Big Tech Wants to Harvest Your Thoughts | WIRED
Overview
Rafael Yuste is in his early sixties and bears a more than passing resemblance to Pablo Picasso—if Picasso had worn glasses and had a trim white goatee. Speaking succinctly and methodically, his accent rich with Spanish inflections, he told me about an experiment he had carried out in his lab at Columbia on the brains of mice, and specifically on that part of the cortex that responds to vision. His mentor had been the Swedish neuroscientist Torsten Wiesel, who won a Nobel Prize for his research into how our visual systems process information.
“He discovered by chance that the strongest stimulus is a pattern of high-contrast dark and light bars.” He held up one hand and waved his fingers back and forth. “If you imagine my fingers were bars of light surrounded by complete blackness—if I move my fingers in front of your eyes, that fires up your whole visual cortex.”
Details
If you buy something using links in our stories, we may earn a commission. This helps support our journalism. Learn more.
To begin with, they used these moving images to train the mice. The bars were projected onto a computer screen in front of them, and when they moved up and down, it was a cue to take a drink from a tube of water. When they moved from side to side, they were to stop drinking. The researchers used a sophisticated laser system to monitor brain activity through the mouse’s skull—identifying exactly which neurons were firing when it was looking at the projected images. “We can see the neurons that are encoding the visual stimulus,” Yuste explains.
Having cracked this neuronal code, Yuste’s group used a second holographic laser system to project a series of points inside the mouse’s brain, with each point activating the very same neurons that represented vertical or horizontal moving bars. “The killer experiment was to turn off the screen,” Yuste says. “Just like when you are playing the piano, you use different fingers on particular keys. So, we are playing the images on the cortex. And when we play them, we make the mouse behave in the way we want it to.” When the team implanted images of bars moving up and down, the mice licked the water. When they implanted images of bars moving side to side, they stopped licking.
In effect, they had read the mind of the mouse, identified exactly what was happening in its brain when it viewed the images—and then used that data to make it see things that were not there.
“The way that the mouse licks the spout when he sees the image that we implanted is identical to when he sees the image with his own eyes. And I mean the same number of licks, the same duration of each lick, the same delay until he starts licking. So, as far as we know, he cannot tell the difference. He thinks that these things are real in front of him.”
It was a clear demonstration, Yuste said, of the power of this new technology—that they could “manipulate the mouse like a puppet” and make it do one thing, or do another, depending on which image they put into its brain.
“And what we can do in a mouse today we can do in a human tomorrow.”
Over the past two decades, researchers using functional magnetic resonance imaging (f MRI), which tracks the iron in the hemoglobin supplying oxygen to neurons, have been building up increasingly detailed maps and inventories of the mammalian cortex. Thanks to huge advances in machine-learning artificial intelligence—computer algorithms that are able to sort through enormous amounts of information and use statistical methods to make classifications and predictions—f MRI scans can now be used to identify everything from depressive thoughts to the nuanced feelings of envy and schadenfreude. Other algorithms have been able to accurately piece together reconstructions of movie clips watched by subjects, just by analyzing their brain scans; or have detected, in probing the brain activity of swing voters in the US presidential election, responding to photographs and videos of presidential candidates, which candidates provoked anxiety or even disgust, and which elicited positive responses or feelings of empathy.
In just the last few years, neuroscience researchers have progressed from decoding images and emotions as they play across the cortex to sounds, words, phrases, and even language. In 2023, in a remarkable demonstration of this emerging technology, a woman called Ann Johnson, who had been paralyzed for 18 years by a brain-stem stroke, was able to speak again through the insertion of a grid of 253 electrodes onto the surface of her brain, which translated her neuronal signals into sentences, in real time, at a rate of 78 words per minute (just about half the speed of standard conversation). The research team at the University of California, led by neurosurgeon Edward Chang, had combined this brain-computer interface with an animated avatar of Johnson’s head, which spoke in her own voice, as reconstructed from a recording of a 15-minute toast she had given at her wedding. Just as the avatar’s mouth spoke Johnson’s words as she thought them, so its expressions were similarly influenced by the nuances of her brain activity, which turned her thoughts about facial gestures into displays of emotion—from smiles to pursed lips and frowns.
“They unlocked her,” Yuste said. “They cloned her mind in a computer. Well, not her whole mind, but this language part. And when they did it, Eddie”—Chang, the study’s lead neuroscientist—“called me up and said, ‘I cannot sleep.’ Because he realized all the power and all the perils. This is incredible for patients that are paralyzed. But imagine you put this on a person for other reasons. There is great responsibility. Look what we have in our hands. We just built you a machine that can decode your language. And in 10 years, we’re going to give you a machine that can interfere with your thoughts the way we do it in mice today.”
All brain-reading technologies work on the same basic principles: They first record the behavior of neurons when a person is engaged in a particular function, such as speech, language, vision, concentration, and so on, to isolate and interpret where this behavior is happening—predominantly expressed through electrical fields, waves, or pulses—and then work out what it means.
Your mind is a prediction machine. Feed it these stories for optimal results.
The more invasive the recording equipment, the richer and more detailed the data. Surgical interventions are at the vanguard of neuroscience and remain very rare—fewer than 100 people on the planet have brain-computer interfaces like Johnson’s embedded beneath their skulls. Yet almost inevitably, a concerted trickle-down effect is occurring. In the summer of 2023, a team at the University of Texas demonstrated that they could use f MRI to translate brain scans into words and sentences, after subjects listened to 16 hours of the storytelling podcasts The Moth Radio Hour and The New York Times’ Modern Love to train an AI model. When the subjects then listened to new podcasts, the algorithm was able to convert the gist of what they heard, as it manifested in their brains, into words, phrases, and sentences that roughly captured the stories. As the team’s lead computational neuroscientist, Alexander Huth, put it in an interview with Science, “Our thought when we actually had this working was, ‘Oh my God, this is kind of terrifying.’”
Now noninvasive, wearable brain scanners are beginning to proliferate beyond the lab, making their way into our workplaces and, through the vast global consumer market, into our homes too.
In conversation with The New Yorker in 2021, Jack Gallant, a professor in cognitive neuroscience at Berkeley whose work is focused on assembling a “complete functional atlas of the human brain,” talked, in a brief aside, about a possible future technology that he described as “a thinking hat.” He imagined companies paying people $30,000 a year to wear the hat, which could incorporate video-recording glasses along with a variety of sensors, to produce brain data on everything the wearer saw, felt, heard, and experienced as they went about their everyday life. The scientific logic was obvious—just imagine the incredible volumes of novel information that a device like this could generate, particularly if your goal was to create a truly comprehensive map of every function of the brain. All the same, when I first read about this thinking hat, my instinctual reaction was a pang of dystopian horror.
You can imagine the progression—research students, many eager to pay off huge university loans, volunteering as the first test subjects. Then a steady flow out from the campus and into the gig economy: the thinking hat a no-brainer—or, rather, an all-brainer—for those already juggling multiple jobs. An extra 30 grand a year, just to think? Imponderably vast reams of data on brain activity would soon be delivered into a digital commons. People—willingly at first, but perhaps increasingly driven by necessity, or desperation, or worse—walking around, going about their daily lives, with wearable scanners constantly “mining” their brains for information. In effect, we would begin trading our neural activity to corporations and data brokers in return for money—or even simply for digital assets or website access, just as we do with our personal information and search engine preferences right now. It is a dangerous calculus. The mind, as Yuste has put it, should be a sanctuary for our identity. “You need to shield that. You cannot just go in and start banking and selling brain data.”
While this may sound alarmist, or like something from the pages of a science fiction novel, it is rooted in already existing and operational tech. There is no one universal, wearable “thinking hat”—and its true realization may still be a long way off—but there are early iterations of the concept. Most of the current devices are worn on the scalp and use a method called electroencephalography (EEG) that detects the tiny electrical fields, generated by the firing of millions of neurons in the cortex, passing through the skull. In recent years, as part of brain-computer interfaces, EEG has become a tool for assessing, and even altering, mental states—from focus and calmness to stress and drowsiness.
The neurotech company Emotiv, for instance, has been trialing a combination of an EEG headset and earbuds, which are worn by office workers to monitor, through brain activity, their levels of attention to tasks throughout the day—including showing when their focus dips or when they are distracted—along with their associated cognitive stress. The aim is increased efficiency and productivity, with the system recommending workers take breaks when stress spikes for too long and attention falls. While the initial concept is for the data to be anonymized, or accessible only by workers themselves as opposed to their employer, it is unlikely that this will remain the case for long. Especially if, as Emotiv’s CEO, Tan Le, predicts, within the next five years brain tracking in the workplace “will be quite ubiquitous.”
EEG headbands are already in use for truck drivers and miners, to warn against fatigue—a system with clear and obvious benefits for preventing workplace accidents. Rather more worrying is the example of the US company Brain Co supplying EEG “Focus” headsets to primary school children in China, to monitor their concentration levels—data that was then uploaded to the company server and accessible by the teacher but not the pupils or their parents. (When a video report on the project appeared in The Wall Street Journal, it was quickly suspended.) In Dubai, the police force has for several years now been using a technology called i Cognative, which scans the brains of suspects, using EEG to detect tiny flashes of object or information recognition. In one case, it is claimed, a man suspected of murder was shown an image of the weapon that was used, and his brain waves produced an involuntary spike. When presented with the data, he immediately confessed. His mind alone had given him up. Or, to put it another way, he had been caught by the thought police.
In the consumer market, EEG devices have predominantly been aimed at the wellness sector. There are headsets like the i Band, which its producers claim can trigger lucid dreaming; and Flow, which offers a home treatment for anxiety or depression, sending electrical pulses into the areas of the brain responsible for regulating mood, sleep, and motivation. Another EEG device, called Brain Bit, 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 without professional help in reading.” The applications range from meditation and sleep to education (“increase schoolchildren’s attention span”), business (“analyze and understand employee brain-state signals”), and online dating (“listen to your brain and swipe based on your instinctive reaction”). It even suggests that you “automatically broadcast your mood to your social media page—or add a mood indicator to a post.” (I wonder if it can capture a mood of existential despair?) One final use suggested for Brain Bit is “neuromarketing”—“collecting neuroinsight” to reveal the “subconscious reaction of consumers and enhancing results of market research.”
It is not hard to see where this is leading: advertisers and corporations delving into our subconscious to find ever more targeted ways to sell their products. In 2022, Emotiv partnered with the world’s largest cosmetics company, L’Oreal, to develop in-store EEG technology as part of personalized fragrance consultations, using neural activity to identify perfume preferences. As Emotiv put it, “We’re simplifying the decisionmaking process by connecting emotion and scent through technology.”
Tech giants Apple, Meta, and Snap are all developing their own neurotechnology products. For Meta and Snap, the aim is to use neural activity to translate thoughts into actions on a computer, “allowing you to push a virtual button simply by focusing on it.” Apple, on the other hand, has filed a patent for a device that presumably will incorporate EEG sensors into a future iteration of its already ubiquitous Air Pods, effectively giving the largest company in the world instant access to our brain activity. While these wearable EEG devices cannot currently “decode” thoughts—at least not in the fashion demonstrated by the neuroscience researchers at California and Texas—the rapid and accelerating pace of development suggests that it is only a matter of time. Already a company called Kernel has built a device (also called the Flow) that combines EEG sensors with infrared light (essentially shining a laser into the brain) and which is able to acquire neural readings of a quality approaching f MRI. Built into a solid, compact headset that looks like a BMX bike helmet, it is perhaps the closest we have come so far to Gallant’s vision of the universal “thinking hat.” It is expected that a consumer version of this technology could reach the market within a decade.
Yuste’s goal in neuroscience, he told me, had always been to root out, and ultimately find a cure for, the brain’s own natural malware. At the start of his career he had worked in a psychiatric hospital in Madrid, treating patients with brain disorders—many of them paranoid schizophrenics, and some so dangerous he had to interview them in the presence of bodyguards. This included one highly intelligent patient who had worked out where Yuste lived—gleaned from small details like his accent—and threatened to come to his home and kill his father. The experience affected him profoundly. Something in the mind of this patient was making them turn against themselves and against society, something that the doctors didn’t understand, because, fundamentally, they did not understand the brain.
When Yuste successfully inserted false images directly into the brains of mice, it was both a breakthrough and an alarm call. It offered a means to understand where and how hallucinations manifested in the brain, and potentially how to remove 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 methods that can help a schizophrenic can also be used on a normal person to reprogram their brain.”
In 2017 Yuste convened a meeting at Columbia, drawing together neuroscientists from around the world who were working at the leading edge of brain research. Joining them were clinical neurosurgeons, bioethicists, lawyers, experts from the tech industry, and pioneers in artificial intelligence.
“There was a group of 25 of us. And we holed ourselves up for three days to ponder on the ethical and social consequences of neurotechnology. And we concluded that this is a human rights problem. Because if this is not a human rights problem, what is? The brain generates the mind, and the mind is what makes us human.”
In the meeting, they coined the term neurorights—proposing the creation of five new core rights, designed to protect the brain from the use and abuse of neurotechnology. First was the right to “mental privacy,” so that data from our brain cannot be decoded, stored, shared, or sold without our explicit consent. Next was the right to “identity,” requiring the establishment of clear boundaries to stop technology from disrupting the sense of self, since if computer interfaces connect with our minds, it could blur the line between our unique consciousness and external inputs. Third was “agency”—that everyone should have control over their own decisionmaking, without unknown interference from neurotechnologies. Then came “fair access to mental augmentation.” If the use of brain-enhancing technologies was seemingly inevitable, it would have to be open to all, to prevent a fundamental schism in humanity between those who can afford implants and those who cannot. Finally, there was “protection from algorithmic bias”—an acknowledgment that machine-learning algorithms often suffer from the innate social or cultural biases of their own creators and the data that trained them. Accompanying the five rights was an ethical framework for anybody working in neurotechnology, from scientists and entrepreneurs to corporations and investors, which was modeled on medicine’s Hippocratic Oath, dubbed the Technocratic Oath.
This initial meeting at Columbia led to Yuste forming the Neurorights Foundation, a professional advocacy group working with governments, policymakers, corporations, and the United Nations to enshrine global protections against the excesses of neurotechnology. The foundation has already notched a number of significant wins. In 2021, after extensive consultation with Yuste and his team, Chile became the first country in the world to recognize neurorights through an amendment to its constitution protecting brain data and brain activity—perhaps not surprisingly, in a nation still so psychically wounded by the oppressive regime of General Pinochet, where numberless people were imprisoned, murdered, and disappeared for dissident thoughts and beliefs.
The foundation is engaged in similar consultations with governments in Brazil, Uruguay, Mexico, Argentina, Spain, and the United States. In April 2024 the state of Colorado worked with Neurorights to enact the world’s first full neurodata law, extending the protections of its existing privacy act to include brain data collected by nonmedical consumer neurotechnology devices, which, the act said, can “collect and process information about an individual that the individual did not even know existed.” California soon followed by enacting similar provisions in its own state law. At a global level, UNESCO’s International Bioethics Committee has recommended an investigation into adapting the existing human rights framework, and even possibly proclaiming new human rights, to recognize the challenges posed by neurotechnology.
For Yuste, these legal expansions are at once necessary and urgent. Technology is finally breaking through to the very core of our beings—and bringing with it the apparatus of “mining.” Extraction is arriving at the coalface of our consciousness, at the very bedrock of our thoughts. As Jared Genser, the lead lawyer for the Neurorights Foundation, puts it, “the need to protect neural data is not a tomorrow problem—it’s a today problem.” The mining of brain data has already begun, and prospectors are ranging hungrily across the landscapes of the mind. In the same month that Colorado passed its new law, Neurorights published the results of the first-ever study into the privacy practices and data policies of current consumer neurotechnologies. Analyzing 30 of the most prominent companies currently in the market, they found that all but one had unlimited access to the consumer’s brain data; two-thirds could share it with third parties; and two implied that they were already selling it to third parties.
“And if you sell something to a third party,” Yuste said, “they are not held by the original consumer agreement. You cannot be less protected. This is predatory. The companies act like this because there is no regulation. They think, ‘Let me just take possession of it all.’ It’s like the Wild West. You go in there and stake a claim for whatever you can.”
If you buy something using links in our stories, we may earn a commission. This helps support our journalism. Learn more.
Key Takeaways
- Rafael Yuste is in his early sixties and bears a more than passing resemblance to Pablo Picasso—if Picasso had worn glasses and had a trim white goatee
- “He discovered by chance that the strongest stimulus is a pattern of high-contrast dark and light bars
- If you buy something using links in our stories, we may earn a commission
- To begin with, they used these moving images to train the mice
- Having cracked this neuronal code, Yuste’s group used a second holographic laser system to project a series of points inside the mouse’s brain, with each point activating the very same neurons that represented vertical or horizontal moving bars
