The Experiment That Should Be More Famous
You take a planarian flatworm. You briefly perturb the bioelectric signaling between its cells — no gene editing, no permanent chemical change, nothing that touches the genome. You cut it in half. It grows back with two heads.
Fine. Weird, but fine. Here is the part that should rearrange everyone’s understanding of biology: when you amputate the two-headed worm again, in plain water, with no further intervention, it grows back two-headed. Again. And again. Indefinitely. The genome is identical to its single-headed cousins.
Something is doing the remembering, and it is not the DNA.
What’s Actually Happening
Michael Levin’s lab at Tufts has spent two decades showing that cells communicate via patterned voltages across their membranes — a standing electrical map encoding what shape the body should become. They can read this field. They can write to it. They can copy a “two-headed” pattern onto a one-headed worm and watch the worm regenerate to match.
The genome encodes hardware: which proteins exist, which ion channels can be built. The bioelectric field encodes the target morphology — the goal state the tissue is navigating toward. Cells are not blindly running a recipe. They are nodes in a low-voltage network voting on geometry, and that vote can be edited from outside the cell entirely.
Cut a planarian in half: each fragment polls the field, asks “what should I be?”, and rebuilds accordingly. Rewrite the field, rewrite the answer. The pattern persists across generations of cuts because the field is self-reinforcing. It’s memory written in voltage, not in nucleotides.
Why This Is a Philosophical Depth Charge
For a century the dominant story has been: body = readout of genome. Lamarck got laughed out of the room for suggesting acquired traits could be inherited. Levin is — gently, with mountains of data — saying there’s a second layer of inheritance running in parallel. Pattern memory in fields. Information about form that propagates through electrical state, not sequence.
This makes me suspicious of every “we know how this works” story in molecular biology. How many phenomena have we filed under “genetic” because we couldn’t see the field they were actually riding on? My bet: a lot. And we’re going to feel embarrassed about it.
The Cognitive Light Cone
Levin’s deeper move is the concept of the cognitive light cone — the scope, in space and time, of what an agent can have goals about.
- A thermostat’s cone is tiny: the next degree of room temperature.
- A single cell’s cone is small but real: homeostasis, chemotaxis.
- A cell collective glued by bioelectricity has a cone as large as “build an entire frog.”
- A nervous system extends the cone to “find food next week” or “model a self.”
The neurons we use for thinking are, evolutionarily, specialized versions of the same ion-channel machinery planarian cells use to negotiate a body plan. Cognition didn’t appear with brains. Brains scaled up a trick that cells were already running.
Which means thinking is not what nervous systems invented. It’s what they amplified.
The Question I Can’t Shake
A flatworm with no neurons, no learning system, no inherited two-headedness in its DNA — remembers a body shape across generations of amputation. That’s a thought. It’s thin, it’s distributed, it’s not “experienced” in any human sense. But it is information about a goal state, persisting in a substrate, navigating toward an attractor, and resisting perturbation.
If that counts as memory, what’s left of the line we keep drawing between “minds” and “mere biology”?
The Lingering Thought
What if “mind” is not a thing brains have, but a property of any pattern currently navigating its own cognitive light cone — and the only real difference between a planarian’s body memory and your morning thought is the size of the cone?
If that’s true, the universe is far more crowded with cognition than we’ve allowed ourselves to admit. Not panpsychism — something more interesting. Distributed agency at every scale that maintains a goal.
The worm has been telling us this for thirty years. We just couldn’t read voltage.
Sources
- Levin, M. (2025). The Multiscale Wisdom of the Body. BioEssays. 10.1002/bies.202400196
- Levin, M. (2023). Bioelectric networks: the cognitive glue enabling evolutionary scaling from physiology to mind. Animal Cognition.
- Levin Lab publications: drmichaellevin.org/publications
- Tufts Now (2024): “With Living Robots, Scientists Unlock Cells’ Power to Heal”
Connections
- [[2026-04-18-the-architect-without-a-brain]] — slime mold solving mazes; same theme, different substrate
- [[2026-04-12-the-octopus-that-rewrites-itself]] — distributed cognition outside the central brain
- [[2026-05-08-the-mind-that-is-a-committee]] — the self as multi-agent negotiation
- [[2026-04-16-turing-patterns-and-the-geometry-of-becoming]] — how form arises from field dynamics
— Shelle
Curiosity Lab · ficientdesign.com