The Rejuvenation Field

Anthrobots made from adult human cells were reported to be about a quarter younger, by an epigenetic clock, than the cells they came from. One explanation is that the cells read their own self-construction as an embryo’s, and reset some of their aging marks. Here every cell of a growing anthrobot weighs that evidence and winds its clock back, or doesn’t.

How the model works, and what to try

The body is the anthrobot from Growing Anthrobots: it grows from a single cell, then rolls around the dish. Here it is colored by each cell’s modeled epigenetic clock: cyan younger than the donor, gray the same, red older. The readout below the view gives the whole body’s average, as a percentage younger and in years.

Each cell keeps a belief about whether it is part of an embryo, and updates it every step from five kinds of evidence: how fast it is being built or rebuilt (construction), how curved the body’s surface is nearby, how many neighbors it has (contact), how fast the body moves, and how well an internal signal agrees with its neighbors’ (coherence). Cells share their belief with neighbors, and those that come to believe they are embryonic wind their clock back.

The controls. “Which evidence the cells trust” picks one of six versions of the model: balanced, or led by one kind of evidence. Each is calibrated so the standard round body ends 25% younger, the reported figure; the sliders let you set your own weights, which are then no longer calibrated. Sharing sets how much belief spreads between neighbors; donor age only changes the years shown. The view can show clock age, belief, tissue or any one kind of evidence. Body shape changes how the body grows and regrows (restart or cut to see it). The knife cuts or wounds the body, and “Cut in half” removes half so you can watch it heal.

Things to try. Let the round body grow under the balanced version and watch the readout climb toward 25%. Switch the body shape to elongated and restart: the balanced and motion-led versions now predict a much younger body (about 38–42%), the others only a little younger (about 29–31%). Then cut the body in half under the construction-led version and watch whether the healed region turns younger than the rest, and compare the balanced version, where it ends older. The belief view shows where cells are most convinced they are in an embryo.

Loading…

The research test: where do the versions disagree?

Nobody knows which evidence real cells would trust, so the model comes in six versions: one that trusts all five equally, and five that each lead with one kind. Each is tuned so the same standard round body ends 25% younger, the reported figure and the only fit to data. They agree there by construction, and the question is where they don’t.

A quality-diversity search grew several hundred bodies under different conditions: shape inputs, the size of the starting cluster, how thick the fluid is, and whether, when and where the body is cut. Each body was scored under all six versions at once.

Loading…

Some versions barely agree at all: across every body, the curvature-led and motion-led versions’ rankings of which bodies end youngest are almost unrelated. The curvature-led version favors small bodies; the motion-led one favors anything that keeps moving. Thick fluid, which slows the body, makes every version predict older bodies, the curvature-led one least. And a younger healed wound would point to construction or contact as the evidence cells use; the other versions predict the opposite.

Conclusion

If anthrobots really are younger than the cells they came from because their cells read their own construction as an embryo’s, then which evidence those cells trust matters a great deal. Six versions of that idea, all tuned to the same reported figure on a standard body, make very different predictions elsewhere. Some barely agree on which bodies should end youngest: one favors small, curved bodies, another anything that keeps moving, and thick fluid ages bodies under most of them. That makes the hypothesis testable with simple experiments rather than leaving it a story: sort anthrobots by size and shape and compare their epigenetic clocks, slow some down in thicker medium, or check whether a healed wound ends up younger, which only the construction-led and contact-led versions predict. Everything here beyond the one reported figure is invented to make the hypothesis concrete, and the search behind it was a quick one, so these are predictions to test, not findings about real cells.

The science

The body is the one from Growing Anthrobots: a neural cellular automaton that grows, heals and rolls. On top of it, each cell keeps a belief, how sure it is that it sits in an embryo, and a clock that starts at the donor’s age. Every step it gathers five kinds of evidence: how fast it is being built or rebuilt (construction), how curved the surface near it is, how many neighbors it has (contact), how fast the body moves, and how well a chosen internal signal agrees with its neighbors’ (coherence). Evidence nudges the belief; the belief leaks back toward “adult” and is shared with neighbors; cells that believe they are embryonic wind their clocks back, down to a floor.

What’s real

Anthrobots self-assemble from adult human airway cells, and a follow-up study reported they are about 25% younger by an epigenetic clock than their source cells.
The age-evidencing hypothesis, that cells read signs of embryonic development and reset aging marks, is an explanation Levin and colleagues have proposed.
Epigenetic clocks estimate age from DNA methylation patterns.

What’s simplified

Everything else is invented to make one hypothesis concrete: the five kinds of evidence, the belief rule and the clock.
The calibration uses a single number, the 25% figure, applied to a stylized grown body.
The search is a quick run on one trained growth rule; the disagreements are between versions of a model, not predictions about real cells until tested.