Here is a method for finding out what someone is thinking, and it does not involve asking them, watching them, or knowing anything about them.
You copy their face. Then you look inside your own head and see what turns up.
"When I wish to find out how wise, or how stupid, or how good, or how wicked is any one, or what are his thoughts at the moment, I fashion the expression of my face, as accurately as possible, in accordance with the expression of his, and then wait to see what thoughts or sentiments arise in my mind or heart, as if to match or correspond with the expression."1
That sits in a marginal column beside the Alcibiades Observance, attributed to Edgar Allan Poe, and Law 44 never refers to it again.
Read it once more and notice where the attention goes at the end. Not to the other person. Back to yourself. You arrange your own face, and then you wait — and the thing you are waiting for arrives in your own chest.
Every other mirror in Law 44 is aimed at somebody.
Fouché reflects Napoleon's methods to become unprovable. Alcibiades reflects a city's values to be adored by it. Marie reflects a king's fantasy to be loved. Ivan reflects the boyars' contempt to teach them. Rikyū reflects a guest's unspoken wish to delight him. Weil reflects an entire bank to rob a man.
All of them are doing something to a target. The reflection is the delivery mechanism, and the payload lands in the other person.
Poe's mirror has no payload. The other person need not notice, need not respond, need not be affected in any way — they can be across a room, asleep, or dead. The information travels in the opposite direction from every other case in the chapter: out of the target's face, into the operator's own nervous system, and it stops there.
That makes this the chapter's only epistemic instrument. Everywhere else the mirror is used to affect; here it is used to learn. And the chapter, which has a four-part typology with seven named mirrors in it, has no category for that at all — because every one of its four effects is defined by what happens to the person being mirrored.
Before going further, the attribution needs looking at, because the printed page and the citation do not say the same thing.
The marginal's header, as it survives in the source text, reads "THE PU RIOINED LLTER" — an obviously mangled The Purloined Letter.2 So the header names one of Poe's short stories.
The attribution line underneath reads, in full: "EDGAR ALLAN POE, 1809-1849."1
Those are two different claims. One says this passage comes from a work of fiction. The other presents it as Poe's own statement, dated by his lifespan the way the chapter dates Freud and Musashi — a man's observation rather than a character's line in a detective story.
🚩 SECONDARY WITHOUT PRIMARY. No edition, no page, no indication of who is speaking. A sentence inside a story is being quoted as though it were its author's considered position on how minds are read, and a reader with only this page cannot tell that a narrator stands between Poe and the claim. [LOW CONFIDENCE] on the attribution; the method is stated clearly enough to examine on its own terms, and that is what this page does.
Take the method seriously as a claim about how bodies and minds work, because it is a real one and it is falsifiable.
Poe's proposal has two moving parts. First, that a face is a reliable index of an inner state — copy the arrangement and you have copied something meaningful. Second, and much stronger, that producing the arrangement in yourself will generate the corresponding state, or something close enough to it to be informative.
The second part is the interesting one, and it runs the ordinary causal story backwards. Common sense says the feeling comes first and the face reports it. Poe says the face can come first and the feeling will follow to match.
That idea has a long and genuinely unsettled empirical history. It was formalised in the nineteenth century, revived in the twentieth, and tested repeatedly — most famously in a study where people held a pen in their teeth, forcing a smile-like configuration, and rated cartoons as funnier.
🚩 REPLICATION RISK — and this one is not hypothetical. That result is among the best-known casualties of the replication crisis: a large multi-laboratory replication failed to reproduce it, and the broader literature since has been mixed rather than settled. The honest tag is [CONTESTED], and it is worth stating plainly because this is exactly the kind of tidy anticipation the vault is prone to over-credit. Poe did not predict a confirmed finding. He described a mechanism that a century and a half of research has been unable to establish or dismiss.
What survives that caveat is still worth having, and it is the part Poe actually needed. Even on a weak version — where arranging your face produces nothing more than a faint shift in what feels available to think — the method has a use, because it is not a truth-detector. It is an idea generator. You are not learning what the other person thinks; you are giving yourself a nudge toward the region their expression occupies, and then seeing what your own mind offers up from there.
Which is a much more modest and much more defensible claim than the one the marginal appears to make. It also means the output requires checking against something else, and Poe says nothing about that.
Two Observances later, in the margin beside Marie Mancini, sits a claim that contradicts the entire chapter, sourced properly, dated, with a named author and a named book.
"Wittgenstein had an extraordinary gift for divining the thoughts of the person with whom he was engaged in discussion. While the other struggled to put his thought into words, Wittgenstein would perceive what it was and state it for him. This power of his, which sometimes seemed uncanny, was made possible, I am sure, by his own prolonged and continuous researches."3
Malcolm is describing exactly the ability Law 44 spends twelve pages teaching — the Courtier's Mirror, the man who knows what you meant before you said it, the "uncanny ability" that made Rikyū famous.
And then he says where it came from, and it is not observation.
Not empathy. Not attunement. Not studying eyes and gestures and tossed-out remarks. Work. Wittgenstein could finish your thought because he had already been down that road himself, alone, for years — so when someone began struggling toward an idea, he recognised the terrain, because he had walked it.
If Malcolm is right, Greene's instruction is aimed at the wrong thing. "Study people's eyes, follow their gestures… soak it all in" treats the ability as a perceptual skill to be trained by looking harder at people. Malcolm's version makes it a residue — the visible by-product of having thought a subject through so thoroughly that other people's half-formed versions of it are legible on sight.
The two accounts predict opposite training regimes. One says: watch people. The other says: go away and think about the thing until you know it, and the reading of people arrives on its own, in the domains you actually understand.
And Malcolm's has an implication Greene's does not: the ability would not transfer. Wittgenstein could finish a philosopher's sentence. Nothing in Malcolm's explanation suggests he could have finished a farmer's.
Directly beneath Malcolm, one line, no context:
"The doctor should be opaque to his patients, and like a mirror, should show them nothing but what is shown to him."4
That is a fifth mirror, and the typology does not have it.
Look at what the four named effects have in common. The Neutralizing Effect supplies a copy of the target's behaviour. The Narcissus Effect supplies an idealised image. The Moral Effect supplies a lesson. The Hallucinatory Effect supplies a counterfeit world. In every case the operator adds something — that addition is the technique.
Freud's mirror adds nothing, on purpose. Its entire discipline is subtraction: the analyst's own personality, opinions, reactions and preferences are withheld so that what appears in the room can be attributed to the patient rather than to the analyst. Opaque is the operative word, and it is a strange one — an opaque mirror is precisely a mirror, since a transparent one is a window. The demand is that nothing pass through.
So a mirror that withholds is being used as an instrument for making someone else's material visible. Like Poe's, its purpose is epistemic. Unlike Poe's, the knowledge is meant to arrive in the other person.
Set the two side by side and you have the two honest mirrors in Law 44, printed a few columns apart, both in the margins, neither one cited by the body of a chapter that is otherwise entirely about using reflections to move people.
Put the three together and something becomes visible that no single one of them shows.
Poe: the mirror aimed inward, for information. Malcolm: the ability is not a mirroring skill at all, it is the sediment of solitary work. Freud: the mirror that supplies nothing, so that what appears belongs to the other person.
Every one of them describes a use of reflection that the chapter's four-effect framework cannot accommodate. And all three sit in the margins of the Observances rather than in the argument — Poe beside Alcibiades, Malcolm and Freud beside Marie Mancini.
⚠ This is the fifth law running in which the marginal column carries named, dated, attributable sources while the body's cases carry none, and the margins keep supplying the mechanism the body needed. [n≥3 PATTERN] on the general finding; what is specific to Law 44 is the direction of the disagreement — here the marginals are not merely better sourced, they are arguing with the typology.
The body of the chapter cites not one of them.
You are reading someone's message for the fourth time and still cannot tell what is behind it.
The words are neutral. The information is complete. Something in it is off, and every reading you try — annoyed? rushed? being careful? — feels equally plausible, which is another way of saying you have no read at all.
Do the thing that feels absurd. Sit up and put the message's posture on your own face. If it reads clipped, set your jaw the way a clipped person's is set. If it reads over-careful, do the small tight mouth of someone choosing words.
Then stop working on it. Do not re-read. Wait, for a few seconds, and notice what turns up.
What you are watching for is not an answer. It is what becomes available. Held one way, the sentence let me know when you have a moment is somebody protecting their afternoon. Held another, it is somebody who has decided not to say the real thing yet. Both were already possible; one of them just arrived with more weight.
Now the essential half, which Poe's marginal does not supply. Treat whatever arrived as a hypothesis, not a finding. The evidence for it is that you produced a feeling in yourself by rearranging your own face — which is a fact about you, not about them. The mechanism this rests on has never been reliably established, and the failure mode is obvious once named: you will generate the reading you were already primed for and it will now feel bodily and therefore true.
So take it and go find something independent. Look at what they did last week. Ask a direct question. Check whether the reading survives one piece of evidence that came from outside your own body.
And keep Malcolm's version in view as the longer game, because it is the one that actually pays. If you can never read a particular person, the answer is probably not more watching. It is that you do not yet know their work well enough to recognise where they are trying to get to. Wittgenstein finished other people's sentences because he had already written those sentences himself, in private, for years. That is a slower method than arranging your face, and it is the only one here with a mechanism that holds up.
Strongest evidence is that the three marginals converge without being placed together. Poe, Malcolm and Freud sit in different columns beside different Observances, drawn from fiction, a philosophical memoir and psychoanalysis, and all three describe a mirror the chapter's framework cannot hold. Convergence across unrelated sources is worth more than any of them individually.
Tension one — Malcolm contradicts the chapter's whole training programme. "His own prolonged and continuous researches"3 makes the ability a by-product of subject-matter mastery; Greene makes it a perceptual skill trained by watching people.5 Both are stated flatly, in the same chapter, and the chapter does not notice. Preserved, not resolved.
Tension two — Freud's mirror is defined by withholding and all four named effects are defined by supplying. The typology cannot absorb it without giving up its organising principle.
🚩 SECONDARY WITHOUT PRIMARY on Poe — a line whose header names a work of fiction, attributed to the author personally, with no edition and no speaker identified.2
🚩 REPLICATION RISK / CONTESTED on the underlying mechanism. The proposition that arranging the face generates the corresponding state has a genuinely unsettled empirical record, including a prominent failed multi-lab replication. Anyone citing this page forward should carry the [CONTESTED] tag with it and must not present Poe as having anticipated an established finding.
Open question — what is the actual output of Poe's method? A read on the other person, or a sample of the operator's own priors given a nudge? The marginal does not distinguish them, and nothing in the chapter offers a way to tell from inside.
Open question — does Malcolm's account generalise beyond a domain? If the gift is the residue of researches, it should be sharply bounded by subject. The memoir describes Wittgenstein in discussion, which is his own territory. Whether the ability travelled is not addressed.
Within Law 44, this page's material is in direct conflict with The Narcissus Effect and with the Courtier's Mirror instruction that closes Sen no Rikyū Falls Into the Ditch.
That instruction is unambiguous about where the skill comes from: "Study people's eyes, follow their gestures—surer barometers of pain and pleasure than any spoken word. Notice and remember the details… Soak it all in."5 Observation, accumulated, deliberately, about people.
Malcolm's sentence, printed thirty lines earlier in the same chapter, says the most celebrated instance of that ability in twentieth-century intellectual life came from somewhere else entirely. Greene has printed the refutation of his own training method in his own margin.
The convergence that runs the other way is with Erickson and the Carpenter. Freud's show them nothing but what is shown to him and Erickson's enter their world and operate inside it, under their rules are two clinicians describing a discipline of self-removal — keep your own material out of the way so the patient's becomes visible and usable. They are the two places in Law 44 where the mirror is not a weapon, and both are practitioners of the same profession. That is not a coincidence the chapter draws out.
And against the corpus's running tally: this page supplies the counter-instance to a pattern the build has been logging since Law 42. Three techniques across Laws 42–44 have no truth condition — they work whether or not the underlying diagnosis is correct. Poe's method is the inverse: it is nothing but a truth condition, an attempt to find out what is actually the case, with no application attached. It is also the least developed thing in the chapter.
The Somatic-Psychological Circuit — the argument that body and mind are not two systems exchanging messages but one nervous system running on two channels, and that the circuit is bidirectional: psychological experience produces somatic states, and somatic states produce psychological experience.
Poe's method is that bidirectionality operated deliberately, as an instrument, and it is the cleanest illustration of the circuit's second direction the vault has.
The first direction is uncontroversial and everybody has felt it — grief slackens the face, fear tightens the gut. The circuit page's real claim is the return leg: that the somatic arrangement is not merely a readout but an input, and that changing it changes what is psychologically available. Poe takes that and turns it into a probe. Set the body to a configuration you did not arrive at honestly, and see what the psychological channel sends back.
What the pairing produces is a caution that neither page issues on its own, and it is sharper than the general replication flag. If the circuit really is bidirectional, then arranging your face does not give you access to another person's state — it manufactures a state in you, indistinguishable from the inside from one you arrived at by understanding. The somatic channel does not label its outputs by origin. So the very mechanism that makes the method possible is the mechanism that makes its results untrustworthy, and the more effectively it works, the more convincing a bad read will feel.
Which turns the circuit page's therapeutic claim inside out. In a clinical setting, bidirectionality is good news: you can reach a stuck psychological state through the body when words cannot get there. Used as an epistemic tool aimed at somebody else, the same property is a fabrication engine — and the confidence it generates is somatic, which is the hardest kind to argue with.
Butcher: Personality as Emergent from Craft Depth — the answer to how do I let my personality come through in the work? that refuses the expected reply. Not be more yourself: the confidence to be yourself "comes from the depth of study, research, repetition, the amount of work you've put in," which then gives you permission to speak candidly.
That is Malcolm's sentence about Wittgenstein, arrived at independently, in a completely different trade.
Both describe a visible interpersonal quality — one man's uncanny ability to finish your thought, another man's distinctive presence in his work — and both refuse to treat it as an interpersonal skill. In both accounts the quality is downstream of solitary technical mastery, and cannot be trained directly because it is not a thing in itself. It is what having done the work looks like from outside.
The pairing produces a diagnostic that is more useful than either page alone, because it tells you what to do when the quality is absent. If a supposedly interpersonal ability is actually the residue of subject-matter depth, then failing at it is a signal about your knowledge, not about your attentiveness. The person who cannot read a room in a domain they know cold, and cannot read one in a domain they are faking, is not equally deficient in both cases — and the standard advice (pay closer attention, watch people more carefully) is only the right advice in the first case.
It also explains something the Greene chapter treats as mysterious and never accounts for. Rikyū could read a glance at a flowerbed because he had spent his life on the tea ceremony, in which every gesture toward an object is meaningful and studied. His "uncanny ability to see into other people" was, on this reading, an ability to see into people doing the one thing he knew everything about. Greene generalises it to a portable power over anyone. Malcolm and Butcher both say it does not travel.
Sharpest implication. The two mirrors in this chapter that are used to learn rather than to move somebody are both in the margins, uncited by the argument, and one of them refutes the chapter's training programme outright. If Malcolm is right that reading people is the sediment of having thought a subject through yourself, then the Courtier's Mirror is not teachable in the form Greene teaches it — the observational surface is a symptom, and the instruction to watch harder is advice to cultivate a symptom. And Poe's method, the one genuinely epistemic tool on offer, rests on a mechanism the last century and a half has failed to settle, and fails in the worst possible direction: it produces the feeling of a read, in the body, where confidence is least available to correction.
Generative questions.
If bodily-generated confidence is indistinguishable from earned understanding, what practices exist anywhere for telling them apart in the moment rather than in retrospect?
Freud's mirror works by subtraction and all four of Greene's work by addition. Are there other domains where the instruments that reveal and the instruments that influence separate that cleanly along the same line?
Malcolm's Wittgenstein and Butcher's craft-depth both say the interpersonal gift does not transfer between domains. How much of what gets called social intelligence is domain expertise being misattributed?