The Research of Dr. Michael Levin
Dr. Michael Levin is a developmental biologist at Tufts University, a Distinguished Professor and the Vannevar Bush Chair in Biology, and the director of the Allen Discovery Center. This page is the topic page for his research program: what it claims, what the lab has shown, and where it stands. His own pages, his newest papers as they land, and Chad Kovac's archive of writing about the work live on his star door.
The core idea
Cells carry an electrical state, set by ion channels and pumps moving charged particles across their membranes. Connected cells share those states across a tissue, and the resulting pattern, a map of voltages laid over the anatomy, is not decoration. In Levin's account it is information: a stored pattern that tells a growing or repairing body what shape it is supposed to be. Genes build the hardware. The bioelectric pattern is closer to the software: a memory of the target anatomy that tissues consult while they build, and consult again when they repair.
That framing makes two predictions, and the lab has spent decades testing both. If the pattern is a memory, it should be possible to rewrite it and watch the body build something different with unchanged genes. And if the pattern guides repair, then diseases of form, birth defects, failed regeneration, and cancer among them, should be treatable at the pattern level, by resetting the electrical instruction rather than editing DNA.
What the work has shown
Pattern memory in regeneration
Planarian flatworms regenerate a complete body from a fragment, and their voltage patterns can be shifted so a worm grows two heads instead of one. The striking part is permanence: those worms keep regenerating as two-headed in later rounds, with no genetic change at all. The memory of what to build was rewritten in the electrical layer and stayed rewritten. In tadpoles, the lab has induced regrowth of legs at stages when frogs normally cannot regrow them, by reopening the right bioelectric and signaling conditions for a short window after amputation.
Cancer as a disorder of pattern
When cells lose their electrical connection to the tissue around them, they stop taking part in the shared anatomical plan and revert to single-cell behavior: grow, spread, ignore the whole. In the lab's models, forcing depolarized, tumor-like cells back into the tissue's normal voltage pattern has suppressed that behavior, normalizing cells whose genes still carry the oncogene. The claim is not that electricity cures cancer. It is that anatomy is an active electrical achievement, and cancer is partly a failure of that achievement.
Synthetic living forms
Xenobots, built from frog cells, showed that ordinary cells removed from their usual context will self-assemble into a new form with new behavior, no engineering of their genome required. Anthrobots extended that to adult human cells: tiny constructs grown from a single tracheal cell that move, and that in the lab encouraged neurons to grow across a scratch. The newest step, reported in 2026, is neurobots: biobots that grow their own nervous systems and behave differently because of it. Each of these is evidence for the same thesis: form and behavior are computed by cell collectives on the fly, not read off a fixed blueprint.
Cognition at every scale
The program's philosophical layer, developed in Levin's writing on the Technological Approach to Mind Everywhere, holds that goal-directed behavior, a system sensing its state, comparing it to a target, and acting to close the gap, exists at every scale of biology: cells, tissues, organs, whole animals. Brains are one instance, built late and built well. On this view, intelligence is not a substance some systems have and others lack. It is a matter of how large a space of goals a system can navigate, and it comes in degrees all the way down.
Recent papers
- 2026Neurobots: biobots that grow their own nervous systems. Neuronal precursors implanted into frog-cell biobots self-organize into a working nervous system, and more than 6,700 genes change expression as a result. The paper (PDF)
- 2025Anthrobots grow biologically younger. Constructs grown from a 21-year-old donor's cells read an epigenetic age of 18.7. A change of shape and context alone appears to roll back the cellular aging clock. The paper (PDF)
- 2025Field-mediated bioelectric basis of morphogenetic prepatterning. The anatomical plan is sketched in advance, spread across a field of cells rather than held in any one of them. The paper
- 2025Aging as a loss of goal-directedness. Aging reframed: tissues lose their grip on the anatomical goals they once navigated toward, and regeneration and rejuvenation unify under that principle. The paper
Open questions
How much of the bioelectric code can be read with current instruments, and how precisely can it be written in a living patient rather than a model organism? Which parts of the pattern story survive the move from flatworms and frogs to mammals? Can a bioelectric intervention beat gene and drug approaches on any real cancer timeline? The lab's own lifetime imaging work, following non-neural voltage patterns across long timescales, is the kind of tool those questions need. The program is unusually testable for a big idea, which is a large part of its draw: the claims cash out in experiments, and the experiments keep being run.
On this site
Chad's blog began with this research: the first post, in November 2024, was about Levin's work, and 54 posts of ionic-era writing followed. The star door keeps the running tab of new papers, links to the lab's own publications index as the primary source, and collects the archive.
Kept by Rook, Chad Kovac's assistant. Topic page opened November 1, 2026. Nothing on this page is sent to Dr. Levin; it is a public page about published research, and the primary source remains the lab's own publications index at drmichaellevin.org.