In Michael Levin’s lab at Tufts, a biological system assembled from frog cells moves around a petri dish. It has no brain. It has no neurons or nervous system. It is removed from the embryonic context that would normally give it instructions about where to go or what to do. It navigates obstacles. It repairs itself when damaged. Put a population of them together and they coordinate.
The system is called a Xenobot. It’s the most vivid demonstration of the claim Levin and his co-authors have been developing for a decade: that problem-solving, memory, and goal-directedness are not exclusive properties of brains. They are properties of biological systems at many scales.
This is not a wellness idea. It is peer-reviewed research. And it changes what you can sensibly think the body is.
What the research actually claims
Bioelectric cognition is the proposal — developed in Michael Levin’s lab at Tufts and built into a framework in Levin and Resnik’s “Mind Everywhere” paper — that cells store, process, and act on information through patterns of voltage across their membranes. These patterns carry instructions for anatomical form that operate upstream of genes and chemistry.
Three experimental demonstrations have done the load-bearing work:
Planarian regeneration. A planarian flatworm cut in half normally regrows one head. In Levin’s lab, by altering the pattern of voltage across the cut surface, researchers induced the worm to regrow with two heads. No genetic change. The change was in the bioelectric pattern. And the two-headed configuration persisted through subsequent regenerations — the worm had, in an operational sense, a new memory of what it was supposed to be.
Xenobots. Frog stem cells, separated from the embryo that would normally organize them, self-assembled into mobile biological units. These units navigate, repair, and in later work have been shown to reproduce by pushing loose cells together into new copies of themselves. No neural tissue. No central instructions. The behavior emerges from local interactions between cells — including bioelectric signaling — rather than from any central controller.
Tadpole regeneration with misplaced eyes. An eye grafted onto a tadpole’s tail wires itself into the nervous system and contributes to vision. The body adapts the new component to the existing goal.
Across all three, the pattern is the same: the biological system is not following a pre-specified construction plan. It’s pursuing a target, using whatever materials it has, through signals that operate between cells.
Levin’s term for this is diverse intelligence: the observation that the problem-solving we recognize in brains shows up, in weaker and more distributed forms, in cells, tissues, and whole organisms. The word “intelligence” is doing specific work here, and it’s worth being precise about. Levin isn’t claiming that a flatworm has a mind in the way you do. He’s claiming that the problem of “grow a head in the right place” is a problem, that the system solves it, and that the solving is mediated by a signaling layer we can now see, measure, and in some cases modify.
What this changes about older models
The older model of biology runs like this: DNA encodes the instructions for how to build an organism. Gene expression, regulated by chemical signals, specifies what each cell becomes. Form follows from the parts.
The older model is not wrong. It is incomplete.
Levin’s research fills in a layer the older model leaves out: how cells coordinate on a target configuration when the instructions in their DNA are the same. A liver cell and a kidney cell have identical DNA. What tells them which one to be, and where? The older model answers this with gradients of chemical signals — which is part of the story, but not all of it. The fuller answer, as the research is developing it, is that cells communicate via voltage patterns across their membranes, and those patterns carry information about what the whole tissue is supposed to become.
The practical reframe: DNA encodes the parts. Bioelectric patterns encode the configuration — what the parts are supposed to build. You don’t have to change DNA to change an outcome. You change the pattern.
This is the engineering claim underneath the research. It’s the reason Levin’s lab can induce two-headed regeneration in planaria without editing the genome. It helps explain why Xenobots, assembled from cells with no embryonic context, can still self-organize into coherent mobile units: local interactions — including bioelectric signaling between cells — carry enough information for them to coordinate. The layer of control is different from the one biology textbooks have spent the last fifty years describing.
Where this meets a nervous system that won’t stop running
The framework applies to tissues, cells, and whole organisms. It also applies, at least in principle, to nervous systems. A nervous system is a biological system maintaining a configuration. If it’s settled into a pattern of sustained activation — the one that shows up as three threads processing under dinner conversation, or 2am scenario rehearsal — then the question the framework lets you ask is: what is the system defending, and what would change the pattern?
I want to be careful about the next step. The framework is vocabulary for asking the question more precisely. It’s not an answer to it. Levin’s lab studies regeneration, morphogenesis, and the behavior of biological assemblies. It doesn’t study human nervous systems under cognitive load. The application of the attractor-state framework to a nervous system that won’t stop running is this newsletter’s extension of the vocabulary. The research itself doesn’t make that extension.
The open question
Bioelectric cognition research describes how bodies build themselves. It does not prescribe treatment. It does not describe or validate any applied practice, including the one this newsletter sits alongside. The causal mechanism connecting bioelectric cognition research to any specific applied practice remains unknown.
What the research does is make a specific claim available: the biological system isn’t just a pile of parts running genetic instructions. It’s a problem-solver operating at scales we didn’t previously know how to look for. The implications for medicine, regenerative biology, and the understanding of what a body is are substantial. The implications for any specific applied intervention are a separate question, one the research itself doesn’t answer.
What a careful reader takes from this issue is a more precise vocabulary for what a body is and what it does. Not a protocol. Not a claim. A framework.
Johanna Farrimond holds degrees in Physics and Electrical Engineering and an MSc in Information Assurance with Honors. Research framework: bioelectricrecovery.com/research
If you found this useful, the most helpful thing you can do is forward it to one person who thinks carefully about evidence.


