Blindsight explained: FDA Breakthrough Device designation, visual cortex stimulation and evidence limits

How cortical vision restoration differs from retinal implants and motor BCIs

Most blindness therapies target the eye itself—gene therapy for retinal photoreceptors or retinal implants for patients with an intact optic nerve. However, when trauma, glaucoma, or optic nerve atrophy destroys the connection between the eye and the brain, signals can no longer travel down the optic nerve.

Even in complete bilateral blindness, neurons in the primary visual cortex (V1) at the back of the occipital lobe remain organized in a spatial map of the visual field (a retinotopic map). For decades, neuroscientists have known that passing a tiny, charge-balanced electrical current through an electrode in V1 causes a person to perceive a localized flash or spot of light in darkness, known as a phosphene. Blindsight aims to pair an external camera mounted on glasses with high-channel V1 implants so software can translate camera scenes into patterns of phosphene stimulation.

What FDA Breakthrough Device Designation means—and what it does not

On 17 September 2024, Neuralink announced that Blindsight received Breakthrough Device Designation from the U.S. Food and Drug Administration (nl_blindsight). Because headlines frequently conflate Breakthrough status with FDA approval, here is the exact regulatory distinction (fda_breakthrough):

Preclinical evidence shown in the Summer 2025 update

In Neuralink's 1-hour Summer 2025 Update (FASMejN_5gs), engineering leads presented the status of Blindsight testing in non-human primates (macaques). In behavioural tasks where monkeys normally move their eyes (saccade) toward a physical point of light on a screen, stimulating specific V1 electrodes with a Blindsight implant caused the animal to saccade reliably toward the corresponding virtual location in visual space.

Official @neuralink Video 27 Jun 2025 · 59:59 Tier 1, 2 & 3 · Flagship 1-Hour Progress Keynote

Neuralink Update, Summer 2025

Click to load the privacy-enhanced youtube-nocookie.com player. No third-party cookies or requests are sent until you press play.

▶ Load embedded video (59:59) Watch on YouTube ↗ Read video analysis →

Key bioengineering challenges before human cortical vision scales

• Charge-injection safety and electrode longevity: Stimulating brain tissue repeatedly can corrode metal electrode contacts or damage neurons if charge density limits are exceeded. Stimulating electrodes require specialized surface coatings and strict charge balancing.
• Cortical folding in the calcarine sulcus: Much of human central vision in V1 lies folded deep inside the calcarine fissure rather than flat on the outer surface of the occipital lobe, requiring longer threads and careful vascular navigation.
• Perceptual fusion vs. pixel grids: Stimulating 1,000 electrodes simultaneously does not automatically look like a 1,000-pixel digital photograph; phosphenes can interact non-linearly, requiring adaptive encoding models calibrated with human feedback.

To see how Blindsight fits into our four-tier evidence framework, read Future possibilities organized by evidence tier.