How a single-celled alga, a pile of frog eggs and a psychiatrist's frustration gave science a light switch for the brain — and won Peter Hegemann, Georg Nagel and Karl Deisseroth the Nobel Prize.
Start with a mouse that has just eaten.
By every measure biology has, it is full. There's food in its stomach, and the hormones in its blood are saying enough. It has no reason in the world to eat again.
Then someone flips a switch.
Within minutes the mouse is eating. Not nibbling: eating, voraciously, like an animal that's been starved. Nothing has changed in its stomach or its blood. What has changed is that about 800 nerve cells at the base of its brain are being flashed with blue light through a thread of glass. Those 800 cells, it turns out, produce the drive to eat. Switch them on and a full mouse is ravenous. Switch them off and the urge fades.
The cells obey the light because they've been given one extra gene. It isn't a mouse gene. It comes from Chlamydomonas, a single-celled green alga with no brain, no nerves and no eyes, which uses that gene to swim toward the sun.
That experiment came from Scott Sternson's lab at the Howard Hughes Medical Institute in 2011. Thousands like it have followed. Over the past twenty years, researchers have used flashes of light to make mice thirsty, to wake them from sleep, to calm their anxiety, to make them attack a rubber glove and to make them relive a specific memory. They've used other colours of light to silence the very same cells.
The method is called optogenetics. This morning in Stockholm, the Nobel Assembly at Karolinska Institutet gave the 2026 Nobel Prize in Physiology or Medicine to the three people who made it possible.
Born 1954, Münster. Humboldt University of Berlin. Spent fifteen years asking how a single-celled alga reacts to light so fast, and guessed the answer a decade before he could prove it. The work began at the Max Planck Institute for Biochemistry in Martinsried and was finished at the University of Regensburg.
Born 1953, Weingarten. University of Würzburg. Put the alga's genes into frog eggs and human cells, showed they made light-gated ion channels, and said in print that this could become a tool. The work was done at the Max Planck Institute of Biophysics, Frankfurt.
Born 1971, Boston. Stanford University and HHMI. A psychiatrist who put the algal gene into neurons, made them fire on command with flashes of light, and took the method into living, behaving animals.
Official citation: “for their discoveries concerning light-gated ion channels and optogenetics” · SEK 12 million, shared equally
Here is the story with the jargon stripped out. Two German scientists found a protein inside a pond alga that almost nobody believed could exist: a molecular door that swings open when light hits it. An American psychiatrist put that door into nerve cells and turned it into a switch. Together they gave neuroscience what it had wanted for a century, which was a way to stop merely watching the brain and start asking it questions.
The story has a hypothesis nobody believed and two genes fished out of a Japanese database. It has a frog-egg rig in Frankfurt, a cold email, a recording made at 1 a.m. in a borrowed lab, and a rejection from Science. It also has a question of credit that the Nobel's rule of three could never fully settle. It begins with a problem that had stumped neuroscience for a hundred years.
For most of the twentieth century, neuroscientists could watch the brain, but they couldn't really touch it.
That doesn't mean they sat still. They worked out which lumps of tissue mattered for what: language here, movement there, a deep knot of cells for hunger and thirst. Mostly they did it the hard way. They studied people with strokes and wounds, destroyed small pieces of animal brains to see what changed, and lowered electrodes to listen to the crackle of cells firing. Later, scanners let them watch whole regions brighten as volunteers thought and felt.
The Nobel Committee compares the resulting picture to a blurry photograph. Researchers knew roughly where things happened. They didn't know which cells did them, and they could almost never prove it.
The trouble was the oldest problem in science. Suppose a group of cells fires every time a mouse is frightened. Those cells might be causing the fear, or reacting to it, or just sitting nearby doing something else. Watching can't tell you which. To find out, you have to intervene: switch the cells on and see whether fear appears, then switch them off and see whether it goes away.
That's where the brain becomes a nightmare. It holds around 90 billion nerve cells, each wired to thousands of others. Cells with completely different jobs are packed together like guests at a crowded party, and next-door neighbours often have opposite jobs. Try it yourself:
An electrode is fast, because nerve cells signal in milliseconds and so can it. But its current fires everything near the tip. Years later, David Anderson's lab at Caltech tried electrically stimulating a tiny region of the mouse hypothalamus to provoke aggression, and nothing happened. When they used light on the same spot, male mice attacked within seconds. (More on that later.) A drug can be aimed at one type of receptor, but it seeps in over minutes, lingers for hours and goes wherever the blood goes. And cutting out a piece of brain is precise about nothing and permanent about everything.
What researchers wanted was something as fast as an electrode, as specific as a gene and as reversible as a light switch. In 1979, Francis Crick had co-discovered the double helix a quarter-century earlier and then, restless, turned to the brain. He wrote that neuroscience needed exactly this: a way to switch all the neurons of just one type on or off while leaving the others alone. Twenty years later he went a step further and guessed what the switch might be.
Light is fast and can be aimed. It doesn't linger, either: switch it off and it's gone. Crick's problem was that brain cells don't respond to light. Why would they? They live in total darkness inside the skull. Somebody would have to make them care.
People tried. In 1971 a Bell Labs laser physicist named Richard Fork used a laser to fire nerve cells of the sea slug Aplysia, which are faintly light-sensitive on their own. In 1983 researchers stained leech neurons with a light-sensitive dye and zapped them. Both approaches tended to damage the cells. In 2002, Gero Miesenböck's lab achieved the first genetically targeted light control of mammalian neurons by borrowing three proteins from the eye of a fruit fly and installing them together. It worked, and it was a landmark. But it was a Rube Goldberg machine. It needed several genes and a chemical relay inside the cell, and its response was too slow and too unpredictable to keep up with real neural signals. Ehud Isacoff, Richard Kramer and Dirk Trauner came up with another clever approach: they bolted a synthetic light-sensitive chemical onto an engineered ion channel. It was fast, but the chemical had to be delivered into the tissue from outside.
The wish list was getting sharper. One gene, one protein: a photon goes in and an electrical current comes out, instantly. No helpers, no add-ons, no chemistry set.
Almost nobody in neuroscience suspected that such a protein existed. A biochemist in Germany had suspected it for years, and he had been hunting for it inside an alga.
Chlamydomonas reinhardtii is about as humble as life gets.
It's a single cell about a hundredth of a millimetre across. Two whip-like flagella at its front beat about sixty times a second in something like a breaststroke, pulling it along at roughly ten body lengths per second. By one Max Planck Society estimate, scaled to body size, that's about twelve times faster than an Olympic breaststroker. It lives in soil and fresh water and photosynthesizes, so like anything solar-powered it cares a great deal about light. Stir a few million of them into a dish and the water turns faintly green. Light one side, and the green drifts toward the lamp.
It finds the light with an eyespot, an orange dot near the cell's equator that works as a minimalist eye. The orange comes from stacked layers of pigment granules that act as a tiny mirror. They bounce light arriving from the front back onto the cell membrane and shade it from light arriving from behind. The alga spins about twice a second as it swims a corkscrew path, so the eyespot sweeps round and round like a lighthouse in reverse. The signal pulses with each turn, and the cell steers toward the brighter side.
Peter Hegemann came from a family of doctors: both his parents, both grandfathers and his brother. He went another way. At his classical high school in Aachen he loathed translating ancient accounts of Roman and Greek wars, played flute in the orchestra, loved motorcycles and idolized polar explorers like Nansen and Amundsen. He has called his school and university record absolutely average.
He studied chemistry, then biochemistry in Munich, and did his doctorate at the Max Planck Institute for Biochemistry in Martinsried, just outside the city, under Dieter Oesterhelt. In 1971, Oesterhelt and Walther Stoeckenius had discovered bacteriorhodopsin, a purple protein that lets a salt-loving microbe turn sunlight directly into a flow of protons across its membrane. It's a gorgeous molecule. A single protein wraps around a molecule of retinal, a form of vitamin A and the same light-catcher your own eyes use, and turns photons into moving ions with no help at all. Hegemann's thesis was on its cousin, halorhodopsin. Scientifically speaking, he grew up in the house of light-driven proteins. Oesterhelt, whom he calls his only real academic teacher, later described him as incredibly tenacious.
Then came a detour. In 1984 Kenneth Foster, a physicist at Syracuse University, had shown that a blind mutant of Chlamydomonas could be made to swim toward light again by feeding it retinal. So the alga "saw" with a rhodopsin, the same family of retinal-based light-catchers. Hegemann went to Syracuse for a year to work with him. His plan, he later told an interviewer, was to purify the alga's rhodopsin, and he thought that would take a year. In his words, "it took me 15 years, actually."
Back in Martinsried, with his own small group, he tried the biochemistry first. In his own blunt summary, the biochemical work failed. So with a graduate student, Hartmann Harz, and advice from an electrophysiologist at the neighbouring Max Planck Institute of Psychiatry, he switched to measuring the electricity. Soviet researchers had already recorded light-triggered electrical signals in another green alga. Harz and Hegemann used a suction pipette on cell-wall-less mutants of Chlamydomonas to record the current that flows when the eyespot is hit by a flash. The current was fast. Absurdly fast. It appeared within about half a millisecond.
To see why that's strange, think about your own eye. When a photon hits a rod cell in your retina, it sets off a bucket brigade. The light-catching rhodopsin activates a G-protein, which activates an enzyme, which destroys a messenger molecule, which makes ion channels in the membrane close. It's a magnificent amplifier, sensitive enough to register a single photon, but every hand-off takes time. Getting from light to an electrical signal takes at least ten milliseconds. The alga was doing it more than twenty times faster.
Hegemann concluded there was no time for a bucket brigade. In the early 1990s, he and his colleagues proposed something that sounded slightly mad to most specialists: in Chlamydomonas, the light sensor and the ion channel were the same protein. Light would hit it, and it would simply open.
The skepticism was reasonable. Biologists knew of channels that opened in response to voltage, to chemicals, to stretch and to heat. Not one was known to open in response to light all by itself. As far as anyone could tell, nature didn't make them.
Proving it meant isolating the protein, and that's where the decade went. The alga carries more than twenty different light sensors, at least eight of them rhodopsins, and the right one was scarce, unstable and fell apart when pulled out of its home in the eyespot. In 1995 the group did clone a rhodopsin from the alga. It turned out to be the wrong one. Hegemann moved to a professorship at the University of Regensburg and kept going. Algal light-sensing, he has noted wryly, was neither the most cited nor the best-funded corner of biology. Failure, he likes to say, is a researcher's daily bread.
The break came from a direction nobody could have planned. In 2000, Japanese researchers at the Kazusa DNA Research Institute published about 26,000 short Chlamydomonas gene sequences: snippets of the genes the cell was actively using. In Regensburg, a young scientist in Hegemann's lab, Suneel Kateriya, went fishing in them for anything that resembled bacteriorhodopsin, the protein Hegemann had grown up with. He caught two.
They named the genes channelopsin-1 and channelopsin-2. Look closely at the name. An "opsin" is the protein part of a rhodopsin. "Channel" was Hegemann's hypothesis, written into the label before anyone had proved it. It was a bet, filed in a public gene database.
The database was public, so the race was on. John Spudich's lab in Houston found the same two genes, called them sensory rhodopsins A and B, and showed that silencing them in the alga knocked out the eyespot's light-triggered currents. They published on 11 June 2002. Spudich's team noticed how fast one of the currents was and suggested the protein might be a channel itself, or wired directly to one. Tetsuo Takahashi's group in Japan found the sequences too, and showed the proteins sat right in the eyespot. Three labs on three continents were circling the same answer. One gene went by three sets of names.
But suspecting that a protein is a channel isn't the same as proving it. To prove it, you have to take the gene out of the alga and put it into a completely unrelated cell that has never had any use for light. Then you have to show that this cell now responds to light, with nothing else from the alga to help. That takes a particular kind of expert. Hegemann knew one. They had met in Frankfurt years before.
Georg Nagel's instrument of choice was the egg of the African clawed frog, Xenopus laevis.
A frog egg is a giant among cells. It's more than a millimetre across, visible to the naked eye, dark brown on top and pale cream underneath. It's also wonderfully obliging. Inject it with the genetic recipe for almost any protein and the egg will build that protein for you, and if it's a membrane protein, the egg will install it in its own surface. Then you can push two fine electrodes into the egg, hold its voltage steady and measure every charged particle that crosses its membrane. For a biophysicist who wants to know what one kind of protein does, it's a clean room the size of a poppy seed.
Nagel took a winding road to the frog eggs. He was the eldest of five children and fell seriously ill at five. While his siblings went skiing, he stayed home building a steam engine and radios from science kits. Many summer holidays were spent in surgery for a bone infection at the university hospital in Würzburg, the city of what he calls his "hospital youth". He studied biology and biophysics at the then-new University of Konstanz, where his first biophysics lecturer, Ernst Bamberg, paid him as an assistant. Then, not feeling ready for a doctorate, he stepped off the track. He taught at a Swiss private school, co-founded a café in downtown Konstanz called Café Chaos and took up hang-gliding.
Bamberg pulled him back. Nagel did his PhD with him at the Max Planck Institute of Biophysics in Frankfurt, learning to record the tiny currents of single membrane proteins. One of the people who came through Bamberg's lab to run experiments was a PhD student from Munich named Peter Hegemann. After postdoctoral years at Yale and Rockefeller University (where he worked on CFTR, the chloride channel that fails in cystic fibrosis), Nagel came back to Frankfurt in 1993. In 1995 he got bacteriorhodopsin pumping protons inside frog eggs. Looking back, he has written that this could have been the start of optogenetics, but the time wasn't ripe. When Hegemann needed someone to test his algal genes, there was really only one call to make.
Nagel injected the first gene, channelopsin-1, added retinal and shone green light on the eggs. The protein was produced only in tiny amounts, but a current flowed in as protons poured through the membrane. He trimmed the gene down to its core, the part that threads through the membrane seven times, and it still worked. There was no mysterious partner and no relay: the protein was the channel. The light-gated ion channel that nature supposedly didn't make existed after all. Nagel gave it its permanent name, channelrhodopsin-1, and in June 2002 he, Hegemann, Bamberg and colleagues published it in Science. Spudich's paper had appeared just seventeen days earlier.
Then they tried the second gene, and for what was coming it was much better. Channelrhodopsin-2 answered to blue light rather than green, and its currents were roughly ten times bigger. It didn't just pass protons. It let through a broad range of positively charged ions, the same kinds of ions that make nerve cells fire. And it was quick: in the frog egg, the channel opened within about 0.2 milliseconds of a flash. With the light held on, the current spiked and then settled to a steady plateau. With the light off, the door swung shut.
Then came the experiment that, in hindsight, contains the whole revolution. Nagel and Hegemann put the channelrhodopsin-2 gene into human embryonic kidney cells and into hamster kidney cells, cells that had never had the slightest use for light. With the gene, they became light-sensitive: flash them with blue light and they produced an electrical signal.
Their 2003 paper in the Proceedings of the National Academy of Sciences said the quiet part out loud. The final sentence of its abstract said that channelrhodopsin-2 could be used to depolarize small or large cells, setting off electrical activity in them, "simply by illumination." The paper's sponsor at the journal was Walther Stoeckenius, the co-discoverer of bacteriorhodopsin, so the family of light-driven proteins had come full circle.
Read that again with Crick in mind. One gene, one protein: light in, current out, in a fraction of a millisecond, in any cell you put it in.
They'd seen it even earlier. On 11 April 2002, before the Science paper was even out, the Max Planck Society filed a patent naming Hegemann, Nagel and Bamberg as inventors. It already imagined nerve cells controlled by light. It even imagined restoring sight by putting channelrhodopsins into the surviving cells of a blind retina. Hold on to that thought for Chapter 7.
The tool existed, and it was sitting in a journal. What it needed now was a nerve cell, and someone with a reason to try.
Karl Deisseroth wanted to be a writer.
He was born in Boston in 1971, finished high school at sixteen and went to Harvard on a scholarship intending to major in creative writing. He switched to biochemistry and got into Stanford's MD-PhD programme at twenty. While doing his doctorate, he kept taking a fiction class two nights a week at a community college. He never really stopped writing. In 2021 he published Projections, a book about human emotion told partly through his patients' stories, written late at night after the children were in bed.
His PhD, in the lab of the neuroscientist Richard Tsien, was about how electrical activity in brain cells reaches into the nucleus and changes which genes are switched on. He was planning to become a neurosurgeon. Then, during medical school, a four-week rotation in psychiatry rearranged his life.
What got him was the gap. He has written about a brilliant young student with psychotic depression, and about an older woman so depressed that she couldn't smile at her grandchildren. A cardiologist, he has pointed out, can show patients the damaged muscle in their heart. A psychiatrist has nothing to point to. There was no physical explanation for what was happening to these people. And the treatments he could offer, in the Nobel Committee's words, were rarely effective and often came with debilitating side effects. He became a psychiatrist anyway, and he still sees patients, specializing in treatment-resistant depression and autism in adults.
That gave him a researcher's version of the same frustration. You can't understand depression by studying a slice of brain in a dish. You need a living brain with its circuits intact, doing what brains do. And you need to be able to intervene precisely. Others were chasing Crick's light switch, but many of them, the Committee notes, worried about wasting years and money on what looked like an unrealistic dream. Deisseroth decided failure was a risk worth taking, because his patients were already living with the status quo.
He didn't start alone. Back in 2000, he and Ed Boyden, then a graduate student in Tsien's lab, had held late-night brainstorming sessions about controlling neurons with light. They even ordered a clone of halorhodopsin, the light-driven pump from salt-loving microbes, to try. That went nowhere at the time. In February 2004, Boyden suggested they contact Georg Nagel. In March, Deisseroth emailed Frankfurt asking for the channelrhodopsin-2 gene. Nagel sent it, along with versions fused to a yellow fluorescent tag, so you could see which cells had taken it up, and one crucial tip.
The tip was about retinal. Like every rhodopsin, channelrhodopsin needs its molecule of retinal to catch light; in the frog eggs, Nagel had simply added it. Would neurons need it too? Nagel had noticed that they might not, because mammalian cells seemed to carry enough retinal of their own. If that held, a single gene would be the entire kit, with no chemicals to infuse into a living brain. The Nobel Committee singles out this "fortuitous observation" as part of why the method could spread so fast.
The worry was that a protein from an alga would poison delicate nerve cells. On 1 July 2004, as he was setting up his own lab, Deisseroth put the gene into cultured rat neurons himself. The cells made the protein and seemed perfectly happy. Two weeks later, a molecular readout suggested they were responding to blue light. Then came the real test. Would light make a neuron fire an actual action potential, the electrical spike that is the basic unit of everything a brain does?
Boyden spent July debugging the optics on a borrowed rig in Tsien's lab during off-hours. At about one o'clock in the morning on 4 August 2004, he patched an electrode onto a glowing neuron and flashed it with blue light.
The very first cell he tried fired in time with the light. Not roughly, and not after a delay: one flash, one spike.
Feng Zhang, a first-year chemistry graduate student, then packaged the gene into a lentivirus, a disabled virus that ferries genes into cells, and that made the whole system far more reliable. A 2013 profile in Nature says Zhang wandered into Deisseroth's basement lab looking for someone else (the door still bore the name of the physicist Steven Chu, its previous occupant) and was talked into staying.
They wrote the work up and sent it to Science in April 2005. Science turned it down. According to Boyden, the reason was that the team hadn't discovered anything about the brain: they had merely shown that channelrhodopsin-2 worked well in neurons. It's worth sitting with that "merely" for a moment. The paper went to Nature, which passed it to its sister journal Nature Neuroscience, and it was published online on 14 August 2005. Its authors were Boyden, Zhang, Bamberg, Nagel and Deisseroth. The title promised millisecond-timescale, genetically targeted optical control of neural activity, and the paper delivered it.
The idea's time had come, and others were close behind. Within about a year, independent groups had channelrhodopsin-2 working in chick spinal cord and rat neurons (Stefan Herlitze and Lynn Landmesser, together with Hegemann). Nagel and Alexander Gottschalk had it in the muscles and neurons of tiny worms, which twitched and recoiled when lit. Hiromu Yawo's group in Japan showed it in neurons too, and Zhuo-Hua Pan at Wayne State University showed it in the retinas of blind mice. Pan had been working on the idea independently, and early; his paper bounced between journals and only appeared in 2006. Science is full of people who were nearly first.
A neuron in a dish is a proof of principle. A neuron in a living, behaving animal is a science.
In 2006, Deisseroth and colleagues, among them Gero Miesenböck, gave the field a name in a review in the Journal of Neuroscience: optogenetics, optics plus genetics. The basic recipe hasn't changed since.
Attach the channelrhodopsin gene to a genetic "address", such as a promoter or a Cre-dependent switch in a specially bred mouse, that is only read in one cell type. Every other cell ignores the instruction. This gives you Crick's specificity.
A harmless, engineered virus carries the gene into the chosen brain region. Within weeks the target cells are studding their membranes with algal light-gated channels, and they need no added chemicals, because the brain already has enough retinal.
A thin optical fibre goes in through a small hole in the skull, coupled to a laser or LED. Pulse it, and only the tagged cells respond, on a millisecond schedule you control. This gives you Crick's timing.
In 2007, Deisseroth's lab built what they called an optical neural interface: a laser diode coupled to an optical fibre, slipped into the brain through a small hole in the skull. They put channelrhodopsin into neurons of the mouse motor cortex that control the whiskers, switched on the light, and the whiskers moved. The same year, working with Luis de Lecea, they put channelrhodopsin into a small cluster of hypocretin (orexin) neurons in the hypothalamus, cells suspected of controlling wakefulness, and flashed them while mice slept. The mice woke up roughly half a minute sooner than controls. The result showed that these cells' firing could actually cause waking, rather than just accompany it.
But an on-switch is only half a tool. To really interrogate a circuit you need an off-switch as well: silence these cells and see what stops. That one came from the salt flats. Halorhodopsin, from the archaeon Natronomonas pharaonis, is a cousin of Oesterhelt's bacteriorhodopsin. Yellow light makes it pump chloride ions into a cell, which pushes the neuron's voltage down and keeps it quiet. In 2007, two papers showed halorhodopsin silencing neurons. One came from Feng Zhang and colleagues in Deisseroth's lab, the other from Xue Han and Ed Boyden, by then running his own lab at MIT. They appeared about two weeks apart; Zhang later said his team hadn't known of the MIT work until they'd sent their own paper off. Because halorhodopsin answers to yellow and channelrhodopsin to blue, you could now put both into the same animal. In worms, blue light made the muscles contract, while yellow light made the worms go limp and stop swimming.
Then the toolbox exploded. In 2012, a team led by Osamu Nureki in Tokyo, working with Hegemann and Deisseroth, solved the atomic structure of a channelrhodopsin. That let researchers re-engineer the protein deliberately: faster versions, versions that stay open after a single flash, versions that let negative ions through instead of positive ones. Others went prospecting in nature, screening the genes of more than a hundred species of algae, and came back with channels tuned to other colours. Red-shifted channels reach deeper into tissue. In 2015, John Spudich's lab found natural light-gated chloride channels in another alga, and they silence neurons far more efficiently than the old pumps. Today there is a light-driven switch for most of the visible spectrum.
And then everyone started using it. One reason is easy to forget: the tool was almost embarrassingly cheap and simple. A gene, a virus and a fibre were things any neuroscience lab could obtain. Deisseroth's lab shared its constructs widely and trained visiting scientists, and by the mid-2010s thousands of research groups worldwide were using optogenetics. The literature shows the curve.
Once you have a switch, you can ask the brain yes-or-no questions. Does this kind of cell cause that behaviour? Flip it and find out.
Some of the most haunting answers came from memory. For more than a century, scientists have had a name for the physical trace a memory leaves in the brain: the engram, a set of cells that changed when the memory was laid down. Nobody had ever caught one in the act. The problem was access. Among millions of cells in the hippocampus, how do you find the few that hold this particular memory, and then control only those?
In 2012, Susumu Tonegawa's lab at MIT came up with an answer. Tonegawa had won the 1987 Nobel Prize for work on antibodies and then, unusually, switched to the brain. His team, led by Xu Liu and Steve Ramirez, used mice with a genetic trick. When a drug was removed from their diet, any neuron that became active would start making channelrhodopsin. In other words, they could tag the cells that were busy while a memory was being formed. Then they could switch those cells back on with light, anywhere and at any time.
Light alone, in a box where nothing bad had ever happened, made the mice freeze as if they were back in the place where they'd been shocked. It was the first time anyone had shown exactly which cells are sufficient to call up a specific memory. The next year Ramirez and Liu went further and attached the fear of a shock to a place where the mice had only ever been safe. They had planted a false memory. By one later account, the first successful run came on Christmas Eve, 2012.
In memoriamXu Liu, first author of the 2012 paper, died in 2015 at the age of 37, just after leaving to start his own lab. Susumu Tonegawa died on 11 July 2026, three months before this prize was announced. The Nobel Committee's scientific background refers to him as "the late Susumu Tonegawa."
A deeper lesson runs through everything optogenetics has touched. Behaviours that look like single things turn out to be built from separate, switchable parts. And what defines a part is often not where a cell sits but where its wires go. Nobody could see that with an electrode. With light, it showed up everywhere:
Shining light only on the endings of amygdala neurons where they reach a neighbouring region (BLA→CeA) calmed mice, which spent more time in exposed, open spaces. Silencing that same connection made them more anxious. The twist: stimulating all the cells in the region did the opposite. The wiring mattered more than the location.
Dopamine neurons in the ventral tegmental area were long lumped together as reward cells. Driving their input from one brainstem region (the laterodorsal tegmentum) made mice prefer a place. Driving input from another (the lateral habenula) made them avoid it. Which wire was active decided the meaning.
Electrically stimulating a tiny hypothalamic region (VMHvl) failed to provoke attack. Light did. Within seconds, male mice attacked females and castrated males, which they normally leave alone, and some went for an inflated rubber glove. Attacks often stopped the moment the light went off. Later work found that gentle stimulation of the same cells drove mounting and stronger stimulation drove attack: one continuum.
One small population of "parenting" neurons (galanin cells in the medial preoptic area) turned out to split into pools by where their wires go. One pool drives grooming of pups. Another drives the motivation to reach them, so that even virgin females crossed a barrier to get to pups. A third dials down interest in other adults.
The team put a light-sensitive channel (ChRmine) into mouse heart muscle and fitted the mice with a tiny vest carrying an amber LED. That let them pace the heart from about 600 to 900 beats a minute without touching the brain. The mice became more anxious, but only in risky situations, and the effect ran through the insular cortex. Deisseroth had wondered about this since his residency.
With a new opsin (ChRmine) and holographic laser beams, the team activated about 20 hand-picked neurons in the visual cortex that normally respond to a striped pattern. Trained mice behaved as if they had seen the stripes. In deep cortical layers, sometimes just a couple of neurons were enough. Deisseroth's follow-up question: if so few cells can create a perception, why aren't we hallucinating all the time?
Sensory cells lining the gut, called neuropod cells, signal the vagus nerve one way for sugar and another way for an artificial sweetener. Silencing them with light erased the mice's preference for real sugar. Doing it meant inventing a flexible optical fibre, because rigid glass punctured the gut.
Deep brain stimulation helps many Parkinson's patients, but nobody knew exactly what it acted on. In parkinsonian rodents, silencing the target region's own cells did nothing. Driving the axons arriving there at high frequency restored movement. That points to the fibres passing through as the key players.
Two intermingled cell types in the striatum had long been theorized to act as "go" and "stop" signals. Light settled it. Driving the indirect-pathway cells made mice freeze and move sluggishly, like a Parkinson's patient. Driving the direct-pathway cells got parkinsonian mice moving again.
The patient was a 58-year-old Frenchman who had been diagnosed with retinitis pigmentosa about forty years earlier.
Retinitis pigmentosa is an inherited disease that slowly kills the retina's light-sensing cells, the rods and cones. What it leaves behind is the interesting part. The ganglion cells, the neurons whose fibres form the optic nerve, are still there and still wired to the brain. They just never hear that light has arrived.
A team led by José-Alain Sahel in Paris and Pittsburgh and Botond Roska in Basel, working with the company GenSight Biologics, injected one of the man's eyes with a virus carrying ChrimsonR. It's a red-shifted relative of channelrhodopsin, and its parent protein was found by Ed Boyden's lab. Blue light bright enough to drive the channel would damage the retina, so the team used amber. The patient was given goggles with a camera that turn the world into pulses of amber light, projected onto his retina.
It took months for the gene to take effect, and about seven more months of training before the man noticed something new on the street: he could see the white stripes of a pedestrian crossing. In the lab, wearing the goggles, he reached for and touched a notebook placed on a table in 36 of 39 tries. He found a smaller box of staples about a third of the time and counted tumblers correctly in nearly two out of three attempts. Electrodes on his scalp picked up activity in his visual cortex when an object was in front of him. Without the goggles, he could do none of it. This wasn't normal sight. It was a coarse, high-contrast glimpse of the world. But it was the first time optogenetics had partially restored a human sense, and it was the use Hegemann, Nagel and Bamberg's patent had imagined nineteen years earlier.
The clinical road has been bumpier than the science. The first-ever human optogenetics trial, launched in 2015 with the original channelrhodopsin-2, ended with its sponsor reporting that efficacy was not demonstrated. GenSight's own programme has been delayed. Others pressed on. On 9 September 2026, less than a month before the prize, the US Food and Drug Administration accepted Nanoscope Therapeutics' application to approve MCO-010, an optogenetic gene therapy for retinitis pigmentosa that targets bipolar cells and needs no goggles. A decision is expected in 2027. In Göttingen, Tobias Moser's team is developing an optical cochlear implant. Light can be focused far more tightly than electric current, so the implant could stimulate the auditory nerve through many more independent channels than today's electrical implants. First human trials are currently planned for 2029.
Georg Nagel, reached this morning on a sunny terrace in Italy, said he hadn't expected the prize. Friends who follow the Nobel announcements every October had been egging him on for years, and he always told them to keep cool. As far as he knew, he said, only one patient had so far benefited from optogenetics, and he had assumed it was still too early. For now, the method's biggest payoff in medicine is knowledge rather than treatment. The circuits it has exposed in depression, anxiety, addiction, Parkinson's disease and pain are the maps that tomorrow's drugs, stimulators and gene therapies will be drawn on.
The Nobel Foundation's statutes say a prize may be divided among no more than three people. Optogenetics had a lot more than three parents.
There is Dieter Oesterhelt, who found the first microbial rhodopsin in 1971 and trained Hegemann. He shared the 2021 Lasker Award with Hegemann and Deisseroth and died in 2022; the Nobel is not given posthumously. There is Ernst Bamberg, Nagel's mentor, a co-author on the 2002, 2003 and 2005 papers and the third inventor on the 2002 patent. He shared the 2010 Wiley Prize and the 2013 Brain Prize with Hegemann and Nagel. There is Gero Miesenböck, who achieved the first genetically targeted light control of neurons in 2002. There are John Spudich and Tetsuo Takahashi, who found the algal genes in parallel. There is Zhuo-Hua Pan, who was nearly first in the retina. There is Feng Zhang, who built the lentivirus and later helped turn CRISPR into a gene-editing tool. And there is Ed Boyden: co-dreamer of the idea in 2000, the man at the rig at 1 a.m., first author of the 2005 paper, and the scientist whose lab found Chrimson, the red-shifted channelrhodopsin family later adapted for the first blind patient.
Credit in this story has always been a little tender. Boyden has written that he regrets the 2005 paper never acknowledged that about half of its channelrhodopsin experiments were done in Tsien's lab. A 2013 Nature profile reported that his relationship with Deisseroth had cooled after he left for MIT, and that neither was keen to discuss the details. Deisseroth's accounts describe a team he formed; Boyden's describe a collaboration the two had planned together since 2000. Both accounts are on the record.
The prize committees of the world have been drawing the line in different places for more than a decade. Look at who got called, and when:
Read down the columns and the logic of each committee shows. The Wiley Prize in 2010 went to the three biophysicists who found and patented the channels. The Lasker jury in 2021 honoured the microbial-rhodopsin lineage: Oesterhelt, Hegemann and Deisseroth, but not Nagel. The Shaw Prize in 2020 went to Miesenböck, Hegemann and Nagel, but not Deisseroth. The Breakthrough Prize went to Boyden and Deisseroth. Count along the rows and Boyden appears in six of these thirteen, one more than Nagel. Stockholm drew its line around a single, clean narrative arc: the discovery of light-gated ion channels (Hegemann and Nagel) and their transformation into a tool for the brain (Deisseroth). The Committee's own scientific background, it should be said, names Boyden, Zhang, Bamberg, Miesenböck, Spudich and Takahashi in its account.
Science is a relay race. Prizes are a photo finish.
At 12:27 a.m. Pacific time, a phone rang at Karl Deisseroth's bedside, and he missed it. Seconds later his wife's phone rang, and a voice from Stockholm asked whether they might speak with Karl.
He told the Nobel Foundation afterwards that for about thirty seconds he had trouble forming words. Later that morning he made sandwiches for his children. Asked how he planned to celebrate, he said he would get back to work.
In Germany, Peter Hegemann found an email asking him to call back. He assumed it was a salesman, he told German television, until he noticed the Swedish number. He was overwhelmed and a little disbelieving. The hype around optogenetics had died down, he said, and you stop expecting it. What pleased him most was that Nagel and Deisseroth were in it with him. Asked what had carried him from an absolutely average student to Stockholm, he named persistence, and the willingness to keep going when older colleagues pat you on the shoulder and say: sure, try it, plenty of people have. In Italy, Georg Nagel took the news on a terrace overlooking olive trees.
The Nobel Committee's press release starts the story not with the brain, or with a disease, or with Crick's wish, but with one person's curiosity: Hegemann wondering how a single-celled alga can swim toward the light.
Think about what that means. Trace the chain backwards from the blind man seeing crosswalk stripes. Before him came a virus carrying a red-shifted channel; before that, a channel found by screening algal genomes; before that, mice whose memories could be switched on with light. Before those came a whisker that twitched, a rat neuron firing at 1 a.m. in a borrowed lab, and a kidney cell that responded to blue light. Before that, a frog egg; before that, two gene fragments in a Japanese database; before that, a puzzling half-millisecond in an alga's eyespot; and before that, a purple protein in a salt-loving microbe, found in 1971.
For most of that chain, nobody involved was trying to cure anything, or even to study the brain. Imagine a grant committee in 1991 choosing between "how does a pond alga swim toward light?" and "restore sight to the blind". It would almost certainly have picked the second. Yet the first question turned out to be the road to the second.
It began with someone wondering
how an alga sees.
This piece sticks to the published record. Where accounts differ, or rest on a single source, the text says so. Scenes are drawn from the participants' own published recollections; no dialogue has been invented.