Episode 294 Ralph Adolphs | Andrew Huberman Transcript

Andrew Huberman interviews neuroscientist Ralph Adolphs on the science of emotions, emotion regulation, and social intelligence

Andrew Huberman speaks with Ralph Adolphs, professor of psychology, neuroscience, and biology at Caltech, about the neurobiology of emotions.

Summary

Andrew Huberman interviews Ralph Adolphs, a world expert in the neurobiology of emotions at Caltech, covering what emotions are, how they function in the brain and body, and how they can be regulated. Adolphs argues that emotions should be understood by their function — as flexible behavioral control states that sit between rigid reflexes and fully deliberate planning — and that tying emotions to conscious experience creates unnecessary scientific confusion. He presents evidence from amygdala lesion patients, amnesia studies, and autism research to challenge common assumptions about how emotions are perceived, stored, and expressed. Adolphs also shares personal reflections on how deliberate cold exposure, ultramarathon training, and a recent health challenge have shaped his own understanding and practice of emotion regulation. A significant portion of the conversation addresses the perception of emotions in others, including why facial expression research has been largely overstated, and how autism spectrum research is revealing stable individual differences in social attention.

Key Takeaways

FULL TRANSCRIPT

What emotions are and how to define them

Andrew Huberman: Welcome. Longtime admirer of your work, because you're one of the few people who has studied these things we call emotions, brain states, cognition, and much more in humans. There have been others in the field of neuroscience of course, but I would argue few people have combined a real grounding in neurocircuits with an interest in what most people think of as a psychological set of issues, in a way that also takes into account the animal data and really the broadest picture of this thing that we call emotions. I'm just going to start with a statement of gratitude. Thank you for doing that, because it's so easy for people to stimulate a brain area, see an animal or a human behave or speak in a certain way, and say, "Oh, this is the brain region for this." And you've long been pushing back on that.

Ralph Adolphs: It's a tall order, and one impetus for the work that I've been doing — and the book that I co-wrote with David Anderson, as well as the book I have forthcoming with Princeton University Press — is precisely to do what you just said: to integrate and pull together findings on emotion from animal studies and findings from social psychology. Typically, people in those fields don't talk to one another. They do very different studies, very different methodologies. So trying to integrate across disciplines that are that far apart, to really get a sense for what emotions are in a way that would satisfy psychologists and lay people about human emotions, but that would also speak to the mechanisms you can study in mice or flies as David Anderson does.

Andrew Huberman: So how do you think about emotions, and how do they relate to brain and bodily states?

Ralph Adolphs: The basic answer is that I think of emotions as functional states of a particular type. They should be understood by what they do — their function — rather than how they happen to be constituted in the human brain, or differently in an octopus nervous system, or in an AI, which we should talk about. The way to situate them is to think of control of behavior in response to particular types of environmental challenges. At the simplest level, you would have reflexes. I put my hand on a hot stove, or you tap the patellar tendon on my knee, and there's a very rigid behavior that's adaptive but just for that particular circumstance — I withdraw my hand. That's a reflex. At the opposite end, you would have goal-directed general volition and planning — everything we do all the time that's extremely flexible. And so if you have a bear or a tiger charging at you, that challenge requires a control of behavior that's intermediate between those two extremes. I can't have a reflex for a bear if it's far away, or if it's close, or if it's a rattlesnake or a tiger or all the different kinds of potential threats out there. Instead, I need something that's more flexible, which emotions are. So that's the big picture: emotions are functional states that have particular features or operating characteristics for dealing with certain types of environmental challenges like running away from a bear.

I think the question "what is an emotion?" is a good and valid question. It's not an ill-posed question. You occasionally run into the view that emotions don't really exist, or that you can't really say anything about them, that the question is just meaningless. I don't think that's the case. Paul Griffiths, a philosopher, wrote a book some years ago — I think 1997 or so — called What Emotions Really Are, where he first made the claim that they don't really exist. You could see how people come to that conclusion if you survey all the different theories out there — Joe LeDoux, Lisa Feldman Barrett, Paul Ekman — they look pretty different. So you might come away with the impression that even the experts can't agree on what emotions are, and that the words we have for emotions predate modern science, so it's just a confused concept. I don't think that's the case. I think there's room for revision, but I think everybody kind of knows what emotions are, and I think it deserves a straight answer that science should provide.

So let's start from the simplest case, which would be to list some examples. You could say emotions are anger, fear, disgust — you could list specific examples that are uncontentious, where pretty much everybody agrees these are examples of emotions. You could do the same if I asked you what's a car, and you would say, well, it's a Mercedes and a Honda and a Toyota. Or if I asked you what's a bird, and you would say it's a robin and a hummingbird and an eagle. That's not an answer yet. What you would then want to say is: what is it in virtue of which these examples are all instances of an emotion? What do fear, anger, etc. share in common? What features do they have that makes them emotions, just like what cars or bird species have in common that makes them cars or birds?

David Anderson and I did that in our book a decade ago — The Neuroscience of Emotion — where we listed some features of emotions that help answer the question. I elaborate on that and have a longer list in my forthcoming book. Maybe I'll just list a couple quickly. One feature of emotions is priority, which is that they can take over your behavior — something they share in common with reflexes. If I put my hand on a hot stove, I can't help but withdraw it. And if you're suddenly confronted by the bear, you can't help but run away or curl into a ball or whatever your adaptive behavior is. So they have priority and interrupt your behavior, which is something the Nobel laureate Herbert Simon already pointed out in a paper back in the '50s or '60s — that they're interrupt mechanisms of a certain kind. If you're just going about your business in normal goal-directed behavior, you always need to be monitoring for the bear or the tiger, the predator that's out there, and if you detect that, emotions have to take over or you're dead.

Another feature is valence, which psychologists ubiquitously mention — the simplest dimension along which you can organize emotions in terms of their similarity. Anger, disgust, and fear you might say are more similar to one another than either of those is with, say, happiness. There's some similarity structure. Some emotions cluster together, some are further apart. The simplest dimension — which Charles Darwin already mentioned in his book The Expression of the Emotions in Man and Animals, calling it the principle of antithesis — is valence: approach or avoid, pleasant or unpleasant, this basic valence dimension.

Andrew Huberman: Are there any neutral emotions? I ask this because your colleague at Caltech, Marcus Meister, once wrote a review where he said a good way to think about sensory systems, and maybe neural systems of all kinds, is that they're encoding for yum, yuck, or meh. And the meh is a very interesting one. Just like an animal, including us, can go forward, backward, or stay put. I sometimes wonder, because algorithms are repurposed across different neural systems, if there is indeed a meh emotion — or maybe the emotion is meh. Does that make sense? Because as long as we're talking about valence, anger and happiness are clearly at opposite ends of the spectrum, but what about apathy? Is there a state of prioritization of doing nothing?

Ralph Adolphs: It's a really good question. When you list these features — like priority and valence — one question is which of these are necessary. Do you need all of them, or are some of them sufficient? I think the answer is some of them are sufficient, but there are some that are necessary. I think priority is necessary. I think valence maybe not, as you were alluding to — I could imagine there's still valence but it's at a neutral region of the spectrum.

But then there are other features which are the ones that David Anderson has studied in particular in the hypothalamus in mice: scalability and temporal persistence. Scalability would be that an emotion can scale in intensity. One of the best examples is for the case of fear and the bear coming at you. Typically it would scale from anxiety and monitoring — there's the bear sort of out there — to fear, where you might run away or hide, to panic. That sort of threat-imminence scalability is something where emotions are different from reflexes, because reflexes are binary. It's all or nothing. I either pull my hand away or nothing. The knee jerks or not. There's not a scalable package in there.

Temporal persistence is even more essential to emotions and speaks to their function. You have this internal state — let's say of aggression or of fear — that has to last some time, because the bear might be around for some time. So you might be running away or attacking in anger. Those are phasic little bouts, but the actual internal state that is motivating those persists for quite a long time. Temporal persistence is really important, and it allows the person or the animal to incorporate lots and lots of information that's coming in and to have a motivational state suited for that kind of environmental challenge. It's not like a hot plate where if you pull your hand away you're done. The bear is out there and it might come back. It might still be lurking in the forest. So you better be afraid for some period of time.

David Anderson studied this a lot in terms of what neural circuits and what properties of neurotransmitters in the brain could actually implement persistence — really slow neurotransmitters like neuropeptides, for instance, which is what he studied. But my colleagues at the University of Iowa actually did a really cool study in humans that showed the temporal persistence of emotions independent of memory.

They took patients with amnesia — four or five people who had medial temporal lobe, hippocampal damage. They live in a time window of maybe a minute or so. I used to study these patients when I first started as a postdoc at the University of Iowa with Damasio. You can talk to them, they can process information, but if you go to the toilet and come back five minutes later, they would stand up and shake your hand and introduce themselves, because they would have no recollection of having met you before. Very strong anterograde amnesia, as it's called.

So they had those and controls. They showed them really sad movie clips. In all cases, the patients can understand and watch the movie clip. They get really sad and weepy. Right after watching these sad movie clips, everybody rates how sad they feel. The amnesic patients feel really sad. The controls feel really sad. You wait five minutes. You then ask the controls how sad they feel, and they say, "Yeah, I still feel a little bit of sadness because I remember having seen the movie." And if you ask them why they feel sad, they say, "Because you showed me the sad movie five minutes ago." The amnesic patients — you ask how they feel, and they say, "I feel really sad and I don't know why. I have no idea." And you ask them, "Do you remember seeing a movie?" They say, "Nope." There's no declarative memory of the movie, but there's still the temporal persistence of the sadness.

I thought this was a really cool study because it showed very clearly that there is temporal persistence in emotions — and sadness probably has one of the longest time constants, as we know from personal experience. But it's separate from just declarative memory. It's not that you're remembering that the bear is there. That's going on too. But independently of that, the emotional state has its own temporal persistence. The state has to last some time. And again, that's something that reflexes don't have. It just goes straight through. There's no temporal persistence.

Do animals and AI have emotions?

Andrew Huberman: It is striking to me that maybe 30 years ago or so, there was discussion about whether animals have emotions based on these criteria — priority, valence, scalability, and persistence. Is it generally agreed upon that animals have emotions? I believe they do because I love my dog and other animals and I want to believe that. But as scientists, if we step back, do they meet the criteria for emotions that you're setting up here?

Ralph Adolphs: Absolutely. Yes. If you have those criteria — those functional criteria — and you add one or two others, many states in animals, and certainly the states that we would ordinarily interpret as emotions in our pets, in our dogs and cats, would meet those criteria and they would have functional emotions.

We should also talk about the claim now from Anthropic that large language models have functional emotions as well.

Andrew Huberman: That's so wild. We have to return to that because it boggles the mind — how could a machine have emotions?

Ralph Adolphs: Well, again, these are functionally defined emotions, and I think there's no problem assigning to AI, to some extent, and certainly to many animals, those functional emotions. Where it gets complicated is that in us — and also by assumption in animals — I have no doubt they also have conscious experiences of emotions. And most people take the conscious experience of the emotion as the ground truth for the emotion. Emotions without conscious experiences of emotions sound like an oxymoron. It's like you can't have that. The conscious experience is the emotion.

Which is what William James said, which is what Joe LeDoux says. And I think it's implicitly there across essentially every psychological theory — the assumption that emotions really are types of conscious experiences, which I think is the intuition that if I ask you how do you know you're having an emotion, you would say because it's a certain type of conscious experience, because of how it feels. And if I asked you more specifically how do you know it's sadness versus anger versus fear, you would say because of how they feel. They feel different. And so we take the conscious experience, our own introspective access to the conscious experience, as the ground truth.

That has a lot of problems associated with it — deep philosophical problems. But just practically it has two big problems that I think science should avoid. One is it gets really problematic to know whether or not animals have emotions, because we don't know how to test whether or not they have conscious experiences. I have no doubt that they do. It's just that there's no scientific way to test that currently. So we should have different criteria. And the other question is: if you tie emotion to the conscious experience of emotion — that is, you conflate emotions with feelings — well, then you're not studying emotions, you're studying consciousness. I have great respect for people that are studying consciousness, but I don't want to study consciousness. That's too hard.

There's lots of debates and disagreements. So if we can avoid that and study emotions in the absence of needing to study conscious experiences of emotions, we can move forward with the science, which is what every other discipline does. If you're a vision researcher, you study vision — you don't need to study consciousness. I mean, you can, but there's a huge amount of work that people do studying vision and how that works in the brain, studying memory, any other cognitive process, without needing to study conscious experiences. They're typically accompanied in humans, and if you talk to the lay person on the street and say you're a vision researcher, they might take you to be studying the conscious experience of seeing. But scientifically, most people working in those fields don't study consciousness, and it's perfectly valid to study vision and memory in horseshoe crabs or Aplysia where nobody has any claims about consciousness. We should do the same thing with emotion, and then we can have an integrated science across species that can move forward without needing to tackle the difficult problem of consciousness.

Again, yes, for sure we and other animals have conscious experiences of emotions, and that's very important. But we also have conscious experiences of seeing and memory. As a scientific discipline, we should separate the conscious experience from the process, from the emotion — and that's the way we make progress. That's one of the main arguments that David and I had in our book, and that I reiterate in the forthcoming book.

Emotion regulation, monitoring, and the role of awareness

Andrew Huberman: At risk of taking us in a philosophical direction, I'd like to actually ask a biological and psychological question. Given these criteria — priority, valence, scalability, and persistence — I think all of us have had the experience of an emotion coming on but not in its overtaking, maximal form. Like, okay, we feel ourselves getting annoyed and we can observe the emotion growing. The arousal state is familiar to us. We understand, let's assume in this example, why it's happening. Maybe you're in line at the store and the person ahead of you is taking everybody's time by doing something that's really unnecessary and kind of obnoxious to us in that moment. And so our frustration is rising and we are observing the emotion even though the emotion is us — it's within us. And then at some point we might blurt something out or feel compelled to say, "Excuse me, could this maybe be taken care of?" But you have some control. And there's this almost like running together of two consciousnesses — like there's two of us in there at the same time, observing self, observing the outside world, and then at some point the prioritization gets to the point where we just do something, and sometimes it's a decision and sometimes it overtakes us.

Now at the opposite extreme, we've all observed — you can just go on X, the social media platform, and for some reason the algorithm likes to serve up a lot of examples of people on planes losing it, people going completely unhinged, just melting down. And it's very, in our language, hypothalamic, right? And so people are drawn to that.

Ralph Adolphs: I have an antidote to those people.

Andrew Huberman: They should take ice baths. I can tell you more. Take ice baths. I think this helps seriously.

Ralph Adolphs: Yeah. Well, ice baths. Let's definitely talk about ice baths and ultra races, because you're training for one.

Ralph Adolphs: No, but I think what you said is exactly right and extremely important for humans, and probably unique to humans as opposed to other animals, which is that we can be conscious of and monitor our own emotions, and by doing so we can regulate them to some extent. So emotion regulation and control, and monitoring emotion, being aware of what emotion you have — those two go hand in hand. Your Stanford colleague James Gross spawned this huge field in the '90s, I guess when it started, of emotion regulation, that is now huge. I just finished my term as co-editor of this journal Affective Science, and I would say like a quarter or a third of the submissions are about emotion regulation. It is so important in humans, and most pathologies and difficulties about emotions are not about the emotion per se. It's about our ability or disability to control them, to regulate them.

Andrew Huberman: If you don't mind — because of your understanding of emotions and states and neuroscience, but also your experience as a human in the world — has anything you've learned changed the way that you watch your emotions grow, decide when to step in, when not to step in? You don't seem like an impulsive person. I'll seed this question by saying I have a few friends who were in tier one special operations, and that's high-risk, high-consequence work. We've had Alex Honnold on the podcast — high-risk, high-consequence endeavors. And they all say the same thing, which is that emotions cloud logical judgment. So they're somehow far from emotions, or their emotions don't get prioritized. But of course all of these people have healthy marriages, healthy families — they're loving people, they have anger, they have emotions — but they've categorized emotions as clouding judgment in the professional or peak performance context. Therefore they don't allow them to be prioritized. Sounds a little bit robotic, a little bit mechanical. But in some ways, provided they're allowed to be emotional and let emotions take over in the right context, that's kind of the ideal circumstance, right? So what do you know that has impacted the way that you think about emotion regulation that perhaps people could consider, if not benefit from?

Ralph Adolphs: I think it's a mixture. On the one hand, you absolutely need emotions. If you have no emotions, as we know from Antonio Damasio's original work in his book Descartes' Error — which he put out just as I was starting as a postdoc in Iowa in 1994 — if you have massive frontal lobe lesions that make it the case that you're no longer emotionally responding to stimuli, the consequences are devastating. You cannot make complex decisions in your life, you can't organize your life, you're not motivated to take the right path through life, because you don't feel one way or the other about the different options that are available. So it's clear that not having emotions is very bad. On the other hand, being able to regulate them and control them is definitely a human skill that's very important and that helps you a lot.

The first part is what you were talking about earlier, which is the ability to monitor your emotions, to identify your emotions. I think Lisa Feldman Barrett actually first did the main work on this. It's a topic called emotional granularity — at what grain can you conceptualize and be aware of your own emotions.

Andrew Huberman: How important is it to use language to provide that granularity? I've certainly experienced feeling a bunch of things around a particular scenario, and if you can't separate out the elements, it can feel overwhelming. People always say if you can name the emotion — but I find that a little tricky because sometimes it's not one emotion. It's a bunch of things. It's confusion, it's frustration. Are those the same thing? I don't know. And I think many people, myself included, sometimes it just feels like too much. Not because the emotion is so big, but because it's such a tangled jungle where there isn't really language to disentangle.

Ralph Adolphs: Absolutely. I don't think you need to put it into language. I think people's strategies vary. Some people think in words, some people think in pictures, most people are a mixture. And if you look at different languages — if you're French or German speaking, like me, bilingual — we have different words for different emotions. But you need some sense of the concept, which I think the concepts for emotions are pretty much universal across languages. All these different examples of emotion words from different languages — Schadenfreude is one that people know now, joy at another person's misfortune, from the German. Gigil, I think, is often trumpeted as another example — I forget what the culture is, but this is the emotion you feel if something is just unbearably cute. There's a great book — Tiffany Watt Smith has a book called The Book of Human Emotions, which is basically like a little dictionary, an entry from A to Z of all the different emotion words in different languages. But the fact that we don't have a word for gigil or for Schadenfreude in English — you perfectly well have the concept. I just described it to you. You have no problem understanding what the emotion is. And so I think as long as you have some concept — which need not be put into words — some way to think about and differentiate the different emotions that are going on in you, that gives you a purchase on your ability to evaluate them and say, "This is an emotion I want," or "This is an emotion that I don't want," and then to act on it through various strategies — cognitive reappraisal, for example, or acceptance, depending on what your strategy is.

Andrew Huberman: Yeah, the psychologists seem to disagree here, because many people — some who've been on this podcast, all of whom are experts — will say if you don't allow yourself to feel the emotion, then it grows. And other people will say, well, if you focus on something, then it grows. And this is where I think neuroscience can be of great help, because probably both things are true and it depends on the context.

Ralph Adolphs: That's exactly right.

Andrew Huberman: Maybe this is a good opportunity to talk about sort of dynamic attractor states — and I don't throw out that language to confuse people, but there is this idea in neuroscience that if you and I both talk about this cup and I slide it forward and we start looking at it, more and more brain circuits are going to be devoted to it. We could spend several hours on it. It's not a very interesting conversation, but if we do that about an emotion we feel because of something somebody said to us at work or online — I think we're all familiar with the idea that we can go down these rabbit holes within our own mind. And it's a very dangerous way to live. Just knowing that it exists perhaps can be helpful, because it's almost like if you go far enough down a path, it's very hard to extract yourself.

Ralph Adolphs: Absolutely. Cognitive reappraisal or any other sort of complex reappraisal strategy for emotions is a double-edged sword, and it's trainable. If I really focus on, okay, I'm starting to feel sad, I really don't want to feel sad, I really don't want to feel sad — the more I think about it, the sadder I get. Same with anxiety. Some emotions are probably more prone to going down this rabbit hole than others. But again, it's a trainable skill, and there are other ways of dealing with it where you can kind of just stop that process from going down at all. You want to stop it at the earliest front end. One quick answer to the question of what's the most successful strategy is: the earlier you can stop that process, the better. In the first place, don't get yourself into a situation where you're going to be anxious or guilty or sad or whatever emotion you want to avoid. Situational avoidance, just in terms of structuring your environment, would be one strategy. And then once you're in it, I think it's idiosyncratic to a large extent. You would have to figure out what works for you, but it's definitely trainable. Two parts are trainable: one is the specific strategy, so you can do cognitive reappraisal without overthinking it and going down the rabbit hole and actually feeling the emotion even more. And the other one is just the strength and automaticity of control that you have. Which brings me to the ice bath.

Ice baths, autonomic training, and emotion regulation

Ralph Adolphs: So briefly — I started doing these a while back. One of my former colleagues, Justin Feinstein, who now runs a float clinic on Maui — I forget how it came about, but we were talking about floating and ice baths, and he mentioned you and mentioned Wim Hof and said, take a look at these guys. And I saw Wim swimming in some frozen lake. Looked kind of interesting. Anyway, so I started doing ice baths some time ago. They weren't that popular yet back then. And now my wife bought from Amazon, like a horse watering trough or some big thing. Insulated it, put on a lid — a lot safer than being out on a lake, too. And I get these 40-pound ice bags every week. They know me very well at Joe's Ice up there. Every time I go in, they say, "Hey boss, having a good party again tonight?" Because I eat a lot of ice.

So this is the interesting thing. The first week it is terrible — super painful. You go in, it's like a whole-body cold pressure task. The second week it's also, you know, your breathing goes up, your heart rate goes up. But gradually, and kind of with some automaticity to it, now I go into the ice bath and immediately my heart rate goes down, my breathing goes down, I get super relaxed. So this is the interesting part that I propose about emotion regulation: it generalizes for me. This is an experiment of one, but for me it generalizes. Previously, prior to the ice bath, driving around Pasadena — and I don't know, in the last year or two this seems to have risen somehow — I'm trying to park on the street. There's a lane next to me. I have my signal out. There's a parking space. It's clear what I'm trying to do. I see somebody come up behind me. They can see I'm trying to park. They can move over. No, they come right up behind me, go on the horn, and you immediately feel just an immediate anger response, right?

Well, after the ice bath, it's flat. So there was an immediate automatic downregulation of my autonomic emotional response to a psychological stressor — somebody honking at me — that was trained and generalized from the ice bath. It's an experiment of one, and I'm not claiming this as a published study. But I think there's a direct autonomic training component as well, in terms of the strength and the automaticity, the ease with which you can regulate emotions — they can become kind of automatically regulated. You don't have to overthink it. It becomes smooth and effortless to some extent. And that's the place where you want to get.

Andrew Huberman: I've known Wim a long time, and although I sometimes unfairly get tacked to the deliberate cold exposure thing a little too much — because we talk about many things on this podcast, not just that — I do find it very useful. Years ago my lab had collaborations with the special operations community, and they'll tell you that one of the reasons it's such a wonderful tool is it's so reliable. You always get an increase in adrenaline, norepinephrine in the brain, and to some extent — and this might surprise you, perhaps not — you get often reductions in cortisol, which is kind of paradoxical. But when people practice, it's very reliable, and provided it's not so cold or it's not an open lake where you're doing hyperventilation, five minutes after a shower, it's extremely efficient.

Ralph Adolphs: That's a good long while, even in a cold shower. And there are now evidence of improved resilience to subsequent stressors of the sort that you described. There are some okay papers. I think it's a wonderful tool, especially for this buffering of the immediate stress or anger response — very powerful.

Ultramarathon training and emotional resilience over time

Andrew Huberman: Before we went on mic, we were talking about a different type of stressor, which is the ultramarathons that you're currently training for — a 100-mile race, which is by the way super impressive even to shoot for. I know you've done others. We were talking about something that I think is relevant to bring up now, which is you were describing how at some points during these races you just feel like complete garbage, but then you start to feel better. And so then the next time you feel like complete garbage, you have the cognitive knowledge that you felt terrible then and better again, and so you're starting to integrate over longer periods of time. So here's another example of emotion regulation but over much longer time scales.

Ralph Adolphs: No, I think this is one reason — there are many reasons — that I love these ultramarathons and sort of identify as an ultra runner, even though I didn't do it for decades and I'm just now getting back into it. It's because of that nonlinearity. You have these major ups and downs, and the sort of mantra that ultra running has co-opted is "relentless forward momentum," but the psychological equivalent is just relentless optimism — optimism in the face of adversity. You have to just keep that mindset, and everybody there is like that. You come around like mile 40, mile 50, mile 60, you come around some curve and you look and feel like crap, and people go, "Looking good, looking good." Nobody says, "Looks like you're not going to make it." Nobody says that. So everybody is super positive. You're hanging in there, and it is really hard. You know, you've just been going downhill, your knees are shot, you have blisters, you've been vomiting, you're super tired, and you extrapolate and you think this is not going to work. But if you refuse to give up — it doesn't always work, but about half the time or so — it levels out and it starts going back up again.

I mean, you feel wrecked at mile 50. I feel wrecked at mile 30. So trying to think forward and say, "Wow, I have to do this over again, this is not fathomable." So it's a really interesting exercise in not giving up and being able to do something that you think is impossible to do. For that matter, all of your listeners would think it's impossible. If any of your listeners went running with me, they would conclude two things: one is they're faster than I am, and two, I'm not going to be able to do 100 miles — which means if I actually finish the 100 miles, all of you would be able to do so as well, which I think is true. I think anybody who's basically healthy, if they only wanted to — which is the key — could actually do it. So it's just a really interesting exercise. It's only 30 hours. Well, "only," but it's only 30 hours. It's not like years. How hard can it be? It's hard, but it's doable. It's definitely doable. And so it's just in that sweet spot of something that seems impossible and is just unbelievably hard but doable, and that's attractive somehow if you can do it — which of course doesn't happen all the time.

Andrew Huberman: Yeah. And I don't know if there are any data on this, but for people that have lived through really hard things and come out better, or have pushed themselves to deliberately do hard things of the sort that you're describing, I do think that the memory of that can recall our body and brain's ability to lean into other hard things.

Ralph Adolphs: I think so. I mean, it must be the case. I don't know of any good experiments on that, but — post-traumatic resilience, I guess they call it. Some people come out of very difficult circumstances, maybe they do the work they need to or not, and they just say, "Okay, I'm better for life because I'm out of it." Other people carry the wounds of that forward in a way that their psychological and sometimes their physical health is a downward trajectory. It's a very interesting thing about how we should conceptualize pain and pain ending. Some people seem to live in the discomfort of the pain having even happened.

You can certainly persevere and overcome severe adversity if it's time-limited, and I think your own sort of belief in yourself and your capability to overcome adversity is one important component. Another one is something like gratitude or awe, and you get to appreciate and just value the fact that I'm here talking to you. I'm actually grateful and feeling some sense of awe right now at just existing. And so getting to that point where every morning you wake up and you have this sense of gratitude and awe, or some combination of those emotions — I think that's a really valuable thing to carry out of all of this. And it takes time to kind of accumulate, but I like living that way.

Andrew Huberman: I want to get back to the emotion system, but I feel inclined to mention — we had Dacher Keltner on the podcast, who studies awe, and he defined it so beautifully. He said it's when we bring our attention from a small spatial scale to a large spatial scale, or from large to small. But then during the course of our discussion, he added that this is also true in the time domain. So when we live our life day-to-day, we start to feel kind of lost in time, and awe becomes less common. But when perhaps we run a 100-mile race across 30 hours, we realize how much can happen in 30 hours. One could say, "Oh, well then you finish the race and the rest of life seems trivial." But it seems the opposite is true — that we realize that time is this very elastic thing, at least in our minds, and all of a sudden we have awe for the fact that we have so many days of life to live forward. It's very interesting, and it almost speaks to the ice bath again — you have this very compressed stressful experience that gives you that gap between stimulus and response that seems very hard to access unless we subject ourselves to these kinds of experiences.

Ralph Adolphs: I think that capability — people think chimpanzees maybe can feel awe, but I think it's uniquely, at least as we have it, a uniquely human emotion. And it's based on a broad sort of metacognitive ability we have, which is to detach our conscious experience from the here and now and do like mental time travel. It's pretty derivative of the need to understand other people. I could put myself into your shoes right now. I can imagine sitting where you're sitting and seeing Ralph Adolphs here. I could imagine zooming around and adopting another person's perspective in space. We certainly zoom around in time and reminisce about stuff from the past, imagine things in the future. And I think awe is like that too — we adopt a different perspective that's not just tied to the here and now, which I think is probably a crucial ability that distinguishes the nature of conscious experience in humans: we have some control over it. I can simulate what it's like to be you. I can imagine being in the past or in the future. I can zoom out broadly when I'm feeling awe.

And I think animals are more just stuck with whatever the senses are delivering to them right now. That's the content of their conscious experience. We can use it flexibly. And that's why we can plan and write novels and fiction and think about things. Most of the time, if I ask you what's in your consciousness, what are you thinking about, it's not stuff that's in front of you. You're daydreaming. You're all kinds of places. And that really gives the human mind the power that distinguishes it from any other animals and is the reason we have everything that makes us so different. So awe, I think, is an example of a uniquely human emotion that derives from that ability to adopt a vast point of view, or a point of view that's not tied to just whatever the senses are delivering to you.

Perceiving emotions in others — faces, dynamics, and the limits of facial expression research

Andrew Huberman: As long as we're talking about assessing the emotions or the perspective of another, what are some favorite experiments — either yours or others' in the field — about perception and assessment of emotions? I've seen a number of studies like, oh, you know, people with lesions here or there see scary faces as not that scary. What's the general conclusion about those studies now in 2026? Because I have a feeling a lot of what was in the textbooks when I was a graduate student may not still hold up.

Ralph Adolphs: When I started as a postdoc in the early '90s with Damasio, I started my career on that topic by studying this famous patient SM, who had selective bilateral amygdala lesions. The finding was that she's unable to perceive fear in emotional facial expressions.

Andrew Huberman: Could she feel fear?

Ralph Adolphs: Subsequently, we also did that. It took some years because it's just harder to do. Justin Feinstein, who has the float clinic in Maui, was the first author on that paper. Some years later, we took that same patient to haunted houses. We took her to an exotic pet store where there were snakes and spiders. We showed her horror movies, a whole bunch of things, and found that she also doesn't seem to consciously experience fear. So there seemed to be this very specific deficit for fear tied to the amygdala, from one patient. That's a big caveat — does this generalize? Well, yes, the amygdala is important for fear, but just saying it's a center for fear is overly simplistic. It's embedded in a big network of structures. There are individual differences and so on.

The perception of emotions has been studied a lot and has probably been the topic of the most acrimonious debates to some extent. Paul Ekman, back in the '70s — famous, and by the way I dedicate my forthcoming book to him because he probably had the biggest impact on modern emotion science of any single person — his claim to fame was he went to New Guinea and these non-western cultures and claimed that universally across all cultures, people recognize these so-called basic emotions from facial expressions. If you Google emotional facial expressions now, you will find pictures of them. Lots of AI companies and computer algorithms are predicated on the idea that you can read out the emotion from somebody's facial expression. He had six of them: happiness, surprise, fear, anger, disgust, and sadness. Sometimes contempt is added in there.

Well, it turns out it's more complicated than that. Actually, Lisa Feldman Barrett of all people co-authored a paper with me and three other people a couple of years back — the title was something like "Emotional Facial Expressions Reconsidered." We took stock of the whole literature and asked: is this really true? Is Ekman's claim really true? Can you tell how someone feels from their facial expression? And the answer is: not nearly as well as you would think, and in general, not. The reasons are methodological, and once you think about it, they make a lot of sense. The emotional facial expressions that Ekman pioneered — I used them in the first 1994 Nature paper with the amygdala lesion patient — those aren't actually what you normally see walking around the world. People don't have those facial expressions. They are posed facial expressions by actors. They're very extreme. And then the task is like, well, here's a list of emotion words — happiness, fear, anger, disgust, etc. — pick the one that matches the face. That's the typical task.

Andrew Huberman: So it's multiple choice.

Ralph Adolphs: Look, if I gave you a bunch of emoticons and a bunch of emotion words and asked you to match them, you could do that very easily. It wouldn't show you anything about emotions. It shows you that there's by convention some association between these things. That's it. So when you actually see real-world facial expressions, and when instead of giving a list you just ask people to freely generate a word that describes what they see, you get something that has much more variation, much more heterogeneity — which then was part of the ingredient for Lisa's whole claim that variation is the norm and that in fact emotions are in some sense constructed rather than packages to be found in nature. I disagree with that part, but emotional facial expression for the perception of emotions in others is much more complex.

I was just talking before coming in here to one of your colleagues about your cute dog, and they've done studies — there are several studies now — where dog owners, if you ask them, "Can you tell if your dog feels guilty when you come home because it did something bad?" Dog owners are pretty convinced they can tell. Well, if you do a control experiment, they're at chance. They think they can tell. So our conviction that we can tell something about people or animals, their emotion states, just from looking at the facial expressions — our conviction vastly outstrips our accuracy. We're pretty convinced we can do it. We're mostly not so accurate. Which makes sense, because ordinarily it's not just some facial expression. If I see you now and you look sad, I wouldn't stop there and say, "Andrew Huberman is sad." I would ask you how you're feeling, what's going on. I would probe you and get a lot more information. Maybe you would say, "Oh, no, I'm just — this drink I had tasted bad or something, that caused me to make a certain facial expression." So in general, these artificial lab situations where all you see is a picture of a face and you're asked for the emotion — we don't do that in the real world anyway. We would probe further. We would take into account the whole circumstances. We would ask the person, and then you can disambiguate these emotions. So it's not that we cannot tell what emotions people or animals feel, but we don't do so normally just from a single slice like a facial expression.

Andrew Huberman: I feel like at least in the context of romantic partnership, so much more of our assessment of another person — and their assessment of us — is actually a mathematical integration of noticing when the frequency of speech or frequency of text is heading in a negative direction relative to the mean, right, at a particular time of day. And I think the brain is really good at saying, like, normally somebody's pretty communicative in the morning but less at night. But suddenly they're a little more terse with us in the morning — maybe not even harsh, there are just fewer words exchanged. And then the brain doesn't think, "Oh, they looked angry or they looked frustrated." You go, "Oh, they were a little quieter this morning than usual." And then if it's a very quiet person, you have fewer data points to work from. But we become exquisitely tuned to this.

Ralph Adolphs: Absolutely. Much more than, like, "Oh, they're carrying their shoulders differently." At the extremes, I think we see this — somebody walks in the door and you can tell they're sad. We're looking for change, like it's a time series. The number of times — or even just like if you've probably had this experience — in an email with a colleague, there's a certain kind of language that's used in an email that's different than the usual language. Like they've never really used these words before. You don't even identify which words those are.

Huberman: Years ago there was this chart that some graduate student or postdoc put out — that if the director of a lab signs off with their name, it means one thing; if they punctuate their email, it means something else; if there's just a dash with a first initial, you know there's all this. And you kind of laugh, and then you realize, wow, they're really paying attention to all these subtle cues. But those seem like the kind of cues that we rely on much more than, like, "Do they look sad? Do they sound happy?" And I think we vastly underestimate this. I don't know if there are any good studies of this, but this feels like the more important signal.

Adolphs: We just finished a study like this where people texted — that's all they had. But it's dynamic. It's a conversation. There's like a whole transcript. And then the question is, what can you tell about a person both in terms of their stable personality and their current mood just from that text? It turns out you can tell a lot. People did a study recently where people generated about a page or so of diary-type written text and asked large language models like GPT or Claude to analyze that text without knowing anything else and just tell them the personality profile. And it's as good as the ground truth, or as another person or a clinician would be.

Huberman: As good as a clinician?

Adolphs: It gets close to the ground truth and it's as good as a good friend or a person who knows the person. So if you score the Big Five — which is what they did in that study, extraversion, conscientiousness, neuroticism, etc. — AI can tell that just from a transcript of text quite well. So the information is clearly there in what we say, what we write, and the order in which that progresses, and it tells a lot about a person — enough that AI can basically diagnose your personality from just text.

It's almost like our brain is very attuned to when we don't get what we were expecting to get in terms of the amount of interaction. One of the key ingredients of Lisa Feldman Barrett's thinking, certainly in recent years, is prediction — that most of the information comes from when there's a mismatch between what you're predicting about a circumstance and something that's just different, because that carries information. Something changed. And Lisa's work, Lisa's thinking on emotions, incorporates prediction in a big way these days. There's something that you're predicting about the world, but suddenly there's a tiger or something unexpected, and you need to incorporate that. And that carries important information. But absolutely, we're always predicting, and we're taking into account things that change.

The original studies — like the ones I started doing as a postdoc, where you just show these static Paul Ekman pictures of facial affect — well, we never see that in people. It's always dynamic. The face is changing. And so Philip Schyns, Rachel Jack, colleagues in Scotland for instance, have studied this in great detail and used dynamic actual videos of people making faces, and asked at what point does the brain start to represent different emotions depending on not how the face looks but how the face is changing. Dynamics are everywhere. It's super important.

Huberman: Because when we look at babies and they start to smile for the first time, or they express what we think is humor or something like that, we are so tuned to faces because they can't really do much. And whether or not they're interacting with us or not is really dependent on our behavior. There's a lot of control there. We're putting them in front of us. As they develop more autonomy, then these sorts of dynamics really emerge.

Adolphs: Right. The face is visible, and the human face has way more muscles and control than other animals do. Great apes have similar sets of muscles, but monkeys or your dog — you can't make those kinds of faces. My dog just sits there every once in a while and I project onto him what that's about. But he's a real sweetie. I had another bulldog mutt who was much more coarse, less affectionate. This one is like pure butter. But of course I don't know what's going on in his head. He could be thinking, "This guy's the biggest jerk in the world. We should be walking right now."

The amygdala, fear, and interoceptive panic

Huberman: This patient SM who had bilateral amygdala lesions — could she experience autonomic arousal? Could she experience stress? Could she experience anxiety under any conditions?

Adolphs: She could. And that's really important. You might think it's just like the whole emotion system is gone, you've just deactivated the entire autonomic nervous system and that's why you don't have emotions. But it's much more specific than that, and fits with thinking about emotions as evolved functional modules for dealing with particular types of statistically recurring environmental challenges that share some challenging feature in common. So she doesn't have fear to external stressors — like we went with her into a haunted house, she poked the monsters and wanted to touch things. Same with the snakes and the spiders in the pet store. But we did a study — again Justin Feinstein was the first author on this one as well — with her and with a couple of patients who had similar lesions, in collaboration with René Hurlemann from Germany, where we put a mask on them and had them inhale carbon dioxide. If you do that, you get a panic attack in about half the population, because it makes you feel like you're suffocating. As soon as you get that carbon dioxide in, it changes the pH in your blood, that's detected by the brain, and you get this massive air hunger — like you suddenly can't breathe. Your brain goes into this alarm mode of suffocation, like a panic. And she and the other amygdala patients had a full-blown panic attack when they inhaled the carbon dioxide. So this interoceptive signal seems not to depend on the amygdala and to depend on other brainstem circuits.

So we had a very important dissociation between different types of fear, which I think is a bottom-line story for the way that the science is informing our folk psychology. These categories we have — like anger or fear or disgust — there are many varieties. It's much more finely differentiated than that once you look at the science. And once you look at the brain systems, they're distinct brain systems — in this instance for fear to something out there in the world versus panic of something happening inside you, like having a heart attack or not getting enough air.

Bodily maps of emotions and the insula

Huberman: Maybe we could talk a little bit about the bodily representation of these things we're calling emotions. Years ago I posted on social media a picture of a figure from your book with David Anderson, which was describing a study where people self-described — so it's purely subjective reporting, as I recall — where in their body they felt different emotions. So it's kind of like a heat map in the forehead for certain emotions and in the whole body for others. And there were a wide range of emotions. Admittedly it's subjective report, but does this hold up? I think to myself, where do I feel sadness — heartache? Well, I tend to feel it around my heart, but maybe that's because it's called heartache. Where do I feel anger? Well, extreme anger I'm going to feel as a whole-body experience. Same thing with awe. So what, if anything, is important about the regularity and/or the variation in where in our body we feel emotions? Some people would say this is their whole experience of life — they feel things in their body all the time. Some people feel more cerebral — they kind of think things more than they feel things.

Adolphs: I think it's a great question. That figure came from a paper in PNAS by Lauri Nummenmaa and colleagues in Finland — he and his colleague Hanna Saarimäki. They've done some great recent studies also on experience of emotions in humans using functional MRI. What you mentioned — the experiment is: I just give you a word. I'm not actually making you feel the emotion. I'm asking you, "Sadness — draw on the body where you think you would feel that emotion." So if you think about it, it's certainly not about actual physiological changes in the body. Nothing's measured in the body. People are just drawing on a little mannequin where they think they would feel the emotion. It's also not about conscious experiences of emotion — I'm not showing you sad movie clips or anything like that. I'm giving you a word, a single word. So it's about concepts. It's about where in the body people's concept of sadness or happiness or fear corresponds to changes in the body.

That's an interesting question, but you have to narrow it down to that, which is a fundamental problem with most emotion research, including that one. I think the title is something like "Bodily Maps of Emotions" or "Bodily Maps of Subjective Experience." Maybe it should be something like "People's Concept of Where in the Body Emotion Words Are Associated with Changes." What's interesting about that study is that Lauri has gone on to look at it in different cultures. He's looked at it developmentally across children and asked how these concepts develop. So it's super interesting work, but it's ultimately about people's concepts of how emotions correspond to changes in the body, not the real actual changes in the body.

But then, as you were saying, of course there are real changes in the body when you really feel an emotion as well. And that's harder to study, but people have studied it quite a bit with a focus on the part of the brain that gets a lot of input from the body, which is a part of cortex called the insula. There's lots and lots of work on the insula. It gets input from all the organs in your body — from your heart, from your liver, from your kidney.

I had a kidney stone not too long ago. For pain — exteroceptive pain, like we were talking about the hot plate before — you pull your hand away, or there's some kind of reflex. For the kidney stone, that's not adaptive. There's nothing to pull away. So my wife — this was in Switzerland, we're in a remote little town in the middle of the night — my wife sees me curled up into a ball right on the floor, just kind of breathing loudly and making rocking motions, and goes, "What's wrong?" And I tell her, "Well, I have a kidney stone." And then she asked me, "Can't you just stop that behavior? Can we get on a plane and fly to the US so you can get ultrasound, maybe get treated for the kidney stone? Can you stop curling up into a ball and making rocking motions?" And I was kind of thinking about that through my haze of pain and said, "Well, no, it's kind of like the hot plate and pulling away. It's a really strongly motivated behavior to just kind of withdraw basically." So it was a very interesting adaptive behavior. She called the equivalent of 911, and in the middle of the night some guy came down and put a bunch of opiates into me, and he kept having the scale of, "On a scale from 0 to 10, how painful is it?" I would sort of go from 10 to 9.5. It was extremely painful. But eventually they got me to a clinic there, I passed the stone, and it was an interesting experience. Pain is not pain. If you have a pain in your eye or you have a pain from a hot stove, this is a very different type of pain, and one that's worse than the hot stove because you can't escape.

Huberman: You can't rub it.

Adolphs: You can't do anything. Just curl up into a ball and pray. It was horrible. At any rate, so at that moment, had I been in an MRI study, my insula would have been strongly activated. This is a region of the brain that, according to theorists — Bud Craig was the first, Antonio Damasio has theories in a similar vein — would be the substrate for the conscious experience of the emotion or of pain in your body. So it's getting all this input from your body and representing the changes in your body as a conscious experience.

This was the original theory that William James had back in 1884. He wrote this paper in Mind — "What Is an Emotion?" — and his theory was that you see the bear, you run away, your heart rate goes up, and then you feel the emotion, which is a consequence of you perceiving all the changes in your body that happened first. Now, that's the conscious experience of emotion. And a lot of that depends on representing and perceiving everything that changes in our body. But to my mind, that leaves the question of what the emotion is that's antecedent to that. You first have to trigger all the changes in your body and run away. What's doing that? That's the emotion. The conscious experience may well come later and unfold over some time. But the emotion state, separate from the conscious experience as I've emphasized, would be the one that actually motivates all the behaviors and all the different changes in your body.

So long answer to your question: there are two different ways of talking about the role of the body. The one that you alluded to in this figure from Nummenmaa and colleagues is about people's concepts of emotions, and they're certainly tied to changes in the body. Whether or not those are accurate, I think we just don't know. Is it the case that those drawings — if you put electrodes on and measured blood flow changes — that's actually where things change when you have awe or anger? I think we just don't know. That's a set of studies that remains to be done.

But then in addition, there is the fact that all of the changes in your body are represented in your brain in the insula, and that seems to be a big part of the substrate for feeling the emotion, for the conscious experience of the emotion, and other things — nausea, pain, kidney stones. So it's not limited to emotions, but the strong feelings that we have — like a tightness in your chest if you're feeling really sad — that would be represented in your insula.

Huberman: In Eastern medicine there are strong beliefs about different organs being associated with different emotional states. But I always have to remind people that a key feature of schizophrenia and psychosis of other kinds is clang associations — two words sound alike and people link them. And so we also are subject to this in a much milder way in the world of health and science and especially in the public discourse. People will say, "Oh, a walnut looks like a brain. Walnuts are good for your brain." Well, they have some fatty acids that are probably good for neurons, but it has absolutely nothing to do with the shape of the walnut. It's amazing how these ideas can perpetuate.

Adolphs: Well, and I think there is some truth to them. I think it just has — the honest answer is it has not been tested.

So if you ask someone — Lisa Feldman Barrett has made this point repeatedly — that there really is no systematic association between specific emotions and specific changes in your body. Well, if you look at those studies, people put a few electrodes on and they measure heart rate changes, blood pressure changes, etc. It is extremely crude. You're looking at a tiny tiny slice of what's going on in your body. What you would want to do is have a detailed high-dimensional readout of your gallbladder and your liver and your kidney and your heart and changes all over your body, and look at that and ask: is there some kind of complex association between specific emotion states and changes in your body? And I would bet that there is, but we just don't have those data yet. Those studies remain to be done.

Huberman: Yeah, they really should be done. Because I could make up a story where if someone or an animal feels extreme anger — of the sort that they would have right before they were in a fight — that certain immune organs would deploy a bunch of things that prepare for wound healing or prepare for resilience in the aftermath. That just stands to reason. I of course have no data for that, but —

Adolphs: Well, and the link between the brain and the body — we now understand that it is so detailed and so intricate that really brain and body are very very tightly coupled. And we do know that there are signatures — they're distributed, they're complicated — but Lisa Feldman Barrett herself, in collaboration with Tor Wager, has published neuroimaging studies showing that for basic emotions like anger and fear and sadness and so on, there are reliable signatures of brain activation. They're distributed, they're spread out, they're complicated, they're not just in one place. But given that association between emotions and patterns of the brain, and the fact that the brain and the body are so intimately tied together, I think it's a completely reasonable hypothesis that you would find a similar high-dimensional signature of changes throughout your body that are systematically associated with certain types of emotions.

Media, attention capture, and the hijacking of emotion systems

Huberman: I had a thought spring to mind based on our earlier discussion. From the perspective of media and advertising, it makes perfect sense why all media and advertising is incentivized to get you to prioritize an emotion to the point where it takes over behavior — to go click and buy something, or to vote a certain way, or to hate a certain group or like a certain group. And so really what they're fighting for is this prioritization thing — the path to behavior is through the prioritization.

Adolphs: In attentional capture. It captures your attention.

Huberman: Which is what emotions are designed to do. And once you've captured people's attention, you have a captive audience and you can control them much more.

Adolphs: And in some sense it's obvious, right? You get people emotional, they're more likely to behave in a certain way versus another. I think we underestimate the extent to which all day long we are bombarded with these signals. Subtle advertising doesn't really work well. Someone said — I don't know who said it — but all A/B testing leads to sex or violence, because if you really just look at what's likely to grab people's attention, eventually you regress into the hypothalamus. The more primitive states inspire more attention, which tells you a lot about the world. And there are ethical lines and societal lines where you say, "No, that's too extreme, too much violence, too much skin." But it's all veering toward the hypothalamus, so to speak. It's all getting more and more primitive. There's a lot of processing that goes on in all these subcortical structures, and certainly what we understand best in terms of emotions are all these subcortical circuits, and a lot of that is below the radar of our conscious awareness.

Art, music, and emotion — what features trigger generalizable responses

Huberman: At the opposite extreme, what happens when I go to a museum and I see a piece of art that just for some reason makes me feel a certain way — maybe sadness, maybe happiness? Do you think that art of that sort is able to capture the circuitry related to particular emotions across people? Not just about my unique experience — and art is subjective, right? So there are things related to taste. But let's assume this is abstract art, so there's no apparent faces or animals or depictions of scenes. But let's say out of a hundred people, more than half say, "Yeah, I look at that and I just feel peace." Is it the case that some art can tap into neural circuits, or some music can tap into neural circuits independent of the lyrics?

Adolphs: Sure. And the interesting question is which particular features in the music or in the abstract art are driving those generalizable emotion responses. And they're certainly there. One of the earliest studies on emotion actually was done by Anne Blood and Robert Zatorre in the mid-'90s or so, using positron emission tomography — PET — rather than fMRI at the time. What they did was they played music to people, and they set it up so that you and I could hear the same piece of music. In your case, the music really does something for you. The particular measure was: does it send shivers down your spine? Do you get chills, goosebumps from listening to the music? And it does that for you, but not for me. Same auditory stimulus, different emotion. And from the imaging, they showed that the insula, the structure we were just talking about, the amygdala, several of these emotion-related structures are indeed activated when you have shivers, chills down your spine listening to music.

For abstract art, my Caltech colleague John O'Doherty had a paper a little while ago where again he found something generalizable. He decomposed abstract art into the constituent visual features — you can feed it into a neural network and get a representation of all the different shades and colors and the basic visual composition — and from that you can reconstitute whether or not it is liked or not liked, whether people find it pleasant or unpleasant. So it doesn't come out of nowhere. Yes, there's something about even abstract music or abstract art — there are some auditory or visual features there that are generalizable and that can be used to evoke emotions.

The big question then is: can we link that back? Obviously, the emotions didn't evolve in order to appreciate abstract art. They evolved to detect tigers and bears and other things. Well, are there similar kinds of visual features? Presumably there are. So there are some visual features in the real world for which particular emotions evolved, that can also find a place in abstract art and evoke similar emotions. That's the basic story. But the answer to your question is yes — even for these very uniquely human, abstract kinds of stimuli, people have done work and found particular features in those stimuli that will trigger emotion states in humans.

Hardwiring versus learning — face processing, neural populations, and the fan-in fan-out architecture

Huberman: Talk a little bit about hardwiring versus learned emotional states at the brain level. We know there are brain areas that process faces — just the organization of two dots and a vertical line and a horizontal line below it. There are neurons that like that organization of features. Are there units in the brain that are tuned to a smiling face versus a frowning face at the level of almost like a very sparse emoji? And of course you'd love to be able to do the experiment in somebody who had never been exposed to them for the first time, but that's impossible to do.

Adolphs: Yeah. With respect to face processing in general, the sort of consensus is it's both hardwired and innate in some sense — there's something there in the brain already at birth that's experience-independent. But then it also depends on normal maturation and development and experience. People like Rebecca Saxe at MIT, for instance, have done functional neuroimaging studies on babies and shown them faces and asked how this processing system develops. It's always a mixture. People tend to think in terms of black and white — it's hardwired and innate, or it's learned. Nothing is either/or. It's always both.

A good example is the visual word form area. There are regions in the brain — like the ones that my former Caltech colleague Doris Tsao studied — that are specialized for faces. But there's also a region in the brain, typically on the left in most people, that's for recognizing words and letters. Now that couldn't have evolved, because we had a long history of brain evolution before written language was ever there. So the question is how does it come about? It can't be hardwired or innate, because there wasn't long enough evolutionary time to evolve a region of the brain for figuring out words. But if you take a look in preverbal infants at the connectivity of this region that's going to become specialized for processing words once the infant can read, it's already wired up in a way to other brain regions that predisposes it to be able to do those computations — that with experience it can later be used to recognize words.

Coming back to your question about smiling versus frowning faces — there are two parts to it. The first part is I can find you a single neuron in the brain that responds selectively for anything you like. There's no shortage of them. But those single neurons don't do anything in general, or aren't important to understand how the brain works, because the rest of the brain that is getting its input from all these neurons is not like a reflex. It's not listening to only that one neuron. It's getting input from tens of thousands of neurons. So at the population level, if we record from thousands of neurons and ask from that: is there information in a thousand neurons from which the rest of the brain listening to those thousand neurons — or we as the scientist — could decode smiles versus frowns? The answer is yes. We've done that in collaboration with Ueli Rutishauser and Shuo Wang and colleagues, where we can record from epilepsy patients — same kind of work as Eddie Chang does. And of course if you think about it, there has to be information in the brain neurons that would distinguish frowns from smiles, because you can distinguish them. So where else would it come from? It has to be there in the brain. But it's not there mechanistically at the level of single neurons. They're very noisy. They typically respond to many different types of cues. So you always have to think about hundreds or thousands — a large ensemble of neurons — because that's really the kind of unit of relevant information, because that's what the rest of the brain listens to.

Huberman: I was thinking about how in a hierarchical system like the visual system, we have neurons that are tuned to circles more or less — center surround. We talk about circles, then bars, then oriented bars of a particular angle, then moving bars, etc. So you can build feature detectors for lots of different things. I guess I was wondering when we come into the world, are there cells in the fusiform area that not just like faces but that are prepared to respond to two dots — the eyes — a vertical thing — the nose — and then a bar that is either bent up at the edges or bent down at the edges, and then we build on that scaffold?

Adolphs: I think that exact experiment hasn't been done, but in broad strokes the answer would be that at birth it's crude, and it becomes much more refined and granular through experience. And then indeed exactly what you said happens. Coming back to emotions — if you think about how the state of fear comes about in the brain from seeing the bear or the tiger, well, at the level of the retina it's just a bunch of pixels. It's some blobby black shape that is looming and coming closer, etc. But then progressively you have to construct more and more abstract representations. Deep convolutional neural networks that do object classification do exactly that. They take just pixel-wise input but then build something abstract — like right now I recognize Andrew Huberman there. If I look over here, I can still recognize you whether you're on the left or the right, whether you're close or far away. I would be able to recognize you invariantly with respect to those details, even though at the level of the retina, like a camera taking a picture, the detailed pixels can be vastly different. You need to abstract from that an invariant representation regardless of the details.

Same thing for emotions. Regardless of whether it's a bear or a tiger or a big armed pickup truck coming at you or lots of other kinds of stimuli, they all need to funnel in to a central abstract representation. That's another key feature of emotions — they require abstraction. And that's what makes them so flexible. That's why David Anderson and I have described it in our book as this sort of fan-in fan-out architecture. I can get to the emotion state from seeing a bear or smelling a bear or hearing a bear or hearing a twig snap or hearing somebody yell "there's a bear." Zillions of different inputs — completely unlike the rigid reflex, which only has one input and that's it. They get to the central state of fear, and then the outputs are also diverse. I can hide, I can run away, I can defend myself. And so that flexibility that the central emotion state affords — you can get to it in many different ways because many different sensory cues are informative about a potential threat, and then you can behave in very flexible ways that are context-dependent. That's exactly what emotions allow you to do and what makes them so different from reflexes. Reflexes are just a single path through. There's only one way to trigger it and there's only one behavior.

Huberman: So if you had only those, you would have to have a separate reflex for the bear and for the tiger and for every possible large threat out there, which you couldn't possibly have. It's not efficient.

Social and emotional intelligence — reading the room and the costs of hyper-attunement

Adolphs: The fan-in fan-out visual is very, very useful. And it gets to the next topic I'd like to ask you about, which is building social intelligence and emotional intelligence, because we define that different ways. People who seem unable to read the room, so to speak, versus people who are hypersensitive to other people's emotions — not necessarily responding to them, but they're really in tune.

Huberman: I'll go out on a limb and say that some of the happiest, most functional people that I know and have observed — and when I say functional, I also include that they take great care of the people around them, their families, their co-workers — are not super emotionally in tune with the emotions of others. Meaning it doesn't seem like they're clocking every little nuance. And some of the more labile people that I've known are really in tune with what everyone in the room is feeling, who did this, who didn't do that. Going back to our earlier conversation — so much of how we gauge other people's emotions is an integration over how often they tend to acknowledge us when we walk down the hall. If the first time I met you, you walked right by me, and the second time and the third time, and then the fourth time you say "Good morning," I'd say, "Oh, this is a really delightful person." If you did the opposite — said hello three days in a row and then just looked at me and brushed right by — I'd probably think, "That's weird." So there's some advantage actually to not being in tune with all the subtle cues of everyone around us. What in the brain is known about social awareness, about reading the room? We talk it up these days because indeed empathy is important, but this is a tricky one. There's also a lot to be said for walking into a room, acknowledging, okay great, and then getting your work done. You don't want to be wrapped up into people's emotional states unnecessarily, because they're labile, they're all over the place. There's a cost.

Adolphs: Well, I think it's extremely heterogeneous. This is not one-size-fits-all, and there's not one type of emotional intelligence skill that fits all occasions. Many different kinds can work. They depend on the context, they depend on the circumstances. Initially when James Gross had emotional regulation, it was sort of focused on some basic mechanisms like suppression, reappraisal, acceptance. And now the big two topics in emotional regulation are variability and flexibility. So there's not just one way, but you need to be able to have many many different options and you can vary your emotion regulation responses.

But the other big part of emotion regulation — which is obvious once you think about it, and any parent would think about it — is it's interpersonal. It's between people. You regulate other people's emotions and you have to pick up on other people's emotions and you influence other people's emotions. It's a bidirectional interaction. And so I think a big part of emotional intelligence comes from being able to read that, probably in a fairly automatic way and in a context-dependent way. And depending on the context, it can be the case that what's most adaptive is if you really get involved with other people and you're very empathic, or you just take over, and that can work too depending on the circumstances.

So there are multiple strategies, and I think for any of these — social intelligence, emotional regulation, any of these metacognitive skills — the most successful skill is if you don't just have one, but you have variety and you can switch between them and you can use context to switch between those. Just like physiological heart rate variability — you don't want to train to just be able to have a low heart rate. You want to be able to train to have a variable, big range. Same thing with emotions. What's most adaptive is being able to have a big range of emotions and being able to switch adaptively depending on the context between them. And I think that's where real long-term, context-dependent social intelligence comes from — for people that really come out ahead. It's not one tiny slice where they do well and don't do well in lots of other circumstances. It's the ability to be really adaptive and flexible and switch around.

Huberman: Which requires a keen sense of oneself — where we are in our internal state — and a keen sense of context. In fact, I observed a really impressive physician recently, moving through the hospital environment. He was just all agency. But then with the patient, it was as if he dropped into a bubble. The patient was understandably a little nervous about what was to come, and it was remarkable. I was like, "Okay, that is superb bedside manner." But then it comes time to do the actual work, and that's not a time for emotion. The life and well-being of the patient relies on being absolutely focused on the task at hand.

Adolphs: It's related again to the presumptive association between emotional granularity — what Lisa Feldman Barrett pioneered decades ago, the ability to differentiate and be aware of all the variety of emotions in you — and emotion regulation flexibility, that you're able to adopt many different emotion regulation strategies. But in order to do so, you need the emotional granularity to be able to be aware of all the different emotions that are going on in the first place. Two things: one is an ability to differentiate at a very fine grain, and the other one is to step back and have some kind of metacognitive oversight and control. So you can see, okay, here's this full spectrum of things I could do, and you can take a look at all of those, incorporate the context, and pick the one path amongst many that's the most adaptive.

It's interesting. Some of my favorite scientists are very emotive people. And then some of my favorite scientists — I'm thinking of one in particular when I was down at UC San Diego when my lab was there — most people would probably label as a little bit on the spectrum. He would walk into a faculty meeting where someone else was presenting and he would say out loud to me, sitting there, "Is this any good?" And you know, it's very disconcerting when you're giving the talk. But I was going to say there's also something incredibly refreshing about it, because it's almost like they're asking the questions they need to ask. You're a little embarrassed for them, but it's liberating. You feel like, ah, we've broken the social mold here a little bit. And someone would say, "Excuse me, so-and-so is presenting," and they'd say, "Oh, sorry," and then just sit down and listen. They're entirely polite, very actively engaged in the conversation. But I sometimes wish — how cool would that be? This person has a little bit less context-dependent flexibility around emotions, and I have to imagine they move through life in a way that's kind of liberated.

Huberman: Well, they wouldn't have this huge cognitive load all the time.

Adolphs: And they're extremely smart. That's correlative, right? But like incredible — actually a physicist turned neuroscientist. And I just thought, man, that's pretty cool. And then the more you go into engineering and science fields, the more of these people you encounter.

My lab studies autism amongst other things. And you certainly would have the case that as a spectrum, yes indeed, you would have mathematicians, engineers, lots of people who are very successful who are on the spectrum to some extent. And I think, coming back to what you were saying, it's sort of liberating or simplifying. If you had to really effortfully, cognitively think about how to regulate your social behavior and make all the right moves and not offend anybody, and you were worried about that all the time, it would be overwhelming. And I think it is overwhelming for many people with autism — you need to somehow internalize it and it has to be relatively effortless and automatic. It has to become automatic at some level, and then you can move through life fairly smoothly in a socially very complex way. And that's what's difficult for many people with autism — they lack that part. They can compensate to some extent, but the cognitive load starts really being an impediment. If you have to track everything — remember to make eye contact, remember turn-taking — if you have to cognitively remember all those things, it's just overwhelming. So you have to be able to somehow have it be automatic at some level.

Autism, social attention, and gaze tracking research

Huberman: Let's talk about autism. And I'll just put the mention up front because this is a topic that sometimes gets people triggered. Let's leave out for the time being — not because it isn't important, it is important — let's leave out severe cases of autism where people need lifelong care. Let's talk about the rest of the spectrum. Something like this former colleague of mine in San Diego — a little bit on the spectrum, by my read highly functional in all aspects of life, but definitely not tuned in to what other people are thinking or doing. And context mattered — he understood what not to do in a broad sense, but he was free of this burden that you described. So there's a lot of emphasis nowadays on kids who have autism of the sort that we're describing and the need, we're told, to train them to make eye contact and read faces. This lands square in your work. Maybe you could fill out a bit of what's now known, what's thought, what's not known.

Adolphs: It's a great topic. The way that I personally have approached it is not with a primary interest in autism. I don't have a family member who has autism. I don't have any personal experience. It's really been just like with the amygdala lesion patient — my main motivation has been to understand the mind, cognition, emotion in general, and with respect to autism by understanding individual differences. Everybody's different from everybody else. And autism is one particular dimension that we just have a lot of tools to look at — social interaction, social awareness, ability to read other people's emotions. And so that dimension of social variability maps onto one of the facets of autism spectrum disorder. That's why we've looked at both people who do not have any diagnosis of autism — there's still a range of autistic-like traits that you can quantify with questionnaires — and then once you get to a DSM-5 diagnosis, you have more and more severe cases of autism with respect to social functioning.

Coming to your question, there's a lot of really interesting work and a lot of opportunities now that we've been actually using in our lab as well, to look at the question of processing about faces, eye contact, etc. Right now — how much eye contact are you making with me? How much eye contact am I making with you? It is extremely variable. The eyes move around. They carry a lot of information. Socially, there's a gigantic range there. It's data that are now fairly easy to quantify and collect in hundreds or thousands of people using their webcams. And so we have a study going on where people watch, for example, a Zoom video of other people interacting and talking, and we use the webcam on their laptop to record where they are looking on the screen when they have that Zoom video going on. So we can quantify across large numbers of people, and the basic finding has been that people who score higher on this dimension of the autism spectrum — of social interaction, with just questionnaire-based measures or a psychiatric diagnosis — make less eye contact, as well as other features. We're looking at contact with the person who's talking, distraction to other things. We have experimental features where there's a person talking, they're making eye contact, there's like TV or something that goes on in the background, and we can quantify to what extent is your visual attention — your gaze, which we can quantify from the webcam — captured by somebody looking, somebody talking, somebody smiling or laughing, versus the TV going on in the background as a distractor.

And we can put all of that into a big model and ask: what is it that normally drives visual attention, and what is it that distinguishes people with autism and where they would look? The basic answer is that the social features — like eyes, or who's talking — have less of a weight on where autistic people would look than the distractors. They tend to be more distracted. And that seems to be a temporally stable, personality trait-like feature of autism spectrum disorder. So it fluctuates to some extent, and we're trying to quantify whether it interacts with whether you've slept badly the night before or you're stressed or other things that can change. Those do have an effect. But the answer seems to be it's fairly temporally stable and it's almost like a signature for individuals. There are some people that are real face-lookers, and there are some people that look at faces hardly at all, and that seems to be stable across time and like a basic personality trait.

Huberman: In this population of people diagnosed —

Adolphs: Across the board for everybody. If I just take the general population, on average it's the case that those people who would have a diagnosis of autism, or who score higher on autistic traits even if they don't have a diagnosis, look less at faces. But across everybody, it's the case that some people look a lot at faces, some people don't. And that's temporally stable. So in general, right now, if you quantify it from this podcast — how often do I make eye contact or look at you? And then you ask, in a classroom or in a lab meeting, how often do I make eye contact? It would be a similar proportion. I'm just like not a face-looker or eye-looker, and you might be. It's like a stable personality trait, and it's associated coarsely with this dimension in autism spectrum disorder.

But of course the million-dollar question is: what else is it associated with? Is it predictive of career success? Is it predictive of anything else? And so we don't know. We're measuring everything else and getting lots of questionnaires on these people to ask what the associations are, and whether there's any causal hypothesis that we would come up with — that if you don't look at faces a lot, there are some consequences, and if you look at faces a lot, there are other consequences. We're not there yet. That would make very interesting predictions.

Huberman: What should we say?

Adolphs: Well, that's another big question which we're looking at in these longitudinal studies — to what extent can you change who you are? We were talking about emotion regulation. You can regulate your emotions. You can have this metacognitive insight. You can have some control. Can you change if you're extroverted or neurotic? Can you change your personality? And the answer seems to be: to some extent. For some variables more than others. But we're trying to quantify that. So we have this big longitudinal data set where we give people tons and tons of questionnaires about their personality, about their mood, about lots of things, and are asking to what extent over time do they get pushed around and change, and which ones stay constant and constitute the core of their personality that's unchangeable. I think it's a super interesting question — what parts of you can change and what parts of you cannot?

Autonomic RPM, spontaneous movement, and the body-mind interface

Huberman: Yeah, I think about this all the time. I feel like some people — this is pseudoscientific what I'm about to describe — they seem to have very high autonomic what I call like a resting RPM. I have a former colleague — you ever see Botond Roska from Basel give a talk? He moves a lot. He's very electric. He's also very thin, and his brain — he's exceedingly smart — has a high basal metabolic rate. Like a hummingbird. He moves a lot. And then other people move less. I think about this all the time because I'm fascinated by it. I love dogs. I go to this AKC Meet the Breeds thing every year. You get to see all the different genetic variation, which of course relates to body size but also temperament. Bulldogs are probably more on the slow mellow side compared to Yorkshire terriers that are more excitable. Some people just seem to have a lot of spontaneous movement. Other people are more stable in their bodily movements. And it does seem to relate to how excitable they are or not. So you can imagine if you're humming at a higher RPM, you're more prone to action-forcing.

Adolphs: It's a super interesting observation, because the question of course is to what extent does it relate to how their mind is working. As we're talking now, we're moving our hands, we're moving our body, we're moving our face — not just sitting there speaking. So to some extent, all of these bodily movements are interfaced with thought. If I had to sit completely still and not move, it would be pretty hard to have this conversation. So to some extent, bodily movements actually help us structure our thoughts. It's not just like a large language model. It's not just language going on. It interfaces with the body, and it punctuates thoughts to some extent. So I think it's super interesting, and it's certainly the case that the way that somebody moves their body does say something about how they're thinking as well.

Huberman: Yeah, I've been reading my audio book and I asked the sound technicians to allow me to stand while I read it. Much easier to read an audio book and to put in inflection. You can actually lean in when you want to emphasize something, and when you need to work through a passage that needs to be very finely structured with little emotion, you clasp your hands and do it. Our friend Eddie Chang tells me that the hand representations in the brain are very close to the language center. So these were probably our first tools of language, or at least they co-evolved.

Adolphs: Well, your mouth and your hands are the most dextrous and over-represented in the brain. You can control the world with them and move them the most.

When the pandemic started in 2020, we weren't allowed to go on campus at all. So we sat around in our apartments. I sat at home. We had lab meetings on Zoom, only Zoom, and lectures. All the classes were on Zoom. Prior to that, everything was face to face. I would stand up and lecture. We would talk face to face. And suddenly switching to just seeing a bunch of often little black squares for the whole classroom — that was a hard switch. That lack of an interactive nature is something that we're looking at now. I think it's going to be super cool. I have a graduate student just starting this. The experiment is: I would see your face on a screen and that's it. I would see a video of you. I would Zoom with you right now and we could be having this podcast over Zoom. I actually thought we were going to have this over Zoom. So I'm very happy it's not over Zoom — actually better in person. And then if you're in person, right? It feels different. It feels different. And so we're trying to parameterize that degree of social realness of social interaction. And then we're doing that with people with autism, asking where does it change and how difficult is it for people with autism to interact with another person? Is it equally difficult over Zoom? Is it equally difficult over Zoom if you have the selfie camera turned off, so the other person can't see you and you know that? We're trying to look at the degree of actual social interaction as a relevant dimension, and I think it makes a huge difference. It feels completely different being face to face than if you're on Zoom, let alone if I just watch a video that's not actually aligned in time.

Flow states, solo time, and the value of internal training

Huberman: The way I described the higher or lower RPM before may have made it sound like one or the other was better. But I think sometimes about Alex Honnold, who was a guest on this podcast, and there have been claims that his amygdala doesn't react as much. I think Alex kind of chuckles a little bit at those studies because he does experience anxiety and fear, just not in the context of this highly trained situation he's put himself in every single day for so long. But I do wonder whether people who go out on long runs — there's not a lot of talking going on — or Alex, he's devoting his physical and mental energetic resources toward climbing and figuring out how to navigate high-risk, high-consequence situations, totally just in flow. You're just totally absorbed in what you're doing. Once he's got it, but like to work out the puzzle of how to maneuver so that he can actually free solo it is a remarkable thing.

But I almost get the impression that if people speak less, they're able to devote more mental and physical resources to the thing they're doing. I never ran ultras, but I recall running — I used to do a long 8 to 10 mile run on Sundays. I come back from those, and there's a fatigue component that confounds what I'm about to say, but one is a lot less reactive when you've spent a bunch of time just in your own head and you've learned to tolerate the voice in your own head. I find that I'm much calmer after I've spent a bunch of time alone, because I've gone through that initial stage of the mind fluttering, and then you want to communicate with people. I'll lock up my phone in a box these days to do real work, and I come out of it and things come my way and they're sort of out there, and unless they really demand my attention, they kind of just don't grab me. Whereas I think when we're highly engaged in our own mental dialogue, it's almost like we're trying to converse with the world. So I think there's real value to doing these solo treks and to devoting time to getting past that need to engage with the world.

Adolphs: Yeah. I mean, I'm sure it's idiosyncratic. It'll vary. But I think that particular aspect that you mentioned — that you can just be comfortable in your own mind and that that's beneficial — I think that's a big part of lots of training techniques and big part of lots of meditative techniques. But again, it's very individual. It's not one-size-fits-all. But the ability to not just be driven by sensory input and be comfortable with your own mind — it's not that you're doing nothing at all. You're training your own mind to be comfortable. And I think the consequence of that is then subsequently, when you encounter challenges in the world, you're actually better equipped to deal with those and better equipped to regulate your emotions. When you're just at peace with yourself, your brain is being trained all the time and it's being trained in a useful way. The ability to detach and decouple from the constant challenges and sensory information that's normally coming in — it's like sleep allows the brain to actually train itself and focus on particular internal regulation strategies that are then subsequently useful for dealing with the world later.

Huberman: Could we say that one form of potentially useful emotional intelligence or social intelligence training is to just spend some time in your own head, either moving or not moving — you could do it meditating or you could do it running — but to just resist the temptation to get pulled out into the world and just get good at sitting with these things that we call emotions and thoughts?

Adolphs: I think so, especially — it's again going to be individual and it's going to vary — but I think so, especially if you can get some kind of metacognitive or reflective awareness and work on what is going on in your mind. And ultimately it's training control. It's training control of your emotions and training control of your thoughts. And if you're constantly bombarded and just dealing with stuff coming in, you don't have time to do that. It's called social media.

Exactly. Yes. Of which I have none. And so if you're just by yourself — if you're running or if you're meditating or if you're in an ice bath — I think you actually get a particular type of internal training that you just don't have the leisure to get if you're bombarded by social media, for instance.

Meditation at the start of lab meetings — a practice born from George Floyd

Adolphs: You know, every beginning of my lab meeting now, people have their laptops open and their phones out. We put them down and we actually have five minutes of silence, a sort of meditative silence, for people to clear their minds at the beginning. Five minutes or three minutes. At the beginning of every lab meeting, we have a short meditative session. I put a picture up of some beautiful scenery on the north shore of Kauai in case people want to look at that, but many people just close their eyes. People can do whatever they want. But the idea is to clear your mind and focus and relax and get into a frame of mind where you're actually most receptive to listen to the research presentation that's coming up, because you come in and your mind is all over the place — there's an email here and the phone is going and your mind is buzzing. So quieting that, finding some focus, and then being able to engage subsequently much better with the real world, listening to the talk. I think that helps, and we do that every lab meeting — Mondays at noon — and it seems to work for people.

Huberman: How long have you implemented this?

Adolphs: We started it in 2020 when George Floyd was killed, specifically to remember George Floyd. So there was a very specific starting point, because many people were upset and we figured we needed some kind of remembrance or just a moment of silence to acknowledge things. And then it morphed into — you can remember whatever you like, you can just relax — and then it sort of morphed into this idea: well, let's just have a moment of silence and just recalibrate from the busy world and everything that's just been going on, to now. It's good to do on average every day in any case, but certainly right before we launch into the research presentation at a lab meeting. So we sort of morphed from a moment of silence to remember somebody to a short meditation essentially.

Huberman: I love it. That my neuroscience colleagues are doing ice baths, training for 100-mile races, and doing meditation.

Adolphs: The question is, is it helping? Am I getting more Nature papers as a consequence of any of these?

Huberman: That's a different metric. That's a different metric. You seem like a happy, vital person, and you're writing books and you're doing public education here. I'd like to actually stay on this meditation thing for a moment. Rick Rubin, famed music producer, came on this podcast. He also happens to be a friend of mine. And right as we sat down, he said, "Can we do some coherence breathing first?" He plays a tone. You do some inhale, exhale. He likes these transition tools, and breathing exercises are great. They will increase your heart rate variability and these kinds of things. But I think that one of the most interesting and understudied things in our field is task switching — whether it's emotional task switching or cognitive task switching. We assume the brain operates in step functions, like: I'm out on the street, I navigate traffic, I park my car, and I come in and I podcast. And what you're doing at the beginning of this lab meeting is on the one hand a meditation, but you're pivoting the brain circuits toward what comes next. I don't think we talk enough as neuroscientists about the fact that the brain doesn't just immediately switch over — that you're carrying residual activity from whatever you were doing before.

Adolphs: Yes. There's a well-known cost to that. As soon as you switch into something, there's a cost carried over from the previous task, and you're much slower and it's more difficult for the couple of next trials on the new task. So there's always a switching cost, which shows you that switching is hard. There's some residue, some sort of inertia being carried over from previously. There's some reconfiguration in the brain required between what you were doing in the previous task and the new task. But all of those would seem to require some effort, and it's definitely a skill. Something that people are looking at a lot in brain recordings in humans, and thinking a lot about how to design into AI — that ability to switch quickly, and knowing how to switch, anticipating when to switch, is a hugely important skill.

Huberman: Yeah. And we don't teach it. Here we are talking about emotional intelligence, social awareness, and these kinds of things. And I think most people would think, "Oh, we should sit down and look at a bunch of faces and learn to detect the subtle cues in faces, or we should make more eye contact." And what I'm hearing today is: while some of that might be true, most things are pointing in a different direction.

We had a guest on this podcast, DJ Shipley, who's a former tier one SEAL team operator. He talked about his routine for coming home. It's like he knows — he's rehearsing as he comes out of the car, just like as if it were an op. "I'm gonna hit the door. My daughter's gonna come running from this direction. My other daughter's gonna come in. My wife's gonna be in the kitchen. Dog's gonna come charging." And he's rehearsed it and he knows exactly where he'll set his bag so he can scoop his daughter up. People either chuckled or were taken aback by that, until he said, "But this is the way that I know I can drop everything from the day and be at my best when I enter that environment." And the comments when we ran this clip on social media — the men were all kind of in awe, like, "You have a system, dude, and you follow it." And the women were like, "Yes, fantastic. This is somebody who's prioritizing how he shows up. He's leaving everything else outside."

Adolphs: That skill to switch is super super important. It is trainable. It's largely trainable. And I think if you have meditative exercises, they train exactly that. It's a huge switch — you're switching from constantly being bombarded and dealing with stuff in the world to suddenly putting all of that aside and just being inside.

And it has health benefits. Richie Davidson, for instance —

Huberman: Yeah, we had him here. Phenomenal.

Adolphs: He studied Tibetan monks and long-term meditators who've meditated 10,000 hours or whatever, which we can't afford to do. But he has a lot of studies on even for the rest of us — short stretches of meditation, regular meditative exercises — he's done well-controlled studies with demonstrable health benefits for sure. So again, it comes back to this issue of variability. Being able to switch and having lots of variability and training that is definitely beneficial. It's beneficial for just basic physiology. It's beneficial for emotional granularity and emotion regulation. It's beneficial for control of thought — when you're being externally oriented versus meditating. All of those are to a large extent trainable skills in variability and control.

Huberman: So the experiment I'd love to see done — maybe your lab will do this — would be: what is the cost of task switching if you do or do not insert this five minutes of silence? Because you're spending five minutes, and we know the transition is choppy into the next thing that you're doing. So there's some overlap there. But by introducing the period of silence, you reduce the amount of time needed to switch your brain over to this new mode. Presumably you're seeing positive effects in these lab meetings.

Adolphs: Yeah, we should do a study. We would have to think about the right controls. We are looking at task switching actually in these brain recordings in epilepsy patients — we're looking at single unit recordings and how those representations change when you switch between tasks. It's a great study we're just working up the analyses on now. But we don't have the intervention that you were just speaking about. So we don't have half the patients getting meditative training and half not. But the extent to which an intervention like that can improve your ability to switch, or some other concrete metric — yes, those are exactly the types of studies that should be done. I think Richie is probably doing studies like that.

They did a study of five minutes of meditation per day and showed that people experienced tremendous increases in anxiety during the initial phase of the study — just having to sit there with your thoughts and let everything stir. I actually think of short meditations like that, at least initially, as another version of the ice bath.

Adolphs: Well, it is. And when we started — we have new summer students come in, new people in the lab — and I say, "Okay, we're going to just have three minutes of silence." And everybody's like this, and these summer students are looking around. "What's going on?" It's very stressful. And if you just tell them, "Well, just don't do anything" — this is obviously extremely difficult for young children — "Don't do anything. Just be completely silent and quiet for one minute or five minutes." It is hard to inhibit behavior. It's really hard to do, but it's trainable. It's completely trainable.

And a big part of it — we're in this same conference room that David Anderson has his lab meetings in, which is why you can always see it goes on for hours and hours. It has glass walls, and so people are walking around outside looking at us sitting there in silence, probably wondering not much progress is being done in that last meeting. So there's that whole social perception aspect. But again, once you get over that and get used to it, it's very easy to do these exercises, and it's definitely a trainable skill for sure.

Huberman: Yeah, we were on the East Coast in Annapolis at the Naval Academy. These are very accomplished students — almost 4,000 of them. And one of the most incredible things about that environment is everyone goes into the cafeteria for lunch and all of a sudden a bell sounds and you could hear a pin drop. But there are thousands of people around you. You hear a bell again. The meal begins. Walking into the auditorium, it's shuffle shuffle shuffle, talking talking talking. Boom. It's a step function. They're training this in themselves. And the other thing — no phones. No one has phones during the day. They get, I think, an hour where they can access their phones.

I normally — my phone is off, or I don't have a phone on me. And zero social media. Not on X, not on Facebook, not on Instagram. Zero. And whenever people ask me, they say, "Oh, can I hook you up with LinkedIn or whatever?" I'm not on LinkedIn. I don't care about LinkedIn. And they're always surprised and they say, "Well, why are you not on social media?" And I say, "Why would I be? I'm already above 100% in terms of people emailing me. I don't need more." Zero.

Huberman: Love that you've given yourself permission to do that. I put social media on a separate phone so that I can really track my time on there, and I just use it only to post, read a little bit, and then I'm out.

Adolphs: I just can't take the mental noise.

Illness, mortality, and the neuroscience of perspective

Huberman: Before we started recording, you mentioned that you fortunately successfully overcame a bout of illness. I'm curious whether the process of going through that — the diagnosis, the treatment, and then emerging from it, thank goodness, healthy — had you at times thinking about the science of emotions. That's an emotional experience to say the least. You don't have to share with us exactly what it was, but were there times when your knowledge of the brain's ability to process emotions, for better or worse, was a useful tool to you in that context?

Adolphs: I guess so. Only in the most indirect sense. I guess it's about two things. I think it's about emotional regulation. Initially when you get a diagnosis that reminds you of your mortality, it's sort of a mixture of fear, sadness, worry — all these negative emotions. And so you're regulating that, and it's interpersonal for sure. Interpersonal emotion regulation between me and my wife, us talking about it. That's a huge part of the process. If you're not regulating those negative emotions anymore, but you're buying into other emotions like awe, gratitude, acceptance — all of these emotions we were talking about before that I think are largely uniquely human — that really makes you appreciate and be just grateful for what you have. So that whole process of regulating the immediate stressors and then coming to having more gratitude and awe in everyday life.

And then also — I think one big consequence of anything like that is that you have this metacognitive perspective on your lifeline. You know where you are, how much time you have left, and how to value that and what to do and what to plan and how to allocate your time to things that are worthwhile. And so in my case, one of those components was signing up for this 100-mile ultramarathon, starting to train for it, and then just pushing through all of that. One of the consequences of the therapy for this was I had lots of joint pain and lots of fatigue. So I tell my doctor about that, and he says, "That's great. That means it's working." It's like, "Yes, but the side effects are preventing me from training for a 100-mile race." And so pushing through that — I think it just makes you grateful for what you have and makes you just be in awe of the fact that there's a world there at all.

You tend to have this perspective that physics gives us — of the way the world works, that the world is going on and when you die you get plucked out of that and everything else goes on. But from a neuroscience perspective, it's kind of the opposite. You get this sense that the whole world is actually a construction of stuff in your brain. And as you get older, little things drop out, various people die, you're unable to do things, and eventually you lose the whole world. So rather than thinking of death as we normally do — as the world loses you — the neuroscience perspective is the inverse: you lose the world. So you have a couple of decades left and then you're going to lose everything, and that's coming. So live it up.

Huberman: I love it. And thank you so much for being willing to answer that question. Throughout today's conversation, I'm struck by the fact that your work plays such a prominent role in your life and your life in your work, even if we weren't necessarily connecting the dots in that way. The fact that you do practices specifically to tap into your autonomic nervous system and emotion regulation, and at the same time the fact that you look at neural circuits and do extremely well-controlled studies to explore the mechanics that translate to everybody — this is, in case anyone hasn't told you, and because I sit at a unique vantage point having done academic research and now in public health communication — people need to hear this and see experienced scientists like you. So thank you so much for coming here today, for being willing to open up about some of this, and frankly for doing the work that you do for so many years and at such a high level. And I'm so grateful that you're writing yet another book.

Adolphs: Thank you very much, Andrew. It was a pleasure being here. We'll stay in touch, and maybe at some point I can tell you that I finished the 100-mile race and the book got published.

Huberman: Excellent. I have no doubt that you will. Whether I join you for that 100-mile race is a completely different story, but thank you so much, Ralph. Appreciate it.

Adolphs: Thank you.