This New Tech Revolutionizes Biology... | Michael Levin
ELI5/TLDR
Your cells talk to each other using electrical signals — not just in the brain, but everywhere in the body. Michael Levin’s lab at Tufts has figured out how to listen to those conversations and, more importantly, how to change them. By tweaking the electrical “software” running on your cells, his team has suppressed tumors without killing cancer cells, fixed brain defects caused by genetic mutations, and regrown legs on frogs that don’t normally regenerate. The big idea: medicine today is like computing in the 1940s — we’re still rewiring hardware when we should be reprogramming software.
Summary
Michael Levin presents his lab’s work on bioelectricity — the electrical signaling system that cells throughout the body use to coordinate what to build and how to maintain it. His central argument is that cells are not dumb building blocks following genetic instructions. They are intelligent agents that store memories about what the body is supposed to look like, communicate those memories electrically, and solve problems creatively when things go wrong.
Levin’s lab has demonstrated this by reading and rewriting these electrical “pattern memories” in several dramatic experiments: preventing cancer by keeping cells electrically connected to their neighbors, repairing genetically caused brain defects using drugs that reset the electrical pattern, regrowing frog legs with a 24-hour chemical cocktail, growing eyes on a frog’s gut by reproducing an eye’s electrical signature, and creating “anthrobots” — tiny self-moving creatures made entirely from human tracheal cells that can heal wounds.
His vision for the future: an “anatomical compiler” where you draw the body part you want on a computer, and it outputs the electrical signals needed to make cells build it. Medicine would shift from chemistry to something more like “somatic psychiatry” — talking to your cells instead of drugging them into submission.
Key Takeaways
- Cells have their own intelligence. Even single cells (and even molecular pathways below the cell level) can learn, form memories, and solve problems. Molecular pathways alone can do Pavlovian conditioning. No brain required.
- Bioelectricity is the body’s coordination language. Every cell has ion channels that create voltage, and electrical synapses that share that voltage with neighbors. This system predates brains by billions of years — it evolved around the time of bacterial biofilms.
- Your body stores a “blueprint” as an electrical pattern. Before a frog embryo’s face forms, there’s already an electrical map showing where each organ will go. Levin calls this the “electric face.”
- Cancer is a communication breakdown, not just a mutation. When a cell’s voltage goes haywire, it disconnects from the electrical network. It “forgets” it’s part of a body and reverts to behaving like a single-celled organism. Levin frames this as a “dissociative identity disorder” of the cellular collective.
- You can suppress tumors without killing cells. By forcing cancer cells to maintain the correct electrical state, Levin’s team kept cells from forming tumors even with active oncogenes blasting away. The mutation was still there. The tumor was not.
- Genetic brain defects can be fixed with electrical reprogramming. Tadpoles with a mutation that destroys normal brain structure were given already-approved drugs that reset the bioelectrical pattern. Result: normal brains with normal learning ability, mutation still present.
- A 24-hour signal triggered 18 months of leg regrowth. Adult frogs (which don’t regenerate) grew functional, touch-sensitive legs after a single day of a chemical cocktail applied to the amputation site. No scaffolds, no stem cell therapy, no micromanagement.
- Human throat cells can be rebooted into tiny wound-healing robots. “Anthrobots” made from tracheal epithelial cells are 100% human DNA, self-moving, capable of stitching wounds together, and biodegrade in weeks. They are nothing like any stage of human development.
- The future of medicine looks more like psychiatry than chemistry. Instead of targeting individual molecules, the goal is to communicate with the body’s intelligence — reset set points, retrain cellular behavior, have conversations (via AI) with your organs.
Detailed Notes
The Anatomical Compiler: The Endgame
Levin opens with his lab’s long-term vision. Imagine sitting at a computer, drawing the anatomy you want — an organ, a limb, a whole organism — and having the system calculate the exact electrical signals to feed your cells so they build it. He calls this the “anatomical compiler.” If it existed, birth defects, traumatic injuries, cancer, aging, and degenerative disease would all become solvable problems. We are, he says, very far from this. But the direction is clear.
Why Medicine Is Stuck in the 1940s
Modern molecular medicine is good at cataloging which genes do what, which proteins bind to which other proteins. That’s the equivalent of knowing how every transistor in a computer is wired. But knowing the wiring diagram doesn’t tell you how to make the computer run a different program.
Levin’s analogy: in the 1940s, to reprogram a computer, you physically rewired it. Today’s medicine is doing the same thing with genes and molecules — manually fiddling with hardware. The breakthrough will come when we realize cells are reprogrammable, like software. You don’t need a soldering iron to switch from Word to PowerPoint.
Cells Are Smarter Than You Think
A single-celled organism called Lacrymaria handles all its own physiology, metabolism, and hunting — no brain, no nervous system. Every one of us started as a single cell that somehow built an entire human body. Even below the cell level, molecular signaling pathways can form six different types of memory, including Pavlovian conditioning. The machinery for learning exists before cells, before neurons, before brains.
The body is what Levin calls a “multiscale competency architecture.” Every level — molecules, cells, tissues, organs — solves problems in its own domain. We just happen to be most familiar with the behavioral level (animals doing animal things), but cells and tissues are navigating problems constantly.
The Picasso Tadpole Experiment
To test whether cell-level intelligence is real, Levin’s team created “Picasso tadpoles” — frog embryos with their facial features scrambled. Eyes on top of the head, mouth off to the side, like a biological Mr. Potato Head. These tadpoles developed into normal-looking frogs anyway. The organs didn’t follow a fixed path. They improvised. They found novel routes to reach the correct configuration and stopped when they got there.
This is a hallmark of intelligence: reaching a goal from unfamiliar starting conditions through paths you’ve never taken before.
How Cells Remember What to Build
The brain stores memories using networks of electrically connected neurons. Ion channels let charged molecules in and out, creating voltages that propagate through the network. Neuroscience bets everything on the idea that decoding these patterns would reveal an animal’s memories, goals, and preferences.
Levin’s insight: this system isn’t unique to brains. It’s ancient. Every cell in your body has ion channels. Most cells have electrical synapses to their neighbors. Your tissues run electrical networks all the time. The brain just repurposed an existing system.
So what do body tissues “think” about? Shape. They think about building the body in the correct shape and defending that shape against aging, injury, and cancer.
The Electric Face
In frog embryos, before any facial features physically form, there’s already an electrical map — a pattern of voltages across cells — that shows exactly where the eyes, mouth, and other structures will appear. Levin calls this the “electric face.” It’s literally reading out the electrical memory that tells cells what a correct face looks like.
This bioelectrical communication also works across multiple embryos. Poke one embryo, and neighboring embryos detect the injury through electrical signaling. The system tends to merge individual units into a coordinated whole.
Reading and Rewriting Pattern Memories
The team monitors these electrical patterns using voltage-sensitive fluorescent dyes (cells glow differently depending on their voltage). But the real prize is rewriting the patterns. They don’t use electrodes or external fields. They manipulate the cells’ own ion channels — the natural interface cells use to control each other — using pharmacology (drugs that open or close specific channels) and optogenetics (light that activates channels).
Story 1: Cancer as a Disconnection Problem
Inject nasty oncogenes (KRAS, mutant p53) into tadpoles, and tumors form. But before the tumor is visible, voltage imaging already shows the affected cells have abnormal voltage. That voltage change causes the cells to electrically disconnect from their neighbors.
Once disconnected, a cell can no longer participate in the collective’s big-picture goals (build and maintain organs). Its world shrinks to just itself. It’s not “more selfish” — it just has a smaller self. Levin calls this a shrinking of the “cognitive light cone.” The cell reverts to single-cell behavior: grow and divide. That’s cancer.
The implication: instead of killing cancer cells, reconnect them. Levin’s team co-injected the oncogene plus an ion channel that forces the correct electrical state. Result: the oncoprotein was expressed everywhere, blazingly strong. But no tumor formed. The mutation was present. The cancer was not. The physiology overrode the genetics.
Story 2: Fixing Brain Defects in Software
Tadpoles with a mutation in the notch gene (critical for brain development) have severely abnormal brains — missing forebrain, malformed midbrain and hindbrain, no meaningful behavior. Levin’s team built a computational model of the electrical pattern a normal brain needs, asked the model what ion channels to adjust, and found a couple of already FDA-approved drugs that did the job.
Result: normal brain structure, normal learning rates, mutation still present. A genetic hardware error fixed by a software patch.
Story 3: Regrowing Frog Legs
Adult frogs don’t regenerate limbs. Lose a leg, and 45 days later there’s basically nothing. Levin’s team designed a chemical cocktail targeting ion channels and applied it for just 24 hours. That single day of signaling triggered about 18 months of leg growth. Regenerative genes switched on immediately. By 45 days, toes appeared. Eventually: a functional leg with toenails, touch sensitivity, and movement.
No scaffolds. No stem cell injections. No ongoing intervention. One brief signal that said “build a leg” and the cells handled the rest. Levin and collaborator David Kaplan founded a company, Morphoceuticals, to push this toward mammalian and eventually human applications using wearable bioreactors.
Growing Eyes on Guts
To prove bioelectric signals are truly instructive at the organ level, the team reproduced the electrical “eye” pattern from the electric face in a different part of the embryo — on the gut. The cells built a complete eye there. Retina, lens, optic nerve, the works. No one told the cells which genes to turn on. The electrical signal was a high-level command: “build an eye here.” The cells figured out the rest.
When only a few cells were injected, they recognized they didn’t have enough members to build an eye — and recruited uninjected neighbor cells to help. The same collective intelligence behavior you see in ant colonies.
Wasps as Nature’s Bioengineers
Oak tree cells normally build oak-shaped things. But parasitic wasps inject signals that hack the plant’s morphogenesis, causing oak cells to build bizarre spiky spheres and other structures (galls) that look nothing like any part of an oak tree. The sophistication of the gall matches the sophistication of the hacker: bacteria make lumpy blobs, insects make intricate structures. What the wasp does by evolutionary accident over millions of years, Levin wants to do deliberately.
Anthrobots: Human Cell Creatures
Take human adult tracheal epithelial cells — the ones that line your throat and move mucus around — and give them the right conditions. They “reboot” into self-moving little creatures that look like pond organisms but are 100% human DNA. These anthrobots can form clusters, crawl over wounds in neuronal cultures, and stitch the two sides together. They biodegrade in weeks and wouldn’t trigger immune rejection because they’re made of your own cells.
Potential applications: cleaning joints, hunting cancer cells, delivering regenerative molecules, repairing neural connections. This is the first thing discovered about what these cells can do in a new form factor. The range of capabilities is likely much wider.
The Future Vision
Levin envisions an AI system that speaks normal language, connected to wearable sensors collecting data about your body. Through it, you (or your doctor) would essentially have conversations with your organs. Treatments would use electroceuticals (ion channel-modifying drugs) and optogenetics (light-based channel control) to retrain cells, reset set points, and provide information — the same kinds of interventions used in behavioral science, applied to cellular collectives.
The Theory of Biology
During Q&A, Levin makes a bold claim: the unified theory of biology won’t look like physics equations or complexity theory. It will look like psychology. The fundamental story of life is the scaling of intelligence — tiny molecular goals getting aggregated into cellular goals, then tissue goals, then organ goals, then organism goals. Bioelectricity is interesting only because it happens to be the “cognitive glue” that enables this scaling. The real theory is a theory of intelligence, and it’s already producing results (regenerated limbs, normalized tumors, repaired birth defects).
On Cancer’s Permanence
Someone asks if cancer can be eliminated entirely. Levin’s answer: no. Cancer is the default state. The real question isn’t “why do we get cancer?” It’s “why isn’t it all cancer, all the time?” The answer is the communication networks that let cells have goals bigger than themselves. You can treat it, prevent it, but you can’t eliminate the failure mode of a system whose entire existence depends on not failing that way.
On Psychiatric Disorders
Levin distinguishes between organic brain problems (fixable through bioelectrical reprogramming) and existential suffering (“the thought that breaks the thinker”). You can fix the hardware. You can’t fix the depression that comes from realizing something true and terrible about the universe. That’s not a brain problem. That’s a being-alive problem.
Quotes/Notable Moments
“Where medicine is today is where computer science was in the 40s and 50s.”
“Cancer is in large part a dissociative identity disorder of the cellular collective intelligence.”
“The oncoprotein is blazingly strongly expressed. It’s all over the place, but there’s no tumor.”
“This is an example of fixing what is fundamentally hardware error — meaning the mutation. You fix it in software.”
“You’re not going to have a conversation with your liver about the movie that you saw. But you absolutely could have a conversation with your liver about what happened yesterday when you drank too much.”
“The question isn’t why do we get cancer? The question is, why isn’t it all cancer all the time?”
“There are problems which are not organic disease. They are ways of thinking or experiences that lead to specific patterns of thought that are harmful. Those things are not going to be handled at the level of repairing the brain. Now you’re into psychoanalysis and environment and conversations and love and whatever else.”
“You’re telling me that these people had to live their entire life subjected to dumb bacteria, viruses, some kind of effect of some stray cosmic ray hitting their cells when they were an embryo. They had to live in whatever body they were given at birth by accident… How did people live this way?”
Claude’s Take
What’s solid:
The experimental results Levin presents are real, peer-reviewed, and reproducible. The cancer suppression via bioelectrical normalization, the brain repair in notch-mutant tadpoles, and the frog leg regeneration have all been published in serious journals. The anthrobots paper was in Advanced Science. These are not speculative claims — they’re demonstrated in animal models. The “electric face” voltage imaging is genuinely striking data.
The core insight — that there’s a layer of information processing between genes and anatomy, and that layer is electrical — is increasingly well-supported and gaining mainstream traction. It’s not fringe anymore.
What’s strong but oversimplified:
The computer science analogy (hardware vs. software, medicine stuck in the 1940s) is useful but hides real complexity. We’re not just missing a programming language for cells. We don’t fully understand the “operating system.” The jump from tadpole experiments to human medicine is enormous, and Levin acknowledges this but breezes past it. Frog legs are not human arms. Tadpole brains are not human brains.
What’s speculative:
The claim that a unified theory of biology will look like psychology rather than physics is a philosophical position, not a demonstrated fact. It’s an interesting bet, and the experimental results give it some credibility, but “biology is fundamentally about intelligence” is still a contested framework. Many biologists would argue the molecular and genetic layers are doing the heavy lifting, and bioelectricity is one signaling mechanism among many.
The “anatomical compiler” is decades away at minimum, if it’s possible at all. Levin presents it as an endgame, which is fair, but the gap between “we regrew a frog leg” and “draw any body part on a computer and cells will build it” is roughly the size of the Grand Canyon.
What to watch for:
The Morphoceuticals mammalian trials are the key test. If the leg regeneration approach works in mice or larger mammals, that changes the conversation substantially. The anthrobots are fascinating but very early-stage — “they can heal a scratch in a petri dish” is a long way from “they can clean your joints.”
Bottom line: Levin is doing some of the most genuinely original work in biology right now. The experimental results are real and the framework is productive. The philosophical claims about intelligence being the fundamental story of life are interesting but unproven at the grand scale he suggests. He’s earned the right to speculate — the data backs him up more than most people in his position — but the distance between current results and the vision he paints is still very large. Worth taking seriously. Not worth taking on faith.