Eagleman, David

neuroscientist, Stanford adjunct professor, author of Livewired and Incognito; founder of the neurotech company Neosensory

David Eagleman: Neuroplasticity and the Livewired Brain

Lex Fridman Podcast · Aug 26, 2020

• David Eagleman
• neuroplasticity
• livewired
• sensory substitution
• Neosensory
• brain-computer interfaces
• in-group bias
• umwelt
• free will
• GPT-3

Core ideas

Eagleman introduces "livewired" as a deliberate replacement for the term "plasticity," arguing the hardware/software metaphor computer scientists reach for fundamentally misdescribes the brain: rather than a fixed substrate running changeable programs, the brain is a system whose physical structure — from synaptic weights down to receptor distributions, cellular biochemistry, and epigenetic gene expression — is continuously reconfiguring itself in response to experience, with no stable endpoint. He illustrates the brain's radical malleability with cases like children who undergo hemispherectomy (removal of half the brain) before age seven and retain near-normal function, and argues plasticity isn't uniform across the brain but calibrated to how stable each domain's input data is: the visual system "hardens" quickly because the visual world is statistically stable, while somatosensory and motor cortices (mapping a body that keeps changing size, shape, and injury status) stay flexible far longer. Drawing on Stewart Brand's "pace layers" concept for cities, he proposes the brain operates as nested layers changing at different speeds — from rapid biochemical cascades to slowly cementing long-term memories — which explains Ribot's Law (older memories outlast newer ones in dementia, since they're encoded in deeper, more stable layers).

Eagleman frames the brain as evolution's solution to an unpredictable world: rather than hard-coding behavior (as in an alligator, whose brain is essentially unchanged across generations), humans are born with a "half-baked" brain built to absorb culture, language, and context — a genetically-encoded capacity for malleability itself, not a blank slate, since substantial innate machinery exists to route sensory inputs correctly and to make the brain "ready" to absorb whatever language or culture it's dropped into. He proposes a "Potato Head theory" of sensory evolution: the brain doesn't need distinct operating principles for each sense (heat pits, electroreception, magnetoreception, etc.) — it simply exploits whatever correlated data streams are plugged in, which is why cochlear implants and retinal implants work despite speaking "the dialect of Silicon Valley" rather than the brain's native code, and why his own company Neosensory can teach deaf users to "hear" via a $399 vibrating wristband that translates sound into patterns felt on the skin (becoming, with practice, a genuine new subjective sensory experience rather than a decoded signal). He extends this to sensory expansion (feeding infrared/ultraviolet sensor data through the same wristband to genuinely perceive normally-invisible radiation) and speculates about sensory addition — feeding entirely novel data streams like the stock market or Earth's magnetic field directly into perception — introducing the concept of "umwelt" (the total slice of reality any given organism, from a tick to a bloodhound, can perceive) to argue humans' sensory window is only a tiny, arbitrary fraction of possible reality. This grounds his skepticism about invasive brain-computer interfaces like Neuralink: while valuable for patient applications (Parkinson's, epilepsy), he doubts most people will accept the risk of skull surgery, and argues the real bottleneck limiting current human-machine interaction isn't bandwidth into the brain (which he calls already close to optimal, evidenced by speed-reading's failure) but getting information in non-invasively — his stated reason for devoting most of his time to Neosensory rather than continued lab research.

On intelligence and AI, Eagleman argues GPT-3, while impressive, differs fundamentally from human cognition because it lacks any model of relevance — an innate drive rooted in survival that lets humans filter overwhelming sensory data down to what matters, and reason explicitly about how specific other people (a boss, a mother, a potential date) will interpret what's said or written, which he says can't be captured by scaling parameters alone. He uses the contrast between a wolf that chews off a trapped leg and adapts to three-legged locomotion versus the Mars rover Curiosity, which failed outright when a wheel got stuck, to argue that true adaptability requires an organism (or system) that has something to care about — a "yen to survive" — suggesting AI research should focus less on synapse-like connection-weight architectures and more on building in analogous drives. On free will, Eagleman takes the mainstream neuroscientific position that the brain, however complex (86 billion neurons, roughly a quadrillion connections), appears to be a physical system with no obvious room for anything outside deterministic causation — while offering, as pure speculation rather than belief, a thought experiment (a Kalahari Bushman deducing a radio's internal wiring without ever imagining distant broadcast towers) to illustrate how thoroughly a science's current framework can miss an entirely different explanatory layer.

On the sources of "evil," Eagleman rejects the idea of localizing moral character to specific brain regions (mocking a Time magazine cover attempting this with Gandhi and Hitler), instead pointing to his own fMRI research on in-group/out-group bias: subjects show a measurably larger neural pain response watching a hand from their own labeled group (e.g., religious affiliation) get "stabbed" with a needle versus an out-group hand, and this in-group preference can be induced within minutes using entirely arbitrary, coin-flip-assigned tribal labels — suggesting figures like Hitler are, in part, products of self-reinforcing crowd dynamics that could form around essentially any shared in-group marker, not some special deficit unique to a "bad brain." He connects this line of research to advocacy for reforming the legal system (via his nonprofit, the Center for Science and Law) — arguing brain differences (schizophrenia, addiction, psychopathy) make uniform sentencing poorly matched to rehabilitation, and proposing specialized courts staffed by judges and juries versed in mental illness and addiction science as a practical, already-adopted reform. He closes on meaning and adaptability, opening his book with a Heidegger-inspired image of a person being born with countless possible life trajectories ("many men") but dying having lived only one, and advising young people to prioritize broad intellectual curiosity and "learning how to learn" over vocational specialization, since livewiring — not fixed skill — is what will let them adapt to jobs and challenges that don't yet exist.

Excerpts