Jarvis Links Human Speech to Birdsong Genetics and Prehistoric Hybridization

贾维斯将人类言语与鸟鸣遗传学及史前杂交关联 神经科学家证实，人类口语直接演化自古老的运动控制通路，而非源于独立的大脑模块，并且与具备发声学习能力的鸟类存在深度的遗传与神经趋同演化。 Erich Jarvis博士阐释了言语和语言的神经架构，对独立的“语言模块”观点提出质疑，转而支持特化的运动和感觉通路假说。对话重点探讨了人类与具备发声学习能力的鸣禽之间的趋同演化，指出了二者共有的遗传机制以及关键期学习限制。此外，Jarvis博士还详细说明了言语是如何从普遍的躯体运动回路演化而来的，并探讨了口吃的神经生物学机制、童年多语言经历的影响，以及舞蹈等身体活动如何帮助人在一生中保持认知敏锐。 科学界尚缺乏存在独立“语言模块”的证据，相反研究揭示言语依赖于特化的运动输出回路，而听觉感知则广泛分布于脊椎动物中。 (0:23) 言语生成通路很可能是直接从负责普遍身体运动和肢体手势的既有运动控制回路演化而来的。 (2:07) 反射性的情绪发声源自低位脑干区域，而发声学习则需要特化的前脑运动通路来控制脑干中枢。 (4:31) 尼安德特人在FOXP2等关键言语基因上拥有与人类完全相同的功能序列，这表明口语的起源可追溯至50万到100万年前。 (7:04) 鸣禽与人类表现出惊人的趋同演化，尽管相隔3亿年的演化历程，双方仍共享特化的基因表达模式和神经回路。 (8:17)

Dr. Erich Jarvis challenges the existence of a standalone linguistic brain module, arguing that speech production evolved directly from general motor control systems and relies on specialized forebrain circuits operating alongside broadly conserved auditory perception pathways. Motor Origins and Hominid Evolution of Speech (0:00) Dr. Jarvis disputes the classical model of an independent language module, distinguishing between widely distributed auditory perception pathways—which allow species such as dogs and apes to comprehend human words—and specialized motor pathways that govern physical vocal articulators. In Jarvis's model, speech production circuits evolved directly from pre-existing motor networks that control general body movements and manual gestures, which sit immediately adjacent in the cortex. Unlike innate brainstem vocalizations such as crying or grunting, learned vocal communication requires specialized forebrain circuits to override lower brainstem motor centers. Addressing the evolutionary timeline, Jarvis estimates that the neural substrates for human speech arose between 500,000 and 1,000,000 years ago. He points to comparative genomics showing that Neanderthals and Denisovans shared identical functional sequences with modern humans in key speech genes such as FOXP2. Furthermore, hominid interbreeding supports this timeline, as contemporary biology demonstrates that vocal-learning species do not successfully produce fertile offspring with non-vocal-learning species. Convergent Genetics in Avian and Human Pathways (8:17) Despite 300 million years of evolutionary divergence, humans and vocal-learning avian orders—songbirds, parrots, and hummingbirds—exhibit striking behavioral, anatomical, and genetic convergence. Specific avian brain structures like Area X and the robust nucleus of the arcopallium directly parallel human speech centers such as Broca's area and the laryngeal motor cortex. Genetically, establishing these pathways requires downregulating repulsive axon guidance molecules, which normally prevent connections, thereby allowing direct cortical projections to laryngeal and syrinx motor neurons. Because vocal articulators contract three to five times faster than locomotory muscles, vocal control neurons fire at unusually high frequencies. To prevent metabolic excitotoxicity, these circuits selectively upregulate protective calcium-buffering proteins such as parvalbumin and heat shock proteins. Striatal basal ganglia damage similarly produces stuttering across both birds and humans; however, while songbirds fully recover within months through robust adult striatal neurogenesis, humans lack equivalent adult neurogenesis and must manage stuttering through sensorimotor behavioral therapy. Critical Periods, Phonemic Repertoires, and Modalities (20:30) During developmental critical periods, the brain manages finite storage constraints by pruning neural connections, discarding unused inputs to protect essential survival memories. Jarvis clarifies that adults raised multilingually learn subsequent languages more easily not because of sustained structural plasticity, but because they retained a wider phonemic repertoire from childhood instead of pruning non-native sounds. Social bonding acts as a critical gatekeeper for this learning, as demonstrated by young finches selectively imitating tutors, and children in multicultural environments synthesizing disparate linguistic inputs into stable pidgin languages. Jarvis also separates vocal communication into affective (emotional expression, mating, territorial defense) and semantic (structured, abstract meaning) modes, proposing that speech originally evolved for singing and emotional connection before adapting to abstract thought. This functional division aligns with hemispheric lateralization: the left hemisphere exhibits dominance for structured speech production, while the right hemisphere engages more heavily with musical tones and emotional cadence. Sensorimotor Loops in Literacy and Physical Movement (27:28) Reading and writing operate via a coordinated four-circuit relay rather than an isolated cognitive center. During silent reading, signals travel from the visual cortex to motor speech circuits (Broca's area), generating measurable electromyographic micro-contractions in the laryngeal muscles as words are silently articulated. The brain then routes these motor signals to the auditory cortex to achieve internal hearing, before transmitting them to the adjacent hand motor cortex to execute writing or typing. Because speech regions remain closely integrated with general bodily motor control, Jarvis argues that complex physical movements, such as dance, stimulate and preserve cognitive function and fluent speech into older age. Regarding digital technology, he views text messaging not as an erosion of linguistic intellect, but as an adaptive shift toward rapid communication that dynamically reorganizes cortical real estate, such as expanding motor regions dedicated to fine thumb control. Jarvis's unified framework grounds speech, literacy, and musical expression in specialized motor adaptations rather than detached cognitive algorithms. A key therapeutic challenge remains finding ways to compensate for limited adult striatal neurogenesis in humans to better treat persistent motor speech disorders like stuttering.

洛克菲勒大学神经遗传学专家 Erich Jarvis 教授指出，人类大脑并不存在独立于运动与感知系统的抽象“语言模块”，语音生成能力本质上源自古老躯体运动控制回路的特化演变。人类与鸣禽等具备发声学习能力的物种在解剖结构、转录组学及感知-运动整合机制上的深度趋同，为这一运动起源假说提供了关键证据。 语言起源：语音回路源于躯体运动控制 (0:00) Jarvis 教授直接挑战了经典认知语言学关于大脑存在孤立语言模块的假说，提出人类大脑依赖专门的输入通路（听觉感知）与输出通路（语音生成）。其中听觉感知通路广泛存在于脊椎动物中，例如犬类能理解数百个单词，类人猿能听懂数千个词汇；然而，能直接协调喉部和颌骨肌群的语音生成通路却极罕见。在解剖层面上，控制语音发声的皮层脑区与支配手势动作的区域紧邻，两者共享复杂的动作序列算法，人类在日常交谈甚至电话沟通中无意识的手势伴随即反映了这一运动连续性。 从演化路径来看，野生非人灵长类的大脑前脑运动皮层能熟练控制肢体动作（如大猩猩 Koko 能习得手语并理解口语），却缺乏直接投射至脑干发声运动神经元的通路。人类、鸣禽及鹦鹉等少数物种演化出了由前脑特化回路接管低阶脑干发声核团的能力，将哭闹、犬吠或呼吸反射等本能情感声音，升级为受前脑运动皮层精密调控的“声音手势”。 趋同演化：鸣禽模型与发声分子遗传学 (8:17) 人类与鸣禽在演化历史上分化超过3亿年，但两者在发声学习回路中展现出惊人的行为与神经机制趋同。人类的布罗卡区（Broca's area）和喉部运动皮层，分别对应鸣禽的X区（Area X）和原脑弓粗壮核（RA）；破坏鸣禽体内的 FOXP2 基因会导致与人类言语障碍高度相似的发声缺陷。化石基因组学分析表明，尼安德特人与丹尼索瓦人拥有与现代人类完全一致的 FOXP2 序列，结合具备发声学习能力的物种无法与不具备该能力的物种产生可育后代的生物学规律，Jarvis 据此推断言语的神经基底可能早在50万至100万年前即已存在。 在微观分子层面，发声回路的形成依赖于特定轴突排斥引导基因的主动下调，从而解除了神经突触连接的分子阻碍，使前脑皮层得以直接投射至喉部（或鸟类鸣管）运动神经元。同时，由于喉部肌群收缩频率比一般行走肌群快3至5倍，发声神经元处于极高频放电状态；大脑通过显著上调小白蛋白（parvalbumin）及热休克蛋白等钙缓冲与神经保护分子，以防止神经元因代谢过载而发生严重的兴奋性神经毒性损伤。 关键期修剪：音位储备与多语言习得机制 (20:30) 针对大脑学习机制，Jarvis 将人脑比作存储容量有限的计算机，在发育关键期必须通过主动遗忘与回路固化来防止核心生存记忆被反复覆盖。人类婴儿出生时具备感知和发出全部语音音位（phonemes）的潜能，但在母语习得过程中会逐步修剪并舍弃非母语音位。儿童期接触多种语言能够保留更广泛的音位储备库，因此成年后学习外语相对轻松，是由于其早期保留了更宽广的声音工具包，而非因为成年大脑维持了更高的整体结构可塑性。 对于数字化通讯（如发短信、简写）是否退化人类语言智力的担忧，Jarvis 依据“用进废退”的可塑性机制指出，科技并未削弱深层语言认知能力，而是促使大脑重构通路以满足高频、快速的交流需求；同时，长期操作移动设备已显著扩大了运动皮层中专管拇指精细动作的神经表征区域。 神经回路整合：读写闭环与口吃病理机制 (27:28) 书写与阅读依赖四重脑回路的跨区域接力：视觉信号首先抵达视觉皮层，随即前向传递至布罗卡区进行隐性发声（肌电图显示此时喉部肌肉会出现微小的隐性收缩）；该信号随后传至听觉皮层形成内部听觉表征，最终映射至相邻的手部运动皮层驱动书写。这种感觉-运动整合在脑半球上存在功能不对称：左半球主导结构化、抽象的语义通信（semantic communication），而右半球则对旋律、语调及情感通信（affective communication）展现出更强的调控偏好。 在发声运动失调方面，Jarvis 实验室在损伤鸣禽负责动作序列协调的基底核纹状体亚区后，观察到了类似人类神经源性口吃的发声表型。鸣禽能够通过成年期纹状体神经发生（adult neurogenesis）整合新神经元并在数月内完全自愈，而成年人类因缺乏这一级别的神经再生能力，口吃往往长期持续。临床上，成年患者需要依赖强化感觉-运动整合的言语疗法，通过有意识地比对听觉反馈与发声动作来恢复言语流畅度。 躯体运动连接：舞蹈与发声对认知的维系 (32:42) 基于大脑语言皮层与躯体运动皮层在解剖上的直接毗邻性，Jarvis 结合自身的舞者经历与神经科学研究提出，复杂的全身性身体活动与高级认知能力紧密耦合。舞蹈、跑步等需要空间定位与躯体协调的复杂运动，能够广泛激活并滋养邻近的神经组织。 Jarvis 主张，将日常肢体运动与主动发声练习（如公开演讲、歌唱）相结合，能够持续刺激控制面部肌群与运动协调的前脑网络。这种针对运动-发声通路的生理与行为锻炼，是维持中枢神经健康、延缓衰老过程中认知与言语功能衰退的重要机制。 Erich Jarvis 教授的研究确立了发声学习建立在广泛的躯体运动回路演化特化与跨脑区感知-运动闭环之上，颠覆了将语言孤立于感知和动作之外的传统假说。然而，人类为何在失去成年纹状体大规模神经再生能力的同时演化出高度复杂的抽象语义能力，以及如何通过行为干预最大化激活神经保护通路，仍有待未来比较基因组学与神经生物学的深入解答。

[20:30] Dr. Erich Jarvis: The brain can only hold so much information, and 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. [28:53] Dr. Erich Jarvis: Bird brain undergoes new neurogenesis in a way that human or mammal brains don't... and it was the new neurons that were coming in into the circuit... that was resulting in this stuttering.

[0:00] Welcome to Huberman Lab Essentials, [0:02] where we revisit past episodes for the [0:04] most potent and actionable science-based [0:06] tools for mental health, physical [0:08] health, and performance. [0:11] I'm Andrew Huberman and I'm a professor [0:13] of neurobiology and ophthalmology at [0:15] Stanford School of Medicine. And now, [0:18] for my discussion with Dr. Eric Jarvis. [0:21] Eric, so great to have you here. [0:22] >> Thank you. [0:23] >> Yeah. Very interested in learning from [0:25] you about speech and language. In terms [0:28] of the study of speech and language and [0:30] thinking about how the brain organizes [0:32] speech and language, uh what are the [0:35] similarities, what are the differences, [0:37] how should we think about speech and [0:38] language? [0:38] >> There really isn't such a sharp [0:41] distinction. Now, let me tell you how [0:43] some people think of it now. That [0:45] there's a separate language module in [0:47] the brain that has all the algorithms [0:49] and computations that influence the [0:52] speech pathway on how to produce sound [0:55] and the auditory pathway on how to [0:57] perceive and interpret it uh for speech [1:00] or for, you know, sound that we call [1:03] speech. I don't think there is any good [1:05] evidence for a separate language module. [1:07] Instead, [1:09] there is a speech production pathway [1:12] that's controlling our larynx, [1:13] controlling our jaw muscles, that has [1:16] built within it all the complex [1:18] algorithms for spoken language. And [1:21] there's the auditory pathway that has [1:24] built within it all the complex [1:25] algorithms for understanding speech, not [1:28] separate from a language module. [1:30] And the speech production pathway is [1:32] specialized to humans [1:34] and parrots and songbirds, whereas this [1:37] auditory perception pathway is more [1:40] ubiquitous amongst the animal kingdom. [1:42] And this is why dogs can understand sit, [1:45] sientese, come here ball boy, get the [1:48] ball, and so forth. Dogs can understand [1:51] several hundred human speech words. [1:53] Great apes, you can teach them for [1:54] several thousand, but they can't say a [1:57] word. [1:57] >> What do we understand about modes of [2:00] communication that are [2:02] like language, but might not be what [2:04] would classically be called language? [2:07] >> So, next to the brain regions that are [2:09] controlling spoken language are the [2:11] brain regions for gesturing with the [2:13] hands. [2:14] And that hand parallel pathway has also [2:17] complex algorithms that we can utilize. [2:20] And some species are more advanced in [2:22] these circuits, whether it's sound or [2:24] gesturing with hands, and some are less [2:27] advanced. [2:28] Humans are the most advanced at spoken [2:30] language, [2:32] but not necessarily as big a difference [2:34] at gestural language compared to some [2:36] other species. So, as you and I are [2:39] talking here today, and people who are [2:41] listening but can't see us, we're [2:42] actually gesturing with our hands as we [2:45] talk uh without knowing it, or doing it [2:48] unconsciously. And if we were talking on [2:51] a telephone, I would have one hand here [2:53] and I'd be gesturing with the other hand [2:55] uh without even you seeing me, right? [2:58] And so, why is that? Uh some have [3:00] argued, and I would agree with based [3:01] upon what we've seen, is that there's an [3:04] evolutionary relationship between the [3:06] brain pathways that control speech [3:08] production and gesturing. [3:11] Uh and and the brain regions I mentioned [3:13] are directly adjacent to each other. And [3:14] why is that? I think that the brain [3:16] pathways that control speech evolved out [3:19] of the brain pathways that control body [3:21] movement. [3:22] All right? And um [3:25] that uh [3:27] when you talk about Italian, French, [3:29] English, and so forth, [3:31] um each one of those languages come with [3:33] a learned set of gestures [3:36] that uh you can communicate with. Now, [3:38] how is that related to other animals? [3:40] Well, Koko, a gorilla who was raised [3:42] with humans for 39 years or more, [3:46] uh learned how to do gesture [3:48] communication. Learned how to sign [3:50] language, so to speak, right? But Koko [3:53] couldn't produce those sounds. Koko [3:55] could understand them as well [3:58] by sign by seeing somebody sign or [4:01] hearing somebody produce speech, but [4:03] Koko couldn't produce it with her voice. [4:06] And so what's going on there is that [4:09] a number of species, not all of them, a [4:11] number of species have motor pathways in [4:13] the brain where you can do learned [4:15] gesturing, rudimentary language if you [4:18] wanted to say with your limbs, even if [4:20] it's not as advanced as humans, but they [4:22] don't have this extra brain pathway for [4:24] the sound. So they can't gesture with [4:27] their voice in the way that they gesture [4:29] with their hands. [4:30] >> One thing that I've wondered about for a [4:32] very long time is whether or not [4:35] primitive emotions [4:38] and primitive sounds are the early [4:41] substrate of language. When I smell [4:43] something delicious, I typically inhale [4:47] more and I might say [4:50] or something like that. Whereas if I [4:52] smell something putrid, I typically turn [4:54] away, I wince, and I will exhale trying [4:57] to not ingest those molecules or inhale [4:59] those molecules. I could imagine that [5:02] these are the basic dark and light [5:04] contrasts of the language system. Is [5:07] this kind of primitive to more [5:08] sophisticated [5:11] pyramid of of sound to language? Is this [5:14] a crazy idea? Do we have any [5:16] Do we have any evidence this is the way [5:18] it works? [5:18] >> No, it's not a crazy idea and in fact [5:21] you hit upon one of the key distinctions [5:24] in the field of research that I had [5:26] started out in, which is vocal learning [5:27] research. Most vertebrate species [5:29] vocalize, but most of them are producing [5:32] innate sounds that they're born with, [5:35] that is babies crying, for example, or [5:38] dogs barking. And only a few species [5:41] have learned vocal communication, the [5:43] ability to imitate sounds. And that's is [5:46] what makes spoken language special. When [5:48] people think of what's special about [5:50] language, it's the learned [5:52] vocalizations. That is what's rare. So, [5:55] all the things you talked about, the [5:56] breathing, the grunting, and so forth, a [5:58] lot of that is handled by the brainstem [6:00] circuits, you know, right around the [6:02] level of your neck and below. Uh [6:05] like a reflex kind of thing. So, or or [6:09] even some emotional aspects of your [6:10] behavior in the hypothalamus and so [6:12] forth. [6:13] But, for a learned behavior, learning [6:15] how to speak, uh learning how to play [6:18] the piano, teaching a dog to learn how [6:20] to do tricks, [6:22] is using the forebrain circuits. [6:24] And what has happened is that there's a [6:27] lot of forebrain circuits that are [6:28] controlling learning how to move body [6:30] parts in these species, but not for the [6:31] vocalizations. But, in humans and in [6:34] parrots and some other species, somehow [6:37] we acquired circuits where the forebrain [6:40] has taken over the brainstem, [6:42] and now using that brainstem not only to [6:45] produce the innate behaviors or vocal [6:47] behaviors, but the learned ones as well. [6:49] >> Do we have any sense of when [6:52] modern or sophisticated language [6:55] evolved? [6:56] >> Amongst the primates, which we humans [6:57] belong to, we are the only ones that [6:59] have this advanced vocal learning [7:01] ability. [7:02] Uh [7:03] Now, [7:05] when you [7:06] It was assumed that it was only Homo [7:07] sapiens. [7:08] Uh then you can go back in time now [7:11] based upon genomic data [7:13] not only of us living humans, but of the [7:15] fossils that have been found for Homo [7:17] sapiens, of Neanderthals, of Denisovan [7:21] uh individuals, and discover that our [7:25] ancestor our human ancestors supposedly [7:29] hybridized with these other [7:31] hominid species. [7:33] And it was assumed that these other [7:35] hominid species don't learn how to [7:37] imitate sounds. [7:39] I don't know of any species today that's [7:41] a vocal learner [7:43] that can have children with a non-vocal [7:44] learning species. [7:46] I I don't see it. It doesn't mean it [7:47] didn't exist. [7:49] Uh and when we look at the genetic data [7:51] from these ancestral hominids that uh [7:55] you know, where we can look at genes [7:57] that are involved in learn vocal [7:59] communication, they have the same [8:01] sequence as we humans do [8:04] for genes that function in speech [8:05] circuits. [8:06] So, I think Neanderthals had spoken [8:08] language. I'm not going to say it's as [8:10] advanced as what it is in humans, I [8:11] don't know. [8:13] Um but I think it's been there for at [8:14] least between 500,000 to a million [8:17] years. [8:17] >> Maybe we could talk a little bit more [8:19] about the overlap between brain circuits [8:21] that control language and speech in [8:23] humans and other animals. You know, I [8:26] was weaned in the neuroscience era where [8:28] bird song and the uh [8:31] the ability of birds to learn their [8:33] tutor song was and still is a prominent [8:35] field and um subfield of neuroscience. [8:38] And this notion of a critical period, a [8:39] time in which language is learned more [8:42] easily than it is later in life. And [8:45] the names of the different brain areas [8:47] were quite different. Um it when one [8:49] opens the textbooks, we hear Wernicke's [8:51] and Broca's for the humans and then you [8:53] look at the birds, so that I remember, [8:55] you know, [8:56] >> HVC [8:56] >> Yeah, a robust archistriatum, area X, [8:59] right? [8:59] >> That's right. Yes. [9:00] >> how similar or different are the brains [9:02] uh brain areas controlling speech and [9:04] language in say a songbird and a and a [9:06] young human child? [9:08] >> Yeah. So, going back to the 1950s or and [9:11] even a little earlier and Peter Marler [9:13] and others who got involved in [9:15] neuroethology, the study of neurobiology [9:19] of behavior in a natural way, right? Um [9:22] you know, they started to find that [9:25] behaviorally [9:26] there are these species of birds like [9:27] songbirds and parrots and now we also [9:29] know hummingbirds, just three of them [9:31] out of the 40-something bird groups out [9:33] there on the planet, orders, [9:35] that they can imitate sounds like we [9:38] And so that was a similarity. In other [9:40] words, they had this kind of behavior [9:41] that's more similar to us than [9:43] chimpanzees have with us or than [9:45] chickens have with them, right? They're [9:47] more closer relatives. And then they [9:49] discovered even more similarities, these [9:51] critical periods that if you remove a [9:54] child you know, this unfortunately [9:56] happens where a child is feral and is [9:59] not raised with human and goes through [10:01] their puberty phase of growth [10:03] becomes hard for them learn a language [10:05] as an adult. So there's this critical [10:07] period where you learn best. And even [10:09] later on when you're in regular society, [10:11] it's hard to learn. Well, the same birds [10:14] undergo these same thing. And then it [10:16] was discovered that if they become deaf, [10:19] we humans become deaf, our speech starts [10:22] to deteriorate without any kind of [10:23] therapy. [10:25] Uh if a non-human primate or [10:28] um you know, or let's say a chicken [10:30] becomes deaf, uh their vocalizations [10:33] don't deteriorate, very little at least. [10:35] Uh well, this happens in the vocal [10:37] learning birds. So there were all these [10:39] behavioral parallels that it came along [10:41] with a package. And then people looked [10:43] into the brain, Fernando Nottebohm, my [10:45] former PhD advisor, and began to [10:47] discover the area X you talked about, uh [10:50] the robust nucleus of the arcopallium. [10:53] And um [10:55] and these brain pathways were not found [10:57] in the species who couldn't imitate. So [10:58] there was a parallel here. [11:00] And then uh jumping many years later, [11:03] you know, I started to dig down into [11:05] these uh brain circuits to discover that [11:08] these brain circuits had parallel [11:09] functions with the brain circuits for [11:11] humans, even though they're by a [11:13] different name like Broca's and [11:15] laryngeal motor cortex. And most [11:17] recently, we discovered not only the [11:19] actual circuitry and the connectivity [11:21] are similar, but the underlying genes [11:23] that are expressed in these brain [11:25] regions in a specialized way different [11:27] from the rest of the brain are also [11:28] similar between humans and songbirds and [11:31] parrots. So all the way down to the [11:33] genes and now we're finding the specific [11:35] mutations [11:36] are also similar. Not always identical, [11:38] but similar. [11:39] Uh which indicates remarkable [11:41] convergence for a so-called complex [11:44] behavior in species separated by 300 [11:46] million years from a common ancestor. [11:48] And not only that, we are discovering [11:50] that [11:51] mutations in these genes that cause [11:54] speech deficits in humans like in FOXP2, [11:58] uh if you put those same mutations or [12:00] similar type of deficits in these vocal [12:03] learning birds, you get similar [12:04] deficits. So, convergence of the [12:06] behavior is associated with similar [12:09] genetic disorders of the behavior. [12:11] >> Do hummingbirds sing or do they hum? [12:14] >> Hummingbirds hum with their wings and [12:16] sing with their syrinx. [12:17] >> In a coordinated way? [12:19] >> In a coordinated way. There is some [12:21] species of hummingbirds um [12:23] that actually will [12:25] um Doug Archler showed this that will [12:27] flap uh their wings and create a [12:30] slapping sound with their wings that's [12:33] in unison with their song. And oh and [12:36] you would not know it, but it sounds [12:37] like a particular syllable in their [12:39] songs uh even though it's their wings [12:43] and their voice at the same time. [12:45] >> Hummingbirds are clapping to their song. [12:47] >> Clapping with their they're snapping [12:49] their wings together uh in unison with a [12:52] song to to make it like if I'm going ba [12:54] da da da da ba da. [12:57] You know, I banged on the table. Except [12:59] they make it almost sound like their [13:00] voice with their wings. What's amazing [13:02] about hummingbirds and I we're going to [13:04] say vocal learning species in general is [13:07] that for whatever reason, they seem to [13:09] evolve multiple complex traits. [13:12] You know, this idea that the evolving [13:14] language, spoken language in particular, [13:17] comes along with a set of [13:19] specializations. [13:20] >> When I was coming up in neuroscience, I [13:24] learned that I think it was the work of [13:26] Peter Marler that um young birds learn [13:29] songbirds learn their tutor's song [13:32] and learn it quite quite well, but that [13:36] they could learn the song of another [13:38] tutor. In other words, they could learn [13:39] a different and for the listeners I'm [13:41] doing air quotes here, a different [13:42] language, a different bird song, [13:44] different than their own species song, [13:46] but never as well as they could learn [13:48] their own natural [13:51] genetically linked song. [13:53] >> Yes. [13:54] >> Genetically linked meaning that it would [13:55] be like me being raised in a different [13:57] culture and um [14:00] that I would learn that the other [14:01] language, but not as well as I would [14:03] have learned English. This this is the [14:05] idea. [14:06] >> Yes. [14:06] >> Is that true? [14:07] >> That is true. Yes, and that's and that's [14:09] what I learned growing up as well and [14:10] and and talked to Peter Marler himself [14:12] about before he passed. Um yeah, this he [14:15] used to call it the innate [14:16] predisposition to learn. All right, so [14:20] um [14:21] which would be kind of the equivalent in [14:23] the linguistic community of universal [14:24] grammar. There is something genetically [14:28] influencing [14:30] our vocal communication on top of what [14:33] we learn culturally. And so there is [14:35] this balance between the genetic control [14:38] of speech or a song in these birds and [14:41] the learned uh cultural control. And so [14:44] so yes, if you were to take um you know, [14:48] um I mean in this case we we actually [14:50] tried this at Rockefeller later on, take [14:52] a zebra finch and raise it with a [14:54] canary, it would sing a song that was [14:57] sort of like a hybrid in between. We [14:59] call it a caninch. Right? [15:02] Uh and vice versa for the canary because [15:04] there's something different about their [15:06] vocal musculature or the or the [15:08] circuitry in the brain. And with a zebra [15:10] finch, even with a closely related [15:12] species, if you would take a zebra finch [15:15] uh young animal and in one cage next to [15:17] it place its own species adult male, [15:20] right? And in the other cage place a [15:22] Bengali finch next to it, it would [15:25] preferably learn the song from the its [15:28] own species neighbor. But if you remove [15:30] its neighbor, it would learn that [15:32] Bengali finch very well. [15:34] >> Fantastic. [15:34] >> So, there's it it has something to do [15:36] with also the social bonding with your [15:38] own species. [15:39] >> That raises a question that I've based [15:41] on something I also heard, but I don't [15:43] have any scientific peer-reviewed [15:45] publication to point to, which is this [15:47] this idea of pigeon, not the bird, but [15:49] this idea of when multiple cultures and [15:53] languages converge in a given geographic [15:55] area that the children of all the [15:56] different native languages will come up [15:58] with their own language. [16:01] I think this was in island culture, [16:03] maybe in Hawaii, called pigeon, which is [16:05] sort of a hybrid of the various [16:07] languages that their parents speak at [16:08] home and that they themselves speak, and [16:11] that somehow pigeon, again, not the [16:14] bird, but a language called pigeon for [16:16] reasons I don't know, [16:18] harbors certain basic elements of all [16:21] language. Mhm. Is that true? Is that not [16:24] true? [16:24] >> What is going on here is [16:27] cultural evolution remarkably tracks [16:30] genetic evolution. So, if you bring [16:34] people from two separate populations [16:35] together that have been in in their [16:37] separate populations evolutionarily at [16:39] least for hundreds of generations. So, [16:42] someone speaking Chinese, someone [16:44] speaking English, [16:45] uh and that child uh then's learning [16:47] from both of them. [16:49] Yes, that child's going to be able to [16:50] pick up and merge uh uh uh phonemes and [16:56] words together in a way that an adult [16:59] wouldn't because why? They're [17:01] experiencing both languages at the same [17:03] time during their critical period uh [17:06] years in a way that um adults would not [17:09] be able to experience. And so, you get a [17:11] hybrid. [17:12] And the lowest common denominator is [17:14] going to be what they share. And so, the [17:17] phonemes that they retained in each of [17:19] their [17:21] languages is what's going to be, I [17:23] imagine, used the most. [17:24] >> So we've got [17:26] brain circuits in songbirds and in [17:29] humans that in many ways are similar, [17:31] perhaps not in their exact wiring, but [17:33] in their basic contour of wiring, and [17:36] genes that are expressed in both sets of [17:38] neural circuits in very distinct species [17:40] that are responsible for these phenomena [17:42] we're calling speech and language. I [17:44] mean, what are What are these genes [17:45] doing? [17:46] >> Uh one of the things that differ in the [17:48] speech pathways of us and these song [17:49] pathways of birds is some of the [17:51] connections are fundamentally different [17:53] than the surrounding circuits. Like a [17:56] a direct cortical connection [17:58] from the areas that control [17:59] vocalizations in the cortex to the motor [18:01] neurons that control the larynx in [18:04] humans or the syrinx in birds. And so we [18:06] actually made a prediction [18:08] that since some of these connections [18:10] differ, we're going to find genes that [18:12] that control neuroconnectivity [18:14] and that specialize in that differ. And [18:17] that's exactly what we found. Uh [18:20] genes that control what we call axon [18:22] guidance and formation of connections. [18:24] And what was interesting, it was sort of [18:26] in the opposite direction that we [18:27] expected. That is [18:30] some of these genes, actually a number [18:32] of them that control neuroconnectivity [18:34] were turned off [18:35] in the speech circuit. [18:37] All right. Uh and it didn't make sense [18:39] to us at first until we started to [18:41] realize the function of these genes are [18:43] to repel connections from forming. So [18:45] repulsive molecules. And so when you [18:48] turn them off, they allow certain [18:50] connections to form that normally would [18:52] have not formed. So it's So by turning [18:54] it off, you got a gain of function for [18:56] speech, right? Um other genes that [18:59] surprised us were genes involved in [19:02] calcium buffering, neuroprotection. [19:05] Like a parvalbumin or heat shock [19:07] protein. So when your brain gets hot, [19:09] these proteins turn on. And we couldn't [19:11] figure out for a long time, why is that [19:12] the case? [19:14] And then the idea popped to me one day. [19:16] I said, "Ah, when I heard the larynx is [19:19] the fastest firing muscles in the body." [19:22] All right? In order to vibrate sound and [19:25] and modulate sound in the way we do, you [19:28] have to control you have to move those [19:30] muscles, you know, three to four to five [19:33] times faster than just regular walking [19:35] or running. [19:36] And so, um [19:38] when you stick electrodes in in the [19:40] brain areas that control learn [19:41] vocalizations in these birds, and I [19:43] think in humans as well, uh those [19:45] neurons are firing at a higher rate to [19:48] control these muscles. [19:50] And so, what is that going to do? You're [19:52] going to have lots of toxicity in those [19:54] neurons unless you upregulate molecules [19:56] that [19:57] take out [19:58] uh the extra load that is needed to [20:00] control the larynx. And then finally, a [20:02] third set of genes that are uh [20:04] specialized in these speech circuit are [20:06] involved in neuroplasticity. [20:08] Uh neuroplasticity meaning allowing the [20:11] brain circuits to be more flexible, uh [20:15] so you can learn better. And why is [20:17] that? I think learning how to produce [20:19] speech is a more complex learning [20:23] ability than say learning how to walk [20:25] or or learning how to do tricks and [20:28] jumps and so forth that dogs do. [20:30] >> In terms of plasticity of speech and the [20:33] ability to learn multiple languages, but [20:35] even just one language, what's going on [20:37] in the so-called critical period? And [20:38] then the second question is, if one can [20:40] already speak more than one language as [20:44] a consequence of childhood learning, is [20:46] it easier to acquire new languages later [20:48] on? [20:49] >> Actually, the entire brain [20:51] uh is undergoing a critical period [20:53] development, not just the speech [20:55] pathways. And uh so, it's easier to [20:58] learn how to play a piano. It's easier [20:59] to learn how to ride a bike for the [21:01] first time and so forth as a young child [21:04] than it is later in life. The brain can [21:07] only hold so much information, and if [21:10] you are undergoing rapid learning to [21:13] learn to acquire new knowledge, you also [21:15] have to put memory or information in in [21:17] the trash, like in a computer. You You [21:19] only have so many gigabases of memory. [21:22] Plus also for survival, you don't want [21:24] to keep forgetting things. [21:26] And so So, the brain is designed, I [21:29] believe, [21:30] to undergo this critical period and [21:33] solidify the circuits with what you [21:35] learned as a child, and you use that for [21:37] the rest of your life. And now, the [21:39] question you asked about if you learned [21:41] more languages as a child, can you Is it [21:44] easier to learn as an adult? And that's [21:46] a common uh finding out there in the [21:48] literature. There's some that argue [21:49] against it, but for those that support [21:51] it, the idea there is [21:54] um you you are born with a set of innate [21:56] sounds you can produce of phonemes, [21:59] and you narrow that down, because not [22:00] all languages use all of them. And so, [22:03] you narrow down the ones you use to [22:05] string the phonemes together in words [22:07] that you learn, and you maintain those [22:10] phonemes as an adult. And here comes [22:12] along another language that's using [22:14] those phonemes or in the in different [22:16] combinations you're not used to, uh and [22:18] therefore, you it's like starting from [22:20] first principles. But if you already [22:22] have them in multiple languages that [22:25] you're using, then it makes it easier to [22:27] use them in another third or fourth [22:28] language. So, it's not like your brain [22:30] has under has maintained greater [22:32] plasticity, is your your brain has [22:35] maintained greater ability to produce [22:37] different sounds that then allows you to [22:40] learn another language faster. [22:41] >> What about modes of speech and language [22:43] that seem to have a depth of [22:45] emotionality and meaning, but for which [22:47] it departs from structured language? I [22:51] think of musicians, like there's some [22:53] Bob Dylan songs that to me [22:55] I understand the individual words. [22:58] I like to think there's an emotion [23:00] associated with it, at least I [23:01] experience some sort of emotion and I [23:03] have a guess about what he was [23:05] experiencing. But if I were to just read [23:07] it linearly without the music and [23:10] without him singing it or somebody [23:12] singing it like him, it wouldn't hold [23:14] any meaning. So in other words, [23:16] uh words that seem to have meaning but [23:18] not associated with language but somehow [23:21] tap into an emotionality. [23:23] >> Absolutely. So So we call this [23:26] difference um semantic communication, [23:29] communication with meaning, and [23:31] effective communication, [23:33] that has more of an emotional [23:36] feeling content to it. I believe, you [23:38] know, based upon imaging work and work [23:41] we see in birds, when when birds are [23:44] communicating semantic information in [23:46] their sounds, which is not too often, [23:47] but it happens, versus uh effective [23:50] communication, sing because I'm trying [23:52] to attract the mate, my courtship song, [23:54] or defend my territory. It's the same [23:57] brain circuits. It's the same [23:58] speech-like or song circuits are being [24:00] used in different ways. There's several [24:03] other points here I think it's important [24:04] for for the those listening out there to [24:06] hear is that when I say also this [24:09] effective and um semantic communication [24:13] um being used by similar brain circuits, [24:16] it also matters the side of the brain. [24:18] Uh in birds and in humans, [24:21] um there's there's left-right dominance [24:24] uh for learned uh communication, learned [24:26] sound communication. Uh so the left in [24:29] us humans is more dominant for speech. [24:32] But the right has a more balance for [24:35] singing or processing musical sounds as [24:38] opposed to processing speech. Both get [24:40] used for both reasons. And so when [24:43] people say your right brain is your [24:45] artistic brain and your left brain is [24:47] your thinking brain, this is what [24:49] they're referring to. Uh and uh so [24:52] that's another distinction. The second [24:54] uh uh thing that's useful to know [24:57] is that all vocal learning species use [24:59] their learned sounds for this emotional [25:02] effective kind of communication. [25:05] But only a few of them, like humans and [25:07] some parrots and dolphins, use it for [25:10] the semantic kind of communication we [25:12] calling speech. [25:13] And and that has led a number of people [25:16] to hypothesize that the evolution of [25:19] spoken language of speech evolved first [25:22] for singing. Uh for this more like [25:26] emotional kind of mate attraction, like [25:28] the Jennifer Lopez, the Ricky Martin [25:30] kind of songs and so forth. Uh and then [25:33] later on it became used for abstract [25:36] communication like we're doing now. [25:38] >> I'd love to chat a moment about facial [25:40] expression, many of which are [25:42] subconscious. We are all familiar with [25:44] the fact that when what somebody says [25:47] doesn't match some specific feature of [25:49] their facial expression, that it can um [25:52] call, you know, that mismatch can cue [25:54] our attention. [25:55] >> Yeah. [25:56] >> So, how does motor circuitry that [25:58] controls facial expression map onto the [26:01] the brain circuits that control [26:03] language, speech, and even bodily and [26:05] hand movements? [26:05] >> Yeah. And you ask a great question [26:07] because we both know some colleagues [26:09] like Win McFall at Rockefeller [26:11] University who study facial expression [26:13] and the neurology behind it. Non-human [26:15] primates have a lot of diversity in [26:17] their facial expression like we humans [26:19] do. [26:20] And what we know about the neurobiology [26:23] of brain regions controlling those [26:25] muscles of the face [26:27] is that these non-human primates and [26:28] some other species that don't learn how [26:30] to imitate vocalizations, they have [26:34] strong connections from the cortical [26:36] regions to the motor neurons that [26:38] control facial expressions. And even [26:40] though it's more diverse in these [26:41] non-human primates, there was already a [26:43] pre-existing diversity of communication, [26:46] whether it's intentional or unconscious, [26:49] through facial expression in our land [26:52] And on top of that, we humans now add [26:55] the voice [26:57] uh along with those facial expressions. [26:59] So, it's like an email, too. You're [27:00] You're emailing, and someone says [27:02] something by email, someone can [27:04] interpret that angrily or or gently, [27:07] uh [27:08] and it it bec- becomes ambiguous. The [27:10] facial expressions get rid of that [27:12] ambiguity. [27:13] >> I'm so glad you brought that out because [27:15] my next question was and is about [27:17] written language. What is the process of [27:19] going from a thought to language to [27:23] written word? And what's going on there? [27:26] What do we know about the neural [27:27] circuitry? [27:28] >> What I think is going on is, to explain [27:30] what you're asking, is about [27:33] that I'm going to take it from the [27:34] perspective reading something. You read [27:36] something on a paper, [27:38] the signal from the paper goes through [27:39] your eyes, it goes to the back of your [27:41] brain to your visual cortical regions, [27:44] eventually. That visual signal then goes [27:47] to your speech pathway in the motor [27:49] cortex in front of you in Broca's area, [27:51] and you silently speak what you read in [27:54] your brain without moving your muscles. [27:57] And sometimes, actually, if you put [27:59] electrodes EMG electrodes on your [28:01] laryngeal muscles, even on birds you can [28:03] do this, you'll see activity there while [28:06] you're reading or or or trying to speak [28:09] silently, even though no sound's coming [28:11] out. [28:12] And so, your [28:14] speech pathway is now speaking what [28:17] you're reading. [28:19] Now, to finish it off, that signal is [28:22] sent to your auditory pathway so you can [28:24] hear what you're speaking in your own [28:25] head. [28:26] >> That's incredible. [28:27] >> And this is why it's complicated. Oh, [28:29] and then you got to write, right? Okay, [28:32] here comes the fourth one. Now, the hand [28:34] area is next to your speech pathway is [28:36] got to take that auditory signal or even [28:38] the adjacent motor signals for speaking [28:41] and translate it into a visual signal on [28:43] paper. [28:44] So, so you're using at least four brain [28:46] circuits, [28:47] um which includes the speech production [28:50] and the speech perception pathways to [28:52] write. [28:53] >> Stutter is a um particularly interesting [28:57] case. What is the current [28:58] neurobiological understanding of stutter [29:01] and are uh what's being developed in [29:03] terms of treatments for stutter? [29:05] >> Yes, so we actually uh [29:07] accidentally came across stuttering in [29:09] songbirds. And we've uh published [29:12] several papers on this. So, try to [29:14] figure out the neurobiological basis. [29:15] The first study we had was a brain area [29:19] uh called the basal ganglia or this [29:20] what's the the striatum part of the [29:22] basal ganglia involved in coordinating [29:25] movements, learning how to make [29:26] movements. [29:27] When it was damaged in these in this in [29:30] the speech-like pathway in these birds, [29:32] what we found is that they started to [29:34] stutter [29:36] as the brain region recovered. [29:38] And unlike humans, they actually [29:42] recovered after three or four months. [29:44] And why is that the case? Because bird [29:46] brains undergoes new neurogenesis in a [29:49] way that human or mammal brains don't. [29:52] Uh and it was the new neurons that were [29:54] coming in into the circuit, uh but not [29:57] quite, you know, with the right proper [29:59] activity [30:01] uh was resulting in this stuttering in [30:03] these birds. [30:05] Uh and after it was repaired, not [30:06] exactly the old song came back as a [30:09] after repair, but still it recovered a [30:11] lot better. [30:13] And it's now known they call this [30:15] neurogen- neurogenics uh stuttering in [30:18] humans. Uh with damage to the basal [30:20] ganglia or some type of disruption to [30:23] the basal ganglia at a young age also [30:25] causes stuttering in humans. And even [30:28] those who are born with stuttering, uh [30:31] um [30:31] it's it's often the basal ganglia uh [30:34] that's disrupted than some other brain [30:36] circuit. And we think the speech part of [30:38] the basal ganglia. [30:39] >> Can adults who maintain a stutter from [30:42] childhood uh repair that stutter? [30:44] >> There are ways to overcome the [30:47] stuttering through um [30:49] through uh you know, behavioral therapy. [30:52] Uh and I think all of the uh tools out [30:55] there [30:57] have something to do with sensory motor [30:59] integration. [31:00] Uh controlling what you hear with what [31:03] you output in a of thoughtful, [31:05] controlled way helps reduce the [31:08] stuttering. [31:09] >> Texting [31:10] is a very, very interesting evolution of [31:14] language. I wonder sometimes whether or [31:16] not we are getting less proficient at [31:19] speech because we are not required to [31:23] write and think in complete sentences. [31:26] >> Mhm. [31:27] >> What do you think's happening to [31:28] language? Are we getting better at [31:30] speaking, worse at speaking, and what do [31:32] you think the role of things like [31:34] texting and tweeting and shorthand [31:36] communication, hashtagging? [31:38] Uh what's that doing to the way that our [31:40] brains work? [31:41] >> Uh texting [31:43] actually has allowed for more rapid [31:45] communication amongst people. It's more [31:48] like a use it or lose it kind of a uh uh [31:52] uh thing with the brain. The more you [31:54] use a particular brain region or [31:56] circuit, the more enhanced. It's like a [31:58] muscle. [31:59] Uh the more you exercise it, the more [32:01] healthier it is, the bigger it becomes, [32:03] and the more space it takes, and the [32:05] more you you lose something else. [32:07] So, I think texting is not decreasing [32:11] the speech powers, or the intellectual [32:13] powers of speech. It's converting it and [32:17] using it a lot in a different way. In a [32:19] way that may not be as rich in in [32:22] regular writing because uh [32:25] you can only communicate so much nuance [32:27] in short term writing. But, um whatever [32:31] that it whatever is being done, you got [32:33] people texting hours and hours and hours [32:36] on the phone. So, whatever your thumb [32:38] circuit is going to get pretty big, [32:40] actually. [32:42] >> For those listening who are interested [32:44] in getting better at speaking and [32:46] understanding languages, are there any [32:49] tools that you recommend? Should kids [32:51] learn how to read hard books and simple [32:54] books? What do you recommend? Should [32:55] adults learn how to do that? Everyone [32:57] wants to know how to keep their brain [32:59] working better, so to speak, but also I [33:01] think people want to be able to speak [33:03] well, and people want to be able to [33:04] understand well. [33:05] >> Yeah. What I've discovered personally, [33:08] right, is that So, when I switched from [33:11] pursuing a career in science from a [33:14] career in dance, I thought one day I [33:16] would stop dancing. Um but I haven't [33:19] because it I find it fulfilling for me. [33:22] And there've been periods of time like [33:24] during the pandemic where I slowed down [33:26] on dancing and so forth. Um And And when [33:29] you do that, you realize, okay, there [33:31] there parts of your body where your [33:32] muscle tone decreases a little bit in [33:34] some way, and or you could start to gain [33:36] weight, or I somehow don't gain weight [33:38] that easily, and I think it's related to [33:39] my dance, if that's that that's [33:41] meaningful to your audience. But what I [33:43] found is in science, we like to think of [33:45] a separation between movement and action [33:48] and cognition. [33:50] And there is a separation between [33:51] between perception and production. [33:53] Cognition being perception, production [33:55] being movement, right? But if the speech [33:59] pathway is next to the movement [34:00] pathways, what I discover is by dancing, [34:04] it is helping me [34:06] think. It is helping keeping my brain [34:08] fresh. It's not just moving my muscles. [34:11] I'm moving the or using the the [34:13] circuitry in my brain to do control a [34:16] whole big body. [34:18] You need a lot of brain tissue to do [34:19] that. And so, I argue, if you want to [34:22] stay [34:23] cognitively [34:24] intact into your old age, you better be [34:27] moving. [34:28] And you better be doing it consistently, [34:30] whatever it's dancing, walking, running, [34:33] and also practicing speech, [34:36] oratory speech and so forth, or singing, [34:38] is controlling the brain circuits that [34:40] are moving your facial musculature, and [34:42] it's going to keep your cognitive [34:44] circuits also in tune. And I'm I'm [34:47] convinced of that from my own personal [34:48] experience. [34:49] >> This has been an incredible conversation [34:51] and opportunity for me to learn. And I'm [34:54] I know I speak for a tremendous number [34:55] of people when I I just really want to [34:57] say thank you for joining us today. You [34:59] are incredibly busy. It's clear from [35:01] your description of your science and [35:03] your knowledge base that you are [35:04] involved in a huge number of things. [35:06] Um very busy. So, thank you for taking [35:08] the time to speak to all of us. [35:10] Thank you for the work that you're [35:12] doing. [35:12] >> Thank you for inviting me here to get [35:14] the word out to the community [35:16] uh [35:17] of what's going on in the science world. [35:19] >> Well, we're honored and very grateful to [35:20] you, Eric. Thank you. Love you.