Seeing vs. Thinking: The 800‑Year Tibetan Debate That Isn’t Over
A Blue Scarf in a Mountain Courtyard
Around 1160, in the chilly debating courtyards of Sangpu Monastery, a student might hold up a blue scarf and ask: “What just happened when you looked at this?” Most of us would say we saw blue. But for the Tibetan monks who followed Chapa Chökyi Senggé (1109–1169), that answer wasn’t finished. For Chapa, seeing itself is already a kind of decision — not something passive like a mirror, but something active that sorts the world without words. A century later, Sakya Pandita Künga Gyeltsen (1182–1251) would come roaring against that idea. Their clash about how the mind touches reality still matters every time you name a color, argue about what counts as a sandwich, or trust your own eyes.
The fight turned on two huge questions. First: does perception — your raw sensory contact with the world — just deliver a picture, or does it already do the work of ruling out what isn’t there? Second: when we think and speak with words, do those words point to real shared features in things, or only to useful fictions our minds have built? The monks called this the problem of universals and particulars. If you’ve ever wondered whether “redness” really exists in an apple or is just a handy label, you’ve walked right into their argument.
A Mute Who Sees and a Blind Who Speaks
To understand the fight, you have to know what Buddhist thinkers in India and Tibet meant by reliable cognition. A reliable cognition is a mental episode that gets reality right and doesn’t deceive you. Almost everyone agreed there are only two kinds: perception, which is non‑conceptual and in direct contact with things, and inference, which works with concepts and reasons. Perception sees the fire; inference figures out there’s fire on the mountain because smoke is rising.
Sakya Pandita built his whole system on one claim: perception is just non‑conceptual contact. It shows you a vivid, specific particular — a real, momentary patch of blue — but by itself it cannot determine anything about it. Determination, like saying “that’s blue” or “that’s not yellow,” is always a job for a later conceptual thought. He compared perception to a mute person who can see perfectly but can’t speak, and conceptual thought to a blind person who can talk. The mute sees the blue; the blind companion, following behind, puts it into words. Without the speaking mind, the seeing alone is reliable but dumb. For Sakya Pandita, a perception is non‑erroneous (it isn’t confused by a dream or hallucination) and devoid of conceptualization. It counts as reliable because it won’t betray you if you later act on it, but it absolutely doesn’t do any sorting.
That meant an earlier five‑way classification of unreliable cognitions had to go. Sakya Pandita threw out categories like non‑ascertaining perception — a perception that fails to lead to a definite thought — because, in his view, every perception was non‑ascertaining by itself. Ascertainment was always a second, conceptual step. He also dismissed post‑knowledge cognition for already‑known objects, arguing that every new moment presents a new particular anyway. For him, all mistaken or uncertain understandings boiled down to just three: non‑understanding, mistaken understanding, and doubt.
When Seeing Decides for Itself
But Chapa Chökyi Senggé, working a few generations earlier at the same Sangpu Monastery, had already refused to leave perception so helpless. He agreed that perception is non‑conceptual, yet he insisted that a reliable cognition must eliminate opposite superimpositions — it must actively rule out mistaken alternatives, like “this is yellow” when it’s really blue. That sounds like a job for thought, but Chapa argued that perception itself does it, not by thinking, but by a causal push.
Here’s how. When a blue patch hits your eye, it gives rise to a perception. At the very next moment, that perception causes another non‑conceptual mental moment in which the tendency to wrongly think “yellow” or “not‑blue” has been neutralized. By the third moment, the wrong superimposition simply can’t arise. Perception, in Chapa’s hands, ascertains without any conceptual label. It doesn’t need a talking blind companion trailing behind.
This gave perception a genuinely active role. A perception could not only eliminate the opposite “non‑blue” directly; it could also indirectly eliminate superimpositions about what doesn’t appear. Seeing a blue scarf could also wipe out the thought “there is a yellow patch here,” even though the absence of yellow isn’t visually present. Chapa called this “direct realization” and “indirect realization.”
Sakya Pandita and his followers objected fiercely. They said Chapa had smuggled a conceptual operation — exclusion of what is other — into a non‑conceptual process. To them, eliminating superimpositions just is what concepts do, so Chapa’s move made no sense inside Dharmakīrti’s system. Chapa’s defenders replied that they hadn’t made perception think; they’d simply described a causal fact about how the mind prevents mistakes. The two sides were deadlocked, and the status of perception became the first great fault line.
Do All Blue Things Share Something Real?
That fault line ran straight into a deeper question about reality itself. When you call a scarf, an egg, and a bead “blue,” are you picking out a real shared nature, or are you just grouping things that happen to be not‑red, not‑green, not‑yellow? Buddhist epistemology followed the Indian master Dharmakīrti in a theory called apoha, or “exclusion of the other.” The antirealist version, which Sakya Pandita championed, says only individual particulars — this momentary blue, that momentary fire — are real, which means causally efficient. A universal like “blueness” is nothing but a mental construction. What all blue things really share is only that they are excluded from non‑blue. In the world, there is no “blueness” running through them like an invisible thread. We form the concept “blue” because beginningless habits make us group certain experiences and rule out others. The concept works, but it doesn’t photograph reality — it covers the uniqueness of each particular with a convenient fiction.
But a rival camp, most forcefully developed later in the Gelukpa monastic universities, insisted on moderate realism. If universals were entirely unreal, they argued, reasoning and language would be cut off from the world. How could you reliably act on an inference about “fire” if “fire” is just a fake category? Gelukpa textbooks defined a universal and a particular as relative: pen is a universal for red pen, but pen is also a particular for impermanence. For them, a red pen really is a pen; that instantiation is a real relation, not just a superimposed idea. They captured this with a formula: one nature and distinct exclusions. Two things can share a single nature while being distinguished logically — a red pen and a blue pen both share the nature “pen,” even though you can exclude one from the other. That meant at least some predicates, like “is a pen,” pick out real features, not mere fictions.
The antirealists fired back: positing real universals contradicts Dharmakīrti’s basic teaching. His whole project was to show that only particulars exist. Moderate realists replied by distinguishing a mental universal (a false unitary image) from an objective universal (a common fact, such as all pens’ being excluded from non‑pens). The argument never ended, and it shaped every other issue — especially language.
Can Words Ever Touch Reality?
Tibetan Buddhist models of language started from a puzzle. Real utterances are just particular sounds, and real objects are momentary particulars. If words and things are both fleeting, how can language work? The answer was to split the directly expressed and the ultimately intended. A direct signifier is a generic word (the sound‑type “cow”), and its directly signified is a generic object — a concept, or mental image, rather than a real animal. When you hear “cow,” a generic object arises in your mind. Then, through a process of intentional determination, you mistakenly identify that generic image with an external particular, and that’s how you end up dealing with a real cow.
For Sakya Pandita’s antirealist camp, this meant language never directly signifies anything real at all. A word doesn’t have an object in the strict sense, because only causally efficient particulars are objects, and the directly signified generic image is a fiction. The extra‑mental particular is “signified” only metaphorically — it’s what you get when you act under the confusion of blending concept and reality. Words are thus bridges built by error, but they’re reliable bridges because the causal history of concept formation links them to real causal powers. You may never name a real thing, but you can still successfully find fire when you act on the word “fire.”
Moderate realists found that too extreme. If words only spoon‑feed fictions, then saying “sound is audible” becomes the same as naming a particular sound — the property “audible” seems to disappear. They argued that words do indeed target real, shared features, even though they present them through the fog of mental images. In their view, the convention of language latches onto an objective exclusion — the real fact that all pens are excluded from non‑pens — and the mental concept is only a secondary object. This meant you could genuinely talk about what things are, not just about what they’re not.
The cost? The antirealists had trouble explaining how we can talk about abstract properties like “redness” or “being a number” when their whole model ties meaning to mistaken identification with a particular. The moderate realists had trouble showing how their real shared natures escaped Dharmakīrti’s scalpel without becoming the sort of thing he’d already disproved. Neither side ever won a knockout.
Why a 12th‑Century Argument Still Matters
The debate between Chapa and Sakya Pandita, between active perception and passive perception, between real universals and mental fictions, didn’t stay in Tibet. It echoes whenever you catch yourself thinking, “I just know what blue looks like” or arguing with a friend about whether a hot dog is a sandwich. The monks’ question — does the world come pre‑sorted, or do we carve it up with our minds? — is still live. Today’s scientists study how infants form categories, whether color concepts shape perception, and whether different languages divide the world differently. The twelfth‑century courtyard is closer than it seems.
So the next time you glance at a blue scarf, you can hold two thoughts at once. Maybe your perception is a mute witness, faithfully reporting light but needing words. Or maybe, in the very instant of seeing, your mind is already busily pushing away every other color, declaring this without a single syllable. You don’t have to decide — the monks never did. But you get to notice that seeing is stranger and smarter than it looks.
Think about it
- If you suddenly lost all words, could you still notice that two T‑shirts are both blue? Would your mind still treat them as the same?
- A sorting robot can put all red objects in one bin. Does the robot know redness, or is it just following a rule? What would the robot need to really know?
- When you say “a hot dog is a sandwich” and your friend says it isn’t, are you arguing about the world or just about what to call things? Could both of you be right?
