Chapter 9 - The Visual System: Retinal Processing and Early Vision
• Retina: thin film of cells lining with the inside back wall of the eyeball. 3 dimensions
• All vertebrates have retina made up of similar type of cells.
• Dominant cells in retina = neurons. Forms an outgrowth of central nervous system
invading eyeball during embryonic development. So, any kind of disease/traumatic
injury to retina destroying these cells = permanent blindness because neurons don’t
regenerate & new neurons cannot be created there.
• Historically: retina was thought as having a nutritional role by providing nourishment to
the inner eyes elements like vitreous and lens. Lens were thought to be the main
element of vision.
A) THE PHOTORECEPTOR ARRAY
• Image is projected on on the outer margin of the retina. Along the margin,
photoreceptors are found where the first events in vision occur.
- They are within the fibrous matrix holding eyeball together
- They are light-absorbing elements (first event in vision; light absorption)
- Function as transducing elements: convert the energy of light into signals later
processed by other neurons in the retina.
• Photoreceptors lie on cells called pigment epithelium.
- Contain pigmented material absorbing stray light (light not captured by photoreceptors)
- Vision can be impaired when the pigment is reduced/absent since unabsorbed stray
light interferes with normal processing of visual info. This leads to albino individuals.
1. Rods and Cones
• Two types of photoreceptors: rods and cones
• Rods are more numerous (120 million rods vs 6 million cones)
• Divided into 2 segments: inner & outer
- in rods, outer segment contains a stack of disks
- in cones, outer segment contains a series of infoldings
• Named after their physical shapes (rod: slender, cylindrical vs cone: conical).
• Front-end of two visual systems:
- Rod photoreceptors are the front-end of the visual mechanism serving night vision
- Cone photoreceptors are the front-end of the visual mechanism serving day vision
- Automatically switch from one mechanism to the other
• Major retinal landmarks - retina can be divided into 2 halves: nasal half (toward the
nose) & temporal half (toward the temple):
- Small pit in the midpoint division = fovea (1st landmark)
- Point in nasal retina where retinal nerve fibres exit eyeball has the optic disk also
called “blind spot” (2nd landmark)
• Distribution of photoreceptors in retina by Gustav Østerberg:
- No photoreceptors in the optic disk meaning that a small part of nasal retina is blind
- Small density of cones in peripheral parts of both divisions ( 7,000 per square mm) vs
high density in the center of the fovea (150,000 per square mm). - because of this concentration in the fovea, it is specialized in image detail
processing since the cones are connected to the other neurons in the retina. Doesn’t
function well in low light conditions.
- Numerous rods in the periphery of both divisions, reaching their peak at 5 mm on both
sides of the fovea. Drop dramatically at center of retina, no rods in the fovea itself.
- peripheral retina is important in low light conditions because of the
concentration of rods.
• Night vision functions only through peripheral parts of retina, starting outside the
• Day vision functions through all parts of the retina.
2. Visual Transduction
• Phototransduction: transformation of energy in stimulus (light) into neural activity
(photoreceptors) at the first stage of visual pathway.
• First, photoreceptors capture light photons to generate an electrical signal.
- accomplished by a special material - photopigment - found in rods & cones.
- photopigment in mammalian photoreceptors is called rhodopsin which are found in
the outer segment of the photoreceptors. The segment is embedded in fibrous matrix
of the eyeball which has a large concentration of rhodopsin molecules.
- George Wald responsible for this understanding.
• In darkness, rhodopsin molecules have an inactive form
• In darkness, steady flow of sodium ions (Na+) in the outer segment by cGMP-gated
- cGMP is an organic molecule highly concentrated in the outer segment.
• These sodium channels bind cGMP to allow sodium flow in meaning they are only
open if enough cGMP is around. This mechanism is called the dark current.
- the entry of + charged sodium ions into photoreceptors happens with an outward flow
of potassium ions (K+) in the inner segment which maintains a balance of ionic
- continued movement of the 2 ions leads to an excess accumulation of Na+ inside & K+
outside the cell.
- to prevent this: sodium-potassium pump ejects sodium ions from the photoreceptor &
brings back potassium ions.
- results in steady level of -40 mV of membrane potential.
- only occurs when rhodopsin remains inactive due to darkness
• Biochemical events of phototransduction:
1. Light protons pass by inner segments of photoreceptors
2. Reach stack of disks in outer segment containing rhodopsin molecules - the large # of
disks ensures that light particles are likely to be absorbed.
3. Energy in photon is imparted to a rhodopsin molecule causing a small change in its
4. The activated rhodopsin interacts with a G protein residing in the disc membrane
5. Activated G protein in turn activates a specific enzyme which convert cGMP within
the outer segment into ordinary GMP which are unable to bind to the sodium channel
(less available) - Cascade of biochemical events results in the removal of cGMP within the outer
segment causing ions channels to shut down = abolishing dark current = membrane
potential more negative since outflow of K+ without inflow of Na+
- photoreceptors become hyperpolarized after the absorption of light. Astonishing
result since usually after neuronal activation in other sensory systems, the receptors
• The greater amount of light is absorbed, the greater the hyperpolarization: membrane
potential can go to -70 mV under bright light conditions.
• Light activation (neural activation) of photoreceptors causes less neurotransmitter
release (again opposite effect in other sensory systems).
- rods & cones produce and release an excitatory neurotransmitter called glutamate.
Under dark conditions: steady release of glutamate, decrease in bright light conditions.
- the next level of neurons in the retina interprets the decrease in glutamate as signal of
activation of photoreceptors.
3. Spectral Relationships
• Amount of light absorbed by photoreceptor depends on 2 factors: intensity of light
• Human vision operates only when wavelength is between 400-700 nm.
• Spectrophotometry: procedure where we measure how much light is absorbed by
rhodopsin at each wavelength.
- Experiment where retinal tissue of nocturnal animals is collected. Tissue made up of
rods. After they’re separated from the other retinal neurons, rhodopsin molecules are
released & collected by the addition of mild detergent. Then, rhodopsin put in a test
tube where shown light with wavelength can be changed. Light that is not absorbed
emerges from the other side of the tube. Measured by detector.
- Results show a curve called absorption spectrum. Describes how much light is
absorbed by rod photopigments across entire visible spectrum.
- Upside-down U-shaped absorption profile shows different efficiencies where rod
rhodopsin molecules absorb light photons of different wavelengths.
- Display data in terms of relative absorption to make comparisons between
photopigments and conditions.
• Light absorption by rods is not equal across all wavelengths & shows a max at around
500 nm. Absorption drops a lot on either side of this peak. This suggests that rods
would produce their greatest signal at this wavelength which also means that vision
under nighttime conditions should be most sensitive to 500 nm.
• Beyond 650 nm; no appreciable absorption of light at all meaning vision at
wavelengths beyond this limit is accomplished by cone rhodopsin.
• Scotopic vision: visual processes that are mediated only by rod photoreceptors.
- Experiment: people have to detect a light spot whose wavelength is systematically
changed. Objective = find he absolute threshold of light detection for each
wavelength. Important to keep only rod vision tested (scotopic vision).
- Experiment has to be in very dim lights conditions and need to apply the light spot in
the peripheral part of the retina (many rods there).
- Typical result: scotopic spectral sensitivity function - visual system’s sensitivity
to light of different wavelengths. - The function is derived from experiments involving people meaning it reflects the
sensitivity of the visual system in behavioral terms.
• Scotopic vision is the most sensitive to bluish-green light and is not functional beyond
650 nm (red light).
• If light is not absorbed by rods beyond a certain wavelength, we should not be anle to
• Our sensitivity to different wavelengths under scotopic conditions can be explained by
the absorption spectrum of rod photopigments.
• Same sequence of experimental approaches applied to cones.
Microspectrophotometry - miniaturized procedure of spectrophotometry applied to
individual, isolated cones - found 3 types of cones, each containing a different type of
photopigments and absorption spectrum.
- The 3 types have a similar shape on graph but are offset.
• Types of cones:
1- Lowest wavelength absorbing cone “short-wavelength” or “S-cone” with peak
absorbance at 440 nm.
2- “Middle-wavelength” or “M-cone” with peak absorbance at 530 nm.
3- “Long-wavelength” or “L-cone” with peak absorbance at 560 nm.
- The rhodopsin in the 3 types is slightly different in molecular terms = responsible for
the differences in the peak absorbance at different wavelengths.
• Three types of cones form the photopic vision - day vision. Only from cones, no
input from rods.
• Brain is able to take separate output of the cone types to construct colour vision.
• Photopic spectral sensitivity function - same idea as scotopic but for day visions.
- Experiment: restrict small light patch to fovea - where only cones are present - and
find minimum intensity required for detection. Subject has to indicate when light is
visible. Detection threshold values are then collected at different wavelengths across
visible spectrum = creation of function.
- Peak around 550 nm suggesting light intensity detection at the behavioral level is
mainly regulated by M- & L-cones.
• Light sensitivity under photopic conditions is due to the light-capturing ability of
• Both scotopic & photopic are independent mechanisms.
4- Sensitivity, Bleaching, and Recovery
• Rods & cones are extremely sensitive of their absolute ability to capture photons.
They are able to reliably absorb a single light photon under ideal conditions.
• Rods more efficient than cones at converting photon absorption into neural signals
meaning that the rod activation threshold is lower allowing them to function at low light
- under these conditions, cones don’t produce neural signals = remain functionally
• Rods can only stay active for a limited range of light levels. Beyond moderate light
intensities, they can’t absorb photons since much of their photopigment is converted to
an inactive form. - after a rhodopsin molecule absorbs light, it temporarily becomes unable to absorb
more photons until the molecule is regenerated.
- this period is called “bleaching”. As the light increases, a greater proportion of
rhodopsin becomes bleached until additional light can’t be detected anymore.
- represents upper limit of of scotopic vision beyond which rods are blind.
• Cones become functional when rod vision tails off & thus the rods stay non-functional.
• Cone rhodopsin becomes increasingly bleached with increasing light levels. However,
they don’t become completely bleached under ordinary daylight conditions. Complete
bleaching occurs under extremely intense light conditions such as looking at the sun.
• Bleaching in cone photopigments = temporary blind spot in the fovea. However, they
have a rapid recovery from bleaching.
- Experiment: find actual time course for the recovery. Dark adaptation experiment -
detection threshold for a small light spot is measured at various times after bleaching.
• Cone profile: initially high threshold levels following bleaching rapidly decline (about 5
minutes). The detection threshold for cone vision is essentially at its lowest level.
• Rod profile: the lowest threshold level are reached only around 20 minutes (or more)
B) NEURAL PROCESSING IN THE RETINA
• Electrical signals are created in the photoreceptors and then processed by a network
of interacting neurons.
• Ultimate result of these interactions: produce final electrical output sent to visual brain.
1. The Retinal Ganglion Cell (RGC)
• Output neuron of the retina.
• Only neurons in retina whose axons leave the eyeball & carry electrical signals to
• First site in retina where action potentials are generated. Electrical activity in other
neurons comes from transient changes in the membrane potential spreading through
cell but never triggering an action potential.
• Site of emergence of parallel visual streams.
• Layout of retinal neurons: three layers.
- Outermost margins of retina: photoreceptors.
- Middle layer: inner nuclear layer - lying within the middle segment of the retina
- Inside margins of retina: ganglion cell layer
• Ganglion cells are located closest to the crystalline lens meaning incoming light
passes by the RGC first & lastly by photoreceptors.
• Each human retina has around 1.25 million ganglion cells.
• Anatomist Stephen Polyak: 1st detailed structural description of RGCs.
- several classes of RGCs differing in shape, size, & pattern of dendrites
- Two particularly important RGCs because of their # & major role in visual processing;
1- Midget ganglion cell characterized by its small size and compact dendritic field.
Represent 70% of ganglion cells. Found in high densities in & around the fovea.
2- Parasol ganglion cell characterized by its large size and dendritic field.
Represents 10% of total ganglion pool. - the axons of both cells form the major output of the retina through the optic nerve and
project to separate segments within the next structure of the visual pathway.
• Retinal ganglion cells generate action potentials even in the absence of light.
Phenomenon known as spontaneous activity.
- occurs around 20-50 action potentials/second.
- random firing of action potentials persists even when retina is flooded with uniform
light meaning that the very purpose of the retina - detect light - doesn’t affect the
electrical state of the ganglion cell since they continue to discharge at a spontaneous
• RGCs aren’t set up to signal overall light levels but are optimized to detect differences
in light stimulation in adjacent parts of the retina
- physiologist Stephen Kuffler: spontaneous firing of action potentials can go in 2
directions - increase (“ON”) or decrease (“OFF”) in activity.
- single ganglion cell can be excited (increase in action potentials) when light falls in
one part of retina & the same neuron can be inhibited (decrease in action potentials)
when the light spot is moved to surrounding parts.
• Receptive field: area of the retina that influences a neuron (excitation or inhibition).
- Kuffler found that every ganglion cell is affected by only a limited region of the retina
that is roughly a circular shape.
• The probing of receptive fields showed that the circular excitatory (“ON”) and inhibitory
(“OFF”) areas overlap each other.
• In one type of cell, a centre circular region is excited by light stimulation (“ON” area) &
the surrounding zone is inhibited by light (“OFF” area). By convention, this cell type is
called ON/OFF. Second type of cell = opposite & called OFF/ON.
- equal # of these cell types in the retina.
• The concentric spatial arrangement of the ON & OFF sub-regions in the receptive field
leads to centre-surround antagonism - 2 fields oppose each other of how they
influence the firing rate of ganglion cells.
• When light spot large enough to cover both parts (ON & OFF) - the cell will fire at its
spontaneous rate since the excitation through its centre is counterbalance by the
inhibition surround. This explains why uniform diffuse light doesn’t lead to increased
neural activity in RGCs.
- retinal output sent to the brain by ganglion cells is driven by light contrast. Best way
to segment an image (light contrast).
- an ON/OFF cell will maximally fire to a light spot precisely filling its center excitatory
region in the presence of a dark surround.
- an OFF/On cell will maximally fire to a dark spot filling its centre in the presence of a
2. Electrical Circuits in the Retina
• Absorption of light by photoreceptors - generation of electrical signal - process of
signal through intermediate set of neurons in the inner nuclear layer - ganglion cells.
•The electrical circuity within the set of intermediate-level neurons is what endows
ganglion cells with their properties.
• Until now saw 2 major retinal cell classes: photoreceptors & ganglion cells. •3rd major class: bipolar cells - transmit signals between photoreceptors and
• A cone photoreceptor is connected to 2 different bipolar cells: ON bipolar & OFF
• Sequence of event when light falls on a cone:
- absorbs light; brief hyperpolarization, drop in the photoreceptor’s release of the
- reduction in glutamate release = odd effect on ON bipolar cells: becomes depolarized
(stimulated). Unexpected result since glutamate is an excitatory neurotransmitter
usually associated with postsynaptic depolarization (like stimulating the next neuron).
- reduction in glutamate release does NOT stimulate the ON bipolar cell since they
have the property of becoming spontaneously depolarized in the absence of
glutamate. Glutamate turns off the spontaneous depolarization. Thus, light ends up
stimulating the ON bipolars by an unconventional manner.
- OFF bipolars are not depolarized in the presence of light but are stimulated when the
light turns off. In this case, photoreceptors aren’t hyperpolarized & return to a state
where more glutamate is released.
- OFF bipolars behave like conventional neurons; increased glutamate =
depolarization. Therefore, they are linked to photoreceptors stimulated only when
they are not absorbing light.
• The 2 types of bipolar cells are linked to a separate ganglion cell.
- when an ON or OFF bipolar is stimulated it will release glutamate stimulating the
ganglion cell it is connected to.
- meaning there are 2 different types of ganglion cells: the ones stimulated by a light
spot on an ON bipolar and the ones stimulated by a dark spot on an OFF bipolar =
centre response property of ganglion cells.
• Ganglion cells behave as typical neurons: sufficient amount of depolarization will
create action potentials that are transmitted through their axons.
• Two other types of intermediate neurons between photoreceptors and ganglions to
generate surround responses (as opposed to center responses described by previous
- horizontal cells: display processes that branch out in a lateral manner & make
contact with adjacent photoreceptors. Their lateral sampling allows them to gather
signals of a large area of the retina = role in generating surround response. Light
absorption signals in the surrounding photoreceptors are passed into horizontal cells
which inhibit the photoreceptors linked to the center response. The firing rate of
ON/OFF ganglion cells is reduced if a light spot falling on its excitatory center is large
enough to overlap the inhibitory surround.
- amacrine cells: make lateral connections at the level of bipolar and ganglion cells.
• These 2 intermediate neurons are responsible for one of the most fundamental
property of the nervous system: lateral inhibition - through this mechanism,
ganglion cells inherit the property of centre-surround antagonism & become endowed
with a receptive field optimized for the detection of light contrast.
• Sequence of events when light falls on a rod:
- similar to cone circuit - rod signals are also transmitted to bipolar cells and there are similarities with in how
horizontal cells mediate the lateral interactions.
- difference: no ganglion cells that are specifically dedicated to transmitting rod signals
out of the retina. The signals merge into the cone pathway and arrive to the same set
of ganglion cells.
- the circuit involves a specific type of amacrine cells to connect both pathways.
- as a result: scotopic visual functions are handled by the same set of ganglion cells
used by the photopic system.
3. Visual Processing Across the Retina
• Cellular components of the retina have a fixed spatial relationship that accounts for its
uniform thickness: 250 micrometers.
• The thickness at the center of the retina is less because of the foveal pit (small
• The pit arises because the neurons are shifted to one side allowing light rays to have
better access to photoreceptors.
• The quality of the image is enhanced at the fovea since there is less chance of light
distortion than in other parts of the retina where light rays have to pass by other
neurons before reaching the photoreceptors. It’s one reason why the fovea is able to
process fine details in an mage.
• Main reason to see fine details in the fovea: way that neurons are wired.
- a single photoreceptor (light detector) connected to a single bipolar cell connected to
a single ganglion cell (output neuron) = spatial sampling.
- the brain is therefore receiving signals by dedicated electrical wires for the smallest
possible patch of the retinal image that can be sampled.
- ability for fine details in the fovea because no information is lost. Only place in the
retina with a one-to-one relationship between photoreceptors and ganglion cells (not
enough ratio between the 2 in the whole retina to maintain this relationship
- solution to the ratio problem = signal convergence - signals from multiple
photoreceptors converging upon a single ganglion cell. Happens outside the fovea.
• The greater the retinal eccentricity, the greater the number of photoreceptors signals
feeding into a ganglion cell. 2 main aspects of this eccentricity:
- the dendritic field of the ganglion cells must increase with eccentricity to allow them to
receive signals from more bipolar cells. Meaning more dendrites in peripheral area,
less in the fovea.
- the receptive field of ganglion cells must show systematic changes with retinal
eccentricity. Expected since the receptive field is defined by the extent of the retina
that influences a visual neuron. Smallest receptive fields are found in the fovea.
• Midget & parasol ganglion cells show different sampling properties:
- since midget ganglion cells have a smaller dendritic field than parasol ganglion cells
at all retinal eccentricities, anywhere in the retina a parasol ganglion cell will have a
larger receptive field than a midget cell.
- This sets the stage for the emergence of 2 distinct & separate signal pathways to the
brain. C) PERCEPTUALASPECTS OF THE RETINAL FUNCTION
• Many aspects of visual perception can be explained in terms of retinal function since
they are regulated by the properties of ganglion cells - especially their predispositions
for contrast detection.
- such fundamental aspects of visual perception explained by retina can be how well
we can see spatial detail (resolution) and how little light is required for a visual
• The perceptual phenomena linked to biological aspects of the retina include two main
factors: the nature of the photoreceptor function (rod vs cone) & the way their signals
arrive to the ganglion cells through an intermediate plexus of retinal neurons.
- both factors impact the output signals produced by ganglion cells which accounts for
a range of perceptual properties referred to as early vision.
• Ability to detect light stimuli
• General rule: trade-off between sensitivity and resolution - the conditions under which
sensitivity is enhanced represent the conditions under which resolution is reduced.
- example: scotopic system shows greater light sensitivity than photopic vision
because of the lower activation threshold of rod photoreceptors. The superior
sensitivity is offset by a poorer ability of the scotopic system at resolving image detail
- something the photopic system is good at.
• Selig Hecht, Simon Schlaer, & Maurice Pirenne: first detailed experiment on light
- tested visual sensitivity under optimal conditions - complete dark adaptation - placing
the light spot in the peripheral retina with the highest density of rods using a stimulus
wavelength close to the peak of the rod absorption spectrum.
- absolute detection threshold = 90 photons on the surface of the cornea.