Huberman Lab Essentials: Andrew Huberman interviews neuroscientist Dr David Anderson on the biology of aggression, mating, and arousal
Andrew Huberman interviews Dr David Anderson, a neuroscientist whose lab has produced landmark research on the neural circuits controlling aggression, mating, and emotional states in animals.
Summary
Andrew Huberman speaks with Dr David Anderson about the neurobiology of emotion, aggression, and mating behavior. Anderson argues that emotions are best understood as internal states — neurobiological processes that persist, generalize, and shape behavior — rather than purely psychological or subjective experiences. He presents findings from his lab showing that aggression and mating behaviors are controlled by distinct but closely interacting neuron populations in the ventromedial hypothalamus, and that estrogen receptors — not testosterone directly — are the key molecular drivers of aggression in male mice. Anderson also discusses the neuropeptide tachykinin, which is upregulated by social isolation and drives increased aggression, fear, and anxiety — and which can be pharmacologically blocked to reverse those effects, even in mice that have become chronically aggressive.
Key Takeaways
FULL TRANSCRIPT
Emotions vs. Internal States: Framing the Neurobiology
Andrew Huberman: I want to start with something fairly basic, and that's the difference between emotions and states. How should we think about them, and why might states be at least as useful a thing to think about, if not more useful?
Dr David Anderson: The short answer is that I see emotions as a type of internal state, in the sense that arousal is also a type of internal state, motivation is a type of internal state, and sleep is a type of internal state. They change the input-to-output transformation of the brain. When you're asleep, you don't hear something that you would hear if you were awake. So from that broad perspective, I see emotion as a class of state that controls behavior.
The reason I think it's useful to think about it as a state is that it puts the focus on it as a neurobiological process rather than as a psychological process. Many people equate emotion with feeling, which is a subjective sense that we can only study in humans, because to find out what someone's feeling you have to ask them, and people are the only animals that can talk that we can understand. That's how I think about emotion. If you think of an iceberg, it's the part of the iceberg that's below the surface of the water. The feeling part is the tip.
Huberman: What are some of the other features of states that represent below the tip of the iceberg?
Dr. Anderson: There have been people who have thought of emotions as having just two dimensions — an arousal dimension and a valence dimension. Ralph Adolphs and I have tried to expand that a little bit to think about components of emotion, particularly those that distinguish emotion states from motivational states, because they are very closely related.
One of those important properties is persistence. This is something that distinguishes state-driven behaviors from simple reflexes. Reflexes tend to terminate when the stimulus turns off — like the doctor hitting your knee with a hammer. It initiates with the stimulus onset and it terminates with the stimulus offset. Emotions tend to outlast, often, the stimulus that evoked them. If you're walking along a trail here in Southern California and you hear a rattlesnake rattling, you're going to jump in the air, your heart is going to continue to beat, and your palms will sweat for a while after it's slithered off into the bush, and you're going to be hypervigilant. If you see something that even remotely looks snake-like — a stick — you're going to stop.
Not all states have persistence. If you think about hunger, once you've eaten, the state is gone — you're not hungry anymore. But if you're really angry and you get into a fight with somebody, even after the fight is over, you may remain riled up for a long time and it takes you a while to calm down.
Generalization is another important component of emotion states. If they have been triggered in one situation, they can apply to another situation. My favorite example of that is: you come home from work and your kid is screaming. If you had a good day at work, you might pick it up and soothe it. If you had a bad day at work, you might react very differently to it.
The Neural Circuits of Aggression
Huberman: I'd like to talk a bit about aggression — the beautiful work of Da Lin and others in your lab. What are your thoughts on aggression, how it's generated, the neural circuit mechanisms, and some of the variation in what we call aggression?
Dr. Anderson: First of all, the word aggression in my mind refers more to a description of behavior than it does to an internal state. Aggression could reflect an internal state that we would call anger in humans, or it could reflect fear, or it could reflect hunger if it's predatory aggression.
The work that Da Lin did when she was in my lab — she found a way to evoke aggression in mice using optogenetics to activate specific neurons in a region of the hypothalamus, the ventromedial hypothalamus, or VMH. This followed the famous Nobel Prize-winning work of Walter Hess. In Hess's original experiments, he describes two types of aggression that he evokes from cats depending on where in the hypothalamus he places his electrode. One he calls defensive rage — ears laid back, teeth bared, and hissing. The other is predatory aggression, where the cat has its ears forward and is batting with its paw at a mouse-like object, as if it wants to catch and eat it.
If you think of the ventromedial hypothalamus like a pear sitting on the ground, the fat part of the pear near the ground is where the aggression neurons are, but the upper part of the pear has fear neurons.
Fast-forward from that — from a lot of work from Da Lin, now on her own at NYU, and with her postdoc Anna Gret Falconer — there's evidence that the type of fighting we elicit when we stimulate VMH is offensive aggression that is actually rewarding to male mice.
Huberman: They like it.
Dr. Anderson: They like it. Male mice will learn to poke their nose or press a bar to get the opportunity to beat up a subordinate male mouse. It has a positive valence. So it's become clear that the state of aggressiveness is multifaceted. It depends on the type of aggression and it involves different sorts of circuits.
Huberman: Why do you think there would be such a close positioning of neurons that can elicit such divergent states and behaviors? You're talking about this pear-shaped structure where the neurons that generate fear are cheek-to-jowl with the neurons that generate offensive aggression.
Dr. Anderson: If you think from an evolutionary perspective, it might have been the case that defensive behaviors and fear arose before offensive aggression, because animals first and foremost have to defend themselves from predation by other animals. And maybe it's only when they're comfortable with having warded off predation and made themselves safe that they can start to think about who's going to be the alpha male in their group. So it could be that brain regions and cell populations evolved by duplication and modification of preexisting cell populations, and that might be the way those regions wound up next to each other.
But I think there must be a functional part as well. One thing we know about offensive aggression is that strong fear shuts it down, whereas defensive aggression — at least in rats — is actually enhanced by fear. It's one of the big differences between defensive aggression and offensive aggression. And maybe these two regions are close to each other to facilitate inhibition of aggression by the fear neurons. We know for a fact that if we deliberately stimulate those fear neurons at the top of the pear, when two animals are involved in a fight, it just stops the fight dead in its tracks and they go off into the corner and freeze. So, at least hierarchically, it seems like fear is the dominant behavior over offensive aggression.
It's not just fight and flight, either. There are also metabolic neurons that are mixed together in VMH as well.
Hydraulic Pressure and the Drive Toward Behavior
Huberman: One of the concepts that's been raised is this idea of a sort of hydraulic pressure toward behavior — Konrad Lorenz talked about a kind of hydraulic pressure toward a given state. What's really driving that hydraulic pressure?
Dr. Anderson: One way that is helpful, at least for me, to break this question apart is to distinguish homeostatic behaviors — that is, need-based behaviors where the pressure is built up because of a need: I'm hungry, I need to eat; I'm thirsty, I need to drink; I'm hot, I need to get to a cold place. It's basically the thermostat model of your brain. You have a set point, and if the temperature gets too hot, you turn on the AC, and if it gets too cold, you turn on the heater, and you put yourself back to the set point.
You can think of this accumulated hydraulic pressure as being based either on something you were deprived of — creating an accumulating need — or something you want to do, building up a drive or a pressure to do that. The natural way to think about that, at least for me, is as gradual increases in neural activity in a particular region of the brain. In the area of the hypothalamus that controls feeding, Scott Sternson and others have shown that the hungrier you get, the higher the level of activity in that region, and then when you eat, the activity goes right back down again.
In the case of aggression, our data and others' show that the more strongly you drive this region of the brain optogenetically, the more of a hair trigger you need to set the animal off to get it to fight. VMH projects to about 30 different regions in the brain and gets input from about 30 different regions. So I see it as both an antenna and a broadcasting center — like a satellite dish that takes in information from different sensory modalities: smell, maybe vision, mechanosensation. Then it synthesizes and integrates that into a fairly low-dimensional representation of this pressure to attack, and it broadcasts that all over the brain to trigger all the systems that have to be brought into play.
Aggression is a very risky thing for an animal to engage in. It could wind up losing and getting killed. So its brain constantly has to make a cost-benefit analysis of whether to continue on that path or to back off.
Hormones, Estrogen Receptors, and Aggression
Huberman: As we're talking about aggression and mating behavior, I think about hormones. One of the common myths that persists is that testosterone makes animals and humans aggressive and estrogen makes animals placid and kind. As we both know, nothing could be further from the truth. The specific hormones involved in generating aggression via VMH are things other than testosterone. Could you tell us a little more about that?
Dr. Anderson: When we finally identified the neurons in VMH that control aggression with a molecular marker, we found out that that marker was the estrogen receptor. Other labs have shown that the estrogen receptor in adult male mice is necessary for aggression — if you knock out the gene in VMH, they don't fight. And it's been shown — a lot of this is work from your colleague Nirao Shah at Stanford, who is one of my former PhD students — that if you castrate a mouse and it loses the ability to fight, not only can you rescue fighting with a testosterone implant, but you can rescue it with an estrogen implant. So you can bypass completely the requirement for testosterone to restore aggressiveness to the mice.
As you say, it's because many of the effects of testosterone — although not all — are mediated by its conversion to estrogen by a process called aromatization, carried out by an enzyme called aromatase. People may have heard of aromatase because aromatase inhibitors are widely used in female humans as adjuvant chemotherapy for breast cancer.
Huberman: What's involved in female aggression that's unique from the pathways that generate male aggression?
Dr. Anderson: We and other labs have studied this in both mice and also in fruit flies. One thing in mice that distinguishes aggression in females from males is that male mice are pretty much ready to fight at the drop of a hat. Female mice only fight when they are nurturing and nursing their pups after they've delivered a litter. There is a window there where they become hyperaggressive. After their pups are weaned, that aggressiveness goes away.
So this is pretty remarkable: you take a virgin female mouse and expose it to a male, and her response is to become sexually receptive and to mate with him. Now you let her have her pups and you put the same male — or another male mouse — in the cage with her, and instead of trying to mate with him, she attacks him.
We recently showed in a paper — this is work from one of my students, Mengyu Liu — that within VMH in females, there are two clearly divisible subsets of estrogen receptor neurons. She showed that one of those subsets controls fighting and the other one controls mating. The male mouse VMH has both male-specific aggression neurons and generic aggression neurons. In the female VMH, the mating cells are only found in females — they are female-specific and not found in the male brain. We're trying to find out what these sex-specific populations of neurons are doing, but that indicates some of the mechanism by which different sexes show different behaviors.
The Overlap Between Mating and Aggression Circuits
Huberman: If one observes the mating behaviors of different animals, we know there's a tremendous range. With that said, when you look at mating behavior of various animals, you see an aggressive component sometimes but not always. Is it species-specific? Is it context-specific? And more generally, do you think there is crosstalk between these different neuronal populations, and that the animal itself might be kind of confused about what's going on?
Dr. Anderson: I can't really speak to whether this is species-specific because I'm not a naturalist or a zoologist. I've seen, as you have, lions when they mate in Africa — there's often a biting component to that as well.
One of the things that surprised us when we identified neurons in VMH that control aggression in males is that within that population there is a subset of neurons that is activated by females during male-female mating encounters. There's some evidence that those female-selective neurons in VMH are part of the mating behavior. If you shut them down, the animals don't mate as effectively as they otherwise would. What happens when you stimulate them, we don't yet know, because we don't have a way to specifically do that without activating the male aggression neurons.
But I think they must be there for a reason, because VMH is not traditionally the brain region to which male sexual behavior has been assigned. That's another area called the medial preoptic area. We have shown that there are neurons there that definitely stimulate mating behavior. In fact, if we activate those mating neurons in a male while it's in the middle of attacking another male, it will stop fighting, start singing to that male, and start to try to mount that male — until we shut those neurons off. So those are the "make love, not war" neurons. And VMH contains the "make war, not love" neurons. There are dense interconnections between these two nuclei, which are very close to each other in the brain. But it's also possible that there are some cooperative interactions between those structures as well as antagonistic interactions. The balance of whether it's the cooperative or antagonistic interactions that are firing at any given moment in a mating encounter may determine whether a moment of bliss among two lions may suddenly turn into a snap or a growl and a baring of fangs.
We don't know that, but certainly the substrate — the wiring — is there for that to happen. When we made that discovery initially, it raised the question in my mind whether some people who are serial rapists and engage in sexual violence might at some level have their wires crossed in some way, such that these states that are supposed to be pretty much separated and mutually antagonistic are not, and are actually more rewarding and reinforcing.
The Periaqueductal Gray, Pain Modulation, and Behavior
Huberman: I'd love to talk about the PAG — the periaqueductal gray. It's been studied in the context of pain, and in the context of the so-called lordosis response — the receptivity or arching of the back of the female to receive intromission during mating. In particular, is there some mechanism of pain modulation and control during fighting and mating? The reason I ask is that years ago I did a little bit of martial arts, and it always impressed me how little it hurt to get punched during a fight and how much it hurt afterwards. There clearly is some endogenous pain control that then wears off. What is the PAG doing vis-à-vis pain, and what is pain doing vis-à-vis these other behaviors?
Dr. Anderson: I think of the PAG like an old-fashioned telephone switchboard. There are calls coming in and then the cables have to be punched into the right hole to get the information routed to the right recipient on the other end. Because pretty much every type of innate behavior you can think of has had the PAG implicated in it.
In cross-section, the PAG kind of looks like the water in a toilet when you're standing over an open toilet bowl. If you imagine a clock face projected onto that, the PAG has sectors from one to twelve, maybe even more. In each of those sectors, you find different neurons from the hypothalamus projecting. So it could turn out that there is a topographic arrangement along the dorsal-ventral axis of the PAG and the medial axis of the PAG that determines the type of behavior that will be emitted when neurons in that region are stimulated. All of the evidence is pointing in that direction, but by no means has it been fully mapped out.
Now, the thing you mentioned about it not hurting when you got beat up during martial arts — there is a well-known phenomenon called fear-induced analgesia, where when an animal is in a high state of fear, as when it's trying to defend itself, there is a suppression of pain responses. The adrenal gland has a peptide in it that is released from the adrenal medulla, which controls the fight-or-flight responses, and that peptide has analgesic activity. It's called bovine adrenal medullary peptide of 22 amino acid residues. I only know about it because it activates a receptor that we discovered many years ago that's involved in pain — we thought it promoted pain, but it turns out it actually inhibits pain. It's like an endogenous analgesic.
Whether this type of analgesia is happening when an animal is engaged in offensive aggression or in mating behavior, I don't know, but it certainly is possible. And I don't know whether these analgesic mechanisms are happening in the PAG specifically. They could also be happening a little further down in the spinal cord. The PAG is really continuous with the spinal cord — if you just follow it down towards the tail of an animal, you will wind up in the spinal cord. So it could be that there are influences acting at many levels on pain, in the PAG and in the spinal cord as well.
I want to distinguish clearly between things that are not known that I know are unknown — which is in a fairly small area where I have expertise — from things that may be known but I'm ignorant of, because I just don't have a broad enough knowledge base to know them.
Tachykinin, Social Isolation, and Aggression
Huberman: Tell us about tachykinin. My understanding is that tachykinin is present in flies, mice, and humans, and may do similar things in those species.
Dr. Anderson: Tachykinin refers to a family of related neuropeptides — brain chemicals that are different from dopamine and serotonin in that they're not small organic molecules. They're actually short pieces of protein that are directly encoded by genes that are active in specific neurons and not in others. When those neurons are active, those neuropeptides are released together with classical transmitters like glutamate.
Tachykinins have been famously implicated in pain — particularly tachykinin 1, which is called substance P, one of the original pain-modulating peptides. This is something that promotes inflammatory pain. We did an unbiased screen of peptides and found that one of the tachykinins — Drosophila tachykinin — when you activate those neurons, strongly promotes aggression, and it depends on the release of tachykinin.
Now, the interesting thing is that in flies, just as in people and practically any other social animal that shows aggression, social isolation increases aggressiveness. So putting a violent prisoner in solitary confinement is absolutely the worst, most counterproductive thing you could do to them. And indeed we found in flies that social isolation increases the level of tachykinin in the brain. If we shut that gene down, it prevents the isolation from increasing aggression.
Since my lab also works on mice, it was natural to see whether tachykinins might be upregulated in social isolation and whether they play a role in aggression. This is work done by a former postdoc, Moriel Zelikowsky, now at the University of Utah in Salt Lake City. She found, remarkably, that when mice are socially isolated for two weeks, there is a massive upregulation of tachykinin 2 in their brain. In fact, if you tag the peptide with a green fluorescent protein from a jellyfish — genetically — the brain looks green when the mice are socially isolated, because there's so much of this stuff released.
She went on to show that that increase in tachykinin is responsible for the effect of social isolation in increasing aggressiveness, increasing fear, and increasing anxiety. There are drugs that block the receptor for tachykinin 2 which were tested in humans and abandoned because they had no efficacy in the conditions they were analyzed for. If you give those drugs to a socially isolated mouse, it blocks all of the effects of social isolation — it blocks the aggression, it blocks the increased fear, and it blocks the increased anxiety. As Moriel described it, the mice just look chill. And it's not a sedative, which is really important — the mice are not going to sleep.
Most remarkably, once you socially isolate a mouse and it becomes aggressive, you can never put it back in its cage with its brothers from its litter because it will kill them all overnight. But if you give it this drug — which is called osanetant, which blocks tachykinin 2 — that mouse can be returned to the cage with its brothers and will not attack them, and seems to be happy about that for the rest of the time. This is an incredibly powerful effect of this drug. I've been really interested in trying to get pharmaceutical companies to test this drug, which has a really good safety profile in humans, in people who are subjected to social isolation stress or bereavement stress. But it's very difficult for economic reasons to find a way to get somebody to test that.
The Brain-Body Connection and the Vagus Nerve
Huberman: As long as we're talking about humans, I'd love to get your thoughts about human studies of emotion. There's a heat map diagram in your book with Ralph Adolphs showing subjective reports of where people experience an emotion or a somatic feeling in their body — or in their head, or both — when they are angry, sad, calm, lonely, and so on. Those heat maps were not generated by any physiological measurement. How should we think about the body in terms of states?
Dr. Anderson: This goes back to something called the somatic marker hypothesis, proposed by Antonio Damasio, who is a neurologist at USC. The idea is that our subjective feeling of a particular emotion is in part associated with a sensation of something happening in a particular part of our body — the gut, the heart.
If there is a physiology underlying these heat maps, it could reflect increased blood flow to these different structures, and that in turn reflects communication between the brain and the body. It's bidirectional communication, mediated by the peripheral nervous system — the sympathetic and the parasympathetic nervous system — which control heart rate and blood pressure. Those neurons receive input from the hypothalamus and other central brain regions that control their activity. When the brain is put in a particular state, it activates sympathetic and parasympathetic neurons, which have effects on the heart and on blood pressure. These in turn feed back onto the brain through the sensory system.
A large part of this bidirectional communication is also mediated through the vagus nerve, which many listeners and viewers may have heard about because it's become a topic of intense activity. The vagus nerve is a bundle of nerve fibers that comes out basically of your skull, out of the central nervous system, and then sends fibers into your heart, your gut, and all sorts of visceral organs. That information is both afferent and efferent. The vagal fibers sense things that are happening in the body — the reason you feel your stomach tied up in knots when you're tense is that those vagal fibers are sensing the contraction of the gut muscles. They're also efferent, meaning that information coming out of the brain can influence those peripheral organs as well.
There's work from a number of labs just in the last six months or so where people are starting to decode the components of the different fibers in the vagus nerve, and it's amazing how much specificity there is. There are specific vagal nerves that go to the lung that control breathing responses, that go to the gut, that go to other organs — it's almost like a set of color-coded, labeled lines for those things. How those vagal afferents play a role in the playing out of emotion states is a fascinating question that people are just beginning to scrape the surface of. What's exciting now is that people are going to be developing tools that will allow us to turn on or turn off specific subsets of fibers within the vagus nerve and ask how that affects particular emotional behaviors.
Darwin recognized this brain-body connection as well. I think it's a central feature of emotion state and underlies our subjective feelings of an emotion.
Huberman: David, as a true fan of the work that your lab has been doing over so many decades, I know I speak on behalf of a tremendous number of people when I say thank you for taking time out of your important schedule to share with us what you've learned.
Dr. Anderson: I really have appreciated your questions. They've all been right on the money. You've hit all of the critical, important issues in this field, and you've uncovered what is known — the little bit that is known — and how much is not known. I think it's important to emphasize the unknown things, because that's what the next generation of neuroscientists has to solve. I hope this will help to attract young people into this field, because it's so important — particularly for our understanding of mental illness and mental health and psychiatry. We've got to figure out how emotion systems are controlled in a causal way if we ever want to improve on the psychiatric treatments that we have now. And that's going to require the next generation of people coming into the field.
Huberman: Absolutely. I second that. Well, thank you. It's been a delight.
Dr. Anderson: Thank you. Really appreciate it.