Shreshth Rajan, May 2024. Written for Dr. Parimal Patil.
Dignāga's conclusion, stated without qualification, is that you have never perceived anything outside your own mind. In the Ālambanaparīkṣā the object of any perception is a form belonging to consciousness itself.
It's a strange conclusion, and Dignāga argues for it carefully. I spent a semester on it, and I think it fails — on a premise he never states, not on any of the steps he does.
He asks two things of whatever you'd call the object of a perception. It has to cause the perception, and it has to be what shows up in the perception. Every candidate manages one of them.
Atoms cause. They're what acts on the eye, and on the Abhidharma picture they're all that's ultimately there. But nothing in your experience has the character of an atom. What you get is a shape with a rim and a handle.
A cup shows up. But a cup, for Dignāga, isn't an extra thing sitting on top of its parts; it's a unity your mind puts on them. A mind-imposed unity does no causal work, because whatever work got done was done by the parts. So the thing appearing in your perception is the one thing that couldn't have caused it.
Neither candidate does both jobs. The cause never appears, and what appears never caused anything, so Dignāga concludes the object was never outside you to begin with.
The case against the cup needs a premise Dignāga never states: that a real cause has to be simple and self-standing, so anything assembled is derivative and inert.
He can't have that premise. Pratītyasamutpāda, dependent origination, is the doctrine the whole tradition runs on, and it says nothing is self-standing. Atoms arise from conditions exactly the way cups do. If dependence is what disqualifies a thing from causing, he hasn't refuted one horn of his dilemma. He's refuted both, and nothing is left to cause anything.
Drop the premise and the cup is an unremarkable cause. It has a shape that holds liquid, a thermal mass your hand registers before you've looked at it, a wall thickness that decides whether it survives being set down hard. Those came from clay and a wheel and a kiln and somebody's judgment about what a cup is for, and they're what the light is bouncing off on its way to you. Nothing in that chain goes around the cup.
The reply I'd expect is about mediation. Grant that the cup causes something. What you perceive is still shaped by your concept of a cup, so two people with different concepts perceive differently, and neither is tracking anything.
But the concept isn't free either. You got "cup" from cups: from gripping them, from the ones that slipped, from watching other people drink. It's as dependently arisen as the object is, and it arose out of the very objects it later helps you sort. That isn't a loop sealing you off from the world. It's how the world got into you.
It also explains the agreement Dignāga says conventional objects can't produce. Two people converge on cups because they were trained by the same kind of thing, not because they signed something.
The objection I can't fully answer is that I've changed the question. Dignāga wants to know what caused this perception, in this instant. A momentary particular can do that. "Cup" is a general character, and a general character isn't a particular, so it isn't the thing acting right now. Showing that the cup is dependently arisen explains why there's a cup to see. It doesn't put the cup in the causal slot.
My answer is that the demand for a bare momentary particular is itself what's in dispute. If causation is a web of conditions rather than a sequence of instantaneous pushes, then "what caused this perception" has no single right answer, and a stable pattern in that web is as good a candidate as any slice through it. You can't use dependence to dissolve the cup and then help yourself to discrete momentary causes.
Whether Dignāga would count that as an answer or as refusing to play, I don't know. But if dependent origination goes all the way down, the dilemma was badly posed, and the cup on the table is doing what it looks like it's doing.