How brain implants are transforming communication

By Avesta Afshari-Mehr


I ask that you imagine losing the ability to tell your family you love them, explain that you are in pain, or simply make a joke with a friend. For people living with conditions that progressively rob them of speech, this is not a hypothetical scenario. But advances in brain-computer interfaces (BCIs) are beginning to offer a remarkable alternative: communicating directly from the brain, bypassing the body’s obstructed connections.

In October 2026, ScienceNews reported on Casey Harrell, a climate activist with amyotrophic lateral sclerosis (ALS), who regained the ability to communicate through an experimental brain implant. After gradually losing his speech following his diagnosis in 2020, Harrell received four electrode devices implanted in his brain in 2023. He now uses the system for hours each day, having communicated almost two million words through it. The technology represents a significant step towards restoring communication for people whose brains can still formulate speech but whose bodies can no longer produce it.

Speaking is truly an extraordinarily complex physical process. The brain coordinates signals to more than 100 muscles involved in breathing, moving the tongue and lips and controlling the vocal cords. Conditions like ALS damage the nerve cells that control these movements, leaving people unable to speak even when they know exactly what they want to say. Brain-computer interfaces offer a way around this breakdown. Rather than relying on the signals reaching the muscles, they can intercept the process at its source.

The prospect is not merely that technology might help them speak again, it is that they might once more be heard

Harrell’s implant contains grids of electrodes positioned over his speech motor cortex, a region involved in controlling the movements required for speech. When he attempts to speak, the electrodes record patterns of electrical activity in his brain. These signals are transmitted to a computer, where machine-learning algorithms analyse them and predict the words he intends to produce.

Such a system learns to associate patterns of neural activity with phonemes, the individual sound units that make up spoken language. By combining these predictions, the software reconstructs words and sentences that appear on a screen, allowing Harrell to communicate without physically articulating them.

But this is not quite mind-reading. The system is trained to interpret specific patterns associated with attempted speech, rather than extracting any arbitrary thought from the brain. Its accuracy depends on the quality of the neural signals, the training data and the algorithms interpreting them.

The potential applications extend beyond ALS to people who have lost speech following strokes or brain injuries, as well as some individuals with other neurological conditions. Parkinson’s disease and cerebral palsy can also affect communication, although the underlying mechanisms differ and not every patient would necessarily benefit from the same technology.

Researchers are working towards interfaces that communicate faster, more accurately and more naturally. Harrell’s system still requires substantial equipment and has limitations in speed and the range of sounds it can decode. Future systems could potentially produce fluent speech in real time, incorporate facial expressions and intonation, or recreate a person’s original voice using recordings made before their illness progressed.

The wider implications reach beyond speech restoration. BCIs are already being investigated for controlling robotic limbs, computer cursors and other assistive technologies. Improved methods of interpreting neural activity could help people with paralysis interact with their environments more independently, potentially reducing barriers to education, employment and social participation.

The system is trained to interpret specific patterns associated with attempted speech, rather than extracting any arbitrary thought from the brain

Nevertheless, important challenges remain. Implanting electrodes requires brain surgery, carrying risks including infection and tissue damage. Devices must remain reliable over time, while the systems interpreting neural activity require extensive calibration and technical support. Their cost and complexity could also limit access, particularly if the technology remains available only through specialised research centres.

And there are countless ethical questions, too. Neural data is exceptionally personal, making privacy, informed consent and control over how information is stored and processed essential considerations. Researchers must ensure that people who depend on these devices retain control over their own communication.

Brain implants are not yet a universal solution for people who cannot speak, and much more research is needed before they become widely available. But their significance is already becoming clear. And for Harrell and others facing the loss of speech, the prospect is not merely that technology might help them speak again. It is that they might once more be heard.

Image: Alireza.g.qre via Wikimedia Commons