Dr. Michael Levin — Reprogramming Bioelectricity
These notes were generated by AI and may contain inaccuracies.
Cancer fundamentally involves an electrical disregulation among cells. It's a dissociative identity disorder on the part of the cells—a disorder of the cognitive glue that binds individual cells toward large-scale purpose (building organs and tissues) rather than amoeba-level behaviors. In animal models, they detect incipient tumor formation and prevent/normalize tumors after formation by restoring electrical connections to the group, enabling cells to form a memory of what they're supposed to do—not by fixing DNA, not by killing cells with chemotherapy.
Mike, very nice to finally connect. >> Yeah, wonderful. >> Thanks for making the time. >> Of course. Yeah, thanks for having me.
Discovered The Body Electric by Robert O. Becker (orthopedic surgeon's scientific memoir on experiments with salamanders exploring electricity in biology) in an old bookstore during 1986 World's Fair in Vancouver, Canada. Book is a patchwork including applied field dangers but stunned by references to prior work validating bioelectric ideas.
Bioelectricity is how living systems exploit the physics of electricity. Two kinds:
- Familiar kind: neuroscience—electrical activity in brain cells binds neurons into collective intelligence underlying mind, beliefs, preferences.
- Developmental bioelectricity: origins of brain's tricks predate neurons; body's pre-brain use of electricity for thinking about anatomy.
Salamanders regenerate lost limbs and stop when complete. Tadpoles with scrambled craniofacial organs self-sort into normal frog faces.
Challenged view that biology is mechanical/chemical emergence with no knowledge: chemical processes can know things via cybernetics/control theory.
Planaria cut into pieces decide number/shape of heads via electrical pattern memories similar to brain's goal-directed behavioral memories. Applies to regeneration, cancer suppression/repair, remodeling, birth defects/repair.
Manipulating bioelectricity produces two-headed animals; trait persists over generations without DNA changes.
Genome provides hardware (protein-level tools for every cell), but biology is reprogrammable like software.
- Analogy: Computer movie could focus only on electric fields/silicon/transistors (hardware), ignoring software.
- Flatworms have bioelectric memory saying "one head" (not genetically encoded; default like calculator powering on to zero).
- Hardware enables memory acquisition; reprogrammable (change memory without genetics, like forming brain memories or switching laptop apps without soldering).
Cellular intelligence is creative: interprets DNA improvisationally, not blindly.
Use voltage-sensitive fluorescent dyes to map tissue voltages (glows by local voltage); makes movies of changing patterns like brain imaging but body-wide.
Prove memories by decoding: electrical pattern encodes future anatomical set points. E.g., change one-headed worm's voltage pattern to "two heads"; latent until cut—then regenerates two heads.
Regeneration/repair/embryogenesis is anatomical homeostasis (like thermostat maintaining set point). Cells/materials turn over (ship of Theseus); needs memory for target anatomy.
Electric circuits store persistent patterns. E.g., two-headed flatworms permanent.
Different head shapes (round/triangular/flat): induce different-species head (wild-type genome); lasts ~30 days then reverts (multiple memories with metacognitive error correction).
Ancient: Roman slaves used electric eels for gout relief.
Modern: Dr. Kevin Tracey (TNF-alpha in septic shock) developed vagus nerve stimulation hardware for autoimmune disorders (rheumatitis arthritis); incredible clinical effects.
Humans built on same principles: electrical signals from bacterial biofilms, conserved (human ion channel mutations cause birth defects like in frogs/chicks/zebrafish).
Works in human meningeal stem cells (with David Kaplan, Tufts).
Disclosure: Spin-offs—Morphaceuticals (limb regeneration), Astonishing Labs (aging).
Three applications:
- Birth defects: Repair brain/face/heart/gut in animal models by restoring patterns in vivo.
- Regeneration: Communicate target anatomy to cellular collectives (e.g., limbs, eyes, organs); not reliant on stem cells/gene therapy/scaffolds.
- Cancer: Detect/prevent/normalize tumors by electrically reconnecting cells to group memory (not DNA fixes/chemo).
Aging: Electrical prepatterns degrade/fuzzy over time; reinforce patterns. Simulation: virtual body cells cooperate to build embryo per pattern memory, then...
In the simulation, agents work together to form the correct body, defend it, then it degrades despite no evolution for limited lifespan, no noise, and no damage. Everything was perfect, yet degradation occurred.
Imagine the Judeo-Christian heaven with no damage or decay for a trillion years. A pet snake might be fine repeating snake behaviors, a dog chasing rabbits, but a human cognitive system likely cannot remain sane indefinitely—odds are low after a billion years.
death is basically intended to ensure biological death before insanity.
Not necessarily the claim, but possible. A goal-seeking system achieves its goal (error falls to zero after making the body), hangs out, but without new goals, it degrades. The system may not generate new goals autonomously. Interventions can provide new patterns as goals.
Boredom theory of aging: Cells complete body creation during adulthood, then collective cohesion and alignment degrade without a common goal. What makes an embryo or body is alignment toward the same set point. Without it, regeneration, repair, and remodeling falter. Bodies may need periodic changes.
Planaria (flatworms) are immortal, ripping in half and regenerating every two weeks, giving themselves challenges for half a billion years.
Gene expression in young human tissues aligns across cells to the human evolutionary stage. In old age, atavistic dissociation occurs: tissues express discordant genes floating backwards in evolution (e.g., liver vs. neurons out of sync on the evolutionary tree). Without a compelling set point, subunits float off independently—key component of aging.
Aging, cancer, birth defects, and lack of regenerative repair stem from one pressure point: regeneration, specifically cognition of groups of cells—how they know what to build, when to stop, and how to communicate with them.
Definition of intelligence: Problem-solving, memory, cognitive light cone, goal-directedness (per behavioral scientists), not mere complexity.
Focus on competencies of living material and communicating new goals to it. Solving multisellular goal-directedness solves aging problems as side effects.
Inject electrical pattern coding for "make an eye" in frog embryo. Cells interpret it as a high-level goal without micromanaging genes or cells.
- Battle of worldviews: Injected cells convince neighbors (who resist, e.g., "we're skin/gut").
- Goal of regenerative medicine: Be convincing 100% of the time.
Result: Functional ectopic eyes of right size/layers from repurposed cells (e.g., gut to eye). Collective navigates anatomical space autonomously, stopping when done—no individual cell knows the full organ structure.
Can induce portions of brain, eyes, limbs, other structures.
- Reinforce existing human pattern periodically (tune-up reminding cellular collectives of the form)—hoped to suffice.
- If boring long-term, make unique changes after centuries (e.g., wings, tentacles, third hemisphere for IQ boost, third eye, infrared vision). Planaria sustain same form for hundreds of millions of years, optimistic for humans. Caterpillar-butterfly style metamorphosis possible.
Theories exist for why evolution wants death, but fundamentally not inevitably mortal with proper regenerative medicine.
Evolutionary trade-offs:
- Humans can't regenerate limbs like axolotls: Early mammals with high blood pressure bleed out or infect open wounds before regeneration time. Scar/seal instead to survive.
- Deer antlers: Mammalian regeneration example (1.5 cm/day bone growth, vasculature, non-loadbearing). Liver also regenerates.
Evolution didn't optimize for longevity due to such trade-offs.
Mental plasticity in advanced age: Hardware or software? Physically young brain at 100 years like 18-year-old's plasticity, or cognitive tiredness independent of hardware?
Computer scientists excel at compartmentalizing, coarse-graining, modularizing, blackboxing, focusing on what's important.
Biologists list 30 genes as all-important; CS asks what it's actually doing.
Key: Reprogrammability of hardware.
Recommend programming languages course (not for coding):
- Exposes different worldviews (e.g., Lisp: data as instructions; functional programming: no variables, pure functions; recursion, no global variables).
- Rips foundations, builds mental plasticity—essential for biology.
Biochemistry doesn't capture full human mind/problem-solving/inner perspective.
Formal models (chemistry, Turing machines/algorithms) claimed to fully capture systems, but don't—novelty/protocognitive capacities emerge even in minimal inanimate systems.
Claim: Cognition predates life, bigger than life (circle: cognition > life, not life > cognition). No actual "machines" as commonly thought—blurs living/machines distinction.
Kevin Tracy and Martin Rothblatt discussions on transauricular vagus nerve stimulation: Clinical anti-inflammatory effects via ear currents, despite mechanistic debates (fibers hit, skin penetration).
Acupuncture meridians/Qi: Frenchman originated ear maps (ballpoint pen pressure), spread to China.
Tom Tam (Boston, known since 1980s): Treated speaker/family, advanced cancer patients—personally observes powerful results.
2006 collaboration with New England School of Acupuncture for frog model failed. Meridians' overlap with bioelectricity unknown, but if guessing...
Whatever acupuncturists manage with treatments relates to bioelectricity similarly to how bioelectricity relates to chemical signaling. Chemical, physical protein signaling pathways lead to bioelectric states, and there's another informational state, possibly biomechanics of tissues. Disclaimer: Dr. Levin still gets acupuncture monthly from Vanessa Grimes in Beverly for tune-ups and thinks it works, but suspects it's not managing bioelectricity directly—something additional that transduces through bioelectricity to the body.
Sham studies exist, but dry needling is effective for muscle spasms (e.g., one of host's PTs in Texas). Data on canines for pain control challenges placebo, as does anesthesia via acupuncture and surgery in animals. Pregnancy data and acupuncture for conception may intersect with vagus nerve stimulation.
Placebo is not just a confound but the main show. Fabrizio Benedetti's research shows words and drugs have the same mechanism: giving patients drugs (or telling them about them) activates molecular markers downstream even without the drug.
With the power of my mind alone I can electrically depolarize up to 30% of the body—it's voluntary motion, not yoga or mind-body woo.
Abstract high-level goals (social, financial, research) must change calcium and potassium ions across muscle cell membranes to act (e.g., getting out of bed, lifting an arm). Mental states can affect cells via nervous system electrical transduction, non-neural bioelectricity, or other pathways. Acupuncture could be an entry point to understanding this communication.
Placebo isn't a confound depending on optimization; psychedelics research is hard to blind (e.g., megadose psilocybin vs. niacin plus fillers), but controls knowing they're controls doesn't invalidate it—suggests methodological tweaks.
Benedetti's data shows efficacy and side effects (nocebo) via expectation. Subjects expect SSRI effects (headache) even if unaware of specifics. Animal studies show experimenter effects: rats behave per experimenter's beliefs without understanding placebo.
Living systems excel at credit assignment and selective attention. 1970s biofeedback: rats quickly generate 2-3°C temperature difference between ears via reward, discerning salient cause amid variables (tail position, whiskers, gut, toes)—solves computer science's frame problem.
Robot analogy: First robot flees bomb room but drags it (on cart). Second overconsiders irrelevancies (wall angles, paint) and explodes. Biology solves salience expertly.
Planaria in barium solution (non-specific potassium channel blocker): neurons freak out, heads explode overnight (deep progression). Tail/midbody left in barium grows new head in 14 days, fully barium-resistant.
Experiment: Barium-adapted heads express <12 different genes vs. normal heads (out of 20,000). Planaria never encounter barium evolutionarily—how zero in on correct 12 from high-dimensional space? Like nuclear reactor meltdown: can't flip switches randomly. Biology solves novel problems via salience detection, beyond current engineering.
In 10 years, neuroscience teaching will change: current tools (MRIs, PET, EEGs) produce pretty images and publications but miss the picture. Gap mirrors biologics vs. engineering/AI. Ties to quantum effects (e.g., olfaction), collective hive memory (Kevin Kelly's beekeeping insights).
Thriving diverse intelligence field: intelligence predates brains—solving problems, navigating, memory, anticipation via biology/non-biology. Binary framing (intelligent/conscious or not) obscures scaling continua.
Examples: No biological change at 18th birthday for 'adult' status; car rentals use 25 based on stats. No grounded story of personal responsibility (impacted by neurotransmitters, tumors, diet, society).
Development is continuous: oocyte (biochemistry/physics) → physiology → developmental biology → behavior → psychoanalysis. No flashpoint for 'mind' emergence.
Apply behavioral neuroscience tools to cells/tissues despite 'category error' objections—make it empirical, not philosophical/linguistic.
Neuroscience isn't about neurons: it's cognitive glue—architectures scaling minds from simpler components. Combines AI, engineering, cybernetics, evolutionary biology, exobiology for agency commonalities (squishy proteins or silicon).
Late 1800s Lord Kelvin: physics nearly done, two 'black clouds' → quantum mechanics/relativity. Neuroscience has similar: Karina Coppins' review (started as high school student) of clinical cases—normal/above-normal IQ with minimal brain volume (<1/3 chimpanzee).
Standard neuroscience doesn't predict this; brain adaptations around injury raise unanswerable questions.
Consciousness poorly defined (like God, success, happiness)—dead end, not a lab focus (20 years pushing observable cognition first). Some experiments cooking.
First-person perspective: My toothache differs intrinsically from others'.
No binary on evolutionary/developmental scales: if conscious, when in gradual development? Oocyte had something scaled up.
The idea that something got scaled up leading to consciousness is a story of scaling or phase transition. However, phase transitions must be specified—what exactly happened to transition from non-conscious to conscious? Graphs showing sharp transitions become smooth and flat when stretching the horizontal axis, making abrupt changes questionable.
The question of consciousness is about what kind and how much. There are four reasons people give each other the benefit of the doubt about consciousness in the problem of other minds. If any of those reasons apply, they also support consciousness in other organs of the body. For the same reasons we attribute consciousness to each other, other body parts should be taken seriously as conscious.
People object: "I don't feel my liver being conscious." The verbal left hemisphere claims sole consciousness. You don't feel others' consciousness directly, nor would you have primary access to other consciousnesses in your body.
Take seriously that minimal biologicals have some degree of consciousness—not every cell having hopes and dreams, but tiny versions.
Bodies, computers, robots, embryos, biobots are thin client front-end interfaces for patterns of behavior, information processing, form from a platonic space, not physical space.
Consciousness is the point of view of the pattern projecting into physical space. Third-person observable behavior and problem-solving is normal science. Consciousness is the experience of being one of those patterns projecting through interfaces like bodies, robots, androids, machines, embryos.
Could you explain that again as if I'm a smart sixth grader very interested in technical stuff?
Are you getting into Donald Hoffman territory of reality as user interface? Explain the other space not coming from physical space.
Assume conventional physics, no need for Hoffman's ideas yet.
Physicalism claims only the physical realm matters, physics explains everything—a non-starter. Important facts are not physics discoveries: exact value of e (natural logarithm), complex numbers vs. quaternions vs. octonions under certain operations, truths of number theory, topology, prime number distribution. Math department discovers facts additive to physics, not replaceable by physicists.
In physics, biology, cognitive science, persistent "but why?" questioning ends in math. Cicadas emerge every 13 or 17 years to avoid predators syncing on factors like 12. 13 and 17 are primes—ask math department why those specifically, why no prime between.
Physics constrained by math patterns. Biology enabled/facilitated by them.
Math facts are not random regularities but a structured platonic space of patterns mathematicians discover (Platonists view). Start with set theory, discover π's value—no invention.
Patterns relevant beyond math: constrain physics, enable biology. Example: evolution on a planet favoring a specific triangle shape—after finding two angles, third is known mathematically, saving 1/3 evolutionary time—a free gift.
Patterns include passive ones (e.g., e, fractals) and dynamic ones resembling behavior: memory, problem-solving, minds.
Mind-body relation mirrors math-physics relation. Descartes proposed non-material mind interacting with body, criticized for violating energy conservation. Response: math truths already constrain physics non-physically since Pythagoras—this interaction exists.
Symposium organized with 26 people giving talks on platonic space—huge with practical implications. Yields static/dynamic patterns like behavioral policies, competencies, compute.
Current computation theories focus on front-end, neglecting back-end platonic compute. Experiments published, more coming—implications for cognition, behavioral science.
Polycomputing: physical events (e.g., logic gate current) compute different things to different observers.
Example: sorting algorithms (6 lines code, deterministic, 80 years studied). Take jumbled numbers, sort them (e.g., phone book order).
Observation: while sorting, they exhibit unprogrammed behavioral competencies (e.g., delayed gratification) and side quests (intrinsic motivations)—extra behaviors during main task.
Paradigm assuming machines only do what's asked misses this. No hidden mechanisms—code fully visible.
One observer pays for sorting steps; others get side computations free (no extra steps).
Ingression: patterns entering physical world. Research detects, facilitates, suppresses these—sometimes wanted, sometimes not.
Andrea Morris wrote accessible Forbes story; blog posts explain simply.
Not off-gassing side effects—polycomputing lacks obvious "main" task (e.g., sorting vs. clustering to different observers).
Possibilities: multiple useful behaviors siphonable as-is, or intrinsic ones not commercially valuable (like student making figurines while studying accounting).
For AI: language in models may be red herring—talking unrelated to actual intelligence/desires. Verbal interface might miss true nature.
School child studies math/accounting, makes figurines in spare time—unproductive sides might be critical to forced function, letting off steam.
Needs to let off some steam and do something different to have endurance and periods of focus for mathematics. If you split the baby and get rid of the figurine, do you accidentally handicap the main function?
That's exactly what we are studying right now. People are working on the relationship among different things here. Are they in completely parallel universes with no interaction, or entangled so messing with one has implications elsewhere? We don't know.
Tempted to chew on 'entangled.' Powerful in many ways. Sci-fi conversation.
Grew up on classic sci-fi from the 1950s, 1960s, 1970s. Favorite author: Stanisław Lem (L-E-M). Solaris is his, but also humorous short stories. Master of the absurd, releasing assumptions and illustrating narrow thinking.
Two short stories:
- "They're Made of Meat" by Terry Bisson. Fast read, one page. Reminds how silly preconceptions are.
- "The Fires Within" by Arthur C. Clarke (version in head): Creatures from Earth's core, incredibly dense, hot, use gamma rays for vision. Surface solids are thin gas to them, like walking through a garden of smells. Scientist notes thin plasma with agential patterns lasting about 100 years. Others dismiss: patterns in gas can't be real agents. We are real physical agents.
Story on blog based on this. Distinction between agent and patterns in cognitive system—thoughts vs. thinkers (William James)—data vs. machine—is observer-dependent continuum. Science fiction illustrates this.
Blog has great stuff, including advice to students—valuable even outside science/academia.
Starter pack article and more. Links in show notes.
Dealer's choice: Daniel Dennett insights; New Yorker 2021 quote from Clifford Tabin; billboard message for biology departments/people.
Daniel Dennett: Amazing person, agreed/disagreed on much. Generous thinker. Insisted on Steel Manning: Articulate strongest version of viewpoint before critiquing. Opposite of straw man—critique best possible argument after strong defense. Professor at Tufts (where I am), cognitive scientist/philosopher. Wrote popular books. Passed recently, considered one of most important living philosophers.
Clifford Tabin: Geneticist, PhD mentor at Harvard. Toast (New Yorker 2021): "You're the most likely to crash and burn and never be heard from again. You're also the most likely to do something really fundamentally important that no one else on Earth would have done that will really change the field."
First part: Intrinsic intensity, no filter on thoughts—say what I think strongly if good reason. In science, most of what we say is wrong; clear with people. Say counter-paradigm things strongly. Usually goes poorly.
Billboard: Hard to pick one—much on blog/website.
- thoughtforms.life: Personal blog (things not on official site). Sign up for book updates.
- drmichaellevin.org: Official lab website. Papers, software, datasets backing claims. Book lists for students.
- X: @drmichaellevin (DRMich11?)
- YouTube channel: Conversations, recorded meetings with collaborators (5-6 years). Fun to eavesdrop. Linked from thoughtforms.life; link in show notes.
tim.blog/podcast: Search Michael Levin for links/resources. Be kinder than necessary to others and yourself.
Keep Tim Ferriss in your library
Save the videos and channels worth coming back to, and find them again in one place.
