the auditory canal is attenuated for sources at high elevations compared to middle or low elevations. During development, the nervous system “learns” the change in transfer function for familiar complex sounds generated by the pinna, which permits localization in elevation. However, the precision of this elevation localization is poorer than for azimuth.

The Inferior Colliculus

Neurons in the medial superior olive and lateral superior olive project to the inferior colliculus in the midbrain. The inferior colliculus consists of a central nucleus surrounded by a dorsal cortex and a lateral external cortex. Elevation- and azimuth-detecting cells in the superior olive converge on colliculus neurons that thus represent location in both horizontal and vertical angles. The inferior colliculus receives reciprocal innervation from the medial geniculate nucleus of the thalamus to which it projects, as well as the auditory cortex. It also receives input from the somatosensory system. The colliculus is thus an integration center for spatial location and reflexive orientation for auditory inputs in the startle reflex.

The Medial Geniculate Nucleus of the Thalamus

The inferior colliculus projects to the medial geniculate nucleus of the thalamus above it in the midbrain (see Figure 13–15). As for vision, somatosensation, and taste, auditory information is relayed through a sensory modality–specific thalamic nucleus that, in turn, projects to a specific “primary” cortical area for that sense. The medial geniculate nucleus consists of 3 main subdivisions: ventral, dorsal, and medial. The ventral nucleus of the medial geniculate contains a tonotopic map with low frequencies being lateral and high frequencies being medial. Neurons in the ventral nucleus represent amplitude and binaural localization information as well as frequency. The dorsal nucleus contains cells that respond to complex environmental stimuli as well as somatosensory input. Neurons in the medial nucleus receive input from intensity and duration coding cells that appear to be part of the azimuth localization pathway.

High Order Auditory Processing

Neurons in the medial geniculate nucleus project to primary auditory cortex in the mid-posterior superior sulcus of the temporal lobe. The primary auditory cortex includes the Heschl gyrus and extends into the lateral fissure of the temporal lobe, including the planum temporale and planum polare. In the Brodmann map of cortex, the auditory cortex comprises areas 41 and 42 and part of area 22. Neurons projecting to the primary auditory cortex explicitly encode frequency, intensity, and some aspects of sound location. The auditory cortex refines discrimination in these dimensions and enables more complex sound identification.

Primary Auditory Cortex

The predominant organization of the primary auditory cortex is a tonotopic map. This map represents the preservation of the initial auditory transduction by the place code of spiral ganglion cell axons along the basilar membrane from oval window to helicotrema. In the auditory cortex, low frequencies are represented anteriorly and high frequencies posteriorly. However, there are multiple maps in the auditory cortex besides frequency, including binaurality (related to localization), intensity, and latency.

Auditory Association Cortex

Neurons in the primary auditory cortex project to other cortical areas for higher order sound processing. Figure 13–17 shows higher order or “association” areas of the auditory cortex and also its projection to the Broca area (44 and part of 45) and other frontal lobe areas. Language and complex sounds are processed in the Wernicke area in the vicinity of Brodmann area 39.

Understanding Language

Language sits at the top of the hierarchy of sound processing and is unique to humans. Other animals can recognize and respond to complex sounds such as keys jangling or doors creaking, but only humans use a complex grammar for communication. Language reception and recognition involve all the early stages of sound processing from the cochlear to auditory primary and association cortical areas, but then extend into the Wernicke area at the junction of the temporal and parietal areas (Figure 13–18). This area projects to the frontal lobe via a tract called the arcuate fasciculus to the Broca area, which is just anterior to the primary motor areas for the larynx, pharynx, and other vocalization apparatus of the mouth, tongue, and lips.

Language processing is also unique in its extreme lateralization. Virtually all right-handers and a majority of left-handers process language primarily in the left temporal lobe and show major deficits from brain damage only on the left side. Language output, or speech, is also controlled by the Broca area on the left side.

THE VESTIBULAR SYSTEM

The vestibular system is responsible for informing us about our orientation in space and acceleration through space and for maintaining balance. The initial sensory transduction in the vestibular system originates close to the cochlea and shares many similarities with it. The transduction mechanisms of the vestibular system for rotation are composed of the semicircular canals (see Figure 13–9). Linear accelerations are transduced by otoliths. The output of the vestibular system is via cranial nerve VIII (see Figure 13–14).

The Vestibular Labyrinth

The cochlea, semicircular canals, and otoliths compose what is called the vestibulum of the labyrinth of the inner ear. Some details of these structures are illustrated in Figure 13–19.

Otolith Organs: Utricle & Saccule

The otolith consists of 2 suborgans called the utricle and saccule, which are oriented at 90 degrees to each other, illustrated in Figure 13–19. The receptors in both of these organs are called maculae, which consist of cilia on whose tips are calcium crystals called otoconia. The individual cilia within each otolith are oriented in multiple directions so that there will be an output from either organ or both organs for acceleration in any direction for any head position. The otoliths sense linear forces.

The Semicircular Canals

There are 3 semicircular canals oriented at right angles to each other. These structures, via their cilia, sense head tilt and rotary accelerations. The output of the semicircular canals is combined in the brainstem with visual information about movement to yield information about movement and rotation in space. This information is used to program compensatory eye movements, such as the vestibulo-ocular reflex, and to control postural compensation for balance.

Vestibular Hair Cells

The transduction mechanisms of the vestibular system are based on hair cells similar to those in the cochlea for both the otolith and semicircular canals.

Vestibular Pathways to Thalamus & Cortex

The output of the vestibular system is via cranial nerve VIII (see Figure 13–14) through the internal auditory meatus with the cochlear nerve, which enters the brainstem at the border between the medulla and pons. The axons project to 4 vestibular nuclei and the cerebellum (via the inferior cerebellar peduncle). The outputs of the vestibular nuclei project via the medial longitudinal fasciculus to several oculomotor nuclei to control eye movements, as well as to the cerebellum.

Balance Problems

Vestibular system dysfunction can lead to balance problems. Infections of the inner ear can cause labyrinthitis, leading to dizziness and balance difficulties. Ménière disease, which is associated with dysregulation of fluid volume in the vestibular system, exhibits symptoms such as dizziness and vertigo. The vestibular nerve itself can become inflamed from viral infections, leading to vestibular neuronities and vertigo. Less severe vestibular problems include benign paroxysmal positional vertigo, which occurs transiently with any rapid change in head position.

SUMMARY

• Sound consists of air vibrations that have the properties of frequency and amplitude.

• Sound is mechanically modified and transduced in the outer and middle ear.

• Sound frequency and amplitude are converted into neural codes in the cochlea.

• The cochlea is tonotopically organized with auditory hair cells transducing sound to neural firing.

• Brainstem nuclei process auditory information and begin coding for location.

• The auditory thalamus relays auditory information to the primary auditory cortex in the temporal lobe.

• Primary and higher order auditory areas in the temporal and parietal lobes further process auditory information.

• The vestibular system originates with the transduction of self-motion by 3 semicircular canals used for balance and spatial orientation.

OBJECTIVES

OVERVIEW

Smell and taste are called chemical senses because molecules activate the receptors in these systems rather than energy, such as photons in vision and vibrations in hearing. Smell and taste have important gateway functions in approach/withdrawal behaviors. The smell given off by a ripe pineapple induces us to consider eating it, and the ripe, sweet taste prompts consumption of the fruit. However, we are repulsed by the smells and tastes of many bitter and sour substances. Those chemical qualities are often (but not always) associated with spoilage or toxicity. Taste receptors, found mostly on the tongue, exist for sweet, sour, salt, bitter, and, according to some researchers, umami, the savory taste induced by the additive monosodium glutamate (MSG). Taste is also strongly influenced by smell because mastication of food also activates olfactory receptors in the nose. The combination of smell and taste gives rise to the complex perception of flavor. The message encoded by receptors for taste relays through several subcortical structures before reaching the thalamus and then the cortex. Olfaction is the exception among the senses in that neurons in the olfactory bulb project directly to cortex in a pathway whose activation is largely subconscious. However, there is an olfactory projection from cortex back to the thalamus, and then back to cortex. This projection reaches, among other loci, the orbitofrontal cortex, where smell and taste inputs are combined to mediate complex olfactory perceptions such as flavor.

ORGANIZATION OF THE OLFACTORY SYSTEM

The nose has a number of functions besides smell, such as filtering, warming, and humidifying the air you breathe. However, here we concentrate on its function of detecting odorants and reporting their characteristics to the brain. Olfactory receptors exist in the roof of the nose. Figure 14–1 shows the overall internal structure of the nose.

The Nose & Olfactory Epithelium

Internally the nose is divided in the middle by a cartilaginous structure called the septum. On each side of the septum are 3 cavities called “turbinates,” shown in Figure 14–1. Within the superior turbinate at the top rear of the nose’s internal chamber is the olfactory mucosa located on its lateral wall. This is where the olfactory cilia sit that transduce odorants into neural signals. These cilia are extensions of the olfactory receptor cells, which are located below the lamina propria, which is below a bone called the cribriform plate, as shown in Figure 14–2.

One basis for the difference in odor sensitivity between animals such as humans and dogs is the vastly different number of olfactory receptors in various species. There are approximately 5 to 10 million olfactory receptor neurons in humans versus approximately a billion in dogs.

Olfactory Receptors

Olfactory cells extend approximately 5 to 10 processes called cilia into the mucous layer where odor molecules entering the nose are entrapped. Olfactory receptors reside on these cilia. In this regard, they resemble receptors for other senses such as sight and hearing. The photoreceptor outer segments that contain the photon-absorbing pigment also are derived from cilia. Auditory hair cells in the organ of Corti are also modified cilia. Olfactory receptors are like G-protein–coupled metabotropic receptors for neurotransmitters, except that they are activated by a molecule from the external world rather than a neurotransmitter released from another neuron. Although there are approximately 1000 distinct olfactory receptor types (and corresponding genes that code for them), a single olfactory receptor cell expresses only 1 of these receptor types. With 5 to 10 million olfactory receptor cells, this means that there are approximately 5000 to 10,000 different receptor cells of each receptor type.

The Transduction of Olfactory Signals

Figure 14-3 shows how the process of smell works. An odorant molecule that has crossed the mucosa binds a receptor recognition site on the olfactory receptor, which is located on one of the olfactory receptor neuron’s cilia. The receptor, like other G-protein–coupled receptors, is a 7-membrane domain spanning molecule. On the external side of the membrane is the recognition site for a specific odorant, whereas on the internal side, the molecule is associated with a G-protein complex. This G-protein complex is in an inactive state when no odorant is bound to the external portion of the receptor. Upon binding an odorant, however, the receptor structure undergoes a conformational change that activates the G-protein complex by causing the subunits to dissociate. The free alpha G-protein subunit binds to and activates adenylyl cyclase, catalyzing the conversion of adenosine triphosphate to cyclic adenosine monophosphate (cAMP). Finally, the cAMP binds to a site on the internal side of Na⁺/Ca²⁺ cation channels that open.

Olfactory receptors constantly turn over, living only a few months. This is the longest known case of adult neurogenesis. In the past, some researchers have attempted to repopulate other brain areas that have experienced cell death with olfactory receptor cells hoping they would multiply and integrate into damaged neural circuits in a manner similar to the use of neural stem cells more recently. Within the past few decades, adult neurogenesis has also been shown in the hippocampus.

The Olfactory Bulb & Olfactory Coding

The odorant-induced cation flux in the olfactory cilia depolarizes the olfactory receptor, generating action potentials that travel down the receptor axons through the cribriform plate to the olfactory bulb. This axon tract is cranial nerve I. The axon terminals of these cells contact the dendrites of mitral and tufted cells, releasing of the excitatory neurotransmitter glutamate. The olfactory receptor axon terminals cluster in several thousand distinct structures called glomeruli, each of which receives input primarily from olfactory receptors of a given odorant receptor type. Glomeruli having similar receptor type inputs tend to be close together, so their organization forms a multidimensional spatial map of odor properties. Dimensions of this map are thought to include properties such as molecular carbon chain length and functional group similarity. Figure 14–4 shows the basic projection scheme from olfactory receptors to the olfactory bulb. In addition to the direct feedforward connections from olfactory receptor cells to mitral and tufted cells, there are also lateral inhibition neurons such as granule cells that mediate mutual inhibition between mitral and tufted cells by releasing γ-aminobutyric acid (GABA). Glomeruli inhibit each other via interneurons called periglomerular cells.

Mitral and granule cells contact each other via dendrodendritic synapses, with the mitral cell dendrites releasing glutamate to the granule cells, and the granule cells releasing GABA onto the dendrites of mitral cells. The net result of this is a feedback inhibition for the mitral cells mediated via their connections to granule cells, as well as mutual inhibition between mitral cells.

The Vomeronasal Organ & Pheromones

Many mammals, including humans, have another distinct olfactory organ toward the front of the septum within the nose called the vomeronasal organ. Olfactory receptors in this organ are involved in the detection of intraspecies communication odors called pheromones. Although pheromones are best known in insects for functions such as trail marking, they also exist in mammals as mediators of social and sexual communication and behavioral triggering. Their detection is usually unconscious, mediating behavioral changes such as menstrual synchrony in women. Vomeronasal receptors project to a structure distinct from the olfactory bulb called the accessory olfactory bulb, which in turn projects to the amygdala and hypothalamus.

CENTRAL PROJECTIONS OF THE OLFACTORY BULB

Direct Projections

The olfactory system is the exception among the major senses in projecting directly to the cortex without a relay in the thalamus (although there is a thalamic circuit in the olfactory system). Specifically, the olfactory bulb projects directly to the pyriform cortex in the frontal lobe and the entorhinal cortex outside the hippocampus (Figure 14–5). Both of these cortical areas are ancient mesocortex, rather than neocortex. Details of the olfactory bulb projections can be seen in Figure 14–6, which shows that the olfactory bulbs on each side of the nose project to each other via the anterior commissure (just anterior to the corpus callosum) and the important direct projections to the amygdala and anterior olfactory nucleus.

(Figure 14–7) shows the overall projection scheme for the olfactory bulb. Of particular note is that several cortical and subcortical structures project olfactory information to the thalamus, which in turn projects to the orbitofrontal cortex. Activity in the orbitofrontal cortex underlies our conscious perception of odor. Activation of the hippocampus is important in forming episodic memories associated with odors, that is, memories associated with a particular event and its context. In addition, neurons in the orbitofrontal cortex that also receive input from the tongue mediate the combination of smell and taste often referred to as flavor.

Olfactory projections to the hippocampus via the entorhinal cortex mediate learning and memory functions associated with smell, whereas the projection to the amygdala adds emotional content to this learning and memory via the amygdala’s projections to both the orbitofrontal cortex and the hippocampus. The amygdala is important in the learning process that associates positive valence with odors that have had positive consequences and negative valence with those that predict negative outcomes. It is particularly important in fear conditioning, for example, as might be the case for the smell of gunpowder for traumatized soldiers.

Projections to the Orbitofrontal Cortex & Olfactory Contributions to Taste

Taste is heavily dependent on smell. When food is masticated, molecules of the injected substance reach the nasal pharynx via what is called the retronasal route from the mouth, where they activate olfactory receptors in a manner similar to odorants. If the nose is pinched shut during chewing, many otherwise easily distinguishable foods are indistinguishable. A major brain locus for the combination of taste and smell is the orbitofrontal cortex, where multimodal neurons occur that have inputs from both the taste and olfactory systems. The sensory addition of olfaction to taste is often referred to as flavor.

The firing of neurons in the orbitofrontal cortex is also influenced by hunger and satiety. As food is consumed, hunger decreases. This phenomenon is called alliesthesia, and it is based on central nervous system mechanisms that receive inputs from receptors in the stomach. A faster acting satiety mechanism is called sensory-specific satiety, which is specific to the food consumed and depends on receptors on the tongue and in the mouth.

Clinical Aspects of Smell

Smell loss can be caused by peripheral or central mechanisms. Chronic rhinosinusitis or other insults to the nasal cavity can kill olfactory receptor cells. Because the olfactory cortex is in the frontal lobe, frontal lobe dysfunctions can affect olfactory processing. Neurodegenerative diseases that impact the sense of smell include schizophrenia and Parkinson, Huntington, and Alzheimer disease. People with Down syndrome also have smell impairment.

TASTE

Humans are probably the most omnivorous vertebrates on the planet, eating almost anything that can be found that has nutritional potential. Our taste receptors, combined with the act of smelling what we masticate, give us sophisticated sensitivity to quickly discriminate nutritious from toxic substances we have ingested.

Classically, taste has been thought to be based on a set of 4 receptor types, namely, sweet, salt, bitter, and sour. However, there is general agreement now for a fifth basic taste, usually referred to as umami (from the Japanese meaning “pleasant and savory taste”), typified by the taste of MSG. The existence of umami as a basic taste is based primarily on the existence of a receptor that responds very specifically to this class of taste substances, such as the amino acid L-glutamate.

Taste receptors are found primarily on the tongue, although a few reside on the roof of the mouth in the pharynx. Taste receptors reside in bump-like structures called papillae. Figures 14–8 and 14–9 show that there are 4 major types of papillae on the tongue: filiform, fungiform, foliate, and circumvallate (vallate). The filiform papillae, which are found across the dorsal tongue surface, particularly in the tongue’s central region, do not contain taste buds, but the other types of papillae do. In some species, filiform papillae make the tongue rough and function mechanically to rasp ingested food.

Humans have several thousand papillae that contain taste cells, that is, taste buds. Each taste bud has >50 cells that include both taste receptor cells and supporting epithelial cells (Figure 14–10). Fungiform papillae tend to be found toward the front of the tongue, foliate on the side toward the rear, and circumvallate at the very rear across the dorsal surface. There are only approximately 10 circumvallate cells at the back of the tongue. Taste substances contact the taste receptor neurons through the taste pore. Within most taste buds are receptor cells for several tastes. There are receptors for all 5 basic tastes throughout the tongue, and these receptors occur in each of the 3 taste bud morphologies (fungiform, foliate, and circumvallate), although the front of the tongue is slightly more sensitive to sweet tastes and the back to bitter. Sensory signals exit the tongue via the chorda tympani and glossopharyngeal axons projecting from the taste cells (see Figure 14–9).

The receptor mechanisms underlying the 5 basic tastes of sweet, salt, bitter, sour, and umami are shown in Figure 14–11. Salt and sour are associated with specific ion channels. Salt (NaCl) is detected by the flux of external Na⁺ through sodium channels. The sour taste is primarily due to acidity, or the presence of H⁺ ions. The detection of acidity is complex and not well understood but may ultimately involve K⁺ flux.

Sweet, bitter, and umami are associated with second messenger receptor systems (see Figure 14–11). The sweet taste, exemplified by glucose, also operates via an MGluR4-like metabotropic receptor via a G-protein second messenger system that causes the release of intracellular calcium from internal stores. Bitter substances, exemplified by quinine, are detected by TAS2R receptors containing a protein called gustducin the activation of which lowers intracellular cyclic nucleotide concentrations and may also activate the IP3-DAG second messenger cascade. These second messenger systems are believed to cause the release of internal calcium from stores and open ion channels. The umami taste is mediated by a receptor that activates an Na⁺/Ca²⁺ cation channel via a G-protein second messenger system.

Taste receptor cells synapse with axon-like processes (see Figure 14–10) in either the chorda tympani or glossopharyngeal nerves that extend from sensory ganglion called the geniculate and petrosal ganglia, respectively (Figures 14–12 and 14–13). There are also a few taste receptors in the pharynx that project to the brain via processes of cells in the nodose ganglion. This arrangement bears some similarity to the connection scheme of sensory dorsal root ganglion cells and auditory hair cells. Taste messages from most of the front of the tongue (mainly fungiform papillae) project to the brain via the chorda tympani nerve, whereas the glossopharyngeal nerve carries messages from foliate, circumvallate, and fungiform papillae. The chorda tympani and glossopharyngeal nerves join other cranial nerves en route to the brain (see Figure 14–12). A few taste receptors in the larynx and mouth project to the medulla via the vagus nerve.

The target of these projections is the solitary nucleus of the medulla (nucleus of the tractus solitarius), which projects to the ventral posterior nucleus (also called the ventral posteromedial [VPM] nucleus) of the thalamus. The gustatory area of the thalamus projects to the anterior insula—frontal operculum area of cortex that further processes taste signals (see Figure 14–13). The location of the gustatory area of neocortex in the area of the insula is illustrated in the coronal section diagram of Figure 14–14.

Axonal endings of fibers in the chorda tympani nerve receive inputs from a number of taste cells in a variably selective manner. Some fibers are activated nearly exclusively by a single receptor type (labeled line coding), such as salt, but others have mixed receptor input (distributed coding). One of the functions of cortical processing is to identify specific tastes from the activation occurring across multiple fibers. However, before reaching the cortex, taste messages are modulated in the nucleus of the solitary tract (NST) in the medulla and in the thalamus. Neurons in the NST receive satiety inputs that modulate their activity. Eating sweet substances will reduce the responses of sweet-responding NST neurons after some time, reducing their appeal.

Neurons in the NST project to the taste portion of the thalamus, the VPM. Damage to the VPM can cause ageusia, the loss of the sense of taste. Neurons in the VPM project to the insula and frontal operculum cortex (see Figures 14–13 and 14–14), the primary cortical taste areas. The insula lies at the junction of the parietal, frontal, and temporal lobes with the opercular cortex (operculum means “lid”) just above it. These 2 evolutionarily old primary gustatory areas are affected by olfactory and visual inputs.

The primary insula and opercular cortical areas project to the amygdala and orbitofrontal cortex. Taste information is combined with olfactory input in the orbitofrontal cortex to represent what is called flavor, the combination of taste and smell. The projection to the amygdala mediates memory for emotionally salient tastes. The amygdala sends a projection to the orbitofrontal cortex, perhaps functioning in its learning function with the orbitofrontal cortex like the hippocampus does with other parts of the frontal lobe. This learning function mediates the association between the sight, smell, and taste of very specific foods and the experience after ingesting it, forming the basis of learned food preferences and various sensory-mediated triggers for hunger. The pleasure derived from eating goes beyond smelling and tasting to include food texture, which is transmitted to the brain by activation of mechanoreceptors while chewing. Despite the fact that some people congenitally lack the senses of taste and smell, they still enjoy eating because of this mechanical component. Texture manipulation is an important aspect of food preparation.

The total lack of taste is called ageusia. More common than ageusia is a poor sense of taste, called hypogeusia, which may be due to damage to the olfactory mucosa that causes partial anosmia. Some cancer chemotherapies temporarily cause hypogeusia. Anosmia, the lack of sense of smell, which can affect taste or flavor, can be congenital or result from brain damage or damage to the olfactory mucosa such as from exposure to noxious substances. Sensitivity to bitter tastes declines with age. Schizophrenia is often associated with deficiencies in smell, which are thought to be linked to generalized frontal lobe neuropathy.

Two main satiety mechanisms modulate the sense of taste, one in the central nervous system and central brain mechanism and the other in the taste receptors. The central mechanism, called alliesthesia (“changed taste”), is a brain mechanism that indicates you are getting full. The mechanism is mediated by reduced firing of orbitofrontal taste and smell neurons to the specific odor of an ingested substance after a substantial consumption, causing a loss of general appetite. The peripheral mechanism is called sensory-specific satiety. Its mechanism involves reduced output in taste and olfactory receptors themselves. This mechanism suppresses appetite specifically for the taste of what is being consumed and acts on a faster time scale than the alliesthesia “fullness” mechanism.

Alliesthesia is mediated by the action of several gastrointestinal hormones, including insulin, ghrelin, and pancreatic polypeptide. The digestion process causes the release of these hormones into the bloodstream, where they act on areas of the hypothalamus and brainstem. These hormones also modulate the transmission of taste information by the vagus nerve to the NST in the medulla. Interactions between the orbitofrontal cortex and the amygdala mediate taste learning.

SUMMARY

• Molecular receptors for smell are found in the nose in the olfactory epithelium.

• There are approximately 1000 distinct types of olfactory receptors.

• Olfactory signals are transduced by several molecular mechanisms.

• Olfactory axons project to the olfactory bulb into structures called glomeruli.

• The olfactory bulb projects directly to cortex and indirectly to other areas of neocortex.

• Taste receptors are found mostly on the front, sides, and back of the tongue in papillae.

• Axons of taste receptors project to the NST, which in turn projects to thalamus.

• Taste and smell inputs are combined in orbitofrontal cortex to produce the complex perception of flavor.

OBJECTIVES

OVERVIEW

Embedded in the skin are receptors for touch, temperature, and pain. These receptors are at the axon endings of neurons whose soma are in the dorsal root ganglia outside the spinal cord, or other ganglia outside the brain for cranial nerves mediating somatosensation in the head. Action potentials produced by somatosensory receptors travel retrogradely past the axonal bifurcation in the dorsal root ganglion to synapses in the spinal gray area. Somatosensory synapses in the spinal gray area participate in local circuits that mediate reflexes and project to the thalamus and other brain areas.

CUTANEOUS & SUBCUTANEOUS SOMATIC SENSORY RECEPTORS

The skin is the boundary between the body and the world, enclosing the body’s tissue, retaining water, excluding bacteria and dirt, and providing thermal insulation. The skin detects the world through the sense of touch, which is called cutaneous or somatosensory perception. The kinds of touch that we can perceive include various kinds of mechanical sensations (pressure, movement, and flutter), as well as temperature and pain. These different kinds of perceptions arise from the activation of different kinds of receptors in the skin.

Detecting something as a skin sensation without or before perceiving what has made this contact is called passive touch. Touch has another, more active function, called haptic perception, which enables us to perform complicated manipulations of objects whose shape and orientation are perceived through touch. This type of perception (active touch) is particularly important for tool use and dexterity skills involving the fingers and fingertips. It has been shown, for example, that people blind from birth who have no visual input to the occipital visual cortex experience activation in the visual cortex during their finger manipulations when reading Braille. Both active and passive touch depend on the activation of a variety of cutaneous receptor types.

The skin has 2 main layers that allow it to perform its functions, the epidermis and dermis, shown in Figure 15–1. The epidermis is the outermost layer of the skin (epi means “above” or “on”; dermis means “skin”). The epidermis consists of layers of dead cell ghosts that provide an insulating barrier to the outside. There are virtually no tactile receptors in the epidermis and none in the superficial epidermis, so moderate abrasions of the skin surface do not kill any living cells and are not felt as painful. The epidermis is formed by the division of cells in the dermis below it that are continually dividing and migrating outward to replace the dead layers as they wear off. As these cells reach the epidermis, they flatten, die, and form the inert epidermis barrier. The dermis is the living layer of skin below the epidermis that includes virtually all the somatosensory receptors. Hairs from hair follicles in the dermis pass through the epidermis before appearing on the skin surface. Below the dermis is the subcutaneous layer that contains vasculature and fat cells.

For all the skin below the neck, somatosensory receptors are specializations of the axons of sensory neurons whose cell bodies are in the spinal cord dorsal root ganglia. The other end of the axons of these cells enters the spinal cord at the dorsal root and makes synapses with local and projection neurons. Cutaneous information is relayed by spinal cord projection neurons to the ventral posterior nucleus of the thalamus, and then to a strip in the parietal lobe where a “touch” map of the body exists. Cutaneous sensation in the face and neck is mediated by cranial nerves in functionally similar pathways.

Mechanoreceptors for Touch

The skin has receptors for several kinds of touch, warm and cold temperature, and several types of pain. Touch receptors are called mechanoreceptors. Four types of mechanoreceptors with distinct transducer structures are specialized to receive tactile information: Pacinian corpuscles, Ruffini endings, Merkel disks, and Meisner corpuscles (Figure 15–2). In addition to these receptor morphologies, different types of so-called “free nerve endings” respond to pain stimuli and temperature. Mechanical displacement of the skin or skin hairs is also detected by free nerve endings in the hair follicle at the hair root.

Mechanoreceptors signal various kinds of touch or mechanical skin displacement. There are 4 main types of mechanoreceptor morphology: Meissner corpuscles, Merkel disks, Ruffini endings, and Pacinian corpuscles (see Figure 15–2). The response properties of these different receptor types fall along 2 major axes, receptive field (RF) size and response transience, illustrated in Figure 15–3. The RF size dimension is related to depth in the skin. The effect of a punctate displacement on the surface of the skin extends farther out laterally for deep versus shallow skin locations. Thus, deep mechanoreceptors (Ruffini endings and Pacinian corpuscles) tend to have larger RFs than shallow ones (Meissner corpuscles and Merkel disks).

The second mechanoreceptor response dimension is how sustained their responses are to a continuous stimulus. The Merkel disk is a shallow, small RF, sustained-responding fiber, whereas the Ruffini ending is a deep, large RF, sustained unit. Meissner corpuscles are shallow, small RF, transiently responding receptors, whereas Pacinian corpuscles have large RF, transient responses.

There is 1 additional response dimension beyond RF size and sustained/transient responsiveness: frequency response. There is a considerable difference among the fibers in the frequency of stimulation to which they respond, which translates into the resulting cutaneous perception. Merkel disks, the small RF sustained units, respond only to very low frequencies, so they tend to report constant force on the skin. The small RF transient Meissner corpuscles respond over a range of 3 to 40 Hz, and their activation is felt as “flutter.” The large RF sustained Ruffini endings respond to stimulus frequencies from 15 to 400 Hz, with their firing interpreted as stretch. Pacinian corpuscles have large RFs and transient responses over a frequency range of 10 to 500 Hz and signal vibration. These frequency ranges overlap, so that at most stimulus frequencies >1 fiber class is active, and the perception of the stimulus is based on the firing of several cutaneous receptor types.

Mechanotransduction

The receptor structures of mechanoreceptors are composed of mechanically gated ion channels in the axonal membranes of dorsal root ganglion cells. The unusual “pseudo-unipolar” morphology of these cells is shown schematically in Figure 15–4. These neurons are composed of a cell body located in the dorsal root ganglion just outside the spinal cord. These cells have no dendrites (typically) and only a single axon. This axon bifurcates close to the cell body in the ganglion and gives rise to 2 processes: 1 projecting out to the periphery in the skin, forming the receptor, and the second axon process extending into the dorsal spinal cord gray area.

The mechanoreceptor at the axonal ending typically consists of a number of mechanically gated channels and, in some types, an enclosing corpuscle that modulates the properties of the transduction. Figure 15–5 shows a highly schematic drawing of a mechanically gated ion channel. Stretch or deflection of the neural membrane in which the channel is embedded causes that channel to open and allow the flux of ions, usually sodium, thus depolarizing the axon and generating action potentials. The corpuscle enclosure in mechanoreceptors such as Pacinian corpuscles mechanically modulates the mechanical input to the axonal ending. The ending by itself responds in a sustained manner, but the corpuscle flexes under pressure so that the axonal ending within experiences a deflection at the onset and offset of applied force, thus becoming an on-off responding receptor despite being based on a sustained on-responding mechanoreceptor within.

The mechanically induced action potentials initiated in the dorsal root ganglion peripheral axon travel in what would be classically considered a retrograde manner toward the cell body and the axonal bifurcation point in the dorsal root ganglion. These action potentials simply continue past the axon bifurcation point now in an orthograde direction into the dorsal root of the spinal cord (see Figure 15–4) to synapse on neurons in the spinal gray area. The second-order neurons can be either spinal interneurons or projection neurons that relay somatosensory information to the thalamus.

Receptive Fields, Sensory Maps, & Dermatomes

The RF of a cutaneous fiber is the area of skin that, if mechanically stimulated, will influence the firing of that fiber. The RF concept can be extended to a particular spinal cord segment dorsal root ganglion as the skin area that influences firing of any of the cells in that ganglia. This is of clinical significance because injury to a specific spinal cord dorsal root will be associated with sensory dysfunction over a particular skin area, called a dermatome. Skin dermatomes associated with particular spinal cord segments (cervical, thoracic, lumbar, sacral) are illustrated in Figure 15–6.

Differences in Mechanosensory Discrimination Across the Body Surface

The ability to resolve small differences in the location of a mechanical stimulus to the skin varies considerably across the body, being very high in locations such as the fingertips and face and low on the body trunk. This resolution ability is related to receptor density and inversely to receptor RF size. The minimum distance needed to distinguish 1 from 2 stimulus points is as high as 2 mm on the fingertips and face, but may exceed 40 mm on the trunk, as shown in Figure 15–7. Mechanoreceptors on high-acuity skin areas such as the fingertips and skin have small RFs and are located very close to each other compared to low somatosensory acuity areas. Because a given area of somatosensory cortex processes information from about the same number of peripheral receptors, the cortical map will be enlarged for high-acuity versus low-acuity areas, as will be seen later in this chapter.

Trigeminal Ganglion & Nerve

The general plan of spinal nerves, as shown in Figure 15–8, is for sensory inputs to enter the dorsal horn and motor outputs to exit the ventral root. These nerves merge into a single bundle peripheral to the dorsal root ganglia (where the cell bodies of the sensory nerves reside) to form what is called a “mixed” sensory and motor nerve. Above the spinal cord, some cranial nerves follow a similar plan, although rather than emanating from the spinal cord, they arise in brainstem areas such as the medulla. The processing of cutaneous perception from the face is mediated largely by the trigeminal nerve, shown in Figure 15–9. The cell bodies of the sensory axons are located in a ganglion outside the brainstem called the trigeminal ganglion. The sensory messages from facial stimulation are conveyed to what is called the main (principle) nucleus in the pons (rather than the spinal gray area for spinal nerves), which in turn relays to the thalamus via the dorsal and ventral trigeminothalamic tracts.

Active Tactile Exploration & Joint Receptors

There is another type of mechanoreceptor that is similar to cutaneous mechanoreceptors but mediates the senses of proprioception (limb position) and kinesthesis (limb motion). Many of these receptors are similar to some of the skin mechanoreceptors such Ruffini corpuscles but are located either in tendons (Golgi tendon organs) or muscles (muscle spindles), as well as in ligaments and joints. Activation of these receptors informs the brain about limb position and movement for motor control. Proprioceptors tend to have sustained responses, whereas kinesthesis receptors tend to be more transient. These receptors allow you to do things such as touch your nose with your eyes closed. Limb movement also stretches the skin around joints in various ways that activate cutaneous mechanoreceptors. This activity is also used by the brain for motor control.

Temperature Receptors

Temperature receptors have the morphology called free nerve endings, which are similar to axon terminals without myelination or other enveloping structures around them. There are 2 basic types of temperature receptors: “warm” receptors that respond with an increase in firing to temperatures above body temperature (37°C), and “cold” receptors that respond with an increase in firing to temperatures below body temperature. The free nerve endings that form warm receptors are unmyelinated C fibers that have a relatively slow conduction velocity, whereas those of cold receptors are a mixture of C and Aδ fibers.

Temperatures below the range of cold fibers and above that of warm fibers are potentially tissue damaging and activate temperature pain receptors. Temperatures below about 10°C or above about 45°C to 50°C are felt as cold pain and heat pain, respectively. These sensations are mediated by a different set of temperature-pain fibers, discussed in the following section. Different warm and cold fibers have different peak temperature sensitivities. Figure 15–10 shows a diagrammatic distribution of the responses of cold, warm, cold pain, and heat pain fibers. The peak of the cold fiber population response is between 20°C and 30°C, but individual fibers respond maximally at temperatures above and below this. The warm fiber population maximum is between 40°C and 50°C, with individual fibers having peak responses over a broader range. Note the minimum response of the entire population at 35°C to 37°C (normal body temperature). As with the broad spectral sensitivity curves for cones in color vision, temperature perception can be quite precise when derived from the ratio of activity of broadly tuned temperature receptors.

The transduction of temperature is mediated primarily by channels called transient receptor potential (TRP) protein complexes. TRPV1 provides the sensation of scalding heat, whereas cold is mediated by TRPM8, whose conductance for Na⁺ and Ca²⁺ is temperature dependent. TRPV1 is also activated by capsaicin, the active ingredient in chili peppers. Interestingly, the TRPA1 (painful cold) and TRPV1 receptors can be found on the same populations of nociceptors, resulting in the “burning cold” sensation with very cold temperatures (<17°C). The percept of cooling can arise from ligands such as menthol that bind the channel and modulate its conductance. Temperature also affects the kinetics of other membrane channels, such as potassium channels.

Pain Receptors & Transduction

There are a variety of stimulus modalities that cause pain, such as extreme temperatures, pressures, or chemical acids and bases. The modalities that are sensed by peripheral pain fibers and interpreted by the brain as pain have the common property of being associated with tissue damage.

Pain Receptor Anatomy & Physiology

Pain neurons, known as nociceptors, have the morphology of free nerve endings. However, unlike thermoreceptors, nociceptors only respond to noxious stimuli. Under normal circumstances, nociceptors are generally responsible for the initiation of pain processing. Two major types of nociceptive fibers are Aδ and C fibers. Aδ fibers are myelinated and have a fast conduction speed. Therefore, they are responsible for the first, sharp pain. As mentioned earlier, C fibers are unmyelinated with slow transmission. Therefore, their activation results in throbbing and burning second pain. Some pain fibers respond by firing action potentials exclusively to stimuli near and above the level of damage, whereas other skin receptors report a nonpain sensation at low stimulus intensities, but pain at intensities above a threshold. An example of this dual function is seen in receptors with the aforementioned TRP channels. In addition, the specific modality that a nociceptor responds to is determined by the expression of ion channels.

Complex Aspects of Pain Perception

There are important differences between pain and other cutaneous senses. Pain activates neurons in many areas of the brain beyond the somatosensory cortex, and compared to other cutaneous senses, the perception of pain is strongly modified by and can alter the mental state. Cognitive factors can strongly mitigate the sensation of pain, such as seen during childbirth or when soldiers ignore significant wounds. Pain can also produce long-lasting changes in mood. For example, chronic pain, or pain that remains after an injury has healed, is often clinically associated with depression. Although pain exists to warn of imminent body damage, it can become disabling in disease conditions such as cancer, for which no behavior can reverse the damage or reduce the pain. The current understanding of chronic pain is that the input to the spinal cord becomes sensitized such that previously nonpainful stimuli elicit a pain response. This process is termed central (as part of the central nervous system) sensitization and is thought to be mechanistically similar to the process of long-term potentiation studied as the basis for memory in the hippocampus.

The attempt to understand the nociceptor and the complex phenomenon that is pain has resulted in a number of theories over the past century. The specificity theory was the belief that there were nociceptors that responded to painful stimuli and only this type of stimulation. This was later replaced by the intensity theory, which stated that a stimulus can be painful or not based on the intensity of the stimulation at the nociceptor. An example would be something akin to a heat stimulus—not painful in the short term, but more painful as the intensity increases. As more neurons were identified in the skin (A and C fibers), the pattern theory emerged, wherein the pattern of cellular activation was the signal to determine whether or not a stimulus was painful. However, none of the previous theories took cognitive factors into account (mentioned earlier). Currently, the only theory that takes both physical and psychological aspects of pain into account is the gate control theory. Briefly, this theory states that nonpainful stimuli can prevent the sensation of pain from traveling to the central nervous system by closing the “gate” to painful stimuli. Put another way, signals from nonpain fibers can block signals from pain fibers, leading to an inhibition in pain. This theory explains why rubbing an area that is in pain seems to provide relief and is a “natural” response to injury. The theory also allows for input from the brain to “close the gate,” and we know that this process is referred to as descending inhibition of pain.

The pain system in the brain uses a unique set of neurotransmitters and antagonists so these receptors can specifically reduce pain without affecting the ability to perceive other sensations. An important class of pain neurotransmitters are the so-called endogenous opioids (opioids generated within the body), such as the endorphins (an abbreviation of endogenous morphine). Endorphins are produced in pain-stress situations, such as running the last few miles of a marathon.

Endorphins contribute to the placebo effect, where patients experience pain relief when given an inert substance they believe will block pain. Opiates such as morphine and heroin reduce pain because they bind the same receptors that endorphins bind; however, when injected in large doses, these drugs produce a “high” and are addictive. The drug naloxone antagonizes the effects of these opioids and is often given to addicts to reverse the effects of heroin they have injected. Naloxone also blocks much of the placebo effects.

There are notable individual and group differences in pain tolerance. Pain tolerance generally increases with age, for example. Some data suggest that men are more tolerant of acute pain but less tolerant of chronic pain than women. Athletic training enhances pain tolerance. Cultures that encourage expression of emotions in general tend to be associated with lower thresholds for reporting pain, but it is not clear that such differences reflect anything but a labeling, rather than an experiential threshold.

An important brain area for processing pain is the anterior cingulate cortex, an anterior mesocortical area immediately above the corpus callosum. This structure has an executive role in allocating cortical processing and priorities according to current goals and results. Consistent with its function of arbitrating between taking different strategies in response to experience, the anterior cingulate is activated by pain, the anticipation of pain, and the production of errors in goal pursuit.

In terms of the brain’s control of pain, an important brain region is the periaqueductal gray (PAG). This area is involved in descending modulation, which ultimately leads to the inhibition of a pain signal before it reaches the brain. This mechanism underlies the gate control theory and is supported by the fact that when the PAG is stimulated, it produces analgesia. The rostral ventromedial medulla receives input from the PAG and is considered the final output pathway for the descending inhibition of pain. Importantly, the dopaminergic, serotonergic, and noradrenergic projection areas also send axons down to the spinal cord to reduce the transmission of pain.

CENTRAL PROCESSING OF CUTANEOUS SENSATION

Skin receptors convey information about skin contact to neural circuits, mediate somatosensory perception, and enable responses to cutaneous stimuli. The touch, temperature, and pain messages encoded by cutaneous receptors are sent via synapses to neurons within the same spinal cord segment, to other spinal cord segments, and to the brain. Somatosensory axons synapse in the spinal gray area, illustrated schematically in Figure 15–4. The neurotransmitter for most mechanoreceptors is glutamate, but pain neurons may also release the neurotransmitter substance P.

Cutaneous Synapses in the Spinal Cord

Local spinal cord circuits mediate reflex arcs through spinal interneurons such as the withdrawal reflex (eg, flexion of the bicep to move the hand away from something hot or an otherwise painful stimulus that has been touched), shown in Figure 15–8. A spinal withdrawal reflex such as this typically consists of monosynaptic and polysynaptic pathways. Recall that cutaneous receptors are the axonal endings of sensory neurons located in the dorsal root ganglia just outside the spinal cord. Action potentials travel in what in most neurons would be considered a retrograde manner from the receptor ending to the axonal bifurcation near the cell body (see Figure 15–4) and then into the spinal cord via the dorsal “root” or axon bundle. The cutaneous axon may synapse directly on lower (α) motor neurons that drive flexor muscles such as the biceps. At the same time, some of these cutaneous axons also synapse on spinal cord interneurons in the gray cell body area of the spinal cord. These interneurons simultaneously inhibit the extensor motor neurons whose relaxation allows the flexor muscle to move the hand away from the painful stimulus. The cutaneous input also projects (sometimes through interneurons) to the contralateral side of the same spinal segment and other spinal segments to allow coordinated withdrawal. For example, in an encounter with a very painful stimulus, not only is the stimulated arm involved in withdrawal, but also the trunk and legs, and this activity is coordinated to maintain balance during the evasive maneuver.

Cutaneous Projections to the Brain

Cutaneous inputs to the spinal cord also synapse on centrally projecting neurons. Projection neurons in the spinal gray area send axons to the thalamus (ventral posterolateral nucleus) and other brain areas via 2 major tracts: the medial lemniscus and spinothalamic tract (illustrated in Figure 15–11). The medial lemniscus is composed of large-diameter mechanoreceptor axons. The spinothalamic tract consists of smaller diameter temperature and pain axons that project to, in addition to the thalamus, the reticular formation. Central projection axons cross within the spinal cord before ascending to the thalamus so that the left side of the brain receives information from the right side of the body and the right side of the brain receives information from the left side of the body. The target of these central projections is the ventral posterolateral nucleus of the thalamus, which in turn relays to Brodmann areas 1, 2, and 3 in the postcentral gyrus of the parietal lobe.

Somatosensory Cortex

From each thalamus, somatosensory signals travel to the somatosensory cortex, which is located just anterior to the central sulcus (Figure 15–12) in the most anterior portion of the parietal lobe (the central sulcus is the border between the frontal and parietal lobes). Somatosensory cortex consists of at least 3 subdivisions (S1, S2, and S3). Although it was once thought that these areas composed a hierarchy in which S1 projected to S2 and S2 to S3, it is now clear that there are direct projections from the thalamus to S2 and S3, which also may receive projections from S1. Thus, consistent with the tradition of referring to all cortical areas that receive direct thalamic sensory projections as “primary,” S1, S2, and S3 would all be considered to be primary somatosensory cortex. These 3 somatosensory areas project to more posterior loci in the parietal lobe, where somatosensory information is linked in terms of location with respect to self and to that from other senses, particularly vision and audition.

This body map in the somatosensory cortex is sometimes called a “homunculus,” and it is very similar to the motor output map just anterior to the central sulcus in the frontal lobe. The somatosensory map or homunculus has several odd features. One is that, contrary to what one might expect, distal, peripheral areas of the body such as the feet are located medially, on the wall of the medial longitudinal fissure, whereas central facial and mouth structures such as the lips and face are laterally located. A second odd feature of the representation is its distortion: Small skin areas of the face and fingertips, for example, occupy cortical areas as large as the entire trunk. This distortion generally follows the principle that each unit of cortical area processes inputs from a relatively constant number of peripheral receptors. In Figure 15–7, one can see that since fingertip skin has 50 to 100 times the receptor density as trunk areas, fingertip representation per skin area will be larger on the somatosensory cortex by that ratio. This constant receptor density per cortical area principle also extends to other senses.

A third oddity of the cortical somatosensory representation is that it exhibits several discontinuities. For example, although, in moving along the skin, the fingers are quite distant from the forehead, their representation in somatosensory cortex is adjacent. This feature of the map has implications for some types of phantom limb pain, where amputees may report pain or touch they perceive to be felt in a limb they no longer have when their faces are touched. Another discontinuity is that there is a small representation of the head near that of the upper trunk, but a separate, much larger discontinuous representation of the face more laterally in the cortex along with the representation of the vocal apparatus.

It is not currently clear the extent to which mechanosensation and other cutaneous senses such as temperature and pain have anatomically segregated representations in somatosensory cortex. Pain sensation is projected from the spinal cord not only through the thalamus, but also via the reticular formation. This latter pathway allows pain inputs to generate autonomic responses such as tachycardia and sweating independent of conscious awareness. Pain inputs to the brain can also activate midbrain descending pathways that module its character, often via opioid neurotransmitters that act within the dorsal gray area of the spinal cord.

Proprioception & Kinesthesis Are Mediated by Tendon & Joint Receptors

Throughout the body are mechanoreceptors whose structure and physiologic characteristics are similar to cutaneous receptors, but whose function is to report limb location (proprioception) and limb movement and force (kinesthesis). Although these topics are treated more fully in the chapters on motor control, it is appropriate to consider some as aspects of proprioception and kinesthesis with cutaneous senses for several reasons: (1) the sensory transducers for proprioception and kinesthesis such as Golgi tendon organs and muscle spindles are mechanoreceptors with similarities in their physiology and morphology to cutaneous mechanoreceptors; (2) the cell bodies for these receptors are located in the dorsal root ganglia, like other mechanoreceptors; (3) the central projections follow a similar scheme; and (4) some information about joint position and movement is obtained from skin cutaneous receptors because skin around joints is stretched during movement. Proprioception and kinesthesis receptors also participate in spinal reflex loops that react to changing loads to maintain limb position, rather than withdrawing a limb in reaction to a painful stimulus, for example.

CONCLUSION

The cutaneous senses inform the brain about what contacts the skin. There are 4 major mechanical transduction pathways for various types of touch. Other fibers mediate cold and hot temperature perception. Pain is transmitted by a variety of other fibers sensitive to extreme mechanical pressure, temperature, or chemical activity (eg, acids and bases). The somatosensory system follows a common sensory plan in which peripheral receptors relay to synapses in a specific area of the thalamus (ventral posterolateral nucleus), which in turn projects to the postcentral gyrus of the cortex. The cortical “primary” receiving areas are comprised of Brodmann areas S1, S2, and S3. These areas in turn project to more posterior areas of the parietal lobe. The perception of pain is more complex, with projection paths reaching the reticular formation. Pain pathways also use endorphin neuropeptide neurotransmitters, and can be modulated by central cognitive states. The anterior cingulate is an important brain area in the perception and modulation of pain.

SUMMARY

• Skin, besides its function in isolating the body from the environment, has receptors for various kinds of touch (mechanoreceptors), as well as receptors for temperature and pain.

• The skin continuously produces new cells below the dermis that migrate, flatten, and eventually die on the surface of the skin in a layer called the epidermis.

• Virtually all skin receptors are found in the dermis with the exception of occasional receptors located in the transition zone between dermis and epidermis.

• Somatosensory receptors originate from dorsal root ganglion cells whose axons bifurcate, with one end going to the skin and elaborating the receptor and the other end entering the spinal dorsal root and synapsing in the spinal gray area on local circuit and thalamic transmission neurons.

• Four types of mechanoreceptors with distinct transducer structures are specialized to receive tactile information: Pacinian corpuscles, Ruffini endings, Merkel disks, and Meisner corpuscles.

• The response properties of different tactile receptor types fall along 2 major axes, RF size and response transience.

• The RF size dimension is related to depth in the skin, such that deep mechanoreceptors (Ruffini endings and Pacinian corpuscles) tend to have larger RFs than shallow ones (Meissner corpuscles and Merkel disks).

• The Merkel disk and Ruffini endings are sustained-responding fibers, whereas the Meissner and Pacinian corpuscles are transiently responding.

• The 4 major tactile receptors also have different stimulation frequency response properties that mediate different perceptual sensations such as stretch, pressure, or vibration.

• Temperature receptors are mostly of the free nerve ending structure and are called cold receptors if their optimum response is below body temperature or heat receptors if their optimum response is above body temperature.

• Pain receptors are primarily free nerve endings but respond to different inputs (pressure, temperature, chemical) whose commonality is immanent tissue damage.

• Local spinal cord circuits mediate reflex arcs through spinal interneurons such as the withdrawal reflex that triggers flexion of the bicep to move the hand away from something hot.

• Cutaneous inputs to the spinal cord synapse on centrally projecting neurons in the spinal gray area that project to the thalamus (ventroposterolateral nucleus) and other brain areas via 2 major tracts: the medial lemniscus and spinothalamic tract.

• The somatosensory thalamic areas project to the most anterior part of the parietal cortex in Brodmann areas 1, 2, and 3.

• Somatosensory cortex is organized in a body map sometimes called a homunculus, which is distorted such that areas of the skin with a high density of receptors project to larger cortical areas than those with low density; to a first approximation, any unit area of somatosensory cortex processes inputs from the same number of receptors.

• Receptors similar to some mechanoreceptors in tendons and muscles that signal limb position, movement, and muscle force, called proprioceptors and kinesthesis receptors, project to the brain in similar pathways as the mechanoreceptors.

OBJECTIVES

HIERARCHICAL CONTROL OF MOVEMENT

Movement distinguishes animals from plants. The sea squirt begins life as a free-swimming larva with sensory receptors like eyes, muscles that enable it to swim, and a small brain that mediates its behavior. At the end of the larval stage, it permanently adheres to a spot on the ocean bottom and commences to digest its eyes and nervous system. The nervous system exists to control movement. Movement must be directed toward some goal (planning), and the component muscles must be controlled to act in concert to achieve that goal.

Muscles, which produce movement by contraction, are called effectors. Different types of movement are subserved by 3 different primary muscle types: striated muscles for voluntary movement of skeletal bones around joints via attachments by tendons, smooth muscle that contracts the walls of organs like the stomach and esophagus, and unique cardiac muscle that has hybrid properties between striated and smooth muscle.

This chapter will focus on the cortical control of striated muscle underlying voluntary control of limb movement. All vertebrate striated muscles are activated by receipt of acetylcholine from motor neurons. Each motor neuron spike releases sufficient acetylcholine to elicit an action potential in the muscle cell. The rate of muscle cell action potentials is generally proportional to the contractile force the muscle generates. The motor neurons emanating from the spinal cord that activate muscles that control the limbs and limb appendages (hands and feet) are called lower or α motor neurons. Lower motor neurons are controlled, in turn, by neural circuits in the spinal cord, brainstem, and motor cortex.

Direct activation of lower motor neurons by neurons in primary motor cortex (upper motor neurons) is phylogenetically new and permits the learning of novel and complex motor sequences such as associated with manual praxis. In contrast, basic, highly evolutionarily conserved movement sequences such as walking and running are controlled by circuits in the spinal cord, with compensation for rough terrain and other perturbations mediated by circuits in the brainstem and cerebellum that receive vestibular and visual input.

The control of movement is done by a hierarchy of neural circuits ranging from the spinal cord to the brainstem to primary motor cortex and prefrontal areas. The focus of this chapter will be on movement control via primary motor cortex. Because the axons of the upper motor neurons emanate from pyramidal cells in layer 5 of this cortical area, it has been named the pyramidal tract, but it is also referred to as the corticospinal tract.

Neocortical Areas Mediating Motor Control

The cerebral neocortex is composed of 4 major lobes: frontal, parietal, temporal, and occipital (Figure 16–1). The parietal, temporal, and occipital lobes are primarily sensory, processing inputs from the 5 major senses. The frontal lobe is concerned with movement. Anterior parts of the frontal lobe, called prefrontal, are involved in goal generation and motor planning for set goals. Middle areas of the frontal lobe (Brodmann area 6), particularly the supplementary motor area (SMA) and premotor cortex (PMC), organize and generate complex motor sequences involving the movement of many limbs in a coordinated manner through time. These areas are shown in more detail in Figure 16–2. SMA is largely more medial and anterior than PMC. The primary motor cortex (Brodmann area 4) is the output of this system via the pyramidal or corticospinal tract.

The Motor Control Hierarchy

The hierarchical nature of motor control is best appreciated from an evolutionary perspective. The behavior of primitive, low-cephalic animals is primarily reflexive, driven by immediate stimuli and basic limbic system–level drives such as hunger and predator avoidance. In developing complex, internal goal–driven behavior, high-cephalic animals, such as humans, have retained many of the circuits and competencies of previously evolved lower level systems but have layered more complex long-range, goal-driven behavior on top of these.

Figure 16–3A is a simplified illustration of the main components of the mammalian motor control system. Prefrontal goal-setting neural circuits project to SMA and PMC, which in turn drive primary motor cortex. The primary motor cortex, via projections of its axons through the corticospinal tract, can directly drive lower motor neurons or indirectly drive them through the brainstem. The primary motor cortex also receives sensory feedback information from the limbs and cerebellar input that helps coordinate multilimb movement.

Spinal Cord Segment Circuits

The lowest level of the control hierarchy is illustrated in Figure 16–3B. At the spinal cord level, the maintenance of limb position uses feedback between sensors for limb position (projected via Ia afferent fibers) to drive α (lower) motor neurons to generate restoring force to perturbation within a single spinal cord segment. However, the motor neuron efferent can be driven directly by upper motor neuron axons from primary motor cortex.

The pyramidal or corticospinal tract is shown in Figure 16–4, along with sensory tracts that send information from the skin, joints, and muscles to the sensory homunculus in somatosensory cortex. This tract of upper motor neurons first passes through the cortical white matter. Upper motor neuron axons that will innervate lower motor neurons in the spinal cord then travel down the brainstem where they cross (“decussate”) in the medulla to the opposite side of the body. The area of the medulla where this occurs is called the medullary pyramid. Some references suggest the name pyramidal tract comes from this fact rather than from the tract’s origin in giant Betz pyramidal cells in motor cortex.

Primary motor neuron axons that will innervate muscles above the spinal cord in the neck and head enter tracts called corticobulbar tracts that terminate on nuclei in the brainstem. Axons emanating from these brainstem nuclei compose the cranial nerves that control muscles of the face and mouth.

Within the spinal cord, the corticospinal axons run in 2 tracts, shown in Figure 16–5. The largest of these is the lateral tract, whereas there is also a ventromedial tract called the ventral tract.

Integrated Control

At the next level of the hierarchy, basic motor behaviors, such as walking or running, can be generated by spinal cord mechanisms embodied across many spinal cord segments that generate motor stimulus patterns. Above that, brainstem mechanisms that receive vestibular and visual input add balance and rough terrain compensation. Above that is direct and indirect control by the primary motor cortex, which receives input from frontal and prefrontal lobe complex goal-action sequence plans. The corticospinal tract not only passes through the brainstem on the way to the spinal cord, but it also innervates several motor control nuclei within the brainstem. These form the nonpyramidal tracts and are discussed in more detail in Chapter 17.

At the top of the hierarchy, motor sequences consequent to goal pursuit are initiated in an interaction between the basal ganglia and prefrontal cortex (see Chapter 17). Prefrontal cortex is the locus of interaction between limbic system goals and working memory. Its output is the choice among the many possible means of accomplishing that goal, gating neural circuits that lead to organized coherent muscle command sequences. This is accomplished in multiple stages through feedforward projections between prefrontal areas and the basal ganglia. The output of these prefrontal computations is sent to cortical areas just anterior to primary motor cortex, namely the SMA and PMC. The output of these areas is to the primary motor cortex, whose output axons (upper motor neurons via the corticospinal tract) drive α or lower motor neurons in the spinal cord. This output is modulated by inputs from the cerebellum (see Chapter 17), which participates in motor learning of timing sequences and in the formation of an internal model of the skeleton-muscle system.

The pyramidal or corticospinal tract originating in the primary motor cortex can bypass lower level motor control circuits that evolved to control basic functions such as balance while standing, movement gaits such as walking, and whole-hand grasping (see Figure 16–3B). Direct primary motor cortex control allows complex, arbitrary learned sequences to be generated, such as typing, which involve movement of the fingers individually and sometimes in opposite directions. Sophisticated motor sequences such as typing or playing a musical instrument could not have evolved as genetically embedded capabilities like grasping have. The ability to control individual muscles through primary motor cortex requires some sort of map that can allow the muscle movement goal generated in the prefrontal cortex to project to and act on that muscle’s upper motor neuron. That map is called a homunculus.

Cortex & Motor Homunculus

Primary sensory cortical areas are typically those that receive a direct input from the area of the thalamus that receives a direct input from the physiologic sensor array, such as the retina for vision or the cochlea for hearing. These primary cortical areas form a map that usually conforms to the sensor layout. In the case of skin senses, the layout is the skin surface. Underneath the skin at various joints and in muscles are receptors for joint position and acceleration and muscle force. The various types of cutaneous and proprioceptive transducers form a map called a homunculus in somatosensory cortex just posterior to the central sulcus. Several kinds of distortions and discontinuities in this map, and the reasons for them, are discussed in Chapter 15 on skin senses.

Anterior to the central sulcus is a very similar map for motor output from the primary motor cortex, shown in Figure 16–6, which strongly resembles the sensory homunculus across the central sulcus in parietal cortex. The proximity of the sensory and motor representations allows direct, fast interactions between cutaneous and proprioceptive inputs from a particular part of the body with motor projections from primary motor cortex. Similar distortions in the 2 maps indicate that muscles that receive many upper motor neuron inputs, for precise control, also tend to have many associated sensory outputs.

Layer V pyramidal cells in the primary motor cortex send their axons through the white matter and then through the brainstem to the spinal cord via the pyramidal or corticospinal tract. Some, but not all, corticospinal tract axons originate from very large pyramidal cells called Betz cells, which are unique to the primary motor cortex. A section of motor cortex is shown in Figure 16–7. The large Betz pyramidal cells are unique to layer V in motor cortex so that it is sometimes called agranular, meaning not dominated by uniformly smaller cells. A small minority of corticospinal axons originate not in the primary motor cortex (Brodmann area 4), but in the supplemental motor cortex (SMA: area 6).

CORTICAL CONTROL OF VOLUNTARY MOVEMENT

Highly evolutionary motor behaviors such as walking, running, and grasping are controlled primarily via subcortical mechanisms. With the addition of vestibular and visual input, balance can be maintained and obstacles avoided through brainstem integrative mechanisms. However, more complex behaviors, such as digging and manual manipulation of seeds, nuts, and eventually tools, were enabled by the neocortex and the prefrontal, frontal, and primary motor cortex systems, which together compose nearly half of the entire human brain.

Premotor Cortex Organizes Muscle Groups During Learned Sequences

The output of the prefrontal–basal ganglia motion selection system is a projection to areas just anterior to the primary motor cortex that encode activation of coordinated groups of muscles, namely the SMA and PMC. These systems have parallel functions that differ primarily in organizing novel versus well-learned tasks and are shown in Figure 16–8. The PMC is activated primarily for coordinating activation of the primary motor cortex for novel tasks, that is, while they are being learned. In doing so, it requires feedback about the state of the limbs from the parietal somatosensory and proprioceptive areas and the state of the motor program sequence from the cerebellum.

Supplementary Motor Cortex Organizes During Monitored Movement

The SMA is activated for task performance when tasks are well-learned, including imagining performing a motor task. Memory input to this system comes from the hippocampus, and there may also be input from multimodal sensory areas at the junction of the parietal, temporal, and occipital lobes. The output of the SMA, like the PMC, is almost entirely to the primary motor cortex, where a map of individual muscles exists. However, unlike the PMC, there are some direct projections into the corticospinal tract. These projections are different from those from the primary motor cortex, however, because virtually all corticospinal axons that are monosynaptic on α (lower) motor neurons arise from primary motor cortex.

Motor Learning & Population Coding

The primary motor cortex is more than just an excitation map for the muscles of the body. It is also involved in learning motor sequence behaviors. Its plasticity is evident in changes in the sizes of areas associated with specific muscles or muscle groups following training, with areas associated with extensive limb use becoming larger with practice. Task learning is also associated with initial increases in motor cortex activity, followed by decreases as the task is mastered, paralleling the switch from PMC- to SMA-dominated input control.

Neurons in the primary motor cortex fire according to the direction a limb is moved. The tuning of individual neurons is broad, however, so that it has been proposed that the limb movement direction is generated by a population code whereby the actual movement is a vector sum of all firing neurons weighted by their activity. Such a code is thought to provide robustness against noise and greater computational speed than rate coding by a smaller number of neurons.

FRONTAL & SUPPLEMENTARY EYE FIELDS

Anterior to SMA and PMC in the frontal lobe is Brodmann area 8, called the frontal eye fields (FEFs). This frontal area appears to receive direct input from the superior colliculus via the medial dorsal nucleus of the thalamus. This area also receives input from the medial pulvinar, an area of the thalamus associated with visual attention. The FEF and the more medial supplemental eye fields (SEFs) are involved in the voluntary control of eye movement, particularly saccades and slow pursuit.

PREFRONTAL CORTEX INSTANTIATES OVERALL MOVEMENT GOALS

In most current function maps of neocortex, the prefrontal area is the most undifferentiated. It is unclear whether this reflects our present ignorance of its distinct functional prefrontal subareas or if perhaps the prefrontal cortex operates somewhat like the “association cortex” postulated in the early 20th century. It is clear, however, that this area is proportionately larger in primates than other mammals and in humans compared to other primates. Research and examination of clinical cases of prefrontal damage show clearly that it is the seat of working memory, where the representation of objects and goals is maintained during the pursuit of a task when the direct sensory input is no longer present.

Why has the prefrontal lobe become so large in primates, particularly humans? One hypothesis is that this enlargement instantiates more levels in the goal-motion control hierarchy that translates into a higher level of abstraction in goal pursuit. For example, if you want to sit down and there is nothing to sit on, you may need to get a job to buy a chair. Another possibility is suggested by the fact that humans perform more varied and more complex motion sequences than any other animal, and virtually all of these are learned.

Primates in general and humans in particular have among the most complex social organizations of any animals. Damage to prefrontal cortex typically results in deficits in working memory, specifically as applied to maintaining contingencies in complex social situations. The appropriateness of human behavior is uniquely dependent on multiple facets of the identity of the person with whom one is interacting. Patients with considerable prefrontal damage can retain higher than normal test-measured intelligence quotients while having extremely poor social function and engaging in inappropriate behavior, such as excessive gambling and treatment of other humans as objects. Appropriate behavior in complex societies clearly places high demands on the use of episodic memory to select the best ends and means in pursuit of limbic system–generated needs.

CONCLUSION

The pyramidal motor system is the most recently evolved circuit in the motor control hierarchy. Lower motor control levels in the spinal cord and brainstem mediate balance, basic gaits such as walking and running, and elementary manual praxis such as grasping. The pyramidal system, through the corticospinal tract and feedback from thalamus and cerebellum, allows frontal control of individual muscles in complex, novel, learned sequences such as dancing the tango or playing the violin.

SUMMARY

• Motor control is mediated by a hierarchical control system, with spinal level reflexes at the lowest level and cortical control via the pyramidal (corticospinal) tract from primary motor cortex at the highest.

• At the spinal cord level, the maintenance of limb position uses feedback between sensors for limb position to drive α (lower) motor neurons to generate restoring force to perturbation within a single spinal cord segment.

• At the intermediate level, brainstem nuclei such as the red nucleus control and modulate stereotyped behavior such as gait.

• The evolutionarily newest level of control is that of direct muscle activation by primary motor cortex via the pyramidal tract.

• Primary motor cortex is organized in a body map sometimes called a homunculus, which is similar in layout to the somatosensory map just across the central sulcus in the parietal lobe.

• The pyramidal tract permits complex motor control of learned motor sequences via projections it receives from the immediately anterior motor regions (supplementary motor and premotor areas), which in turn are sequenced and modulated by the basal ganglia and cerebellum.

• The prefrontal cortex, which is the locus of the most general representation of action goals, projects to supplementary and premotor cortical areas that in turn project to primary motor cortex and the pyramidal tract to activate specific action sequences to meet current goals.

• The pyramidal motor system, as the most recently evolved circuit in the motor control hierarchy, works with lower motor control levels in the spinal cord and brainstem to mediate balance, basic gaits such as walking and running, and elementary manual praxis such as grasping, whereas the pyramidal system, through the corticospinal tract and feedback from the thalamus and cerebellum, allows frontal control of individual muscles in complex, novel, learned sequences such as dancing the tango or playing the violin.

OBJECTIVES

BASAL GANGLIA INITIATE MOVEMENT

What structures in the brain cause movement? Bacteria and plants exhibit phototaxis, movement toward or away from the sun. Virtually all animals execute reflex movements such as moving a limb away from a painful stimulus. Vertebrate animals clearly have higher, limbic system types of goals such as pursuit of food, water, and mates and avoidance of predators and temperature extremes. Humans are thought to consciously make movement decisions based on high-level goals. The likely locus of these high-level goals and their translation into motor sequences were postulated in Chapter 16 to be the prefrontal cortex. However, as any Parkinson disease sufferer can tell you, an intact prefrontal cortex alone is not sufficient to initiate and control movement.

The initiation and modulation of movement in vertebrates generally depends on 2 phylogenetically old systems: the basal ganglia and cerebellum. In mammals, the basal ganglia system interacts with the frontal lobe to produce motor behavior (Figure 17–1). The large, complex frontal neocortex provides the basal ganglia with a wealth of behavioral options. The basal ganglia, acting through the thalamus, act as a central program that calls frontal and cerebellar subroutines that contain the expertise for specific, complex motor sequences. Much of this high-level control relies on lower level motor expertise in the brainstem and spinal cord that controls balance and limb alteration and other basic locomotion procedures.

Organization of the Basal Ganglia: Nuclei

Figure 17–2 shows a pseudo-3-dimensional rendering of the basal ganglia on the top left and a coronal (frontal) section through the basal ganglia and thalamus on the right. The basal ganglia proper consist of an interconnected system of several nuclei. The main processing nuclei of the basal ganglia are structures called the globus pallidus, which is composed of internal and external segments (GPi and GPe). The globus pallidus is lateral to the thalamus. Lateral to the globus pallidus is the putamen, an input area to the globus pallidus. The caudate nucleus passes above the thalamus and has direct inputs to the globus pallidus and indirect inputs through the putamen. The caudate and putamen together are called the striatum. Associated with the basal ganglia and often described as part of it are 2 midbrain nuclei: the substantia nigra (SN) and subthalamic nucleus (STN), both of which project to the globus pallidus.