Two people are cooking the same dinner. Same recipe, same evening, same amount of effort, and both of them will be at the table.
One is cooking because they want to be the person who cooked it. They're already imagining the moment the plates land, the pause, the thing that gets said. They've chosen a dish that photographs well. If nobody notices, the evening is a loss.
The other is cooking because the guest hates coriander and loves things that are slightly too rich, and there is a specific face that person makes at the first mouthful, and getting that face is the entire object. If the cook is ignored completely, the evening was a success.
Nothing on the outside separates them. Same pans, same hours, same food, same smile at the door. You could film both and not tell.
This chapter says that's the only distinction that matters, and it says it about the most scandalous relationship in Indian religion.
Ādi 4 states it as a definition, not an argument:
"The desire to gratify one's own senses is kāma, but the desire to please the senses of Lord Kṛṣṇa is prema."1
One axis. Whose satisfaction is the point. There is no second criterion — no test of restraint, no permitted-acts list, no threshold of intensity. The verse names the variable and stops.
And it's doing serious work, because the thing it's classifying is a set of married women leaving their houses at night to meet a man who isn't their husband. The tradition knows exactly how that reads. It has spent centuries fielding the objection. And its answer is not it didn't happen, and not it was a metaphor, and not it was allegory about the soul. Its answer is: look at the direction of benefit and you'll find it's the opposite of what it resembles.
The gloss immediately after is the part people miss: "Whenever there is unselfish love, that is its style. The reservoir of love derives pleasure when the lovable object is pleased."2 The pleasure isn't renounced. It's rerouted. The lover does get satisfaction — from the other person having it. That's a different claim from self-denial, and much odder.
The chapter then does something clever. Having made the criterion invisible, it produces visible evidence for it — not evidence of what the gopīs felt, but evidence of what they paid.
The list is itemised: for his sake they abandoned religious principles, their patience, their bodily comfort, their relatives, their husbands and sons, and their shyness — and, at the top of the list, the one that costs most in that society, their reputation for chastity.3
Read as an argument, that's an accounting exercise. Anybody pursuing their own pleasure buys it with something, and the price tells you what the purchase was worth to them. Somebody who gives up religion, family, and the ability to walk through their own village is not making a good trade for an evening's enjoyment. The costs are too high for the pleasure hypothesis to explain them.
It's not a proof. Compulsions run at a loss too, and the text does not consider that. But it's a real move, and it's the only externally checkable thing the chapter offers.
Here's where the corpus adds a mechanism, and the mechanism is stranger than the doctrine.
The problem it has to solve: if Kṛṣṇa is the actual husband of everyone and everything, and the gopīs are his own energies, then there's no adultery available — they're technically his. But the whole emotional charge of the story depends on the transgression. Remove the illicitness and the love drops back to the ordinary devotional temperature the previous page says cripples it.
So the text has the illicitness manufactured. Yogamāyā — his internal deluding potency — induces the paramour sentiment, arranging the appearance of a forbidden relationship where none metaphysically exists.4
Sit with that. The transgression is staged, by the deity, so that the love aimed at him can be of the higher, forbidden kind. It's an engineered scandal — the theological equivalent of building an obstacle so somebody can climb it, except the climbers don't know and the intensity depends on them not knowing.
The vault already holds the jurisdictional side of this argument; this page's contribution is only to note that the direction test and the yogamāyā mechanism are doing opposite jobs. The direction test says externals don't matter. Yogamāyā exists specifically to arrange the externals. If the internal direction were really sufficient, no staging would be needed.
Take the criterion seriously for a moment and ask what you'd actually do with it.
Somebody comes to you and says: I did this for them, not for me. How do you check?
You can't ask them — the answer is the claim under examination. You can't watch the act, because the doctrine has already told you the act is identical either way. You can't use the outcome, because a self-serving act can produce a good outcome and a genuinely selfless one can end in disaster. You can look at the price paid, which is the gopī argument, and that only distinguishes serious from casual, not for you from for me.
What you're left with is: the only reliable witness is the person being examined.
That is not a small consequence, and it propagates. Once the decisive variable is invisible, every judgement in the system becomes a judgement about interiors, and interiors are reported by their owners or declared by an authority. The corpus does exactly this later — Antya 5 licenses outrageous behaviour on an interior state "He alone, and no one else, can understand"; Madhya 23 settles authenticity by expert declaration of what is real feeling and what is its shadow.5
So this verse is not just a definition of love. It's the moment the tradition adopts an unfalsifiable criterion, and the rest of the corpus's most difficult episodes are downstream of it. The definition is beautiful and it is also the load-bearing wall of everything that goes wrong later.
And here is the tension the corpus never places on one page.
The theology on this page has no content rules. The discipline has nothing else. Madhya 22 names association with women in an enjoying spirit and association with non-devotees as the two formidable enemies, in flat behavioural terms. Antya 2 permanently expels a monk for talking with a woman while collecting alms, and refuses every appeal until the man drowns himself. Antya 13 has Caitanya run through thorn bushes to escape the sound of a woman singing, and stop the instant he learns whose voice it was.6
None of those rulings ask about direction of benefit. They ask what was done.
The book, on the same authority, in the same voice, says intention is the whole of the matter and behaviour is the whole of the matter. It never once puts the two sentences beside each other. Per this vault's constitution that stays open, and every discipline page in this corpus inherits it.
The honest reading is not that the text is confused. It's that the two rules are doing different jobs — one describes the summit and one runs a community — and the corpus keeps them in separate rooms because they cannot be in the same room.
Layer B handles this verse nervously, and the nervousness is informative.
The purports around Ādi 4 repeatedly stress that this material is not for general discussion, that imitating the gopīs is disastrous, and that the reader's business is regulated practice rather than the sentiment being described.7 Some of that is in the verse — the chapter has its own esotericism gate. But the volume of the warning is Layer B's.
That's a commentator with an expanding twentieth-century organisation looking at a doctrine which says the highest love is defined by nothing you can observe, and quietly bolting a behavioural fence around it. Which is, in miniature, the same manoeuvre the corpus itself performs across Ādi 4 and Antya 2 — and it's worth noticing that the commentator repeats his text's structure rather than resolving it.
You are helping someone move house. Third flight of stairs, and you catch yourself hoping they'll mention it later, to the others. Don't scold yourself — just note which way the hope points. The box is the same box either way. The doctrine's whole claim is that this small directional fact is the only thing about the afternoon that will be true tomorrow.
You give a genuinely good piece of advice and they ignore it and it goes badly. Notice the flare of satisfaction. That's the test running in reverse and returning a result you didn't want. The advice was real; the flare says who it was for.
Someone you love is happy about something you find slightly embarrassing. You have a choice available in the next two seconds: be pleased that they're pleased, or manage them toward a version you'd enjoy more. Both are available; both look like interest from the outside. This is the whole doctrine, at four seconds' resolution.
You are cooking that dinner. Halfway through you realise you've chosen a dish you like over the one they'd prefer, and it's too late to change. Finish it well anyway. The corpus's position, harshly stated, is that you made a good dinner and not a loving one — and that these are different achievements, and that people confuse them constantly because they taste the same from the outside.
The definition is a single verse, stated without hedging and immediately glossed with the reservoir-of-love line, then supported by the enumerated list of what the gopīs abandoned.8 That's a definition, a mechanism and an evidential argument in about twenty verses — dense even by this chapter's standards.
The tension the source doesn't close is the corpus's central one. Intention decides everything here; behaviour decides everything in the disciplinary chapters. Both are asserted absolutely, by the same figure, and never compared.
A second, internal to this page. Yogamāyā's staged illicitness undercuts the sufficiency of the direction test. If the direction of benefit is genuinely the whole criterion, the paramour sentiment is decorative — and the text treats it as anything but.
Open: the direction test is stated for prema toward Kṛṣṇa specifically. Does it generalise? Nothing in the range read (Ādi 4, Madhya 8, Madhya 19, Antya 20) applies the same criterion to love between two human beings, and the corpus's disciplinary chapters imply it does not — human-directed devotion of the same shape is exactly what gets Junior Haridāsa expelled. The definition may be narrower than it sounds.
This page underwrites Bhakti Transcends Dharma: The Gopī Defense, and reading the two together shows where each stops. That page argues the exemption jurisdictionally — the gopīs' conduct isn't judged by dharma because dharma has no authority over that transaction. This one argues it definitionally: their conduct isn't the same kind of thing it resembles, because the variable that classifies it isn't visible in it. Two different acquittals for the same defendants, offered by the same tradition, on incompatible grounds — one says the court lacks jurisdiction, the other says the charge misdescribes the act.
Against Parakīyā-vāda and the Jurisdictional Exemption the yogamāyā material here is straightforward enrichment rather than dispute — Ādi 4 supplies the induced-sentiment mechanism that page's argument assumes.
There's a sharper divergence with Ahaitukī Bhakti: The Motiveless Ideal. Motiveless devotion is usually framed as devotion with the motive removed — nothing wanted in return. Ādi 4's version is not motiveless at all. The gopīs want something enormous: his pleasure, and their own pleasure taken through his. Prema here is not the absence of desire but desire pointed the other way, which is a considerably more workable psychology and a considerably less austere one.
Judge by Effects, Not Intentions — this is a head-on collision, and it's worth having rather than smoothing. Greene's rule is that treating "they meant well" as exculpatory is a measurement error: intentions are unobservable, self-reported, and systematically flattering, so the only honest instrument is what actually happened. Ādi 4 says the reverse in as many words — the act and its effects are identical across the two categories, and the invisible intention is the sole classifier.
They can't both be right about the same question, but notice they aren't asking the same question. Greene is asking how to assess somebody from outside. Ādi 4 is telling you what your own act is. And that's where the two together produce something neither has: the direction test is a first-person instrument that becomes worthless the moment it's applied in the third person. Used on yourself, mid-act, it is the sharpest tool available — you genuinely can catch which way the hope is pointing while you're carrying the box. Used on somebody else, it is unfalsifiable and immediately becomes a licence, because the only witness is the defendant.
Which predicts precisely what happens to this doctrine downstream. The corpus applies it in the third person — Antya 5's prerogative doctrine, Madhya 23's expert declaration of genuine feeling — and every time it does, the result is a ruling nobody can check. Greene's rule isn't a refutation of the direction test; it's a statement of where it stops working, and the Caitanya-caritāmṛta crosses that line and never comes back.
Sharpest implication. A criterion that is genuinely correct can still be the wrong criterion to institutionalise. The direction of benefit really does seem to be what separates love from appetite — the cooking example works, everyone recognises it. And it is also completely unauditable, which means that the moment a community adopts it as a standard rather than a private examination, it converts into whatever the authority says it is. The tradition got the definition right and the governance catastrophically wrong, and both came from the same verse.
Generative questions.