A child can lose half a brain. The rest will rewire around the job

Lex Fridman opens with David Eagleman's new book, a ten-year attempt to capture how the brain changes. The metaphors we already have, Eagleman says, are the problem. Engineers praise clean hardware with software running on top; the brain does something else. He coined livewired for a system that reconfigures itself physically as it learns. The hardware and software layers are blended, and the circuitry is not a finished board.

The example he reaches for is not a metaphor. Children under about seven can have one hemisphere removed in a hemispherectomy and still function. They walk with a slight limp on the other side of the body, and that is largely it. Some children are even born without a hemisphere, and the remaining visual system rewires so that everything maps onto the single leftover side. You cannot tear half the circuitry out of a phone and expect it to boot. The brain has done this in thousands of cases.

What those cases teach, Eagleman argues, is that the organ is trying to accomplish the tasks in front of it with the real estate it still has. Evolution took a long time to arrive there, and tens of thousands of neuroscientists are still poking at three pounds of tissue. His bet is the same one Crick and Watson made looking into a noisy cell: it will stay impossibly complicated until the general principles snap into focus.

Part of the public confusion about plasticity, Eagleman says, is that studies at different ages get mashed together. Remove half of Fred's adult brain and he does not recover; do it in a child and the system is still soft. Plasticity is the ability to mold into a new shape and hold it, which is why we call molded toys plastic. William James coined the term more than a hundred years ago, impressed by manufacturing. Eagleman uses it sparingly, because that is not what the brain does. It keeps rewiring for an entire life. There is no finished shape.

Different regions also close on different clocks, and the clock, he argues, tracks how stable the incoming data are. Visual cortex cements itself over a few years because the world's angles, colors, and shapes are essentially stable. Somatosensory cortex and motor cortex sit next to the body, which keeps changing: you get taller, fatter, thinner; you break a leg and limp; you learn a bicycle or a surfboard. Because that data stream never sits still, those maps stay more malleable. The genetic trick is flexibility itself. An alligator's brain does roughly the same job every generation. A human drops in half-baked and has to absorb language, customs, and belief.

It is not a blank slate. Fibers from the eye are already routed toward visual cortex; the machinery of audition is already aimed. What evolution engineered is a sponge with just enough structure to fill in the rest. That is also why Eagleman wants to drop the hardware/software split. Activity passing through the system changes receptor distributions, the shape of the neuron, biochemical cascades, even the epigenome that turns genes on and off. Researchers talk so much about synaptic weights, he says, mainly because that is what they can measure.

His better picture is Stewart Brand's pace layers for cities: fashion changes fast, governance slower, buildings slower still, nature slowest of all. Map that onto the brain and you get biochemical cascades that flip in seconds sitting on top of structures that take decades to settle. The oldest rule in neurology, Ribot's law, suddenly looks like a feature. Older memories are more stable than newer ones, which is why a person with dementia may remember a childhood podcast and not last month. No other machine works this way. Time has simply written the older traces into deeper layers.

The system stays malleable into old age, which is lucky, Eagleman notes, because the technological world does not stop. He sent his mother an Alexa; she found the settings. The interesting limit is not the tissue so much as the goal. Locked in silence and darkness, the brain is trying to build an internal model of how the world works: how to function, how to speak, when to be kind or aggressive. He keeps a movie poster in mind, Matt Damon alone on Mars in *The Martian*. That, he says, is what it is like inside a skull.

You and I have absorbed different lives up to this moment, and we communicate over a thin bandwidth. He will say a peach tastes just so, and you will say you know what he means, and the inner experiences may still be nothing alike. As people get older, that private model can decide it already understands the world well enough. Eagleman suspects the familiar claim that plasticity fades with age is partly a story about missing motivation. Give an eighty-year-old a firing and a WordPress site that has to go up, and the capability is still there.

Fridman turns to Neuralink and brain-computer interfaces. Eagleman is glad people are jumping in, and he thinks invasive electrodes will matter for Parkinson's, epilepsy, and other patient cases, especially at higher density. He is less sanguine about a consumer business. Thumbs on a phone and photons in an eye already move a lot of data. Cracking the skull for a twenty percent faster text is a hard sale when neurosurgeons talk about death, infection, and a saying that the person is never quite the same after the bone is opened.

The part that already works, he argues, does not require a perfect common language. Hundreds of thousands of people walk around with cochlear implants or retinal implants: a digital microphone and an electrode strip in the inner ear, or a grid on the back of the eye. Those devices do not speak the natural biological language. They speak the dialect of Silicon Valley. As recently as twenty-five years ago, plenty of people thought that mismatch would kill the whole idea. The brain figured it out by chasing correlations: a clap, a mouth moving, a signal arriving together, until any incoming language becomes usable.

He calls this the potato-head theory of evolution. Eyes, nose, mouth, ears, and fingertips are plug-and-play peripherals. Across the animal kingdom you find heat pits, electroreceptors, cryptochrome for Earth's magnetic field, and the brain does not reinvent its operating principles each time. It asks whether the data are useful and molds around them. Kevin Kelly's line is the cultural version: technology is what gets invented after you are born; what already exists is background furniture. An iPad, to Eagleman's children, is furniture. The livewired brain was ready for the smartphone, and it will be ready for whatever comes next.

If there is a silver lining in 2020, Eagleman says, it is plasticity. People were on gerbil wheels, optimizing an internal model for a familiar circumstance, and then the year knocked them off. Large studies of older adults who stay cognitively active show a grim and hopeful split: some fraction have Alzheimer's physically chewing the brain, and nobody knows it while they are alive because they keep building new roadways as old ones degrade. Novelty, responsibility, and a social life are the construction crew. Being forced out of automaton mode is, on this view, the only good news.

The same year also made the world's best explainers try to teach on a screen. Fridman talks about Grant Sanderson's 3Blue1Brown videos and the chance that a Nobel-winning faculty member becomes a kind of couturier of canonical lectures. Eagleman has already met thirteen-year-olds after book talks who cite a TED talk as if it were mother's milk. His own generation grew up on just-in-case learning: Mongolian dates, stored against a day that might never come. His eight-year-old son and five-year-old daughter live on just-in-time learning. They get curious, they ask Alexa, and the answer arrives inside the curiosity.

That timing is not a convenience. For plasticity to happen, you need to care. The ancient Romans, Eagleman notes, already ranked curiosity as the highest of seven levels of learning. Knowledge that lands at the moment of interest is the knowledge that gets written in. He thinks this generation will be much smarter than ours, not because their attention spans are simple, but because a livewired brain will absorb tweets, four-hour podcasts, and Netflix alike and still assemble a world view. Nature versus nurture is the wrong fight here. You arrive with a genome and a womb, and then experiences write the epigenome. Rat pups licked and groomed take a different path, for life, than pups left aside.

Asked whether there is free will, Eagleman does not dodge. Most neuroscientists, he says, cannot see where it would fit. The brain looks like a machine: 86 billion neurons, about as many glial cells, each neuron as complicated as the city of San Francisco, each expressing millions of gene products and talking to some 10,000 neighbors, for on the order of half a quadrillion connections. If nothing in it is not driven by something else, free will is hard to place. That is the camp almost everyone falls into.

He immediately adds that science is still young, and that every era has felt one puzzle piece short of a finished picture. The speculation he offers, and he asks not to be podcasted as believing it, is the parable at the end of *Incognito*. Imagine a Kalahari Bushman who finds a radio in the sand, turns the knob, hears voices, and as a radio materialist opens the back. Pull some wires and the voices garble or stop, so he builds a full theory of how that pattern of wires produces speech. It would never occur to him that a distant city has a tower, and that invisible radiation is the origin of the voices.

What we know for certain, Eagleman says, is that damaging pieces of the brain jumbles functions. How would you know if something you cannot see, like electromagnetic radiation, is what is actually generating the show? He is not offering a conclusion. He is reminding Fridman that the known unknowns fill the labs, and that a larger space of unknown unknowns has not even been asked yet.

Fridman asks about genius and evil, Hitler and Einstein, in a livewired brain. Eagleman bristles at a *Time* cover that showed a sagittal slice with spots labeled Gandhi and Hitler, as if good and evil were addresses. Because we are livewired, we take the world around us. Hitler received positive feedback as the ideas got crazier; crowds spun each other up the same way in the Cultural Revolution and the Russian Revolution. Tolstoy's line about the king as slave to the people is the warning. Pinning evil on one brain misses the crowd that trained it.

His lab's scanner study makes the mechanism uncomfortably cheap. People see six hands on a screen. A computer picks one at random and a syringe needle goes in and out, which lights up the pain matrix even though it is not your hand. That is empathy, raw and low-level, unlike a Q-tip poke that looks similar. Then each hand gets a one-word label: Christian, Jewish, Muslim, atheist, Scientologist, Hindu. Across religions, the brain cares more when the label matches. Same hands. Same needle. One word.

They push it further. Tell subjects the year is 2025 and three religions are at war with three others, randomized, and former out-groups become allies; care rises. Then a new cohort tosses a coin: heads Augustinian, tails Justinian, tribes invented in the hallway. A wristband, a story about warring sects, and the pain matrix already plays favorites. Studies of monkey troops show the same stickiness and the same speed when an animal is switched to the other side. The livewired cost is that a brain growing up in hatred will grow hatred the way it grows surfing.

The light-bulb joke in *Livewired* is: how many psychiatrists does it take to change a light bulb? Only one, but the bulb has to want to change. Friends, communities, and psychiatrists matter, Eagleman says, because each of us is trapped on that private planet and can see only our own angle. Psychiatry's larger gift is killing the childhood assumption that everyone is like us. Schizophrenia is a different head. Psychopathy is a different head. A million other configurations exist, and watching behavior is still how we guess at the inside.

That is the argument of his *Atlantic* essay The Brain on Trial, and of the national nonprofit he runs, the Center for Science and Law. Five people stand before the same bench for the same crime, and fairness currently means the same sentence. Their brains may not be the same instrument. One has schizophrenia, one is a psychopath, one is deep in drugs. The United States, he notes, already locks up a higher share of its population than any other country, as if incarceration were a one-size-fits-all tool.

The point is not to wave anyone off the hook. It is to change the question from how blameworthy you are to what is the best thing to do from here. Breaking rocks in the sun does not treat schizophrenia. Jail does not treat an addiction. Neuroscience and psychiatry already have more useful moves, and the simplest policy step is specialized courts: mental-health courts whose judges and juries know the illness, drug courts that know rehabilitation. In a regular courtroom, a defense lawyer saying schizophrenia often meets a jury that calls it an excuse, because they have never known a patient. How do counties adopt this? Eagleman has watched the sequence. The jail fills, they cannot afford a new one, and only then do they want something smarter. A crisis, including a pandemic, is when backs hit the wall.

Fridman describes GPT-3: 175 billion parameters, unsupervised on unstructured internet text, generating impressive passages. Eagleman grants the impressiveness and then draws a line. The wow examples have already been filtered by humans who threw out the rest. Human creativity also remixes what it has absorbed, but it does so with good models of other humans. He will not switch into French mid-sentence, because Fridman would not follow. Everything leaving his mouth is aimed at those ears, in real time, inside a relationship that is itself learning how to talk.

GPT-3, in his telling, remixed sentences and popped something out. It does not know what Lex needed to hear next. Raising the parameter count will not, by itself, add a model of the department chair who might read the tweet, the mother who might see it, the date who might judge it, or the political weather and the fear of mortality sitting under the whole calculation. Artificial nets with units and synapses have done astonishing work. They are still not running the same algorithms. His infant children can navigate a room, lie, and hold a social conversation that no 2020 language model could.

The missing piece he cares about is relevance. Humans drown in data and encode only what matters, first for survival and then for the particular goals of a life. Computers do not have a yen to survive. The penultimate chapter of *Livewired* is titled The Wolf and the Mars Rover: a wolf chews off a trapped leg and learns to walk on three; Curiosity stuck a wheel in Martian soil and a billions-dollar project died. The wolf can do that because motor and somatosensory maps are livewired and because it wants to eat and get back to the pack. He is trying to write a piece called If You Want to Build a Robot, Start with the Stomach. No hunger, no goal; no goal, no relevance; no relevance, no intelligence. A robot that rolls into a room with nothing it wants will freeze.

Ninety percent of Eagleman's time now goes to a company called Neosensory. The product is a wristband of vibratory motors. As he talks, it captures his voice, runs algorithms, and turns the result into patterns of vibration. Deaf users learn to hear through the skin. On day one they are already better than you would expect at tagging a dog, a baby, a doorbell. Months later the experience is no longer a label. They say there is a dog barking somewhere. That, he says, is a new quale: a new subjective inner experience, built the same way an infant builds hearing from eardrum spikes plus a mother's moving mouth.

Skin is the unused channel. Eyes and ears are overtaxed; the vest he first built was a joke about wasting a computational surface. Sensory substitution, sound into skin, already works. Sensory expansion is next: detectors for infrared and ultraviolet on a slice of the electromagnetic spectrum vision never sees. The first night an engineer built the infrared version, Eagleman walked in the dark between two houses, felt a wash of vibration, followed his wrist, and found a night-vision camera. It became part of his reality. Sensory addition is the step after that: Earth's magnetic field, Twitter, the stock market, other people's moods, anything a machine can measure, fed in until it joins the world you inhabit. Take it off and it will not kill you. At the time of the conversation the band costs $399, a fraction of a hearing aid and a couple of hundred times cheaper than a cochlear implant.

The concept underneath is umwelt: the slice of reality an animal can pick up. A tick has butyric acid, temperature, and touch. A bat has returning air-compression waves. A weakly electric fish has the field around its body. A bloodhound has some two hundred million scent receptors and a world of odor in which a human nose looks pitiful: you cannot know a cat is a hundred yards away, or that a friend was here six hours ago. Visible light is less than a tenth of a trillion of the spectrum, and that fraction is already a convention, because the spectrum runs infinite in both directions. For the first time, Eagleman says, a species can choose to include other channels in its own world. He still teaches at Stanford. The fire is here.

*Livewired* opens with Heidegger: every man is born as many men and dies as a single one. Eagleman means it as a plasticity diagram, a space-time cone. You dropped into the world with the potential to be a great surfer, a chess player, thousands of other people. Chance and choice narrowed the path, and on the day you die you will be exactly one person. Asked for the meaning of it all, he refuses a verdict. Cultures answer that question differently; a secular scientific childhood is one more angle, not a proof.

What he will say to a young person in a chaotic year is stay adaptable. The jobs that will exist in twenty years do not have names. College should be less and less vocational, because the model of learn-a-trade-and-do-it-forever no longer matches the speed of the world. Digest widely. Every paper, book, and Wikipedia hole becomes the raw storehouse you later remix. Then run a gut check: does this page make you care? When something sparks, follow it, and do not be afraid to commit. Options are beautiful; fear of choosing is how the cone closes without a fight.

Fridman lands on Confucius as Eagleman had quoted it: each person has two lives, and the second begins when you realize you have just one. Eagleman does not live day to day inside a meditation on dying in the next five minutes. Grant deadlines and co-authors do the steering. Still, the trajectory does not last forever, and the raw material of a life is whatever you were curious enough to take in. The closing lines Fridman reads from Eagleman's *Sum* are the numinous version of the same claim: if we are only algorithms on a fabric of cells, understanding that would be better than anything in a holy text.