一、What this concept is saying
1.1 Where it sits: a complete picture of how we validly know
To get 自利比量 (svārthānumāna, "inference for one's own sake") in focus, we need to see the table it sits on. The Indian and Tibetan Buddhist tradition inherited a long-running debate about what counts as a valid means of knowledge (pramāṇa, 量). Dignāga (c. 6th c.) and his commentator Dharmakīrti (c. 7th c.) cut that debate down to two pramāṇas, then re-cut it differently for two different purposes.
This is exactly how 《正理滴论》(Nyāyabindu, *Drop of Logic*), translated into Chinese from the Tibetan, structures itself. The whole text is three chapters: 现量品 (perception), 自利比量品 (inference-for-oneself), and 利他比量品 (inference-for-others). The first asks, "what is direct seeing?" The second asks, "how do I validly conclude something I cannot directly see?" The third asks, "how do I lay out such a conclusion for another person?"
A later Chinese master, Śaṅkarasvāmin (商羯罗主), summarized this whole architecture in one of the most quoted lines in Buddhist logic:
「能立與能破，及似唯悟他；現量與比量，及似唯自悟。」 *Arguments that establish or refute, and their look-alikes, are for the realization of others. Perception and inference, and their look-alikes, are for self-realization.* — 《因明入正理論》(Nyāyapraveśa, *Introduction to Logic*)
This verse is the cleanest classical statement of what 《正理滴论》 spends three chapters unpacking. 自利比量 is the second half of "現量與比量...唯自悟" — the inference half of "things that produce self-realization."
So 自利比量 names a very specific thing: the cognitive event that happens inside your own mind when, by following a valid reason, you arrive at a conclusion you had not directly seen. It is not the argument you spoke to someone else. It is the silent "ah, so that's why" that follows.
1.2 The classical definition
The definition from 《正理滴论》 is unusually crisp:
「自利比量者，謂由具三相因，所生諸認識。」 *Svārthānumāna is the cognition that arises from a reason (hetu, 因) possessing the three characteristics (trirūpa, 三相).*
Three things to notice:
- 所生諸認識 — "the cognitions that arise." Svārthānumāna is not the syllogism itself, not the words, not the formal steps. It is the cognition in your mind that is *produced* by following those steps. The measure of success is internal: did understanding actually occur?
- 因 — "reason." Not any reason, and not any belief. Specifically a reason that has the three characteristics. If the reason fails the test, you don't get svārthānumāna — you get a fallacious "non-inference" (似比量).
- 具三相 — "possesses the three characteristics." This is the gate. Everything else in the text is exposition of how the gate works.
So when the text later adds, "the establishment of the pramāṇa-fruit (量果) in svārthānumāna is the same as in perception (現量)," it means: when a three-characteristic reason actually does its work, the resulting cognition is as direct and as reliable as seeing. It does not feel like "inference" anymore. It feels like acquaintance.
That is the goal-state of the whole system.
1.3 The three characteristics (三相): what a valid reason must do
The reason you use must satisfy three demands, all at once:
「於所推論比度之事上必須具備，於同品上定有，於異品上遍無。」 *It must be present in the subject (paksha, 所比度之事) being investigated; it must invariably be present in similar instances (sapaksha, 同品); it must be entirely absent in dissimilar instances (vipaksha, 異品).*
In plainer words. Suppose you are trying to conclude something about X (the paksha), using some property P (the reason).
- P must actually be in X. You cannot conclude "this is a sandalwood tree" using "has sandalwood-nature" if the thing in front of you is not in fact a sandalwood. The reason has to be present in the subject (paksadharmatva).
- P must be in everything that is X-like (all sandalwoods, all trees of that class). If some sandalwood trees lack "has sandalwood-nature," your reason is unreliable — the conclusion won't hold for them. The reason must hold in similar instances without exception (sapakṣe sattvam).
- P must be absent in everything that is NOT X-like (every non-sandalwood, every non-tree of that class). If a non-tree also has sandalwood-nature (a wooden chair made of sandalwood, say), you cannot use "has sandalwood-nature" to prove "is a tree" — because it would also prove the chair is a tree. The reason must be absent in dissimilar instances without exception (vipakṣe 'sattvam).
The text further subdivides "dissimilar instances" (異品) into three: those outside the class (彼外之異品, things clearly not in the class), those that contradict the class (與彼相違之異品, things that are explicitly the opposite), and those that lack the class entirely (彼上沒有之異品, things that simply have no relation to it). For our purposes, the practical takeaway is: the reason must not let any non-target thing slip through.
This three-characteristic discipline is the heart of svārthānumāna. It is not a bureaucratic checklist. It is a discipline of self-honesty: the reason you are using — does it really satisfy all three? Or are you reaching?
1.4 The three kinds of reasons (三因): how reasoning actually moves
Once you understand that a reason must possess the three characteristics, the next question is: what kinds of reasons can possibly do this? 《正理滴论》 answers: only three, and no more.
「未緣到因」「自性因」「果因」
(a) Self-nature reason (自性因, svabhāva-hetu). Example: 「此是樹木，是沉香樹故」 — "This is a tree, because it is a sandalwood tree." Here the reason and the predicate share the same essential nature — to be a sandalwood tree just *is* a way of being a tree. This reason easily passes the three characteristics: it is in this subject, in all sandalwoods, and in nothing that is not a tree (well, sandalwood is a kind of tree, so this is rock-solid).
(b) Effect reason (果因, kārya-hetu). Example: 「彼處有火，以有烟故」 — "There is fire there, because there is smoke." Smoke is the visible effect of fire. Wherever smoke is observed under ordinary conditions, fire is present. The reason "smoke" passes the three characteristics because of a stable causal law: smoke is the effect of fire, and not of anything else in the same context (say, dry ice in a stage set aside). The conclusion is reached through a *causal* bridge.
(c) Non-apprehension reason (未緣到因, anupalabdhi-hetu). Example: 「於各別地方無瓶，若有，則當為緣慮之所明瞭，但未緣到故」 — "There is no pot in that location, because, if there were, it would be apprehensible to cognition, but it is not apprehended." This one is the Dignāgan revolution. The reason is the absence of apprehension under a conditional. If a pot were there, you would apprehend it; you do not apprehend it; therefore it is not there.
These three can be folded into two: affirming reasons (立物因) — self-nature and effect reasons, both positive — and negating reasons (否定因) — non-apprehension reasons, which establish *absence*. Both kinds produce svārthānumāna when their three characteristics hold. Both arrive at valid cognition of either "there is X" or "there is no X."
1.5 The eleven forms (十一量式) and why they all collapse into one
The text then enumerates eleven specific configurations of these three reasons. They are given canonical names:
自性未緣到、果未緣到、能遍未緣到、緣到自性相違、緣到相違果、緣到相違遍、緣到果相違、緣到能遍相違、因未緣到、緣到因相違、緣到因相違果
You do not need to memorize the list. What you need to understand is the punchline the text delivers immediately after:
「果未緣到等未緣到之十種量式，皆攝於自性未緣到之中。」 *The ten non-apprehension forms other than self-nature non-apprehension all reduce into self-nature non-apprehension.*
This is Dignāga's signature move: the radical simplification of inference. All valid inference, however varied its surface form, reduces to one operation — excluding (遮) what does not belong. The conclusion is reached by *removing* what is incompatible with the reason.
In Dignāga's broader framework (well-documented in the Pramāṇasamuccaya and developed by Dharmakīrti), this is why the cognition arising from svārthānumāna is "the same as perception": both arrive at exclusion. Perception excludes "this is not blue" by directly confronting blue. Inference excludes "fire is not absent there" by following the reason "smoke" through the three characteristics to its proper scope. Both are acts of establishing what cannot be the case.
1.6 Negation (遮, abhāva): how absence is known
This deserves its own short paragraph because it surprises newcomers. Buddhist logic treats absence as a legitimate object of cognition. When you say "there is no pot here," you are not making a vague negative; you are establishing a real fact.
「凡所說之否定，唯從緣慮之所明瞭而言緣到及未緣到者。」 *Every stated negation is established only by what is or is not apprehended to cognition.*
The mechanism: to know "no X here," you need a conditional — "if X were here, it would be apprehensible in this context." If the conditional is correct (and the text's three kinds of reasons help verify it), and you do not apprehend X, then you validly conclude absence. This is the 未緣到因 in action, and it is how svārthānumāna handles a huge portion of what we actually want to know — that a friend has no intention, that a claim has no evidence, that a career path has no future.
Dignāga's deeper claim (not stated explicitly in this text but foundational to the tradition) is that all cognition is fundamentally exclusion. We know things not by grasping some inner essence, but by successfully excluding what they are not. This is the bridge between pramāṇa theory and the Mahāyāna doctrine of emptiness — but we will not extend the bridge here; the text itself stays inside the epistemological frame.
1.7 The practical character: how svārthānumāna is actually acquired
The most under-appreciated sentence in the text is the one about habituation:
「由於屢見立量而串習，使自己亦如是領悟其遮破，故於自利比量時，亦分別敘之。」 *Because, by repeatedly seeing arguments set out, one trains through familiarity so that one also arrives oneself at the exclusion, svārthānumāna is also set out separately.*
Three implications:
- svārthānumāna is not innate. It develops through repeated exposure to properly-formed arguments (parārthānumāna). You learn to reason-for-yourself by watching others reason-for-others.
- The "fruits" are identical. Whether you learn through being argued with or through arguing with yourself, the resulting cognition is the same kind — an act of exclusion that yields direct acquaintance.
- Form matters, even when no one else is watching. The text "sets out svārthānumāna separately" precisely because the silent act still benefits from being structured. Following the three characteristics, recognizing which of the three reasons you are using, noticing when one reduces to another — these disciplines sharpen the cognition even when there is no audience.
This is why Dignāga and Dharmakīrti wrote technical treatises at all. They are not just for philosophers. They are training manuals for the inner act of reasoning well.
二、Life walkthrough: where this shows up
The danger in any technical concept is that it stays in the seminar room. Svārthānumāna names something you already do, badly, all the time. Let's walk through four scenes and watch the classical structure show up — and watch where most people map it wrong.
2.1 The job offer
Your friend says, "this job is great, take it." You want to know: is this actually great for me? You try to reason it out.
- Paksha (所比度之事): This specific job, in this specific company, for my specific situation. - The reason you consider: "people in this role seem happy." Or: "the salary is high." Or: "the founder seems thoughtful." - Sapaksha test (同品 test): Is the reason present in all happy outcomes for people like me? Are there happy former employees in *similar* roles? Or did you hear about one person who liked it and assume the rest? - Vipaksha test (異品 test): Is the reason absent in *all* unhappy outcomes? Do unhappy former employees also have the same salary, the same thoughtful founder, the same kind of role? Or did you filter out the complaints as "not like me"?
Now, the place most people map svārthānumāna wrong here: they treat it as listing pros and cons. They collect data but do not check whether their reason actually meets the three characteristics. They end up with information, but no conclusion — or worse, a confident conclusion that the data does not actually support. The classical structure is not "collect evidence." It is "stress-test the *one* reason you are about to commit to." What is the single reason you are using? Does it pass all three tests? If not, you are not yet reasoning; you are rationalizing.
A subtler Dignāgan move is the non-apprehension reason. You might realize: "I have asked three careful questions and received no substantive answer from the recruiter. If they had a substantive answer, they would have given it. Therefore, no substantive answer exists." This is not a hostile inference. It is a valid svārthānumāna — a cognition of absence produced by a properly-conditioned non-apprehension.
2.2 The news headline
"Studies show that drinking three cups of coffee a day extends lifespan."
You want to know whether this is true. You are not yet arguing with anyone. You are doing svārthānumāna — internal reasoning for yourself.
- Paksha: Coffee extends lifespan in humans. - Reason: "Studies show..." (this is itself a reason — that academic research supports the claim). - Sapaksha: Is the property "supported by studies" present in all similar cases where academic research actually establishes a health claim? In many cases, yes — replications, large samples, effect sizes. In many cases, no — single studies, observational confounds, publication bias. - Vipaksha: Is "studies show" absent from all cases where the claim does NOT hold? No — plenty of "studies show" claims have turned out to be false (the spinach-iron myth, the beta-carotene-vitamin story).
So the reason "studies show" does NOT pass the three characteristics as deployed. You cannot validly conclude. This is not a failure of intelligence; it is a recognition that the headline did not give you a reason strong enough to support its conclusion. The Dignāgan phrase for what you have just done is: "the exclusion (遮) has not been established; the cognition of 'coffee extends lifespan' has not arisen as svārthānumāna."
Where people map it wrong: they confuse finding a study with concluding. They think "if I can cite a study, the reasoning is done." But the three characteristics demand more — the cited study's reason must in fact support the conclusion under the proper conditions. The intermediate step is what most people skip, and it is the entire heart of svārthānumāna.
2.3 The friend's silence
He said he would text back by evening. He hasn't. You want to know: is something wrong, or is he just busy?
This is where the non-apprehension reason (未緣到因) shows its power. The classical structure:
"If he had something to communicate, and was in a position to do so, his message would be apprehensible to me. But it is not apprehended."
The conditional "if X, then Y" must itself hold. Here it requires testing: when he has been busy before, has he texted? Has he texted during prior periods of busyness? If yes, then "busy" is not a sufficient condition for silence, and the conditional fails. If no — when he is busy, he is consistently silent; when he has something to say, he consistently speaks — then the conditional holds, and the non-apprehension gives you a valid cognition of absence.
You may then conclude: "there is no message." This is a real fact about your inbox. It is a piece of svārthānumāna that holds.
Where people map it wrong: they let the conditional smuggle in unexamined. They assume "if he cared, he would text," which often encodes a self-flattering sapaksha/vipaksha confusion — they count only past behaviors that fit their preferred conclusion as "real" data and discount the rest as "atypical." The Dignāgan discipline is: do not let your preferred conclusion determine what counts as a similar or dissimilar instance. Check the sapaksha/vipaksha honestly, or the conclusion is not yours — it is a fantasy wearing your name.
2.4 The label on your feeling
You say, "I am angry." But are you? You have an impulse to snap at someone. Is that anger, fear, hunger, exhaustion, grief?
The Buddhist introspective tradition — which Dignāga's epistemology underwrites, even though the text does not say so explicitly — runs the three-characteristic test on its own states.
- Paksha: This specific impulse, in this moment. - Reason: The candidate label — "anger." - Sapaksha: Is the property "having this felt quality" present in all past instances where you were genuinely angry? Was it always hot, urgent, tied to a sense of being wronged? - Vipaksha: Is the property absent in all non-anger states? Does hunger never produce a hot urgent feeling of being wronged? Does exhaustion never?
Often the vipaksha test fails. Hunger, exhaustion, fear, and grief can all produce anger-shaped impulses. The label "anger" was a premature sapaksha compression — you treated the first familiar feature as the whole story. The Dignāgan discipline is to slow down, apply the three tests, and let the correct label arise (or fail to arise).
When you do this — when you actually arrive at the right label through the three characteristics — the resulting cognition is "the same as perception." It is not an inference anymore. It is direct acquaintance with what is happening. This is svārthānumāna at its most intimate: knowing your own mind with the precision of seeing.
三、Why contemporary people need this concept
3.1 We are drowning in arguments
The defining feature of contemporary information ecology is unlimited access to parārthānumāna — arguments set out for us. Every feed, every podcast, every Substack, every WhatsApp forward is an inference-for-others, polished and pitched. The supply of arguments vastly exceeds our capacity to evaluate them.
The Dignāgan response to this is not "evaluate every argument." It is "develop svārthānumāna so that you, by yourself, know which arguments yield cognition and which yield only the appearance of cognition." The skill is not reading more carefully; it is internally running the three-characteristic test so that the conclusion either arises as exclusion (and you know it) or does not arise (and you know you do not yet know).
In an age of infinite persuasion, this is one of the few genuine epistemic skills. It is the difference between being well-read and being well-concluded.
2.2 The confusion between knowing and defending
Modern discourse, especially online, has trained us to conflate arriving at a conclusion with being able to defend one. People who can argue well are assumed to know; people who cannot are dismissed.
But svārthānumāna and parārthānumāna are different acts. You can argue persuasively for a conclusion you do not actually hold as cognition. You can also silently arrive at a conclusion you cannot articulate to others. Svārthānumāna is silent and private; parārthānumāna is vocal and public. Conflating them is a category error that the 《因明入正理论》 is built precisely to prevent.
The contemporary lesson: when you cannot articulate why you believe something, that does not mean the belief is wrong. It may mean you have not yet developed parārthānumāna for it. Conversely, when you can articulate why you believe something, that does not mean you have actually arrived at it. It may mean you have only memorized an argument.
The Dignāgan test is internal: has the cognition of exclusion actually arisen in your own mind, by your own reason, possessing the three characteristics? If not, the belief is not yet yours — regardless of how eloquently you can perform it.
2.3 The hidden skill of habituated clarity
The text's most overlooked line is about habituation: "由屢見立量而串習." Svārthānumāna develops through repeated exposure to well-formed arguments. This is a beautiful practical insight for an age of intellectual formation.
If you read well-argued essays regularly — essays where the author carefully distinguishes sapaksha from vipaksha, where they flag the conditional in their non-apprehension arguments, where they notice when their reason is self-nature versus effect — you will, over time, internalize those moves. Your own internal reasoning will start to use them without conscious effort. This is what the text means by "串習" (familiarization-training): not memorization, but the gradual absorption of a discipline into the shape of your thinking.
The implication for contemporary self-cultivation: choose what you read and listen to with an eye to whether the reasoning is well-formed, not just whether you agree. The point is not the conclusion. The point is that you are training your own svārthānumāna by witnessing parārthānumāna done well or poorly.
2.4 As an antidote to motivated reasoning
Modern cognitive science has documented what the Buddhist tradition already knew well: we routinely generate reasons *after* arriving at conclusions. We use reasons to defend what we wanted to believe. This is the failure mode that the three characteristics are designed to expose.
The three characteristics are not designed for the easy cases. They are designed for the moments when you want to believe X. They force you to ask: is the reason I am about to use actually present in the subject? Is it present in all similar instances? Is it really absent in all dissimilar instances? The questions are uncomfortable by design. They are uncomfortable because motivated reasoning wants to skip them.
In Dignāga's language, motivated reasoning is parārthānumāna aimed at oneself — the rhetorical argument you set out, fully formed, just to convince yourself. The three characteristics are the discipline that prevents the rhetoric from counting as cognition.
四、Common misreadings and clarifications
4.1 "自利比量 is just subjective opinion"
This is the most common collapse, and it is wrong on every term.
Svārthānumāna is not "what I personally think." It is the cognition that arises from a reason possessing the three characteristics. The three characteristics are not subjective; they are structural conditions on the reason. Sapaksha and vipaksha are public: all sandalwoods, all non-sandalwoods, all fires with smoke, all smokes-without-fire-context, and so on. You check these by observing the world, not by checking your feelings.
Subjective opinion is precisely what svārthānumāna is not. Subjective opinion is an attitude prior to reasoning. Svārthānumāna is the cognition that arises *from* a properly structured reason — it is post-reason, not pre-reason.
4.2 "自利比量 is the same as parārthānumāna, just without speaking"
This is closer to the truth, but still collapses an important distinction.
The text's own definition: 「自利系就个人进行思惟而言，利他则就对他人以语言建立为论式所收效果而言」 — "svārthānumāna concerns personal contemplation; parārthānumāna concerns the effect produced by setting up an argument in language for another person."
The distinction is not just about speaking. It is about the orientation of the cognition. In parārthānumāna, the cognition is structured around producing a specific effect in a listener — it has rhetorical commitments (clarity, sequence, persuasion). In svārthānumāna, the cognition has no rhetorical commitments; it has only epistemic ones.
In practice, the two often occur together: you set out an argument for another, and as a side effect, you yourself also arrive at the conclusion (or vice versa). But they are conceptually distinct, and conflating them obscures why the 《因明入正理论》 bothered to separate them in the first place.
4.3 "The three characteristics are a bureaucratic checklist to apply mechanically"
This is the academic failure mode. Students of logic often treat the three characteristics as a procedure to run on candidate reasons, the way one runs a syntactic check on a program.
The Dignāgan text itself suggests something different: it is a discipline of attention. The three characteristics are conditions you must internalize until they shape what you spontaneously notice. Once internalized, you do not "apply" them — you simply do not see a reason as valid unless it passes them. The transition is from checking to seeing.
This is also what the text means by 「串習」 — the gradual transformation of procedure into perception. The goal is that svārthānumāna "is the same as perception" — the three-characteristic check has been absorbed into the structure of the cognition itself.
4.4 "自利比量 only applies to Buddhist topics"
A common misreading, given that the text comes out of a Buddhist tradition. But the formal structure of svārthānumāna is general. You can run the three-characteristic test on reasons about chemistry, career choices, friendships, weather predictions. The Buddhist tradition uses it to discuss doctrine, but the discipline is not confined to doctrine.
The restriction, if any, is on the *purpose*. Dignāga explicitly develops pramāṇa theory as the foundation for the path to liberation — valid cognition is what enables correct practice. So while the formal tool is general, the motivation within the tradition is soteriological: you want svārthānumāna about Buddhist claims because you want your liberation-practice to be grounded in valid cognition rather than wishful thinking.
4.5 "The eleven forms are esoteric knowledge that only specialists need"
You do not need to memorize the eleven forms. What you need to understand is the punchline: they all reduce to self-nature non-apprehension — the act of excluding what does not belong.
This is actually an encouraging conclusion, not an obscure one. It means that no matter how elaborate an inference looks, it is doing one simple thing: holding up the reason and seeing what must be excluded. Once you see that, the elaborate lists become unnecessary scaffolding. You can return to the simple operation.
In contemporary language: every good argument, no matter how many moves it makes, is doing one thing — drawing a line that excludes what does not belong to the conclusion. The eleven forms are just different ways of drawing that line.
4.6 "Self-nature reasons are circular"
A persistent worry: "This is a tree, because it is a sandalwood tree" — isn't that circular? Doesn't the conclusion just repeat the reason?
The classical answer is no. The reason and the conclusion are not the same proposition. "Has sandalwood-nature" and "is a tree" are co-extensive in this case, but they are different categorizations — one is a more specific identity, the other is the broader class. Knowing the more specific identity logically entails knowing the broader class, but it is not the same as knowing it from the start. The self-nature reason is informative when you are reasoning from a particular observation to a class membership. The cycle is broken by the asymmetry of specificity.
In other words: "this is a tree" can be a genuine new cognition for someone who knew only "this is a sandalwood-natured thing." The reason carries them from particular to general. It is not circular.
A closing thought
The Dignāgan project of svārthānumāna is, at heart, a project about the inner integrity of cognition. It is not enough to argue well; one must also *arrive*. Not enough to defend; one must *know*. Not enough to perform exclusion in language; one must perform it in one's own mind.
In an age of infinite argument and exhausted attention, this is a quietly radical teaching. The classical texts give us the formal apparatus — three characteristics, three reasons, the reduction to one act of exclusion, the cultivation through habituation. The contemporary task is to take that apparatus seriously as a discipline of self-knowledge, not as a museum exhibit.
The result, when it works, is what the text calls "the same as perception": a cognition that has the cleanness and directness of seeing, but on behalf of what could not be seen. That is what svārthānumāna is for.