IKS/Part 2·Knowledge: Framework and ClassificationChapter 7 of 14
Part 2 · Chapter 7
Knowledge
प्रमाणम् · पदार्थः · तर्कः
Before you can add to what is known, you need an account of what knowing is, what kinds of things there are to know, which routes to knowledge are legitimate, and how a disagreement gets settled. Two schools — Nyāya and Vaiśeṣika — spent two millennia working on exactly those four questions, and produced the apparatus every other Indian discipline argues within.
This is the methodological chapter. Everything in Part 3 was produced by people trained in these rules.
1A misconception, firstभ्रान्तिनिरासः
The standing assumption about Indian intellectual life is that it proceeds by authority — that assertions are settled by scripture rather than by argument. It is worth disposing of this before going further, because it is close to the reverse of the truth.
What the record shows is an intensely adversarial culture of debate, with formal rules, recognised fallacies, appointed moderators, and conventions for conceding. Schools that agreed on almost nothing shared a technical vocabulary precise enough to argue in. A Buddhist logician and a Naiyāyika could not agree on whether there is a self, but they agreed exactly on what would count as a valid inference — and spent centuries refining that agreement by attacking each other's use of it.
The characteristic form of an Indian philosophical text makes this visible. A treatise typically states the opponent's position first, in its strongest form (the पूर्वपक्ष, pūrvapakṣa), often more persuasively than the opponent had managed; then refutes it; then states its own conclusion (सिद्धान्त, siddhānta). A position that cannot survive its own author's best attempt to defeat it does not get published.
Nyāya's specific brief
Among the six Vedic darśanas (chapter 3), Nyāya is the one whose subject is method: how to enquire into a problem systematically and establish something new. Vaiśeṣika, its sister school, supplies the inventory of what there is to enquire about. They were treated as a pair for most of their history, and are usually taught together.
2The four terms of knowingचतुष्टयम्
Any episode of knowledge involves four things, and the tradition names all four. Missing one is how enquiry goes wrong.
New knowledge, on this account, always begins from something already known and moves outward. The interesting question is therefore never “how do we know anything?” but “which moves from the known to the unknown are permitted?”
3Vaiśeṣika's projectवैशेषिकम्
Take the second term, prameya, seriously and you get a question: what kinds of things are there? Vaiśeṣika's answer is the most sustained attempt at an ontology in Indian thought.
The name comes from विशेष (viśeṣa), individuality or particularity — the school is named for its insistence that what ultimately exists is particular. It is attributed to the sage Kaṇāda, placed somewhere between the sixth and second centuries BCE, and its aim is stated as describing everything “nameable and knowable”.
4The six categoriesषट् पदार्थाः
The whole system is announced in the fourth sūtra of the text:
Vaiśeṣika-sūtra 1.1.4 — “The highest good comes from knowing the truth of the categories — substance, attribute, action, genus, particularity and inherence — through their similarities and differences.”
The word पदार्थ (padārtha) is worth pausing on: literally “the meaning of a word”, and used to mean a category of the real. The premise is that the ways language carves things up correspond to the ways reality is jointed — and that “through their similarities and differences” names the method: you know what a thing is by what it resembles and what it does not.
The sixth is the subtle one and the most attacked. Samavāya is what licenses the intuition “this is in that” — the redness is in the pot, the pot's parts are in the pot. Two things merely placed together can be separated; things joined by inherence cannot, because separating them destroys one of them.
5The nine substancesनव द्रव्याणि
The list is broader than “substance” suggests in English. Time, space, self and mind are all on it — anything that can bear attributes and actions qualifies, whether or not you can weigh it.
6The elemental staircaseगुणविभागः
The most elegant piece of the system. Each element is distinguished from the others by which sense-qualities it carries — and the qualities drop away one at a time as you ascend from earth to ether.
The reasoning is inferential and worth following. Sound propagates; propagation requires a medium; the medium is not any of the four elements, since sound is heard where none of them is manifest. Therefore a fifth substance exists whose sole distinguishing quality is sound. Whether or not one accepts the conclusion, the argument has a clean shape.
7Attributesगुणाः
Twenty-four attributes are enumerated, and they sort neatly by what kind of substance can carry them.
8Actionकर्म
The third category is motion, treated as a distinct kind of thing rather than as a change in attributes — because motion, unlike colour, produces effects in places other than where it is.
Five kinds
upward movement — utkṣepaṇa
downward movement — avakṣepaṇa
contraction — ākuñcana
expansion — prasāraṇa
general motion — gamana: rotation, flow, percolation and the rest
Three properties
Occurs only in finite substances — the infinite ones do not move.
Produces an effect in its own substratum and elsewhere.
Cannot itself originate another action or a substance; and is destroyed by the conjunction it brings about.
Seven causes
gravity — gurutva
volition — prayatna
conjunction — saṃyoga
fluidity — dravatva
impact — abhighāta
impulse — nodana
the unseen — adṛṣṭa
The last cause on that list is worth marking honestly. Adṛṣṭa — “the unperceived” — is the school's placeholder for motions whose cause is not evident: the rising of sap, the circulation of the body, the first motion of the atoms. It functions as an admission of ignorance with a name attached, which is both intellectually honest and a place where enquiry tends to stop.
9Genus, species, inherenceसामान्यविशेषसमवायाः
The last three categories are what turn an inventory into a structure.
Substance, attribute and action all admit of this treatment, so the same apparatus generates a class hierarchy over everything in the system. What holds each level to the one above it is inherence — a relation between things “inseparably associated by nature”, as distinct from mere conjunction, which is contingent and reversible.
The multiplicity thesis
Four sūtras state how the many arise from the few, and they are worth having in order:
Sūtra
Text
What it asserts
1.1.6
द्रव्यगुणयोः सजातीयारम्भकत्वं साधर्म्यम्
Substances and attributes alike originate things of their own kind. Like begets like.
But they also originate things of other kinds. Novelty is possible.
4.1.6
महत्यनेकद्रव्यवत्त्वात् रूपाच्चोपलब्धिः
A thing is perceptible only when it is large, composed of many substances, and has colour. Perceptibility is a consequence of composition.
4.2.4
अणुसंयोगः तु अप्रतिषिद्धः
The conjunction of atoms is not thereby ruled out — we cannot see it, and that is not an argument against it.
The last is the significant one methodologically. Sūtra 4.2.2 has just conceded that the joining of the elemental atoms is not perceptible; 4.2.4 refuses to let that concession become a refutation. Unobservability is not disproof — an explicit statement, in a text of the last centuries BCE, of a principle that atomic theory needed again two thousand years later.
10Falling and flyingगुरुत्वम्
Several Vaiśeṣika sūtras concern weight and motion, and they are among the most over-claimed passages in the whole corpus. Here they are, with a careful statement of what they do and do not amount to.
What these are, and what they are not
They are not Newton's laws. There is no quantitative statement, no proportionality between force and acceleration, no mass distinguished from weight, no universal gravitation, and no mathematics anywhere near them. Presenting them as a pre-emption of the Principia is exactly the inflation warned against in chapter 1, and a physicist will dismantle the claim in one question.
They are, nonetheless, genuinely notable. Four things are being asserted that are correct and not obvious: that falling has a cause, named and treated as a property of the body; that a body left unsupported falls because of that property; that motion in any other direction requires an impulse; and — the sharpest of the four — that a projectile's continued motion is produced by a disposition imparted by the previous motion, iterated. That last is a theory of impetus, and it is the same move Philoponus makes in the sixth century CE and Buridan in the fourteenth, on the road that eventually leads to inertia.
Stated at that level, the claim is defensible, interesting, and needs no exaggeration.
11The four pramāṇasचत्वारि प्रमाणानि
Which routes to knowledge are legitimate? Nyāya accepts four. Every school in India can be located by which of these it accepts and which it rejects — it is the single most useful diagnostic in Indian philosophy.
Testimony is the one modern readers resist, and the resistance does not survive scrutiny. You have not personally verified the speed of light, the existence of Antarctica or your own date of birth. Nyāya's position is that testimony is a genuine source of knowledge with genuine conditions on it — the speaker must be competent, honest and informed — and that pretending otherwise is not scepticism but self-deception.
Where the schools divide
Cārvāka accepts perception only, and is therefore committed to denying that inference yields knowledge — a hard position, which it defended vigorously. Vaiśeṣika and Buddhist logicians accept two: perception and inference. Sāṅkhya and Yoga accept three, adding testimony. Nyāya accepts these four. Mīmāṃsā adds two more — presumption (arthāpatti) and non-apprehension (anupalabdhi) — reaching six. The list is not decoration: how many pramāṇas a school accepts determines what it can argue for.
12The five-step inferenceपञ्चावयवः
Nyāya's most famous contribution: the canonical form in which an inference must be presented if it is to be assessed by an opponent.
To a reader trained on the Aristotelian syllogism this looks redundant — steps 1 and 5 say the same thing, and steps 3 and 4 look like one move split in two. The redundancy is deliberate and is about debate rather than about logic: step 1 puts the claim on record before evidence is offered, step 5 closes it, and separating the rule from its application forces the arguer to defend the universal claim and the particular claim on separate occasions. It is a protocol for adversarial use, not a minimal proof form.
The load-bearing concept: vyāpti
Step 3 asserts an invariable concomitance — व्याप्ति, vyāpti — between smoke and fire. Every inference stands or falls on whether such a relation genuinely holds, and the whole later literature of Indian logic is largely about how a vyāpti can be established, since no finite number of observed cases entails it. This is the problem of induction, recognised early and worried at for a thousand years. Cārvāka's rejection of inference as a pramāṇa rests precisely here: they argue that vyāpti can never be established, and therefore that inference is never certain.
13The Nyāya frameworkन्यायपद्धतिः
Put the pieces together and what emerges is a complete procedure for enquiry — recognisably a research method, with a place for objectives, for the problem statement, for evidence, for adversarial testing and for publication.
Map this onto a modern research paper and the correspondence is close: motivation, gap in the literature, method, results, discussion under peer criticism, conclusion. What Nyāya adds — and what the modern version handles informally — is an explicit taxonomy of dishonest argument, so that a debate can be lost on procedure.
14Where doubt comes fromसंशयकारणानि
Enquiry begins with saṃśaya, and Nyāya classifies its sources. A researcher who can say which kind of doubt they are in knows what would resolve it.
Each source suggests a different remedy: find a discriminating observation; extend the theory; resolve the conflict by locating the error; acquire the missing basics; improve the measurement. The classification is diagnostic, not decorative.
15Six ways to argueकथाभेदाः
Nyāya distinguishes six modes of debate, and ranks them frankly. Only the first is an honest search for truth; the tradition names the other five so that they can be recognised and called out.
The realism here is bracing. A tradition that only described honest debate would be useless; the point of naming chala and jāti is that they happen, and that a debater who can name what has just been done to them can say so out loud and be believed.
16Fallaciesहेत्वाभासाः
A हेत्वाभास (hetvābhāsa) is “a reason that only appears to be one” — a step-2 that fails. Five are classically listed.
Savyabhicāra — erratic सव्यभिचार
The reason occurs in cases where the conclusion does not hold. “The hill has fire, because it is knowable” — everything is knowable, including hills without fire. The mark does not discriminate.
Viruddha — contradictory विरुद्ध
The reason actually establishes the opposite of what is claimed. Offered as support, it is evidence against.
Prakaraṇa-sama — counterbalanced प्रकरणसम
An equally good reason exists for the contrary conclusion, so the argument merely restates the problem rather than settling it.
Sādhya-sama — unproved साध्यसम
The reason is itself as much in need of proof as the conclusion. Circularity, named.
Kālātīta — mistimed कालातीत
The reason is offered at a point where it no longer bears — the relation it depends on does not hold at the relevant time. Contextual irrelevance.
Why five and not more
Each corresponds to a way the concomitance in step 3 can fail: it may not be universal, may run the other way, may be matched by a rival, may be unestablished, or may not obtain now. The list is derived from the inference form, not collected from experience.
17Four kinds of conclusionचतुर्विधसिद्धान्तः
Nyāya also classifies what a debate can conclude with — because “everyone agrees” and “our school holds” are different claims and should not be confused.
Type
Devanāgarī
What it is
Example
Sarvatantra-siddhānta
सर्वतन्त्रसिद्धान्त
Accepted by every school; common ground that can be appealed to without argument.
All physical substances are subject to weight.
Pratitantra-siddhānta
प्रतितन्त्रसिद्धान्त
Held by one school and denied by another; must be argued for, not assumed.
The Āyurvedic account of health, as against a rival medical system's.
Adhikaraṇa-siddhānta
अधिकरणसिद्धान्त
A conclusion whose acceptance carries other conclusions with it — implication chains.
Accepting that there is water on Mars commits you to a set of further claims about what water permits.
Abhyupagama-siddhānta
अभ्युपगमसिद्धान्त
Not asserted directly, but following from the details examined; a working assumption made explicit.
Analysing eclipses and day-length presupposes relative motion of sun, moon and earth — even where nobody says so.
The fourth is the useful one
Abhyupagama-siddhānta names the assumptions a practice carries without stating them. Every technical tradition has these, and they are the hardest things to see from inside. Naming the category is what makes it possible to ask, of any body of work: what must be true for this to make sense, that nobody here has ever written down?
18Self-checkपरीक्षा
Check your reading
1 · What does samavāya name, and how does it differ from conjunction?
Threads and cloth, not sand and sand. Separate the threads and there is no cloth; separate two grains of sand and you still have two grains.
2 · Why does Vaiśeṣika posit ether (ākāśa) as a distinct substance?
The reasoning is inferential: propagation needs a medium, the medium is not any of the four, therefore a fifth exists whose sole quality is sound.
3 · Which claim about the Vaiśeṣika motion sūtras is defensible?
Neither dismissal nor inflation. The impetus reading is textually supported by 5.1.17 and is genuinely notable; the mathematics that would make it Newtonian is entirely absent.
4 · Steps 1 and 5 of the Nyāya inference restate the same proposition. Why?
It is not a minimal proof form and was never meant to be. The redundancy does procedural work in an adversarial setting.
5 · A Cārvāka rejects inference as a means of knowledge. On what grounds?
This is the problem of induction, and Cārvāka pressed it hard enough that the rest of Indian philosophy had to answer. Note that Cārvāka accepts perception only, so testimony is not available to them either.
Questions worth arguing about
Is treating universals as real entities a strength or an embarrassment?
The classic objection is a regress: if cowness is a real thing over and above cows, what relates cowness to the cows — and is that relation a further thing? Vaiśeṣika answers that inherence is self-linking and needs no further relation, which opponents call special pleading. The Buddhist alternative — that universals are constructions, and that what a word picks out is only the exclusion of what it is not — is one of the great rival positions in Indian philosophy, and the debate ran for a thousand years without either side collapsing.
Does naming the dishonest modes of debate legitimise them?
Nyāya's own answer is that jalpa and vitaṇḍā are permissible defensively — against an opponent already using them, or to protect a true position from a skilled bad-faith attacker. That answer has always been controversial within the tradition, and later Naiyāyikas restricted it. It is the same argument that surrounds adversarial legal systems today.
What would it take to add a sixth category to Vaiśeṣika?
Later Vaiśeṣika did exactly this, adding abhāva — absence — as a seventh padārtha, on the grounds that “there is no pot here” is a genuine cognition of something. Whether an absence can be a category of the real is a live question, and the debate over it is a good place to watch the school reasoning about its own foundations rather than about the world.
19Glossaryशब्दकोशः
IAST
Devanāgarī
Sense
adṛṣṭa
अदृष्ट
“The unseen”: the named placeholder for a cause not evident to observation.
anumāna
अनुमान
Inference; the second pramāṇa.
avayava
अवयव
A member of the five-step inference.
dravya
द्रव्य
Substance; the first category, of nine kinds.
guṇa
गुण
Attribute; the second category, twenty-four in number.
hetu
हेतु
The reason offered in an inference — step 2.
hetvābhāsa
हेत्वाभास
A reason that only appears to be one; a fallacy.
karma
कर्म
Action or motion; the third category. (In ethics the word means something else.)
nirṇaya
निर्णय
Settled conclusion; the end of an enquiry.
padārtha
पदार्थ
“The meaning of a word”; a category of the real.
pramā
प्रमा
Valid knowledge.
pramāṇa
प्रमाण
A legitimate means of knowing.
pramātā / prameya
प्रमाता / प्रमेय
The knower; the knowable.
pratyakṣa
प्रत्यक्ष
Perception; the first pramāṇa.
pūrvapakṣa
पूर्वपक्ष
The opponent's position, stated first and at its strongest.
śabda
शब्द
Reliable testimony; the fourth pramāṇa. Also “sound”, the quality of ether.
samavāya
समवाय
Inherence; the inseparable relation.
sāmānya / viśeṣa
सामान्य / विशेष
Genus and particularity; assimilation and differentiation.
saṃśaya
संशय
Doubt; the recognised ambiguity that begins an enquiry.
siddhānta
सिद्धान्त
An established tenet; of four kinds.
tarka
तर्क
Reasoned argumentation under ambiguity.
upamāna
उपमान
Comparison; the third pramāṇa.
vyāpti
व्याप्ति
Invariable concomitance; the relation that makes an inference work.