Vision

Topics

• To introduce the general issues related to the transduction of light into vision.
• To describe the anatomy and physiology of the eye and retina.
• To discuss two forms of neural coding: lateral inhibition and color opponency.
• To describe the major sub-modality pathways (M, P, and K) leaving the retina from specialized ganglion cells.
• To describe the targets of optic nerve projections.
• To discuss different form of brain damage on visual perception.

Preparation for Class

• Reading: PN6 Chapter 12

Topic Slide

Hunayn Ibn Ishaq (809-873) wrote "Ten Treatises On Ophthalmology". Ishaq was born in southern Iraq and went to Baghdad to study medicine. He became well known as a translator of Greek texts into Arabic. However, he also published original studies, including the Ten Treatises. This occurred during the early middle ages (or, Dark Ages) in Europe. The major work of Ishaq was later plagiarized and attributed to a European writer during the Enlightenment.

Ishaq's work overcame early conceptualizations of Galen, among others, that vision was related to a type of fire that emanated from the eyes. The evidence for this fire was the “seeing stars” phenomena that occurs with a blow to the head. These were interpreted as sparks related to this ‘fire’.

Light and vision

The Euglena (a eukaryote protist that can both make its own food from photosynthesis like a plant but also eat other organisms like an animal) has an 'eye spot' with photosensitive opsin molecules that permits it to phototax (orient and move to light).

The mollusk shows wide variation in eye form and function. What advantage do you think a recessed (cup) photoreceptor array provides?

The human visual system transduces electromagnetic energy within a restricted range referred to as visible light. The visible light spectrum is bounded on the lower end by the infrared spectrum and on the upper end by ultraviolet light. Some snakes can 'see' infrared light (which is sensitive to heat, and which can be transformed into a visible light signal by night vision goggles). Some birds can 'see' ultraviolet light.

Issues in vision

Objects have visual properties that are represented in vision.

• Luminance
• Color
• Motion
• Contrast
• Edges and contours
• Location in space (including depth)

There are issues in representing the world about which the visual system must contend.

• Inverse problem
  • The 3D world is projected upon a 2D retina. How can we distinguish among object sizes at different depths?
• Compression problem
  • There is a greater range of input presented to the retina than can be transmitted to the brain. There are many fewer axons in the optic nerve (~1M) than there are photoreceptors (~106M).
• Visual perception is constructive
  • The visual system guesses or fills in and is sometimes tricked. This can be the basis of visual illusions.

Anatomy and physiology

Anatomy of the human eye

An anatomical drawing of the eye can be found here. Some of the main components are listed below:

• Cornea – the transparent layer that covers the iris and pupil.
• Sclera – the tough white covering of the external eye that is continuous with the cornea.
• Iris – contains sphincter muscles that controls the opening into eye.
• Pupil – – black hole through the sclera that passes light and stands in front of the lens. The size of the pupil is determined by the muscles of the Iris.
• Lens – – double convex lens that inverts image as it projects image onto retina.
• Retina – where the light entering through the pupil is focused by the lens and where the photoreceptors are located.
• Photoreceptors – light sensitive receptors that transduce photons of light into neural signals.
• Blind Spot – an area of the retina where axons from each ganglion cell form the optic nerve (see below). There are no photoreceptors in the blind spot and, thus, no visual signal.

The human retina is inverted in that the photoreceptors are at the back of the retina and are covered by layers of neurons and their fibers that make up the optic nerve. The fibers exit the retina in mass and create a blind spot since there are no photoreceptors where the nerves exit. This inverted organization is found in all vertebrates. However, the cephalopod (octopus) retina is not inverted.

The loss of acuity of an inverted retina is partially compensated by the lack of obscuring fibers and blood vessels over the fovea (macula), the most sensitive portion of the retina.

The inverted retina may be an adaptation that provides a double blood supply to the highly energetic photoreceptors (higher oxygen consumption than the rest of the brain). There is a blood supply behind and in front of the photoreceptors.

Anatomy of the retina

The retina contains sensory receptors (photoreceptors) and neurons. You can find a detailed discussion of the retina here.

Photoreceptors

• Rods (100,000,000) are extremely sensitive to light (they can respond to a single photon in the visible light spectrum). Rods have large receptive fields, thus low spatial acuity.
• Cones (6,000,000) require several photons of light to activate. They have very small receptive fields, thus higher spatial acuity than rods.
  • Three kinds of cones. S, M, L for short, medium, and long wavelengths of visible light. This corresponds to blue, green, and red light, although the spectral sensitivities are broad and overlapping.

The photoreceptors respond to light intensity over a range of 11 log units:

• Only rods can respond to the weakest 3-4 log units of light. This is referred to as scotopic vision.
• Only cones can respond to the highest 4-5 log units of light. This is referred to as photopic vision.
• The pupil dilation/constriction can modulate 2 log units of light.
• The large range of light requires processing within the retina to convert intensity signals to contrast signals – i.e., emphasize differences in light intensity, not absolute intensity.
• There are some ganglion cells that have melanopsin and respond directly to absolute light intensity. As we will see later, these ganglion cells are not involved in pattern vision, but in circadian rhythms and the pupil dilation/constriction response.

Neurons

The photoreceptors transduce photons of light in the visible light spectrum to neural signals. The ganglion cells are the output cells of the retina, and project the signals from the retina to the visual thalamus (the dorsal lateral geniculate nucleus, or dLGN) and other targets. However, there are other neurons within the retina that form a vertical (bipolar cells) and horizontal (horizontal and amacrine cells) organization. Retinal process is responsible for a first level analysis and compression of visual information. The organization of the retina is well described in this figure.

Vertical organization

There is a vertical signal path in the eye in which the photoreceptor stimulates bipolar cell which stimulates a ganglion cell (photoreceptor -> bipolar cell -> ganglion cell). The axons from the ganglion cells make up the optic nerve (cranial nerve II) which exits the eye and carries signals to the visual thalamus (dorsal lateral geniculate nucleus) and other targets described further below.

The bipolar cells are stimulated by the photoreceptors. There are two types of bipolar cells:

• Off cells activate in the dark.
  • Glutamate ionotropic receptors
• On cells activate in light.
  • Glutamate metabotropic receptors

The output of the bipolar cells is transmitted to the ganglion cells.

Horizontal organization

There is also a horizontal organization within the eye (photoreceptor -> horizontal/amacrine cell -> bipolar cell).

The Horizontal cells and Amacrine cells are key for lateral inhibition. The bipolar cells (and consequently ganglion cells) responsive to light/dark within a small region of the visual field (the receptive field) are inhibited by cells responsive to light/dark within an adjacent region of the visual field.

Output layer

Ganglion cells are the only cells in the retina that generate action potentials. The photoreceptors, bipolar cells, and horizontal and amacrine cells transmit information using graded, passive potentials only (like we discussed earlier for the dendrites). The optic nerve (cranial nerve II) is made up of ganglion cell axons. There are ~ 1 million axons in the optic nerve.

Energy

The retina is considered one of the highest oxygen-consuming tissues of the body, exceeding even that of the brain. Blood supply to the outer retina (mainly the photoreceptor cells) comes from choroidal capillaries that originate from ciliary arteries, whose source is the ophthalmic artery. Blood supply to the inner retina, on the other hand, is provided by the central retinal artery, also a branch of the ophthalmic artery.

Neural coding

The retina solves the compression problem by performing information reduction on the incoming visual signals. With some exceptions, the absolute intensity of visual input to the retina is reduced to differences and contrasts.

Lateral inhibitions and center/surround organization

Lateral inhibition leads to center/surround organization.

There are 'On' center and 'Off' surround ganglion cells:

• Inner receptive field surrounded by an annulus.
• Center of receptive field responds strongly to light.
• Surround annulus inhibits response.
• Strongest response when light falls on center, and no light falls on surround.
• If light falls on both center and surround, the response is muted and is similar to background firing rate.
• If light falls on surround only, and no light falls on center, then cell is strongly inhibited and does not fire at all.

There are also 'Off' center and 'On' surround ganglion cell.

• Just like On/Off cell described above, but opposite in sign. That is, the cell is inhibited by light falling on the center, and activate by light falling on the surround.

What are the functions of the Center/Surround organization?

• Spatial filter – emphasizes edges and contrast.
• Temporal filter – emphasizes motion.

Thus, the retina is sensitive to change – edges and motion. This helps with the compression problem. By differencing light intensity across space, this also helps with the huge range in absolute light intensity for which the visual system is sensitive. Differencing subtracts out absolute light intensity.

The video embedded below presents a nicely illustrated discussion of center-surround organization in the retina. It provides more detail than I did in lecture.

Color and color opponency

Color coding depends upon two components: Trichromatic photoreceptors, and color opponency.

• The trichromatic component relies on different spectral sensitivities of S, M, and L (short, medium, and long wavelength) cone receptors (Young-Helmholtz). Recall the figures used in lecture that showed their spectral sensitivity curves. The S curve is maximally sensitive to blue, the M to green, and the L to red. You might think this is sufficient for color vision, but it isn't.
  • These curves substantially overlap (particularly M and L). Where is yellow?
  • The precise wavelength of the stimulating photon is lost once it stimulates the photoreceptor. The photoreceptor uses an intensity code, not wavelength code. So there is an inherent ambiguity about the precise wavelength of the photon to which the photo receptor responded. So, individual photoreceptors do NOT signal color per say, but rather intensity of light stimulation across a broad range of wavelengths.
• The color opponency model (Hering) is also necessary.
  • Afterimages are consistent with opponent colors.
  • Colors we never see – e.g., reddish-green.
  • (May have, depending upon time) Provided a demonstration in class whereby staring at red and green patches produce, respectively, green and red afterimages. And blue and yellow patches produce, respectively, yellow and blue afterimages.
• It appears that both the Young-Helmholtz and Hering models are correct in that the different spectral sensitivities of cones are combined to create opponent color cells.

How does color opponency work?

• The overall luminance (light intensity) can be represented by a neuron that sums the outputs of cones and rods.
• Red and green can be represents by adding the output of L and S cones and subtracting the output of M (i.e., R+B-G). The resulting opponent spectrum has a peak at red and a trough at green. Thus the activity of a single opponent cell can reflect a location on that opponent spectrum. It is also why we can't see reddish-green, since they have opposite signs on the spectrum.
• Blue and yellow can be similarly represented by summing L and M and subtracting S (i.e., R+G-B).

The embedded video below provides an excellent tutorial on color opponent processes in the retina and visual system.

The concept of opponency is used in other realms of neural coding.

Ganglion cells and sub-modality pathways

There are about 20 different types of ganglion cells that have different output properties. Here are the 3 major types that give rise to three major submodality pathways:

• Parasol ganglion cells give rise to the magnocellular, or M-pathway. These ganglion cells have large receptive fields (and thus low acuity) and high sensitivity. Rods are the source of much of the magnocellular pathway. 10% of ganglion cells are of this type.
• Midget ganglion cells give rise to the parvocellular, or P-pathway. These ganglion cells have small receptive fields (and thus high acuity) and less sensitivity. Cones are the source of much of the parvocellular pathway. 80% of ganglion cells are of this type.
• Short wavelength (blue) cones activate bistratified ganglion cells and give rise to the K (konicellular) pathway. 8% of ganglion cells are of this type.
• Melanopsin ganglion cells signal overall illumination.

Retinal disparity and depth perception

Although not computed within the retina, one important cue for depth perception is the fact that a stimulus in space simultaneously falls upon a different part of the retina of each eye. Retinal disparity is used at higher levels of the visual system as an important cue for localizing objects in space.

Projections of the optic nerve

To where do the optic nerve fibers project in the brain?

• Dorsal Lateral Geniculate Nucleus (dLGN): the visual thalamus. The dLGN projects to primary visual cortex. (Note: the dLGN is often referred to as the LGN, as nobody seems to care that much about the ventral or vLGN).
• Superior colliculus: the SC, or optic tectum (recall that the inferior and superior colliculi make up the midbrain, or mesencephalon), projects to motor areas that control head and eye movements, and to the pulvinar (another thalamic nucleus) that itself projects to other regions of cortex.
  • There is evidence that the superior colliculus also projects directly to visual regions outside of primary visual cortex, such as area MT (as we will discuss in the case of blindsight in next lecture).
• Suprachiasmatic nucleus: a small nucleus above the optic chiasm that plays an important role in regulating circadian rhythms.
• Pretectum and ventral part of the LGN (vLGN): these regions are involved in setting circadian rhythm phases and in controlling pupil dilation/constriction. The do not receive visual pattern (i.e., object) information, but rather light/dark information.

Projections of optic nerve to dorsal lateral geniculate nucleus (dLGN)

The dLGN is a remarkably layered structure. Inputs from each individual eye are segregated to alternate layers.

• M pathway (layers 1 and 2)
• P pathway (layers 3, 4, 5, 6)
• K pathway exists in the pale stripes between adjacent M layers and P layers

Projections of dLGN to V1

Primary visual cortex is known as V1, but is also referred to as striate cortex (for its appearance under the microscope) and as Brodmann's Area 19.

Unlike other sensory areas, layer 4 (thalamic input) has clear subdivisions (4A, 4B, 4C-alpha, 4C-beta) that reflect the submodality pathway the innervates it. Note also that layer 4 is monocular and represents ocular dominance columns – while layers 2-3 are binocular.

• M pathway (4C-alpha)
• P pathway (4C-beta)
• White matter layer (4B)
• K Pathway (4A) – also projects up to layers 2 and 3 (blobs)

There is structure in V1 that can be seen with with cytochrome oxidase staining:

• Ocular dominance columns
  • In layer 4, due to segregation of eyes to alternating layers of dLGN.
• Blobs
  • In layers 1, 2, blobs are color sensitive.
• Inter-blobs (defined by the absence of blobs)
  • In layers 1 and 2, represent form vision.

Damage and deficits

Visual field defects

• The nerve fibers from the right and left nasal retinas (halves of the retina close to the nose) cross in the optic chiasm (X, or 'chi') and are carried to the contralateral hemisphere.
• The nerve fibers from the right and left temporal retinas (halves of the retina close the temples) do not cross in the optic chiasm, and remain in the same hemisphere.
• This arrangement above means that each eye gets input from the left and right visual fields (so, if you close one eye, you don't lose a visual field, although you do lose depth perception). The consequence of this arrangement is that the right visual field projects to the left hemisphere, and the left visual field projects to the right hemisphere.
  • This is the basis for a homonymous hemianopsia – if the left visual cortex is damaged, then the right visual field is blind. If the right visual cortex is damaged, then the left visual field is blind.
    • If only part of the left visual cortex is damaged, then the visual loss may only occur to the upper or lower quadrant of the right visual field. If the upper left visual cortex is damaged, then the lower right visual field quadrant is blind. If the lower right visual cortex is damaged, then the upper left visual field quadrant is blind. Upper visual field projects to lower brain. Lower visual field projects to upper brain.
  • If the optic chiasm itself is damaged, then one suffers a bitemporal hemianopia. This means that both the left and right temporal visual field (the outer part of the each visual field) is blind. Subjects see only the nasal part of the visual field.

Scotoma and visual field disturbances

The blind spot can be thought of as a natural scotoma.

Studies have shown that the visual system 'fills in' the blind spot. I presented examples of offset lines being seen as continuous when these lines cross the blind spot.

Some individuals (due to damage to the optic nerve or to parts of visual cortex) have missing areas of vision. In some cases, these individuals experience filling in of the scotoma by surrounding patterns – such as wallpaper patterns. These individuals may also experience lines filling in the scotoma if the line extends through the scotoma. Scotoma filling does not appear to be high level – that is, it seems to fill in textures but not numbers in a series, or body parts.

Some individuals, however, experience visual hallucinations in their scotomas or hemianopsias. Charles Bonnet syndrome is the name given to this phenomena. Those experiencing these hallucinations know they are not real.

Videos

Prerecorded lectures for 2020

Live lectures from previous semesters

The video embedded below is lecture 10 from fall, 2019.