The school that asked what makes a belief count as knowledge, and then spent eighteen centuries refining the answer. Its four means of knowing, its five-membered inference and its catalogue of bad arguments became the shared property of Indian philosophy — you cannot read a Buddhist, a Jaina or a Vedāntin without meeting Nyāya's vocabulary, usually being turned against Nyāya.
Also called
Ānvīkṣikī आन्वीक्षिकी, “the science of enquiry”; Pramāṇa-śāstra; Tarka-śāstra; Hetu-vidyā
Root text
Nyāya-sūtraन्यायसूत्र of Gautama (Akṣapāda), c. 2nd century CE, in 5 books of 2 chapters each — 528 sūtras
Read the sūtra backwards: “Release follows from the successive removal of suffering, birth, activity, faults and false knowledge, each disappearing on the disappearance of the one after it.”
Why Nyāya's tools became everyone's tools
A school whose central claim is that correct reasoning liberates has to make its account of correct reasoning public, explicit and defensible against anyone. So Nyāya wrote down what an argument is, what counts as evidence, and what makes an argument fail — in a form that its opponents could use. They promptly did. By the seventh century a Buddhist logician and a Naiyāyika disagreed about almost every substantive question while sharing a technical vocabulary precise enough to conduct the disagreement in.
2The lineageपरम्परा
Note the shape: eight centuries of commentary on one text, then a restart. Gaṅgeśa does not reject the Nyāya-sūtra; he simply stops commenting on it and writes a fresh treatise on the four pramāṇas, and after him almost nobody comments on Gautama again.
3The sixteen categoriesषोडश पदार्थाः
The very first sūtra lists sixteen things, knowledge of which produces the highest good. The list is unlike any Western catalogue: it mixes ontology, method, and the etiquette of debate in a single enumeration, because for Nyāya these are one subject.
The twelve knowables
The second category, prameya, gets its own list of twelve — and this list is Nyāya's ontology-in-brief, arranged as the anatomy of a person in bondage.
ātman the selfśarīra the bodyindriya the sensesartha their objectsbuddhi cognitionmanas the inner organpravṛtti activitydoṣa the faultspretya-bhāva rebirthphala the fruit of actionduḥkha sufferingapavarga release
Read in order, the twelve run from what you are, through the equipment you have and what it meets, to what goes wrong and how it ends. It is the same chain as §1, unpacked.
Indeterminate perception: bare awareness of a thing and its qualities, before they are related to one another or to a name. Not itself expressible, and not itself doubted or affirmed.
The problem: nobody ever reports having one. Nyāya's reply is that it is inferred, not observed — the determinate perception “that is a cow” presupposes prior acquaintance with cowness and with the animal, and that prior acquaintance is the indeterminate stage.
Savikalpaka सविकल्पक
Determinate perception: the same object now grasped as qualified — this is a cow, where a universal, a name and a substance have been brought into one structure.
Nyāya insists this is still perception, not judgment added to perception. What is presented is a structured fact, and its structure is out there, not supplied by the mind. Buddhists deny exactly this, and the disagreement runs through everything else.
Three kinds of extraordinary perception
Beyond the ordinary, Nyāya admits three alaukika contacts, and each solves a specific technical problem.
Sāmānya-lakṣaṇa — perceiving all smoke whatsoever through the universal smokeness present in this smoke. This is Nyāya's answer to the problem of §7: how a finite observer can grasp a universal relation.
Jñāna-lakṣaṇa — perceiving a property through a revived memory, as when you see that the sandalwood is fragrant. This is what carries the whole weight of Nyāya's theory of error in §11.
Yogaja — the perception attributed to advanced practitioners, of distant, past, future and subtle things. Formally required so that omniscience is possible; rarely doing work in an argument.
5Six ways to touch an objectषट् सन्निकर्षाः
If perception is defined by contact, the theory must say what contacts what. Nyāya's answer is a list of six, and it is a good test of whether the Vaiśeṣika ontology of the next page actually works.
The list is generated by the ontology, not collected from experience: given substances, qualities, universals, ether and absences, exactly these six relations are available. That the list closes neatly is the argument for the ontology.
The Buddhist objects
Absence is nothing. There is nothing there to be contacted, so there can be no perception of it. What you call “seeing the absence of the pot” is an inference: you look, you fail to find, and you conclude.
Nyāya replies
Then say what the inference is. An inference needs a perceived mark. If the mark is “failure to perceive the pot”, that failure is itself an absence, and you have helped yourself to exactly what you denied. Better to admit that a bare floor looks different from a floor with a pot on it, and that the difference is seen.
Mīmāṃsā intervenes
Neither. Absence is known by its own means, anupalabdhi — non-apprehension — which is a sixth pramāṇa and reduces to neither of yours.
6Inferenceअनुमानम्
Nyāya's most influential work. The account has three parts: what an inference is made of, what makes it valid, and how it is stated to another person.
The isolation of parāmarśa is characteristic. Nyāya is not describing a logical form but a causal process in a knower, and it wants to name the exact mental event without which the conclusion does not occur.
Three ways to classify an inference
Scheme
Kinds
The distinction
By purpose
svārtha · parārtha
For oneself, where no statement is needed; or for another, where the five members of §8 must be set out.
By direction Nyāya-sūtra 1.1.5
pūrvavat · śeṣavat · sāmānyato-dṛṣṭa
From cause to effect (clouds, therefore rain); from effect to cause (a swollen river upstream, therefore rain); or from a general correlation with no causal link at all (the sun has moved, therefore it travelled).
By the vyāpti available
anvaya-vyatireki · kevalānvayi · kevala-vyatireki
Whether both positive and negative examples exist; only positive (nothing anywhere lacks nameability); or only negative (nothing else can be produced to compare with).
The third scheme matters most, because it is where the theory strains. A kevalānvayi inference has no counterexample class at all — there is nothing that is not nameable, so “whatever is not nameable is not knowable” has no instances to check. A kevala-vyatireki inference has no positive instance — Nyāya's own argument that earth is distinguished from everything else by smell has only negative support. Buddhist logicians attacked both, and Navya-nyāya spent centuries shoring them up.
7The problem of vyāptiव्याप्तिग्रहः
Everything depends on a universal relation. You have observed finitely many cases. This is the problem of induction, and Indian philosophy found it early and never let it go.
Cārvāka states the difficulty
You have seen smoke with fire in kitchens and in a few hills. You have not seen every case of smoke. To move from “all observed smoke” to “all smoke” you would need another inference, whose own pervasion would need establishing, and so on. Inference never gets off the ground.
Nyāya's first answer: repeated observation
Repeated concomitance with no observed exception (bhūyo-darśana), supported by the failure to find any defeating condition. Not proof, but reasonable.
And the difficulty returns
“No exception observed” is not “no exception”. This is the counting argument again, in a smaller font.
Nyāya's real answer: tarka and the universal
Two moves. First, tarka: suppose there were smoke without fire. Then smoke would arise without its cause, and effects would arise causelessly, and nothing would be reliably produced by anything — an unacceptable consequence. Tarka cannot prove the pervasion, but it removes the live doubt about it. Second, sāmānya-lakṣaṇa perception: in perceiving this smoke you perceive smokeness, and through smokeness all smoke whatsoever. The relation is grasped between the universals, not tallied over the instances.
Gaṅgeśa on defining pervasion
The opening of the second book of the Tattva-cintāmaṇi is one of the most concentrated pieces of philosophical writing in any language. Gaṅgeśa states a definition of vyāpti, finds a counterexample, repairs it, finds another, and repeats — through roughly twenty-one attempts, each rejected for a stated reason — before offering his own. The section is called the Siddhānta-lakṣaṇa, and it is where Navya-nyāya's technical language was forged: the successive repairs needed a vocabulary for saying exactly which property, qualified by exactly which limitor, is being related to which other. §14.
8Why five membersपञ्चावयवः
Chapter 7 of this series introduced the five-membered statement. Here is why the number is five, and why Nyāya refused to reduce it.
9The five fallaciesहेत्वाभासाः
A hetvābhāsa is a reason that only looks like one. Nyāya lists exactly five, and the list is not collected from experience — it is derived from the ways a pervasion can fail.
Each corresponds to one way the pervasion in §7 can fail: it may be untrue, run the other way, be matched by a rival, be unavailable because the reason is absent, or be overridden by a stronger pramāṇa. The list is complete because those are the only options.
10Comparison and testimonyउपमानं शब्दश्च
Upamāna उपमान — comparison
The standard case: a townsman is told that a gavaya is like a cow. He goes to the forest, sees an animal resembling a cow, and knows — this is what the word gavaya denotes.
What exactly is known? Not that the animal exists, which is perception. Not that it resembles a cow, which is also perception. What is new is the relation between a word and a thing, and Nyāya argues that no other pramāṇa delivers it.
Why others disagree: Vaiśeṣika and Buddhism fold it into inference; Mīmāṃsā accepts it but analyses it differently, as knowledge of the cow's similarity to the gavaya rather than the reverse.
Śabda शब्द — testimony
The statement of a reliable speaker — āpta: one who knows the matter, and reports it without distortion.
The Nyāya twist: for Nyāya the Veda is authoritative because it has a reliable author, namely God. This is the exact reverse of Mīmāṃsā, which holds that the Veda is authoritative precisely because it has no author and so no source of error. The two schools agree the Veda is a pramāṇa and disagree completely about why.
Four requirements for a sentence to convey meaning: expectancy (ākāṅkṣā), compatibility (yogyatā), proximity (sannidhi), and — disputed — intention (tātparya).
11What happens in a mistakeख्यातिवादः
You see mother-of-pearl and take it for silver. Something is presented to you. What? Every school had to answer, and the answers divide them more cleanly than almost any other question.
Nyāya's own position is the most conservative and, for that reason, the most exposed. Anyathā-khyāti keeps everything real — the shell is real, the silver is real, perception genuinely presents both — and locates the error entirely in a wrongly asserted relation. The cost is the machinery of §4's extraordinary perception, which the rivals regard as invented for the occasion. The benefit is that Nyāya never has to say that anything appearing to anyone is unreal, which is precisely the concession Advaita needs and Nyāya refuses to give.
12The proofs of Godईश्वरसिद्धिः
Nyāya is the Indian school that argues for God's existence, in the ordinary philosophical sense of offering premises. Udayana's Nyāya-kusumāñjali — “a handful of flowers of reason” — collects the arguments, and a single verse lists them.
Nyāya's Īśvara is an efficient cause only. He does not create the atoms, which are eternal, and he does not create souls, which are eternal, and he does not create karma, whose fruits he administers rather than decides. He arranges eternal materials according to the deserts of eternal souls. That is a much smaller claim than the one the word “God” usually carries in English, and it is why Nyāya's theism could coexist with an ontology in which almost everything is uncreated.
13Self, body and releaseआत्मापवर्गौ
The self is a substance
Not consciousness, but a thing that has consciousness as an adventitious quality — one it acquires when in contact with a body and a mind, and loses when it is not. Selves are many, eternal and all-pervading.
The argument: desire, aversion, effort, pleasure, pain and cognition are qualities; qualities inhere in substances; none of the eight other substances will do; therefore a ninth.
Manas is an atom
The inner organ is a partless, atomic substance, and this is doing real work: because it is atomic it can be in contact with only one sense at a time, which explains why attention is serial and why you cannot fully attend to two things at once.
An empirical claim, derived from an ontological commitment, and testable.
Release is the end of pain
Since consciousness is adventitious, in apavarga the self loses it along with everything else. The liberated self is a bare substance, without knowledge, without bliss, without experience of any kind.
Opponents found this appalling and said so: “Who would want it?” — the state of a stone, said the Advaitins. Nyāya's reply is that the objection assumes bliss is worth having, when the whole diagnosis of §1 was that experience is the problem.
14Navya-nyāyaनव्यन्यायः
After Gaṅgeśa, Nyāya stops being a philosophy with a technical vocabulary and becomes a technical vocabulary that philosophy is done in. Vedāntins, Mīmāṃsakas and grammarians all adopt it, including those attacking Nyāya's doctrines.
a formal language is the ambition, not the method.
Raghunātha revises the categories
Most Naiyāyikas defend the inherited list. Raghunātha Śiromaṇi's Padārtha-tattva-nirūpaṇa does not: he argues that several of the seven categories are redundant, that time and space should be identified with God, that there is no separate category of number, and that new categories are needed that nobody had proposed. He is writing in the sixteenth century, from inside the most conservative philosophical tradition in India, and he is dismantling its foundations. It is the best evidence that the tradition remained genuinely alive rather than merely repetitive.
15The attacksआक्षेपाः
Nyāya's development is legible as a sequence of responses. Two assaults were serious enough to change the school.
Dignāga and Dharmakīrti — the Buddhist logicians
Your universals are fictions. There is no cowness over and above cows; what the word “cow” picks out is only the exclusion of non-cows — apoha. Nothing positive is shared, and you have built an ontology on a shadow.
Your perception is confused, too. Perception grasps a bare particular and nothing more; the moment you have “this is a cow” you have conceptualisation, which is construction, which is not perception. Your savikalpaka perception is a contradiction in terms.
And two pramāṇas suffice. Perception delivers particulars, inference delivers universals. Your comparison collapses into inference and your testimony into either inference or nothing.
Uddyotakara and Udayana reply
Exclusion presupposes what it excludes. To exclude non-cows you must already be able to identify cows, so apoha either smuggles in the universal or explains nothing.
As to momentariness: Udayana's Ātma-tattva-viveka devotes a whole book to it. If nothing lasts, recognition is impossible — yet I recognise this as the pot I saw yesterday. You must call recognition an error, and then you have made memory, testimony and inference all unreliable, including your own arguments for momentariness.
Śrīharṣa — the Advaitin, twelfth century
I offer no thesis of my own. I claim only this: you cannot define anything. Take your definition of any term. I will show it is over-extended, or under-extended, or circular, or dependent on a term you have not defined. Do it for pramāṇa. Do it for cognition. Do it for difference.
Khaṇḍana-khaṇḍa-khādya — “sweets for the crushing of crushings” — is four hundred pages of exactly this and nothing else.
The Naiyāyika replies — and is changed by it
Two answers were given. The debater's answer: pure refutation without a counter-thesis is vitaṇḍā, and its practitioner is defeated by his own rules, since he must use definitions to attack definitions.
The philosopher's answer was better, and it is Navya-nyāya. If our definitions can be broken, we will build definitions that cannot be — and to do that we need a language precise enough to state exactly what a definition claims. Śrīharṣa did not defeat Nyāya. He provoked its second founding.
16Legacyपरिणामः
The common language
By the fifteenth century a Vedāntin arguing against Nyāya wrote in Navya-nyāya. Madhusūdana Sarasvatī's Advaita-siddhi and Vyāsatīrtha's Nyāyāmṛta — the great Advaita–Dvaita exchange of tier 3 — are conducted entirely in it, by two people who reject Nyāya's metaphysics wholesale.
Into other disciplines
Navya-nyāya's apparatus was adopted by grammarians for stating semantic conditions, by Mīmāṃsakas for stating the scope of injunctions, by Dharmaśāstra authors for stating the conditions of a legal rule, and by poeticians. It became the technical prose of the Sanskrit intellectual world.
The reading it invites now
Modern interest has focused on Navya-nyāya's expressive devices — the limitor, the counterpositive, the nesting of descriptions — as an attempt at an unambiguous technical language independent of formal notation. Comparisons with predicate logic have to be made carefully, and are more interesting when they are.
What was never resolved
The status of universals; whether pervasion can be established at all; whether a definition can be given that survives Śrīharṣa. Nyāya's answers are good and its opponents' answers are good, and the argument stopped because the institutions did, not because anybody conceded.
17Self-checkपरीक्षा
Check your reading
1 · Why does a school of logic state a soteriology in its second sūtra?
Nyāya-sūtra 1.1.2: remove mithyā-jñāna and doṣa, pravṛtti, janma and duḥkha all fall in turn.
2 · What is parāmarśa?
“This hill has smoke, which is pervaded by fire.” Nyāya names the mental event, not just the logical form, because it is describing a causal process in a knower.
3 · An upādhi defeats an inference when it —
Wet fuel accompanies all smoke but not all fire, which is why fire does not prove smoke. Naming what would defeat a pervasion and then failing to find it is Nyāya's actual method.
4 · On the Prābhākara theory of error, what is false in the shell-silver illusion?
Akhyāti is the boldest of the six positions: all cognition whatsoever is true, and error is a failure of discrimination rather than a false content.
5 · What did Śrīharṣa's attack actually produce?
The most productive assault in Indian philosophy. Gaṅgeśa writes about a century and a half after the Khaṇḍana.
Questions worth arguing about
Is nirvikalpaka perception a real stage or a theoretical artefact?
Nobody reports one, and Nyāya concedes this — it argues that the stage is inferred from the structure of determinate perception rather than observed. That is a legitimate move, and it is also exactly what one would say of a posit invented to fill a gap. The Buddhist inversion is instructive: Dignāga agrees that bare non-conceptual awareness exists, and concludes that only it is perception, everything else being construction. Same datum, opposite systems.
Does Nyāya's account of release make the whole system self-undermining?
The objection is sharp. If liberation is the permanent loss of consciousness, and the argument for pursuing it is that experience is unsatisfactory, then a school that spends eighteen centuries refining the pleasures of argument has an odd relationship to its own stated goal. Naiyāyikas answered that clarity about ends is itself the first step. Whether that answers it is worth deciding for yourself.
Is Navya-nyāya usefully compared to symbolic logic?
Only with care. It has no formalism, no derivation rules, no calculus, no notion of validity as a syntactic property. What it does have is a rigorously controlled technical register for stating relational structures of arbitrary depth without ambiguity — closer in spirit to a specification language than to a proof system. The comparison illuminates when it is used to ask what problem each was solving, and misleads when it is used to award marks.
18Glossaryशब्दकोशः
IAST
Devanāgarī
Sense
anumāna
अनुमान
Inference; the second pramāṇa.
anyathā-khyāti
अन्यथाख्याति
Nyāya's theory of error: apprehending one real thing as another.
apavarga
अपवर्ग
Release, as the final cessation of pain.
avacchedaka
अवच्छेदक
Limitor: the property under which something is being considered.
hetu / liṅga
हेतु / लिङ्ग
The reason or mark from which an inference proceeds.
hetvābhāsa
हेत्वाभास
A reason that only appears to be one; a fallacy.
nirvikalpaka / savikalpaka
निर्विकल्पक / सविकल्पक
Indeterminate and determinate perception.
pakṣa · sādhya · hetu
पक्ष · साध्य · हेतु
The three terms of an inference: subject, probandum, reason.
parāmarśa
परामर्श
The consideration that immediately causes an inferential conclusion.
pratiyogin
प्रतियोगिन्
The counterpositive: what an absence is an absence of.
pratyakṣa
प्रत्यक्ष
Perception; the first pramāṇa.
sannikarṣa
सन्निकर्ष
Sense–object contact; of six kinds.
śabda
शब्द
Reliable testimony; the fourth pramāṇa.
tarka
तर्क
Hypothetical reasoning that eliminates an alternative without establishing anything.
upādhi
उपाधि
A defeating condition: pervades the probandum but not the reason.
upamāna
उपमान
Comparison; the third pramāṇa.
vyāpti
व्याप्ति
Invariable pervasion; the relation on which every inference depends.