Chapter 2: Examining Cognition
Chapter 1 defined the object; this one defines the subject. The defining characteristic of cognition is clear awareness, and it holds across every division: conceptual or non-conceptual, erroneous or not, valid or non-valid. Divided by its objects cognition is many, but taken from within it is one thing only, self-awareness. The second half sets out valid cognition and then, at considerable length, the non-valid kinds.
Condensed from Khenpo Sodargye’s Chinese lecture transcripts on the Pramāṇayuktanidhi, and rendered into English. Root verses are given as the lectures give them; the prose beneath summarises the exposition. On the provenance and the limits of this rendering, see the series introduction.
Chapters: 1. Objects · 2. Cognition · 3. Universals and Particulars · 4. Proof and Exclusion · 5. Expressed and Expresser · 6. Connection · 7. Contradiction · 8. Definitions · 9. Perception · 10. Inference for Oneself · 11. Inference for Others
The Cognition that Knows
Two parts: (1) definition and classification; (2) ascertaining its meaning.
1. Definition and Classification
The defining characteristic of cognition is clear awareness.
Having defined the object in Chapter One, the text now defines the subject. Cognition is clear, luminous knowing. This holds across every division — conceptual or non-conceptual, erroneous or not, perceptual or inferential, valid or non-valid: a cognition lacking clear awareness does not exist.
Against the older definition. Some earlier pramāṇa scholars defined cognition as “the subject that knows an object.” Two overextension objections were raised against them. First, an infinite regress: cognition would be defined by the object, and the object by the subject, so knowing either definition would require knowing the other first, endlessly. Second, if only what cognises an existing object counts as cognition, the apprehension of a double moon would not be cognition, since no double moon exists. Rongtön Paṇḍita answered the second by parallel reasoning: on that logic, the cognition of no-self would also fail to be valid, since no self exists and hence no “no-self” object either. This parallel overturns the objection, and is the Sakya position. (Sakya study of the Pramāṇayuktanidhi relies chiefly on Rongtön’s Commentary on the Treasury of Reasoning and on Gorampa.)
Objection from Cittamātra. If cognition is by definition clear awareness, and all phenomena are mind-only, would pillars and vases — being of the nature of cognition — become sentient? No. On the Cittamātra view no object exists apart from cognition at all, so there is no insentient stone “outside the mind” whose awareness one could ask about. As in a dream, the outer object is merely the appearance-aspect of the clearly-aware mind. The same point is taught in the Commentary on the Madhyamakālaṃkāra: the essence of cognition is luminous awareness.
In terms of objects, cognition divides into many kinds; In terms of cognition itself, it is one — self-awareness alone.
Divisions of cognition run several ways: by object, into non-conceptual cognition apprehending the particular and conceptual cognition apprehending the universal; by mode of engagement, into valid and non-valid; by its own nature, into erroneous and non-erroneous; and into primary mind and mental factors. Every division shares the one defining characteristic of clear awareness.
Gathered inwardly, all of them reduce to a single self-awareness. Luminous clarity pervades every cognition, so eye-, ear-, or nose-cognition, erroneous or conceptual, looked at from within is self-aware luminosity and nothing else. As the Commentary on the Madhyamakālaṃkāra notes, an eye seeing a multicoloured object has cognitions taking red, blue, white, and shape, yet all fall within one eye-cognition; likewise all cognitions fall within self-awareness. The practical consequence: penetrate one’s own mind and one penetrates all phenomena; fail to penetrate the mind’s essence and neither outer forms nor conceptual universals are truly understood.
2. Ascertaining the Meaning
Two parts: (a) definition of valid cognition; (b) non-valid cognition.
a. Definition of Valid Cognition
Deferred — given later in connection with perception and inference.
b. Non-Valid Cognition
Two parts: (i) refuting others’ positions; (ii) establishing one’s own position.
i. Refuting Others’ Positions
Two parts: (1) stating the opponents’ view; (2) refuting that view.
1) Stating the Opponents’ View
All cognition falls under four types: non-realisation, doubt, erroneous cognition, and valid cognition — the first three being non-valid. Sakya Paṇḍita accordingly holds non-valid cognition to be threefold. The earlier Tibetan pramāṇa tradition (Chapa Chökyi Sengé, Tsang Nakpa and others) instead counts five.
Appearance-without-ascertainment, investigative thought, Subsequent cognition, erroneous cognition, and doubt.
The five: investigative thought (ཡིད་དཔྱོད་), appearance-without-ascertainment (སྣང་ལ་མ་ངེས་པ་), subsequent cognition (བཅད་ཤེས་), erroneous cognition (ལོག་ཤེས་), and doubt (ཐེ་ཚོམ་). The last two agree with Sakya Paṇḍita’s own system and are not refuted; the first three are the targets.
i. Investigative thought. [Earlier tradition] A cognition ascertaining a previously unknown hidden fact in a way that happens to agree with reality, without depending on inference — in modern terms, a lucky guess: “I reckon a guest may come today”; “there may be treasure in that spot.” (This differs from the Abhidharmakośa’s use of the term.) Jigme Phuntsok glossed it simply as the ordinary person’s guess-based judgement. It is threefold:
- Reasonless. “This was my grandparents’ well, so there must be water in it” — no genuine ground, since water centuries ago does not entail water now. A guess without reason about something that does not in fact exist is not investigative thought at all; only a groundless guess that happens to hit an existing object qualifies. (Some object that the well case is not truly reasonless; the answer is that past water is not a reason for present water.)
- Faulty-reasoned. Asserting a thesis on an inconclusive or unestablished reason. Inconclusive: “there is fire on the hill, because of something grey” — the grey may be smoke or fog; likewise “I hear gushing, so there is a stream.” Unestablished: “there is fire on the hill, because I see a yak.”
- Inconclusive-reasoned. Someone ignorant of three-aspect inference says “a pillar must be impermanent, because it is produced.” The words are those of a correct inference, but he is guessing, not running the three aspects.
Positions compared. Sakya Paṇḍita: none of the three exists; all collapse into doubt. Mipham Rinpoche: investigative thought exists and falls under non-realisation. Does a “guessing” state of mind exist in the world? Certainly — and it is neither valid cognition nor doubt, since doubt’s mark is that while one thought (“there might be water”) is still active its opposite arises. On Mipham Rinpoche’s analysis, from the side of how the certainty is generated (no real reason) it is non-realisation; from the side of the certainty happening to match fact it looks like valid cognition — but a validity that is neither perception nor three-aspect inference, so it cannot be counted among the two pramāṇas.
ii. Appearance-without-ascertainment (seeing-without-ascertainment). [Earlier tradition] Its mark: manifesting the object’s own-characteristic without contradicting the superimposition of the opposite. It is non-conceptual, whereas investigative thought is wholly conceptual. Superimposition is removed only by certainty; this cognition is a perception, and perception does not remove superimposition. The own-characteristic has appeared before it, yet no genuine certainty was generated — which is what marks it off from real perception and inference. Threefold:
- Object not appearing clearly. Lightning flashing red across the eye, unregistered; or, as tantra teaches, the dharmatā-bardo flashing past at death. Eye-cognition did occur, since object, attention and faculty were complete in the preceding instant, so cognition must arise in the second — but an ordinary being cannot register it.
- Mind not attentive. Absorbed in an item in a shop, one looks straight at an acquaintance and later cannot recall it (“Did I? I didn’t see you!”). Dharmakīrti is said to have eaten food a princess had laced with a bitter herb while composing the Seven Treatises, and tasted nothing — he did taste it, but absorbed in composition did not register it.
- Deranged faculty. Taking a yellow conch for gold, a shell for silver. Something appeared, but the derangement blocked its true essence.
It falls within the domain of perception. From the side of the object actually appearing it can be called genuine perception; since no ascertaining certainty was produced, from that side it is pseudo-perception. The test: after genuine perception one can afterwards generate certainty (“I just saw a red pillar”); after appearance-without-ascertainment one cannot, however much a friend insists one looked at him. Hence Mipham Rinpoche places it under non-realisation; Sakya Paṇḍita does not accept it as a separate non-valid cognition at all.
Warning for practice: many treat investigative thought as valid — “it must be so, because I feel it is so.” “I feel” is not an establishing reason.
iii. Subsequent cognition. [Earlier tradition] A cognition taking its object in a manner wholly opposed to the opposite superimposition while the force of a prior knowing has not yet faded. It has a perceptual portion and an inferential portion; the inferential portion lets it remove superimposition entirely. It means knowing again what one already knew, and can follow either perception or inference — the certainty has been taken as object once already, so a second taking is no longer fresh. Threefold:
- Perceptual. Seeing the blue pillar in the first instant and the same pillar in the second; the continuum from the second instant on.
- Post-perceptual. First instant, perceiving the red pillar; second instant, the thought “oh, I just saw a red pillar,” and all subsequent conceptual thoughts.
- Post-inferential. Inferring fire from smoke in the first instant; all later thoughts of fire’s presence.
Positions compared. Sakya Paṇḍita: non-valid, because its object contains nothing new — re-knowing what is fully known serves no purpose. Mipham Rinpoche: valid, and specifically the result of valid cognition (perception and inference being its causes), because it accords with fact — post-perceptual subsequent cognition agrees with the perceived object, post-inferential with the inferred principle. These do not conflict; Jigme Phuntsok’s ruling was that the two omniscient masters refute and establish from different focal points, and the content is the same. Khedrup Je holds subsequent cognition to be recollection.
2) Refuting That View
Three parts: (a) refuting the essence of investigative thought (Sakya Paṇḍita: apart from doubt there is no such thing); (b) refuting that appearance-without-ascertainment is non-valid (it should be classed as valid perception); (c) refuting that subsequent cognition is distinct from non-realisation.
a) Refuting the Essence of Investigative Thought
Two parts: (i) general refutation; (ii) specific refutation.
(i) General Refutation
Three parts: (1) refutation by examination; (2) pointing out overextension; (3) refutation by parallel reasoning.
(1) Refutation by Examination
Investigative thought, in the end, does not depend on a reason — It is mere thesis-assertion and so collapses into doubt. If it does depend on a reason, that reason cannot go beyond A genuine reason or one of the three pseudo-reasons.
The dilemma. Does the alleged investigative thought depend on a reason or not? If not, it is bare thesis-assertion — a guess — and collapses into doubt. If it does, that reason must be either genuine or one of exactly three pseudo-reasons; no fourth kind exists.
A genuine reason has the three aspects complete: “there is fire on the hill, because of smoke,” smoke pervading similar instances and absent from dissimilar ones. The three pseudo-reasons:
- Contradictory — the reason is established of the subject but contradicts the thesis: “pillar is permanent, because it is impermanent.” One cannot call this unestablished, since the pillar really is impermanent.
- Inconclusive — “there is fire on the hill, because there is something grey”; “you must be able to speak, because you are an animal.”
- Unestablished — “there is fire on the hill, because there is a yak there.”
Each pseudo-reason already has its home: cognitions from contradictory reasons are erroneous cognition, from inconclusive reasons doubt, from unestablished reasons non-realisation. So on either horn there is nothing left for “investigative thought” to be: a correct reason yields inference (valid cognition), a faulty reason yields one of the three non-valid cognitions. Positing it separately serves no purpose whatsoever.
(2) Pointing Out Overextension
If the two reason-based investigative thoughts Are not subsumed under the three pseudo-reasons, There would come to be a fourth pseudo-reason.
The question is put to the earlier tradition about their two reason-based types (faulty-reasoned and inconclusive-reasoned): can these be subsumed under unestablished, inconclusive and contradictory? If not, a fourth pseudo-reason follows — which contradicts Dignāga’s Pramāṇasamuccaya and Dharmakīrti’s Seven Treatises, neither of which posits one. The opponent cannot deny the subsumption either, since he himself admits the faulty reason is unestablished or inconclusive, and the inconclusive-reasoned case uses a reason he grants is insufficient.
If investigative thought is not a pseudo-reason, Then a third pramāṇa would have appeared.
[Earlier tradition] Then let the reason be genuine, not a pseudo-reason. [Reply] But you explicitly deny that investigative thought is either perception or inference. A genuine-reasoned cognition that is neither would be a third pramāṇa alongside perception and inference — which no pramāṇa treatise admits.
If, being genuinely ascertaining, it is still not a pramāṇa, Then inference too would not be a pramāṇa.
[Earlier tradition] Investigative thought genuinely ascertains and judges things rightly (“there is certainly water in the old well,” and there is), yet is not a pramāṇa. [Reply] Parallel reasoning: inference also ascertains — “there is certainly fire on the hill, because of smoke.” If ascertainment is compatible with non-validity, inference too must be non-valid. This closes the opponent’s emergency repair.
If, being ascertaining, it still did not remove superimposition, They would not stand in the relation of eliminator and eliminated.
[Earlier tradition] Investigative thought (“a guest will surely come today”) ascertains but does not remove the opposite superimposition — that is how it differs from inference. [Reply] Ascertainment and superimposition are mutually exclusive, standing as eliminator and eliminated; this is stated in the Pramāṇavārttika (“ascertainment and superimposition are eliminator and eliminated”). Reasoning confirms it: where unhesitating certainty about fire has arisen, the opposite superimposition cannot coexist — they are as incompatible as fire and water. The claim contradicts Dharmakīrti, reasoning, and everyday experience alike.
[Objection:] Though it lacks ascertainment, The true [investigation], the false [erroneous cognition], and doubt — these three do not contradict each other.
[Earlier tradition] Genuine investigative thought, though not decisive, accords with fact. Thinking a guest will come and one comes — genuine investigative thought; thinking none will come and one comes — erroneous cognition; thinking “perhaps, perhaps not” — doubt. All three coexist without contradiction, so the ground for positing investigative thought is solid. (The verse opens with the standard marker of an opponent’s voice.)
[Reply:] Without ascertainment, your investigative thought is lost; And if those two are established, doubt cannot be established.
Two consequences. First, granting that it lacks ascertainment destroys the opponent’s own definition, which required accordance with fact and ascertainment. Second, if both the genuine (“water ought to exist”) and the erroneous (“water does not exist”) are established, doubt has no room left to stand. The opponent can reply that his doubt is precisely the double-minded state in which the first thought has not ceased when its opposite arises — but the Autocommentary answers: “an examination that cannot generate certainty is what the wise call investigative thought” (“the wise” being masters such as Devendrabuddhi). Guessing is all it amounts to, and guessing has no sufficient reason.
Where it finally belongs. Sakya Paṇḍita: doubt — one takes oneself to be affirming or denying, but the reasons are insufficient. Mipham Rinpoche: non-realisation — no real certainty was generated, existence was not settled by perception, and the inferential ground is insufficient. Both agree it cannot be valid cognition: not perception, since perception is non-conceptual and guessing is conceptual; not inference, since inference needs reasons anyone would accept. If a guessed guest arrives it is luck; from the object-side it looks valid, from the subject-side there is no reason and no ascertainment. Valid cognition requires ascertainment and the removal of the opposite superimposition.
(3) Refutation by Parallel Reasoning
Or: just as genuine investigative thought exists, Why should erroneous investigative thought be unreasonable?
The earlier tradition calls all three of its types “genuine” because the guess matches fact. But if ascertaining a previously unknown hidden fact in agreement with fact is genuine investigative thought, then ascertaining one contrary to fact should be erroneous investigative thought. “I shall go there tomorrow” and I go — genuine; and I do not go — erroneous. Why should the latter be unreasonable?
If you accept it, you lose your genuine investigative thought.
[Earlier tradition] Very well, we accept erroneous investigative thought too. [Reply] Then you have destroyed your own definition, which requires accordance with fact.
If not, the former is likewise unreasonable.
[Earlier tradition] Then we deny it. [Reply] If erroneous investigative thought is unreasonable, genuine investigative thought is unreasonable on exactly the same grounds.
A two-sided wavering becomes doubt; And if admitted, the fixed number is ruined.
[Earlier tradition] Erroneous investigative thought is unreasonable because “a guest will / will not come” is two-sided and therefore doubt. [Reply] Then genuine investigative thought equally entails doubt and falls with it. And if you do admit it, your settled enumeration is ruined: a sixth non-valid cognition has been born and the number five is lost.
If erroneous investigative thought is erroneous cognition, By parallel reasoning, the genuine should become a pramāṇa.
[Earlier tradition] Erroneous investigative thought is simply erroneous cognition. [Reply] By the same parallel, genuine investigative thought is correct; if correct, it should be a pramāṇa and fall under inference — not the non-valid cognition you claim it to be.
(ii) Specific Refutation
Three parts: (1) refuting reasonless investigative thought; (2) refuting faulty-reasoned investigative thought; (3) refuting that scripturally-derived cognition is investigative thought.
(1) Refuting Reasonless Investigative Thought
If reasonless investigative thought is genuine, Then everything whatsoever becomes genuine.
If a groundless claim that water is in the well counts as genuine, so does any groundless claim at all: “there must be treasure under my house”; “tomorrow at seven I shall fly through the sky displaying miracles.” Baseless fabrications would all become genuine. In debate, “I think so” and “I feel it is so” establish nothing; assert a thesis only with a reliable reason. Applied inwardly: “perhaps I shall soon attain realisation” — on what ground, if one does not practise? Guess, not reason. Pith-instructions describe this as taking the sharp sword of wisdom to one’s own mind and unravelling the net of conceptual thought.
(2) Refuting Faulty-Reasoned Investigative Thought
If investigative thought rests on a faulty reason, It contradicts the intent [of Dharmakīrti].
“Intent” means the intent of Dharmakīrti’s treatises. In the Seven Treatises the certainty generated from an inconclusive reason is placed under doubt, and that from an unestablished reason under erroneous cognition. To call these certainties “investigative thought” contradicts him outright. There is also an internal incoherence: the earlier tradition says the object judged by investigative thought is real, while also saying inconclusive and unestablished reasons produce it — but a conclusion drawn from such reasons cannot be counted on to land on the truth.
Distinguishing the pseudo-reasons in practice: establishing impermanence by “being an object of cognition” is inconclusive; “this vase is impermanent, because it is a pillar” is unestablished; establishing impermanence via permanence is contradictory.
Gorampa adds that so-called reasonless investigative thought is not really reasonless — the earlier presence of water in the well is being used as a reason — so classing it as “no reason whatsoever” is itself wrong.
(3) Refuting That Scripturally-Derived Cognition Is Investigative Thought
[Earlier tradition] Investigative thought must exist, since the treatises of both Buddhist and non-Buddhist traditions mention it: the Great Commentary on the Abhidharmakośa describes certain stages of the path of preparation as investigative thought or subsequent cognition, and the Abhisamayālaṃkāra likewise calls some path-of-preparation wisdoms investigative thought.
Cognitions arising from one’s own or others’ scriptural treatises Are subsumed under the four cognitions — they are not investigative thought.
Every cognition, scripturally derived or not, falls among the four. Whatever in scripture is fully reasonable — the Four Noble Truths, the two no-selves, Śākyamuni’s being a genuine valid person — is known by perception or inference, hence valid cognition. The non-Buddhist claim that bathing in the Ganges purifies sin is erroneous cognition. Wondering whether one’s own obscurations can be purified by practice is doubt. A thought that simply lacks knowledge of the truth is non-realisation. No fifth is needed.
The same fourfold test applies to one’s stance towards Buddhism itself: finding the teaching on past and future lives reasonable through perception or inference is valid cognition; rejecting them (like holding impermanent sound to be permanent) is erroneous cognition; failing to understand without adopting a perverse view is non-realisation; wavering is doubt. Sakya Paṇḍita later notes that mental states analysed in detail are limitless but gathered by kind are few — which is why cognitions failing to know the true principle all reduce to non-realisation. Buddhism has both characteristics, summary and extension.
b) Refuting That Appearance-without-Ascertainment Is Non-Valid Cognition
If appearance-without-ascertainment Is not a genuine pramāṇa, All perceptions would become non-pramāṇa, Since perception has been excluded from having ascertainment.
The consequence: if lack of ascertainment is what makes appearance-without-ascertainment non-valid, then sensory, mental, self-aware and yogic perception are all non-valid too — because no perception is ascertaining, a point Dharmakīrti stresses throughout the Pramāṇavārttika. Perception would drop out of valid cognition entirely, leaving inference alone. Sakya Paṇḍita’s positive claim: what you call appearance-without-ascertainment is nothing over and above perception (specifically pseudo-perception, like seeing two moons, where the faculty is impaired and so cannot veridically deliver the own-characteristic).
Why Mipham Rinpoche can still call it non-valid. Because it can never produce certainty from its own power. Ordinary sensory and mental perception can: certainty follows afterwards by their force. Appearance-without-ascertainment cannot — someone brushed past in a flash, and however much others insist afterwards, that perception’s own strength yields nothing; likewise the shell-seen-as-silver produces no certainty on its own. Certainty may arrive by another’s strength, but that does not belong to the perception itself.
No real conflict. From the standpoint of seeing, both masters call it perception — Mipham Rinpoche granting that the object was veridically seen, only too briefly (or through a deranged faculty) to yield certainty; and that is exactly why Sakya Paṇḍita refutes the earlier tradition. From the standpoint of generating certainty, no certainty can ever arise from it, and that is why Mipham Rinpoche files it under non-realisation. Three positions, finely separated: earlier tradition, Sakya Paṇḍita, Mipham Rinpoche.
A single verse suffices here; the refutation is not extended.
c) Refuting That Subsequent Cognition Is a Non-Valid Cognition Distinct from Non-Realisation
Two parts: (i) general refutation; (ii) specific refutation.
(i) General Refutation
If subsequent cognition is examined by reasoning, It becomes either non-realisation, Erroneous cognition, or valid cognition — No other cognition beyond these is admissible.
Since only four cognitions exist, subsequent cognition must fall among them. It is not doubt, since doubt is double-minded while subsequent cognition re-cognises an already-ascertained object. The Autocommentary, with Jamyang Loter Wangpo’s commentary agreeing, places it under non-realisation, on two grounds: it merely recollects what was already experienced, and it generates no newly apprehended object. Reusing a used object is no longer fresh.
Not therefore worthless. Subsequent cognition is indispensable in practice, especially as presented in the Abhisamayālaṃkāra: much of the path of preparation is subsequent cognition, since the non-dual wisdom of the first bhūmi is not yet directly realised and the practitioner repeatedly familiarises himself with views already ascertained on the path of accumulation — no-self, the impurity contemplation, the impermanence of all phenomena. To dismiss it as mere improper attention would be wrong. This is why Mipham Rinpoche calls it the fruit of valid cognition.
How the two views combine. On Mipham Rinpoche’s reading it is pramāṇa from one side (produced by the force of perception and inference, not self-produced) and non-valid from the other (no new information). Sakya Paṇḍita in fact grants the same: the second-instant thought “ah, the pillar is red,” or the continued knowledge that the pillar is impermanent, does accord with the object, and from the subject-side he too accepts it as pramāṇa; from the object-side its content is a universal already known, hence non-valid.
Rongtön’s Commentary sharpens the point with an unestablished-reason analogy: telling Dharmakīrti “pillar is impermanent, because produced” proves nothing to him, since the thesis is long established for him. Establishing again what is already established for the interlocutor is superfluous — as is re-cognising for oneself what one has already ascertained.
A note on “instant.” The instant meant here is the siddhi-kṣaṇa (completion-instant), not the kāla-kṣaṇa (time-atom-instant), in which no inference could occur at all. The Mahāyāna Abhidharma posits the completion of objects of abandonment and their antidotes in siddhi-kṣaṇa for the same reason. So: first siddhi-kṣaṇa, direct perception; second, conceptual thought arises; thereafter, subsequent cognition.
A caution against misusing the category: one hearing of a teaching does not make further hearings redundant “subsequent cognition.” Patrul Rinpoche heard the Words of My Perfect Teacher twenty-five times from Rigdzin Gyalwa’i Nyugu.
(ii) Specific Refutation
Two parts: (1) genuine refutation; (2) removing objections.
(1) Genuine Refutation
Two parts: (a) refuting the co-substantiality of contradictory factors; (b) refuting that a continuum-type of sense cognition is a pramāṇa.
(a) Refuting the Co-substantiality of Contradictory Factors
[Earlier tradition] Only they hold this. Of ten instants seeing a red pillar, the first is pure perception with no subsequent-cognition component; from the second on, while the perception-continuum has not ceased, each instant has both a perceptual component and a re-knowing component. So perception and subsequent cognition are co-substantial.
Perception is present and non-conceptual; Subsequent cognition is past and conceptual. Object, time, and mode of apprehension are opposed — How could these two share a single substance?
Four contrasts drawn from essence, object, time and mode. Perception: non-conceptual in essence; its object present; its object the own-characteristic; its mode clear and distinct. Subsequent cognition on the opponent’s own account: conceptual; its object the universal; its time the recollection of past events; its mode blurred and mixed, since concepts apprehend through universals. One non-conceptual, one conceptual; one present, one past; one own-characteristic, one universal; one clear, one blurred. Contradictories cannot share a substance.
Karmapa Chödrak Gyatso’s objection (in his Ocean of Reasoning): perception arises dependent on the preceding instant’s faculty, attention and object, so is its object past or present? If past, it is no different from subsequent cognition; if present, subject and object are simultaneous, which is the earlier tradition’s own error.
Reply. The analysis of objects here stands on the Sautrāntika standpoint, and no fault follows. Conceptual thought ranges widely in time — the preceding instant, yesterday, last year, centuries ago; anything recollectable can be its object. Perception’s object is effectively simultaneous: although the current eye-cognition is produced by the preceding instant’s faculty, attention and object, and so is separated from the outer object by one instant, conventional usage does not fix “the present” at the finest instant. Khenpo Lowo pressed this against the Karmapa at length: fixing the present at the finest instant accords with neither Dharmakīrti’s tradition nor any master’s conventional positing. Conceptual thought admits no such latitude — if its object were “present,” events decades past would be present.
(b) Refuting That a Continuum-Type of Sense Cognition Is a Pramāṇa
[Opponent:] “The apprehension of the continuum-type is perception; Later knowing of it is subsequent cognition.”
[Earlier tradition] A repair: seeing red in the first instant is perception, and because the kind-continuum persists, perceptions of red continue while a grasping conceptual thought also runs alongside — that continuum is “perceptual subsequent cognition,” so no single instant has to be two things at once.
The [general] type is not the object of perception; There is no subsequent cognition of the object seen [by perception].
The “type” is a universal: earlier and later blues strung together by conceptual thought and grasped as one “blue.” As yesterday’s river has long since flowed to the sea, the previous instant’s blue has passed and cannot be the next instant’s perceptual object — though it can serve as a conceptual universal. The opponent has confused perception’s object with conceptual thought’s object. (Two senses of “type” should be distinguished: the non-Buddhist type pervading all its instances, which stands up under no cognition at all; and the ordinary universal, which is a conceptual object only.)
The second line: a perceptual object is never the object of the alleged subsequent cognition. The first instant’s pillar has ceased by the second; the eye seeing “the pillar” again sees a new pillar. If the same pillar remained, it would not be impermanent.
Concluding verses:
Because they are contradictory, there is no shared substance.
Present object and subsequent-cognition object wholly contradict, as fire and water — no shared substance is possible.
Because it is perception, it is not recollection.
Perception is present seeing. Post-perceptual subsequent cognition (“I recall the big apple I wanted at the market but could not afford”) and post-inferential subsequent cognition (recalling the selflessness established earlier by the neither-one-nor-many reasoning) are both recollection, reaching back over years or even past lives. The two cannot be confused.
The [general] type contradicts the object of the senses.
The type is a conceptual construct, not really existent; the object of eye and ear is the perceptual object. A type cannot be a faculty’s object.
In direct seeing, there is no difference between earlier and later.
[Earlier tradition] The first seeing is pure perception; everything after is subsequent cognition — so, absurdly, only the first time one saw the red pillar in the classroom was perception, and every day’s seeing since has been non-valid. [Reply] Every instant, earlier or later, is equally perception. Not only does the object differ instant by instant — the past object does not remain in the slightest, and the present arises and ceases momentarily — the subject too changes continuously. The “continuum” is only a conventional label. Ordinary people do say that re-cognising the already-cognised is superfluous, but that is conceptual thought speaking; new objects genuinely arise.
Whatever has ascertainment is not perception.
Whatever weighs, ascertains and thereby removes superimposition is conceptual, and perception has no conceptual thought. Perception’s power may produce conceptual thought, but its own essence is not conceptual. So subsequent cognition, having ascertainment, cannot be perception.
Grasping things as identical is conceptual.
Taking the earlier red pillar and the later red pillar to be entirely the same is conceptual grasping, not a perceived object.
Therefore, by cognition alone, one can refute The notion that perception and subsequent cognition share one substance.
No external authority is needed: check one’s own continuum. Whenever perception is present, recollection is not; whenever recollection is present, perception is not. Perception is sensory or self-aware, and it is through self-aware perception that recollecting conceptual thought later arises. Jamyang Loter Wangpo’s commentary makes the same point.
(2) Removing Objections
When the object’s and the sense-faculty’s atoms gather together, Across however many moments occur, The experiencing cognition that arises Is the perception of ordinary beings.
[Objection] Perception sees the own-characteristic, but things change momentarily and ordinary beings cannot see a momentary pillar — the instant is too brief to be their object. So do ordinary beings have no perception at all?
[Reply] Under ultimate Madhyamaka analysis no pramāṇa withstands examination; but at the conventional, unanalysed level — “worldly-seeing pramāṇa” — valid cognition must be posited, and ordinary perception is not in danger. Its conditions: the object must accumulate a certain number of atoms; the faculty is likewise an atomic aggregate (Sautrāntika holds it cannot perceive partless atoms); the two must combine, with attention, and an instant-continuum must pass before eye-cognition arises.
On the gathering of instants some masters object that instants cannot accumulate as atoms do, since each arises and ceases immediately — otherwise the prior instant would be permanent. Even so, no fault follows for ordinary perception. Analogy: a single drop cannot fill a vase, but many drops can and the result is plainly visible. Suppose twenty instants are needed: eye-cognition arises in the twentieth, not because twenty instants stood together, but because twenty instants’ worth of time has passed. Each instant should yield its own eye-cognition, yet no one generates the certainty “I saw the pillar twenty times in those twenty instants” — ordinary minds are too coarse for that fineness, though every feature has in fact appeared to the faculty. How many atoms and instants at minimum has no fixed standard; it varies by individual eyesight.
The working standard of perception: when an atom-aggregate appears to the eye as a vase, one says “I see a vase perceptually” — the mark being that certainty can be produced on one’s own without relying on another.
Appearing before one does not by itself make perception, however. Seeing two moons is pseudo-perception: within the domain of perception and so labelled, but not genuine, as an imputed person is not a real person. It still falls inside the four cognitions — under erroneous cognition.
An ordinary being’s perception also differs greatly from a noble being’s. Analysed deeply, partless atoms do not stand and all is emptiness; but pramāṇa works conventionally and asks only how seeing, hearing and knowing are established.
Pause-verse
All enumerations must be gathered By means of kind-differentiations; All internal groupings Are subsumed within that respective scope.
Classification must proceed by kind. The definition of “human” is knowing language and meaning; humans divide into men, women and paṇḍakas, which is exhaustive. Subdividing men into eastern and western, then eastern into eastern-eastern and eastern-western, never settles, and no sub-class can be returned to a root class. Dharmakīrti says beings’ wrong views are boundless; without grouping by kind, the boundless could never be gathered into five, three, or one. The Abhidharmakośa’s long lists of mental factors make the same demand.
An application: is there a cognition that is neither perceptual nor inferential — say, faith in the Three Jewels? It is not non-valid; it is conceptual, so not sensory perception; it uses no reasoning, so not inference. It is subsumed under self-aware perception (perception having four kinds). Every cognition finds a chief category this way.
The earlier tradition’s own failure is here: they accept grouping by kind, but since they do not accept non-realisation, their five cannot in fact be gathered under any one. On Sakya Paṇḍita’s arrangement, non-valid cognition falls under three (non-realisation, erroneous cognition, doubt), and those three finally under non-realisation. Why keep three rather than one? Because the threefold division answers a specific need (the analogy: dividing a monastery into men’s and women’s regions, and subdividing further where there is reason to). Divisions made for a purpose are fine; divisions multiplied without purpose are endless. The blade drawn from a sheath belongs to that sheath and must be returned to it.
ii. Establishing One’s Own Position
Three parts: (1) general definition of non-valid cognition; (2) enumeration; (3) individual accounts.
1) General Definition of Non-Valid Cognition
That which is not non-deceptive is non-valid cognition.
The double negation means simply: a deceptive cognition. Acting on doubt, erroneous cognition or non-realisation to find their object, one is deceived — things do not turn out as imagined. Pramāṇa will be correspondingly defined as non-deceptive cognition.
The three-sided test. Why is “there is water in the old well” non-valid when there really is water? Because non-deceptiveness, treated in Chapter Nine and in Chapter Two of the Pramāṇavārttika, has three sides — agent, action and object. Object-side: no deception, since water is there and acting on it succeeds. Agent-side: deception, since the agent had no perception and no genuine inferential reason. One deceptive side is enough to make the cognition non-valid.
This definition pervades every non-valid cognition, conceptual or not. Measured by it, much ordinary judgement is non-valid: denying karma, denying the Three Jewels’ blessing, or treating one’s own untested view as right — like a madman who takes the sane for mad. Such views persist until a genuine teacher and sublime treatises meet the right conditions, at which point non-valid cognition is displaced by valid.
2) Enumeration
Non-realisation, erroneous cognition, and doubt — These three are the opposites of valid cognition. They are divided by their mode of apprehension; In essence they are subsumed as one.
Two questions: how many, and into what they reduce. The earlier tradition’s five omit non-realisation, which is precisely why their five cannot be gathered under any one head. Sakya Paṇḍita’s three reduce conveniently to non-realisation.
The division is by mode of apprehension, not essence: failing to understand the object truly is non-realisation; grasping the opposite’s superimposition is erroneous cognition; double-minded apprehension is doubt. By essence they are one. Doubt does not penetrate reality; erroneous cognition (the pillar grasped as permanent) does not penetrate reality; so both fall under non-realisation, which some treatises simply call ignorance. The same applies to the countless non-valid states of ordinary minds — including aesthetic judgements, since beauty and ugliness have no intrinsic basis and are fabricated by conceptual thought. The Buddha’s teaching corresponds: all the infinite conceptual thoughts of beings reduce to afflicted ignorance, eliminated by wisdom; every concept falls either on the side that penetrates reality (wisdom) or the side that does not (ignorance).
The chain of reductions — pramāṇa’s “secrets,” also given in the Autocommentary: all non-valid cognition reduces to non-realisation; all valid cognition to perception and inference; inference ultimately to perception; perception to self-aware perception; outer objects to the own-characteristic (also one of the five characteristic features of Śāntarakṣita’s Madhyamakālaṃkāra).
3) Individual Accounts
The three are kept distinct because each has its own definition and classification.
a) Non-Realisation
Neither grasping [the object] as itself nor as other — Such non-realisation is of three kinds: Not yet having engaged the object; having engaged but not completed; Having completed, yet still not attained.
Definition: a cognition that neither truly grasps the object as it is nor misgrasps it as its opposite. The vase is impermanent; grasping it as impermanent would be valid (perception or inference), grasping it as permanent would be erroneous; non-realisation is neither, merely seeking the essence without finding it. Everyday “I don’t know” is a form of it. Examples: taking the Buddha for a compassionate old man, a god, or an immortal — missing the truth, but without denying him or calling him bad.
A distinction to keep clear. The final-subsuming non-realisation (under which doubt and erroneous cognition are gathered) is not the same as non-realisation as one of the three classifications. The former is general, posited from the standpoint of failing to know reality. Their definitions differ; their essence is the same. The parallel: by essence all object-awarenesses are subsumed under self-awareness, while by object-apprehension eye-cognition takes form and ear-cognition sound, unconfused. Hence the verse’s “divided by mode of apprehension.”
“Realising” means penetrating, accomplished by perception or inference. “The Buddha’s teaching is unreasonable” rests on neither — hence non-realisation.
(i) Not yet having engaged the object. The mind has not yet taken up the object at all: telling someone unfamiliar with three-aspect inference that the pillar is impermanent, or explaining the virtues of the Three Jewels to someone who has never taken refuge.
(ii) Having engaged but not completed. The examination has begun but not finished: having studied pramāṇa incompletely, one ponders the pillar’s essence with three-aspect reasoning without yet reaching the conclusion. Likewise one may have taken refuge while the real principles remain unknown.
(iii) Having completed, yet still not attained. The examination is finished but the result has not been obtained. Per the Autocommentary and Rongtön’s Day-Storehouse Commentary, three subtypes:
Objectless non-realisation. The object’s essence does not exist — a barren woman’s son, a rabbit’s horns. The Autocommentary places subsequent cognition here. Does subsequent cognition then have no object? It has one, but not a newly known one. If re-knowing counted as knowing, an overextension follows: one would have to prove the pillar’s impermanence to Dharmakīrti; a child once born would have to be born again, one who has died would have to die again. The second-instant recollection is just that — recollection, as Khedrup Je and many Indian masters also say. On this Mipham Rinpoche and Sakya Paṇḍita agree: subsequent cognition is non-valid, and belongs to objectless non-realisation.
Mind-defiled non-realisation. One perceives directly, but conceptual thought defiles the cognition — the pillar is perceived and is impermanent, yet thought insists “today’s river is yesterday’s, today’s pillar is yesterday’s.” The object’s true continuum goes unpenetrated.
Non-engageable-object non-realisation. The object is simply not within that cognition’s range, in three ways. Time — ordinary beings perceive only the present pillar, while the Buddha perceives past, future and present. Distance — vultures see meat at distances beyond human sight, akin to one of the four things unknowable to an arhat. Nature — the ordinary being cannot see the pillar’s momentary impermanence or the selflessness of phenomena, however the object appears.
Two related points. Some early Sakya disciples held that perception and non-valid cognition can coexist, using the case of a vase in a dark room: one sees the darkness and not the vase, so within one cognition there is seeing and not-seeing. Later Sakya masters refuted this — no such coexistence is found in the seven treatises.
And relativity: what appears to the ordinary being as a vase is an aggregate of atoms held together by everyday language, tradition and latent habits, so his perception is not ultimate perception. Facing the same vase, an arhat free of self-grasping knows its essence as empty while an ordinary being takes it as truly existent or “mine.” As the Jewel Lamp of Certainty notes, humans see a bowl of water while a hungry ghost beside them sees pus and blood. Mipham Rinpoche’s formulation: relative to one’s own faculty there is a provisional valid cognition; as one’s realm rises the higher is valid and the lower non-valid; ultimately only the Buddha’s seeing is valid.
b) Erroneous Cognition
Grasping it in a way that invalidating evidence for its opposite exists — Such erroneous cognition divides into two: Conceptual and non-conceptual; Each of these subdivides, making five kinds in all.
Definition: a cognition that apprehends its object in a way invalidated by valid cognition of the opposite — grasping the vase as permanent, for instance. The difference from non-realisation: non-realisation merely fails to know; erroneous cognition positively apprehends the opposite essence and is refuted by perception or inference. The Four Hundred Verses’ four inversions belong here — taking the impermanent as permanent, the unclean as clean, suffering as bliss, and the selfless as having a self. Erroneous cognition is essentially wrong view: denying karma, the Three Jewels’ blessing, past and future lives.
Conceptual erroneous cognition — three kinds:
- Appearance-wise. A similar appearance produces an illusion and one thing is taken for another: mistaking a stranger for an acquaintance and greeting him; a patterned rope for a snake; a blue mirage for a river; red cloth for fire.
- Time-wise. Today’s river taken as yesterday’s; today’s person as the same person as before. Similarity across a continuum obscures the difference of times. (Grasping the impermanent pillar as permanent arguably belongs here, though earlier pramāṇa texts do not say so.)
- Location-wise. Someone on a mountain top taken by those below to be standing in the sky; the vase at the door taken as the one inside.
The general mark of the conceptual: it always apprehends “this is such-and-such.” The non-conceptual mirrors the object’s features without judgement or analysis.
Non-conceptual erroneous cognition — two kinds:
- Sense-faculty-defiled. Eye disease producing two moons or floaters; a white conch seen as yellow; a body itching taken for lice that are not there; ghosts seen through faulty eyes. If one goes on to grasp “this is two moons,” that grasping is no longer non-conceptual — only the bare arising of the imagery is. The Great Commentary on the Pramāṇavārttika classes a magician’s audience here: wood, stones and mantras manifest elephants and horses, and the proof that the faculties are deranged is that a bystander with undefiled faculties sees nothing. Television and film, by contrast, are not derangement but genuine perception: the camera captured the object, the transformed image is a form produced by material causes, and the undefiled eye-cognition sees it — as the Abhidharmakośa treats a mirror reflection as a form reflected by the mirror-surface.
- Mind-deranged. Chiefly the non-conceptual mind in dream. Per the Abhidharmakośa the five faculties do not function in dream — they remain supporting-faculties only — and no sense-cognitions arise; the mental consciousness, stirred by beginningless or recent latent seeds, arises deranged and produces dream phenomena. From the appearance-side this is perception, but pseudo-perception. Dream also has a conceptual derangement: grasping “this is an elephant, that is red fire” within the dream, which the Great Commentary classes as conceptual mind-derangement. Longchenpa’s Great Chariot Commentary on Resting in the Nature of Mind joins tantric and pramāṇa instruction here, especially on recognising luminosity in dream. Per pramāṇa’s conventional analysis there is no real cognition in dream, the six consciousnesses being inactive; the Cittamātra reading of the dream as an illusion-like manifestation of mind differs somewhat.
c) Doubt
Grasping it while another side — “not that” — also arises: Such is the mark of doubt. It has two kinds: Manifest doubt and latent doubt, Apprehending in either an even or a weighted manner.
Definition: while apprehending an object, and before that apprehension’s force has faded, the grasping of “not that” also arises. “The pillar is impermanent” — then “perhaps not impermanent.”
The one-continuum objection. Pramāṇa and Madhyamaka hold that a being has one mental continuum, not two. Do “permanent” and “impermanent” then arise at once, making two continua? No: the two do not arise simultaneously. The second arises while the first’s force and latent seeds have not yet gone. Neither side is affirmed, since “impermanent” has no reason offered and “not-impermanent” has no invalidating evidence.
Against a “third product.” Some masters posited a third category alongside valid and non-valid — neither right nor wrong, on the analogy of neutral karma among good, bad and neutral, or a middling person neither bad nor good. Since in doubt neither side has a reason, they held it a middle state. This is unreasonable: a state in which no side has evidence is not neutral but simply non-valid. There is no direct-seeing evidence as in perception and no irrefutable reason as in inference, so nothing is left to be neutral about.
Gorampa’s formulation: while a prior mental function has not been destroyed or faded, a contrary present conceptual thought arises. “Function” matters — it is the latent seed remaining after the thought itself has ceased. Sakya Paṇḍita’s repeated argument that investigative thought falls under doubt rests on exactly this: both sides are groundless. “The old well has water” is said in daily speech, but no evidence stands on either side.
The same test applies to faith: asked about the Buddha’s virtues, some can produce evidence for neither existence nor non-existence and repeat what others say, while doubt remains in the heart. Practice therefore demands certainty — investigate on one’s own or, failing that, through the scriptures of the buddhas, bodhisattvas and masters, until the thought “the Buddha’s teachings are entirely correct” arises as certainty, which is valid cognition. Without it, recognition is only guessing.
A note on scope. All cognitions produced in beings’ continua fall under valid and non-valid, including meditative wisdoms — but this is pramāṇa’s conventional analysis. From the ultimate side even a tenth-ground bodhisattva’s absorption-wisdom is non-valid, a slight latent defilement remaining and not all phenomena’s reality being penetrated. Hence pramāṇa’s repeated statement that the ultimate pramāṇa is the Buddha, the only valid person; and yet, below that level, the ordinary being still has his valid cognition. Both angles are needed.
Classification. Manifest doubt and latent doubt; manifest doubt subdivides into even and weighted.
- Even doubt (“equal-share doubt” in some works). The pillar may be impermanent or permanent, the two graspings exactly equal with no leaning — “one catty” each way; or “perhaps the Buddha’s blessing is there, perhaps not.”
- Weighted doubt. Leaning to one side, and subdivided by whether the lean is reasonable. Reasonable: “this pillar is probably impermanent” — and it really is impermanent, so the lean runs with the facts (discussed in the Teaching on Pure Land). Unreasonable: leaning towards the pillar’s permanence, or towards giving having no fruit.
- Latent doubt (also “hidden doubt”). A doubt lying in the continuum, to mature later but not yet manifest. Investigative thought is placed here, as are groundless opinions and conclusions drawn from inconclusive or unestablished reasons. The Autocommentary’s criterion: whatever has neither establishing evidence nor invalidating evidence is latent doubt. Summarily, Mipham Rinpoche also places investigative thought under doubt.
The three-sided test again: a guest is expected and arrives, so the thought accorded with fact — yet non-deceptiveness requires object, action and agent all to be non-deceptive, and here the agent produced no genuine perceptual or inferential certainty. The agent is deceptive, so the cognition is non-valid.
Concluding Verse
Though valid cognitions are two, All come down to the single pramāṇa of self-awareness; Though non-valid cognitions are three, There is in truth none other than non-realisation.
Sakya Paṇḍita’s summary of the chapter. Valid cognition is perception and inference — perception subdividing into sensory, mental and the rest, inference into effect-inference, nature-inference and so on — and all gather into self-aware perception, since cognition looked at inwardly is self-awareness throughout. Non-valid cognition is non-realisation, erroneous cognition and doubt, and all gather into non-realisation.
A final objection. If classification must go by kind, and doubt and erroneous cognition are kinds of non-realisation, is the threefold division not as arbitrary as the earlier tradition’s fivefold? No — because it answers a specific need, treated in Chapter Eleven. The need is in debate over three-aspect inference. If all non-valid cognition were gathered into non-realisation alone, the only available reply to a faulty reason would be “unestablished,” and the replies “contradictory” and “inconclusive” would be lost. Against “sound is permanent, because produced,” one cannot say unestablished — produced is established of sound — one must say contradictory, since the thesis and the proof are wholly opposed. Against “sound is impermanent, because it is an object of cognition,” the correct reply is inconclusive, and “unestablished” would again be wrong, since being an object of cognition is established of sound. Hence the provisional need for the division, though the ultimate subsumption stands. And no, the single non-realisation need not carry the definitions of doubt and erroneous cognition, since it is posited only from the side of failing to penetrate the object’s reality.
Closing. Mipham Rinpoche’s Great Commentary on the Pramāṇavārttika: apprehender and apprehended are ultimately unestablished, and at the end only individual self-awareness’s domain manifests. Seeing a pillar, the pillar is provisionally the apprehended and the faculty the apprehender; ultimately no duality is found. So valid and non-valid are provisionally posited within cognition; cognition is individual self-awareness; and the outer objects that self-awareness sees do not in fact exist. This is pramāṇa’s ultimate intent.
End of Chapter Two: Examining Cognition
Object and subject have both now been examined. What follows treats how the subject takes the object — the relations through which the object is known — in four aspects.
← 1. Examining Objects | All chapters | 3. Universals and Particulars →