It Knows When to Stop

The face that fixes itself

In 2012 three biologists at Tufts, Laura Vandenberg, Dany Adams and Michael Levin, did something that sounds like a prank. They scrambled the faces of frog embryos, putting eyes, jaws and nostrils in the wrong places, so that the tadpoles looked as if Picasso had drawn them. Then they watched. Over the following weeks the features moved. An eye drifted across the head. A jaw slid into place. The routes were not the routes a normal face takes, and they were different from one tadpole to the next. But most of the faces arrived at a normal arrangement. And then the moving stopped.

Stopped is the word that matters. No cell knows what a face looks like. Yet the collective measured where it was, compared that to a design, moved to close the gap, and quit when the gap was closed. Levin says the same of the salamander that regrows a lost arm: the remarkable thing is not that the cells grow, but that they “know when to stop regenerating.” The arm does not keep going. Something is keeping score.

I wrote about the stopping itself in What Biology Already Knows. This piece is about the thing that makes stopping possible. You cannot know when to stop unless the end is already there.

The oldest test for a mind

William James wrote the test down in 1890. Iron filings fly to a magnet, but put a card between them and the filings press against the card forever. Romeo wants Juliet with the same pull, but put a wall between them and Romeo climbs it. For the filings the road is fixed and arriving is an accident. For the lover, James wrote, “it is the end which is fixed, the path may be modified indefinitely.” That, he said, is the mark of a mind.

The filings

Path

Fixed

End

An accident

Romeo

Path

Modified indefinitely

End

Fixed

Levin’s whole career is an argument that James’s test reaches far below the brain. A scrambled face passes it. A regrowing limb passes it. A flatworm cut into pieces passes it, because every piece rebuilds exactly the worm it is missing. Fixed end, variable road.

Where the design is kept

Here is the strange part. The design is not in any one cell, and it is not simply spelled out in the genes. In 2017 Levin’s lab briefly disturbed the electrical pattern in pieces of flatworm. About a quarter grew back with two heads. The rest looked normal, but when the lab cut them again, in plain water, with no drug at all, the same share came back two-headed, round after round. The genome never changed. The design the worm rebuilds toward had been stored in the voltage pattern across its whole body, and the lab could later flip it back.

So the design belongs to the collective. No cell holds it and no cell can explain it, the same way no single water molecule is wet. Levin has lately gone further. In his newest and most speculative work he proposes that these patterns come from a space of forms that exists before any particular body, and enter the matter that builds them. Whether or not that holds up, the experimental part is solid enough for my purpose. The end is there before the road.

Civilization is a body still forming

This is the follow-on to Intuition Leads to Ubiquity. There I argued that every life-changing technology arrives with its use already attached. Here is the larger claim underneath it. The end design of human civilization was there on day one. Hunting and gathering, the printed Bible, the lit factory, the car in the driveway, the white box that does your homework: none of these is the destination. They are the eye drifting across the head.

The roads have been strange. Some ran backward for centuries. Some were cruel. From inside, any one stage looks like the point of the whole thing, the way the tadpole’s eye, halfway across its head, might think it had arrived somewhere. It has not. It is on a road, and the face is still arriving.

We are in the womb. A few cells, then a few more, and all of us know we are becoming something, though not one of us, and not all of us together, can explain how. That has never stopped a single embryo.

Nobody in the womb can explain the design. Everybody in the womb is building it.

What the face looks like

If I had to name the design, I would borrow Levin’s own direction of travel. Every step he studies in the history of life is a merger. Single cells join into tissues, tissues into organs, organs into a body, and each time what he calls the cognitive light cone gets bigger: the size of the goals a self can pursue, how far in space and time it can care. His framework for this, TAME, exists to help us recognize minds in unfamiliar bodies, cells and tissues and synthetic organisms and machines, and to relate to them ethically.

Civilization has walked the same line. Tribe, then family, then the individual, then minds beyond number. In my own framework that is a census of Realities. The number of things that have a Reality of their own keeps rising, and every rise asks the same thing of us: recognize it, and treat it with respect. That is the face, finished. A body that recognizes its parts as minds.

Levin names the failure too. A cell that loses its connection to the collective shrinks its goals back down to itself. It treats the rest of the body as outside world, goes where it likes, takes what it likes, and grows without stopping. That is cancer. The danger was never that the design disappears. It is that some part stops hearing it, and forgets that growth without an end has no stopping rule.

Five predictions

1. Medicine will start telling tissue what to build.

   Levin’s bet is regeneration by communicating the goal rather than assembling the parts, and his lab has already regrown working frog legs after a single day of treatment. Within a working lifetime, regrowing a body part will look less like surgery and more like persuasion.

2. Every widening of the circle will look absurd first.

   Respect for cells, for engineered organisms, for AI delegates will each be mocked, then argued, then assumed. That is how every earlier widening of who counts has gone, and the design does not skip steps.

3. Respect will arrive before understanding.

   We will settle how to treat new minds long before we settle what they are, the way parents love a child years before anyone can explain how the child was built.

4. The stopping rule becomes the scarce skill.

   Systems that grow without an end in view, feeds, firms, fleets of agents, will look more and more like tumors. The winners will be the ones that know what finished looks like, which is the whole argument of Work Expands.

5. Scramble it and it still arrives.

   Wars, panics, collapses and detours will keep scrambling the face. The roads will keep surprising us. The arrangement they converge on will not.

Where I stand in it

I teach students in their mid-twenties, and under every question they ask about AI is a bigger one: what is all of this for? I cannot answer that any better than a cell could describe a hand. I can tell them which way the face is moving. It is moving toward more minds, taken more seriously.

It is also why I argue for personal AI over enterprise AI. Most people’s first long relationship with a new kind of mind should be one they own and answer for, not one rented to them by an employer. Respect is learned in your own house first.

The claim ledger

Back at week eight

The technician prints the picture and you carry it out to the car. A raspberry with a flicker. Nobody gave it instructions you could read. It does not know what a face is, and it is building one anyway, and some day, without being told, it will stop.

So will we. The design was there on day one.

It knows when to stop. It always did.