§2.1 Three instruments
Before asking what a passage means, ask a prior question: by what instrument do you propose to find out? Indian philosophy makes this question explicit and technical. A pramāṇa is a means of valid knowing. The Vedāntic schools accept three.
| Instrument | Range | Failure mode |
|---|---|---|
| Pratyakṣa direct perception | Objects present to a functioning sense organ | Silent outside its range. A telescope does not report on justice; it reports nothing, which is not the same as reporting absence. |
| Anumāna inference | Whatever is reliably linked to something perceived | Inherits every defect of its premises, and its reach is bounded by the perceived instance that grounds the invariable concomitance. |
| Śabda testimony | Whatever a reliable speaker reports | Only as reliable as the speaker. The entire weight of the tradition's epistemology therefore lands on the question of which speaker. |
Madhva organises these under a distinction worth carrying: kevala-pramāṇa, valid knowledge itself, as against anu-pramāṇa, the means that produce it. The three above are means. Their product is a cognition that corresponds to its object — and correspondence, not coherence or utility, is Madhva's criterion of truth. He is, in modern terms, a thoroughgoing realist, and this matters: his hermeneutics is downstream of his epistemology, not the reverse.
§2.2 The four defects
The tradition's argument for the primacy of śabda in a specific domain runs through a diagnosis of the conditioned knower. Four defects (doṣa) are enumerated:
- Bhrama — illusion. The tendency to mistake one thing for another. Not occasional but structural; the standard illustration is the rope taken for a snake.
- Pramāda — inattention. Error by oversight. Not a moral failing; a bandwidth limit.
- Vipralipsā — the disposition to deceive. Including, and especially, self-deception. The interpreter who finds in a text exactly the conclusion he brought to it is exhibiting this defect, usually without noticing.
- Karaṇāpāṭava — defective instruments. The senses have a bandwidth. Every extension we have built — telescope, interferometer, detector — extends the band without changing its nature, because each still terminates in the same narrow-band observer.
The argument is not that these defects make empirical knowledge impossible. That would be self-refuting and the tradition does not say it. The argument is that they make a specific class of object — atīndriya, beyond the reach of the senses — permanently inaccessible to the first two instruments. For that class, and only for that class, śabda is not a supplement but the sole available channel.
What counts as atīndriya — and what does not
Within the class: the ontological status of the self; the existence and nature of Bhagavān; the mechanism of transmigration; the structure and inhabitants of realms not reachable by any instrument; the ultimate purpose, if any, of the cosmos.
Outside the class: the boiling point of water; the age of a mineral deposit; the aerodynamic feasibility of a described craft; the manuscript history of a text; the phonetics of Vedic accent. For all of these, pratyakṣa and anumāna are the appropriate instruments and śruti is not a competing authority — a point developed at length in §15, where it does the most work.
The line between the classes is not always obvious, and drawing it wrongly in either direction is the source of most bad argument at the science–scripture boundary. Where we are unsure which side a question falls on, we say so.
§2.3 Jīva Gosvāmī's formulation
In the Tattva-sandarbha, the first of the six Sandarbhas, Jīva Gosvāmī sets the epistemology out with unusual economy. He surveys the candidate pramāṇas, argues that for the subject at hand only śabda survives, identifies śruti as the śabda in question, notes the practical difficulty of the Vedic corpus for a fallen age — its vastness, its dispersal, the loss of śākhās — and concludes that the Purāṇic literature, and specifically the Bhāgavata, functions as the accessible and self-evident form of that same authority.
Whatever one thinks of the conclusion, the structure is worth noting, because it is the structure this document imitates: state the instrument, state its range, state why the alternatives fail for this object, then and only then start reading texts. Beginning with the texts and reverse-engineering an epistemology to justify the desired reading is the standard failure, and it is detectable.
"Whether it enjoins Me, names Me, or is argued and refuted concerning Me — this much is the entire meaning of all the Vedas: the sound itself is otherwise composed of Me."