What This Concept Is About
Imagine you taste a spoonful of sugar. The sweetness is not something *added* to the sugar after the fact; it is not an *effect* the sugar produces, nor a *correlation* your tongue has been trained on. Sweetness is the very nature of what sugar is. You can know sugar is sweet because the sweetness is *already* in the sugar itself; you do not need to wait for sugar to do something else to be detected.
This pattern of reasoning — where the evidence is not a downstream signal of the thing, but the very nature of the thing made visible — is what the Indian Buddhist logician Dharmakīrti (法称), in his *Nyāyabindu* (《正理滴論》), calls 自性因 (svabhāva-hetu): the *self-nature reason*, or *intrinsic-property reason*. It is one of only three kinds of valid reason (正因) the text recognizes, alongside 未緣到因 (anupalabdhi-hetu, the *non-apprehension reason*) and 果因 (kārya-hetu, the *effect reason*).
All three must pass the same test, the famous 三相 (triśaṇā, "three characteristics") — formulated classically in *Nyāyapraveśa* (《因明入正理論》) by Śaṅkarasvāmin (商羯罗主) as 「遍是宗法性，同品定有性，异品遍无性」 and rendered in the *Nyāyabindu* as:
1. 於所比度之事上必須具備 — the reason must be present in the subject of the inference (the *paksā*). 2. 於同品定有 — the reason must invariably be present in all similar instances (*sapakṣa*). 3. 於異品遍無 — the reason must be entirely absent in all dissimilar instances (*vipakṣa*).
What distinguishes *svabhāva-hetu* from the other two is not whether it passes this threefold test — all valid reasons do — but *how* the reason relates to the thing being proved. That relation is the entire game.
The Core Definition
The classical formulation is stark: 「唯自具有所成立法之因」 — *"the reason that, by itself alone, possesses the property to be established."* The *hetu* (the evidence you cite) is, by its own nature, already an instance of the *sādhya* (the thing to be established). You do not need it to *produce* or *signal* that property; you cite it because the property is *contained in it* as part of its own self-nature (svabhāva).
The root principle is stated even more bluntly: 「彼為彼之自性，其自性亦為因」 — *"That is the self-nature of that, and its self-nature is also the reason."* The relation is one of identity-of-nature, not sequence, not signal.
The Canonical Example
The text offers this as the foundational instance:
「此是樹木，是沉香樹故。」 *"This is a tree, because it is a sandalwood tree."*
Why is this valid? Because *being-a-tree* is part of the *svabhāva* (self-nature) of a sandalwood tree. To point to "sandalwood-treeness" is already to point to "tree-ness." The latter is *contained in* the former, not produced by it, not correlated with it. If you know the subject is a sandalwood tree, you already know — without any further inference-step about causation — that it is a tree.
This is in fact the simplest and most uncontroversial form of inference in the Buddhist logical tradition. It is what happens when the evidence is the thing itself, viewed from a narrower angle.
The Three Articulations of the Same Reason
The *Nyāyabindu* distinguishes three technical ways the *same* logical form — *svabhāva-hetu* — can be expressed in language. They form a progression of increasing specification:
1. 「凡有，皆是無常，猶如瓶等」 — the 純自性因量式 ("pure self-nature syllogism"): *"Whatever exists is impermanent, like a pot."* Here *existence* (有, bhāva) is cited as the reason; impermanence is contained in the svabhāva of whatever exists. 2. 「凡有生，皆是無常」 — the 自性差別法之自性因量式 ("self-nature-as-differentiator syllogism"): *"Whatever has birth is impermanent."* *Birthedness* (有生) is a more specified form of *existence* — every existent has birth, and impermanence is contained in birthedness. 3. 「凡所作，皆是無常」 — the form using 所作性 (saṃskāra, "the conditioned / the made") as a special differentiator: *"Whatever is made is impermanent."* The text explains that *saṃskāra* is singled out in this way because, in establishing its own nature, it further requires reference to *a maker*. There is one more layer of specification, but the svabhāva-relation to impermanence remains intact.
All three forms do the same work: they cite something whose svabhāva already contains impermanence.
The Pervasion Principle
The text says of *svabhāva-hetu*: 「自性因，乃由自性所遍」 — *"the self-nature reason is pervaded by self-nature."* This is the principle of 遍充 (*vyāpti*, pervasion): wherever the reason is, the property-to-be-established must be. You cannot talk loosely about whether the reason appears in similar or dissimilar cases. The text's phrase 「於何物而無何性者」 ("that on which some nature is absent") states the contrapositive: a nature that does not pervade its domain cannot be invoked as a self-nature reason.
The canonical positive and negative expressions of this:
- 隨因後行 ("following upon the reason," the positive direction): 「凡所作性，必有無常性」 — *"Whatever has saṃskāra, must have impermanence."* - 隨因遣行 ("removed along with the reason," the negative direction): 「若無無常，有、有生或所作性亦必然無」 — *"If there is no impermanence, then existence, birth, or saṃskāra is necessarily absent as well."*
Notice that the second direction works only because of the svabhāva relation. The text is explicit: *if two things are not related as svabhāva, the removal of one does not necessitate the removal of the other.* In *svabhāva-hetu*, the biconditional is real; in mere correlation or causation, it is not.
The Distinction From Effect Reason
This is where *svabhāva-hetu* earns its precision. The *Nyāyabindu* states that the relation between a reason and the property-to-be-established can only be one of two kinds:
1. The reason is directly the svabhāva of the property-to-be-established → *svabhāva-hetu*. 2. The reason arises from the property-to-be-established → *kārya-hetu* (effect reason).
The smoke-from-fire case is the textbook *kārya-hetu*: smoke is not the *svabhāva* of fire; smoke is *produced by* fire. Fire's nature is not "having-smoke"; smoke is its downstream product. The two are related, but as producer and produced, not as nature and its intrinsic feature.
Anything that is neither the svabhāva of X nor produced from X is, the text says, "without *相系屬* (relevant connection)" to X and cannot be used to establish it.
This is precisely why the *sandarbha*/tree example is so clean: there is no temporal or causal sequence involved. The reasoning is instantaneous and identity-based.
Walking Through This in Life
Classical Indian logicians called the application of a principle *prayoga*, literally "going forward" or "running through." Let us run *svabhāva-hetu* through scenes unmistakably contemporary, holding strictly to the classical definition: the reason cited must, by its own svabhāva, already contain the property-to-be-established, and the three characteristics must hold.
Scene 1: Browsing a bookstore
You pick up a paperback. You read the back cover: "a noir detective novel." You conclude: *this is a work of fiction.*
Is this *svabhāva-hetu*? Yes. Reasoning: *"This is fiction, because it is a novel."* Being-a-novel is the kind of thing that *already contains* being-fiction as part of its self-nature. A novel, by what a novel *is*, is fiction. You do not need to read the book to know; the genre-label itself carries the property.
Three characteristics: - *On the subject* (this book): being-a-novel is present. ✓ - *In similar cases* (all novels): fiction is invariably present. ✓ - *In dissimilar cases* (all non-fiction): being-a-novel is absent. ✓
You have just used *svabhāva-hetu* without naming it.
Scene 2: Why "smoke means fire" is *not* this kind of reason
Now contrast with the smoke case. You smell smoke in a stairwell. You conclude: there is fire.
This is *kārya-hetu*, the effect reason. "Smoke" is not the svabhāva of "fire." Smoke is *produced from* fire. You are inferring fire from its downstream effect. The relation is causal, not identity-based.
This matters in daily life: when someone says "X causes Y," they are usually not making a *svabhāva* claim; they are making a *kārya* claim. The two kinds of evidence need different scrutiny. In *svabhāva-hetu*, the inverse always holds: if the property is absent, the reason must be absent too. In *kārya-hetu*, the inverse is trickier — absence of smoke does not always entail absence of fire (a fire can be smokeless).
Scene 3: A doctor's bedside
A physician palpates a fixed, hard, irregular mass in the breast. She writes in the chart: "likely malignancy." Her reasoning: "This is likely malignant, because it is a fixed, hard, irregular mass."
Is this *svabhāva-hetu*? Strictly, no. The clinical features she palpates do not *by themselves* constitute malignancy; they are *signals that correlate with* malignancy. This is closer to *kārya-hetu* (the mass-features as effect of malignancy) or, in some framings, *anupalabdhi-hetu* (the non-apprehension of normal-tissue features as the reason). Most clinical reasoning is in fact *kārya*-style or *anupalabdhi*-style, not *svabhāva*-style. This is why medical inference is probabilistic and revisable.
Where *svabhāva-hetu* does appear in medicine: in *definitional* claims. "This is a fracture, because it is a discontinuity in bone." A discontinuity in bone is part of what *fracture* means; cite one, you cite the other. That is *svabhāva-hetu*.
Scene 4: Diagnosing a software bug
You open source code. You see `var ratio = y / 0;`. You conclude: *"this will throw a runtime error."*
Is this *svabhāva-hetu*? Yes — in a language where division by zero is *definitionally* an error-producing operation. The error is contained in the self-nature of the operation as defined by the language. Cite the operation, and the consequence is already known, without running the code.
But suppose instead you see the program crash and infer "there's a bug somewhere." That is not *svabhāva-hetu*. The crash is an *effect* of an unknown cause. You are doing *kārya*-style reasoning under uncertainty, and the three characteristics do not robustly hold.
Scene 5: Spotting your own emotion
You notice your jaw clenching, your shoulders rising, your breathing shallow. You conclude: *"I am anxious."*
Here the trap is real. If you say "I am anxious *because* my jaw is clenched," you risk treating the jaw-clenching as a downstream *signal* of anxiety, which would be *kārya-hetu*. But if you instead say "I am anxious, *as evidenced by* the very way my body is holding itself right now" — where the way-of-holding is taken as part of the *self-nature* of the anxiety-state — you are doing something closer to *svabhāva-hetu*.
In contemplative traditions this matters a great deal. The instruction "notice anxiety in the body, not as a signal of something else, but as anxiety itself" is precisely a training in recognizing *svabhāva*-relations within one's own experience.
Scene 6: An AI gives you a confident answer
You ask a large language model a factual question. It produces a fluent, confident answer. You conclude: *"this is probably correct."*
This is *kārya-hetu* (the fluent production as downstream effect of training on correct material) — and a dangerously weak one, because fluent production can equally be the effect of training on plausible-but-wrong material. The three characteristics do not hold: fluent-form is *not* invariably present only in correct cases. "Confident" is not the svabhāva of "true."
By contrast, when a program *executes* code you have verified and you observe its deterministic output, the relation *is* svabhāva-style: that output *is* the svabhāva of running that code on that input. This is why execution-based testing is more reliable than asking an LLM to inspect the code.
Why Contemporary People Need This
We live in an information environment saturated with inferences, most of them never examined. The *Nyāyabindu*'s framework, and *svabhāva-hetu* specifically, gives a vocabulary for sorting them.
1. To tell when evidence is "from the inside" versus "from downstream"
This is the most practical use. When someone offers evidence for a claim, the first question is: *is the evidence part of what the claim is, or is it a downstream effect of the claim?*
If someone says "He is a good father because his kids are well-adjusted," the kids' adjustment is a *kārya* (effect) of fathering, not part of the svabhāva of good-fathering. The evidence is downstream; it can fail (well-adjusted kids can have terrible fathers; well-adjusted kids reflect many factors). The inference is weak.
If someone says "He is a good father because he reads to his children every night," that is different. Reading nightly is more directly part of what good-fathering *is*. The evidence is closer to *svabhāva*. (Even here, one can quibble: reading nightly is not the totality of good-fathering, and one could mechanically read without engaging. The text would call this a *partial-svabhāva* use — legitimate but limited.)
2. To recognize when a claim is unfalsifiable in the wrong way
A contemporary anxiety: people make claims that no evidence could touch. The *Nyāyabindu*'s 隨因遣行 principle (the contrapositive: if X is absent, the reason is absent too) is a powerful diagnostic. If you cannot even articulate *what would show the claim to be false*, you do not have a *svabhāva* reason; you have something much weaker — an unanchored assertion. *Svabhāva-hetu* therefore functions as a kind of discipline for whether your claim is the kind of claim that evidence can, even in principle, speak to.
3. To use the three characteristics as a debugging tool
Before accepting a piece of evidence, walk it through: - Is it really present *in the case at hand*? - Is it really present *in all similar cases*? - Is it really *absent in all dissimilar cases*?
Most everyday reasoning fails one of these. "He must be guilty — he looks shifty." Present in the case? Maybe. Present in all similar cases (all guilty people)? No — most guilty people do not look shifty, and many innocent people do. So the second characteristic fails. The inference is invalid.
4. To ground identity-claims in a non-tautological way
It might seem that *svabhāva-hetu* is just saying "X is X" — a tautology. It is not. The *sandarbha* example shows why: "this is a tree *because* it is a sandalwood tree" tells you something — that *sandarbha*-property *contains* tree-property in its svabhāva. This is only useful when the broader property (tree-ness) was not directly evident but the narrower one (*sandarbha*-ness) was. It is, in fact, a model for how we actually come to know many things in the world: by spotting the specific and recognizing the general.
5. As a corrective to the algorithmic age
Recommendation algorithms, news feeds, and large language models all blur *svabhāva*, *kārya*, and mere correlation. When a recommendation system shows you an item "because you liked X," it is performing a *kārya*-style inference from a behavior to a preference — and that inference is weak. When an AI confidently produces an answer, it is performing no *svabhāva*-style inference at all: the answer is generated from patterns, not from svabhāva. *Knowing the difference* between evidence that *is* the thing and evidence that is merely *produced by* the thing is a contemporary epistemic hygiene skill.
A note of caution: this framework comes from 7th-century India. It can illuminate, but it should not be over-stretched. Cognitive science does not currently endorse any clean identity of "svabhāva" with neural states, and there is no scientific consensus yet on how concepts like "tree" or "impermanence" are realized in the brain. The Nyāya framework is a *logical* tool — it tells you the structure of your inference, not the metaphysics of what is being inferred.
Common Misreadings and Clarifications
Misreading 1: "Self-nature reason" = "self-evident reason"
In contemporary English, "self-nature" gets misread as "self-evident" or "intuitive." This is wrong. *Svabhāva* in *svabhāva-hetu* is not about *your psychological certainty*; it is about the *ontological relation* between the reason and the thing-to-be-established. Sandalwood being a tree is not "self-evident" in any psychological sense — many people do not know what sandalwood is. But it is a *svabhāva* relation: sandalwood-treeness *contains* tree-ness as part of what it is. The evidence does not have to feel obvious; it has to *be* part of the thing's own nature.
Misreading 2: "It is just a definition / just a tautology"
This confuses a partial truth with the whole. *Svabhāva-hetu* is *more than* definition: it is a *use* of definition in inference, where you knew the definition but not the application. "Sandalwood tree ⇒ tree" looks tautological in retrospect, but the very point of the *Nyāyabindu* is that we frequently *don't* have direct access to the broader property and must reason through the narrower one. The inference is real, even if it looks like a tautology when written out.
Misreading 3: Conflation with *kārya-hetu* (effect reason)
The most common contemporary error. Many arguments that *feel* like self-nature reasons are actually effect reasons. "He must be lying because he is sweating" — sweating is an *effect* of lying (and of many other things); the svabhāva of lying is not sweating. "She is in love because she is glowing" — same issue. The text's insistence that the relation must be one of exactly two kinds (*svabhāva* or *produced-from*) is meant precisely to prevent this collapse.
Misreading 4: Treating *svabhāva* as a permanent metaphysical essence
In Indian philosophy, *svabhāva* has many uses, and in some Brahmanical schools it denotes an *eternal, intrinsic essence* (the *svabhāva* of fire is to be hot, eternally). Buddhist logicians like Dharmakīrti use the term more carefully: as the *defining nature* of a thing, *including* its impermanence. The text's use of impermanence as the property-to-be-established (*sādhya*) shows this clearly: impermanence is *part of* the svabhāva of conditioned things, not a contingent property. So when you hear *svabhāva*, do not read it as "eternal essence" — read it as "defining nature," which may include change.
Misreading 5: Assuming it requires no empirical work
Because the *sandarbha*/tree inference looks immediate, one might think *svabhāva-hetu* bypasses experience. Not so. To know that sandalwood trees are a *kind* of tree, you have had to observe. To know that "whatever is made is impermanent" requires that you have actually watched things age and break. The *svabhāva* relation, once articulated, *feels* instantaneous, but it is grounded in prior observation. The *Nyāyabindu* is not anti-empirical — it is describing how empirical content gets packaged into inferential form.
Misreading 6: Treating the three articulations as three different reasons
The three forms (純自性因量式, 自性差別法之自性因量式, and the *saṃskāra* form) are not three different kinds of reason. They are three *linguistic articulations* of the same logical form — *svabhāva-hetu* — with increasing specification of the reason-property. *Existence → birth → conditioned-ness* — each is a narrower specification that still carries impermanence in its svabhāva. The progression shows that the framework is sensitive to *how* a property is articulated in language without losing its *svabhāva*-relation to the property-to-be-established.
A Final Image
Imagine a glass prism. White light enters. Colors emerge — not because the prism *added* them, but because the colors were *already there*, contained in the nature of white light. To see "red" coming out of the prism and conclude "the light was white-with-red-in-it" is to use a kind of *svabhāva*-reasoning. The color is not an *effect* of the prism's action; it is a *manifestation* of something already present.
*Svabhāva-hetu* is the logic of manifestation, not the logic of production. It tells you when you are looking at the unfolding of something that was already there, rather than at the consequence of something that just happened. In a world that constantly offers us consequences — smoke, crashes, glowing faces, fluent AI — the discipline of asking "is this evidence a manifestation of the thing itself, or a downstream effect?" may be the most quietly radical gift of a 7th-century logician to a contemporary attention.