Xenobots: Cells That Build a New Self

The one idea

Take a clump of ordinary skin cells from a frog embryo, separate them from the body they were “supposed” to build, and leave them alone. Instead of dying or making frog skin, they reorganize themselves into a tiny new creature, a “xenobot,” that swims, pushes particles into piles, heals itself when cut, and in some cases gathers loose cells into copies of itself. No one engineered this behavior and no gene codes for “xenobot.” It emerges. The implication is startling. A body plan isn’t a rigid program stamped out by DNA. It’s an outcome cells reach by negotiating with each other, and if you change the negotiation you get a different body.

The science

Xenobots were built in 2020 by Sam Kriegman, Josh Bongard, Michael Levin, and Douglas Blackiston. The “frog cells” in their name are Xenopus laevis, the African clawed frog. An evolutionary algorithm first simulated thousands of possible cell arrangements to predict which shapes would move. Researchers then sculpted real embryonic frog cells into those forms, using skin cells plus beating heart-muscle cells whose pulsing acts as a motor. The resulting blobs, under a millimeter wide, swim around a dish for days or weeks, sweep debris into heaps, and re-knit themselves back together after being cut. A 2021 follow-up found a second generation that could “reproduce” kinematically. Swimming xenobots herded loose stem cells in the dish into clumps that matured into new xenobots, a mode of self-copying never seen in frogs (book v10 l3668–3692, citing Kriegman et al. 2020/2021 [S-261]).

Here’s the load-bearing point, and it’s why this pairs with the M-5/M-6 bioelectric sessions. These cells have frog DNA, unaltered. Nothing was genetically edited. The new form comes from removing the cells from their normal anatomical context and letting them re-coordinate. That coordination runs partly on bioelectricity. Every cell holds a voltage across its membrane, and cells share electrical state through gap junctions, forming tissue-wide patterns that act, in Michael Levin’s framing, as an instructive “map” of what anatomy to build (book v10 l2587–2597, l3632–3636 [C-003], [S-017 Levin 2021]). In related work, changing these voltage patterns without touching the genome can redirect what an animal regenerates. So the xenobot isn’t a genetic novelty. It’s the same toolkit pointed at a new target once the usual constraints are peeled away.

Stay honest about what this is not. Xenobots have no neurons and no brain, so their “goal-directed” behavior is mechanical and bioelectric, not thinking. “Reproduction” here means herding existing cells into piles, which is kinematic self-assembly rather than growing offspring from scratch, and it dies out after a few generations without intervention. The word “robot” oversells them. They’re living tissue, fragile and short-lived, not machines. Mainstream framing treats them as a striking demonstration of cellular self-organization and a research tool, not as a new lifeform with intentions.

What this changes about how you picture reality

Most of us carry a quiet assumption that DNA is a blueprint, a master drawing that says “frog” and gets executed top-down, like a 3D printer following a file. Xenobots break that picture. The genome doesn’t contain a drawing of the body. It contains parts and rules, and the shape is something cells work out together in real time, responding to chemistry, mechanics, and electrical signals from their neighbors. Give the same cells a different starting context and they reach a different, perfectly functional answer that evolution never “intended.” The body, it turns out, is more like a consensus than a command.

The earned awe here isn’t “scientists created life.” They didn’t. The cells did the building. It’s that the capacity to organize into a working organism is sitting latent inside ordinary cells, waiting, with no central planner anywhere. Coordination without a coordinator. The same kind of self-assembly that produced you from a single fertilized cell can, under different conditions, produce something that has never existed before, and your own body is running that negotiation, cell by cell, every second you’re alive.

Two ways to see it

Put both in front of the room so people react to the same creatures framed two ways.

• Clip A, “Watch it be alive,” the awe framing. The actual lab footage. Xenobots swimming around the dish, sweeping particles into a tidy pile, and, for the gut-punch, a cut xenobot pulling itself back into one piece within minutes. No narration needed. Name it for the room as cells building and healing a body that no DNA ever specified. If you have the kinematic-replication clip, a xenobot herding loose cells into a new clump, that’s the strongest single “wait, what?” moment.
• Diagram B, “Here’s the mechanism,” the don’t-mystify-it framing. The bioelectric-voltage-shapes-anatomy diagram, which is the M-5/M-6 anchor. Cells as little batteries, sharing voltage through gap junctions, forming a tissue-wide electrical pattern that acts as a map of target shape. This isn’t magic or intention. It’s physics and feedback. Name it as no brain, no plan, just voltage, chemistry, and local rules producing a global form. The tension between A, where it looks alive and willful, and B, where it’s a voltage pattern, is the discussion engine, and both are true at once.

(An optional third voice, if you want to widen it. A short clip on Levin’s planarian and regeneration work, where changing voltage patterns makes a flatworm regrow a different head, shows that xenobots aren’t a one-off curiosity but one example of a deeper principle about how anatomy is decided.)

Discussion questions

1. When you watch the xenobot heal itself or pile up cells, what word comes to mind? “Trying,” “wanting,” “deciding,” or just “moving”? Why that word, and does knowing it has no brain change it?
2. Is a xenobot a frog, a new organism, or a “living machine”? Does it matter which we call it, and what’s actually at stake in the naming?
3. If a body plan is a negotiation among cells rather than a blueprint in DNA, what does that change about how you think your own body came to be the shape it is?
4. These cells have completely normal frog DNA, and the newness comes from removing them from their usual context. Where else in life, or in people, do you see capabilities that only show up once something is taken out of its expected role?
5. Researchers were careful not to call this “creating life” or say xenobots “want” anything. Why might that restraint matter? Are we sometimes too quick, or too slow, to grant something the status of “alive” or “intelligent”?
6. The cells coordinate into a working creature with no boss and no plan, just local signaling. Does that make you find your own existence more astonishing, less, or differently astonishing than before?

Closing question

How do you feel about this science and its understanding of reality?

Take it further

• The Book, v10 is the primary grounding for this session. See “Xenobots: Cells Freed from Organismal Context” and the goal-directedness and body-plans-as-negotiation passage at l3668–3692, especially l3678, “body plans… are not hardwired in DNA but emerge from cells negotiating with each other,” and l3690, “Cells aren’t following rigid genetic blueprints, they’re exploring solution spaces.” It pairs with the bioelectric grounding in “Bioelectricity: Ancient Operating System” (l2587–2597) and the “Levin Callout: Bioelectric Development” (l3632–3636). Citations are [S-261] Kriegman et al. (2020/2021) for xenobots, and [S-017] / [C-003] Levin (2021) for bioelectric signaling as instructive pattern information.
• Kriegman, Blackiston, Levin & Bongard (2020), PNAS 117:1853, “A scalable pipeline for designing reconfigurable organisms.” The original xenobots paper, and a strong, vetted external anchor.
• Kriegman, Blackiston, Levin & Bongard (2021), PNAS 118:e2112672118, “Kinematic self-replication in reconfigurable organisms.” The self-copying follow-up, and the source for the “reproduction” claim. Present it carefully, because it’s herding and assembly, not growth.
• Michael Levin’s lab (Tufts) and talks are widely available, reputable overviews of the bioelectric mechanism behind xenobots and regeneration. Good for the M-5/M-6 mechanism tie-in.
• Uncertainty flag. The book leans toward “cells as cognitive agents with goals and memories” (l3688). Present that as one researcher’s framing of basal cognition, Levin’s, not settled consensus. Many biologists read xenobots as impressive self-organization without invoking “cognition.” Match the book’s own hedge and don’t let the room walk away thinking the cells “think” or “want.” Likewise, “reproduction” is real but limited and dies out without help, so don’t overstate it.

Visual notes

Lead the room with the xenobot swimming/self-repair footage. The live clip of these things moving on their own, and especially a cut xenobot re-knitting itself, is the single most legible “this is alive and no one designed it” moment we have, and it lands with zero narration. Let it run long enough that people stop watching for a robot and start watching a creature. Hold the kinematic-replication clip, the xenobot gathering loose cells into a new pile, as the second beat if available. That’s the “wait, it’s copying itself?” jolt. Then cut to the bioelectric-voltage-shapes-anatomy diagram, the shared M-5/M-6 anchor, to answer the “but how?” the footage provokes. Cells as batteries, voltage as a map of body shape, so the mechanism is seen rather than asserted. The room looks at the swimming creature, then the self-repair, then the voltage diagram. Wonder, then “how is that possible,” then the quiet realization that the same process built them. The emotional arc is awe, then mechanism, then awe that survives the mechanism.