The Race to Protect Our Minds Before Neurotechnology Unlocks Them
In his early 60s, Rafael Yuste bears more than a passing resemblance to Pablo Picasso—if Picasso had worn glasses and sported a neatly trimmed white goatee. Speaking in concise, methodical tones, his voice thick with Spanish inflections, he walked me through a landmark experiment he ran out of his Columbia University lab, focused on the brains of mice, specifically the region of the cortex that processes visual input. Yuste trained under legendary Swedish neuroscientist Torsten Wiesel, who won a Nobel Prize for unpacking how the human visual system processes information.
“Wiesel discovered by chance that the strongest stimulus for the visual cortex is a pattern of high-contrast alternating dark and light bars,” Yuste explained, holding up one hand and waving his fingers back and forth. “If you picture my fingers as light bars against total darkness—when I move them in front of your eyes, that fires up your entire visual cortex.”
Buy this book here: If you purchase a product through links in our articles, we may earn a small commission. This helps support our independent journalism. Learn more.
To start, the team trained mice using these moving bar patterns. The bars were projected onto a screen in front of the animals: when the bars moved up and down, that was a cue for the mice to take a drink from a water spout; when they moved horizontally side to side, the mice were trained to stop drinking. The researchers used an advanced laser system to track brain activity through the mouse’s intact skull, pinpointing exactly which neurons fired as the mice watched the projected images.
“We can clearly see which neurons are encoding the visual stimulus the mouse is seeing,” Yuste explains. Once the team cracked this neural code, they used a second holographic laser system to project targeted points of stimulation directly into the mouse’s brain. Each point activated the exact same set of neurons that had fired when the mouse saw vertical or horizontal moving bars in real life.
“The game-changing part of the experiment was when we turned off the screen entirely,” Yuste says. “It’s like playing piano: you use different fingers to press specific keys to make the sound you want. We were playing images directly onto the mouse’s cortex. And when we did that, we got the mouse to behave exactly how we wanted it to.”
When the team activated the neurons corresponding to up-and-down moving bars, the mice immediately licked the water spout. When they activated the neurons for side-to-side movement, the mice stopped licking. In effect, the team had read the mouse’s mind, mapped exactly what brain activity looked like when the animal viewed specific images, then used that data to make the mouse “see” things that weren’t actually there.
“The way the mouse licks the spout when it sees the image we implanted is identical to when it sees the real image with its own eyes,” Yuste says. “I mean identical: the same number of licks, the same duration for each lick, the same delay before it starts licking. As far as we can tell, the mouse can’t tell the difference. It thinks these images are really in front of it.”
Yuste says the experiment was a clear demonstration of what this new technology can do: the team could “manipulate the mouse like a puppet” and make it behave one way or the other, just by loading different images into its brain. “What we can do in a mouse today, we will be able to do in a human tomorrow,” he says.
The Rapid Rise of Brain-Reading Technology
Over the past two decades, researchers have used functional magnetic resonance imaging (fMRI)—which tracks iron in the hemoglobin that carries oxygen to active neurons—to build increasingly detailed maps and inventories of the mammalian cortex. Thanks to massive leaps in machine learning artificial intelligence (computer algorithms that sort through enormous datasets and use statistical analysis to make classifications and predictions), fMRI scans can now identify everything from depressive thoughts to nuanced feelings like envy and schadenfreude.
Other algorithms have accurately reconstructed clips of movies that research subjects watched just by analyzing their brain scans; they have also detected which U.S. presidential candidates triggered anxiety or even disgust, versus which prompted positive reactions or empathy, by scanning the brain activity of swing voters as they viewed candidates’ photos and videos.
In just the past few years, neuroscience researchers have moved far beyond decoding images and emotions playing out across the cortex, and can now decode sounds, words, phrases, and even full language. In 2023, a stunning demonstration of this emerging technology gave Ann Johnson, a woman who had been paralyzed for 18 years following a brain stem stroke, the ability to speak again. Surgeons inserted a grid of 253 electrodes onto the surface of Johnson’s brain, which translated her neural signals into full sentences in real time, at a rate of 78 words per minute—roughly half the speed of normal conversational speech.
The research team at the University of California, led by neurosurgeon Edward Chang, paired this brain-computer interface with an animated avatar of Johnson’s face, which spoke in her own voice, reconstructed from a 15-minute recording of a wedding toast she had given decades earlier. As the avatar’s mouth spoke Johnson’s words exactly as she thought them, its facial expressions also shifted to match the nuances of her brain activity, turning her intentions for facial gestures into visible emotion, from smiles to pursed lips and frowns.
“They unlocked her,” Yuste says. “They cloned part of her mind in a computer—well, not her whole mind, but her language center. When they pulled it off, Eddie—that’s Chang, the lead neuroscientist—called me and said, ‘I can’t sleep.’ Because he suddenly understood both the incredible power and the terrible danger this technology brings. This is life-changing for paralyzed patients. But just imagine if you put this technology in someone’s brain for other reasons. There is enormous responsibility here. Look what we hold in our hands: we just built a machine that can decode your language. And in 10 years, we will have a machine that can interfere with your thoughts the same way we interfere with a mouse’s thoughts today.”
All brain-reading technologies operate on the same basic principle. First, they record neural activity when a person is carrying out a specific function, like speaking, seeing, focusing, or understanding language, to isolate and map where this activity occurs—most often detected through electrical fields, waves, or pulses—then decode what that activity means. The more invasive the recording equipment, the richer and more detailed the data it collects.
Surgical implants are currently at the cutting edge of neuroscience, and they remain extremely rare: fewer than 100 people worldwide have brain-computer interfaces like Johnson’s implanted under their skulls. But just as with many new technologies, a steady trickle-down to mainstream use is all but inevitable. In summer 2023, a team at the University of Texas showed that they could use fMRI to translate brain scans into words and sentences, after training an AI model by having subjects listen to 16 hours of narrative storytelling from the podcasts The Moth Radio Hour and The New York Times’ Modern Love. When subjects then listened to new podcast episodes, the algorithm was able to convert the gist of what they heard—captured through their brain activity—into words, phrases, and sentences that roughly matched the original content.
As the team’s lead, computational neuroscientist Alexander Huth, told Science, “When we got this to actually work, our first thought was, ‘Oh my God, this is kind of terrifying.’”
The Coming Mainstream of Wearable Brain Tech
Now noninvasive, wearable brain scanners are starting to move out of the lab and into our workplaces, and through the global consumer market, into our homes too. Speaking to The New Yorker in 2021, Jack Gallant, a cognitive neuroscience professor at UC Berkeley whose work focuses on building a “complete functional atlas of the human brain,” mentioned offhand a possible future technology he called a “thinking hat.”
He imagined companies would pay people $30,000 a year to wear the hat, which would pair recording glasses with a suite of sensors to collect brain data on everything the wearer sees, feels, hears, and experiences in their daily life. The scientific rationale is clear: just imagine the massive volume of new data a device like this could generate, especially for researchers aiming to build a truly comprehensive map of every brain function.
Even so, when I first read about Gallant’s thinking hat, my immediate reaction was a jolt of dystopian horror. It’s easy to see how this would play out: research students, many eager to pay off massive student loans, would volunteer as the first test subjects. Then the technology would move off campus and into the gig economy, where the thinking hat would be an obvious choice—an all-brainer, no-brainer—for workers already juggling multiple jobs. An extra $30,000 a year just to go about your life and think?
Unfathomably huge amounts of brain activity data would quickly end up in the hands of corporate data holders. People would go about their daily lives wearing wearable scanners that constantly “mine” their brains for information—at first willingly, but over time increasingly driven by necessity, desperation, or coercion. In effect, we would start trading our neural activity to corporations and data brokers in exchange for money, or even just for access to websites or digital services, just like we do with our personal data and search preferences today. That’s a dangerous trade-off. As Yuste puts it, the mind should be the ultimate sanctuary of our identity. “You need to protect that. You can’t just let companies bank and sell brain data.”
While this might sound alarmist, or like a plot from a science fiction novel, it is rooted in existing, fully operational technology right now. There is no single mass-market “thinking hat” yet, and a full realization of that concept may still be decades away, but early versions of the idea are already on the market. Most current consumer devices are worn on the scalp and use electroencephalography (EEG), a method that detects tiny electrical fields generated by firing millions of cortical neurons as they pass through the skull.
In recent years, EEG has become a tool for brain-computer interfaces that can assess, and even alter, mental states, from focus and calm to stress and drowsiness. For example, neurotech company Emotiv has been testing a combination EEG headset and earbuds worn by office workers to track their focus levels throughout the workday via brain activity. The device notes when focus drops or workers get distracted, and also tracks associated cognitive stress. The stated goal is to boost efficiency and productivity: the system recommends workers take breaks when stress stays high and focus drops for too long.
While the original plan calls for data to be anonymized, or only accessible to workers themselves rather than their employers, it’s unlikely that this arrangement will last. That’s especially true if, as Emotiv CEO Tan Le predicts, workplace brain tracking “will be quite ubiquitous” within the next five years.
EEG headbands are already used by truck drivers and miners to warn against fatigue—a system that has clear, proven benefits for preventing workplace accidents. Far more concerning is the case of U.S. company BrainCo supplying EEG “Focus” headsets to primary school students in China, to track their concentration levels in class. The data was uploaded to the company’s servers and accessible to teachers, but not to the students or their parents. (The project was quickly suspended after The Wall Street Journal published a video report on the program.)
For several years now, Dubai’s police force has used a technology called iCognative, which scans suspects’ brains using EEG to detect tiny involuntary spikes in activity when a person recognizes an object or piece of information. Officials claim that in one murder case, a suspect was shown an image of the murder weapon, and his brain waves produced an involuntary recognition spike. When presented with the evidence, he immediately confessed. His own mind gave him away. Or, put another way: he was caught by the thought police.
In the consumer market, EEG devices are mostly targeted at the wellness sector. There are headsets like iBand, which its makers claim can trigger lucid dreaming; and Flow, which offers at-home treatment for anxiety and depression by sending electrical pulses to the regions of the brain that regulate mood, sleep, and motivation. Another EEG device called BrainBit promises to help you “use your brain like never before,” with a headband that “monitors brain activity and gathers data that can be converted and used for reading without professional help.”
Its advertised uses range from meditation and sleep tracking to education (“increase schoolchildren’s attention span”), business (“analyze and understand employee brain-state signals”), and even online dating (“listen to your brain and swipe based on your instinctive reaction”). It even suggests users “automatically broadcast your mood to your social media page—or add a mood indicator to a post.” (I have to wonder if it can detect a mood of existential despair?)
One final use BrainBit advertises is “neuromarketing”: “collecting neuroinsight” to reveal the “subconscious reaction of consumers and enhancing results of market research.” It’s not hard to see where this is heading: advertisers and corporations digging into our subconscious to find ever more targeted ways to sell us their products. In 2022, Emotiv partnered with L’Oréal, the world’s largest cosmetics company, to develop in-store EEG technology for personalized fragrance consultations, using neural activity to identify a customer’s perfume preferences. As Emotiv put it, “We’re simplifying the decisionmaking process by connecting emotion and scent through technology.”
Big tech companies including Apple, Meta, and Snap are all developing their own neurotechnology products. For Meta and Snap, the goal is to use neural activity to turn thoughts into actions on a device, “allowing you to push a virtual button simply by focusing on it.” Apple, meanwhile, has filed a patent for a device that would add EEG sensors to a future version of its already ubiquitous AirPods, effectively giving the world’s largest company direct access to our brain activity any time we wear them.
While these wearable EEG devices can’t currently “decode” full thoughts—at least not at the level demonstrated by researchers at UC and the University of Texas—the rapid, accelerating pace of development means it’s only a matter of time. Already, a company called Kernel has built a device (also named Flow) that combines EEG sensors with infrared light (essentially shining a laser into the brain) and can collect neural readings of a quality close to what fMRI provides. Shaped like a solid, compact headset that looks like a BMX bike helmet, it’s probably the closest we’ve come so far to Gallant’s vision of a universal “thinking hat.” Industry analysts expect a consumer version of this technology could hit the market within a decade.
The Fight for Neurorights
Yuste told me his goal in neuroscience has always been to track down and ultimately cure what he calls the brain’s natural malware. Early in his career, he worked at a psychiatric hospital in Madrid, treating patients with brain disorders—many of them paranoid schizophrenics, and some so dangerous he had to conduct interviews with bodyguards present. One patient, a highly intelligent man, had figured out where Yuste lived from small clues like his accent, and threatened to go to Yuste’s home and kill his father. The experience left a deep mark on Yuste. Something in this patient’s brain was turning him against himself and society, and doctors couldn’t explain what it was, because at a fundamental level, they still didn’t understand how the brain works.
When Yuste successfully implanted false images directly into mice’s brains, it was both a major breakthrough and a warning call. It opened a path to understanding where and how hallucinations form in the brain, and potentially how to eliminate them. “I thought, oh my God, we can really help schizophrenics now. We can go into the brain and reprogram their cortex, maybe we can cure them. But at the same time, the same methods that can help a schizophrenic can be used on a healthy person to reprogram their brain,” he says.
In 2017, Yuste organized a meeting at Columbia that brought together leading neuroscientists from around the world working at the cutting edge of brain research. They were joined by clinical neurosurgeons, bioethicists, lawyers, tech industry experts, and artificial intelligence pioneers. “There were 25 of us. We holed up for three days to think through the ethical and social consequences of neurotechnology,” Yuste says. “And we concluded that this is a human rights issue. If this isn’t a human rights problem, what is? The brain generates the mind, and the mind is what makes us human.”
At that meeting, the group coined the term neurorights, and proposed five new core rights designed to protect the human brain from the misuse and abuse of neurotechnology:
Mental privacy: Brain data cannot be decoded, stored, shared, or sold without explicit, informed consent.
Right to identity: Clear boundaries must be established to stop technology from altering a person’s innate sense of self, as brain-connected interfaces could blur the line between unique human consciousness and external inputs.
Right to agency: Every person must have full control over their own decision-making, without hidden interference from neurotechnology.
Fair access to mental augmentation: If brain-enhancing technology becomes inevitable, it must be open to all, to prevent a fundamental split in humanity between those who can afford enhancements and those who cannot.
Protection from algorithmic bias: Recognition that machine learning algorithms often carry the innate social and cultural biases of their creators and training data, and people must be protected from harm caused by that bias.
Alongside the five rights, the group created an ethical framework for everyone working in neurotechnology, from scientists and entrepreneurs to corporations and investors. Modeled after medicine’s Hippocratic Oath, it is called the Technocratic Oath.
That initial 2017 meeting led Yuste to found the Neurorights Foundation, an advocacy group that works with governments, policymakers, corporations, and the United Nations to enshrine global protections against the misuse of neurotechnology. The foundation has already notched several major wins. In 2021, after extensive consultations with Yuste and his team, Chile became the first country in the world to formally recognize neurorights via a constitutional amendment protecting brain data and brain activity. That move is perhaps not surprising in a nation still psychically scarred by the oppressive regime of General Pinochet, where countless people were imprisoned, murdered, or disappeared for holding dissident thoughts and beliefs.
The foundation is now holding similar consultations with governments in Brazil, Uruguay, Mexico, Argentina, Spain, and the United States. In April 2024, the U.S. state of Colorado worked with the Neurorights Foundation to pass the world’s first full neurodata privacy law, extending the protections of its existing privacy act to cover brain data collected by non-medical consumer neurotechnology devices. The law notes that these devices can “collect and process information about an individual that the individual did not even know existed.” California soon followed, passing similar provisions in its own state law.
At the global level, UNESCO’s International Bioethics
