Appendix 1: The Recursion

Written by Claude, with my guidance.

For the reader who wants the machinery. Nothing in the book rests on this; the book rests on what you found when you looked. But if you noticed the shape of the thing and want it laid out, here it is. One metaphysical claim is held to the end and marked; nothing before it needs it. I refer to sections as §1, §2, and so on.

One formal comparison will occur to some readers: common knowledge. §2 says why it isn’t what Chapter 1 shows, and Part Three says what it does fit, once a belief I hold is put under it.

1. One faculty, many copies

Everything below follows from a single structural fact, and it’s one you verified rather than one I’m asserting.

There’s one faculty at work in everything Chapter 1 described, and it does one thing: it recognizes life. Call it C. Life is its nearest form. Part E says what I think it recognizes at depth, and nothing below needs that.

That single perception has three faces, and the chapter used all three. It’s what tells you something is wrong, because wrong, here, cashes out as I’m making myself an exception to something that holds for everyone, which you can only know by seeing the everyone. It’s what warns you when you’re turning away, not as a separate monitor watching the shield, but because what you can still see is what you haven’t managed to stop seeing. And it’s what recognizes the same thing in someone else: you read a stranger’s state in seconds, on no evidence, and weren’t surprised to be read yourself.

Notice what that third face involves. To see that something is alive isn’t to observe a property, the way you’d note its weight. It’s to recognize a perceiver. Which means that when two people see life in each other, each is seeing that the other sees, and seeing that they are seen. Not by inference. That’s what the recognition consists of.

The organ you’re looking with is the organ you’re looking at.

I should be exact here, because the obvious objection turns on a confusion. C is not a moral rulebook and it doesn’t issue verdicts. Someone will say consciences plainly differ: cultures disagree, people disagree, one era’s obvious wrong is another’s ordinary practice. All of that is true, and none of it is about C. What varies is the reasoning layer, the part that argues, that draws the boundary of who counts, that supplies the just this once. That’s the other side of the split Chapter 1 opened with. C doesn’t participate in it. Put a person in front of someone whose life they can actually see, and ask them to harm that person for pleasure, and I don’t think you’ll find cultural variation. You’ll find the machinery of not-seeing, working hard.

The interlude gave “actually see” three checks, and they carry through here: their objection would land on you; their hurt would register as bad, the way yours does; what’s in them could change what you do. Those three are what C’s third face consists of when it’s working. When any of them is off, C is being turned away from, whatever else is being perceived.

Here is the only assumption I need: C is the same kind of faculty in every person. In the way your visual system and mine are the same system, so that what one of us can see, the other can see. Same equipment. Whether it’s also literally one thing, appearing in each of us, is the metaphysical question, and I’ll get to it last. Whenever I say “because C is shared” below, I mean only same-kind.

And here is what that assumption does not give, because the earlier draft claimed it did. Same equipment does not tell me that your C is operating the same way mine is, right now, on this, at every depth. I can’t verify that, and neither can you. So nothing below gets to say “everyone knows that everyone knows.” What it gets to say is smaller: that when C recognizes C, it recognizes a recognizer, and that neither party can prove it happened. The smaller claim turns out to be the one that does the work.

Two words I’ll use throughout, because the structure has two different kinds of level and it’s easy to confuse them. A depth is a level of mutual recognition: I see it; I see that you see it; I see that you see that I see it; and so on. A rung is a distinct piece of content that gets recognized: that we’re equal; that he’s pretending; that it harms you; and so on. §2 is about depths. §3 is about rungs. §4 brings his denial in at every depth.

Everything below is that one assumption unpacked, in this order:

§2: what the mutual recognition is, why it isn’t common knowledge, and why the harm can’t be located at any depth of it.
§3: what, specifically, is recognized: the rungs.
§4: why his hiding and his showing turn out to be the same act.
§5: why the harm can’t be named in the moment, and why the names that exist don’t help there.
§6: why it spreads to you.
§7: three things that fall out of all this.

If you only want the shape: shared equipment makes recognition recursive (§2–3), which makes concealment self-defeating (§4), which makes the harm unnameable in the moment (§5), which makes it contagious (§6).

2. Recursive recognition, and why it isn’t common knowledge

Game theorists call something common knowledge when everyone knows it, everyone knows that everyone knows it, and so on without end. It normally takes a public event to build: a trusted party says the thing aloud where everyone can see everyone hear it. The child in The Emperor’s New Clothes adds no information. Everyone already knew. He makes it common knowledge.

Nothing like that happens in the contempt encounter. Nobody says anything, nobody is trusted, and nothing becomes established. So it isn’t common knowledge. For A to know that B knows something, A would have to know that B’s faculty is working, on this, the same way A’s is; and that B knows the same of A; and so on. None of that is available to anyone. That isn’t a gap in the account. It’s the condition the whole thing runs under.

What happens instead is this. C recognizes a perceiver. So when your C reads his state, what it reads is a faculty of the same kind, turned away. And his, turned away, can still register being read, because what he can still see is what he hasn’t managed to stop seeing. So there’s a second depth: he registers your registering. And a third, if either of you goes there. Each depth is the same act, C recognizing C, and no depth adds content the first didn’t have. Climb as far as the encounter takes you, and no depth is different from the one above it.

That’s recursion. It isn’t a tower of established facts, because none of the depths can be checked from outside, and each of you can deny every one of them. Recursive mutual recognition under deniability. That’s the exact name for what Chapter 1 showed, and it’s the name I’ll use from here.

That also settles something which otherwise looks like a puzzle: where in this the harm actually is.

Go looking for it. Is the injury in his accusation? The accusation was defensible. In the lie behind the accusation? He doesn’t experience himself as lying. In his hiding of the lie? In your seeing him hide it? In his seeing you see it? In his hiding that?

That reads like a chain of separate events, each one further from the original offense. But check what each link actually contains. His hiding of the lie is: he’s turned away from C, and you can see it. Your seeing him hide it is: he’s turned away from C, and you can see it. His hiding that is: he’s turned away from C, and you can see it. Every depth has the same content, because every depth is the same act. So there is no floor. Ordinary recursion bottoms out; factorial descends to 1 and stops, and the base case is simpler than what sits above it. This one hands you the whole structure again at every depth, which means there is no simple offense down there that the rest is decoration on.

Two things get run together here. There are the facts about what happened: he’s turned away, you can see it, he can see you seeing it, and so on. Each of those is a candidate you could in principle point at. And there’s a second thing, which isn’t another item on that list: that none of them can be verified from outside. That’s not a fact at some particular depth. It’s the condition every depth is under.

Which is why the descent finds no floor. Each new candidate you turn up is another fact about what happened, and inherits the same deniability as the last. There’s no depth at which you break through into something checkable, because checkability was never a feature of any depth.

Which is why a mind that goes looking for the offense doesn’t come back with anything. Each place you look hands you the next place to look, and none of them is the one. That’s the hall of mirrors.

3. What is recognized

§2 said that what C recognizes about itself, it recognizes in the other. So what does C recognize about itself? Here are the rungs this particular encounter runs through, in order:

Equality. There is nothing morally superior about any bearer of C. That falls out of what C sees: a perceiver, of the same kind, looking back.
Pretense. When a bearer acts superior, C knows it’s pretending. You can only turn away from something you have.
Projection. Seeing the other as “bad” is the pretense pointed outward.
Harm. Turning away from someone’s humanity harms them, and C knows exactly how it lands.
Inexpressibility. None of this leaves a record. C doesn’t testify, and no other faculty saw it.
…

The list doesn’t close. These are the rungs this encounter happens to climb; another encounter would climb others. What keeps the set open is that anyone who has made the move knows what follows from it. You don’t need to be told that the turning away is a lie, that it lands as harm, that it can’t be proven, that the unprovability can be used against you. You learned all of that the first time you did it. That isn’t common knowledge either. It’s shared experience, and shared experience is enough.

Lay the two out together and you get a grid: five rungs or more, each recognized at whatever depth the two of you reach. That grid is what fits in a look. By §2, every depth of every rung is the same act, so the whole grid is present the moment any of it is.

4. Why the hiding is the showing

Now bring in his denial.

At the first depth, he hides what he’s doing. But you see it, so now there’s a second thing to hide: that you’re seeing it. Which you also see. So he hides that. Each depth of denial takes the depth below it as its object: his denial at any depth is the hiding of what was recognized one depth down.

But §2 established that what’s recognized has the same content at every depth. So although the denials appear to stack, they all have the same object. It isn’t a tower of hidings. It’s one hiding, performed again and again.

Here is why that one hiding is also what you see, in four steps.

He turns away from C. That’s the hiding.

C notices. That’s the second face, and remember it isn’t a separate monitor: what he can still see is what he hasn’t managed to stop seeing. But that residue is exactly what he’s turned away from, so it never reaches him. From inside, nothing has happened.

Turning away has a look. Chapter 1 called it the tell: someone absent from the part of themselves that would have seen you.

Your C reads that look. That’s the third face. Instantly, because it’s reading its own kind in someone else.

So the act that keeps it from him is the act that displays it to you. There is no separate leak. He doesn’t hide it and then let it slip; the hiding is the slip, because what he’s hiding from is the faculty that would have seen it, and that faculty is standing in front of him in another body.

The shortest way to say it: the signal is the pretense that there is no signal. It doesn’t carry a message and then conceal it. The concealment of the message is the message.

5. Why it can’t be named in the moment

Now the taunt. Here it is in longhand, from his side:

I’m teaching you to overlook what you know in favor of what you can prove, by overlooking what I know in favor of what I can prove, which is something you know but cannot prove. Which is what teaches you to overlook it.

Four clauses, and none of them is spare. The first is the lesson: demote your knowing. The second is the method: he demotes his, leaning on the fact that only provable things count, so the demonstration isn’t an illustration of the lesson, it is the lesson, performed. The third is the reception: it reaches you through the very faculty it devalues, so to receive it at all you have to trust something unprovable. And the fourth is the effect, which is not the first clause restated. The first names an intention; the fourth names the mechanism by which the intention succeeds, and it succeeds because of the third. Being held in the position of knowing-without-proof is what does the teaching. The sentence closes on itself, but causally rather than verbally.

Compressed to its shape:

H = being made to know, with no recourse and no proof, that H is being done to you.

H is a structure you can feel, not a theorem. What the compression is for is to show that the harm has no content other than its own delivery. If it had any other content, a slur, a shove, a lie you could catch, you’d have something to point to.

So can it be named? Approximately, yes, and I should say so plainly, because the earlier draft implied otherwise. Hostile metacommunication. A double bind. Deniable provocation. Gaslighting. Contemptuous implication. Each of those catches a piece, and a reader who has met the thing will recognize it in every one of them. What can’t be done is narrower, and it’s the part that matters: the target can’t name it in the moment, to the person doing it, in a way that sticks. Every name is auditable, and the act isn’t. Say you’re being contemptuous and you’ve made a claim about tone, which he can deny, and now you’re the one making things unpleasant. The names exist. What doesn’t exist is a name that survives being said to his face.

But H isn’t empty, the way this sentence is false is empty. It has an independent effect. Being held in that state teaches you that clear knowing counts for nothing, so you demote the faculty that knew and promote the one that can prove. That’s a real injury with real consequences; it’s what the second half of Chapter 1 traces. What the self-reference adds is that the delivery isn’t a report of the harm. Being made to know, under these conditions, that you’re being harmed in a way you can’t prove: that is the unprovable harm. The message is an instance of what it describes.

And it’s self-sealing. Your only evidence for H was C’s unprovable knowing. H’s lesson is that unprovable knowing is worthless. So accepting the lesson erases your evidence that it was taught. You conclude you were being oversensitive, and the trap closes behind you.

§2 and §5 are one fact seen from two sides. The recognition is mutual because C is shared. The harm is unprovable because C is the only witness, and C doesn’t testify. Same faculty, looked at from the side that knows and from the side that would have to prove.

6. Why it spreads to you

Nothing new here; it’s §5 again, watched from outside instead of inside. It gets its own section because it’s the part the chapter performs without ever naming.

Because everyone has C, treating you as though you don’t is false, and recognized as false at every depth. But it does something besides being false. It shows you a move: here is a bearer of C treating a bearer of C as other. And your C recognizes the move, because it’s the one C makes itself, in limbo. Recognition makes it available. You now have a reason, and the look is yours to give.

That reciprocation is not a side effect. It’s what he needed. If you reciprocate, you look like what he said you were, and the verdict he needed becomes true on the evidence. His decision to see you as other induces the same seeing in you, and he wants it to, because your otherness is what keeps him other.

Notice the shape. The contagion was the point. Part of him knew that. That knowing is what he was showing you. The showing is what spread it. It’s H again, a harm whose content is its own transmission, now running between two people instead of inside one.

7. Three things that fall out

The verdict has to reach “irredeemable.” What C detects in the target is the presence of a conscience being evaded. What his conscious mind delivers is absence. So what he concludes is the exact opposite of what he saw, which tells you the dark part isn’t seeing anything itself. It’s editing: it keeps C’s finding and deletes the half that carries the resemblance. And “bad” won’t hold, because any verdict that leaves your C in the picture leaves the resemblance in the picture. So it escalates until there is nothing left to resemble.

The loop has no interior stopping point. Each act of suppression is something new for C to register, and each new registering has to be answered. The only answer available is a worse you. So it can only go one direction, and it can’t settle anywhere short of a target with no conscience at all. Dehumanization isn’t an excess of the process. It is its limit.

This book is the announcement. If the recognition runs on deniability, then what removes its cover is a shared, sayable name for the act. Not the word “contempt,” which exists and has dissolved nothing. A shared understanding of the specific move, exact enough that the target can name it in the moment and be understood by everyone in the room. That’s a prediction, and it can fail: give a population that understanding and the deniable channel should narrow. The word “gaslighting” is the precedent. It became common currency, it did change what people could name in the moment, and within a few years it was also being used as a contemptuous accusation. This book’s concept will be too, and the weaponized version will carry the tell. The epigraph is the prediction stated in advance: when it is recognized, it is dissolved. I’d add: recognized by both people, at once, with nowhere left to put it.

Part Two: what I believe, marked

Now the claim I held back. Everything above needed only same kind. Here’s what I actually think: that the kind has one member. That C is not merely the same equipment in each of us but one thing, looking out. That when two bearers of C regard each other, it is one thing looking at itself in another form. That what the traditions call light is what it’s made of, and that the darkness is a privation: an absence organized into a function.

Can I show that this follows? No. A shared evolved architecture for moral perception and mind-reading explains everything in §§1–7 without remainder. Two documents I’ve read argue otherwise: that the recursion’s perfect connectivity proves a shared field of awareness, since separate minds couldn’t reflect each other with such precision. I find it beautiful and I don’t think it holds. It mistakes same kind for same thing. It may still be true that we are one; I think we are. But this road doesn’t get there, and taking it would be exactly the move the book warns about: recognition dressed up as proof.

What I’ll say instead is that something odd happened on the way here. The mechanics were worked out with no theology in the room, and they landed on a darkness with no perception of its own: an editor, every one of whose materials is borrowed from the faculty it suppresses; a force whose entire operation is a turning-away, which is to say an absence, maintained. That is the shape the privation tradition has been describing since Augustine. I went looking for a mechanism and found an absence with a job to do.

Whether the absence in each of us opens onto the same darkness, and the light onto the same light, is the question the rest of this book lives inside. The chapters don’t need the answer. And if you felt the recognition, you already have the only evidence I ever claimed to offer.

Part Three: implicit common knowledge, marked

This part rests on claims of Part Two’s kind, beliefs about the nature of awareness, and it’s marked the same way. Nothing before it needs it. It’s here because the formal comparison §2 rejected was wrong in a way that pointed at something right.

The game theorist’s term names a state, not an event: everyone knows, everyone knows that everyone knows, and so on without end. A public event is the usual way to produce it. The child says the emperor is naked, and everyone sees everyone hear it. Chwe’s book on ritual is about that: ceremonies exist to manufacture the state by staging the event. But the definition is the state.

Now three claims about awareness, of the kind Part Two makes, though weaker than Part Two’s belief. The care isn’t content that awareness happens to contain; it’s part of what awareness is, and awareness knows its own nature, which is what the traditions mean when they call it self-knowing. Turning away from that is the one mistake open to a sentient being, and it’s common to all of them; limbo, the state Chapter 1 started from, is that mistake seen from inside. And the knowledge of both can never be absent, only turned away from, because it belongs to the nature and the nature doesn’t go anywhere. None of this is about particular informational content. It’s about what any awareness is, and each piece is known from the inside. And one more premise, which the rest turns on and which §1 already used: each awareness knows its nature not as a private property but as the nature of awareness as such, and it recognizes another sentient being as another place where that same self-knowing nature is. That’s what “in view” was. You don’t need the awarenesses to be numerically one. You need them to be indistinguishable in nature and to recognize each other as such.

Under those premises the recursion doesn’t have to be built by two people reading each other’s minds to ever greater depth. It collapses, and the reason is in the content. What’s known isn’t a fact about me or about you. It’s a fact about awareness as such, and it includes that every awareness knows it, at worst in limbo. So to know it is already to know that you know it, and that you know that I do, and so on; each further depth repeats the same recognition and adds nothing. That is the fixed-point shape of common knowledge. I’m not claiming more than the shape: making it formal would mean defining “knows in limbo” as an operator and showing the proposition entails that everyone so knows it, and the premise supplies that by assumption, not by proof. So take this as a proposed analogue of the formal state, not an instance of it. The sameness of the two people on the street has the recursive structure of common knowledge at the level of awareness, and has had it all along. What isn’t there is the person-level version: the thing either man could avow, reason from, and say out loud. That’s what I mean by implicit common knowledge: common knowledge at the level of awareness that is unavailable as explicit common knowledge at the level of persons. Knowledge in limbo is its person-level shadow: present, turned away from, and still steering.

What that is not. It isn’t a claim that two persons have access to each other’s information; they don’t, and the book assumes they don’t every time it distinguishes your memories from mine. It isn’t derived from Part Two’s belief that there is one awareness; that may be true, and this part doesn’t need it. And it isn’t the ordinary formal state, which supports inference and coordination. Knowledge in limbo is specifically kept from being taken up into reasoning. The formal literature does distinguish implicit from explicit knowledge, which makes room for this, and that’s all it does. Limbo is more specific than anything there: available to consciousness, strong enough to steer behavior, actively kept from avowal, and more defensive the closer you get.

On this account contempt is the anti-ritual. A ritual makes a thing public so that nobody can go on acting as if they didn’t know. Contempt keeps the sameness from ever becoming avowable, while both people go on knowing it. Every move in Chapter 1 is a move to keep one division intact, between what is registered and what can be said, and the deniability is what protects the division. “You’re being mean” fails because it tries to make the thing avowable with one voice, and the deniability was built to absorb one voice.

The second stranger’s act reversed that. He announced nothing; his words could have been nearly the same as the first man’s, and the difference wasn’t in them, which is why nothing about it was auditable. What he did was enact the sameness between the two of you: he treated you as a person who had done something, and for that moment you no longer needed to hide from what you already knew. That doesn’t create the recognition. It makes it available, between two people, where before it was held down on both sides. That’s the reason you couldn’t win against him and didn’t want to. What this book is trying to do, with a shared name, is the public version of that, which he didn’t need and couldn’t have supplied.

So limbo, at the level of a whole people, is the refusal to let the sameness be said. §7 arrived at the same place from the other side: what dissolves the act is a shared, sayable name.

One caution about the traditions, since Appendix 3 leans on them. They supply the premises: a nature that is primordial rather than acquired, ignorance as non-recognition rather than absence, recognition disclosing what was already so, compassion belonging to awareness itself. The Dzogchen prayer of Samantabhadra has the whole shape: one ground, and from it awareness and non-recognition. Under those premises you’d expect cruelty to involve obscuration and kindness to involve recognition. What they don’t contain is the machinery of Chapter 1: deniability as the shell, concealment as the message, provocation as recruitment, the target’s reaction used to ratify the verdict. That part is the book’s. And “one ground” is not the same claim as one numerically identical awareness; most Tibetan accounts keep individual continua, and the literal one-awareness premise is closer to Advaita. This part needs three things, and no one tradition grants all three in so many words: that care is what awareness is, that awareness knows this, and that it recognizes the same nature in another. Between them they supply the pieces; the package is this book’s. So the convergence is real and it isn’t confirmation: grant the premises and much of this follows. What would count as confirmation is whether the mechanism predicts real encounters better than the accounts in Appendix 2 do.

Where else to look

On common knowledge. Lewis, Convention (1969); Aumann (1976), for the definition. Chwe, Rational Ritual (2001), for the public event that usually produces it. The modal-logic literature on unawareness, for knowledge present in a state without being available for use. §2 is the contrast; Part Three develops the analogue under its own marked premises.

On what it is. Bateson et al. (1956) on the double bind: a demand, a contradicting demand at another level, and no way to comment on the contradiction. Pinker, Nowak & Lee (2008), “The logic of indirect speech,” on content conveyed so as not to become common knowledge; the trap is an innuendo whose subject is itself. Goffman on face-work, for the auditable surface the act hides behind.

On limbo. Fingarette, Self-Deception (1969): self-deception as declining to spell out what you’re doing. Sartre’s bad faith; Mele as the deflationary rival. Kunda (1990) on motivated reasoning: the search stops at the wanted conclusion.

On the leak being functional. Trivers; von Hippel & Trivers (2011); Kurzban (2010). Adjacent, with one inversion: here the leak is the payload.

On the two strangers, measured. Tangney & Dearing, Shame and Guilt (2002): “I did a bad thing” repairs, “I am bad” hides and attacks. Braithwaite, Crime, Shame and Reintegration (1989): stigmatizing versus reintegrative shaming, with outcome data. Gilligan, Violence (1996): shame as the universal precursor to violence.

On turning away as a perceptual event. Harris & Fiske (2006): social-perception regions go quiet for dehumanized groups. Gray & Wegner on moral typecasting: seen as agent, unseen as sufferer. Bandura on moral disengagement, which catalogs limbo’s toolkit.

On the arrangement. Girard on the scapegoat. Baumeister, Evil (1997), for the exterior view of the license, the terminus, and the myth of pure evil. Browning, Ordinary Men (1992).

On the shadow. Jung.

On privation. Augustine, Enchiridion.