Session 4 · 自義比量品, the inference chapter

What Makes an Inference Actually Valid

The three-part test behind every "that doesn't prove it" you've ever said in this course

Before class: Collected Topics (all six sessions) and Pramāṇavārttika Session 3. Today formalizes what "no pervasion" has meant this whole time.

Part 1

Lecture Notes

The question you've been begging since Session 1 of Collected Topics

Every single time you said "that doesn't prove it — there is no pervasion," across two entire courses, you were relying on an intuition nobody had formally defined for you yet. Why does smoke prove fire, but a rooster's crow doesn't prove sunrise? Why does "impermanent" prove "product," but "has a color" doesn't prove "green"? You could always feel the difference. This chapter is where Dharmakīrti finally writes down the actual test.

The three modes of a valid reason

A reason (a piece of evidence offered to prove something) only counts as logically valid if it passes all three of these, together — not two out of three, all three:

It genuinely applies to the case at hand. If you're arguing about this mountain, there actually has to be smoke on this mountain — not smoke somewhere else, not smoke you're assuming. (This is where "the reason is not established" lives — a reason that fails this mode is simply false.)
Positive pervasion: everywhere the reason occurs, the thing you're proving occurs too. Every genuinely smoky place is a fiery place, no exceptions. (This is where "no pervasion" catches a reason that's real but doesn't force the conclusion.)
Negative pervasion: everywhere the thing you're proving is absent, the reason is absent too. Every non-fiery place is a non-smoky place. (The mirror image of mode 2 — checking the connection holds in reverse as well, not just forward.)

Mode 1 asks: is the evidence real? Modes 2 and 3 ask: does the evidence actually force the conclusion, in both directions?

Notice something: modes 2 and 3 are exactly the two failure points you've been testing this whole time with "no pervasion." You were doing this formal check informally, by feel, from the very first color-and-shape debate. This chapter just gives it its full name: pervasion (khyab pa) — the genuine, checkable, two-directional connection between a reason and what it proves.

Three types of valid reason

Dharmakīrti sorts every reason that actually passes all three modes into one of three families:

Effect Reason

Inferring a cause from its effect.Smoke → fire (Collected Topics Session 6)

Nature Reason

The reason and the conclusion share one entity by very definition."Impermanent, because it's a product" (Collected Topics Session 4)

Non-Perception Reason

Proving an absence, because something that should be perceivable simply isn't perceived."No pot on the floor, because if there were one, in this light, I'd see it — and I don't"

Two threads finally tied together

"Pervasion" was never just a Collected Topics word. It's the technical center of the entire logical system — modes 2 and 3, applied to whatever subject you're debating.
You already knew the three reason-types from experience — smoke/fire (effect), product/impermanent (nature) — before you had names for them. This chapter is the payoff for paying attention back then.

Debate Check — running the full three-mode test

Challenger

"So, it rained last night, because the grass is wet."

Defender

"That doesn't prove it! There is no pervasion."

Mode 1 passes (the grass really is wet). Mode 2 fails: wet grass does not always mean rain — a sprinkler produces the same evidence. A reason that fails mode 2 is called inconclusive, and this is exactly that.

Part 2

Discussion Questions

Go back to Collected Topics Session 1's "has a color, therefore green" trap. Which specific mode does it fail — 1, 2, or 3?
Is mode 3 (negative pervasion) doing real independent work, or is it automatically guaranteed once mode 2 holds? Argue it either way.
Of the three reason-types (effect, nature, non-perception), which feels most intuitively solid to you, and which feels shakiest?
Try classifying "this room is empty of elephants, because if one were here, I'd see it, and I don't" — which type of reason is this, and does it remind you of anything from Collected Topics Session 5?

Part 3

AssignmentChoose one · half to one page · bring it next session

Option A — Run the full test

Take a claim from your own everyday reasoning ("you have a fever, because your forehead feels hot" or similar) and formally check it against all three modes. Does it survive?

Option B — Sort your own examples

Come up with one original example each of an effect reason, a nature reason, and a non-perception reason — none copied from this session — and explain why each belongs in its category.

Part 4

Self-Check

1. Name the three modes a reason must pass to count as logically valid.

(1) It genuinely applies to the subject, (2) positive pervasion — wherever the reason occurs, the conclusion occurs, (3) negative pervasion — wherever the conclusion is absent, the reason is absent.

2. Classify: "sound is impermanent, because it is a product."

A nature reason — "product" and "impermanent" are one entity by definition, not cause-and-effect or an absence.

3. Classify: "there is fire on the mountain, because there is smoke."

An effect reason — smoke is an effect of fire, and the effect is used to infer the cause.

Looking ahead

Session 5 — Proving It to Someone Else

This chapter turns from reasoning-for-yourself to putting a valid inference into words someone else can check (為他比量品) — the difference between knowing something is true and actually proving it to a skeptic sitting across from you. It's also this pilot's capstone, closing the four-chapter arc that began with Session 1.

← Course overview Session 5 →