Indian epistemology is unusual in being conducted, from the beginning, as a debate. Every school has a view about what counts as a valid means of knowledge—a pramāṇa (प्रमाण)—and every other school is in the room arguing.
The classical list, accepted in some form by most schools, contains six items. Schools differ in which they accept and how they define them. The disagreements are productive. Here is the reader's map.
1. Pratyakṣa (प्रत्यक्ष)—perception. Direct sensory awareness. Accepted by everyone, defined differently by each. The Nyāya school requires that the contact between sense and object be specific and present; the Buddhist Dignāga argues perception is "mere awareness, free of conceptual construction".
2. Anumāna (अनुमान)—inference. Reasoning from what is known to what follows. The classical example: parvato vahnimān dhūmāt—the mountain has fire because of smoke. The structure of valid inference is worked out with enormous care in Nyāya, especially the five-membered syllogism (pañcāvayava-vākya).
3. Upamāna (उपमान)—analogy. Recognizing the unknown by similarity to the known. You have never seen a gavaya (a wild ox), but someone tells you it is "like a cow". When you encounter one, the analogy gives you knowledge of what you are seeing. Accepted by Nyāya and Mīmāṃsā, rejected by Vaiśeṣika as reducible to inference.
4. Śabda (शब्द)—testimony. Knowledge from a reliable verbal source. This is the most philosophically interesting of the six because it is the only one not based on direct experience. All the Vedānta schools accept Śruti (revealed scripture) as a valid śabda. Modern empiricism mostly does not accept śabda as a basic pramāṇa, treating testimony as reducible to perception plus inference. The classical argument against this reduction is sharp: most of what any of us knows—history, geography, science we have not personally verified—is in fact testimony.
5. Arthāpatti (अर्थापत्ति)—postulation. Inferring an unstated cause from an observed effect. The classic example: Devadatta is fat but does not eat by day. We postulate that he eats by night. Accepted by Mīmāṃsā, treated by Nyāya as a form of inference. The category survives in modern thought as "inference to the best explanation" or abductive reasoning.
6. Anupalabdhi (अनुपलब्धि)—non-apprehension. Knowledge by absence. You enter the room and know the elephant is not there. The way you know is not by seeing nothing; it is by the specific absence of what you would have seen if it were there. Accepted by Bhaṭṭa Mīmāṃsā and Advaita, contested elsewhere.
The six together constitute a remarkably comprehensive map. They cover direct experience, reasoning, analogy, testimony, abduction, and absence—most of how human knowledge actually works.
The schools' disagreements about which are basic and which are reducible are the engine of Indian philosophical progress. Buddhists collapse the list to two (perception and inference) by treating the rest as derived. Vedāntins resist this because śabda must remain irreducible for revelation to do its work. Mīmāṃsakas accept all six because they need the maximum machinery to handle Vedic exegesis.
What the modern reader takes from all this is not a doctrinal commitment to one school's list. It is a vocabulary. The questions "how do you know?" and "what counts as evidence here?" become more precise when one has the classical map.