Scientists don’t believe in Elon Musk’s dream of superior vision for people using the Blindsight implant

In the journal Nature, an article was published in which scientists from the University of Washington (UW) refuted Elon Musk’s claim about the future superior capabilities of the Blindsight implant for restoring or giving vision to blind or visually impaired people. Musk is not familiar with the workings of the brain, and vision is not a series of pixels, scientists said. Everything is much more complicated and technology is unable to rise to the level of living tissues, much less surpass them.

Image source: Computerra

In March of this year, it became known that Elon Musk’s Neuralink is developing and even testing a cortical implant called Blindsight (literally, “Blindsight”). Following the apparent success of the Neuralink brain implant to restore communication functions to patients with spinal cord injuries, the announcement of a new miracle implant to restore vision has been met with enthusiasm. Loss of vision is a fairly common occurrence in medical practice, not to mention congenital defects of this kind. Vision often deteriorates steadily as people age, and this is now perceived as a necessary evil.

On March 20, 2024, Musk posted a short message on his social network X account about the company’s next product, the Blindsight implant. “The [artificial vision] resolution will be low at first, like early Nintendo graphics, but could eventually exceed normal human vision,” Musk later explained in X. Scientists at the University of Washington latched on to the phrase and tried to figure out if this was true.

Patterns of electrode arrays and how cat videos will be processed in the visual cortex. Image Source: Fine, I & Boynton, G/CC By 4.0

To study the possibilities of an artificial stimulus to the cortical cells of the visual cortex of the brain – without the participation of vision and channels for transmitting signals directly through the nervous tissue to the brain – a simple computer model was used as “virtual patients”. Scientists have shown that the position of engineers that the brain receives signals like pixels on a matrix or display screen does not correspond to the real state of affairs. The simulation clearly showed how a patient with a certain number of pixels would see, for example, a cat.

In a regular image of a cat from an array of 44 thousand pixels, the animal is visible with sufficient quality. But if the same array irritates neurons in the visual cortex of the brain, then the situation changes dramatically for the worse. In the video, scientists showed what it would look like. One can only guess about the presence of a cat in the frame.

The problem, the researchers explain, is that the image is fed to neurons with overlap and encoding. Each neuron responsible for vision receives a complex (encoded) signal from an array of sensory cells in the eyes – the so-called receptive field. Also, information reaches more than one neuron. Simply stimulating brain cells is not the same as normal vision and is unlikely to ever approach, let alone surpass, the normal quality of human vision. Even millions of “pixels”—the high resolution of a neuron-stimulating implant—won’t solve the problem of encoding visual information to above-average quality.

Scientists have not ruled out a breakthrough in this area, but urge caution when talking today about returning people to excellent vision. This can create unjustified expectations that are dashed by reality and leave patients vulnerable.