The visual system is the sensory apparatus responsible for detecting and processing light to enable vision, encompassing the eyes, neural pathways, and brain regions that convert electromagnetic radiation into perceptual experiences such as color, form, motion, and depth.[1] It begins with photoreception in the retina and extends through interconnected neural structures to higher cortical areas, allowing organisms to interpret their environment.[2]In humans, the visual system relies on the eye as its primary organ, where light enters through the cornea and is focused by the crystalline lens onto the retina—a thin, multilayered neural tissue lining the back of the eye.[3] The retina contains approximately 120 million rod photoreceptors for low-light (scotopic) vision and 6 million cone photoreceptors for high-acuity, color (photopic) vision, with cones concentrated in the fovea centralis for sharp central vision.[2][4] Phototransduction occurs when light activates photopigments like rhodopsin in rods or opsins in cones (sensitive to long-, medium-, and short-wavelength light for red, green, and blueperception, respectively), triggering a cascade that hyperpolarizes photoreceptors and modulates neurotransmitter release to bipolar and ganglion cells.[3]Ganglion cells, numbering about 1 million per eye, integrate signals from the retina and form axons that constitute the optic nerve, transmitting action potentials to the brain.[1][5]The afferent visual pathway routes these signals from each optic nerve through the optic chiasm—where nasal fibers cross to the contralateral side—into the optic tracts, synapsing in the lateral geniculate nucleus (LGN) of the thalamus, a six-layered structure that relays information via the optic radiations to the primary visual cortex (V1, Brodmann area 17) in the occipital lobe.[1] The LGN organizes inputs into magnocellular (M) layers for motion and depth processing and parvocellular (P) layers for color and fine detail, preserving retinotopic mapping throughout the pathway.[3] Beyond V1, visual information disperses to extrastriate areas like V2, V4 (color processing), and V5/MT (motion detection), enabling complex functions such as object recognition, binocular disparity for depth, and attentional modulation.[1]Accessory components include efferent pathways for eye movement and pupillary reflexes: the oculomotor (III), trochlear (IV), and abducens (VI) cranial nerves coordinate conjugate gaze, while parasympathetic fibers from the Edinger-Westphal nucleus constrict pupils in response to light via the optic nerve's afferent limb.[1] Sympathetic innervation dilates pupils for low-light adaptation.[1] Disruptions in this system, such as lesions at the chiasm causing bitemporal hemianopia, underscore its precision, with the overall architecture supporting trichromatic color vision evolved from ancestral dichromacy around 30 million years ago.[2]
Overview
Optical principles
The visual system begins with the optical components of the eye, which refract incoming light to form a focused image on the retina. Light rays from an object enter the eye primarily through the cornea, the transparent anterior surface that accounts for approximately two-thirds of the eye's total refractive power, about 43 diopters (D) in the relaxed state.[6] The rays then pass through the aqueous humor, a clear fluid in the anterior chamber between the cornea and the lens, which maintains intraocular pressure and contributes minimally to refraction while nourishing avascular tissues like the cornea.[7] Next, the crystalline lens, with a relaxed power of around 20 D, further bends the light, followed by transmission through the vitreous humor, a gel-like substance filling the posterior chamber that helps maintain the eye's shape and allows undistorted passage of light to the retina. Together, these elements create an optical system with a total power of about 60 D and a focal length of roughly 17 mm, projecting an inverted, real image onto the retina in the emmetropic (normal) eye, where parallel rays from distant objects converge precisely on the fovea.[8]To focus on objects at varying distances, the eye employs accommodation, the dynamic adjustment of the lens's curvature via contraction of the ciliary muscle, which relaxes the zonular fibers and allows the lens to thicken and increase its power by 8–12 D.[9] In young adults, this process enables a near point of accommodation at about 10 cm, corresponding to approximately 10 D of accommodative amplitude, allowing clear vision from infinity to this minimum distance.[10] As aging progresses, the lens loses elasticity, leading to presbyopia, which typically begins around age 40 and reduces accommodative amplitude to less than 2 D by age 50, shifting the near point farther away and necessitating corrective lenses for close work.[11]The eye's optics are not perfect and exhibit aberrations that degrade image quality. Spherical aberration occurs because peripheral rays refract more strongly than central rays, with the cornea contributing positive aberration and the lens negative, partially balancing in youth but worsening with age. Chromatic aberration arises from the lens and cornea refracting shorter wavelengths (e.g., blue light) more than longer ones (e.g., red), causing color fringing; the eye compensates partly through the fovea's insensitivity to blue and by constricting the pupil, which reduces the effective aperture and minimizes both spherical and chromatic aberrations by limiting peripheral ray entry.[12]Deviations from emmetropia result in refractive errors, quantified in diopters as the reciprocal of the focal length in meters (D = \frac{1}{f}, where f is in meters). In myopia (nearsightedness), the eyeball is elongated or the refractive power excessive, causing distant rays to focus anterior to the retina; correction requires a concavelens with negative power (e.g., -3 D for a focal length of -0.33 m). Hyperopia (farsightedness) involves a shorter eyeball or insufficient power, focusing rays posterior to the retina, corrected by a convex positive lens (e.g., +2 D). Astigmatism stems from irregular curvature of the cornea or lens, creating different refractive powers in principal meridians (e.g., +1.00 D sphere and -0.50 D cylinder at 90°), blurring images in specific orientations unless compensated by cylindrical lenses.[13]
Neural organization
The visual system's neural organization exhibits a hierarchical structure that processes visual information from the retina through successive stages to the cerebral cortex. At the retinal level, photoreceptors (rods and cones) detect light and transmit signals via bipolar and amacrine cells to retinal ganglion cells, whose axons form the optic nerve. These fibers converge at the optic chiasm, where nasal retinal fibers from each eye cross to the contralateral side, ensuring that visual field information is bilaterally represented in the brain. Post-chiasm, the optic tract carries these signals primarily to the lateral geniculate nucleus (LGN) of the thalamus, a key relay station organized into six layers that maintain retinotopic mapping of the visual field. From the LGN, geniculocalcarine radiations project to the primary visual cortex (V1, or striate cortex) in the occipital lobe, where initial feature extraction occurs, before signals diverge to higher extrastriate areas for advanced processing.[1]The retinogeniculate pathway is divided into parallel magnocellular (M) and parvocellular (P) streams, which segregate early in the retina and are preserved through the LGN to V1. The M pathway, originating from large parasol ganglion cells, conveys low-spatial-resolution, high-contrast, motion-sensitive signals through the ventral LGN layers (1 and 2), supporting detection of fast-moving or low-contrast stimuli. In contrast, the P pathway, from smaller midget ganglion cells, transmits high-spatial-resolution, color-opponent information via the dorsal LGN layers (3-6), enabling fine detail and chromatic discrimination. These streams partially converge in V1 but maintain functional separation, with additional koniocellular (K) projections handling blue-yellow color signals through the LGN's interlaminar regions. Extrastriate projections from V1 fan out to secondary areas like V2, V3, and beyond, integrating inputs for complex visual analysis.[14]Beyond V1, the visual system employs two major parallel processing streams: the ventral "what" pathway and the dorsal "where" pathway. The ventral stream extends from V1 through the occipitotemporal cortex to the inferotemporal lobe, specializing in object recognition, form perception, and visual identification by analyzing invariant features like shape and color. The dorsal stream, projecting from V1 to the parietal cortex, focuses on spatial localization, motion guidance, and visuomotor coordination, facilitating actions such as reaching and grasping by computing egocentric representations. These streams originate post-V1 and interact bidirectionally, allowing flexible integration of perceptual and action-oriented processing.[15]A hallmark of this organization is neural convergence, which enhances signal efficiency and receptive field properties. The human retina contains approximately 126 million photoreceptors that synapse onto about 1 million retinal ganglion cells, resulting in a convergence ratio of roughly 126:1 and reducing redundancy while amplifying sensitivity in peripheral vision. Ganglion cell receptive fields are circular and center-surround organized, with M cells featuring large fields for broad motion detection and P cells having smaller fields for precise acuity; this structure is preserved retinotopically through the LGN and V1, enabling a compressed yet spatially mapped representation of the visual world.[3]
Anatomy
Eye structure
The human eye is a roughly spherical organ with an axial length of approximately 24 mm in adults, consisting of external protective layers and internal chambers that maintain its structure and optical function.[16]External structures protect and interface with the environment. The eyelids are movable folds of skin and muscle that close reflexively to shield the eye from injury, distribute tears across the surface, and prevent corneal drying.[17] The conjunctiva is a thin, vascularized mucous membrane that lines the inner surfaces of the eyelids (palpebral conjunctiva) and covers the anterior sclera (bulbar conjunctiva), producing mucus to lubricate the ocular surface and serving as a barrier against pathogens.[18] The sclera forms the opaque, fibrous outer coat of the eye, comprising tough collagen fibers that provide structural integrity and rigidity to the posterior five-sixths of the globe, extending from the cornea to the optic nerve.[19] Anteriorly, the sclera transitions into the cornea, a transparent, avascular dome-shaped structure with a horizontal diameter of about 11.7 mm in adults, which contributes significantly to the eye's refractive power by bending incoming light rays.[20]Internally, the eye is divided into fluid-filled chambers that support its shape and nourishment. The anterior chamber lies between the cornea and the iris, while the posterior chamber is a narrow space between the iris and the lens; both are filled with aqueous humor, a clear fluid produced by the ciliary body processes to maintain intraocular pressure and provide nutrients to avascular tissues like the cornea and lens.[21] The larger posterior chamber, or vitreous chamber, occupies the space between the lens and the retina, filled with the gel-like vitreous humor that helps maintain the eye's spherical form and transmits light.[21]The lens is a biconvex, transparent structure suspended within the posterior chamber by zonular fibers (zonules of Zinn) that anchor it to the ciliary body, allowing for accommodation.[22] It consists primarily of elongated fiber cells filled with high concentrations of soluble crystallin proteins, which ensure optical clarity and refractive index gradation for focusing light.[22]Vascular supply to the eye arises mainly from branches of the ophthalmic artery. The choroid, a highly vascularized layer between the sclera and retina, nourishes the outer eye structures through its dense capillary network supplied by short posterior ciliary arteries.[23] The inner retina and optic nerve head receive blood via the central retinal artery and corresponding vein, which enter and exit through the optic disc.[23]
Retinal layers and cells
The retina is a multilayered neural tissue lining the posterior inner surface of the eye, consisting of ten distinct layers that facilitate the conversion of light into neural signals. These layers, organized from the innermost (vitreous-facing) to the outermost, include: the inner limiting membrane, a thin basement membrane formed by the footplates of Müller glial cells that separates the retina from the vitreous humor; the nerve fiber layer, comprising unmyelinated axons of retinal ganglion cells that converge to form the optic nerve; the ganglion cell layer, containing the cell bodies of these ganglion cells; the inner plexiform layer, where synapses occur between bipolar and amacrine cell processes and ganglion cell dendrites; the inner nuclear layer, housing the nuclei of bipolar cells, horizontal cells, and amacrine cells; the outer plexiform layer, the site of ribbon synapses between photoreceptor terminals and the dendrites of bipolar and horizontal cells; the outer nuclear layer, consisting of the nuclei of rod and cone photoreceptors; the external limiting membrane, a fenestrated layer of adherens junctions between Müller cell processes and photoreceptors; the photoreceptor layer, including the inner and outer segments of rods and cones where phototransduction occurs; and the retinal pigment epithelium, a single layer of cuboidal cells that absorbs stray light, recycles photopigments, and forms part of the blood-retina barrier.[24]Photoreceptor cells are the primary light-detecting elements, located in the photoreceptor layer, with rods and cones differing in distribution, sensitivity, and function. Rods, numbering approximately 120 million per retina, are specialized for scotopic (low-light) vision and exhibit peak sensitivity at 498 nm, enabling detection in dim conditions but without color discrimination.[4][25]Cones, totaling about 6 million, support photopic (bright-light) vision and color perception, with three subtypes: long-wavelength-sensitive (L-cones) peaking at around 564 nm for red light, medium-wavelength-sensitive (M-cones) at 534 nm for green, and short-wavelength-sensitive (S-cones) at 420 nm for blue.[4][25][26] These photoreceptors hyperpolarize in response to light, initiating signal transmission.Beyond photoreceptors, the retina features several neuronal and glial cell types that process and relay visual information. Bipolar cells, numbering around 10 million, form direct synaptic connections with photoreceptors and transmit graded potentials to ganglion cells, with subtypes specialized for on/off responses or specific cone inputs.[3] Horizontal cells provide lateral inhibition to enhance contrast by feedback to photoreceptors and feedforward to bipolar cells. Amacrine cells, diverse in morphology and neurotransmitter use, modulate bipolar-ganglion synapses for temporal and spatial refinement, including direction selectivity. Retinal ganglion cells, about 1 million in total, integrate inputs in the inner plexiform layer and generate action potentials that travel via their axons in the optic nerve; subtypes like midget and parasol cells correspond to parvocellular and magnocellular pathways. Müller glial cells span all layers, offering structural support, metabolic aid, and ion homeostasis while contributing to the inner and external limiting membranes.[24][3]The fovea centralis, a specialized depression in the macula lutea approximately 1.5 mm in diameter, optimizes high-acuity vision by featuring a cone-only region with minimal overlying layers, allowing direct light access to a density of approximately 150,000–200,000 cones per mm² (or ~15–20 cones per 100 µm²).[27][28][29] Its central portion includes an avascular zone about 0.5 mm in diameter, ensuring unobstructed photon capture without vascular interference.Retinal blood supply is dual: the inner layers (from inner limiting membrane to outer plexiform layer) receive oxygenated blood via the central retinal artery, a branch of the ophthalmic artery that enters through the optic disc and forms superficial and deep capillary networks; the outer layers (photoreceptors and pigment epithelium) are nourished by the choriocapillaris of the choroid, supplied by short and long posterior ciliary arteries, supporting high metabolic demands through diffusion.[30][24]
Central visual pathways
The central visual pathways begin with the formation of the optic nerve, which consists of axons from retinalganglion cells that converge at the optic disc and exit the eye.[31] This nerve contains approximately 1.2 million myelinated and unmyelinated fibers in humans, transmitting visual signals from the retina toward the brain.[32] Myelination of these axons occurs primarily after the optic chiasm, beginning in the optic tract and progressing retrogradely toward the chiasm during late gestation and early postnatal development.[33]At the optic chiasm, located at the base of the brain anterior to the pituitary gland, the optic nerves from both eyes partially decussate. Fibers originating from the nasal retina of each eye cross to the contralateral side, while temporal retinal fibers remain uncrossed, resulting in approximately 50% of fibers decussating in humans to enable binocular vision.[34] This partial crossing ensures that each cerebral hemisphere receives input from the contralateral visual field, integrating monocular signals for depth perception and stereopsis.[35]Post-chiasm, the bundled axons form the optic tracts, which carry segregated visual information to multiple targets, including the lateral geniculate nucleus (LGN) of the thalamus, the superior colliculus for orienting responses, and the pretectum for pupillary light reflexes.[36] The optic tracts maintain a retinotopic organization, preserving the spatial mapping of the visual field from the retina.[37]The LGN serves as the primary thalamic relay station for visual signals en route to the cortex, featuring a layered structure with six distinct laminae in primates. Layers 1 and 2 comprise magnocellular (M) cells, which process low-spatial-frequency, motion-sensitive information, while layers 3 through 6 contain parvocellular (P) cells specialized for high-spatial-frequency, color-opponent signals.[38] These layers exhibit strict retinotopic organization, with upper layers representing the inferior visual field and lower layers the superior field, and alternate ocular dominance for ipsilateral and contralateral inputs.[39] Local interneurons within the LGN modulate relayneuron activity, providing inhibitory feedback to refine signal transmission.[40]Anatomical variations in these pathways can occur, notably in albinism, where reduced melanin during development leads to excessive decussation at the optic chiasm, resulting in abnormal routing of over 90% of fibers to the contralateral hemisphere and disrupted binocular representation.[41] Such anomalies highlight the role of pigmentation in guiding axonal pathfinding during embryogenesis.[42]
Visual cortex regions
The visual cortex, located primarily in the occipital lobe, comprises a hierarchy of specialized regions that process visual information relayed from the lateral geniculate nucleus (LGN) of the thalamus. These regions form a retinotopic organization, where adjacent neurons respond to adjacent parts of the visual field, enabling precise spatial mapping. The primary visual cortex (V1) serves as the initial cortical entry point, followed by secondary areas like V2, V3, V4, and V5/MT, which handle increasingly complex features such as form, color, and motion. Higher-order regions in the temporal and parietal lobes integrate these signals for object recognition and spatial awareness, respectively.The primary visual cortex, also known as V1 or the striate cortex, corresponds to Brodmann area 17 and is situated along the calcarine sulcus in the occipital lobe. It receives direct thalamocortical inputs from the LGN, primarily terminating in layer 4. V1 exhibits a precise retinotopic map of the visual field, with disproportionate representation of the fovea—a phenomenon called cortical magnification—where central vision occupies a larger cortical area due to higher acuity demands. This magnification factor can exceed 10 times that of peripheral vision, underscoring V1's role in fine-grained spatial analysis.Adjacent to V1, area V2 processes more integrated features, including contours and simple forms, while maintaining a retinotopic organization divided into thin and thick stripes for color and disparity processing, respectively. Area V3 (or VP in some nomenclatures) extends this by handling global form, depth perception via binocular disparity, and coarse color information. Area V4, located anteriorly in the ventral stream, specializes in color constancy and object shape invariance, enabling perception of hues independent of illumination changes. In the dorsal stream, area V5 (also MT) is dedicated to motion processing, with neurons selectively responsive to direction and speed, crucial for tracking moving objects.Beyond these early extrastriate areas, the inferotemporal cortex in the ventral pathway supports advanced object recognition, where neurons respond to complex shapes and faces, achieving viewpoint-invariant identification through hierarchical feature integration. In contrast, parietal regions, such as the intraparietal sulcus, contribute to spatial attention by modulating visual salience and directing gaze, facilitating the selection of relevant stimuli in cluttered scenes.Cytoarchitecturally, V1 is distinguished by its layered structure, particularly layer 4C, which is subdivided into 4Ca and 4Cb. Layer 4Ca receives magnocellular (magno) inputs from the LGN, conveying low-spatial-frequency information for motion and luminance, while 4Cb receives parvocellular (parvo) inputs for high-resolution color and form details. These segregated inputs preserve parallel processing streams from the retina onward.Hemispheric specialization in the visual cortex reflects asymmetric processing: the left hemisphere excels at detailed, local feature analysis, such as fine textures and letters, whereas the right hemisphere prioritizes global, holistic configurations, like overall scene layout. This dichotomy arises from interhemispheric differences in connectivity and receptive field sizes, influencing tasks from reading to navigation.
Function
Phototransduction
Phototransduction is the process by which photoreceptor cells in the retina convert light energy into electrical signals through a series of biochemical reactions. In the outer segments of rods and cones, light absorption by visual pigments initiates a G-protein-coupled cascade that ultimately modulates membrane potential. This mechanism enables the detection of photons across a wide range of intensities and wavelengths, with rods specialized for low-light vision and cones for color and high-acuity tasks.[43]In rod photoreceptors, the visual pigment rhodopsin consists of the protein opsin bound to the chromophore 11-cis-retinal. Upon absorption of a photon, 11-cis-retinal isomerizes to all-trans-retinal, inducing a conformational change in rhodopsin to its active form, metarhodopsin II (R*). This activated rhodopsin catalyzes the exchange of GDP for GTP on the G-protein transducin, activating approximately 20 transducin molecules per R*.[43][44]The activated transducin then stimulates phosphodiesterase (PDE), which hydrolyzes cyclic guanosine monophosphate (cGMP) to 5'-GMP, rapidly reducing cytosolic cGMP levels. In the dark, high cGMP concentrations keep cation channels open, allowing Na⁺ and Ca²⁺ influx that depolarizes the rod to approximately -40 mV. Light-induced cGMP decline closes these channels, halting the "dark current" and hyperpolarizing the membrane to about -70 mV, which decreases glutamate release at the synapse. This amplification—one photon activates approximately 20 transducin molecules—enhances sensitivity.[44][43]Cone photoreceptors employ similar mechanisms but with distinct photopigments called iodopsins, each comprising opsin variants (short-, medium-, and long-wavelength sensitive) covalently linked to 11-cis-retinal derivatives from vitamin A. These pigments exhibit faster response kinetics and recovery times than rhodopsin, enabling cones to adapt quickly to varying light levels. The transducin and PDE cascade in cones mirrors that in rods, leading to comparable cGMP-gated channel closure and hyperpolarization, though with lower amplification suited for brighter conditions.[43]Dark and light adaptation maintain phototransduction sensitivity across illumination changes. In darkness, guanylate cyclase (GC) synthesizes cGMP to reopen channels, while light reduces Ca²⁺ influx through closed channels. This Ca²⁺ decline activates guanylate cyclase-activating proteins (GCAPs), which stimulate GC to restore cGMP levels and terminate the response. Calcium feedback via GCAPs provides gain control, compressing the response range in bright light and extending it in dim conditions. Rhodopsin deactivation by phosphorylation and arrestin binding further regulates adaptation.[45][43]Rods display peak spectral sensitivity around 500 nm in the blue-green range, following the absorption spectrum of rhodopsin, while cones peak at approximately 420 nm (S-cones), 530 nm (M-cones), and 560 nm (L-cones). The quantum efficiency of rhodopsin, the probability that an absorbed photon triggers a detectable response, is approximately 0.67 in rods, contributing to their single-photon detection capability.[43][46]
Visual signal processing
Visual signal processing begins in the retina, where retinal ganglion cells encode spatial patterns of light into action potentials that are transmitted via the optic nerve. These cells exhibit center-surround receptive fields, characterized by a central region that responds oppositely to a surrounding annular region, enabling contrast detection. For instance, ON-center ganglion cells increase firing when light stimulates the center while the surround is dark, whereas OFF-center cells respond to darkness in the center. This organization, first described in cat retina, sharpens edges by emphasizing differences in luminance across the visual field.[47]Ganglion cells are classified into major types based on their properties and projections. Parvocellular (P) ganglion cells, which constitute about 90% of the population in primates, have small receptive fields and are sensitive to color differences through opponent processes (e.g., red-green or blue-yellow), contributing to fine spatial detail and form perception. In contrast, magnocellular (M) ganglion cells have larger receptive fields, respond transiently to low-contrast stimuli, and are tuned to motion and coarse luminance changes, supporting depth and dynamic vision. These distinctions arise from inputs from distinct bipolar and amacrine cells, preserving parallel processing streams from the retina.[48]Lateral inhibition, mediated by horizontal cells in the outer retina and amacrine cells in the inner retina, enhances edge detection by suppressing activity in neighboring regions. Horizontal cells provide feedback to photoreceptors and bipolar cells, reducing responses to uniform illumination and amplifying boundaries between light and dark areas. Amacrine cells similarly inhibit bipolar and ganglion cells laterally, contributing to surround antagonism and temporal sharpening of signals. This mechanism underlies the center-surround structure and improves contrast sensitivity across the visual field.Axons from ganglion cells project to the lateral geniculate nucleus (LGN) of the thalamus, which acts primarily as a relay station while introducing subtle modulations. LGN neurons maintain center-surround receptive fields similar to those of ganglion cells, with parvocellular layers preserving color opponency and magnocellular layers emphasizing achromatic contrast and motion. Retinotopic organization ensures precise spatial mapping, and inputs from brainstem and cortical areas provide top-down modulation for attention and arousal. Weak orientation selectivity begins to emerge in some LGN cells, particularly in koniocellular layers, though it remains rudimentary compared to cortical processing.[49]In the primary visual cortex (V1), LGN inputs drive neurons with more specialized receptive fields. Simple cells, located mainly in layer 4, respond to oriented edges or bars at specific positions within their receptive fields, exhibiting elongated excitatory and inhibitory subregions. Complex cells, found in other layers, respond to oriented stimuli across a broader area without precise positional specificity, integrating inputs from simple cells. According to the Hubel-Wiesel model, these properties arise from convergent wiring of LGN afferents aligned along ocular dominance columns and orientation preferences, forming the basis of feature detection. The receptive fields of simple cells resemble Gabor functions, which are Gaussian-modulated sinusoids optimal for detecting edges in natural images.Parallel processing streams are maintained through the visual pathway, with the P pathway (via parvocellular LGN) specializing in color and high-acuity form, and the M pathway (via magnocellular LGN) handling motion, depth, and low-spatial-frequency information. These streams partially converge in V1 but remain segregated into ventral (form/color) and dorsal (motion/depth) cortical pathways. Contrast sensitivity in both streams is quantified using Michelson contrast, defined asC = \frac{L_{\max} - L_{\min}}{L_{\max} + L_{\min}}where L_{\max} and L_{\min} are the maximum and minimum luminances in the stimulus; this measure highlights how M cells detect low contrasts for global scene analysis, while P cells resolve finer details. Processed signals from V1 project to higher cortical areas for further integration.[48][50]
Visual perception mechanisms
Binocular vision enables depth perception through the slight differences in the images projected onto each retina, known as binocular disparity. This disparity provides cues for stereopsis, where the visual system computes relative depths by comparing corresponding points across the two eyes. The horopter represents the locus of points in space that project zero disparity onto corresponding retinal locations, forming a theoretical curve—often approximated as the Vieth-Müller circle—beyond which points elicit uncrossed or crossed disparities for perceived depth.[51] Under optimal conditions, the human stereopsis threshold for detecting depth via disparity is approximately 10 arcseconds, allowing fine discrimination of distances as small as a few centimeters at arm's length.[52]Color vision arises from the combined action of receptor and neural processing mechanisms in the visual pathway. The trichromatic theory, proposed by Thomas Young and elaborated by Hermann von Helmholtz, posits that color perception results from the relative stimulation of three types of cone photoreceptors sensitive to short (blue), medium (green), and long (red) wavelengths.[53] This retinal stage is complemented by the opponent-process theory, developed by Ewald Hering and quantitatively formulated by Leo Hurvich and Dorothea Jameson, which describes post-receptoral channels encoding color differences along red-green, blue-yellow, and black-white axes to account for phenomena like afterimages and color contrast.[54] Anomalies in these systems lead to color vision deficiencies; for instance, protanopia involves the absence of functional long-wavelength cones, resulting in confusion between reds and greens due to reliance on medium- and short-wavelength signals alone.[55]Motion perception involves resolving ambiguities in local motion signals to form coherent global representations. The aperture problem occurs when a limited visual field restricts observation of an object's full motion trajectory, allowing multiple possible directions perpendicular to the visible edge, as neurons in early visual areas respond only to the component normal to their orientation tuning.[56] Optic flow patterns, first conceptualized by James J. Gibson, describe the radial expansion or contraction of visual motion during self-movement, providing cues for heading direction and environmental layout through the focus of expansion where motion vectors converge.[57] In the middle temporal (MT) area of the extrastriate cortex, neurons exhibit robust direction selectivity for complex stimuli, integrating inputs from primary visual cortex to disambiguate local motions and support perception of object trajectories and speeds.[58]Visual illusions reveal how perceptual mechanisms can misinterpret sensory inputs, often due to incomplete or conflicting cues. The Müller-Lyer illusion, where lines flanked by inward- or outward-pointing arrows appear unequal in length despite being identical, arises from the visual system's probabilistic inference of depth from angular cues, biasing length estimation as if viewing corners in a three-dimensional scene.[59] Similarly, the phi phenomenon, described by Max Wertheimer, induces the perception of smooth motion from sequentially flashing stationary lights, driven by low-level temporal integration in early visual areas that fills spatial gaps to create apparent continuity.[60] These effects are modulated by feature binding, where attention links disparate attributes like position and motion into unified objects, and top-down influences from higher cortical areas, which impose expectations to resolve ambiguities in ambiguous scenes.[61][62]Gestalt principles describe innate organizational rules that the visual system uses to segment and interpret complex scenes into meaningful wholes. The principle of proximity groups elements based on spatial nearness, such that dots clustered closely are perceived as forming patterns or objects separate from more distant ones, facilitating scene parsing without explicit computation.[63] Similarity promotes grouping of elements sharing attributes like color, shape, or orientation, overriding minor positional differences to bind features into coherent entities, as seen in camouflage breakdown when uniform patterns disrupt matches.[63] Closure completes incomplete contours into enclosed shapes, with the brain inferring missing segments to perceive a whole figure, such as recognizing a circle from a partial arc, enhancing object recognition amid clutter.[63] These principles operate primarily in early to mid-level visual processing, aiding efficient segmentation before integration in higher cortical areas.
Development
Embryonic formation
The embryonic development of the visual system begins in the third week of gestation, when optic grooves appear in the ventral neural folds of the developing forebrain, marking the initial site of eye formation. These grooves rapidly evaginate outward as optic vesicles, which consist of neuroectoderm and protrude laterally from the diencephalon. The optic vesicles induce the overlying surface ectoderm to thicken into a lens placode, while the proximal portion of each vesicle remains connected to the brain via the optic stalk, which will later develop into the optic nerve. This evagination process is regulated by genes such as Rx, which is essential for optic vesicle formation and retinal differentiation.[64]By the fourth week, the optic vesicle invaginates to form the double-layered optic cup, where the inner layer differentiates into the neural retina and the outer layer into the retinal pigment epithelium. The optic stalk narrows and becomes the optic nerve precursor, facilitating axonal outgrowth from retinal ganglion cells. Concurrently, the lens placode, induced by signals from the optic vesicle including the transcription factorPax6, deepens into a lens pit and detaches to form the lens vesicle, which fills with elongating primary lens fibers. Mutations in Pax6, a master regulator of eye development, disrupt lens placode induction and can lead to congenital cataracts due to impaired lens fiber differentiation. The ventral optic cup also develops the choroid fissure, a groove through which the hyaloid artery enters to nourish the lens and retina.[64][65][66]Retinal lamination proceeds in an inside-out manner starting around week 5, with retinal progenitor cells in the inner neuroblastic layer generating neurons in a sequential order: ganglion cells differentiate first, followed by amacrine, horizontal, and cone photoreceptor cells, while rod photoreceptors and bipolar cells emerge later. By week 20 of gestation, the major retinal layers are established, including the ganglion cell layer, inner plexiform layer, inner nuclear layer, outer plexiform layer, outer nuclear layer containing photoreceptors, and the retinal pigment epithelium, though synaptic refinement continues. Vascularization of the eye involves the hyaloid artery, which supplies the avascular lens and inner retina during early development; the choroid begins forming around weeks 6-7 from mesenchyme surrounding the optic cup, developing capillaries by week 12 and mature vessels by week 22. The hyaloid artery subsequently regresses postnatally, leaving remnants like the central artery of the retina.[67][68][69]Critical periods for visual system morphogenesis occur between weeks 3 and 8, during which teratogens can disrupt key processes such as optic fissure closure. The choroid fissure must fuse by week 7 to seal the ventral optic cup; failure leads to coloboma, a gap in ocular structures like the retina, choroid, or iris. Thalidomide exposure during this window interferes with fissure closure, resulting in coloboma and other ocular defects in affected embryos, highlighting the vulnerability of these early developmental stages. Genes like Pax2 are crucial for proper fissure closure and optic stalk development.[64][70][71]
Postnatal maturation
At birth, human infants exhibit limited visual capabilities, with acuity estimated at approximately 20/400, allowing them to detect only large, high-contrast features at close range. Newborns show a preference for high-contrast edges and patterns, which guide their initial visual exploration and support early perceptual learning. Foveal development, including pit formation and cone specialization, progresses rapidly postnatally, reaching significant maturity by around 6 months of age, thereby improving central acuity and fixation stability.The visual system undergoes critical periods of heightened plasticity during early infancy, particularly in the primary visual cortex (V1), where ocular dominance columns segregate inputs from each eye between 3 and 8 weeks in animal models like cats, establishing binocular organization. In humans, this sensitivity window extends into the first few years; disruptions such as eye misalignment (strabismus) during this period can lead to amblyopia, or "lazy eye," due to competitive imbalances in cortical representation.Myelination of the optic pathways, which enhances signal conduction speed, begins prenatally but continues postnatally, with completion in the optic nerve and radiations typically by age 2 years, coinciding with refinements in visual processing efficiency.[72]Synaptic pruning in the visual cortex refines neural circuits by eliminating excess connections, with synaptic density peaking around 8 months to 2 years before progressive elimination, stabilizing toward adult levels by puberty to optimize visual function. This process, driven by activity-dependent mechanisms, sharpens receptive fields and enhances feature selectivity in higher visual areas.During puberty, surges in sex hormones such as estrogen and testosterone modulate visual processing, with evidence indicating influences on color discrimination abilities, where females often exhibit superior performance potentially linked to estrogen's role in cortical plasticity.[73] These hormonal changes contribute to subtle sex differences in visual perception that emerge or consolidate in adolescence.[74]
Clinical aspects
Common disorders
The visual system is susceptible to a range of disorders that impair various stages of visual processing, from refractive errors affecting light focus to degenerative conditions damaging neural components. Globally, at least 2.2 billion people experience near or distance vision impairment, with many cases linked to preventable or treatable conditions such as uncorrected refractive errors and cataracts.[75]Cataracts involve opacification of the eye's crystalline lens, leading to progressive blurring, glare, and reduced visual acuity, and represent the leading cause of blindness worldwide, affecting over 94 million people with moderate or worse vision impairment as of 2020. Primarily age-related, they result from protein aggregation and oxidative stress, though congenital, traumatic, or secondary forms (e.g., from diabetes or steroids) also occur; surgical removal with intraocular lens implantation restores vision in most cases and is highly effective when accessible.[75][76]Refractive errors, including myopia, hyperopia, and astigmatism, occur when the eye's shape prevents proper light focusing on the retina, leading to blurred vision. Myopia, or nearsightedness, affects approximately 30% of the global population currently, with projections estimating nearly 50% by 2050 due to increasing prevalence driven by factors like prolonged near work and reduced outdoor time.[77][78] This rise is particularly evident in urbanized regions with high educational demands, where extended reading or screen use correlates with axial eye elongation and myopia progression.[79]Diabetic retinopathy (DR) arises in individuals with diabetes mellitus due to microvascular damage from chronic hyperglycemia, leading to retinal hemorrhages, exudates, neovascularization, and macular edema that impair vision. Globally, DR affects about 22% of people with diabetes, contributing to over 1 million cases of blindness and 3 million of moderate-to-severe visual impairment as of 2020, with prevalence rising alongside the diabetes epidemic. Proliferative DR can cause vitreous hemorrhage or retinal detachment, while non-proliferative forms progress variably; early screening and glycemic control are key to prevention.[80][81]Glaucoma encompasses a group of disorders characterized by progressive damage to the optic nerve, often resulting from elevated intraocular pressure exceeding 21 mmHg, which compresses nerve fibers and leads to peripheral vision loss.[82] While primary open-angle glaucoma is the most common form, affecting aqueous humor drainage, the condition can also arise from normal intraocular pressure in susceptible individuals, ultimately causing irreversible visual field defects if undetected.[82]Retinal diseases significantly impact photoreceptor function and central vision. Age-related macular degeneration (AMD) is a leading cause of vision loss in older adults, manifesting in two forms: dry AMD, involving gradual atrophy of the macula and retinal pigment epithelium, and wet AMD, where vascular endothelial growth factor (VEGF) promotes abnormal choroidal neovascularization, leading to fluid leakage and rapid central vision deterioration.[83]Retinitis pigmentosa (RP) represents a heterogeneous group of inherited disorders primarily affecting rod photoreceptors, causing initial night blindness and peripheral vision loss due to genetic mutations that trigger progressive rod degeneration followed by secondary cone death.[84]Cortical disorders arise from damage to higher visual processing areas in the brain. Homonymous hemianopia results from lesions in the optic tract, lateral geniculate nucleus, or visual cortex, often due to stroke, causing loss of the same half of the visual field in both eyes and impairing spatial awareness.[85]Akinetopsia, or cerebral motion blindness, is a rare condition stemming from bilateral lesions in the middle temporal (MT) area, disrupting the perception of moving objects while preserving static vision, as documented in cases of stroke or hypoxic injury.[86]Color vision deficiencies impair the discrimination of hues due to cone photoreceptor anomalies. Red-green color blindness, the most prevalent form, affects about 8% of males due to X-linked recessive mutations in opsin genes on the X chromosome, resulting in altered spectral sensitivity.[87]Achromatopsia, a rarer autosomal recessive disorder, involves complete or near-total loss of color perception from absent or dysfunctional cones, often accompanied by reduced visual acuity and photophobia.[87]
Diagnostic methods
Diagnostic methods for assessing the integrity of the visual system encompass a range of techniques that evaluate acuity, visual fields, structural integrity, and electrophysiological responses, enabling clinicians to identify impairments from retinal to cortical levels.[88] These methods are essential for detecting conditions such as glaucoma, macular degeneration, and optic neuropathies, with selections based on suspected pathology.[89]Visual acuity tests measure the clarity of central vision by determining the smallest letters or symbols a patient can resolve at a standardized distance. The Snellen chart, introduced as a clinical standard, uses rows of letters decreasing in size, where 20/20 vision indicates the ability to resolve details subtending 1 arcminute of visual angle, equivalent to normal resolution at 20 feet.[90] For greater precision, especially in research and low-vision assessments, the logMAR chart employs a logarithmic scale with evenly spaced letter sizes and consistent spacing, offering advantages in repeatability and sensitivity to small changes in acuity over traditional Snellen testing.[91] These tests are typically conducted monocularly with refractive correction to isolate central retinal and optic pathway function.Visual field testing, or perimetry, maps the extent and sensitivity of peripheral vision to detect defects like scotomas, which are blind spots arising from localized damage. Goldmann perimetry, a kinetic manual method, uses a moving stimulus to delineate field boundaries and is particularly useful for patients with low vision or unreliable fixation, providing qualitative isopters for overall field shape.[89] In contrast, automated static perimetry, such as the Humphrey Field Analyzer, presents fixed-intensity stimuli at predefined locations to quantify threshold sensitivities, offering higher reproducibility and quantitative data for monitoring progressive losses, though it may miss subtle peripheral defects compared to kinetic approaches.[88]Imaging techniques provide detailed structural evaluation of the visual pathway. Optical coherence tomography (OCT) delivers non-invasive, high-resolution cross-sectional images of retinal layers, achieving axial resolutions of 5-10 μm to quantify thicknesses of the nerve fiber layer and detect early thinning indicative of axonal loss.[92]Fundus photography captures wide-field color images of the retina, optic disc, and vasculature, facilitating documentation of abnormalities like hemorrhages or drusen for longitudinal comparison.[93] For posterior pathway assessment, magnetic resonance imaging (MRI) excels in visualizing the optic nerve, chiasm, and tracts, with gadolinium enhancement highlighting inflammation or compression, though it is less sensitive for subtle retinal changes.[94]Electrophysiological tests objectively measure neural responses along the visual pathway. Visual evoked potentials (VEP) record cortical responses to patterned stimuli via scalp electrodes, with the P100 component—a positive peak around 100 ms post-stimulus—reflecting conduction time from retina to visual cortex; delays beyond 115-120 ms suggest demyelination or axonal damage.[95]Electroretinography (ERG) assesses retinal function by detecting electrical potentials from photoreceptors and bipolar cells in response to full-field flashes, with standardized protocols like the ISCEV guidelines distinguishing rod versus cone contributions to diagnose widespread retinal dysfunction.[96]Emerging advancements in the 2020s incorporate artificial intelligence (AI) for automated analysis of fundus images, enhancing early detection of age-related macular degeneration (AMD) by identifying subtle drusen or pigment changes with sensitivities exceeding 90% in validation studies, thus supporting triage in screening programs.[97] These AI tools, often based on deep learning convolutional neural networks, integrate with OCT and photography to predict progression risks, thereby aiding identification of disorders like AMD before symptomatic vision loss.
Comparative aspects
Invertebrate visual systems
Invertebrate visual systems exhibit remarkable diversity, ranging from simple photoreceptive structures to complex organs that rival vertebrate capabilities in specific functions, such as motion detection and polarization sensitivity. Unlike the centralized camera-type eyes of vertebrates, many invertebrates rely on distributed arrays or specialized detectors adapted to their ecological niches, enabling behaviors like rapid flight navigation or underwater predation.[98]A prominent example is the compound eye found in insects and crustaceans, composed of numerous repeating units called ommatidia that collectively form a mosaic image. In the fruit fly Drosophila melanogaster, each compound eye contains approximately 800 ommatidia, each functioning as an independent optical unit with a corneal lens, crystalline cone, and photoreceptor cluster.[99] These eyes operate via two primary optical mechanisms: apposition optics, where screening pigment isolates light to individual ommatidia for high spatial resolution in bright conditions, and superposition optics, which allows overlapping light paths from multiple ommatidia to enhance sensitivity in dim light.[100] Compound eyes provide exceptional temporal resolution, with flicker fusion rates exceeding 200 Hz in some insects, facilitating precise motion detection during high-speed activities like flight.[101]In addition to compound eyes, many insects possess ocelli—simple, non-imaging photoreceptors that detect light intensity and direction. In bees, such as bumblebees (Bombus spp.), the three dorsal ocelli serve primarily as light sensors for sky polarization navigation, particularly in low-light conditions like dusk, where they help maintain stable flight orientation by processing polarized skylight patterns.[102]Cephalopods, including octopuses and cuttlefish, possess camera-like eyes that convergently resemble those of vertebrates but with key structural differences, such as an everted retina where photoreceptors face the incoming light, mirroring the vertebrate inverted configuration in reverse.[103] These eyes feature a dynamic pupil; in cuttlefish (Sepia officinalis), it adopts a W-shaped form in bright light, which projects a blurred pattern onto the retina to balance vertically uneven illumination from above-water sources, enhancing contrast in shallow aquatic environments.[104]At the cellular level, invertebrate photoreceptors typically employ rhabdomeric opsins embedded in microvillar membranes, contrasting with the ciliary opsins in ciliated photoreceptors of vertebrates; this distinction reflects ancient evolutionary divergence in phototransduction pathways.These anatomical features underpin specialized behavioral adaptations, such as ultraviolet (UV) vision in bees, where trichromatic photoreceptors sensitive to 300–650 nm wavelengths allow detection of UV-reflective nectar guides on flowers, guiding foraging efficiency.[105] Similarly, mantis shrimps (Stomatopoda) exhibit advanced polarization vision through up to 16 spectral channels in their compound eyes, enabling dynamic processing of linear and circular polarized light for prey detection and intraspecific signaling in complex underwater scenes.[106] This diversity highlights convergent evolution with vertebrate systems, particularly in cephalopod camera eyes, yet underscores unique invertebrate solutions to visual challenges.
Vertebrate variations
The visual systems of vertebrates exhibit diverse adaptations shaped by ecological niches, ranging from aquatic environments to aerial and terrestrial habitats. In aquatic vertebrates like fish, the cornea is notably large and contributes to enhanced light gathering, compensating for the minimal refractive power it provides underwater due to the similar refractive indices of corneal tissue and water; instead, focusing relies primarily on a spherical lens that can shift position for accommodation.[107] Many fish species also possess tetrachromatic vision, incorporating ultraviolet-sensitive cones alongside red-, green-, and blue-sensitive ones, which enables detection of UV-reflecting patterns for communication and foraging in underwater light spectra.[108]Avian visual systems are characterized by tetrachromacy, with four cone types including a UV-sensitive variant, allowing birds to perceive a broader color spectrum than trichromatic mammals. These cones contain colored oil droplets that act as spectral filters, sharpening color discrimination by reducing chromatic aberration and enhancing contrast in bright daylight environments. Additionally, birds feature a unique pecten oculi, a vascular structure projecting into the vitreous humor that nourishes the avascular retina and may stabilize visual acuity during head movements.[109][110]Nocturnal mammals have evolved structures to maximize sensitivity in low-light conditions, including the tapetum lucidum, a reflective layer behind the retina that recycles unabsorbed light to increase photon detection efficiency. Their retinas are dominated by rods over cones; for example, domestic cats exhibit a rod-to-cone ratio of approximately 95:5, prioritizing scotopic vision for hunting at dusk or night while sacrificing color acuity.[111]Primate visual evolution reflects adaptations for diurnal frugivory and arboreal life, with Old World primates (Catarrhini) achieving routine trichromacy through duplication of the long-wavelength-sensitive (LWS) opsin gene on the X chromosome, enabling separate medium- and long-wavelength cones for red-green color discrimination centered in the fovea. In contrast, New World monkeys (Platyrrhini) exhibit polymorphic vision: males and homozygous females are dichromats relying on short- and medium-wavelength opsins, while heterozygous females achieve trichromacy via allelic variation in a single opsin locus, a mechanism that arose independently after the divergence from Old World lineages around 40 million years ago.[112][113]
Historical perspectives
Early discoveries
The earliest insights into the visual system emerged in ancient Greece, where philosophers and early anatomists began to conceptualize the eye's role in perception. Around the 5th century BCE, Alcmaeon of Croton proposed that the eye served as a pathway for light and sensory impressions to reach the brain through channels known as poroi, marking one of the first attempts to link vision to internal anatomy rather than external emanations from the eye. This idea was advanced in the 2nd century CE by the Roman physician Galen, who provided a detailed description of the optic nerve as a conduit for visual "spirits" or pneuma, emphasizing its role in transmitting sensory information from the eye to the brain while integrating it into his broader theory of extramission, where visual rays emanated from the eye.During the Renaissance, anatomical studies benefited from direct dissections and improved illustrations, leading to more accurate depictions of ocular structures. In 1543, Andreas Vesalius published De humani corporis fabrica, featuring precise woodcut illustrations of the eye's layers, including the cornea, sclera, choroid, and retina, which challenged Galenic errors and established a foundation for modern ocular anatomy through empirical observation. Building on this, in 1619, Jesuit scholar Christoph Scheiner demonstrated the inversion of the retinal image in his treatise Oculus hoc est: Fundamentum opticum, using pinhole experiments on animal eyes to show that light rays cross at the lens, forming an upside-down image on the retina—a key physiological insight that aligned the eye with optical principles.The 19th century saw further physiological explorations, often leveraging emerging technologies like the microscope to reveal entoptic phenomena and cellular details. In 1825, Czech physiologist Jan Evangelista Purkinje documented entoptic images, such as the shadows of retinal blood vessels visible against bright light (now known as the Purkinje tree), providing early evidence of internal ocular structures influencing perception without external aids. Hermann von Helmholtz, in his Handbuch der physiologischen Optik during the 1850s, modeled the eye as a camera obscura, with the lens focusing light onto the retina to form an image, integrating optics and physiology to explain accommodation and refraction.Theoretical advancements in color vision also took shape in the 1800s. Thomas Young proposed the trichromatic theory in 1802, suggesting three distinct retinal receptors sensitive to red, green, and blue-violet wavelengths, later refined by Helmholtz in the 1850s through quantitative analyses of color mixing. In the 1870s, Ewald Hering introduced the opponent-process theory, positing paired color channels (red-green, blue-yellow, and black-white) that explained phenomena like afterimages, challenging and complementing the trichromatic model.The advent of the compound microscope in the mid-19th century enabled histological breakthroughs, such as those by Albert von Kölliker in the 1850s, who identified distinct retinal cell types—including rods, cones, and supporting elements—through detailed examinations of fixed tissues, laying groundwork for understanding photoreceptor diversity.
Modern advancements
In the mid-20th century, David H. Hubel and Torsten N. Wiesel's groundbreaking electrophysiological studies in the 1960s revealed the functional organization of the primary visual cortex (V1), demonstrating that neurons exhibit orientation selectivity, responding preferentially to lines or edges at specific angles.[114] Their work established the hierarchical processing of visual information, where simple cells detect oriented edges and complex cells integrate motion and position, laying the foundation for understanding cortical feature detection; this research earned them the Nobel Prize in Physiology or Medicine in 1981 alongside Roger Sperry.[114]Advancements in neuroimaging during the 1990s enabled non-invasive mapping of visual areas in humans, with functional magnetic resonance imaging (fMRI) allowing researchers to delineate retinotopic organization—the orderly representation of the visual field on the cortical surface—in areas V1 through V4.[115]Roger B. H. Tootell and colleagues' 1995 study used fMRI to precisely identify borders of multiple visual areas by measuring responses to visual stimuli, confirming the retinotopic maps previously observed in animals and extending these findings to awake human subjects without invasive procedures.[115] Complementary techniques like electroencephalography (EEG) further supported these mappings by capturing temporal dynamics of visual processing, enhancing the spatial precision of fMRI.[116]Optogenetics emerged in the early 21st century as a transformative tool for manipulating neural activity with light, pioneered by Edward S. Boyden and colleagues in 2005, who introduced channelrhodopsin-2—a light-sensitive ion channel from algae—into mammalian neurons to achieve millisecond-precision control of spiking and synaptic transmission.[117] This technique has been applied to vision restoration, enabling targeted activation of surviving retinal ganglion cells in degenerative diseases like retinitis pigmentosa.[118] Building on this, retinal prostheses such as the Argus II system, approved by the U.S. Food and Drug Administration in 2013 for humanitarian use but whose manufacturer ceased support in 2020 and filed for bankruptcy in 2022, delivered electrical stimulation to the retina via an epiretinal array, restoring basic light perception and object recognition in profoundly blind patients with severe retinitis pigmentosa.[119][120] Subsequent optogenetic therapies, such as GS030, have advanced to clinical stages, with phase 1/2 data reported in 2023 showing restored light perception in patients with retinitis pigmentosa.[121]Genetic research in the late 1990s identified mutations in the cone-rod homeobox (CRX) gene as a cause of cone-rod dystrophy, a progressive retinal disorder affecting both cone and rod photoreceptors, with Charles L. Freund and colleagues linking CRX variants to impaired photoreceptor maintenance and differentiation in 1997. More recently, CRISPR-Cas9 gene editing has advanced toward clinical application for Leber congenital amaurosis (LCA), a severe form of inherited blindness; the EDIT-101 trial (BRILLIANCE), initiated in 2020, directly injects CRISPR components into the eye to disrupt a pathogenic mutation in the CEP290 gene, marking the first in vivo human use of this technology for retinal disease and, as reported in 2024, demonstrating safety with vision and quality-of-life improvements in 79% of treated participants (11 of 14) in at least one outcome measure.[122][123]In the 2010s, computational models using deep neural networks (DNNs) provided insights into visual processing by mimicking hierarchical representations in cortical areas V1 to V4, with Daniel L. K. Yamins and colleagues developing performance-optimized DNNs in 2014 that predicted neural responses in the inferior temporal cortex with accuracy rivaling biological systems during object recognition tasks.[124] These models, trained on natural images, replicate orientation selectivity in early layers analogous to V1 and invariant object features in later layers akin to V4, bridging neuroscience and artificial intelligence to test hypotheses about visual computation.[124]