Eyes — What They Are, What Breaks Them, How to Build New Ones¶

The eye is a biological camera + first-stage neural network: two cubic centimetres wired into a quarter of the cortex. Every eye disease is a failure of one of four subsystems — optics (cornea/lens), pressure/fluid, photoreceptors (rods/cones/RPE), or wiring (ganglion cells/optic nerve). Name the four and the full disease catalog collapses into a handful of failure modes.

🌿 budding tended 2026-05-19 research health eyes vision anatomy ophthalmology

flowchart LR
  light[photons] --> optics[cornea · lens · iris]
  optics --> photo[rods · cones · RPE]
  photo --> wire[ganglion cells]
  wire --> nerve[optic nerve]
  nerve --> cortex[V1 → V4]

The two-sentence definition¶

An eye is a biological camera + first-stage neural network that converts photons into spike trains and ships them down a million-axon cable to the brain. Almost every disease of the eye is a failure of one of four things: the optics (cornea, lens), the pressure/fluid system (aqueous and vitreous humor), the photoreceptor layer (rods, cones, RPE), or the wiring (ganglion cells, optic nerve, cortex).

If you keep those four subsystems in mind, the long catalog of eye diseases collapses into a handful of failure modes, and the engineering question — "can you replace the eye?" — becomes a question of which of those four you can rebuild.

Anatomy as engineering parts list¶

You can't talk about how to fix or replace an eye without naming the parts the way an engineer would.

Part	Function	Failure mode
Cornea	front lens, ~⅔ of total focusing power; transparent avascular collagen	scarring, keratoconus, infection, abrasion
Aqueous humor	clear fluid in front chamber; made by ciliary body, drained through trabecular meshwork	drainage failure → high intraocular pressure → glaucoma
Iris + pupil	aperture (f/2 to f/8 in humans)	not a common disease site, but pupil shape is a neuro sign
Lens	fine focus + accommodation; crystallins, originally transparent	cataract (oxidation, glycation); presbyopia (stiffening with age)
Vitreous humor	gel that holds the eye's shape; ~99% water, 1% collagen/HA	floaters; posterior vitreous detachment; vitreoretinal traction
Retina	sensor + first ~5 layers of neural processing	tears, detachment, vascular disease, degeneration
Photoreceptors	rods (~120M, low-light, monochrome) + cones (~6M, color, fovea)	retinitis pigmentosa, AMD, Stargardt, cone dystrophies
Retinal pigment epithelium (RPE)	nurse cells under photoreceptors; recycle visual pigment, phagocytose outer segments	AMD (dry and wet), inherited dystrophies
Bipolar / amacrine / horizontal cells	inner retina; first edge detection and gain control	spared early in most photoreceptor diseases — this is why optogenetic restoration is possible
Retinal ganglion cells (RGCs)	output neurons; ~1.2M per eye; axons form the optic nerve	glaucoma kills these; also optic neuritis, ischemic optic neuropathy
Optic nerve	bundle of RGC axons → LGN	does not regenerate in mammals (the central problem)
Optic chiasm	nasal fibers cross; this gives binocular cortex maps	tumors (pituitary) cause bitemporal hemianopia
LGN (lateral geniculate nucleus)	thalamic relay; six layers, M/P/K	strokes here are rare but devastating
V1 primary visual cortex	retinotopic map; orientation and ocular-dominance columns	strokes cause hemianopia; lesions cause cortical blindness
Higher visual areas (V2–V5, MT, IT, FFA, PPA)	motion, color, faces, places	prosopagnosia, achromatopsia, akinetopsia from local lesions

Note the asymmetry: everything in front of the retina is replaceable today (corneal transplant, IOLs, vitrectomy fluid). Everything from the photoreceptors backward is what 21st-century vision science is fighting over. And the optic nerve in the middle is the central unsolved problem — without nerve regeneration, no eye transplant can carry vision.

The disease catalog — sorted by where in the pipeline it breaks¶

Optics failures (front of eye)¶

Disease	What it is	Lifetime risk	Treatment
Refractive errors (myopia, hyperopia, astigmatism, presbyopia)	eye too long/short or cornea not spherical; lens stiff with age	~80% of adults eventually need correction	glasses, contacts, LASIK/SMILE, IOLs, ortho-K
Cataract	lens proteins oxidize / aggregate → opacity	~50% by age 75, ~70% by 85; #1 cause of blindness worldwide	phacoemulsification + IOL; one of the highest-volume, most successful surgeries in medicine
Keratoconus	cornea thins and bulges; usually bilateral, starts in teens	~1 in 2,000 (higher in some populations)	rigid contacts; corneal cross-linking (riboflavin + UV) early; transplant late
Corneal dystrophies / scars / infections	many causes — Fuchs', HSV keratitis, bacterial ulcers, Acanthamoeba	varies	drops, DSAEK/DMEK (partial transplants), full PKP
Pterygium	conjunctival overgrowth onto cornea, sun-driven	high in sunny climates	excision if visual axis threatened

Pressure / fluid failures¶

Disease	What it is	Why it matters
Glaucoma (open-angle, angle-closure, normal-tension)	IOP elevated or RGC vulnerability; progressive death of ganglion cells from the periphery in	#2 cause of irreversible blindness worldwide; ~80M people; mostly painless and asymptomatic until advanced
Ocular hypertension	IOP > 21 mmHg with no damage yet	a risk factor, not yet a disease — but the trigger to start watching
Hypotony	IOP too low (post-surgery, leak)	flattens the eye, ruins focus

The mean trabecular meshwork drains ~2.5 µL/min of aqueous; lose ~20% drainage capacity and pressure climbs enough to start killing RGC axons at the optic disc. The killing fact: by the time a glaucoma patient notices peripheral loss, 30–50% of their retinal ganglion cells are already dead. They do not come back.

Retinal / photoreceptor failures¶

Disease	What it is	Population scale
Age-related macular degeneration (AMD) — dry (geographic atrophy, drusen) and wet (choroidal neovascularization)	RPE and photoreceptors die in the macula	#1 cause of central vision loss in over-60s in high-income countries; ~200M globally
Diabetic retinopathy	hyperglycemia → microvascular damage → ischemia → neovascularization	leading cause of working-age blindness; almost all diabetics develop some after 20 years
Retinitis pigmentosa (RP)	inherited rod-cone dystrophy; >80 genes implicated	~1 in 4,000; night blindness → tunnel vision → eventual central loss
Stargardt disease (ABCA4)	juvenile macular dystrophy	~1 in 10,000
Leber congenital amaurosis (LCA)	severe infantile retinal dystrophy; RPE65 and others	rare; one of the first targets of FDA-approved gene therapy
Retinal detachment	neural retina lifts off RPE — rhegmatogenous, tractional, exudative	medical emergency; ~1 in 10,000/yr; surgery (vitrectomy, scleral buckle) within hours/days
Retinal vein/artery occlusions	clot or ischemia	sudden painless vision loss; stroke of the eye
Retinopathy of prematurity (ROP)	abnormal retinal vascularization in premature infants on supplemental oxygen	leading cause of childhood blindness historically — now largely managed
Uveitis	intraocular inflammation; many causes (autoimmune, infectious)	~10% of legal blindness if poorly managed

Optic nerve / brain failures¶

Disease	What it is
Optic neuritis	inflammation of the optic nerve; classic MS herald
Ischemic optic neuropathy (NAION, AION)	stroke of the optic nerve head; giant cell arteritis is the emergency variant
Leber's hereditary optic neuropathy (LHON)	mitochondrial; sudden bilateral central vision loss in young men typically
Compressive optic neuropathy	tumors, thyroid eye disease, aneurysms
Cortical visual impairment	brain damage at LGN, optic radiations, or V1 — eye is fine, the brain can't see
Functional / amblyopia	"lazy eye" — developmental cortical suppression of one eye; reversible only in childhood

The killing facts about prevalence¶

2.2 billion people have some vision impairment (WHO 2019 World Report on Vision).
1 billion of those are preventable or unaddressed — the gap is mostly uncorrected refractive error and unoperated cataract.
Cataract is the single biggest treatable cause of blindness on Earth. A 15-minute surgery costing $50–$100 in low-resource settings restores sight to tens of millions of people. The global ophthalmology effort that has scaled this is one of the great unrecognized public health wins of the last 40 years.
Glaucoma + AMD + diabetic retinopathy are the three large irreversible causes in high-income countries. Detection is the lever, because none of them are loud at the start.
Myopia is exploding in East Asia — 80–90% of urban young adults now. This is the largest secular shift in eye biology in the historical record, and it tracks education intensity and indoor time in childhood, not screens specifically.

Reading the eye — symptoms and what they actually mean¶

The same logic as weather and cancer signs: one sign in isolation is usually nothing; two unrelated signs together, or any "sudden + painless + persistent" sign, is the threshold to be seen the same day.

The same-day list — go to an ophthalmologist or A&E now¶

Sign	What it usually means
Sudden painless loss of vision in one eye	retinal artery/vein occlusion, retinal detachment, ischemic optic neuropathy, vitreous hemorrhage, stroke
New onset of "curtain coming down" or "shower of floaters + flashes"	retinal tear or detachment — hours-to-days matter
Sudden severe eye pain + nausea + halos + red eye	angle-closure glaucoma — can blind the eye within hours
New double vision (esp. binocular, not resolved by covering one eye)	nerve palsy, MS, aneurysm, thyroid eye, myasthenia
Vision loss with a tender scalp / jaw claudication / new headache in someone over 50	giant cell arteritis — start steroids before the biopsy, the other eye is hours away
Chemical splash	irrigate immediately and at length — copious water for ≥15 min before transport
Penetrating injury / suspected open globe	shield (do not patch); to surgery
Sudden loss of vision in pregnancy / postpartum	preeclampsia, central serous, pituitary apoplexy

The this-week list¶

Sign	Suspicion
Persistent floaters without flashes	PVD (often benign in over-50s, but should be examined to rule out tear)
Gradual loss of peripheral vision	glaucoma, RP
Lines look bent (Amsler grid)	macular disease — wet AMD or CSC
New, persistent eye redness with discharge / photophobia	infectious or autoimmune — uveitis is masquerade-prone
Headaches + transient visual obscurations	idiopathic intracranial hypertension, optic disc edema

The annual / biennial baseline¶

Adults 40+: comprehensive dilated exam every 1–2 years. This is where asymptomatic glaucoma is caught.
Diabetics: annual dilated retinal exam from diagnosis (type 2) or 5 years after (type 1). Non-negotiable.
High myopia (>6D): dilated exam to monitor for detachment and myopic maculopathy.
Family history of glaucoma, AMD, RP: start screening earlier and more often.
Children: vision check at age 3–5; catching amblyopia before ~7 is critical, after that the cortex stops being plastic enough.

How to keep your eyes safe — the actionable ladder¶

In rough order of evidence quality and effect size:

Don't smoke. Smoking roughly doubles AMD risk and accelerates cataract. Single biggest modifiable factor.
Wear UV-blocking sunglasses outdoors. UV-A and UV-B drive cataract, pterygium, and ocular surface squamous neoplasia. Cheap polarized lenses with UV400 are fine; the price tag mostly buys you optics, not protection.
Wear safety glasses for the right tasks. Drilling, grinding, hammering, chemistry, weeds-whackers, fireworks, paintball, racquet sports. The ER eye-injury rate is dominated by amateur DIY and sports; pros wear protection.
Control blood sugar, blood pressure, and cholesterol. Diabetic retinopathy and retinal vein occlusion both track systemic vascular health. HbA1c < 7% if you can manage it without hypos.
Have a baseline eye exam. Once at age 40 for everyone; annually for diabetics; every 1–2 years for adults 40+; immediately for any same-day sign above.
Contact lens hygiene. Never sleep in lenses you aren't certified to sleep in; never rinse with tap water (Acanthamoeba keratitis is rare but blinding); replace as prescribed; throw out solution at the date. Tap water + contacts is the single most preventable cause of catastrophic young-adult eye infections.
Outdoor time for children. ~2 hours of outdoor time per day in childhood substantially reduces myopia incidence and progression. The active ingredient appears to be light intensity (≥1,000 lux), not "looking far away."
Children with detected myopia: atropine 0.01–0.05% drops at night and/or orthokeratology slow progression by ~50% (talk to a pediatric ophthalmologist).
20-20-20 rule (mild evidence) for screen fatigue. Every 20 minutes, look at something 20 feet away for 20 seconds. Helps dryness and accommodative fatigue. Does not cause myopia or AMD, despite the popular fear — screen time is correlated with myopia but the operative variable is reduced outdoor time, not screens themselves.
Use eye drops sanely. Artificial tears are fine; "get-the-red-out" vasoconstrictors (naphazoline, tetrahydrozoline) used chronically cause rebound redness. Steroid drops without supervision can trigger glaucoma and cataracts in susceptible people.
Sleep, hydration, and dry-eye basics. Lid hygiene (warm compresses, lid scrubs) for blepharitis and meibomian gland dysfunction prevents most chronic dry eye. Omega-3 evidence is modest but reasonable.
Helmets in cycling, seatbelts in cars. Most of the most-blinding ocular trauma is preventable head/face trauma; the eye is along for the ride.

What does not help: blue-light-blocking glasses for screen use (no good evidence for retinal protection at screen intensities); "eye exercises" to fix refractive error; bilberry; megadose vitamins beyond the AREDS2 formulation for intermediate or unilateral AMD specifically (which is a real, narrowly indicated intervention — not a general supplement).

Engineering a new eye — what's possible, in increasing order of difficulty¶

This is where eyes get interesting from a builder's perspective. The optic nerve is the gate: anything that bypasses or substitutes for it is plausible; anything that requires regrowing it is hard. We can group the engineering options by which layer of the pipeline they intervene at.

Layer 1 — replace the optics (solved)¶

Intervention	What it does	Status
Glasses, contacts	external refractive correction	trivial; the most-used medical device on Earth
LASIK / SMILE / PRK	reshape cornea with laser	mature; ~40 years of data
Phakic IOLs (ICL)	implanted lens in front of native lens	mature
Cataract surgery + IOL	replace cloudy crystalline lens with synthetic acrylic lens	the most-performed surgery in many countries; ~28 million/year worldwide
Multifocal / accommodating / EDOF IOLs	restore some near vision	mature; trade-offs in contrast and halos
Corneal transplant (PKP, DSAEK, DMEK)	swap in donor cornea (full thickness or layered)	mature; ~180k/yr globally; cornea is the only routinely transplanted "tissue" because it is avascular
Bioengineered cornea (recombinant collagen, decellularized scaffolds)	grown corneal substitute	early human trials (LinkoCare BPCDX, 2022 Lancet — restored vision in keratoconus patients in Iran/India)
Keratoprosthesis (Boston KPro, OOKP)	plastic cornea anchored in tissue (osteo-odonto for OOKP — literally a tooth as scaffold)	last-resort but successful in otherwise-blinding scarring

The optics layer is essentially a solved engineering problem. The lens we already replace as routinely as a dental filling. The cornea is harder, mostly because the supply of donor tissue is the bottleneck — but bioengineered corneas are emerging fast.

Layer 2 — rescue the photoreceptors (active frontier)¶

This is the layer where gene therapy, optogenetics, and stem cell biology are converging.

Approach	Mechanism	Status
Voretigene neparvovec (Luxturna)	AAV delivering RPE65 to the retina in biallelic-RPE65 LCA	FDA-approved 2017 — first gene therapy for an inherited disease in the US
AAV gene therapies for other IRDs	one gene per AAV delivered subretinally or intravitreally	dozens in trials — choroideremia (CHM), USH1, XLRP (RPGR), ABCA4 (Stargardt) — mixed but improving
CRISPR/base editing in vivo	Editas EDIT-101 for CEP290 LCA, etc.	early human; Casgevy is the proof that genome editing works in humans for hemoglobinopathies; eye is next
Antisense oligonucleotides	Sepofarsen (CEP290), etc.	intravitreal injections; modest but real benefit
Optogenetics — Sahel/GenSight (botaretigene/GS030)	AAV puts a microbial opsin (ChrimsonR) into surviving retinal ganglion cells; goggles transform the world into amber-wavelength stimulation matched to the opsin	2021 Nature Medicine: one RP patient regained ability to detect, count, locate objects — first human optogenetic vision restoration
Stem-cell-derived RPE transplants	iPSC- or ESC-derived RPE sheets implanted under the macula	Mandai et al. 2017 (RIKEN, iPSC) and Astellas/OCATA trials in dry AMD; safety established, efficacy modest so far
Stem-cell-derived photoreceptor transplants	iPSC/ESC cones into degenerate retinas	preclinical; material-transfer between donor and host photoreceptors is a real complication (Pearson 2016)
Retinal organoids ("mini-retinas in a dish")	self-organizing 3D structures from PSCs	research tool today; potential transplant source
Neuroprotection (small molecules, neurotrophic factors)	slow photoreceptor loss without restoring vision	CNTF implants, complement inhibitors (pegcetacoplan, avacincaptad pegol) — FDA-approved 2023 for geographic atrophy, modest slowing

The cleanest big-picture summary: if you catch a photoreceptor disease early enough, gene therapy can often arrest it. Optogenetics is the first route to give partial vision back to people who've lost it, by repurposing the surviving inner retina. Stem-cell-derived transplants are inching forward but face an integration problem — even healthy donor photoreceptors don't always wire into the host's bipolar cells, and the recent finding is that some apparent "transplant function" was material exchange, not engraftment.

Layer 3 — bypass the retina (retinal prostheses)¶

The classic "bionic eye" idea: put a camera on glasses, send signals to an implanted chip that stimulates the surviving inner retina.

Device	Approach	Status
Argus II (Second Sight)	60-electrode epiretinal implant, external camera-glasses	FDA-approved 2013; company collapsed 2020; implanted patients stranded — a cautionary tale for any neural implant
Alpha-AMS / Alpha IMS (Retina Implant AG)	1,500-pixel subretinal microphotodiode array	CE-marked; company wound down 2019
PRIMA (Pixium / Stanford)	photovoltaic 378-pixel subretinal chip, no external wires; goggles project NIR-encoded image	active trials; the most promising current implant; ~20/420–20/500 letter reading reported in dry AMD patients
POLYRETINA (EPFL)	wide-field photovoltaic	preclinical / early human

Retinal prostheses today give hundreds of pixels, not thousands; that is enough for object localization, face detection at very low resolution, and reading large print, but it is not "vision" in any rich sense. The hard limits are: - Pixel density — electrodes can't be packed closer than ~25 µm without current crosstalk - Spatial selectivity — each electrode stimulates a blob of ganglion cells, not a single one; spatial acuity is bounded by stimulus spread, not pixel count - Cell-type selectivity — natural retina has 30+ ganglion cell types encoding different things; electrodes hit them all in parallel, scrambling the code

This is why the field is shifting toward optogenetic + projection systems (more naturalistic stimulation through the surviving inner retina) and toward cortical implants (skip the retina entirely).

Layer 4 — bypass the eye entirely (cortical visual prostheses)¶

If everything in front of V1 is destroyed — bilateral enucleation, optic nerve transection, end-stage glaucoma — the only option is to write directly to visual cortex.

Device	Approach	Status
Orion (Second Sight)	60-electrode surface array on V1	small human trial; users perceive phosphenes whose position roughly maps the camera's field
Gennaris (Monash Vision Group, Australia)	array of 9 mm² tiles each with 43 penetrating microelectrodes	early human implantation 2022–2024
CORTIVIS (Univ. Miguel Hernández, Spain)	Utah-array-based; Fernández et al. 2021 — 16 years blind, 6 months implanted, recognized letters at ~10 pixel spatial resolution	published proof of concept
Neuralink and the high-density-array generation	not yet in vision; the platform (1,000+ electrodes per probe, flexible threads) is the relevant one to watch	speculative for vision specifically

Cortical visual prostheses face a different version of the same density problem, plus a placement problem: V1 is folded into the calcarine sulcus, much of it inaccessible to surface arrays. Penetrating arrays reach more of it but trigger more glial response and lose channels over months to years. The honest current state: a blind person with a modern cortical implant can perceive phosphenes that roughly map a camera, can recognize simple shapes and letters, and can navigate marginally better — they are not seeing.

Layer 5 — replace the whole eye (the open frontier)¶

In May 2024, the team at NYU Langone (Rodriguez, Ceradini, et al.) performed the first whole-eye transplant in a human as part of a partial face transplant. The eye survived — it stayed pressurized, the retina remained perfused, the cornea stayed clear — and the patient never recovered any vision, exactly as expected.

The reason is the optic nerve. In mammals, retinal ganglion cell axons are part of the central nervous system, and severed CNS axons do not regrow under default conditions — they're stopped by: - intrinsic developmental loss of growth competence (PTEN, SOCS3, KLF transcription factors) - glial scar formation at the lesion site - myelin-associated inhibitors (Nogo, MAG, OMgp) - absence of guidance cues to find the LGN

The He lab (Harvard) and Goldberg and Benowitz labs have shown that co-deleting PTEN and SOCS3, plus inflammation cues (zymosan, oncomodulin) and electrical activity, can drive substantial mouse RGC axon regeneration through the chiasm and to LGN targets. There is functional reconnection in mice. There is not yet functional reconnection in primates, and the path from "many axons regrew" to "the axons hit the right targets and the brain rewires meaningfully" is the hard part — vision is a topographic map, and any axon hitting "somewhere in LGN" is not the same as hitting the right column.

So the realistic decomposition of whole-eye transplant is:

Vascular and structural transplant — solved, NYU 2024.
Immune tolerance — manageable with current regimens, plus the eye's immune privilege buys margin.
Optic nerve regeneration — actively researched, mouse-stage success, primate-stage missing.
Topographic remapping — even if axons regrow, the adult brain has to make sense of an arbitrarily scrambled visual map. Some plasticity exists; whether it's enough is unknown.

The most plausible path to functional whole-eye transplant in the next 20 years is: donor eye + bridging scaffold + small-molecule cocktail (PTEN/SOCS3 modulators, c-myc, mTOR) + electrical stimulation + a long period of perceptual learning. None of those four parts is impossible; none is currently good enough alone.

Layer 6 — grow an eye from scratch (lab-grown organs)¶

Sasai's lab at RIKEN showed in 2011–2012 that mouse and human embryonic stem cells can spontaneously self-organize into optic-cup organoids — a recognizable retinal cup with stratified neural retina and RPE. These are now standard tools. Yoshiki Sasai's optic-cup paper is one of the most beautiful demonstrations of self-organization in developmental biology.

The leap from "optic cup in a dish" to "transplantable functional eye" requires: - vascularization (organoids necrose past ~1–2 mm without blood supply) - size scaling (mouse-scale to human-scale) - assembly with a host cornea, lens, sclera - a connected optic nerve that finds the brain

Xenotransplant + decellularized scaffolds (pig eye chassis seeded with human cells) is one possible scaffold. None of this is close to clinical, but the developmental-biology side of it is genuinely promising.

Engineering MORE eyes — adding a new visual organ to a human¶

This is the speculative end of the page, and it deserves to be treated honestly. Three layers of difficulty stack:

Difficulty 1 — placing a sensor¶

A camera on the back of your head is trivial as hardware. GoPros do this for $200. The real question is what happens after the camera.

Difficulty 2 — getting the signal to the brain¶

You have three options:

Sensory substitution — route the signal through an existing sense.
vOICe (Meijer 1992): camera → sonified soundscape via headphones. Trained blind users can identify objects and even recognize faces at very low resolution. The auditory cortex partially reorganizes to handle it.
BrainPort V100 (Wicab): 400-pixel tongue-stimulation array. FDA-cleared 2015. The tongue is a high-density sensor; users learn to navigate with practice.
Haptic vests, vibrating belts (David Eagleman's Neosensory work): cardinal direction (NorthPaw), audio environment for the deaf (Buzz).
What this proves: the brain can repurpose input streams from one modality to do something like vision after weeks of training. It is not "a third eye" — it is the existing brain doing more with a new channel.
Cortical stimulation — bypass the substituted-sense step and write to visual cortex directly. Same engineering as Layer 4 above, just with a second image source. This is theoretically straightforward if you've already accepted the cortical prosthesis premise, but you'd need either:
a separate cortical patch dedicated to the new eye — there is no spare cortex, you'd have to take it from somewhere
or, integration with the existing V1 in a way that doesn't corrupt the native input — extremely hard, because V1's retinotopic map is built around two eyes, not three
Peripheral nerve splicing — wire a camera's output directly into an existing optic nerve or a cranial nerve. Currently infeasible: cranial nerves carry a few thousand fibers each, and we cannot yet write a million-channel signal into one in a structured way.

Difficulty 3 — integration¶

The brain's visual system was built by 500 million years of evolution around two forward-facing eyes with overlapping fields. Adding a third channel runs into:

No spare cortex. The visual system already uses ~25–30% of human cortex. Adding a new sensor either requires repurposing cortex from another function (sensory substitution shows the cortex can do this, slowly and partially) or it remains a second-class signal that is interpreted but never "seen" in the phenomenal sense.
Binocular integration assumes two. Stereopsis, vergence, conjugate eye movements — these are wired for two. A third eye would not contribute to depth via disparity the way the other two do.
Selective attention is one-pointer. You have one attentional spotlight; a third eye gives you data, not perception.
Developmental plasticity is largely closed by ~7 years. Adult-acquired senses can be added, but they get integrated like a wearable, not like a native modality.

Tetrachromat humans (rare individuals with a fourth functional cone) are the cleanest natural experiment for "extra channel in an existing modality": most do not actually perceive richer color, because their downstream wiring is built around three. The brain can sometimes use the extra channel, sometimes not. The lesson generalizes — a new sensor without supporting brain real estate is a data stream, not an experience.

What's actually plausible in the medium term¶

Goal	Realistic path	Likely timeline
Restore some functional vision in profoundly blind	gene therapy (where applicable) + optogenetics + retinal/cortical prosthesis	now to ~10 years for steady improvement
Restore color, detail, motion in macular degeneration	RPE/photoreceptor stem-cell transplant + complement inhibitors + AAV neuroprotection	~10–20 years
Whole-eye transplant with vision	optic nerve regeneration in primates first; clinical maybe 20–30 years
Tetrachromatic / IR / UV vision in sighted humans	AAV-delivered novel opsins (proof of concept in macaques and color-blind mice exists) + brain learning to use it	~20+ years; legally and ethically restricted
A third eye on the head with cortical integration	technologically and neurologically extremely hard; the substitution route (back-camera → tongue/auditory/haptic) is the realistic version	already partially possible via sensory substitution; "real" third eye not on any near horizon
Drone-mounted "remote eye" feeding cortical implant	this is the most tractable form of "extra eye" — uses Layer 4 tech to write a new camera's output into V1	feasible in principle once cortical prostheses mature; novel but plausible 10–30 years

The unifying picture¶

The eye is the cleanest case in medicine of a sense organ that is both understood as engineering (it really is a camera-plus-neural-network) and constrained by neurobiology (the optic nerve is CNS and won't regrow on demand). Almost every solved problem in ophthalmology has come from the engineering side — replace a lens, reshape a cornea, drain some fluid, kill a vessel with a laser. Almost every unsolved problem lives in the neurobiology side — make a ganglion cell axon find its way back to LGN, persuade a transplanted photoreceptor to wire into the right bipolar cell, write a million-channel signal into a folded sheet of cortex.

That asymmetry is also the prediction. The next decade of progress in vision will be: - more gene therapy wins as AAV capsids improve and more single-gene targets clear regulatory bars, - the first generation of optogenetic vision moving from one patient to a few hundred, - stem-cell-derived RPE becoming a routine intervention in dry AMD, - a cortical prosthesis that gives a few thousand pixels of usable image to people who would otherwise see nothing, - and photoreceptor regeneration moving from proof-of-concept to small clinical wins.

Whole-eye transplant with vision and "extra eyes" are real research directions but on a 20+ year horizon, and they will follow optic-nerve regeneration as a precondition. The most important short-term lever for global vision is still less glamorous than any of this: find the people with cataracts and operate on them, find the people with diabetes and screen their retinas, find the people with glaucoma before they lose 30% of their RGCs. Most of the world's blindness is preventable today, with 19th-century optics and 20th-century surgery.

Cross-references¶

SIGNS-AND-LEVELS.md — the "one sign is noise; stacked signs are signal" framework
CANCER.md — same read-the-world logic, different organ
WEATHER.md — the same pattern in a non-medical domain

References¶

Kandel, E. R. et al., Principles of Neural Science (6th ed., 2021). Chapters 25–29: visual system wiring from retina to V1 through dorsal/ventral streams.
Hubel, D. H., Eye, Brain, and Vision (1988). Scientific American Library. Cortical vision processing; readably explains receptive fields, orientation columns, and ocular dominance.
Sahel, J.-A. et al. (2021). Partial recovery of visual function in a blind patient after optogenetic therapy. Nature Medicine 27. First clinical optogenetic vision restoration; ChrimsonR channelrhodopsin in retinal ganglion cells.
Fernández, E. et al. (2021). Visual percepts evoked with an intracortical 96-channel microelectrode array inserted in human occipital cortex. Journal of Clinical Investigation 131(23). Cortical prosthesis restoring rudimentary vision after 16 years of blindness.
He, Z. & Jin, Y. (2016). Intrinsic control of axon regeneration. Neuron 90(3). PTEN/mTOR pathways for retinal ganglion cell regeneration after optic nerve injury; central to glaucoma repair prospects.
WHO, World Report on Vision (2019). Global blindness epidemiology; 2.2 billion people with vision impairment, 1 billion preventable; public health framing for the disease prevalence section.