Robert Sapolsky · The biology of our best and worst selves
A neurobiologist walks onto the TED stage and opens by describing how he plans to torture Hitler.
He overpowers the elite guard, bursts into the bunker, grapples, pins, cuffs. "Adolf Hitler," I say, "I arrest you for crimes against humanity." Then, he says, the Medal of Honor version of the fantasy ends and the imagery darkens. What follows, delivered in the flat voice of someone reading a grocery list: sever the spine at the neck, take out the eyes with a blunt instrument, puncture the eardrums, cut out the tongue, leave him alive on a respirator — able only to feel — then inject something cancerous until every cell in his body is screaming. He's had this fantasy since he was a kid, he says, and when he does, his heart speeds up. Then he takes it apart himself:
But there's a problem, which is I don't actually believe in souls or evil, and I think wicked belongs in a musical.
The next thirty seconds are the engine of the whole talk. There are people he'd like to see killed, but he's against the death penalty; he likes schlocky violent movies, but he's for strict gun control. "In other words, I'm your basic confused human when it comes to violence." The confusion isn't a flaw — it's the subject. We don't hate violence, we hate the wrong kind; the right kind we cheer, decorate, vote for and mate with. And the same species that delivers poison gas through shower heads is also extraordinarily altruistic and compassionate.
So he gives us a move: walk backwards. The scene is a riot; a stranger runs at you, expression unreadable, holding something that kind of looks like a handgun. You pull the trigger. It was a cell phone. Now rewind, one layer at a time. One second before: how active was your amygdala? Months before: had chronic stress already enlarged that amygdala and thinned your frontal cortex? Further back: adolescence, the epigenetic marks laid down in fetal life, the genes you started as, the culture of honor your ancestors invented while herding camels and goats centuries ago, millions of years of primate evolution. None of these layers is a competing explanation; each one is the cause of the next.
Which makes his official conclusion sound like a surrender: "Officially, it's complicated. Wow, that's really helpful." But the half-sentence right after it is the blade: you had better be very careful before concluding you know what causes a behavior — especially a behavior you are judging harshly.
He closes on change, with examples that get faster and faster: John Newton took decades; at Christmas 1914 British and German soldiers spent a few hours in no-man's-land carrying bodies, digging graves, praying and playing football together, and redrew who counted as "us"; at My Lai, helicopter pilot Hugh Thompson turned his machine guns on his fellow Americans in a matter of seconds.
And afterwards? The multi-level-causation framework won outright — it became the skeleton of the 800-page Behave, and today it is close to common sense in behavioral science. But some of the bricks he used on stage have worked loose. The MAO-A study rests on 442 white boys, and the largest independent replication failed. The number "brain isn't mature until 25" turns out to be one that even the most-cited researchers can't trace — and it has already reached courtroom sentencing. And in 2023 Sapolsky pushed the same logic all the way to a book arguing there is no free will; a philosopher's response was that a book about free will somehow never discusses free will.
"The fantasy always runs something like this."
A sixty-year-old Stanford neurobiologist, standing on TED's red dot, is describing how he overpowers Hitler's elite guard and bursts into the secret bunker with a machine gun. Hitler lunges for his Luger; knocked away. Lunges for his cyanide pill; knocked away. "He snarls, comes at me with otherworldly strength. We grapple, we fight, I manage to pin him down and put on handcuffs."
"Adolf Hitler," I say, "I arrest you for crimes against humanity."— Robert Sapolsky, TED2017
It's a handsome opening, and the room knows what to do with it. Then Sapolsky says: "Here's where the Medal of Honor version of the fantasy ends and the imagery darkens."
For the next twenty seconds, in a voice close to flat narration, he reads out a list. Sever his spine at the neck. Take out his eyes with a blunt instrument. Puncture his eardrums. Cut out his tongue. Leave him alive on a respirator, tube-fed, unable to speak or move or see or hear, just to feel — then inject him with something cancerous that will fester and pustulate until every cell in his body is screaming in agony, until every second feels like an eternity in hell. "That's what I would do to Hitler."
No laugh line has been set up, and there is nowhere for the audience to put a laugh. He's had this fantasy since he was a kid, he says, still does sometimes, and when he does, his heart speeds up. Then he reaches over and knocks the whole thing down himself — "But there's a problem."
Robert Sapolsky was born in Brooklyn in 1957, spent his childhood reading about and imagining living with silverback gorillas, was writing fan letters to primatologists by twelve, and taught himself Swahili in high school. He graduated from Harvard in 1978 and was in Kenya a week later — not with gorillas, as it turned out, but with baboons. For the next two decades-plus he returned every summer to the Masai Mara, spending four months a year, eight to ten hours a day, inside one troop of savanna baboons, recording who bullied whom, who groomed whom, whose blood ran high in stress hormones. Ph.D. in neuroendocrinology from Rockefeller in 1984, MacArthur Fellowship in 1987, now professor of biology, neurology and neurosurgery at Stanford. The shape of the talk is hiding in that résumé: he is one of very few people who counts baboons in the field, measures stress hormones in the lab, and reads anthropology and legal scholarship.
And the timing of April 2017 is almost too neat. The talk was recorded on 24 April; his 800-page Behave: The Biology of Humans at Our Best and Worst was published on 2 May — as he stood up to speak, the book was somewhere between the printer and the bookstore. Chris Anderson's introduction says Robert has spent the last few years thinking about how weird human behavior is and how inadequate most of our language for explaining it is, and that it is exciting to hear him explain the thinking behind it "in public for the first time." At the close, besides "Good luck with the book," Anderson adds a telling line: "we're going to try and get you to come here in person one year."
So this is not an "I just discovered something" talk. It is a stitching job: in fifteen minutes he has to sew neuroscience, endocrinology, developmental psychology, behavioral genetics, cultural anthropology and evolutionary biology into a single piece of cloth. That is what decides which seams later pull too tight.
Having dismantled the Hitler fantasy, Sapolsky spends half a minute pinning himself to his own contradictions: there are people he'd like to see killed, but he's against the death penalty; he likes schlocky violent movies, but he's for strict gun control. Then he scales the confusion up to the species: we use shower heads to deliver poison gas, letters with anthrax, airplanes as weapons, mass rape as a military strategy — "we're a miserably violent species." But there's a complication:
we don't hate violence, we hate the wrong kind. And when it's the right kind, we cheer it on, we hand out medals, we vote for, we mate with our champions of it.— Robert Sapolsky, TED2017
And a second complication: the same species is also extraordinarily altruistic and compassionate. So the real question is not where violence comes from but how one biology produces both ends of that range. The hard part was never the motor side of behavior — your brain tells your spine, your spine tells your muscles, and you've behaved. The hard part is the meaning: the same trigger pull is an appalling act in one setting and heroically self-sacrificial in another. And if you think there's going to be the brain region or the hormone or the gene or the childhood experience that explains everything, you're not going to get anywhere.
Most of the rest of the talk is that sentence turned into a move. The scene: there's a crisis going on, rioting; a stranger is running at you in an agitated state — you can't quite tell if the expression is frightened, threatening or angry — holding something that kind of looks like a handgun. You're not sure. He comes running at you and you pull the trigger. The thing in his hand was a cell phone.
Then he starts walking backwards, one layer at a time:
What makes this clever is that the layers interlock rather than sit side by side — months of stress enlarge the amygdala and thin the frontal cortex, which is exactly why that amygdala is so easy to light one second before. Reaching the genetic layer, he makes the hardest claim of the talk, and the brick most worth chasing later:
genes don't determine anything, because genes work differently in different environments. Key example here: there's a variant of a gene called MAO-A, and if you have that variant, you are far more likely to commit antisocial violence if, and only if, you were abused as a child.— Robert Sapolsky, TED2017
Further back are the ancestors of centuries ago: nomadic pastoralists in deserts and grasslands would likely have invented "what's called a culture of honor filled with warrior classes, retributive violence, clan vendettas," still shaping the values you were raised with centuries later. The last layer is millions of years of evolution, where humans float in between the primate extremes by every measure. Here he lets himself be dry — "Officially, it's complicated. Wow, that's really helpful" — but the second half of the thought tightens immediately:
It's complicated, and you'd better be real careful, real cautious before you conclude you know what causes a behavior, especially if it's a behavior you're judging harshly.— Robert Sapolsky, TED2017
In the last three minutes he turns around and talks about change: every bit of the biology he just described can change in different circumstances. Ecosystems change — thousands of years ago the Sahara was a lush grassland. Cultures change — in the 17th century the most terrifying people in Europe were the Swedes, rampaging all over the place; "this is what the Swedish military does now. They haven't had a war in 200 years." Most importantly, brains change.
Then four examples on ever-shorter time scales. Decades: the British theologian John Newton, a slave-ship captain and later an investor in slavery as a younger man, who ended up central to abolition in the early 1800s (we take this one apart in the next section). Fifty years: the Japanese pilot Zenji Abe, about to lead a squadron of bombers toward Pearl Harbor on the morning of 6 December 1941, who fifty years later to the day embraced a survivor at a memorial ceremony and apologized in halting English. Hours: the Christmas truce of 1914, where British and German soldiers went out to collect bodies, then helped each other carry them, dig graves in the frozen ground, pray, spend Christmas together, exchange gifts, and the next day play football and exchange addresses — in a matter of hours they had built a brand-new "us" (everyone in the trenches on both sides, dying for no damn reason) and a brand-new "them" (the faceless powers behind the lines using them as pawns), until the officers arrived and said they'd be shot unless they went back to trying to kill each other. Seconds: at My Lai, helicopter pilot Hugh Thompson landed, saw American soldiers shooting babies and old women, set his aircraft down between the surviving villagers and the soldiers and trained his machine guns on his own countrymen — "If you don't stop the killing, I will mow you down."
He ends by inverting the tired maxim. Those who don't study history are destined to repeat it — whereas those who don't study the history of extraordinary human change, and the biology of what can transform us from our worst to our best, are destined not to be able to repeat those incandescent, magnificent moments.
1. The MAO-A brick is much narrower than that "if, and only if." The study he means is Caspi et al., Role of genotype in the cycle of violence in maltreated children, Science, 2002. Its foundation: the Dunedin birth cohort in New Zealand, 442 participants, 100% white, and all male. Maltreatment was measured between ages 3 and 11, outcomes between 11 and 18, using a composite of conduct disorder, violent conviction, antisocial personality symptoms and disposition toward violence. In the 2014 meta-analysis that pooled 27 studies, Caspi's own paper contributes a one-tailed p = .005.
Three things the stage version leaves out. First, this is an interaction, not a risk factor: people carrying the low-activity variant who were not maltreated do not score high on antisocial measures — maltreatment is what does the work; the gene only bends the slope. Second, the sample is entirely male and entirely white; the same meta-analysis found no such interaction across 11 female cohorts. Third, "far more likely" corresponds to a statistical interaction term, not a prediction you can drop onto an individual — and "if, and only if" turns a noisy slope into a switch. The irony is that this sentence arrives inside the passage where he insists genes don't determine anything.
2. "The frontal cortex doesn't fully mature until around 25" — a number he himself didn't use thirteen years earlier. On stage: "It doesn't fully mature until you're around 25." But in his own 2004 paper in Philosophical Transactions of the Royal Society B, "The frontal cortex and the criminal justice system," his wording is that the PFC shows "remarkably late development (in which full myelination may not occur until early adulthood)" — no 25. The difference isn't tone: "early adulthood" is an observation, "25" is a line. In 2016, Harvard's Leah Somerville devoted a piece in Neuron to asking what exactly we are searching for when we say a brain is mature; her starting point is precisely that evidence of continued neurobiological maturation through adolescence is increasingly invoked in discussions of youth-focused policies. In 2022, Laurence Steinberg — among the most-cited developmental psychologists alive — put it to Slate more bluntly: "I honestly don't know why people picked 25. It's a nice-sounding number? It's divisible by five?"
3. His strongest evidence that cultures change is the one he cut. On stage he proves cultures change with the Swedes — someone else's history. His own baboons are the first-hand version. Between 1983 and 1986, the most aggressive males in the Forest Troop he had been watching died of bovine tuberculosis contracted from a garbage pit at a tourist lodge. By 1993 and after, the troop's males had turned over completely (male baboons leave their natal troop and join new ones as adults), and yet the low-aggression pattern persisted: dominant males were significantly less aggressive toward subordinates, and they lacked the elevated stress-hormone markers seen in the control troops. He and Lisa Share published it in PLoS Biology in 2004, and left themselves an opening: teasing out the mechanisms of such complex behaviors will require future study — and the sample is one troop. In fifteen minutes, the uncertain example that was his own lost out to the certain one that belonged to somebody else.
4. John Newton's change was decades slower than the stage version. Sapolsky says Newton "spent decades as a younger man as the captain of a slave ship, and then as an investor in slavery, growing rich from this. And then something changed" — something Newton celebrated in "Amazing Grace." The real timeline is more awkward. Newton marked his prayer during a storm in March 1748 as his conversion — and it was after that conversion that he became a captain: the Duke of Argyle in 1750, then two voyages on the African in 1752–53 and 1753–54. After a stroke in 1754 forced him off the sea, he continued to invest in slaving operations. "Amazing Grace" was written in late 1772 under the title "Faith's Review and Expectation," for a New Year's prayer meeting on 1 January 1773, and published in the Olney Hymns in 1779 — it is about Newton's own conversion, and he is not known to have connected it to anti-slavery sentiment. His first public opposition to the trade came in 1788, in Thoughts upon the African Slave Trade: thirty-four years after he left the business, forty years after the storm. He did become an ally of the abolitionists and did live to see 1807 — but this "change" was not a lightning bolt. It was a very slow, very ugly slope. And the entire second half of the talk is about how fast change can be.
PARTLY HELD UP · as of 2026-09
The framework — that explaining a behavior means asking about nine time scales at once — won decisively and is close to common sense in behavioral science today. But several of the bricks he used on stage to hold it up, MAO-A and "25" among them, were sanded thin over the following decade; and the conclusion he drew by pushing the same logic to its end (that there is no free will) is still taking hits.
The framework: won. Behave, published a week after the talk was recorded, follows the talk's own order, running from "One Second Before" all the way to "Centuries to Millennia Before." Today essentially no serious behavioral scientist argues that the gene explains aggression — in a sense, the straw man Sapolsky was swinging at on stage has become hard to find.
The bricks: sanded thin. Six years before the talk, in 2011, Laramie Duncan and Matthew Keller systematically reviewed 103 candidate gene-by-environment studies published between 2000 and 2009, and the result was a blow to the head: 96% (45/47) of novel findings were positive, against only 27% (10/37) of replication attempts; median sample size 345, with less than 10% power to detect moderate effects.
MAO-A specifically cuts both ways. In 2013, Brian Haberstick and colleagues ran a large-sample test on the US Add Health cohort — 3,356 white men and 960 black men, aged 24–34 — and reported one sentence: "We could not confirm the hypothesis that MAOA genotype moderates the relationship between childhood maltreatment and adult antisocial behaviors." They noted explicitly that this was not a power problem. The following year, Byrd and Manuck's meta-analysis of 27 studies did confirm the interaction across 20 male cohorts (p = 8.2×10⁻⁷ when narrowed to maltreatment), calculating that it would take 93 unpublished null studies to overturn it — and found nothing across 11 female cohorts.
So MAO-A itself broadly held up, but only in males, and much more narrowly than advertised. The real problem is the ground it stands on. In 2019, Richard Border and colleagues tested the 18 most-studied candidate genes for depression against data on 620,000 people and found them no more associated with depression than randomly chosen genes. The whole candidate gene-by-environment paradigm was largely abandoned in those years. MAO-A was one of the sturdier rooms in that building — but the building came down.
The "25": misused. That is exactly where Somerville's 2016 Neuron piece lands: the fact that neurobiological maturation continues past adolescence is increasingly being carried into youth-focused policy, while neuroscientists themselves have not settled how to define or quantify "maturity." A decade on, the number has not become better founded, yet it has reached sentencing considerations for defendants under 25 in Scottish courts and "emerging adult" sentencing reforms for 18-to-25-year-olds in several US states. Sapolsky invoked it in order to make judgment less harsh, and that direction is right; but a number with no traceable origin, once treated as a line, gets taken up by both sides for their own purposes.
The man himself: no retraction — he doubled down. In October 2023 Sapolsky published Determined: A Science of Life Without Free Will, taking the "rewind to a million years" logic to its terminus: if every thought is determined by prior events you did not control, free will does not exist and the whole apparatus of responsibility and punishment has to be rebuilt. On 3 November that year, the philosopher John Martin Fischer gave it a distinctly unfriendly review in Notre Dame Philosophical Reviews, charging that Sapolsky uses definitions to bypass argument: "He offers no arguments in the entire book against compatibilism," and the line that has been quoted ever since — "From my perspective as a philosopher, it is jarring that a book on free will would not discuss free will." The same year, the neuroscientist Kevin Mitchell published Free Agents with the opposite answer: living things evolved agency, and eventually metacognition, and so evolved at least partial free will.