How We Know
The Pramāṇas: perception, inference, testimony and more. Before any school could tell you what is real, it had to say how anything is known at all.
How do you know there is a phone in your pocket? You might feel it — perception. You might hear it buzz and conclude it must be there — inference. Your friend might tell you she just saw you put it there — testimony. Every day we rely on several different ways of knowing without thinking about it. Classical Indian philosophers thought about it very hard. Before a school could argue about the Self, God or the world, it had to declare its pramāṇas: the means by which true knowledge is gained.
The question behind every school
The word pramāṇa comes from pra- + mā, “to measure”: it is the measuring rod of knowledge. Philosophers analysed every act of knowing into four parts:
- Pramātā प्रमाताThe knower — the self who comes to know.
- Pramāṇa प्रमाणThe means — the instrument by which knowledge arises, such as the eye or a valid argument.
- Prameya प्रमेयThe object known — the pot, the fire on the hill, the meaning of a Vedic command.
- Pramā प्रमाThe result — a true cognition, which presents its object as it really is.
The stakes were high. If testimony is a genuine pramāṇa, the Veda can teach us truths no eye can see. If only perception counts, as the Cārvākas held, then heaven, karma and rebirth are beyond knowledge — and probably fictions. So the list of pramāṇas a school accepted shaped everything it could go on to say.
Six ways of knowing
The fullest list has six means. Each comes with a classic example that every student learned:
👁️PratyakṣaPerception
Pratyakṣa, “before the eye”: knowledge from the contact of a sense with its object. You see the pot on the table. Schools debated whether perception includes a first, wordless stage (nirvikalpaka) before we recognise “this is a pot” (savikalpaka) — and whether yogis can perceive what ordinary senses cannot.
🔥AnumānaInference
Anumāna, “measuring after”: knowing something unseen through a sign that is invariably connected with it. You see smoke rising from a distant hill and know there is fire there, because wherever there is smoke there is fire. That invariable connection is called vyāpti, “pervasion.”
🐂UpamānaComparison
Upamāna: learning what a word refers to through similarity. A town-dweller who has never seen a gavaya (a wild ox) is told by a forester, “a gavaya is like a cow.” Later, in the forest, he sees a cow-like animal and knows: “this is a gavaya.”
शब्दŚabdaTestimony, the word
Śabda: knowledge from the words of a reliable speaker (āpta) — or, for Mīmāṃsā and Vedānta, from the authorless Veda itself. Most of what you know about history, distant countries or your own birth, you know this way.
🍽️ArthāpattiPresumption
Arthāpatti: assuming the only fact that resolves an apparent conflict. “Devadatta is fat, but he does not eat by day.” Both are known to be true; the only way they can both be true is that he eats at night. Or: Devadatta is alive and not at home, so he must be somewhere else.
∅AnupalabdhiNon-apprehension
Anupalabdhi: knowing an absence by not finding something that would be found if it were there. You look at the floor in good light and know: “there is no jar here.” Bhāṭṭa Mīmāṃsakas and Advaitins argued that absence cannot be perceived by the senses, so this needs its own pramāṇa.
Nyāya’s own founding text lists four:
Who accepts what
The schools agreed on perception and disagreed on almost everything else. Tap a number to see who stopped there — and why they reduced the others:
One — the Cārvākas: perception only. Inference, they argued, depends on a universal rule (“all smoke goes with fire”) that no one has observed in every case, so it can never be certain. Later Cārvākas such as Purandara seem to have allowed inference about things that can in principle be seen.
Two — the Buddhist logicians (Dignāga, Dharmakīrti) and Vaiśeṣika: perception and inference. Testimony is only trustworthy because we infer that the speaker is reliable, so it is a kind of inference; comparison and presumption are likewise reduced.
Sāṃkhya, Yoga and the Jains, along with Ramanuja and Madhva among the Vedāntins, accept perception, inference and reliable testimony. Patañjali puts it in one line: pratyakṣānumānāgamāḥ pramāṇāni, “perception, inference and scripture are the means of knowledge” (Yoga Sūtra 1.7).
Jain thinkers usually frame things differently, dividing knowledge into direct (pratyakṣa) and indirect (parokṣa), with memory, recognition, reasoning, inference and testimony under the second; the count of “three” is a simplification.
Four — Nyāya: perception, inference, comparison and testimony. Presumption is a form of inference, and absence is perceived along with the bare floor.
Five — Prābhākara Mīmāṃsā: adds presumption, but rejects non-apprehension: for Prabhākara, “absence” is simply the bare place perceived on its own.
Six — Bhāṭṭa Mīmāṃsā (after Kumārila) and Advaita Vedānta: all six. Advaita’s classic manual of method, Dharmarāja’s Vedānta-paribhāṣā (17th century), is organised chapter by chapter on them. Some Paurāṇika writers added even more, such as “inclusion” (sambhava) and tradition (aitihya), which Nyāya folded into inference and testimony.
Sāṃkhya explains why three are enough:
त्रिविधं प्रमाणमिष्टं प्रमेयसिद्धिः प्रमाणाद्धि ॥ dṛṣṭam anumānam āptavacanaṃ ca sarva-pramāṇa-siddhatvāt |
trividhaṃ pramāṇam iṣṭaṃ prameya-siddhiḥ pramāṇād dhi || Perception, inference and the word of a reliable person — since every other means is included in these, the means of knowledge are held to be threefold. For it is through a means of knowledge that what is to be known is established. Īśvarakṛṣṇa, Sāṃkhyakārikā 4
Inference in five steps
Nyāya’s great contribution was to set out inference in a fixed public form, the “inference for others” (parārthānumāna) used in debate. It has five members (avayavas). Step through the most famous example in Indian logic:
Western readers often compare this with Aristotle’s syllogism, but the purpose is different. The Nyāya inference is a tool of persuasion between people, which is why it repeats itself: a listener in a debate must be led step by step. And the example matters: an Indian inference must always be grounded in something observed, never in a purely formal rule. Buddhist logicians, led by Dignāga, argued that the rule and the reason are enough, and cut the form down to two or three members.
When you reason silently for yourself (svārthānumāna), Nyāya says, you go through the same moves in your own mind: seeing smoke, remembering the rule, and concluding.
When reasons go wrong
A reason that looks like a reason but is not is a hetvābhāsa, a “pseudo-reason.” The standard later list, taught in Annambhaṭṭa’s student primer Tarkasaṅgraha (17th century), has five. Tap each:
The Nyāya Sūtra itself (1.2.4) gives an older list of five under slightly different names. Alongside fallacies it discusses tarka, hypothetical reasoning (“if there were no fire, there would be no smoke”), which does not prove anything by itself but clears the ground for a proof.
Does knowledge prove itself?
One debate cut deeper than the rest. When I have a cognition — “that is water” — how do I know it is true? This is the question of prāmāṇya, validity.
Mīmāṃsā view Intrinsic validity (svataḥ prāmāṇya). Every cognition is valid by its own nature, from the moment it arises. We do not need to check it; we act on it. It is overturned only if a later cognition defeats it (“that was a mirage”) or if we find a fault in its source (bad light, jaundiced eyes). Otherwise, the Mīmāṃsakas warned, we would need a second cognition to validate the first, a third for the second, and so on forever.
This was the foundation for the Veda’s authority: an authorless text has no author whose faults could make it false, so its teaching stands. Advaita Vedānta adopted the same view.
Nyāya view Extrinsic validity (parataḥ prāmāṇya). A cognition is made valid by the soundness of its causes, and we come to know it is valid by something outside it — above all, by successful action. You see water, walk over and drink: the quenched thirst confirms the cognition. No infinite regress follows, because most of the time we act first and only check when doubt arises.
For Nyāya, the Veda is trustworthy because its author, Īśvara, is perfectly reliable — a view the Mīmāṃsakas rejected (see The Four Vedas).
Later doxographers mapped four positions: Sāṃkhya making both truth and falsity intrinsic, Nyāya both extrinsic, Mīmāṃsā truth intrinsic and falsity extrinsic, and the Buddhists the reverse. The scheme is tidy; the actual arguments were subtler.
- Cārvāka: since pervasion can never be checked in every case, inference rests on a leap. The sceptic Jayarāśi Bhaṭṭa (c. 8th century) went further and attacked every pramāṇa in his Tattvopaplavasiṃha, “The Lion that Overthrows All Principles.”
- Nāgārjuna (Madhyamaka Buddhist) asked: what establishes the pramāṇas themselves? If another pramāṇa, there is a regress; if nothing, they are unproven — so the whole system has no final foundation. Nyāya replied that a lamp lights itself as well as the room.
- Advaita: all six pramāṇas are fully valid in everyday experience (vyavahāra), but only the Upanishadic word can reveal Brahman, which is the knower and can never be an object.
- Shaiva Siddhānta, Ramanuja and Madhva accept three — perception, inference and scripture — with the Āgamas or the Pāñcarātra counted as scripture alongside the Veda. Gauḍīya writers discuss more but make scripture supreme.
Navya-Nyāya: logic sharpened
In the 14th century, Gaṅgeśa Upādhyāya of Mithilā wrote the Tattvacintāmaṇi, “The Wish-Jewel of Truth,” in four books — one for each Nyāya pramāṇa. It began Navya-Nyāya, “New Nyāya,” with a technical language for describing exactly what qualifies what in a cognition. Its precision was such that later writers in grammar, law, Mīmāṃsā and Vedānta adopted it. See Nyāya & Vaiśeṣika.
Check your knowledge:
“Fat Devadatta does not eat by day, so he must eat at night” illustrates…
Presumption posits the one fact that reconciles two known facts.
How many pramāṇas does the Nyāya Sūtra accept?
Perception, inference, comparison and testimony (Nyāya Sūtra 1.1.3).
In the five-membered inference, “wherever there is smoke there is fire, as in a kitchen” is the…
The example states the pervasion (vyāpti) and anchors it in an accepted case.
Which school held that every cognition is valid by its own nature?
Svataḥ prāmāṇya — valid until defeated. Advaita followed it.
Questions people often ask
Is “testimony” just believing what you are told?
No. Every school that accepts śabda sets conditions: the speaker must know the matter, be truthful and intend to communicate it, and the sentence must be coherent. Testimony from an unreliable source is not a pramāṇa at all. The real debate was whether scripture is reliable, and why.
Did Tamil thinkers discuss pramāṇas?
Yes. The Tamil Buddhist epic Maṇimēkalai has its heroine hear exponents of rival schools set out their views, and its later cantos summarise Buddhist logic. Shaiva Siddhanta treatises in Tamil, such as the Civañāṉacittiyār, open with the means of knowledge before turning to God, soul and bondage. See Shaiva Siddhanta.
Does this matter for a practising Hindu?
It explains why teachers so often say that the Self cannot be found by argument alone, why scripture and the guru are given such weight, and why a claim based on experience still has to be tested. The pramāṇas are the grammar of every philosophical conversation in the tradition.
॥ Pause and reflect
- Think of three things you are sure of. By which pramāṇa did you come to know each?
- How much of what you believe rests on testimony? Whose word do you trust, and why?
- When did a “that is water” of your own turn out to be a mirage?
Your notes are saved privately in this browser.
Key terms on this page
Sources & further reading
- D. M. Datta, The Six Ways of Knowing: A Critical Study of the Vedānta Theory of Knowledge (George Allen & Unwin, 1932).
- Bimal Krishna Matilal, Perception: An Essay on Classical Indian Theories of Knowledge (Clarendon Press, 1986).
- Bimal Krishna Matilal, The Character of Logic in India (SUNY Press, 1998).
- Stephen H. Phillips, Epistemology in Classical India: The Knowledge Sources of the Nyāya School (Routledge, 2012).
- Jonardon Ganeri, Philosophy in Classical India: The Proper Work of Reason (Routledge, 2001).
- Daniel Arnold, Buddhists, Brahmins, and Belief: Epistemology in South Asian Philosophy of Religion (Columbia University Press, 2005).
- Ganganatha Jha (trans.), The Nyāya-Sūtras of Gautama, with the Bhāṣya of Vātsyāyana and the Vārttika of Uddyotakara (4 vols, Indian Thought, 1912–19).
- J. N. Mohanty, Classical Indian Philosophy (Rowman & Littlefield, 2000).