Retina

Abstract

The retina is the thin sheet of neural tissue lining the back of the eye where vision begins. Far more than a passive screen, it is an outgrowth of the brain that captures light with its photoreceptors and then processes the image through several layers of neurons before sending it onward. Within this circuit the raw pattern of light is transduced, sharpened into local contrast, split into parallel channels for brightness, color, and motion, and compressed roughly a hundredfold onto the output cells whose axons form the optic nerve. Because most of what the mind knows of the visible world passes first through this sheet, the retina sets the terms of visual perception. This article covers its structure, its cell types and receptive fields, and the duplex rod and cone systems, with a worked example and three demonstrations.

Keywords: retina, photoreceptor, retinal ganglion cell, receptive field, duplex retina

What the Retina Is

The *retina* is the light-sensitive inner layer of the eye, a sheet of neural tissue roughly a quarter of a millimeter thick spread across the back of the eyeball. It is where the optical image formed by the cornea and lens is converted into neural signals. Unlike a photographic film, the retina does not merely record light: it is a piece of the central nervous system, developing as an outpouching of the embryonic forebrain, and it contains a complete miniature circuit of neurons that begins to interpret the image before any signal reaches the brain proper.

That circuit is built from five principal classes of neuron arranged in layers. Light is absorbed by the *photoreceptors*, the *rod* and *cone* cells, which transduce photons into an electrical signal through a biochemical cascade called *phototransduction*. That signal passes to the *bipolar cells*, which relay it inward to the *ganglion cells*, whose axons gather into the optic nerve and leave the eye. Woven across this vertical pathway are two classes of laterally connecting interneuron, the *horizontal cells* and the *amacrine cells*, which compare signals across neighboring regions and sculpt the response. The upshot is that the retina forwards to the brain not a copy of the image but a heavily recoded summary of it (Masland, 2001).

Types of Retina

The Medical Subject Headings vocabulary files *Retina* (D012160) beneath *Eye* in tree A09.371.729 and lists the narrower descriptors shown in Table 1. This placement reflects indexing practice rather than a functional taxonomy: MeSH is a controlled vocabulary for cataloguing literature, so its subtypes gather the anatomical regions, cell populations, and boundaries associated with the retina for retrieval, not distinct kinds of retina. The parent kind, *Eye*, is the whole organ of sight of which the retina is the innermost sensory coat. The entries below are glossed from their ordinary anatomical meaning and do not yet have their own articles on this site.

The Layered Circuit

The retina's neurons are stacked in a fixed order that, curiously, runs backward with respect to the light. Incoming light must pass through the transparent ganglion cell and bipolar cell layers before it reaches the photoreceptors at the very back, which lie against the *retinal pigment epithelium*, a dark cell layer that absorbs stray light and maintains the receptors. From there the signal travels forward again: photoreceptor to bipolar cell to ganglion cell, the *vertical pathway* through which the image flows toward the optic nerve. The bipolar cells are the pivotal middle stage, and they come in more than a dozen types that sample the receptors in systematically different ways (Euler, Haverkamp, Schubert, & Baden, 2014).

How the signal is transformed as it descends these layers was first read out directly by intracellular recording. Werblin and Dowling penetrated the individual neurons of the mudpuppy retina and found that photoreceptors, horizontal cells, and bipolar cells respond not with spikes but with smoothly graded changes in voltage, and that the *amacrine cells* and ganglion cells are the first to fire the action potentials that will travel down the nerve (Werblin & Dowling, 1969). This established the retina as a place where analog processing precedes the digital spike code, and it revealed the two lateral interneuron classes as the source of the inhibition that shapes each cell's response.

Figure 1

The Layered Circuit of the Retina

Parallel Channels and Receptive Fields

A retinal neuron does not report the light falling on a single point. Each ganglion cell answers to a small patch of retina, its *receptive field*, and Kuffler discovered that this patch is organized as an antagonistic center and surround: light in the center of an on-center cell excites it, while light in the surrounding ring suppresses it, so the cell signals the difference between the two (Kuffler, 1953). Barlow, recording in the frog, had shown the same year that summation within a central region and inhibition from beyond it govern the ganglion cell's output, framing the retina as a device tuned to spatial and temporal change rather than to steady illumination (Barlow, 1953). The center-surround arrangement means the retina emphasizes edges and contrast and discards uniform expanses of light, an early form of the redundancy reduction that makes vision efficient.

This local computation is not performed once but many times over in parallel. The single optic nerve fibers that Hartline first isolated came in distinct kinds, some firing at light onset, some at offset, some at both, the earliest evidence that the retina transmits several different messages about the same scene (Hartline, 1938). Modern accounts treat the mammalian retina as a stack of parallel channels: roughly a dozen bipolar types feed into some thirty or forty ganglion cell types, each tiling the retina and each extracting a different feature such as motion in one direction, a bright edge, or a change in color (Wässle, 2004). Large-scale functional imaging in the mouse resolved more than thirty distinct ganglion cell response types covering the visual field, confirming that the retina hands the brain a richly parsed image rather than a raw one (Baden et al., 2016). In the primate, where color vision is elaborate, the channels that leave the eye carry a specific, decodable code that population studies have begun to read directly (Field & Chichilnisky, 2007).

Rods, Cones, and the Duplex Retina

The photoreceptor layer is not uniform but doubled. The *duplex retina* combines two receptor systems with complementary jobs: the *rods*, extraordinarily sensitive and able to signal the absorption of a single photon, serve vision in dim light but are slow, colorblind, and pooled together too coarsely for fine detail; the *cones*, far less sensitive, work in daylight and provide color and high acuity. That the rod is sensitive enough to count single quanta of light was established by recording the tiny electrical current of an individual outer segment (Baylor, Lamb, & Yau, 1979). The two systems are distributed differently across the sheet: cones are concentrated in the central *fovea*, at the middle of the *macula lutea*, where acuity peaks, while rods dominate the periphery and are absent from the very center.

Because the two systems overlap in one retina but operate over different ranges of light, the eye's sensitivity shifts as conditions change. On moving from bright light into darkness, vision recovers in two stages: the cones adapt first and quickly but reach a modest floor, and then, after several minutes, the far more sensitive rods take over and the threshold drops much further, a course known as dark adaptation. The break between the two branches marks the handover from cone to rod vision and is a direct behavioral signature of the duplex design (Masland, 2012).

The wealth of cell types underlying these systems is not incidental clutter but the retina's fundamental plan: a small, repeated set of circuits, each built from a defined complement of interneurons, that is broadly conserved across the vertebrates (Sanes & Masland, 2015). Understanding how those circuits assemble the signal is the aim of comprehensive surveys of the primate retina's cells and their connections (Grünert & Martin, 2020).

Worked Example: How Steeply the Retina Compresses

Consider the degree of convergence between the photoreceptors that catch light and the ganglion cells that report to the brain. A human retina holds on the order of 120 million rods and cones, yet only about 1 million ganglion cells carry its output. Dividing the input by the output gives an average convergence of 120 million over 1 million, or roughly 120 photoreceptors funneling onto each ganglion cell.

This average, however, hides the crucial fact that convergence is wildly uneven. In the peripheral retina, where rods dominate, hundreds of receptors may pool onto a single ganglion cell; that heavy summation is what lets the periphery detect a faint light at night, at the cost of blurring detail. In the central fovea the ratio inverts: a single cone can drive its own midget bipolar and midget ganglion cell almost privately, a convergence approaching one to one, which is what preserves the fine acuity of central vision.

The trade-off is quantitative. If about 100 rods converge on one output cell in the periphery, their pooled signal is roughly 100 times more likely to cross a detection threshold in dim light than a lone cone acting alone, but the pooled cell can no longer say which of its 100 receptors was struck, so spatial resolution falls by the same factor. The retina thus spends its fixed budget of one million output lines differently in different places, lavishing them on the fovea for acuity and pooling them in the periphery for sensitivity. The first demonstration shows the center-surround computation each of those output cells performs, and the second shows how receptor density, and therefore this convergence, changes from the fovea outward.

Discussion

The retina overturns the intuition that the eye simply relays a picture to the brain. By the time signals leave through the optic nerve, a full stage of neural computation has already occurred: the image has been transduced, its redundant uniform regions discarded, its edges and movements emphasized, and its content divided among dozens of parallel channels that each report a different feature. What the visual brain receives is therefore an interpretation, not a photograph, and much of the character of visual perception is set by decisions the retina has already made (Wässle, 2004).

This matters for cognition well beyond the eye. The concentration of acuity in the tiny fovea forces the eyes to move constantly to sample a scene, tying vision tightly to the machinery of attention. The center-surround receptive field is the elementary operation from which the more elaborate feature detectors of the cortex are built, so the retina is where the study of sensation and neural coding begins. And because the retina is accessible brain tissue, transparent and directly viewable, it has served as the model system in which general principles of neural circuitry, from parallel processing to efficient coding, were first worked out and can still be tested against the whole of perception.

Current Directions

The most active frontier is a complete parts list of the retina. Single-cell transcriptomics now assigns every neuron to a molecular type, and a cell atlas of the human fovea and peripheral retina has begun to match the physiological channels to defined genetic identities, resolving the receptors, bipolar, amacrine, and ganglion classes as distinct populations (Yan et al., 2020). Comparative work is asking which of these types and circuits are shared across species and which are specializations, turning the retina into a testbed for how a neural computation is conserved or reinvented in evolution (Baden, Euler, & Berens, 2020).

A second front is clinical and translational. Because the retina is an accessible piece of the central nervous system, it is a leading target for therapies that aim to restore vision lost to photoreceptor degeneration, whether by gene therapy that reactivates surviving cells, by optogenetic tools that confer light sensitivity on downstream neurons, or by electronic prostheses that attempt to reproduce the ganglion cell code directly. Progress on all three depends on the detailed circuit maps now emerging from primate work (Grünert & Martin, 2020). Open questions remain about how faithfully the mouse retina models the human one, and about whether an artificial device can drive the many parallel channels in their correct proportions rather than stimulating them indiscriminately.

Common Misconceptions

The retina is like film in a camera.

It is active neural tissue, part of the brain, that processes the image through several layers before transmitting it, not a passive surface that merely records light (Masland, 2001).

The photoreceptors face the incoming light.

The retina is inverted: light must pass through the ganglion cell and bipolar cell layers before reaching the rods and cones at the very back, against the pigment epithelium (Werblin & Dowling, 1969).

A ganglion cell reports how much light hits it.

Its center-surround receptive field makes it report the contrast between a central spot and its surround, so uniform light produces little response (Kuffler, 1953).

One kind of cell carries the whole image.

The retina splits the scene among dozens of parallel ganglion cell types, each extracting a different feature and tiling the visual field independently (Baden et al., 2016).

Glossary

Amacrine cell.

A laterally connecting interneuron of the inner retina that shapes the signal passing from bipolar to ganglion cells, contributing inhibition and timing.

Bipolar cell.

The middle neuron of the vertical retinal pathway, relaying the photoreceptor signal to the ganglion cells; it comes in more than a dozen functional types.

Cone.

A photoreceptor specialized for bright light, color, and fine detail, concentrated in the fovea at the center of the retina.

Duplex retina.

The combination in one retina of two receptor systems, rods for dim light and cones for bright light, with complementary sensitivities and roles.

Fovea.

The central pit of the retina, densely packed with cones and served by near one-to-one convergence, responsible for high-acuity central vision.

Ganglion cell.

The output neuron of the retina whose axon enters the optic nerve; ganglion cells come in dozens of types with center-surround receptive fields.

Horizontal cell.

A laterally connecting interneuron of the outer retina that compares signals across neighboring photoreceptors, helping to build the antagonistic surround.

Macula lutea.

The yellow-pigmented central region of the retina containing the fovea and responsible for detailed central vision.

Photoreceptor.

A retinal neuron, either a rod or a cone, that absorbs light and transduces it into an electrical signal through phototransduction.

Phototransduction.

The biochemical cascade by which a photoreceptor converts an absorbed photon into a change in its membrane current.

Receptive field.

The region of the retina in which light changes the firing of a given neuron, classically organized as an antagonistic center and surround.

Retinal pigment epithelium.

The layer of pigmented cells behind the photoreceptors that absorbs stray light, recycles visual pigment, and renews the receptors' outer segments.

Rod.

A photoreceptor of extreme sensitivity, able to signal a single photon, that serves dim-light vision but provides no color and coarse detail.

Vertical pathway.

The through-route of the retina from photoreceptor to bipolar cell to ganglion cell, along which the image flows toward the optic nerve.

Key Researchers

Tom Baden

A neuroscientist at the University of Sussex who uses large-scale functional imaging to map the many response types of retinal ganglion and bipolar cells and to compare the retinal basis of vision across species.
ORCID - Faculty page - Lab

Horace B. Barlow

(1921-2020). A visual neuroscientist at the University of Cambridge who demonstrated summation and inhibition within retinal receptive fields and framed the retina as an efficient coder that strips redundancy from the image.
Wikipedia - Wikidata

E. J. Chichilnisky

A neuroscientist at Stanford University who studies how populations of retinal ganglion cells encode the visual image and how that code might be reproduced by a retinal prosthesis.
ORCID - Wikipedia - Faculty page

John E. Dowling

A neuroscientist at Harvard University who, with Frank Werblin, used intracellular recording to trace the flow of signals through the layers of the retina and reveal the graded responses of its interneurons.
Wikipedia - Faculty page

Stephen W. Kuffler

(1913-1980). The founder of Harvard's Department of Neurobiology, who first described the antagonistic center-surround organization of the receptive fields of mammalian retinal ganglion cells.
Wikipedia - Wikidata

Richard H. Masland

(1942-2019). A neuroscientist at Harvard Medical School who argued that the retina is assembled from a fixed set of parallel circuits and whose syntheses catalogued its cell types and its fundamental plan.
Faculty page

Botond Roska

A neuroscientist at the Institute of Molecular and Clinical Ophthalmology Basel who dissects the parallel circuits of the retina and develops optogenetic gene therapies to restore vision in blinding disease.
Wikipedia - Wikidata

Heinz Wässle

A neuroscientist emeritus at the Max Planck Institute for Brain Research who established that the mammalian retina processes the image through parallel channels of separately tiling cell types.
Wikidata - Faculty page

Frequently Asked Questions

What is the retina?

The retina is the thin, light-sensitive sheet of neural tissue lining the back of the eye, where the optical image is converted into neural signals. It is part of the central nervous system and processes the image through several layers of neurons before sending it to the brain along the optic nerve (Masland, 2001).

What are the main cell types in the retina?

The retina contains five principal classes of neuron: photoreceptors (rods and cones), bipolar cells, ganglion cells along the vertical pathway, and horizontal and amacrine cells connecting sideways. Each class comes in multiple functional types, so the retina holds many dozens of distinct neurons in all (Masland, 2012).

What is the difference between rods and cones?

Rods are extremely sensitive and support vision in dim light but give no color and coarse detail; cones are far less sensitive, work in bright light, and provide color and fine acuity. This division into two receptor systems is called the duplex retina, and rods can signal the capture of even a single photon (Baylor, Lamb, & Yau, 1979).

What is a receptive field?

A receptive field is the patch of retina in which light alters a given neuron's firing. For a ganglion cell it is organized as an antagonistic center and surround, so the cell reports the contrast between the two regions rather than the overall light level (Kuffler, 1953).

Why does light have to pass through the retina to reach the photoreceptors?

The retina is inverted: the ganglion cell and bipolar cell layers lie toward the incoming light, and the rods and cones sit at the back against the pigment epithelium. Light therefore crosses the nearly transparent inner layers before it is absorbed, and the neural signal then flows the other way (Werblin & Dowling, 1969).

Does the retina simply relay the image to the brain?

No. It performs a full stage of processing, discarding uniform regions, emphasizing edges and motion, and dividing the scene among dozens of parallel channels. What leaves the eye is a recoded, feature-parsed summary of the image, not a copy of it (Wässle, 2004).

How many photoreceptors and ganglion cells does the retina have?

A human retina has on the order of 120 million rods and cones but only about 1 million ganglion cells, an average convergence of roughly a hundred to one. That convergence is heavy in the periphery, favoring sensitivity, and close to one to one in the fovea, favoring acuity (Field & Chichilnisky, 2007).

Can retinal disease be treated by repairing the retina?

It is an active field. Because the retina is accessible neural tissue, gene therapies, optogenetic tools, and electronic prostheses are all being developed to restore vision lost to photoreceptor degeneration, guided by increasingly detailed maps of its cell types and circuits (Baden, Euler, & Berens, 2020).

References

Hartline, H. K. (1938). The response of single optic nerve fibers of the vertebrate eye to illumination of the retina. American Journal of Physiology, 121(2), 400-415. https://doi.org/10.1152/ajplegacy.1938.121.2.400

Barlow, H. B. (1953). Summation and inhibition in the frog's retina. The Journal of Physiology, 119(1), 69-88. https://doi.org/10.1113/jphysiol.1953.sp004829

Kuffler, S. W. (1953). Discharge patterns and functional organization of mammalian retina. Journal of Neurophysiology, 16(1), 37-68. https://doi.org/10.1152/jn.1953.16.1.37

Werblin, F. S., & Dowling, J. E. (1969). Organization of the retina of the mudpuppy, Necturus maculosus. II. Intracellular recording. Journal of Neurophysiology, 32(3), 339-355. https://doi.org/10.1152/jn.1969.32.3.339

Baylor, D. A., Lamb, T. D., & Yau, K. W. (1979). The membrane current of single rod outer segments. The Journal of Physiology, 288, 589-611. https://doi.org/10.1113/jphysiol.1979.sp012716

Masland, R. H. (2001). The fundamental plan of the retina. Nature Neuroscience, 4(9), 877-886. https://doi.org/10.1038/nn0901-877

Wässle, H. (2004). Parallel processing in the mammalian retina. Nature Reviews Neuroscience, 5(10), 747-757. https://doi.org/10.1038/nrn1497

Field, G. D., & Chichilnisky, E. J. (2007). Information processing in the primate retina: Circuitry and coding. Annual Review of Neuroscience, 30, 1-30. https://doi.org/10.1146/annurev.neuro.30.051606.094252

Euler, T., Haverkamp, S., Schubert, T., & Baden, T. (2014). Retinal bipolar cells: Elementary building blocks of vision. Nature Reviews Neuroscience, 15(8), 507-519. https://doi.org/10.1038/nrn3783

Masland, R. H. (2012). The neuronal organization of the retina. Neuron, 76(2), 266-280. https://doi.org/10.1016/j.neuron.2012.10.002

Sanes, J. R., & Masland, R. H. (2015). The types of retinal ganglion cells: Current status and implications for neuronal classification. Annual Review of Neuroscience, 38, 221-246. https://doi.org/10.1146/annurev-neuro-071714-034120

Baden, T., Berens, P., Franke, K., Román Rosón, M., Bethge, M., & Euler, T. (2016). The functional diversity of retinal ganglion cells in the mouse. Nature, 529(7586), 345-350. https://doi.org/10.1038/nature16468

Grünert, U., & Martin, P. R. (2020). Cell types and cell circuits in human and non-human primate retina. Progress in Retinal and Eye Research, 78, 100844. https://doi.org/10.1016/j.preteyeres.2020.100844

Yan, W., Peng, Y. R., van Zyl, T., Regev, A., Shekhar, K., Juric, D., & Sanes, J. R. (2020). Cell atlas of the human fovea and peripheral retina. Scientific Reports, 10, 9802. https://doi.org/10.1038/s41598-020-66092-9

Baden, T., Euler, T., & Berens, P. (2020). Understanding the retinal basis of vision across species. Nature Reviews Neuroscience, 21(1), 5-20. https://doi.org/10.1038/s41583-019-0242-1