Testosterone doesn't create aggression, it amplifies what's already there

The Short Version

Testosterone does not create aggression. That's the verdict of Stanford neuroendocrinologist Robert Sapolsky, Caltech neurobiologist David Anderson and Andrew Huberman, and chemosensation researcher Noam Sobel adds evidence that points the same way [1][2][3][4]. Sapolsky's version is that testosterone turns up the volume on whatever a person was already inclined to do, and pushes them toward whatever earns status in their world, which can be fighting in a baboon troop or generosity at a charity auction [1]. The circuit that actually triggers attack is a small cluster of estrogen-receptor neurons in the ventromedial hypothalamus. Testosterone mostly reaches that circuit after the brain converts it into estrogen, and stress, day length, isolation and even odorless chemicals in tears decide whether the circuit fires [2][3][4]. The popular idea that men with high testosterone are naturally violent confuses a hormone that is necessary, and that amplifies, with a cause.

Table of contents

What They Said

Robert Sapolsky: "upping the volume," not flipping a switch

Sapolsky opens bluntly: "almost everybody out there has a completely wrong idea as to what testosterone does" [1]. He admits the myth has an easy logic. In nearly every species males have more testosterone and more aggression. Seasonal breeders get a surge of testosterone just as they start fighting over territory. Castration lowers aggression [1]. From that, it's natural to conclude testosterone causes aggression. It doesn't, he says [1].

What it actually does is lower the threshold for things that would already provoke you, so systems that are already switched on run louder [1]. He takes this down to the level of single neurons. Testosterone doesn't make amygdala neurons fire on their own. It raises the firing rate of neurons that are already being stimulated, partly by shortening their after-hyperpolarization, the brief quiet period after a neuron fires [1]. His summary: "It's just upping the volume of whatever aggression is already there." [1]

The monkey experiment. His best-known illustration: put five male monkeys together and they form a hierarchy. Give number three a large dose of testosterone and he gets into more fights. But he doesn't challenge numbers one and two. He keeps deferring to them as before and becomes a terror to numbers four and five [1]. The hormone didn't create a new urge to dominate. It amplified patterns and social learning that were already there [1].

Response more than cause. Sapolsky points out that the causal arrow in the famous correlations often runs backward. Sexual behavior and aggression raise testosterone, while baseline levels barely predict what someone will do [1]. Raising a person's testosterone by 30 percent, male or female, has essentially no effect on sexual behavior, because the brain isn't that sensitive to fluctuations in the normal range [1]. Even a devoted fan watching their team from an armchair with potato chips gets a testosterone rise, which shows it's the psychological framing, not physical fighting, that drives the hormone [1]. The same holds in marathon studies. A winner's testosterone can fall because he finished slower than he expected, while the runner in 73rd place gets a surge because he expected to collapse and managed to finish [1].

Castration and prior history. After castration, sex and aggression drop but not to zero, and how much remains is predicted by how much there was before [1]. Social learning and context carry the behavior forward without the hormone [1].

Developing brain versus adult brain. Sapolsky separates two roles. In the developing fetus, the amount of testosterone makes a huge difference to how the brain is built. In adulthood it acts more like an on-off signal [1]. He points to the ratio of the second to fourth fingers, which reflects prenatal testosterone and estrogen exposure and predicts a range of subtle adult behaviors. Huberman, who worked in Marc Breedlove's lab on this research, says the finding has been replicated at least five times [1]. In females, Sapolsky says, aggression outside the postpartum period involves testosterone at much lower average levels, and the conclusion is the same: testosterone is necessary for typical levels of aggression and sex, but it doesn't cause them, and small individual differences don't matter much [1]. Maternal aggression after birth is driven by estrogen and progesterone instead [1].

The challenge hypothesis: status, whatever it takes. Sapolsky calls John Wingfield's challenge hypothesis his favorite finding. Testosterone is released when your status is challenged, and it makes you more likely to do whatever is needed to keep that status [1]. For a baboon that means aggression. Humans earn status in many ways, like the half-drunk alpha males at a private school auction competing to give away the most money [1]. That led to what he calls a nutty prediction that turned out to be true: in economic games where status comes from being trustworthy and generous, giving people testosterone makes them more generous [1]. If you gave Buddhist monks testosterone, he jokes, they'd compete over random acts of kindness [1]. His policy conclusion: if a society has too much aggression, testosterone isn't the first suspect. "The first to look at is that we hand out so much damn elevated status for aggression in so many circumstances." [1]

Where testosterone does cause harm. Sapolsky doesn't call the hormone harmless. It makes people more confident, which is bad when the confidence is wrong. In economic games it makes people less cooperative because they feel they can handle things alone. "Testosterone makes people cocky and impulsive." [1] He suggests testosterone-driven misjudgments of risk, like being sure an army will overrun another country in three days, may have played a part in decisions such as starting World War One [1]. It also increases energy, alertness and motivation even in non-aggressive people. That, he says, is why testosterone replacement often helps aging men, and within minutes it increases glucose uptake into skeletal muscle [1]. Whether that motivation goes into serving soup to homeless people or into something terrible depends on who you were before [1]. Lab rats will press a lever for testosterone infusions in the range that optimizes dopamine release [1].

Andrew Huberman: estrogen in the brain triggers the attack

In his episode on aggression, Huberman sets out "to dispel a still, unfortunately, very common myth": "testosterone does not increase aggressiveness. Testosterone increases proactivity and the willingness to lean into effort." [2] People given testosterone, or who naturally have more, become more competitive, "but not just in aggressive scenarios" [2]. Then comes the line that matches Sapolsky's amplifier idea: "if somebody is already aggressive, giving them testosterone will have the tendency to make them more aggressive. If somebody, however, is very benevolent and altruistic, giving them testosterone will make them more benevolent and altruistic, at least up to a point." [2]

His mechanism is aromatization. The enzyme aromatase converts testosterone into estrogen inside the brain, and that estrogen, acting on estrogen-receptor neurons in the ventromedial hypothalamus (VMH), is what triggers attack [2]. In mice or humans lacking aromatase, "It doesn't matter how much you increase testosterone or any of its other derivatives. You do not observe this aggression." [2] He admits that "This runs counter to everything that we know and think about the role of testosterone," and suggests a comeback for the next time someone blames testosterone: "ah, no, actually, it's estrogen that makes people aggressive and animals aggressive, for that matter." [2] His summary: "it's not testosterone, but testosterone converted into estrogen that activates these circuits for aggression." [2]

This also explains a sex difference. Males have relatively little circulating estrogen, so testosterone is needed as the raw material. Females generally have enough estrogen already to activate the circuit [2]. "So both males and females are primed for aggression, but that's riding on a context." [2]

The circuit. Huberman traces the history. Walter Hess lowered electrodes into awake cats and found a site where stimulation turned a passive, purring cat into one in absolute rage, calm again within seconds of switching it off [2]. Later work in mice and humans showed that stimulating the same area produces subjective feelings of anger, not just aggressive behavior [2]. That area is the VMH, only about 1,500 neurons on each side of the brain [2]. David Anderson's lab at Caltech showed the VMH is both necessary and sufficient for aggression, with estrogen-receptor neurons as the key population [2]. Dayu Lin used optogenetics to switch just those neurons on with blue light. A male mouse mating with a receptive female immediately stopped and tried to kill her when the light came on, then went back to mating when it went off [2]. A male alone in a cage attacked an inflated rubber glove whenever the neurons were stimulated [2]. Downstream, the VMH connects to the periaqueductal gray, which can release the body's own opioids for pain relief and drives biting and uncontrolled punching [2]. His takeaway: "neural circuits, not individual brain areas, evoke the constellation of behaviors that we call aggression." [2]

Aggression is a process. Following Konrad Lorenz, Huberman argues that "aggression is a verb. It has a beginning, a middle, and an end. And it's a process. It's not an event." [2] Lorenz proposed a hydraulic pressure built up by many factors, like hormones, neurotransmitter states and childhood history, "almost like a hydraulic, pressure," like fluid pushed into a container "until the can or the container is ready to explode" [2]. Huberman says this is how the system works: no single switch turns aggression on [2]. He also rejects the pop-psychology idea that aggression is amplified sadness. The brain has separate, non-overlapping circuits for aggression and for grief, and irritability and aggression aren't the same thing either [2].

Context sets the dial. Estrogen alone doesn't guarantee aggression either: "turns out it's all contextual" [2]. On long, sunny days, melatonin drops, dopamine rises and cortisol falls, and experimentally raising estrogen doesn't produce aggression. On short days, with more melatonin, less dopamine and more stress hormones, raising estrogen does increase the tendency to aggression [2]. High cortisol and low serotonin tilt Lorenz's pressure toward attack [2]. He cites a paper by Trainor and colleagues in the Proceedings of the National Academy of Sciences, "photo period reverses the effects of estrogens on male aggression via genomic and non-genomic pathways" [2]. He also notes that some people carry a genetic variant that changes estrogen-receptor sensitivity and can make them considerably more irritable and aggressive, and that day length strongly affects even that [2].

David Anderson: the estrogen receptor is the marker, and the target decides

Anderson, whose lab found the VMH aggression neurons, provides the circuit-level evidence. Huberman introduces the topic as a myth: "that testosterone makes Animals and humans aggressive and estrogen makes animals placid," and "Nothing could be further from the truth" [3]. Anderson explains that the molecular marker for the VMH aggression neurons turned out to be the estrogen receptor [3]. Knocking out that receptor in the VMH stops male mice from fighting. Work by Nirao Shah at Stanford showed that fighting in castrated mice can be restored with estrogen as well as with testosterone [3]. Many of testosterone's effects depend on aromatase converting it to estrogen, and males given aromatase inhibitors in animal experiments stop fighting and mating [3]. Progesterone receptors are involved too [3].

Females and context. Female mice aren't aggressive until they're nursing pups, when they attack any intruder, male or female [3]. Anderson's lab found two subsets of estrogen-receptor neurons in the female VMH, one for fighting and one for mating. Stimulating the fighting subset makes even a virgin female attack. Stimulating the whole population never did, because the two effects cancelled out [3]. In virgins the mating neurons dominate; in mothers the fighting neurons do [3]. The same hormone receptor, in the same small nucleus, produces opposite behavior depending on life stage.

Aggression describes behavior, not one state. Anderson warns against treating aggression as one thing: "the word aggression In my mind refers more to a description of behavior than it does to an internal state" [3]. It could reflect anger, fear, or hunger in the case of predatory aggression [3]. The VMH-driven kind is offensive aggression, and male mice find it rewarding. "They like it," and they'll press a bar for the chance to beat up a subordinate male [3]. Defensive aggression, the kind you feel when attacked or cheated, lives somewhere else that is still unknown [3]. Fear neurons sit just above the aggression neurons in the VMH, and strong fear shuts down offensive aggression. Stimulating the fear neurons during a fight stops it immediately [3].

The target pulls the trigger. Huberman raised Lorenz's hydraulic pressure. Anderson doesn't think aggression builds up as an accumulating need the way hunger does [3]. Stronger VMH stimulation puts a mouse on more of a hair trigger. But with nothing there to attack, the stimulated mouse mostly wanders around. Put something in front of it, even a glove, and it attacks [3]. How sensory information about a target combines with the drive to "pull the trigger" is what his lab is studying now [3]. He calls the VMH a satellite dish that takes input from about 30 brain regions and broadcasts to about 30. Because "aggression is a very risky thing for an animal to engage in," the brain "constantly has to make a cost benefit analysis Of whether to continue on that path or to back off" [3].

Social isolation. Anderson's most practical finding is about environment, not sex hormones. His lab found that tachykinin neurons promote aggression and that social isolation raises tachykinin levels and aggressiveness [3]. Moriel Zelikowsky showed that two weeks of isolation causes a massive rise in tachykinin 2 in the mouse brain, driving more aggression, fear and anxiety. A drug called osanetant, which blocks its receptor, reversed all of it, and "It's not a sedative, which is really important" [3]. Isolated mice that would normally kill their littermates overnight could be put back in the cage safely [3]. His comment: "So putting a violent prisoner in solitary confinement is absolutely the worst most counterproductive thing you could do to them" [3]. Huberman noted that the mass shooters he'd read about were consistently socially isolated [3].

Anderson also points out a limit of hormonal thinking in humans: "there is no treatment that we have for a violent sexual offender that eliminates the violence but not the sexual desire," whether physical or chemical castration [3]. Removing the hormone is a blunt tool that removes both behaviors.

Noam Sobel: chemical cues shift hormones and aggression

Sobel, who studies smell and chemical signals at the Weizmann Institute, adds a different kind of evidence: aggression can be shifted by chemical cues from other people, sometimes through testosterone and sometimes not [4].

Tears. Sobel wondered whether emotional tears might be a chemical signal. Of roughly a hundred volunteers who said they cried easily, about six could reliably produce tears in the lab while watching sad films. The tears were completely odorless. Yet when men sniffed them, their free testosterone dropped by about 14 percent within about half an hour, and brain activity in the hypothalamus and fusiform gyrus dropped too [4]. An independent group in South Korea later reproduced the testosterone result almost exactly. Unpublished work from his lab shows that sniffing tears lowers aggression in men on a laboratory aggression task [4]. A later Nature paper found that mouse pup tears reduce aggression in adult males [4]. Sobel describes tears as a kind of chemical blanket against aggression [4]. There's a caveat: a Dutch lab reported a failed replication of the behavioral part of his original study, and it declined his offer to host and fully fund a joint replication [4].

A molecule from babies' heads. Graduate student Eva Mishor tested hexadecanal, a molecule that promotes social buffering in mice. Using the Taylor Aggression Paradigm, where participants provoked by an unfair partner can blast them with loud noise, hexadecanal made men less aggressive and women more aggressive [4]. The result held up in a second experiment inside a brain scanner. There, hexadecanal activated the left angular gyrus, a region involved in social appraisal, in both sexes. But its connectivity with aggression areas like the amygdala and temporal pole shifted in opposite directions in men and women [4]. Hexadecanal turns out to be the most abundant semi-volatile compound given off by babies' heads. For an infant, calming fathers and making mothers more protective is a win-win. Or, as Sobel puts it, "babies are conducting chemical warfare." [4]

Fear. People in a fearful state give off a body odor that raises arousal in those who smell it, so fear is somewhat contagious [4]. Sobel's lab is building a "fear bank" of body odor from first-time paratroopers to identify the molecules involved [4].

Where They Agree, and Where They Don't

They agree on the main point. All four sources reject the simple equation testosterone = aggression. Sapolsky calls it a nearly universal misunderstanding [1]. Huberman calls it a myth outright, in his own episode and again with Anderson [2][3]. Anderson's lab restored fighting in castrated mice with estrogen and stopped it by knocking out the estrogen receptor [3]. Sobel shows that aggression and testosterone can both be shifted by chemical signals from other people [4].

They agree testosterone amplifies what's already there. Sapolsky's monkey number three [1] and Huberman's line that testosterone makes aggressive people more aggressive and altruistic people more altruistic [2] make the same claim in different words. Both also link testosterone to competitiveness and effort rather than violence: Huberman says proactivity and willingness to lean into effort [2], Sapolsky says energy, motivation and status-seeking [1].

They agree context decides. For Sapolsky, context means what earns status in your social world and how you frame events [1]. For Huberman, it's day length, cortisol, serotonin and genes [2]. For Anderson, it's the presence of a target, fear, motherhood and isolation [3]. For Sobel, it's the chemical signals of the people around you [4].

Where they differ: what the real cause is. The sources stress different mechanisms, and the difference matters. Huberman goes further than Sapolsky, telling listeners to answer "it's estrogen that makes people aggressive" [2]. Sapolsky's account centers on testosterone itself acting on amygdala neurons, turning up activity that's already there, and on the social meaning of status [1]. These views can be reconciled, since Huberman's estrogen is made from testosterone and needs a context too [2]. But one is a story about brain chemistry and the other about social incentives. Huberman's shortcut also risks swapping one oversimplification for another. Anderson, whose lab found the estrogen-receptor neurons, is more careful. He says aggression names a behavior with several different internal causes [3], and he notes that defensive aggression is controlled somewhere else entirely, which is still unknown [3].

Where they differ: does aggression build up? Huberman accepts Lorenz's hydraulic pressure as exactly how the system works [2]. Anderson is more skeptical. He doesn't think aggression builds up as an accumulating need like hunger, and he points out that a maximally stimulated mouse with no target mostly just wanders [3]. Both agree that internal state raises readiness. They differ on whether that readiness spills over on its own.

Where they differ: does testosterone matter at all? None of them says testosterone is irrelevant. Sapolsky calls it necessary for typical levels of aggression and says it makes people cocky, impulsive and uncooperative [1]. Huberman says males need it as raw material for the estrogen that triggers attack [2]. Anderson notes castration removes both violence and sexual desire [3]. The disagreement with the popular view is about whether testosterone is a cause, not whether it plays a role.

How strong the evidence is. Much of the circuit evidence comes from mice [2][3], and the summary of Anderson's episode notes that both scientists warn against mapping mouse findings directly onto human behavior [3]. Sobel's tears-and-aggression behavioral result is unpublished and has been disputed by another lab [4].

Why It Works

The sources describe four layers of mechanism.

1. At the level of neurons, testosterone amplifies. In the amygdala, testosterone doesn't start firing. It increases the firing rate of neurons that are already stimulated, partly by shortening their after-hyperpolarization [1]. That's why it can only turn up signals that are already present: no provocation, no amplified response [1].

2. At the level of circuits, estrogen receptors trigger attack. The VMH, about 1,500 neurons per side, is both necessary and sufficient for offensive aggression, and its key cells carry estrogen receptors [2][3]. Testosterone reaches them largely through aromatization into estrogen in the brain, which is why aromatase-deficient animals don't become aggressive however much testosterone they have [2], and why castrated mice fight again when given estrogen [3]. The VMH then routes the attack through the periaqueductal gray, which drives biting and striking and releases the body's own painkillers [2]. Fear neurons nearby can shut it down [3]. The same neurons mix fighting with mating, and in female mice separate subsets for each compete, with motherhood tipping the balance [3].

3. At the level of state, stress and environment set the threshold. High cortisol and low serotonin, and short days with more melatonin and less dopamine, make it more likely estrogen will trigger aggression. Long sunny days prevent it [2]. Social isolation raises tachykinin 2 and aggression, and blocking its receptor reverses the effect [3]. Strong stimulation of the VMH lowers the threshold but still needs a target [3].

4. At the level of society, status decides what gets amplified. Under the challenge hypothesis, testosterone rises when status is challenged and pushes you toward whatever restores it [1]. Where status comes from fighting, it promotes fighting. Where it comes from generosity, as in the economic games, it promotes generosity [1]. Testosterone also responds to how you interpret events: the armchair fan and the runner in 73rd place both get surges without any physical contest [1].

An outside chemical layer. Chemical signals from other people can change the system from outside. Odorless tears lower men's free testosterone by about 14 percent and reduce activity in the hypothalamus [4]. Hexadecanal reshapes the connection between a social-appraisal region and aggression areas, in opposite directions for men and women [4].

Where It Applies, and Where It Doesn't

It applies to the stereotype about men. The idea that a man's testosterone level predicts his violence doesn't hold up. Baseline levels barely predict behavior, and a 30 percent change has essentially no effect on sexual behavior [1]. Testosterone makes people more of what they already are [1][2].

It applies to thinking about testosterone therapy. Sapolsky says testosterone replacement often helps aging men through energy, alertness and motivation [1]. The amplifier view suggests the result depends heavily on who the person already is [1][2]. It also comes with the costs Sapolsky names: overconfidence and impulsiveness [1]. Anyone considering testosterone therapy, aromatase-related drugs or supplements should decide with a doctor, not a podcast.

It applies to policy. If aggression is amplified by status rewards and made worse by isolation, the useful levers are social. Stop handing out "so much damn elevated status for aggression" [1], and reconsider practices like solitary confinement [3].

It doesn't apply to prenatal development. Sapolsky is explicit that before birth, the amount of testosterone makes a huge difference to how the brain is built. The adult on-off picture doesn't describe that period [1].

It doesn't mean hormones are irrelevant. Testosterone is needed for typical levels of aggression [1], males need it as the raw material for brain estrogen [2], and removing it reduces both sex and violence [3]. The claim is that it isn't the cause, not that it has no effect.

It doesn't cover every kind of aggression. The VMH findings are about offensive aggression. Defensive and predatory aggression use other circuits, and defensive aggression isn't yet mapped [3]. Postpartum aggression is driven by estrogen and progesterone [1][3].

It doesn't transfer automatically from mice to people. Much of the circuit work is in mice, and the researchers themselves caution against direct extrapolation [3]. Some human findings, like Sobel's on tears and aggression, are preliminary or disputed [4].

What To Do With It

1. Stop using too much testosterone as an explanation for someone's violence. Ask instead what that person's environment rewards. Sapolsky's first suspect for excess aggression is how much status a culture gives it [1].

2. Change what earns status in the groups you run. In teams, classrooms and families, make generosity, trustworthiness and effort the way to gain standing. In economic games, testosterone made people more generous when generosity earned status [1].

3. Watch your stress chemistry when you feel reactive. Huberman links high cortisol and low serotonin to a tilt toward aggression [2]. His tools: get sunlight in your eyes early and throughout the day, and use a sauna or hot bath to lower cortisol [2]. He mentions ashwagandha in two-weeks-on, two-weeks-off cycles, but only after checking with a doctor [2].

4. Notice the season. Short, dark days change the context so estrogen is more likely to produce aggression [2]. If you're more irritable in winter, or during long stretches indoors under artificial light, treat that as information rather than character [2].

5. Treat aggression as a process you can interrupt early. Because it has a beginning, middle and end, learn to recognize the early stages in yourself and others, verbal or physical, before the pressure peaks [2].

6. Take isolation seriously as a risk factor. Anderson's work links social isolation to more aggression, fear and anxiety through tachykinin [3]. Checking in on isolated people, and questioning isolation as punishment, fits the evidence [3].

7. Expect higher confidence, not higher aggression, if testosterone rises. Sapolsky's warning is overconfidence: testosterone makes people "cocky and impulsive" and less cooperative [1]. If you're on testosterone therapy or in a high-challenge period, double-check your risk estimates and deliberately ask for others' input.

8. Remember that interpretation moves the hormone. Testosterone rises and falls with how you frame an outcome, as the marathon runners show [1]. Reframing a challenge changes the chemistry, not just the mood.

9. Don't dismiss the quieter human signals. Sobel's work suggests emotional tears and babies' scent carry chemical messages that can lower male testosterone and aggression [4]. These findings are early and partly disputed, but they support the broader point that aggression responds to social context in ways people don't consciously notice [4].

Sources

1. Dr Robert Sapolsky Science Of Stress Testosterone Free Will · Huberman Lab
2. Understanding & Controlling Aggression · Huberman Lab
3. Dr. David Anderson: The Biology of Aggression, Mating, & Arousal · Huberman Lab
4. Dr Noam Sobel How Smells Influence Our Hormones Health Behavior · Huberman Lab