A Machine That Reads the Language of the Mind: The New Return of Lost Voices

In the 1990s, French journalist Jean-Dominique Bauby became almost fully paralyzed after a brainstem stroke. Though unable to speak or write, he told his remarkable story using an extraordinary method—a helper would repeatedly recite the alphabet, and he would choose the correct letter by blinking his left eye. The tale of such patience and hardship has come to us as the book The Diving Bell and the Butterfly. But today, three decades later, technology has advanced immensely. Those who have lost their ability to speak due to ALS (Amyotrophic Lateral Sclerosis) or brainstem stroke can now communicate much faster and more comfortably—sometimes with just their eye movements, sometimes with subtle muscle twitches. At the forefront of these advances lies the brain-computer interface, or BCI technology.

Recently, scientists have taken it a step further—now, not even the effort to speak is necessary; the computer can transcribe what you simply think in real time. This new “inner speech” mind-reading technology could fundamentally change how we communicate. Until now, BCI devices have relied on “attempted speech,” or the effort to vocalize, which could often be exhausting for users. These devices pick up signals from sensors implanted in the motor cortex and translate those into words. Users had to simulate behaviors like taking breaths and moving facial muscles as if to speak, even though no sound came out. Due to shortness of breath or physical limitations, it could take several breaths to get out a single word. Unintended facial expressions or sounds could also cause discomfort for the user.

But the motor cortex doesn’t just activate during efforts to speak aloud; some parts also become active during imagined speech or silent self-talk. Erin Kunz and her colleagues at Stanford University have used this trait to develop a new “inner speech” decoder. In a recent study published in Cell journal, four participants—three ALS patients and one person affected by a brainstem stroke, all of whom already had sensors implanted in their motor cortices—took part. The trials showed that by merely thinking a sentence, it would instantly appear on the screen. Where earlier technology was limited to vocabularies of just a few dozen words, the new system could utilize a dictionary of 125,000 words.

The greatest breakthrough has come in terms of speed. While before, users could only utter a handful of words per minute, the new technology allows for 120 to 150 words per minute—almost as fast as natural conversation. And this requires only composing the sentence mentally, without any extra effort using facial muscles or breathing. One participant jokingly said that, because it used to be so slow, they could never interject in a conversation, but now they can finally do so again.

However, this technology is not effective for everyone. As Alexander Huth, a researcher at Berkeley, explains, the normal process of speaking begins with forming an idea in the mind, which then converts into a plan to move the necessary muscles, and finally, that plan is executed. In many patients, the planning and preparation stage remains intact, but the step converting the plan into muscle movements is damaged. The new “inner speech” technology is able to capture that first step, so those who can successfully transform ideas into speech plans will benefit most from it.

Kunz’s relationship to this research is deeply personal as well. Her father lost his ability to speak due to ALS. In his final days, Kunz became her father’s personal “speech translator” because she understood his language best. This experience propelled her into brain prosthetics research. So she knows firsthand the value of regaining a lost voice.

Of course, the ability to read the mind raises questions about mental privacy. For this reason, the research team has incorporated a special “code phrase”—such as “chitty chitty bang bang”—which, when thought of, will start or stop the device from transcribing. This allows users to control when they want to express their thoughts.

According to researchers, the greatest contributions to this research have come from the participants themselves. Though they may not benefit instantly, they have willingly given their time and effort to bring about change in the lives of people with paralysis in the future. In Kunz’s words, “They are extraordinary people, and their contributions to the development of this technology are unique.”

This innovation is not merely a milestone for medical technology; it is also reconstructing our idea of human connection. In the future, we may reach a world where voices are no longer needed for conversation—thoughts alone will reach others directly. But with that will come complex ethical and privacy questions, for which scientists and society will have to seek answers together. The more advanced the technology becomes, the greater the challenge will be to maintain this balance.