Andrew Huberman interviews neuroscientist Dr. Erich Jarvis on the biology of speech, language, and music
Andrew Huberman speaks with Dr. Erich Jarvis, a neuroscientist at Rockefeller University, about the neuroscience of speech, language, and music.
Summary
Dr. Erich Jarvis, a neuroscientist at Rockefeller University, joins the Huberman Lab podcast to discuss the neuroscience of speech, language, and music. He argues that there is no separate "language module" in the brain — instead, spoken language is built into the speech production pathway itself, a system shared in parallel form by humans, parrots, songbirds, and hummingbirds, but absent in most other species including non-human primates. He presents evidence that the genes, neural circuits, and even genetic mutations underlying speech disorders are strikingly convergent across species separated by 300 million years of evolution. The conversation also covers the neurobiology of stuttering, the critical period for language acquisition, the relationship between movement and cognition, and what texting and modern communication habits may — or may not — be doing to the human brain.
Key Takeaways
FULL TRANSCRIPT
What Speech and Language Are — and Why the Distinction Matters
Andrew Huberman: I'm very interested in learning from Dr Dr Erich Jarvis about speech and language — the study of speech and language, how the brain organizes it, what the similarities and differences are, and how we should think about speech and language.
Dr Dr Erich Jarvis: There really isn't such a sharp distinction. Now let me tell you how some people think of it. There's a view that there is a separate language module in the brain that has all the algorithms and computations that influence the speech pathway on how to produce sound, and the auditory pathway on how to perceive and interpret it for speech — for sound that we call speech. I don't think there is any good evidence for a separate language module. Instead, there is a speech production pathway that's controlling our larynx, controlling our jaw muscles, that has built within it all the complex algorithms for spoken language. And there's the auditory pathway that has built within it all the complex algorithms for understanding speech — not separate from a language module.
This speech production pathway is specialized to humans and parrots and songbirds, whereas the auditory perception pathway is more ubiquitous amongst the animal kingdom. This is why dogs can understand "sit," "good boy," "get the ball," and so forth. Dogs can understand several hundred human speech words. Great apes you can teach several thousand words, but they can't say a word.
Andrew Huberman: What do we understand about modes of communication that are like language but might not be what would classically be called language?
Dr Dr Erich Jarvis: Next to the brain regions that are controlling spoken language are the brain regions for gesturing with the hands. That hand parallel pathway also has complex algorithms that we can utilize. Some species are more advanced in these circuits, whether it's sound or gesturing with hands, and some are less advanced. Humans are the most advanced at spoken language, but not necessarily as big a difference at gestural language compared to some other species.
As you and I are talking here today — and for people who are listening but can't see us — we're actually gesturing with our hands as we talk without knowing it. We're doing it unconsciously. If we were talking on a telephone, I would have one hand here and I would be gesturing with the other hand without you even seeing me. Why is that? Some have argued, and I would agree based upon what we've seen, that there is an evolutionary relationship between the brain pathways that control speech production and gesturing. The brain regions I mentioned are directly adjacent to each other. I think the brain pathways that control speech evolved out of the brain pathways that control body movement.
When you talk about Italian, French, English, and so forth, each one of those languages comes with a learned set of gestures that you can communicate with. Koko, a gorilla who was raised with humans for 39 years or more, learned how to do gesture communication — learned how to sign, so to speak. But Koko couldn't produce those sounds with her voice. Koko could understand them by seeing somebody sign or hearing somebody produce speech, but couldn't produce it vocally. What's going on there is that a number of species — not all of them — have motor pathways in the brain where you can do learned gesturing, rudimentary language if you wanted, even if it's not as advanced as humans, but they don't have this extra brain pathway for sound. So they can't gesture with their voice in the way that they gesture with their hands.
Innate Versus Learned Vocalizations
Andrew Huberman: One thing I've wondered about for a very long time is whether primitive emotions and primitive sounds are the early substrate of language. When I smell something delicious, I typically inhale more and I might make a sound. Whereas if I smell something putrid, I turn away, I wince, and I exhale, trying not to inhale those molecules. I could imagine that these are the basic dark and light contrasts of the language system — a kind of primitive-to-sophisticated pyramid of sound to language. Is this a crazy idea? Do we have any evidence this is the way it works?
Dr Dr Erich Jarvis: No, it's not a crazy idea. In fact, you hit upon one of the key distinctions in the field of research that I started out in, which is vocal learning research. Most vertebrate species vocalize, but most of them are producing innate sounds that they're born with — babies crying, for example, or dogs barking. Only a few species have learned vocal communication, the ability to imitate sounds. That is what makes spoken language special. When people think of what's special about language, it's the learned vocalizations. That is what's rare.
All the things you talked about — the breathing, the grunting — a lot of that is handled by the brain stem circuits, right around the level of your neck and below, like a reflex. Even some emotional aspects of behavior are handled in the hypothalamus and so forth. But for a learned behavior — learning how to speak, learning how to play the piano, teaching a dog to do tricks — that uses the forebrain circuits. What has happened is that there are a lot of forebrain circuits controlling how to move body parts in these species, but not for the vocalizations. In humans and in parrots and some other species, somehow we acquired circuits where the forebrain has taken over the brain stem, and now uses that brain stem not only to produce innate vocal behaviors but the learned ones as well.
When Did Spoken Language Evolve?
Andrew Huberman: Do we have any sense of when modern or sophisticated language evolved?
Dr Dr Erich Jarvis: Amongst the primates, which we humans belong to, we are the only ones that have this advanced vocal learning ability. It was assumed that it was only Homo sapiens. But you can go back in time now based upon genomic data — not only of living humans but of fossils that have been found for Homo sapiens, Neanderthals, and Denisovan individuals — and discover that our human ancestors supposedly hybridized with these other hominid species. It was assumed that these other hominid species don't learn how to imitate sounds. I don't know of any species today that's a vocal learner that can have children with a non-vocal learning species. It doesn't mean it didn't exist.
When we look at the genetic data from these ancestral hominids and examine genes that are involved in learned vocal communication, they have the same sequence as we humans do for genes that function in speech circuits. So I think Neanderthals had spoken language. I'm not going to say it was as advanced as what it is in humans — I don't know. But I think it's been there for at least between 500,000 to a million years.
Parallel Brain Circuits in Songbirds and Humans
Andrew Huberman: Maybe we could talk a little more about the overlap between brain circuits that control language and speech in humans and other animals. I was trained in the neuroscience era where birdsong and the ability of birds to learn their tutor song was a prominent subfield of neuroscience — this notion of a critical period, a time in which language is learned more easily than later in life. The names of the different brain areas were quite different. In the human literature we hear Wernicke's and Broca's, and in the bird literature I remember HVC, the robust nucleus of the arcopallium, Area X. How similar or different are the brain areas controlling speech and language in, say, a songbird and a young human child?
Dr Dr Erich Jarvis: Going back to the 1950s and even a little earlier, Peter Marler and others who got involved in neuroethology — the study of neurobiology of behavior in a natural way — began to find that behaviorally there are species of birds like songbirds and parrots, and now we also know hummingbirds — just three of the forty-some bird groups out there — that can imitate sounds like we do. That was the similarity. They had a kind of behavior more similar to us than chimpanzees have with us, or than chickens have with them, even though chickens are closer relatives.
Then they discovered even more similarities — these critical periods. If a child is feral and not raised with humans and goes through their puberty phase of growth, it becomes hard for them to learn a language as an adult. There's this critical period where you learn best, and even later when you're in regular society it's hard to learn. Well, birds undergo the same thing. And then it was discovered that if humans become deaf, our speech starts to deteriorate without any kind of therapy. If a non-human primate or a chicken becomes deaf, their vocalizations deteriorate very little. But this deterioration does happen in the vocal learning birds. So there were all these behavioral parallels that came along as a package.
Then people looked into the brain. Fernando Nottebohm, my former PhD adviser, began to discover Area X and the robust nucleus of the arcopallium, and these brain pathways were not found in species who couldn't imitate. Then, jumping many years later, I started to dig down into these brain circuits and discovered that they have parallel functions with the brain circuits for humans, even though they go by different names like Broca's and the laryngeal motor cortex. Most recently we discovered not only that the actual circuitry and connectivity are similar, but the underlying genes expressed in these brain regions in a specialized way — different from the rest of the brain — are also similar between humans and songbirds and parrots. All the way down to the genes. And now we're finding that the specific mutations are also similar — not always identical, but similar — which indicates remarkable convergence for a so-called complex behavior in species separated by 300 million years from a common ancestor.
Not only that, we are discovering that mutations in these genes that cause speech deficits in humans — like in FOXP2 — if you put those same mutations or similar types of deficits in these vocal learning birds, you get similar deficits. So convergence of the behavior is associated with similar genetic disorders of the behavior.
Andrew Huberman: Do hummingbirds sing or do they hum?
Dr Dr Erich Jarvis: Hummingbirds hum with their wings and sing with their syrinx in a coordinated way. There are some species of hummingbirds — Doug Altshuler showed this — that will flap their wings and create a slapping sound with their wings that's in unison with their song. You would not know it, but it sounds like a particular syllable in their songs, even though it's their wings and their voice at the same time.
Andrew Huberman: Hummingbirds are clapping to their song.
Dr Dr Erich Jarvis: Snapping their wings together in unison with a song to make it — if I'm going "da da da da da da" and I bang on the table — except they make it almost sound like their voice with their wings. What's amazing about hummingbirds, and vocal learning species in general, is that for whatever reason they seem to evolve multiple complex traits together. There's this idea that evolving spoken language in particular comes along with a set of specializations.
The Innate Predisposition to Learn and Cultural Hybrids
Andrew Huberman: When I was coming up in neuroscience, I learned — I think it was from the work of Peter Marler — that young songbirds learn their tutor song quite well, but that they could learn the song of another tutor, a different bird song different from their own species song, but never as well as they could learn their own natural, genetically linked song. This would be like me being raised in a different culture and learning that other language, but not as well as I would have learned English. Is that true?
Dr Dr Erich Jarvis: That is true. And that's what I learned growing up as well, and discussed with Peter Marler himself before he passed. He used to call it the innate predisposition to learn — which would be the equivalent in the linguistic community of universal grammar. There is something genetically influencing our vocal communication on top of what we learn culturally. There's this balance between the genetic control of speech or song in these birds and the learned cultural control.
We actually tried this at Rockefeller later on. Take a zebra finch and raise it with a canary — it would sing a song that was sort of a hybrid in between. We call it a "caninch." And vice versa for the canary, because there's something different about their vocal musculature or the circuitry in the brain. With a zebra finch, even with a closely related species, if you took a young zebra finch and placed its own species adult male in one cage next to it and a Bengal finch in another cage next to it, it would preferably learn the song from its own species neighbor. But if you removed its neighbor, it would learn the Bengal finch song very well. So it also has something to do with social bonding with your own species.
Andrew Huberman: That raises a question based on something I've heard but don't have a peer-reviewed publication to point to — this idea of pidgin. Not the bird, but the idea that when multiple cultures and languages converge in a given geographic area, the children of all the different native language speakers will come up with their own language. I think this was in island culture, maybe in Hawaii — called pidgin — which is sort of a hybrid of the various languages their parents speak at home. And somehow pidgin harbors certain basic elements of all language. Is that true?
Dr Dr Erich Jarvis: What is going on is that cultural evolution remarkably tracks genetic evolution. So if you bring people from two separate populations together — someone speaking Chinese, someone speaking English — and a child is learning from both of them, that child is going to be able to pick up and merge phonemes and words together in a way that an adult wouldn't. Why? They're experiencing both languages at the same time during their critical period years in a way that adults would not be able to experience. So you get a hybrid, and the lowest common denominator is going to be what they share. The phonemes that have been retained in each of their languages are what's going to be used the most.
What Genes Are Doing in Speech Circuits
Andrew Huberman: We've got brain circuits in songbirds and in humans that in many ways are similar — perhaps not in their exact wiring but in their basic contour of wiring — and genes that are expressed in both sets of neural circuits in very distinct species that are responsible for these phenomena we're calling speech and language. What are these genes doing?
Dr Dr Erich Jarvis: One of the things that differ in the speech pathways of us and the song pathways of birds is that some of the connections are fundamentally different from the surrounding circuits — like a direct cortical connection from the areas that control vocalizations in the cortex to the motor neurons that control the larynx in humans or the syrinx in birds. We made a prediction that since some of these connections differ, we're going to find genes that control neural connectivity and specialize in that function that also differ. And that's exactly what we found — genes that control what we call axon guidance and form connections.
What was interesting is that it was sort of in the opposite direction from what we expected. A number of these genes that control neural connectivity were actually turned off in the speech circuit. It didn't make sense to us at first until we started to realize that the function of these genes is to repel connections from forming. They are repulsive molecules. When you turn them off, you allow certain connections to form that normally would not have formed. So by turning them off, you got a gain of function for speech.
Other genes that surprised us were genes involved in calcium buffering and neuroprotection — like parvalbumin or heat shock proteins. When your brain gets hot, these proteins turn on. We couldn't figure out for a long time why that would be the case in speech circuits. Then the idea came to me one day: the larynx has the fastest-firing muscles in the body. In order to vibrate and modulate sound in the way we do, you have to move those muscles three to four to five times faster than regular walking or running. When you stick electrodes into the brain areas that control learned vocalizations in these birds — and I think in humans as well — those neurons are firing at a higher rate to control these muscles. What does that do? You're going to have a lot of toxicity in those neurons unless you upregulate molecules that take out the extra load needed to control the larynx.
And then a third set of genes specialized in these speech circuits are involved in neuroplasticity — allowing the brain circuits to be more flexible so you can learn better. I think learning how to produce speech is a more complex learning ability than, say, learning how to walk or learning how to do tricks and jumps.
The Critical Period and Multilingualism
Andrew Huberman: In terms of plasticity of speech and the ability to learn multiple languages — or even just one language — what's going on in the so-called critical period? And then, if one can already speak more than one language as a consequence of childhood learning, is it easier to acquire new languages later on?
Dr Dr Erich Jarvis: The entire brain is undergoing critical period development, not just the speech pathways. It's easier to learn how to play a piano, easier to learn how to ride a bike for the first time as a young child than later in life. The brain can only hold so much information. If you are undergoing rapid learning to acquire new knowledge, you also have to put memory or information in the trash — like in a computer, you only have so many gigabytes of memory. Plus, for survival, you don't want to keep forgetting things. So the brain is designed, I believe, to undergo this critical period and solidify the circuits with what you learned as a child, and you use that for the rest of your life.
Now, the question about whether learning more languages as a child makes it easier to learn as an adult — that is a common finding in the literature. There are some that argue against it, but for those that support it, the idea is this: you are born with a set of innate sounds you can produce — phonemes — and you narrow that down because not all languages use all of them. You narrow down the ones you use to string phonemes together into words, and you maintain those phonemes as an adult. When along comes another language using those phonemes or in combinations you're not used to, it's like starting from first principles. But if you already have them in multiple languages, it makes it easier to use them in a third or fourth language. So it's not that your brain has maintained greater plasticity — it's that your brain has maintained a greater ability to produce different sounds, which then allows you to learn another language faster.
Music, Emotion, and the Two Sides of the Brain
Andrew Huberman: What about modes of speech and language that seem to have a depth of emotionality and meaning but depart from structured language? I think of musicians — there are some Bob Dylan songs that I understand the individual words of, and I experience some sort of emotion, and I have a guess about what he was experiencing. But if I were to just read the lyrics linearly without the music and without him singing it, it wouldn't hold the same meaning. Words that seem to have meaning but not associated with language in the traditional sense, but somehow tap into an emotionality.
Dr Dr Erich Jarvis: We call this difference semantic communication — communication with meaning — and affective communication — communication that has more of an emotional feeling content to it. I believe, based upon imaging work and work we see in birds, that when birds are communicating semantic information in their sounds versus affective communication — singing because I'm trying to attract a mate, my courtship song, or defend my territory — it's the same brain circuits, the same speech-like or song circuits, being used in different ways.
There are several other points important for those listening. When I say affective and semantic communication are being used by similar brain circuits, it also matters which side of the brain. In birds and in humans there's left-right dominance for learned sound communication. The left in us humans is more dominant for speech, but the right has more balance for singing or processing musical sounds as opposed to processing speech. Both get used for both reasons. When people say your right brain is your artistic brain and your left brain is your thinking brain, this is what they're referring to.
The second thing useful to know is that all vocal learning species use their learned sounds for this emotional, affective kind of communication, but only a few of them — like humans and some parrots and dolphins — use it for the semantic kind of communication we call speech. That has led a number of people to hypothesize that the evolution of spoken language evolved first for singing — for this more emotional, mate-attraction kind of communication — and then later on it became used for abstract communication like we're doing now.
Facial Expression and Its Relationship to Speech
Andrew Huberman: I'd love to chat a moment about facial expression, many of which are subconscious. We're all familiar with the fact that when what somebody says doesn't match some specific feature of their facial expression, that mismatch can cue our attention. How does the motor circuitry that controls facial expression map on to the brain circuits that control language, speech, and even bodily and hand movement?
Dr Dr Erich Jarvis: We both know colleagues like Winrich Freiwald at Rockefeller University who study facial expression and the neurobiology behind it. Non-human primates have a lot of diversity in their facial expression like we humans do. What we know about the neurobiology of brain regions controlling those muscles of the face is that non-human primates and some other species that don't learn how to imitate vocalizations already have strong connections from the cortical regions to the motor neurons that control facial expressions. Even though it's more diverse in these non-human primates, there was already a pre-existing diversity of communication — whether intentional or unconscious — through facial expression in our ancestors. On top of that, we humans now add the voice along with those facial expressions. It's like email — someone can interpret an email angrily or gently and it becomes ambiguous. The facial expressions get rid of that ambiguity.
Written Language and the Four Brain Circuits It Requires
Andrew Huberman: I'm so glad you brought that out, because my next question is about written language. What is the process of going from a thought to language to written word, and what's going on there? What do we know about the neural circuitry?
Dr Dr Erich Jarvis: What I think is going on — and I'll take it from the perspective of reading something — is this: you read something on paper. The signal from the paper goes through your eyes to the back of your brain, to your visual cortical regions. That visual signal then goes to your speech pathway in the motor cortex, in Broca's area. And you silently speak what you read in your brain without moving your muscles. Sometimes, if you put EMG electrodes on your laryngeal muscles — even in birds you can do this — you'll see activity there while reading or trying to speak silently, even though no sound is coming out. So your speech pathway is now speaking what you're reading. Then that signal is sent to your auditory pathway so you can hear what you're speaking in your own head.
And then you've got to write. Now the hand areas next to your speech pathway have to take that auditory signal, or even the adjacent motor signals for speaking, and translate it into a visual signal on paper. So you're using at least four brain circuits — which includes the speech production and the speech perception pathways — just to write.
Stuttering: A Neurobiological Perspective
Andrew Huberman: Stuttering is a particularly interesting case. What is the current neurobiological understanding of stuttering, and what's being developed in terms of treatments?
Dr Dr Erich Jarvis: We actually accidentally came across stuttering in songbirds and have published several papers on this to try to figure out the neurobiological basis. The first study we had involved a brain area called the basal ganglia — specifically the striatum part of the basal ganglia, which is involved in coordinating movements and learning how to make movements. When it was damaged in the speech-like pathway in these birds, what we found is that they started to stutter as the brain region recovered. Unlike humans, they actually recovered after three or four months. Why? Because bird brains undergo neurogenesis in a way that human or mammal brains don't. It was the new neurons coming into the circuit — but not quite with the right proper activity — that was resulting in this stuttering. After repair, not exactly the old song came back, but it recovered a lot better.
It's now known — they call this neurogenic stuttering in humans — that damage to the basal ganglia or some type of disruption to the basal ganglia at a young age also causes stuttering in humans. Even those who are born with stuttering often have disruption in the basal ganglia rather than some other brain circuit — specifically the speech part of the basal ganglia.
Andrew Huberman: Can adults who maintain a stutter from childhood repair that stutter?
Dr Dr Erich Jarvis: There are ways to overcome stuttering through behavioral therapy. I think all of the tools out there have something to do with sensory-motor integration — controlling what you hear with what you output in a thoughtful, controlled way helps reduce the stuttering.
What Texting and Modern Communication Are Doing to the Brain
Andrew Huberman: Texting is a very interesting evolution of language. I wonder sometimes whether we are getting less proficient at speech because we are not required to write and think in complete sentences. What do you think is happening to language? Are we getting better at speaking, worse at speaking? And what do you think the role of things like texting, tweeting, shorthand communication, and hashtagging is doing to the way our brains work?
Dr Dr Erich Jarvis: Texting has actually allowed for more rapid communication amongst people. It's more like a use-it-or-lose-it thing with the brain. The more you use a particular brain region or circuit, the more enhanced it becomes — like a muscle. The more you exercise it, the healthier it is, the bigger it becomes, and the more space it takes, and the more you lose something else. So I think texting is not decreasing the speech prowess or the intellectual prowess of speech. It's converting it and using it a lot in a different way — in a way that may not be as rich as regular writing, because you can only communicate so much nuance in short-form writing. But whatever is being done, you've got people texting hours and hours on the phone. Whatever your thumb circuit is, it's going to get pretty big.
Movement, Dance, and Keeping the Brain Sharp
Andrew Huberman: For those who are interested in getting better at speaking and understanding languages — are there any tools you recommend? Should kids learn how to read hard books and simple books? Should adults do that? Everyone wants to know how to keep their brain working better, but also people want to be able to speak well and understand well.
Dr Dr Erich Jarvis: What I've discovered personally is that when I switched from pursuing a career in dance to a career in science, I thought one day I would stop dancing. But I haven't, because I find it fulfilling. There have been periods of time — like during the pandemic — where I slowed down on dancing. When you do that, you realize there are parts of your body where your muscle tone decreases somewhat, or you could start to gain weight. I somehow don't gain weight that easily, and I think it's related to my dancing.
In science we like to think of a separation between movement and action and cognition. There is a separation between perception and production — cognition being perception, production being movement. But if the speech pathway is next to the movement pathway, what I discover is that by dancing, it is helping me think. It is keeping my brain fresh. It's not just moving my muscles — I'm using the circuitry in my brain to control a whole big body. You need a lot of brain tissue to do that. So I argue: if you want to stay cognitively intact into your old age, you better be moving, and you better be doing it consistently — whether it's dancing, walking, or running. And also practicing oratory speech or singing is controlling the brain circuits that are moving your facial musculature, and it's going to keep your cognitive circuits also in tune. I'm convinced of that from my own personal experience.
Andrew Huberman: This has been an incredible conversation and an opportunity for me to learn. I know I speak for a tremendous number of people. I really want to say thank you for joining us today. It's clear from your description of your science and your knowledge base that you are involved in a huge number of things. Thank you for taking the time to speak to all of us, and thank you for the work that you're doing.
Dr Dr Erich Jarvis: Thank you for inviting me here to get the word out to the community about what's going on in the science world.
Andrew Huberman: We're honored and very grateful to you, Erich. Thank you.
Dr Dr Erich Jarvis: You're welcome.