# The Retina

> Light quiets each photoreceptor, but one receptor cannot separate wavelength from intensity, so the retina's circuit compares signals to report lighter or darker, edges and colour. You follow the signal from the photoreceptor through bipolar, horizontal and amacrine cells to the ganglion cells that send spikes to the brain. Afterwards you can explain why ON and OFF cells exist, build a centre-surround receptive field from the circuit, compute photon statistics from a rod recording, and say why colour needs more than one kind of cone.

Course: Neuroscience (Sensory systems). Level: university elective. About 34 minutes.
Lesson page (link this one when citing): https://stemhop.com/lessons/retina/

Builds on: The Resting Membrane and Passive Signals · Principles of Sensory Coding · Synaptic Transmission and Quantal Release

## What you'll learn

- Trace phototransduction from the isomerization of 11-cis retinal through transducin, PDE6 and falling cGMP to the closure of CNG channels, the hyperpolarization of the photoreceptor and the fall in glutamate release; explain with the balance point why closing a cation channel makes the cell more negative; and explain calcium feedback through GCAPs.
- Follow a signal through photoreceptors, bipolar, horizontal, amacrine and ganglion cells, and say which of them signal with graded potentials and which fire action potentials.
- Explain ON and OFF centre-surround receptive fields from the circuit: sign inversion by mGluR6 and TRPM1 in ON bipolar cells, sign preservation by AMPA and kainate receptors in OFF bipolar cells, and surrounds built by horizontal and amacrine cells.
- Compare rods and cones in sensitivity, speed and convergence, trace the AII rod pathway, and explain colour as a comparison across three cone classes.

**A rod sitting in the dark is depolarized, with current flowing steadily into it, and it releases glutamate all the time. Light turns that current and that release down.**

Most sensory receptors are excited by their stimulus. Rods and cones are turned down by theirs. The retina then reads that drop in two opposite ways, compares each spot of the image with its neighbours, and sends the brain spikes through many parallel types of ganglion cell. Most report contrast, colour or motion more than raw brightness, though a few, such as melanopsin ganglion cells, track overall light level. This lesson follows the signal from the photon to the optic nerve.

## The layout

Light enters the eye and passes through the transparent inner retina, including the ganglion cells, before it reaches the photoreceptors at the back. The signal then runs the other way. Hold three lines in mind.

*Figure (AI-generated illustration).* Light crosses the inner retina to reach the rods and cones at the back, and the signal runs back through bipolar cells to the ganglion cells, with horizontal and amacrine cells working sideways. Image: a section through the retina with light entering from the top and passing through ganglion, amacrine, bipolar and horizontal cell layers to the rods and cones at the back, while an arrow shows the signal running back up to the ganglion cells.

- **The through path.** Photoreceptors (rods and cones) drive **bipolar cells**, which drive **ganglion cells**. Ganglion cell axons gather at the optic disk (the blind spot) to form the optic nerve.
- **The sideways cells.** **Horizontal cells** spread across many photoreceptor terminals in the outer retina. **Amacrine cells** connect bipolar and ganglion cells in the inner retina.
- **The signals.** Photoreceptors, horizontal cells and most bipolar cells signal with **graded potentials**, voltage changes that vary smoothly with input and need no spike over these short distances (see [synaptic transmission](https://stemhop.com/lessons/synaptic-transmission/) for graded release). Ganglion cells fire action potentials. **The ganglion cells are the retina’s only output.**

The layout raises four puzzles.

1. **Why does light make a photoreceptor more negative?**
2. **How can one transmitter** make some bipolar cells report “lighter” and others “darker”?
3. **Why do ganglion cells answer a small spot** better than light spread evenly over the whole retina?
4. **How can colour come from cells** that cannot tell one wavelength from another?

> **The one idea to carry through this lesson.** One receptor cannot separate wavelength from intensity; the circuit compares. Each rod or cone turns the photons it absorbs into a graded signal, a fall in its glutamate release below the dark level, and that signal carries no label for which wavelengths the photons had. The classic pathways build their reports by comparing these signals: the bipolar cell’s receptor sets the sign of the comparison (lighter or darker), the surround subtracts the neighbourhood (where), and opponent circuits compare cone classes (what colour).

**What the rule covers.** It describes the classic ON and OFF centre-surround pathways over their roughly linear range. Three limits matter. First, the subtraction is usually partial: in this lesson’s examples the surround is weaker than the centre, and how much it cancels varies between cell types and light levels. Second, a receptor’s signal is a noisy time course, not a tally of photons: the same number of photons arriving at different times need not give the same response, and adaptation (below) changes the response to each photon. Two lights are confused only when they give the same effective catch with the same timing, under the same adaptation. Third, a cone’s release is already shaped by horizontal-cell feedback from its neighbours, the circuit’s first comparison. A few ganglion cell types need more than this rule (see Going deeper).

## Light turns the photoreceptor down

### The dark current

In darkness, cGMP made by guanylyl cyclase holds open **cyclic-nucleotide-gated (CNG) channels** in the outer segment. Sodium and calcium flow in, the inward **dark current**, which leaves again at the inner segment in a steady loop. It keeps the cell depolarized and releasing glutamate at its terminal.

### Why closing channels makes the cell more negative

Depolarized compared with what? Use the rule from [the action potential](https://stemhop.com/lessons/action-potential/): in a sketch with fixed conductances and no current injected, the membrane potential moves toward a balance point, the average of the reversal potentials (the voltage where each channel’s current is zero), weighted by their conductances.

Two conductances matter most in a photoreceptor.

- The **CNG channels** in the outer segment pass cations, reverse near 0 mV and pull the voltage up toward it.
- **Potassium channels** in the inner segment pull the voltage down toward EK, about −80 mV in this sketch.

In the dark the two pulls are comparable; take the conductances as equal, a convenient sketch rather than a measurement. The balance point then sits halfway, at about −40 mV, the textbook dark potential of a photoreceptor.

> **Vm = (gCNG × 0 mV + gK × EK) / (gCNG + gK)**

**Predict.** Light closes half the CNG channels, and the potassium conductance stays the same. Which way does the balance point move?

> **Check your prediction.** Toward EK, so the cell hyperpolarizes. With half the CNG conductance left, the sketch gives (0.5 × 0 + 1 × −80) / (0.5 + 1) ≈ −53 mV. Close all the CNG channels and the sketch gives −80 mV. Real rods reach about −65 to −70 mV in bright light, because other conductances remain. A less depolarized terminal releases less glutamate.

So closing a channel can make a cell more negative: removing the inward cation pull leaves potassium in charge. The response is graded: dim light closes a few channels, brighter light more. Only saturating light shuts the whole dark current.

### The cascade

What closes the channels? A five-step chain, each step pushing the next.

1. **A photon** absorbed by rhodopsin (in rods) or a cone opsin isomerizes its chromophore, the light-absorbing retinal, changing its shape from 11-cis to all-trans. The pigment is now active.
2. **The active pigment switches on the G protein transducin.** One activated pigment activates many transducins, which is where much of the amplification, or gain, comes from.
3. **Transducin activates PDE6**, a phosphodiesterase.
4. **PDE6 hydrolyzes cGMP**, so cGMP falls.
5. **CNG channels close.** The dark current shrinks, the photoreceptor hyperpolarizes, and glutamate release falls.

The molecular steps belong to [senses and cGMP](https://stemhop.com/lessons/senses-and-cgmp/) and [G proteins](https://stemhop.com/lessons/gpcrs-and-g-proteins/). Shutoff, of rhodopsin by rhodopsin kinase and arrestin and of transducin by RGS9-1, is in [turning signals off](https://stemhop.com/lessons/turning-signals-off/).

Calcium closes the loop. As the channels close, less calcium enters while extrusion continues, so outer-segment calcium falls. Low calcium lets **GCAPs** (guanylyl cyclase-activating proteins) stimulate the cyclase, which remakes cGMP and speeds recovery. In steadier, brighter light the same feedback turns the gain down, part of **light adaptation**, the photoreceptor’s adjustment to the prevailing light. So the response to each photon depends on recent light.

One more fact gives the first half of the rule. **Once absorbed, every photon has the same effect, whatever its wavelength.** The response keeps no record of which wavelengths were caught; the colour section puts this to work.

## Counting photons

How small can the signal be? Picture the record that answered it.

Baylor, Lamb and Yau (1979) drew the outer segment of a single toad rod into a snug glass pipette, a **suction electrode**, so the current entering the outer segment had to pass through the pipette. In darkness the trace is flat: the steady dark current. A very dim flash, repeated many times at the same strength, gives a scatter of results. Some flashes give nothing. Many give a dip of about the same size. The responses do not vary smoothly: a histogram of their sizes has a peak at zero and a peak near 1 pA. So the responses come in discrete units, or **quanta**. Each unit, with a mean size of about 1 pA, is the current suppressed by one photoisomerization, the activation of one pigment molecule by one photon.

**Predict.** Double the strength of the dim flash. Does each response double in size, or do more trials respond?

> **Check your prediction.** More trials respond, and more of them carry two events instead of one. The unit stays about 1 pA, because each event is still one photoisomerization. The mean response doubles. Single events scatter in size around 1 pA and the recording adds noise, but each trial holds a whole number of events, never a fraction of one. That is what quantal means.

**Worked problem: counting photons.** Take a toad rod, suction-electrode recording, and dim flashes that give on average m = 0.5 photoisomerizations per rod. Assume photon capture is random and independent, so the chance that a flash gives exactly k events follows a Poisson distribution, P(k) = mk × e−m / k!, and that each event gives 1 pA and events add linearly.

*Step 1: failures, the chance of zero events.* P(0) = e−0.5 = 0.607. About 61% of flashes give no response at all.

*Step 2: singles.* P(1) = 0.5 × e−0.5 = 0.303.

*Step 3: doubles or more.* P(2 or more) = 1 − 0.607 − 0.303 = 0.090.

*Step 4: the mean.* Mean response = 0.5 events × 1 pA = 0.5 pA, though no single trial gives 0.5 pA; in this ideal model each gives 0, 1, 2 pA and so on.

Run it backwards and the failure rate gives the mean: m = −ln P(0). This is the same Poisson logic used for quantal release at a synapse. It is the current of one toad rod in vitro, not a statement that a person sees a single photon. Dim-flash rod recordings resolve single photoisomerizations, but the fixed 1-pA event is an idealization, and a cone cannot reliably signal single photons through its noise.

## One transmitter, two signs

**Film transcript: One Transmitter, Two Signs**

### Film transcript: One Transmitter, Two Signs

Photoreceptors release one transmitter, glutamate. More in the dark, less in the light. Yet some bipolar cells depolarize to light, and others hyperpolarize. How? Picture one cone terminal feeding two bipolar cells. On the left, darkness. On the right, light. The OFF bipolar cell has ionotropic glutamate receptors. Glutamate opens their cation channels, so in the dark the cell is depolarized. Light lowers glutamate, so the excitation fades, and the cell hyperpolarizes. The sign is kept. The ON bipolar cell uses a different receptor, a metabotropic one. In the dark, glutamate on this receptor activates a G protein. That G protein holds a cation channel shut.

So in the dark, the ON cell is hyperpolarized. Light lowers glutamate. The G protein's activity falls, the brake comes off, and the channel opens. The cell depolarizes. The sign is inverted. Keep a ledger of the flips. The photoreceptor flips the sign once. The OFF cell adds no flip. The ON cell adds a second. An even number of flips means the cell responds like light. That makes it an ON cell. How do we know the receptors differ? Slaughter and Miller recorded inside cells of the mudpuppy retina. They bathed the retina in a glutamate mimic that acts on the ON receptor. Add the drug, then turn on a light. What does the ON cell do? And the OFF cell?

The ON responses vanished. The drug held the ON cells in their dark state, so light had nothing to remove. The OFF responses carried on, in the same retina. So the two cells read glutamate with different receptors. Each cone feeds two streams, one for lighter, one for darker. The takeaway: the receptor, not the molecule, sets the sign.

From the film

- Glutamate's sign is set by the receptor, not the molecule: OFF bipolar cells use ionotropic receptors and keep the sign, while ON bipolar cells use mGluR6 and invert it.
- Counting the sign inversions from the photoreceptor shows which cells are ON: an even number means the cell responds like light.
- In mudpuppy retina, a glutamate analogue abolished ON responses and spared OFF responses, showing that the two cell types read glutamate with different receptors.

Photoreceptors release one transmitter, glutamate, more in the dark and less in the light. Yet some bipolar cells depolarize to light and others hyperpolarize. Glutamate’s sign is set by the receptor, not the molecule ([transmitter systems](https://stemhop.com/lessons/transmitter-systems/) covers receptor classes).

**OFF bipolar cells** use ionotropic AMPA or kainate receptors. Glutamate opens cation channels and depolarizes them in the dark; light removes that excitation and they hyperpolarize. The sign is preserved.

**ON bipolar cells** use the metabotropic receptor **mGluR6**. In the dark, glutamate-bound mGluR6 activates the G protein Go, which keeps the cation channel **TRPM1** closed. Light lowers glutamate, Go activity declines, TRPM1 opens and the cell depolarizes. The sign is inverted.

*Figure (AI-generated illustration).* One transmitter, two signs: when light lowers a cone's glutamate release, OFF bipolar cells hyperpolarize, keeping the sign, and ON bipolar cells, whose TRPM1 channels open, depolarize, inverting it. Image: a cone terminal in light releasing little glutamate onto two bipolar cells: the OFF cell, reading glutamate through AMPA or kainate receptors, hyperpolarizes, while the ON cell, whose mGluR6 and Go no longer hold TRPM1 shut, depolarizes.

Two decreases and then an opening are easy to lose, so keep a **sign ledger**, following one flash from the photons to each bipolar cell.

| Step | In light | Sign so far |
| --- | --- | --- |
| Photons on the cone | more | + |
| Cone voltage | hyperpolarizes | − (first flip) |
| Glutamate release | falls | − (follows the voltage) |
| OFF bipolar (AMPA or kainate: glutamate opens cation channels) | less excitation, hyperpolarizes | − → **OFF** (no flip) |
| ON bipolar (mGluR6 → Go keeps TRPM1 shut) | Go activity falls, TRPM1 opens, depolarizes | + → **ON** (second flip) |

A rule of thumb: count the inversions. The photoreceptor gives one. The OFF bipolar cell adds none, so it responds like darkness. The ON bipolar cell adds a second, and an even number means the cell responds like light.

### The evidence: one drug, one pathway

How do we know the two cells read glutamate with different receptors? Slaughter and Miller (1981) worked with the retina of the mudpuppy, a salamander with large retinal cells, kept as a perfused eyecup so they could change the bathing solution while recording intracellularly. They applied APB, a glutamate analogue that mimics glutamate at the ON receptor.

**Predict.** Apply APB, then turn on a light. What does an ON bipolar cell do? An OFF bipolar cell?

> **Check your prediction.** ON responses were abolished and OFF responses were spared. Acting like glutamate at the ON synapse, APB holds ON bipolar cells in their dark state, so light has nothing to remove. OFF cells read glutamate through other receptors and respond as before. The OFF responses in the same retina, and recovery after washout, are the controls that show a receptor effect rather than a damaged retina.

APB shows only that the ON pathway uses a distinct receptor; it does not name the receptor or the channel (the design problem at the end takes that step).

So each cone feeds two parallel streams, one signalling “lighter” and one “darker”. Rods reach both streams by a different route, described below. ON bipolar cells drive ON ganglion cells, and OFF bipolar cells drive OFF ganglion cells.

## Centre minus surround

**Film transcript: Centre Minus Surround**

### Film transcript: Centre Minus Surround

Why does a ganglion cell answer a small spot better than light spread over the whole retina? In 1953, Stephen Kuffler recorded single ganglion cells in cats. He shone small spots of light onto the retina. A spot in the middle of the field made one cell fire faster. A spot on a ring around the middle made the same cell fall silent. When the light went off, it fired. That is an ON-centre cell. An OFF-centre cell does the mirror image. The centre comes from the bipolar cells that contact the ganglion cell. The surround comes from cells that act sideways. Horizontal cells do this in the outer retina, and amacrine cells add more in the inner retina. Both push the response the opposite way from the centre.

Now light the whole field, centre and surround together. More spikes than the best spot, fewer, or no change from the background? In this example, fewer: 45 spikes a second instead of 60. Its surround cancels only part of the centre. So the cell reports contrast, a spot brighter than its surroundings, more than overall light level. Now line up a row of ON-centre cells across an edge, where light meets dark. The cell right on the edge is not the winner. In this model, cells just on the bright side fire most. Their surrounds reach into the dark and subtract less. Cells just on the dark side fire least. The edge stands out across the population.

The same logic gives colour. One cone cannot tell a change in wavelength from a change in intensity. Compare two cones with overlapping curves, and their ratio changes with wavelength. Turn the light up, and the ratio holds. But in pigment models the ratio can repeat. Set two wavelengths to the same L catch, and their M catches match too. The S cone catches far more of the shorter one, so it tells them apart. Here is the whole rule. One receptor cannot separate wavelength from intensity. The circuit compares. The takeaway: ganglion cells send those comparisons to the brain as spikes.

From the film

- A ganglion cell has a centre and an opposing surround, from bipolar cells and from lateral horizontal and amacrine cells, so it reports contrast more than overall light level.
- In the lesson's linear example, light on both centre and surround gives fewer spikes than the best centre spot, because the surround subtracts only part of the centre.
- Across an edge, in a linear model, ON-centre cells just on the bright side fire most, so the edge stands out across the population.
- Colour comes from comparing cones with overlapping sensitivity curves, and a third cone class breaks ties that two classes would confuse.

### Kuffler’s spots

A ganglion cell’s **receptive field** is the patch of retina where light changes its firing. Stephen Kuffler (1953) recorded single ganglion cells in cats while he projected small spots of light onto the retina.

- **A spot in the middle of the field:** the cell fires faster.
- **A spot a little farther out, on a ring around the middle:** the same cell falls silent, then fires when the light goes off.

That is an **ON-centre** cell: light on the centre excites it, and light on the ring around it, the **surround**, suppresses it. An **OFF-centre** cell does the mirror image: light in the centre suppresses it and light in the surround excites it.

### Where centre and surround come from

The centre comes from the bipolar cells that contact the ganglion cell’s dendrites. The surround comes from lateral cells. Horizontal cells, which collect from many photoreceptors, feed back onto cone terminals, so light in the surround pushes the centre cones’ release the opposite way from light on the centre. How that feedback works is still debated (see Going deeper). Amacrine cells add further surround inhibition in the inner retina.

**Predict.** Light the whole receptive field of an ON-centre cell, centre and surround together. Does it fire more than for the best centre spot, less, or no more than on the grey background?

> **Check your prediction.** Less than for the spot, but more than on the background: in this lesson’s example the surround subtracts but is weaker than the centre. The worked problem below puts numbers on it.

**Worked problem: reading a receptive field.** In a made-up recording, an ON-centre ganglion cell fires 20 spikes/s on a uniform grey background. A spot filling the centre raises it to 60 spikes/s; an annulus over the surround alone lowers it to 5 spikes/s. Assume the two effects add linearly.

*Step 1.* Centre effect = 60 − 20 = +40 spikes/s.

*Step 2.* Surround effect = 5 − 20 = −15 spikes/s.

*Step 3.* Light covering both: 20 + 40 − 15 = 45 spikes/s, less than the spot’s 60.

The cell reports contrast, a spot brighter than its surroundings, more than overall light level. In this linear model, uniform light would leave the cell at its 20 spikes/s background only if the surround cancelled the centre exactly. Here the net change is 40 − 15 = +25 spikes/s. An OFF-centre cell would give the mirror image, firing best to a dark spot on a light background.

Grow a spot from tiny to the full field, and in a standard model of the field the response rises, peaks for a spot about the size of the centre, then falls as the spot spreads into the surround. With this example’s weaker surround, it settles above the background rate rather than returning to it.

### Edges stand out across the population

What about an edge, where light and dark meet? Picture a row of ON-centre cells lined up across it. A cell centred exactly on the edge does not fire the most. In a linear model with a symmetric field, half its centre and half its surround are lit, so it fires halfway between its rates for uniform bright and uniform dark: its change from the uniform-dark rate is half the full bright-field change. (If bright and dark sit equally above and below a grey background, its change from the grey rate is zero.) **Cells whose centres sit just on the bright side of an edge fire more than cells in the middle of a bright area**, because their surrounds reach into the dark and subtract less. ON-centre cells just on the dark side fire least of all. OFF-centre cells show the mirror image: they fire most just on the dark side and least just on the bright side. [Sensory coding](https://stemhop.com/lessons/sensory-coding/) treats receptive fields in general; [visual cortex](https://stemhop.com/lessons/visual-cortex/) follows the signal through the optic chiasm.

### Try it

Start with a small spot on the centre and enlarge it. Watch the rate rise, peak and fall. Then try the annulus alone, then the full field. Switch to an OFF-centre cell, which the simulation tests with dark spots and rings, and repeat.

*Interactive on the lesson page:* Centre and surround — one ganglion cell’s receptive field

## Rods borrow the cone circuits

Rods are more sensitive than cones and respond more slowly, trading temporal resolution for sensitivity. Human cone responses to dim flashes peak in 50 to 100 ms (Schnapf, Kraft and Baylor, 1987). Many rods also converge onto each downstream cell, pooling scarce photons at the cost of spatial detail. In a human retina there are about 20 rods for every cone, and rods make up about 95% of photoreceptors (the counts are in Going deeper). Cones dominate the fovea, the central pit of the retina used for fine detail, and rods are absent from its very centre.

Rods have no ganglion cells of their own. In the main mammalian rod circuit, their signal joins the ON and OFF lines that the cones use. Following one dim flash through five stops is a good test of the sign ledger.

1. **Rod.** Light hyperpolarizes the rod, and its glutamate release falls.
2. **Rod bipolar cell.** Rod bipolar cells read glutamate through mGluR6, so they are ON cells. Less glutamate means they depolarize.
3. **AII amacrine cell.** In the classical primary pathway, rod bipolar cells do not contact ganglion cells directly. Their glutamate excites a narrow-field amacrine cell called the **AII**, which depolarizes and relays the signal into the cone pathways. Their ribbon synapses also commonly contact A17 amacrine cells, which feed inhibition back. Many rods feed each rod bipolar cell, and many rod bipolar cells feed each AII: this is where the pooling happens.
4. **Into the cone circuits, two ways.**
  - To **ON cone bipolar cells** through **gap junctions**. Current passes straight through, so the ON cone bipolar cell depolarizes too. The sign is kept.
  - To **OFF cone bipolar cells** through **glycine synapses**. Glycine is inhibitory, so the OFF cone bipolar cell hyperpolarizes. The sign is inverted. AII cells also inhibit some OFF ganglion cells directly.
5. **Ganglion cells.** The cone bipolar cells drive ON and OFF ganglion cells as they do in daylight. In dim light, ON ganglion cells fire more and OFF ganglion cells fire less, the same pattern that cone signals produce.

| Stop | Response to dim light | Sign |
| --- | --- | --- |
| Rod | hyperpolarizes | − |
| Rod bipolar (mGluR6) | depolarizes | + |
| AII amacrine | depolarizes | + |
| ON cone bipolar (gap junction) | depolarizes | + |
| OFF cone bipolar (glycine) | hyperpolarizes | − |

So one ganglion cell can carry rod signals at night and cone signals by day, and the brain receives ON and OFF reports either way. This is the primary rod pathway, mapped synapse by synapse in rabbit retina and found across mammals; rods also reach cone circuits by secondary routes (Going deeper).

**Predict.** A drug blocks glycine receptors. In very dim light, which ganglion cells lose most of their rod-driven response, ON or OFF?

> **Check your prediction.** OFF cells. In the circuit as drawn, the AII reaches OFF cone bipolar cells, and some OFF ganglion cells directly, through glycine synapses. Its route to ON cone bipolar cells is a gap junction, which needs no transmitter. Some rod signal can still reach OFF cells by the secondary routes, especially at the brighter end of dim light.

*Figure (AI-generated illustration).* In the primary rod pathway, as mapped in rabbit retina, many rods feed one rod bipolar cell, and the AII amacrine cell passes the signal to the cone ON line by a gap junction and to the OFF line by a sign-inverting glycine synapse. Image: four rods converge on a rod bipolar cell that drives an AII amacrine cell, which passes the signal to an ON cone bipolar cell through a gap junction and to an OFF cone bipolar cell through a glycine synapse, each feeding its own ganglion cell.

## Colour is a comparison

A single cone cannot report colour. Wavelength changes only the probability that a photon is caught, and a change in intensity can mimic that. This is the **principle of univariance**: given the same timing and the same adaptation, an M cone responds to a dim light near its peak and a brighter light far from it in ways it cannot tell apart (the same on average; each trial is noisy). In standard pigment templates, an L cone, with its peak near 560 nm, needs about 2.6 times as many photons at 620 nm as at 560 nm to give the same catch.

Colour comes from comparing cones with different, broad sensitivity curves. Humans with normal colour vision have three classes: short-wavelength (S), medium (M) and long (L). Schnapf, Kraft and Baylor (1987) recorded from a small number of human cones and found M and L sensitivity peaks near 530 and 560 nm. The two curves overlap heavily, so it is the ratio of M to L catch (photons absorbed) that varies with wavelength. Turn a light up or down and both catches change together, so the ratio holds: that is the comparison doing what a single cone cannot. **Colour-opponent** circuits read such comparisons; a cell excited through L cones and inhibited through M cones, for example, signals the balance between their catches. That balance is a difference, not a ratio, so it still grows as the light gets brighter; adaptation and later normalization reduce, but do not remove, its dependence on intensity.

**What does the third cone add?** Two cone classes already give some colour vision: people with only two (dichromats) see colours but confuse some lights that others tell apart. In standard pigment templates the M/L ratio does not change steadily with wavelength. It rises to a peak near 466 nm and then falls, so a light near 427 nm and a light near 529 nm give the same M/L ratio. Adjust their intensities so their L catches match, and their M catches match too, so M and L cones alone would confuse them. The S cone breaks the tie: it catches far more at 427 nm than at 529 nm.

In dim, mesopic light, between night and daylight levels, rods also interact with cone signals and can shift perceived colour.

So the rule from the start holds to the end: **one receptor cannot separate wavelength from intensity; the circuit compares.** Ganglion cells send those comparisons to the brain as spikes.

Where next: how the retina’s maps in the brain are refined during development, including by spontaneous retinal waves before eye opening, is covered in [spontaneous activity](https://stemhop.com/lessons/spontaneous-activity/) and [topographic maps](https://stemhop.com/lessons/topographic-maps/).

### Design problem

Slaughter and Miller’s APB result shows that the ON pathway uses a distinct receptor, but not which channel ON bipolar cells open to light. Design experiments to name it.

> **One way to do it.** Start from what APB means. Because it holds ON bipolar cells in their dark state, the channel must be one that glutamate’s receptor keeps shut and light allows to open. Keep the same controls: OFF responses in the same retina, and recovery after washout.
>
> Next, test a candidate channel genetically. Morgans and colleagues (2009) compared TRPM1-deficient mice with controls by electroretinogram. The a-wave, mainly photoreceptor current, was normal; the b-wave, mainly ON bipolar activity, was absent. The defect lies after photoreceptor activation.
>
> The ERG sums the whole retina, so localize the defect: record whole-cell from rod bipolar cells in retinal slices from knockout and control mice. Expect no light-evoked or mGluR6-dependent current in the knockout, with normal responses in OFF bipolar cells.
>
> Then rescue: restore TRPM1 in ON bipolar cells of the knockout and look for the current and the b-wave to return.
>
> Two caveats. A germline knockout can act during development, for example by changing synapse formation, so a missing current might not mean the channel itself is gone from a normal circuit. Check that the synapse is still built, or delete the gene only in adults. And loss of function shows the channel is needed; expressing TRPM1 with mGluR6 and Go in another cell would test whether the parts are enough.

## Going deeper

### Cells that break the signalling pattern

Most bipolar cells signal with graded potentials, but some fast mouse bipolar types also spike at their terminals. Some amacrine cells fire action potentials too. Neither changes which cells carry the output.

### How horizontal-cell feedback works

In [whole-cell recordings](https://stemhop.com/methods/patch-clamp/) from macaque cones, Verweij, Hornstein and Schnapf (2003) saw surround antagonism when surrounding cones were lit, and found it inconsistent with a GABA-gated chloride conductance. Which of the other proposed feedback mechanisms dominates, and in which species, is still unsettled.

### Beyond adding and subtracting

The lesson’s rule does not cover every ganglion cell. Melanopsin ganglion cells have their own pigment (see [intrinsically photosensitive ganglion cells](https://stemhop.com/lessons/iprgcs/)). Direction selectivity needs more than adding and subtracting signals (see [computing motion in the retina](https://stemhop.com/lessons/direction-selectivity/)).

### What the third cone class adds, in numbers

The colour section’s numbers, including the 2.6 factor, come from the A1 pigment template of Govardovskii and colleagues (2000), including its beta band, with peaks at 420, 530 and 560 nm, counting photons absorbed by the pigment and ignoring the lens and macular pigment. The matched M/L ratio at 427 and 529 nm is about 1.2. The S/L ratio is about 6 at 427 nm against about 0.001 at 529 nm. At any one wavelength the lens and macular pigment dim all three cone classes alike, so they change how much light must enter the eye, not the ratios between cones.

### How many rods?

Curcio and colleagues (1990) counted photoreceptors in eight whole-mounted adult human retinas from seven people aged 27 to 44. The means were 92 million rods and 4.6 million cones per retina.

*Step 1.* Rods per cone = 92 / 4.6 = 20.

*Step 2.* Rods as a share of photoreceptors = 92 / (92 + 4.6) = 92 / 96.6 = 0.952, or 95.2%.

These are ratios of sample means, and they exclude every other retinal cell: 95.2% of photoreceptors are rods, not 95% of the retina.

### The AII hub in numbers

In the rabbit RC1 [connectome](https://stemhop.com/methods/serial-section-em-connectomics/), an electron-microscopy reconstruction (Marc and colleagues, 2014), each of five fully analysed AII cells received input from 9 to 17 rod bipolar cells. These are rabbit numbers.

### Secondary rod routes

Rods also reach cone circuits through gap junctions between rods and cones, and through direct contacts onto some OFF cone bipolar cells. In mouse serial EM (Behrens and colleagues, 2016), rod contacts onto OFF cone bipolar types 3A, 3B and 4 were confirmed, and cones were found contacting rod bipolar cells, though the staining showed no explicit synaptic markers.

### Rods in daylight

Mouse rods, after an initial drop, recovered responses to moderate contrast in bright light, at backgrounds nominally up to 107 photoisomerizations per rod per second (Tikidji-Hamburyan and colleagues, 2017; the nominal values ignore bleaching, so the true rate was lower). Rods are not simply silent in daylight. The mouse result gives no value for humans, where estimates of rod saturation depend on viewing conditions.

---

From STEMhop (https://stemhop.com), free lessons by Christopher Deppmann. Lesson text © the authors. Practice questions: https://stemhop.com/lessons/retina/
