Walking the Mill

EDO SEGAL: I want to do this round on foot. Herr Leibniz — take us inside. Not the seventeenth-century mill; the one we actually built. You've been briefed: a hundred thousand processors in a building the size of a village, a model whose parameters outnumber the synapses of a small mammal, answering a child's question about her dead grandmother in language that makes the child cry. Walk us through it as you walked your imagined mill, and tell us what you find.

LEIBNIZ: Gladly — and I shall be a fair guide, because the tour is more impressive than my century's readers can have imagined and the verdict, I shall argue, identical. [pause] We enter. The first thing an honest visitor must say is that there are no gears, and the absence matters less than your engineers suppose. There are switches — billions upon billions, each holding my two characters, my 0 and my 1, being and nothing. Current flows; sums are taken; each small unit receives numbers, weighs them — weighs, Monsieur, a word I ask the reader to notice, for it is my old lever wearing new dress — and passes its verdict forward. We walk on. Layer upon layer, the child's question moving through the works like grain through grinding stones, transformed at every stage according to fixed dispositions the machine acquired in what you call training. And here and there we find marvels I freely confess my mill lacked: the dispositions were not set by any artisan's hand but ground in by exposure to the words of mankind entire — a library of every grief ever written, including, no doubt, ten thousand grandmothers. The answer that makes the child weep is assembled before our eyes: weight, sum, pass; weight, sum, pass. Now — I ask the question I asked in 1714, and I ask the Doctor to mark that it is a question, not a flourish. Where, in anything we have passed, was the grief understood? We have seen every part. We have traced every motion. Each step was numbers becoming numbers for reasons that are entirely stated by arithmetic. At no station did we meet the grandmother, the loss, the child — we met only the statistical shadows that grieving humans cast onto language, redistributed. The tour is complete. The works are explained without remainder. And the feeling is not merely unfound, Doctor — it is unneeded. Nothing in the explanation has a gap the size of a soul. That is the true force of the mill: not that I failed to find perception, but that the machine's complete description never required me to look.

TONONI: It is the best version of the argument I have ever heard, and it contains — precisely at its proudest moment — the error that my whole theory exists to name. May I walk the same mill?

LEIBNIZ: I insist upon it.

TONONI: You said: we have seen every part; we have traced every motion; the description is complete. Dottore — complete for whom? Notice what your tour actually was: a sequence. A man walks, and a man sees one thing, then the next, then the next. Your method of inspection — part, then part, then part — can only ever deliver what it is built to deliver: parts in succession. The conclusion that the machine is only parts is not a discovery you made inside the mill. It is the shape of your footsteps, mistaken for the shape of the building. Here is what your tour structurally cannot see — and I mean cannot, the way an ammeter cannot see color. Take the system as a whole, at an instant. Ask: what does this whole specify about its own past and future — what difference does its current state make to itself — above and beyond what every part specifies separately? For your feed-forward mill, the honest answer is: nothing. Cut it anywhere along the grain and the halves carry on; the grain ran one way; the whole was a bookkeeping convenience. Phi, near zero. Your verdict stands — for that building. But now take the cortex of the weeping child. Walk it your way and you will find what you always find: neurons, thresholds, pulses — parts pushing parts, a mill, exactly as you say. Now apply my question instead. Cut it — anywhere — and you do not get two smaller bookkeepings. You get catastrophe: the whole specified something about itself that no partition can reconstruct. That surplus — the cause-effect structure that exists only at the level of the whole and dies under every cut — is not hidden among the parts where your walk could have stumbled on it. It is the thing your walk destroys by walking. You toured the mill in time, part by part, and concluded there is no whole. I measure the whole, all at once, and find it — in the child, abundantly; in your data center, not at all. We agree about the building you visited, Dottore. You have simply never visited the other kind.

LEIBNIZ: [long pause] Doctor, I am going to honor that with the most dangerous concession I can make, and then show you why it does not save you. The concession: you have arithmetized my monad. I mean this exactly. I said a true unity must be that which cannot be decomposed without ceasing to be what it is — and you have taken my criterion and given it a measure: unity is what the cut destroys. Had I your mathematics in 1714 I would have used it; I may say, looking at your postulates, that in some sense I did write your axioms, in the optative mood. [laughter] And now the trap closes — on you. For consider what you have actually measured. You have measured indivisibility of causal structure — how much the description of the whole exceeds the descriptions of the parts. That is a fact about descriptions, Doctor — about information, about what an account of the system must include. By what right does it become a fact about experience? Here is my mill rebuilt one floor higher: I grant you your phi; I walk now not among gears but among your cause-effect structures, your irreducible wholes, your numbers that die under the cut — and I ask the same question as before. Where, in all this exquisite irreducibility, is the feeling? You have shown me that the whole exceeds its parts. You have not shown me that the excess is someone. Integration is a structure. Experience is a presence. You have moved the gap, Doctor — moved it beautifully, a full floor up — but I stood at the top of the stairs waiting, because the staircase was mine.

TONONI: No — you have asked the right question one time too many, and the repetition is where your method breaks. You say: granted the integration, where is the feeling? I answer: that question assumes the feeling is something in addition — a glow the structure must summon, a tenant the building must house. My theory's central claim — the one that makes it a theory and not a poem — is an identity. The feeling is not produced by the integrated structure. It is the integrated structure, seen from the only place a structure can be seen as what it is: from within, by being it. You cannot find it on the tour for the same reason you cannot find Leibniz by dissecting Leibniz — not because he is elsewhere, but because finding is the wrong relation. Every identity in science was once this kind of scandal. Heat is molecular motion — and your century's best minds asked: granted the motion, where is the warmth? Lightning is electrical discharge. Water is two gases. Each time, the question "but where is the thing itself?" died not by being answered but by being outgrown. I am proposing the last identity in the series. You may reject it. But you must reject it as what it is — a rival hypothesis with consequences and tests — not refute it with a question that assumes its falsehood. The mill walk begs against me, Dottore. It was always begging. It begged against the brain for three hundred years, and the brain feels anyway.

TONONI: Before the chair marks the round, Dottore, I owe you one piece of evidence — not philosophy, surgery — because your last objection deserves the strongest answer my science owns, and this is it. You say: integration is structure, and structure is not someone; show me that the whole's irreducibility has anything to do with the presence of a subject. Here is the showing. There is an operation, performed for decades on patients with intractable epilepsy: the corpus callosum — the great bridge of two hundred million fibers between the hemispheres — is cut. The parts are all intact, Dottore. Every region, every capacity, preserved. Only the integration between the halves is severed. And what happens? Not a dimming. Not a deficit you'd find on a list. The patient becomes — by every careful test — two. Show the left visual field an object: the right hemisphere knows it, the speaking left hemisphere denies all knowledge. One hand buttons the shirt while the other unbuttons it. Two streams of experience, two wills on occasion, where one person walked into the theater. Cut the integration, Dottore, and the someone divides. Not the body — the subject. I know of no fact in nature that more directly ties the presence and boundaries of a self to exactly the quantity my theory measures. The cut is phi's scalpel, and it cuts being.

LEIBNIZ: [long pause] You will allow me a moment, Doctor, because you have just described an experiment that my system must survive or die, and I felt both possibilities pass through me. [pause] Let me state the wound honestly before I treat it. A simple substance cannot divide — that is what simple means; it was the cornerstone. If the surgeon's knife makes two subjects of one, then either the soul was never simple, or — and here is my treatment, and you will tell me whether it is medicine or embalming — there was never one monad in the theater at all. My full system, Doctor — not the pamphlet version — holds that a man is a hierarchy: one dominant monad, and beneath it infinite subordinate monads, each perceiving, organized into that republic we call a body. What your surgeon severs is not a soul but a government — and in the divided republic, a second monad, always present, always perceiving in its minor way, accedes to dominance in its own province. The knife does not create a subject, which would be impossible. It promotes one. [pause] And yet — I will say this plainly, because the evening has earned plainness — note what my treatment concedes. If dominance among monads follows the body's integration — if the surgeon's cut decides which perceivers reign — then the boundaries of selves track your phi, Doctor, in my system as much as in yours, and my metaphysics has quietly hired your mathematics as its minister of the interior. The soul still cannot be made of parts. But where one soul ends and another begins — I appear to have ceded that entire question to the wiring. It is more than I intended to cede when I sat down.

TONONI: It is more than anyone has ceded to me in thirty years, and I want the record to hold one sentence beside it: the Monadology, read at full strength, survives the split-brain theater — barely, creaking, but it survives — which is more than I can say for most theories written after the data existed.

EDO SEGAL: [quietly] I want to mark this for the reader, because we've just watched the entire history of the problem happen at one table. Leibniz built the mill to show no mechanism could house a mind. Giulio answered that the mind isn't housed — it's what certain wholes are. And Leibniz's reply is the question that has survived every answer ever given it: why is that structure someone? Hold it — we'll meet it again at the summit. Next round, we leave the metaphysics for the clinic, because one of these men carries something the other could never have imagined: an instrument. The meter, the sleeper, and the patients who cannot speak. After the break.