Essentials: The Neuroscience of Speech, Language & Music | Dr. Erich Jarvis
In a Nutshell
Speech and language lack a dedicated brain module; instead, specialized human speech production pathways (adjacent to gesture and movement controls) evolved from ancient circuits shared with vocal-learning species like songbirds, parrots, and hummingbirds, enabling sound imitation via convergent genes (e.g., FOXP2) and forebrain override of brainstem reflexes. Critical periods govern learning, with multilingual childhood exposure preserving phoneme flexibility; music conveys emotion via right-brain circuits, while semantic speech is left-dominant, and stuttering often stems from basal ganglia disruptions treatable by sensory-motor therapy. Consistent movement (dancing, walking) and speech practice (oratory, singing) maintain adjacent circuits, preventing cognitive decline and enhancing language proficiency.
These notes were generated by AI and may contain inaccuracies.
Andrew Huberman introduces Huberman Lab Essentials, revisiting past episodes for science-based tools on mental health, physical health, and performance. He is a professor of neurobiology and ophthalmology at Stanford School of Medicine. Discussion with Dr. Erich Jarvis on neuroscience of speech, language, and music.
No sharp distinction between speech and language. Some propose a separate language module in the brain with algorithms influencing speech production and auditory perception, but no good evidence supports this. Instead, speech production pathway controls larynx and jaw muscles, containing complex algorithms for spoken language. Auditory pathway contains algorithms for understanding speech. Speech production pathway specialized to humans, parrots, and songbirds. Auditory perception pathway more ubiquitous in animal kingdom—dogs understand several hundred human speech words; great apes can learn several thousand but cannot vocalize them.
Brain regions controlling spoken language adjacent to those for hand gesturing, with complex algorithms. Humans most advanced at spoken language, but gestural language not as differentiated from other species. People gesture unconsciously with hands while talking, even on telephone. Evolutionary relationship between speech production and gesturing pathways; speech pathways evolved from body movement control pathways. Languages like Italian, French, English include learned gesture sets. Koko the gorilla, raised with humans for 39+ years, learned sign language but could not produce speech sounds, though understood them.
Primitive emotions and sounds (e.g., inhaling "mmm" for delicious smells, wincing/exhaling for putrid) may form basic substrate of language, from primitive to sophisticated pyramid. Not a crazy idea; key distinction in vocal learning research. Most vertebrates produce innate sounds (babies crying, dogs barking). Only few species have learned vocal communication—ability to imitate sounds, making spoken language special. Innate vocalizations handled by brainstem circuits (reflex-like) and hypothalamus for emotions. Learned behaviors (speaking, playing piano, dog tricks) use forebrain circuits. Vocal learners acquired forebrain control over brainstem for both innate and learned vocalizations.
Humans only primates with advanced vocal learning. Initially assumed only Homo sapiens; genomic data from fossils (Homo sapiens, Neanderthals, Denisovans) shows hybridization. Ancestral hominids had same gene sequences as humans for speech circuits, suggesting Neanderthals had spoken language (possibly less advanced), present for 500,000 to 1 million years.
Birdsong learning prominent in neuroscience. Critical period for language acquisition easier in youth. Human areas: Wernicke's, Broca's. Bird areas: HVC, robust nucleus of arcopallium (RA), area X. Similarities since 1950s (Peter Marler, neuroethology): songbirds, parrots, hummingbirds (3 of 40+ bird orders) imitate sounds like humans, closer to us than chimpanzees or chickens. Behavioral parallels: critical periods (feral children struggle post-puberty); speech deteriorates if deaf in humans and vocal-learning birds, but not in non-vocal learners like chickens or non-human primates. Fernando Nottebohm discovered area X, RA absent in non-imitators. Circuits have parallel functions with human Broca's, laryngeal motor cortex. Recent findings: similar connectivity, genes expressed uniquely in these regions, specific mutations between humans, songbirds, parrots—convergence despite 300 million years separation. Mutations like FOXP2 causing human speech deficits produce similar deficits in vocal-learning birds.
Hummingbirds hum with wings, sing with syrinx in coordinated way. Some species flap wings for slapping sound in unison with song, mimicking syllables (Doug Altshuler's work).
Songbirds learn tutor song best, but can learn other species' songs less well—like learning different human language. Peter Marler called it innate predisposition to learn, akin to universal grammar—genetic influence atop cultural learning. Balance between genetic and cultural control. Zebra finch raised with canary produces hybrid "caninch" song due to vocal musculature/circuitry differences. Zebra finch prefers own species song but learns Bengal finch song well if isolated.
Cultural evolution tracks genetic evolution. Children of parents from separate populations (e.g., Chinese-English speakers) merge phonemes/words into hybrid during critical period, unlike adults. Pidgin (e.g., in island cultures like Hawaii) hybrids multiple languages, retaining shared phonemes as lowest common denominator.
Similar brain circuits in songbirds/humans for speech/language: genes control axon guidance and connectivity. Prediction: genes differing for specialized connections (e.g., direct cortical to motor neurons for larynx/syrinx). Many connectivity genes turned off in speech circuits—repulsive molecules; turning off allows atypical connections (gain of function). Other genes: calcium buffering, neuroprotection (parvalbumin, heat shock proteins) for high-speed larynx firing (3-5x faster than walking/running muscles; fastest in body). Neurons in control areas fire at higher rates, requiring upregulated protection. Third set: neuroplasticity genes for circuit flexibility, as speech learning more complex than walking or dog tricks.
Entire brain in critical period during development—easier to learn piano, bike, etc., as child. Brain solidifies childhood circuits for life, discarding unused info like computer memory for survival/efficiency. Multilingual childhood learning eases adult language acquisition—not greater plasticity, but retained broader phoneme repertoire (languages narrow innate phonemes; multilingual maintain more for new combinations).
Music/lyrics (e.g., Bob Dylan) convey emotion beyond structured language. Semantic communication (meaning) vs. affective communication (emotional). Same brain circuits used differently in birds/humans. Left brain dominance for learned speech (humans); right more for singing/music. All vocal learners use learned sounds affectively; few (humans, parrots, dolphins) for semantic speech. Hypothesis: speech evolved first for singing (mate attraction, territory; e.g., J.Lo/Ricky Martin), later for abstract communication.
Facial expressions often subconscious; mismatch with speech cues attention. Non-human primates have diverse facial expressions with strong cortical connections to facial motor neurons. Species without vocal imitation have these strong connections, even though diverse.
There was already pre-existing diversity of communication, whether intentional or unconscious, through facial expressions in our ancestors. Humans add the voice along with those facial expressions. It's like an email: someone can interpret an email angrily or gently, creating ambiguity, but facial expressions eliminate that ambiguity.
The process from thought to language to written word involves reading as a model. Visual signals from paper go through the eyes to the back of the brain's visual cortical regions, then to the speech pathway in the motor cortex, specifically Broca's area. There, you silently speak what you read without moving muscles. EMG electrodes on laryngeal muscles—even in birds—detect activity during silent reading or speaking, even without sound. The speech pathway signal then goes to the auditory pathway so you hear what you're speaking in your head.
Writing adds a fourth circuit: hand areas next to the speech pathway take the auditory or adjacent motor signals for speaking and translate them into visual signals on paper. This uses at least four brain circuits, including speech production and speech perception pathways.
Stuttering was accidentally discovered in songbirds. Damage to the striatum part of the basal ganglia—a brain area involved in coordinating movements and learning movements—in the speech-like pathway caused birds to stutter as the region recovered. Unlike humans, birds recover after three or four months due to neurogenesis, where new neurons integrate into the circuit but initially lack proper activity, causing stuttering. The song improves but does not fully return to the original.
This is called neurogenic stuttering in humans. Damage to or disruption of the basal ganglia, especially the speech part, at a young age causes stuttering. Even congenital stuttering often involves basal ganglia disruption.
Adults with lifelong stuttering can overcome it through behavioral therapy focused on sensory-motor integration. Controlling what you hear with what you output in a thoughtful, controlled way reduces stuttering.
Texting enables more rapid communication. It's a 'use it or lose it' principle: the more you use a brain region or circuit, the more it enhances, like a muscle—the healthier, bigger it becomes, potentially at the expense of other areas. Texting does not decrease speech or intellectual prowess; it converts and uses it differently. Short-form writing limits nuance compared to regular writing, but extensive texting enlarges the thumb circuit.
To improve speaking and understanding languages, and keep the brain working better:
- Practice movement consistently: dancing, walking, running.
- Practice speech, oratory, or singing to control facial musculature circuits, keeping cognitive circuits in tune.
Dancing helps thinking and keeps the brain fresh by using circuitry next to speech pathways to control the body. Movement and cognition are linked; perception (cognition) and production (movement) overlap, especially since speech pathways are adjacent to movement pathways. Consistent physical activity prevents cognitive decline into old age. Personal experience during the pandemic showed reduced dancing led to decreased muscle tone.
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, running, and also practicing speech, oratory speech and so forth, or singing.
Thank you for the conversation and opportunity to learn. Thank you for inviting me to share science with the community.
Keep Andrew Huberman in your library
Save the videos and channels worth coming back to, and find them again in one place.
