Twenty-Two Ways to Lose an Argument

The Nyāya-sūtra contains a numbered list of 22 nigraha-sthānas — literally "grounds of defeat". Not 22 fallacies. Twenty-two named conditions under which a participant in a formal debate has lost, and the audience is entitled to say so out loud.

Some are what you would expect: contradicting yourself, giving a reason that proves the opposite of your thesis, arguing in a circle. Others are procedural in a way that has almost no Greek counterpart — shifting your thesis mid-argument, repeating yourself instead of answering, failing to reply within the allotted turns, saying something the audience cannot understand, or conceding your opponent's point while continuing to argue as if you hadn't. Losing was a *defined event*, and the definition was written down.

That tells you what Indian logic actually is. It is not primarily a theory of which sentence forms preserve truth. It is a theory of what happens when two trained people disagree in public and something has to be settled. Run that test on your own meetings: your team almost certainly has an unwritten sense of who "won" a discussion, and almost certainly no shared definition of it. Nyāya wrote the definition down two thousand years ago.

This article is part of the Roots of Reasoning series on the world's argumentation traditions.

Logic as a Branch of Epistemology

Start where the Indian authors started. The organizing question is not "does this conclusion follow from these premises?" but "by what process can a cognition count as knowledge?" Those processes are the pramāṇas. Nyāya recognizes four:

• Pratyakṣa (perception): direct sensory contact with an object
• Anumāna (inference): moving from something known to something not yet known
• Upamāna (comparison): knowledge acquired through similarity
• Śabda (testimony): knowledge from a competent, reliable speaker

Other schools adjust the list. Some add presumption (arthāpatti) or non-perception (anupalabdhi); Buddhist logicians reduce it to perception and inference; materialists accept perception alone. The lists differ, but the framing does not: logic lives inside epistemology. An inference is not an arrangement of sentences that happens to be truth-preserving. It is a process by which a particular person comes to know something they did not know before.

That single decision explains most of the differences that follow. Aristotle can call a syllogism valid while its premises are false, because validity is a property of form. For Nyāya that would be a strange thing to be interested in. The interesting question is whether the reason you gave actually delivers the knowledge you claim it delivers.

The Five-Member Argument — and Why the Extra Members Are There

The canonical presentation, with the standard fire-on-the-hill example:

• Pratijñā (thesis): "There is fire on the hill."
• Hetu (reason): "Because there is smoke."
• Udāharaṇa (example / general rule): "Wherever there is smoke, there is fire — as in a kitchen."
• Upanaya (application): "The hill has smoke of that kind."
• Nigamana (conclusion): "Therefore there is fire on the hill."

One of the earliest surviving statements of this five-part structure is not in a philosophical text at all — it is in the Caraka-saṃhitā, a medical compendium. Debate training was part of a physician's education, because diagnosis is inference under uncertainty in front of people who may disagree with you.

The obvious Western reaction is that this is Aristotle's syllogism padded out to five steps. That reading misses what the extra members do. They are not logical redundancy; they are dialogical moves. The thesis announces what is in dispute so the opponent knows what to attack. The example supplies a case both parties already accept, which is how you establish shared ground rather than asserting it. The application connects that agreed case to the contested one. The conclusion restates what has now been established, for the benefit of an audience that has been listening rather than reading.

The form was initially strongly analogical — the kitchen is offered as a *case like this one*. Later authors made the universal connection between reason and conclusion progressively more explicit and more deductively stringent. Dignāga and Dharmakīrti eventually argued that three members suffice, converging on something closer to the Greek shape. But the five-member form stayed the pedagogical standard, because teaching people to argue and teaching them to prove are not the same task.

Three Kinds of Debate — and 22 Ways to Lose

Nyāya authors classified debates by purpose, and treated the classification as load-bearing:

• Vāda: cooperative, truth-oriented discussion — both parties want the right answer
• Jalpa: competitive debate aimed at victory, using any legitimate means
• Vitaṇḍā: attacking the opponent's position without defending a positive thesis of one's own

This is a meta-level move Aristotle never quite makes. The Greek tradition separates dialectic from eristic, but Nyāya turns dialogue type into a formal category with different admissible moves in each — which is exactly what contemporary argumentation theory rediscovered when Douglas Walton catalogued persuasion, inquiry, negotiation, deliberation, information-seeking, and quarrel as distinct dialogue types with distinct norms.

The manuals then catalogue what can go wrong. There are hetvābhāsa — "pseudo-reasons", the fallacy list proper, including the inconclusive reason, the contradictory reason, the reason contradicted by perception, the unestablished reason, and the reason that would equally prove the opposite. There are illegitimate rejoinders (jāti), the sophistical moves an opponent uses when they have nothing better. And there are the nigraha-sthānas: the 22 grounds of defeat.

The nigraha-sthāna list is the part with no clean Western analogue, because it mixes two categories the West keeps separate. Some entries are logical defects. Others are procedural: shifting your thesis, shifting your reason, saying something meaningless or incoherent, saying something the audience cannot follow, saying things out of order, omitting a required member of the argument, adding superfluous ones, mere repetition, silence, evasion, admitting your opponent's point, or failing to notice that your own position has already been refuted.

Put plainly: Nyāya had written down burden of proof, relevance, admissible challenge, concession, procedural violation, and defeat — the machinery of a *protocol*, not merely a calculus. This is the part of the tradition that generalizes most directly to a modern meeting, and the part most often left out of the summary.

Vyāpti and the Problem of Induction

Underneath the trairūpya sits vyāpti — invariable concomitance, or pervasion. Smoke is a good reason for fire because smoke pervades only fire-bearing places. The inference works because that relation holds without exception.

Which raises the obvious problem: how would you ever establish a universal from finite observation? Indian logicians met this head-on. Their answer combined repeated observation (bhūyodarśana), the failure to observe counterexamples (vyabhicāra-adarśana), and — the decisive move — a demand for a grounding relation. Dharmakīrti argued that vyāpti is trustworthy when it rests on either a causal connection (smoke is an effect of fire) or an identity-of-nature connection (this is an oak, therefore it is a tree). Accidental correlations do not qualify.

So the Indian answer to the problem of induction is: don't accept a generalization merely because you have seen many instances; accept it when you can say <em>why</em> the connection cannot fail. That is a stronger standard than most boardroom reasoning meets. It is also, roughly, what a modern reader means by asking for a mechanism rather than a correlation.

Prasaṅga and the Tetralemma

Buddhist authors developed prasaṅga — reasoning that draws out a consequence of the opponent's own position that the opponent cannot accept. It is close to reductio, with one important difference in the Madhyamaka use of it: the arguer need not be committed to any thesis themselves. They are showing that your commitments do not survive contact with each other.

Then there is the catuṣkoṭi, the tetralemma, associated above all with Nāgārjuna (c. 2nd century CE): a proposition is examined under four alternatives — it is, it is not, both, neither.

It is tempting to file this as "four-valued logic," and that is where popular accounts usually go wrong. In Madhyamaka usage the four corners are typically not four truth values on offer, one of which you pick. They are four positions the arguer works through and rejects, in order to show that the conceptual presupposition generating the question was defective in the first place. The point is not to land on corner three; it is to stop asking the question in that shape. The precise logical interpretation of the catuṣkoṭi remains actively disputed among specialists — paraconsistent readings, presupposition-failure readings, and purely dialectical readings all have serious defenders. That dispute is a reason to treat it carefully, not a reason to flatten it.

What survives any interpretation is the practical technique: when both the affirmation and the denial of a question look wrong, suspect the question. Anyone who has watched a team argue itself in circles over "is this a build or a buy decision?" — when the real answer is that the framing hides a third option — has met the tetralemma without knowing its name.

The Jain Contribution: Say Which Respect You Mean

Jain philosophers pressed a different point. Anekāntavāda — many-sidedness — holds that a complex reality cannot be exhaustively captured from a single unqualified standpoint. Their syādvāda tradition turns that into a discipline of speech: a sevenfold pattern of conditional predication, each form prefixed with syāt, "in some respect".

Roughly: in some respect it is; in some respect it is not; in some respect it both is and is not; in some respect it is inexpressible; and three further combinations pairing inexpressibility with the first three.

This is routinely misread as woolly relativism. It is closer to the opposite — a demand for precision about scope. You may not assert a partial description as though it were exhaustive. State the standpoint, the condition, or the respect under which your claim holds. The associated parable of the blind men and the elephant is usually told as a story about the limits of knowledge; in Jain hands it is a story about the obligation to qualify your report.

Translate it into a meeting and it becomes a rule with teeth: "the migration is risky" is not a claim until you say risky in what respect — schedule, cost, security, morale. Most unresolvable disagreements dissolve the moment both parties are forced to name the respect they are talking about, because they turn out to have been describing different faces of the same elephant.

But Isn't This Just a Clumsier Syllogism?

It is worth stating the skeptical case properly, because it is the one most Western-trained readers arrive with. It runs: the five-member form is padded; the tetralemma is either trivially about ambiguity or straightforwardly incoherent; Navya-Nyāya never produced a symbolic calculus; and no Indian school proved anything like a completeness result. On the narrow question of formal logic as the twentieth century defines it, the Greek line really does lead somewhere the Indian line did not.

That is fair, and the answer is not to deny it but to notice that it scores one event and calls it the whole meet. Judged as a theory of *formal validity*, Indian logic is not where you go. Judged as a theory of **regulated disagreement** — dialogue types, burden of proof, admissible challenge, grounds of defeat, the separation of reasoning from arguing, the obligation to qualify a claim's scope — it is far ahead of anything in the Organon, and much closer to what modern argumentation theory ended up needing.

The traditions were not answering the same question badly. They were answering different questions well. And the question Indian logicians answered is the one that comes up in every meeting you will attend this week.

Navya-Nyāya: The New Logic

The tradition's technical peak is Navya-Nyāya ("New Nyāya"), founded by Gaṅgeśa Upādhyāya in 14th-century Mithila. His Tattvacintāmaṇi built a specialized Sanskrit idiom for stating logical relations without ambiguity — relational abstracts (the property of being a pot, distinct from a pot), quantification, nested negation (the absence of the absence of X), and precise conditions under which a term applies.

How close is this to predicate logic? Closer than most Western histories admit and further than enthusiasts claim. The apparatus handles quantification and nested relations with real rigour, but it stays inside natural language: there is no symbolic notation, and therefore no calculus you can manipulate mechanically. Whether that is a limitation or a design choice depends on whether you think the point of logic is to be computed or to be argued.

Navya-Nyāya dominated Sanskrit scholarship — spilling into grammar, poetics, law, and theology — until colonial-era disruption of traditional education. Its serious comparison with Western logic is barely a century old.

What Indian Logic Offers a Modern Team

Strip away the Sanskrit and five things transfer directly:

• Name the dialogue type first. Vāda, jalpa, or vitaṇḍā — are we solving this together, competing, or is someone just attacking? Most bad meetings are two people playing different games without saying so.
• Define defeat in advance. The 22 grounds exist so that "you've lost this point" is an observation rather than an insult. Agree what counts as a conceded point before you need the rule.
• Hunt the dissimilar case. Dignāga's third condition is a standing instruction to look for the counterexample rather than pile up confirmations. It is the discipline behind steelmanning.
• Demand a mechanism, not a correlation. Vyāpti is only trustworthy when grounded in cause or nature — the ancient version of "correlation is not causation", with a stated remedy.
• Qualify the respect. Syādvāda's discipline: no unqualified claims about complex things. Say which respect you mean, and half your disagreements evaporate.

Modern argumentation theory arrived at strikingly similar places by a different route. Toulmin's "backing" plays the role of the Nyāya udāharaṇa. Walton's critical questions test, case by case, what vyāpti tests in general. Pragma-dialectics' rules of critical discussion are a nigraha-sthāna list with different manners. And the whole enterprise of argument mapping is an attempt to make the structure visible enough to check — which is what a five-member form was doing out loud.

None of this replaces Aristotle. It completes the picture. The Greek tradition asked whether a conclusion follows and answered brilliantly. The Indian tradition asked how disagreement gets settled between people who both think they are right — and wrote down the answer, including the part about how to lose.

The Roots of Reasoning Series

This article is part of a series exploring the world's argumentation traditions: