Cells know what body to build. Michael Levin wants to learn how to change it

Twenty four hours set off a process lasting months

An adult frog does not normally regrow a lost leg. In an experiment by Michael Levin and David Kaplan, researchers placed a small device containing a drug mixture over the stump for just twenty four hours. That brief treatment initiated a process lasting months and produced a functional leg. The researchers did not construct the limb one cell at a time. They tried to give the tissue a signal, then let it do most of the work itself.

That result captures the idea driving Levin's research programme. Rather than treat the body as a collection of parts to be repaired separately, he studies how many cells coordinate a goal larger than any one of them: where an eye belongs, how many fingers to build and when growth should stop. The medical challenge, in his view, is learning to communicate with that system without specifying every step.

No single cell knows what a finger is

Cells outside the nervous system also maintain an electrical voltage across their membranes. Ion channels let charged particles move in and out, while connections between cells allow voltage patterns to spread through tissue. These signals are slower and subtler than those in the brain, but they appear during development, repair and the continuing maintenance of body structure.

Levin describes this coordination as collective intelligence. The term does not mean that each cell thinks like a person. It means that a group of cells can integrate information, retain a preferred state and respond to damage in ways unavailable to a cell on its own. In a 2024 paper, he presents bioelectric networks as the glue connecting local actions to large anatomical goals.

When a cell stops listening to the group

The connection with cancer follows from the same picture. A cancer cell keeps multiplying when the body does not need it. Levin and colleagues are investigating whether some of that behaviour reflects a loss of the electrical communication that links a cell to tissue level goals. In animal experiments, altering ion channels and restoring voltage patterns has suppressed tumour like behaviour even without removing the mutations inside the cells.

That is an experimental result, not a proven treatment for people. It raises an important possibility: changing a cell may sometimes require reconnecting it to the group's information rather than destroying it or correcting every gene separately. Laboratory studies have also detected unusual voltage patterns before visible tumours emerged in tadpoles, a line of work that could eventually contribute to earlier diagnosis.

Ageing as a loss of instructions for form

In a 2024 review, Léo Pio Lopez and Levin propose adding another layer to the biology of ageing. Alongside molecular damage, telomere shortening and metabolic change, tissues may gradually lose information that helps them maintain their intended form and function. They call this a loss of morphostatic information: a weakening ability to preserve structure while the material composing it is continually replaced and damaged.

This remains a model and a research programme. No bioelectric treatment has been shown to return a human body to a younger state. The proposal's value lies in the tests it suggests: measure how voltage patterns change with age, determine whether targeted changes improve tissue repair, and establish whether any improvement lasts without triggering uncontrolled growth.

The anatomical compiler Levin wants to build

Levin describes a future goal he calls an anatomical compiler. A person would specify a desired form, such as a replacement organ, and the system would translate it into stimuli that persuade cells to build it. This would not be a biological printer placing every cell in position. A closer image is a conductor giving a cue to an orchestra that already knows how to play.

The vision is distant. Researchers still cannot read most of the electrical language of tissues, and a result in a frog does not promise the same response in a mammal. Control over growth also needs dependable brakes. A system capable of starting regeneration must be equally capable of telling it when to stop.

The test that would move the idea into medicine

The next development to watch is not another metaphor about smart cells but a defined experiment: a known bioelectric signal, a target anatomy chosen in advance and a lasting result without abnormal growth. Repeated success in mammals would mark an important transition. Until then, Levin's work puts forward a measurable possibility: the body may not need complete construction instructions from us, only the right message at the right time.