Tether Evo Advances Brain-Computer Interface Research

Tether Evo is taking a major step forward in brain-computer interface (BCI) research, with three new peer-reviewed papers accepted for publication in the Journal of Neural Engineering, Imaging Neuroscience, and Neural Networks.

Two of the studies were conducted by Tether Evo in partnership with researchers at the University of Rome Tor Vergata (UniTOV). Together, the three papers explore how artificial intelligence can interpret brain activity across different people in areas including speech, vision, and music.

The research tackles a major obstacle in BCI development: no two brains produce exactly the same neural signals. Traditionally, this has meant that systems often need extensive training and calibration for each individual. Tether Evo’s latest work suggests that a single model could instead learn patterns shared across people and then be adapted to a new user much faster.

One AI Model Could Work Across Different Brains

Tether Evo is Tether’s frontier technology division, focused on the intersection of biology and machine intelligence. Its work includes brain-computer interfaces and neuroprosthetics, with an emphasis on local-first, high-performance technology designed to support individual autonomy.

One of the papers, “Cross-subject decoding of human neural data for speech brain computer interfaces,” was accepted by the Journal of Neural Engineering. The research focuses on people who have lost the ability to speak because of conditions such as ALS, stroke, or brain injury.

BCIs can potentially help these individuals communicate by recording brain activity and converting neural signals into text.

The challenge is that speech-related brain activity can vary considerably between people. Differences can come from implantation locations, functional brain activity, and individual learning experiences.

Researchers tested a cross-subject neural-to-phoneme decoding model using invasive brain recordings from multiple participants whose implants were placed in different cortical regions.

Instead of developing a completely separate system for every patient, the researchers used shared speech patterns across participants. A lightweight mathematical realignment technique brings signals from different brains into a common space, while a new layered decoding network handles the translation.

According to the research, the approach can match or outperform existing single-patient systems while allowing a new user’s data to be adapted in minutes to hours, rather than requiring lengthy calibration.

Reconstructing What the Brain Sees

Tether Evo and UniTOV also explored whether brain activity could be used to reconstruct visual information.

Researchers recorded neural signals from macaques while the animals viewed thousands of images. The resulting model analyzed just 200 milliseconds of neural data and identified the correct image from thousands of possibilities with 70% accuracy.

It also generated a plausible reconstruction that captured key elements of the image, including its shape, colour, and content.

The paper, “A Modular Semantic-Structural Pipeline for Visual Decoding from Primate Spiking Data via Selective Temporal Integration,” was accepted at Imaging Neuroscience.

The findings could contribute to future research into visual restoration, including cortical visual prostheses and closed-loop BCIs for people experiencing vision loss.

Brain Activity Can Also Reveal Music

The third study examined how the brain responds to music.

In “R&B – rhythm and brain: Cross subject decoding of music from human brain activity,” researchers recorded fMRI brain scans from five people while they listened to 540 songs covering 10 genres.

The researchers then trained a model to translate brain activity into an AI representation of the music being heard.

Using an alignment technique designed to account for differences between people’s brains, the system identified the correct music genre around 61% of the time, compared with 10% by chance. It also identified the exact song from a group of 60 choices roughly 25% of the time, compared with less than 2% by chance.

The study also provided insights into the brain regions involved in music perception. Auditory areas associated with processing music played an important role, while classical and jazz produced some of the most distinctive neural patterns. Metal and disco, meanwhile, were more likely to be confused with other genres.

Why Faster BCI Calibration Matters

Calibration remains one of the biggest hurdles standing between BCI research and practical, real-world applications.

If a model can learn from data collected from many people and then adapt to a new individual within minutes rather than weeks, researchers could potentially shorten the path from laboratory experiments to usable BCI technology.

The same cross-person learning approach demonstrated with speech, images, and music could ultimately be applied to technologies designed to restore communication and other functions.

Privacy is another important part of Tether Evo’s broader approach. Tether says personal information should remain under individual control. Its QVAC open-source, on-device AI stack is designed to run intelligence locally and privately without requiring permission on a device.

The research reflects Tether’s wider focus on decentralized technologies, wallets, and tools intended to give individuals and communities greater control over their own technology and information.

Also read : Instagram Unveils a Sharper, More Modern Wordmark