Open Śaṅkara's commentary on the Īśā Upaniṣad and watch a magician's hands. The verse says tena tyaktena bhuñjīthāḥ — and the root of bhuñjīthāḥ means "enjoy." Enjoy what is given up. Eat it. That is what the word does in Sanskrit; it is the same root that gives you bhoga, enjoyment, and bhojana, a meal.1 A plain reading lands somewhere world-affirming: renounce, and in renouncing, enjoy.
Śaṅkara cannot have that. His whole system needs the world held at arm's length, not enjoyed. So he takes the verb and, very quietly, adds a word that is not in the verse — ātman — and re-reads the line as "protect the self by renouncing."1 Enjoy becomes protect. The shift is invisible unless you know the grammar, and most readers do not. A word was slipped in. The verse now says what the commentator already believed.
Selvalingam tells this story not to attack Śaṅkara but to confess. Because if Śaṅkara did it, so did everyone — Rāmānuja, Madhva, Abhinavagupta, every towering interpreter — and, he says, "arguably what I'm doing right now." Every one of them read their own meaning into the text and called it the text's meaning. There is a word for that, and it is not flattering.1
Two Greek words mark the line. Exegesis is drawing the meaning out of a text — leading it out, letting the words say what they say, even when you do not like it. Eisegesis is reading your meaning in — leading your conclusion into the text and discovering, with great relief, that the text agrees with you.1 One word has ex-, out. The other has eis-, in. The whole difference between honest reading and self-flattery hides in that one prefix.
Selvalingam's claim is bracing: the great commentaries are eisegesis, and so is he, and so are you.1 He gives it a second name borrowed from psychology so no one mistakes it for a compliment — confirmation bias. You go to the text looking for what you already hold, and the text, being old and rich and ambiguous, obliges. Then he gives it a third name, the broadest: all translation is interpretation.1 There is no neutral rendering. To carry a word from one language, one century, one inner world into another is already to choose. The choice is unavoidable. The dishonesty is only in pretending you did not make it.
This is a corrosive thing to say inside a tradition that runs on commentary. It dissolves the floor the great teachers stand on — including the floor under the speaker's own feet. That is the point.
The mechanism is small and almost never caught. It is not lying. A liar knows the text says one thing and reports another. The eisegete does something subtler and more sincere: he sees in the text what his commitments have already taught him to see, and he genuinely cannot see otherwise. Śaṅkara did not feel himself smuggling ātman into the verse. To him it was obviously implied. The conclusion came first; the reading bent to fit; and the bending felt like discovery.1
Watch how the move works. Step one: a prior commitment — "the world must be renounced, not enjoyed." Step two: a text that, read plainly, threatens the commitment — enjoy what is given up. Step three: a small grammatical adjustment that relieves the threat — supply the absent word, re-root the verb, and the danger is gone. Step four, the crucial one: the adjustment is reported not as an adjustment but as the text's own meaning, signed with the authority of the great ācārya. The reader inherits the conclusion already laundered as scripture.1
What makes it so durable is the authority stack. By the time the reading reaches you it is no longer "Śaṅkara preferred this"; it is "the Upaniṣad teaches this," with a thousand years of repetition behind it. The insertion has become invisible because it has become canonical. Selvalingam's instrument for making it visible again is just the prefix: ask of every commentary, including his own, did you draw this out, or did you lead it in?
This page hands the vault a solvent for false authority. The commentarial pages can now be read with one extra question running underneath them. Kashmir Śaivism as a Commentarial Corpus describes a tradition that exists almost entirely as layered commentary on prior commentary; this page tells you that every layer is also an act of eisegesis, which does not discredit the corpus but changes how you stand toward it — as a record of brilliant, motivated readings rather than a relay of fixed truth. It also sharpens the indigenous definition of Tantra: when Selvalingam de-flattens a word, he is doing deliberate, confessed exegesis against the grain of inherited eisegesis. The honesty about interpretation is what licenses the whole method.
Take the single verb apart, because the whole argument lives in it. Bhuñjīthāḥ is a form of bhuj, and bhuj means to enjoy, to consume, to partake.1 It is not a rare or contested root. A first-year student would gloss it "enjoy." The Īśā Upaniṣad's opening, on a plain reading, makes a startling and beautiful claim: the whole world is enveloped by the divine; therefore renounce it — tena tyaktena — and by that very renunciation, bhuñjīthāḥ, enjoy. Give it up and you get to taste it. Renunciation as the doorway to enjoyment, not its opposite.1
That reading is intolerable to a strict non-dualist for whom the world is finally a problem to be transcended, not a feast to be tasted. So Śaṅkara reaches for the verb and reinterprets. He cannot change the word on the page, but he can change what it must mean, and he does it by importing a word that is simply not there: ātman. Now the line is no longer "enjoy" but "protect the self." The renunciation no longer opens onto enjoyment; it closes onto self-preservation. The world-affirming charge is defused. And the absent word does all the work.1
Selvalingam's phrase for this is that Śaṅkara tortures the verse — and he means torture in the old sense, racking a body until it confesses what the inquisitor needed to hear.1 The verse, stretched, confesses non-dualism. But the confession was scripted before the questioning began. This is the cleanest possible specimen of eisegesis: a prior system, a threatening text, a silent insertion, and a re-reading presented as the meaning rather than as the maneuver.
Here is why the case is devastating rather than merely clever. The example chosen is not a marginal teacher but Śaṅkara — the most authoritative commentator in the whole Vedāntic tradition, the gold standard of exegetical rigor. If even he is doing eisegesis on a five-word verse, then no commentary is safe from the charge, and the reader cannot outsource trust to authority. You are thrown back on the prefix and made to ask it of everyone, forever — which is precisely the burden Selvalingam wants the serious student to carry.
You sit with a verse and its commentary open side by side, the way a student of the tradition is supposed to. You read the verse first, slowly, in the bare grammar — before the great teacher tells you what it means. You let the words say their plainest thing. Bhuñjīthāḥ — enjoy. You sit with the discomfort if the plain meaning cuts against what you wanted it to say. You do not reach for the commentary yet, because the commentary is a hand that will gently turn your head.
Then you read the commentary, and you keep one question lit the whole time: out, or in? Every place the commentator adds, qualifies, "clarifies," you mark it and ask whether the addition was already in the text or carried in from his system. When you find an inserted ātman — a word supplied, a root re-bent, a "this obviously means" doing heavy lifting — you do not sneer and you do not swallow. You simply note: here the reading turned inward. The teacher led his conclusion in. Maybe rightly. But it was led.
And the last move, the one that keeps the practice honest, you turn the prefix on yourself. You notice which reading you were rooting for before you opened the page — and you watch your own attention reward the gloss that flatters your commitment and resist the one that threatens it. You catch yourself doing to the verse exactly what Śaṅkara did, in real time. That catch is the whole practice. Not neutrality — there is none — but the end of pretending you are neutral.
The first sign is the feeling of relief. You read a hard text, hit a line that threatens you, and feel a small click of relief when the commentary "shows" it means something safer. That relief is the tell. Truth does not usually arrive feeling exactly like reassurance. When a reading lands as comfort precisely where you needed comfort, suspect that you led it in.
Second sign: you only ever cite the commentators who agree with you. The non-dualist quotes Śaṅkara; the world-affirmer quotes the world-affirmers; and each treats his chosen authority as simply "the tradition." A whole shelf of brilliant minds, and you keep pulling down the three that confirm you. That is confirmation bias wearing a scholarly robe.1
Third sign, the deepest: you cannot name a single text that proved you wrong. If years of reading scripture have never once forced you to abandon a position you walked in with, you have not been reading the texts — you have been using them as a mirror. Real exegesis leaves scars; it makes you give up readings you loved. Pure eisegesis never costs you anything, because it never tells you anything you did not already believe.
Selvalingam lays this out across a single transcript span — the exegesis/eisegesis pair, the confsession naming Rāmānuja, Śaṅkara, Madhva, Abhinavagupta and himself, the equation with confirmation bias, the bhuñjīthāḥ case, and the silently inserted ātman.1 The honest tension is reflexive and he names it himself: if all translation is interpretation, then this very page is interpretation too, and Selvalingam's reading of Śaṅkara as "torturing" the verse is itself a motivated reading by a world-affirming Śaiva with a stake in the world-affirming gloss. The page does not resolve this; it cannot, on its own terms. What it can do is hold the screw turned on every party at once, including the speaker and the reader. The vault's scholarly warrant (Wallis, Samuel, White) treats the commentarial tradition historically rather than adjudicating the Īśā verse, so this page rests on Selvalingam's transcript for the specific bhuñjīthāḥ claim — treat the grammatical point (the root means "enjoy," ātman is absent from the verse) as checkable and firm, and the charge of deliberate torture as Selvalingam's argued, self-implicating reading rather than neutral fact.
The plain reason these domains meet here: reading your own conclusion into a text is the same act as reading your own conclusion into the world, and psychology and behavioral mechanics have both already mapped it — one as a flaw to be guarded against, the other as a lever to be pulled.
First, into psychology. Self-Deception is the study of how a mind hides a truth from itself while sincerely believing it is being honest. That is eisegesis turned inward on experience instead of on a verse. Śaṅkara did not feel himself cheating, and the self-deceiver never does — the prior commitment edits perception upstream of awareness, so the conclusion arrives feeling like discovery. Lay the two side by side and you get something neither states alone: that scriptural commentary and ordinary self-deception are the same cognitive engine pointed at different objects. The commentator deceives himself about what a text says; the patient deceives himself about what his life means; both supply the absent ātman that makes the threatening data safe. The tension is instructive too — self-deception is treated as a wound to be healed, while Selvalingam treats eisegesis as unavoidable, something to confess rather than cure. Holding both keeps the reader from the fantasy of a fully honest reading and from the despair of giving up on honesty altogether.
Second, into behavioral mechanics, where the same seam becomes a tool used on others. People Want to Believe is the operator's recognition that a person will reach for the reading that serves them and resist the one that costs them — and that you can win by handing them the gloss they already wanted. That is eisegesis weaponized. Where Śaṅkara supplies the absent word to comfort himself, the persuader supplies it to move you: he gives your prior commitment a text to stand on, and you do the rest, sincerely, calling his framing your own conclusion. The structural parallel is exact — both exploit the mind's hunger to find outside confirmation for an inside commitment. Set them side by side and one uncomfortable fact surfaces: the same mechanism that makes a thousand-year-old commentary persuasive is the mechanism a con artist runs in an afternoon. The defense is identical to the spiritual practice on this page — keep the prefix running, ask of every offered meaning whether it was drawn out or led in, and watch most sharply when the offered meaning is exactly what you hoped to hear.
The Sharpest Implication. If all translation is interpretation, then there is no safe outsourcing of truth — not to Śaṅkara, not to your guru, not to the tradition, and not to this page. Every authority you trusted to "just tell you what it says" was, somewhere, supplying an absent ātman. The comfort of standing on a settled commentary is the comfort of not noticing whose conclusion you inherited. You are reading interpretations of interpretations, all the way down, and the only honesty available is to know it.
Generative Questions.