Two entirely separate research breakthroughs, published weeks apart in 2025 and 2026, have moved the science of artificial vision further in a single year than the preceding decade managed to achieve.
One bypasses the damaged eye entirely and talks directly to the brain.
The other teaches the brain to talk back.
On May 7, 2026, Illinois Institute of Technology announced that a third blind patient had received the Intracortical Visual Prosthesis (ICVP), a wireless brain implant developed over nearly three decades that skips the retina and optic nerve altogether and delivers electrical signals directly to the visual cortex, the brain region responsible for processing sight.
The ICVP features 34 wireless stimulators and 544 electrodes implanted permanently into the visual cortex, with a camera mounted on a pair of glasses capturing the scene and translating visual information into signals transmitted directly to the brain.
Meanwhile, a study published in Science Advances in November 2025 by researchers at Miguel Hernández University in Spain introduced something the field had never had before: a closed-loop visual neuroprosthesis that does not just stimulate the brain but listens to how the brain responds, and adjusts accordingly.
The system establishes a direct dialogue with the visual cortex, and in two blind volunteers, this approach safely induced stable visual perceptions enabling recognition of shapes, movement, and letters.
Both systems are early-stage.
Neither is available to patients yet.
But together they represent a meaningful shift in how researchers are thinking about what is possible, and for whom.
There are currently 43 million blind people globally, with an additional 295 million living with moderate to severe visual impairment.
For most of them, once the retina or optic nerve fails, medicine has had almost nothing to offer.
These devices are beginning to change that calculation.
What makes the ICVP different from everything before it
Before understanding what the ICVP does, it helps to understand why blindness has been so difficult to treat technologically.
Most existing visual implants, including the now-discontinued Argus II and the more recent PRIMA chip, work by stimulating the retina or the cells immediately adjacent to it.
That approach works for a specific category of patient: people who have lost the photoreceptor cells that convert light to neural signals, but still have functional downstream pathways, meaning the optic nerve and visual cortex remain intact and capable of processing information.
Many patients with total blindness, however, have no light perception vision because their retinas or optic nerves are damaged or absent entirely, making retinal stimulation impossible and leaving direct visual cortex stimulation as potentially the only advanced sensory option available.
The ICVP was designed specifically for this population.
The ICVP system bypasses the eyes and optic nerve completely, targeting the visual cortex directly and making it the first intracortical visual implant to use a group of fully implanted miniaturized wireless stimulators.
That last word matters: wireless.
Previous cortical visual implants required external cables running through the skull to connect the implanted electrodes to external hardware.
Cables create infection risk, limit mobility, and introduce mechanical vulnerabilities every time the patient moves.
The ICVP’s permanently implanted design gives researchers and recipients extended time to explore how the device can work effectively, and allows the recipient to learn how the device can be useful over a long follow-up period.
The third implantation, performed at Rush University Medical Center in Chicago, was described by the surgical team as demonstrating the scalability and robustness of the system across multiple patients.
One successful implantation could be an outlier.
Three, in succession, using the same surgical protocol, is evidence of a reproducible procedure.
How the ICVP works
A small camera is embedded in a conventional-looking pair of glasses.
The camera captures the visual scene in front of the patient in real time.
That visual information is processed and converted into a pattern of electrical signals.
Those signals are transmitted wirelessly to the 34 implanted stimulators sitting on the surface of the visual cortex.
The stimulators deliver electrical pulses to 544 electrodes in contact with the cortex, creating patterns of neural activation that the brain begins to interpret as elementary visual percepts, small points of light called phosphenes that can combine to form simple shapes.
The brain does not passively receive these signals like a television receiving a broadcast.
It learns from them.
With repeated use, patients’ brains develop the ability to interpret the patterns of phosphenes more efficiently, a process analogous to how the brain learns to interpret other sensory inputs it has not encountered before.
The clinical goal at this stage is not photographic vision.
The clinical phase will test whether the implant provides patients with improved ability to navigate and perform basic, visually guided tasks, which represents a meaningful threshold of functional independence for people living with complete blindness.
Being able to detect where a door is.
Navigating a corridor without a guide.
Reaching for an object placed on a table.
These are the targets, and for someone who has lived in total darkness for years, they represent an enormous quality-of-life shift.
The breakthrough hiding inside the Spanish research
Two weeks before the third ICVP implantation made headlines, a separate finding from the same Spanish research team that developed the UMH neuroprosthesis added an unexpected dimension to the conversation.
During a clinical trial of cortical electrical stimulation, one of their blind patients, a person with complete blindness caused by irreversible optic nerve damage, began to partially recover natural vision after participating in the study.
The improvement was spontaneous, sustained over time, and independent of the implanted device itself, observed during a study designed to evaluate safety and feasibility rather than to test vision recovery.
The researchers were not expecting it.
They had not designed the trial to produce it.
The suggestion embedded in that result is striking: electrical stimulation of the visual cortex may not just create artificial perception, it may in some cases reactivate dormant biological circuits that were suppressed by chronic sensory deprivation rather than destroyed by the original injury.
That possibility is speculative.
It requires much more investigation before any clinical conclusions can be drawn.
But it opens a question that was not on anyone’s research agenda before: could regular stimulation of the visual cortex, delivered as a form of neural rehabilitation, help recover residual natural vision in patients who were presumed to have none?
How the closed-loop system changes the game
The UMH neuroprosthesis, described in the November 2025 Science Advances paper, addresses a problem that has quietly limited every cortical visual implant that came before it.
Until now, all visual neuroprostheses have been open-loop systems, meaning they did not take into account how neurons respond to electrical stimulation at all.
That is a more significant limitation than it might initially sound.
The brain is not a static receiver.
When you stimulate a neuron electrically, the neuron adapts.
Its threshold for activation changes.
The surrounding neurons reorganize their connectivity.
The cortical map of what electrical patterns correspond to what perceptions shifts over time.
An open-loop system delivers the same stimulation pattern regardless of what the brain does in response.
It cannot adjust when neurons adapt.
It cannot compensate when the relationship between stimulus and perception drifts.
Over time, open-loop stimulation becomes less reliable and less perceptually meaningful as the brain changes around it.
The closed-loop approach developed at UMH enables real-time two-way communication between the brain implant and the visual cortex, allowing adaptation of stimulation based on neural responses, and through this dynamic interaction participants were able to recognize complex patterns, movements, shapes, and even some letters.
The system reads what the visual cortex is doing after each stimulation pulse, and adjusts its next output based on that response.
It treats the brain as an active partner in the communication, not a passive endpoint.
As Professor Eduardo Fernández Jover explains, vision is not a passive process: it is a constant exchange of information between the eyes and the brain, and artificial systems must also reproduce this feedback loop to better mimic how the visual system truly functions.
The assumption this field needs to challenge
Here is the part of this story that the headlines almost always miss.
The conventional framing of brain implants for vision typically presents the technology as a gift flowing one direction: from the device to the brain.
The device stimulates.
The brain receives.
The patient sees.
But the neuroscience of vision does not work that way, and the UMH closed-loop system is the first clinical device that takes that seriously.
The visual cortex is not a screen.
It is a prediction engine.
Natural vision works because the brain continuously generates predictions about what it expects to see, based on context, memory, and prior experience, and compares those predictions to the incoming sensory data.
When the incoming signal matches the prediction, it is processed efficiently.
When it does not, the brain recalibrates.
This predictive, bidirectional architecture is why vision is so fast, so robust, and so resilient to noise.
A device that stimulates the brain without listening to its responses is fundamentally misaligned with how the visual system processes information, which is why open-loop systems have always produced perceptions that are less stable, less reliable, and harder to interpret than natural vision.
The closed-loop approach does not just improve a technical parameter.
It aligns the artificial system with the biological logic of visual perception in a way that no previous device has done.
That alignment is why the UMH volunteers could recognize letters, not just blobs of light.
Letters require fine spatial discrimination and pattern recognition.
That kind of perception requires a brain that is actively engaged in interpreting the signal, not just passively receiving stimulation.
A convergence of approaches arriving at the same time
The ICVP and the UMH neuroprosthesis are not the only developments in artificial vision arriving in rapid succession.
In January 2026, Science Corporation’s PRIMA retinal implant had results published in the New England Journal of Medicine showing that in 38 patients with geographic atrophy due to age-related macular degeneration, the device enabled reading of letters, numbers, and words for the first time since vision loss.
The NEJM study demonstrated a mean improvement of 25.5 letters on the ETDRS letter chart and found that 84% of patients regained the ability to read letters, numbers, and words, with no decline in patients’ existing peripheral natural vision.
In March 2026, Belgian biotech ReVision Implant received FDA Breakthrough Device Designation for its visual cortex prosthesis, with a first human clinical trial scheduled for October 2026 and broader early-stage trials planned for summer 2027.
ReVision’s CEO noted that while cochlear implants have transformed treatment for hearing loss, there remains no widely available neuroprosthetic solution for restoring vision, and their aim is to provide functional vision that improves independence and quality of life for people with severe blindness.
And in October 2025, researchers at University College London published results from the AI-powered PRIMA chip in AMD patients, showing that pairing the implant with augmented reality glasses and artificial intelligence processing allowed patients to read again after losing central vision to macular degeneration.
What these parallel developments have in common is a convergence toward the same architectural principle: bypass what is broken and connect directly to what still works.
When the retina fails, stimulate the next layer.
When the optic nerve fails, bypass it entirely.
When the brain’s own feedback mechanisms are available, use them.
Each approach reaches the visual cortex through a different route, and each one is extending the range of patients who could potentially benefit.
What the surgery actually involves
One of the most remarkable engineering details of the ICVP is the surgical delivery system.
The entire ICVP implantation can be done through an opening just 8 to 10 millimeters wide, avoiding the need for a full craniotomy, which means patients can be discharged early and experience far less postoperative discomfort.
A surgical robot and advanced neuronavigation system guide electrode insertion with millimetre-level precision in real time.
The 34 stimulators are placed on the surface of the visual cortex with their electrodes making contact with the tissue, transmitting pulses wirelessly from an external coil worn near the head.
There are no transcutaneous wires.
No external hardware physically connected to the skull.
The system is designed to function continuously, adapting as the brain changes.
And because the stimulators are permanently implanted rather than temporarily placed, the research team can track how the brain responds to stimulation across months and years, a timeline that was impossible with earlier cable-tethered systems that could only be used in clinical settings.
Who this technology is actually for
The ICVP is specifically designed for people with no-light-perception blindness, meaning people who cannot detect even the presence of light, which affects a significant proportion of the blind population and for whom retinal and optic nerve-based implants offer nothing.
Causes include severe glaucoma that has destroyed the optic nerve, traumatic injury to the visual pathway, certain rare genetic conditions, and late-stage retinal degenerative diseases where the downstream neural pathway has also been compromised.
The UMH system is designed for a similar population: people who once had sight and lost it due to retinal or optic nerve disease, whose visual cortex remains capable of processing visual information if a signal can reach it.
In people who are congenitally blind, the visual cortex never fully develops its capacity to process visual information in the way it does in sighted individuals, which is why the current generation of cortical implants is targeted at people who once had sight rather than those who were born without it.
That distinction defines an addressable population that is both large and currently without meaningful options.
What comes next
Neither system is available to patients today.
The ICVP is in an ongoing clinical trial, actively seeking new participants, with the current phase focused on establishing that the implant reliably produces useful visual guidance across multiple patients over extended periods.
The UMH closed-loop system has demonstrated proof of concept in two volunteers, and the team’s next challenge is refining stimulation protocols, improving the resolution of the artificial perceptions, and extending testing to a larger cohort.
Both research teams are clear that what exists today is a foundation, not a product.
Professor Fernández Jover emphasizes that although the results are highly promising, many challenges remain, and it is essential to advance carefully and avoid creating false expectations, as this is still ongoing research.
The challenges are real.
Increasing the resolution of cortical stimulation, so that patients can perceive complex scenes rather than simple shapes, requires a density of electrodes that current manufacturing and surgical techniques are still working toward.
Long-term biocompatibility, ensuring that implanted hardware remains stable and effective across years or decades of use, is a significant engineering challenge.
And the learning curve for patients, who must train their brains to interpret novel artificial signals, requires time and rehabilitation support that existing healthcare systems are not yet designed to provide.
Philip Troyk, who has led the ICVP project for nearly three decades, describes the third successful implantation as demonstrating the immense possibilities of neurotechnology and the potential of translating decades of research into real-world applications that can enhance lives.
The human visual system took hundreds of millions of years of evolution to reach its current state.
Artificial systems that interface with it are necessarily approximate, limited, and incomplete relative to what biology achieved.
But approximation, when the alternative is total darkness, is not a small thing.
For the person who can navigate a room, recognize a shape, or read a letter for the first time in years, the gap between what the device provides and what natural vision provides is far less important than the gap it closes between darkness and light.
That gap, for the first time in history, is finally narrowing from both sides at once.
Sources:
MassDevice. ReVision Implant wins FDA breakthrough nod for vision-restoring BCI. March 9, 2026.
BrightFocus Foundation. The latest developments in retinal implants. September 2025.
