If you tried, at some point this week, to test the ground you had reached — I suggested it at the end of last week — you may have done it almost without deciding to.
You had settled on something new as the most basic level of yourself — the point below which there is nothing further. Not a soul, not an owner, but the parts: the body, the feelings, the passing moments of seeing and hearing. Those, you had decided, are the real things, the bottom. And then, at some quiet moment — washing up, or driving, or lying awake — the question I had left you with came back on its own. Are the parts really the bottom? Is this any more secure than what it replaced?
Notice what you reached for to answer it. You did the obvious thing, the only honest thing it seemed. You looked harder — you tried to catch the parts directly, to turn your attention on them and see whether they really are as solid as you had taken them to be. Or you tried to reason it out: if the self is only the parts, then the parts must be such-and-such, and you followed the thread. Looking, and reasoning. That is what anyone does when they want to know something. You look, or you work it out.
Stop there. Tonight is about those two things you just did, and about a question that has never once been asked of them. This whole enquiry has trusted that if you only looked hard enough, or reasoned carefully enough, you would reach the bottom and recognise it when you arrived. We have questioned the self. We have questioned the soul, the owner, the parts. We have never questioned the looking. We have never examined the reasoning itself.
This is Philosophy for Us — philosophy for everyone, no degree required.
This is the fifth of eight episodes. Last week the tradition paid the realist what it owed him, using a distinction so plain it sounded at first like a trick: there are two kinds of true. A banknote, at the bottom, is only paper and ink — and it is also, completely, money. It pays the rent, settles debts, is owed to you when you are owed. Both true, at two different levels. The same holds for you. You are real in the everyday sense — you suffer, you remember, you are owed an apology — without there being a separate self standing behind the parts. The realist was not defeated. He was answered. But he was paid with a claim that went unexamined. To dissolve the self into the parts, we treated the parts as what is finally real. We kept a bottom again, one level down.
Tonight we ask how anyone could ever check that. Not whether the parts are the bottom — that question is a fortnight off. The question beneath it: by what means would anyone ever settle what is finally, really there? You would look, or you would reason. Tonight we examine those two directly. And before we begin, two things this is not, so the direction is clear. It is not going to turn out that nothing can be known, that it is all opinion and one may as well choose whatever answer suits — that is a cheap exit, it is wrong, and I will show you why. And it is not mysticism. What is coming is exact and careful: two ways to know, and a test for when one of them works — a test precise enough to use on a hillside too far away to see.
Start with a fair question. How many ways are there to know anything at all?
It sounds as though it should have an enormous answer, or no answer. The Buddhist logicians who worked on this — and they worked on it as hard as almost anyone, for centuries, building a whole science of it — gave a famously small answer. Two. There are two ways, they said, that knowledge ever reaches you. You perceive it, or you infer it. You see it for yourself, or you work it out from something else you already know.
That two-item list came from a logician named Dignāga, and from the man who followed him and developed it, Dharmakīrti. Perception and inference, and nothing else — that was their claim. But notice who is making it. That is the Buddhist side's count. The realists on the other side — last week's realists — did not agree even about this. They counted more ways than two. They added, among others, the word of a reliable witness: most of what you know, someone in a position to know told you, and you never checked it for yourself. So before we have asked a single question about the self, the two sides already disagree about how many ways there are to know at all. Hold on to that. It matters later.
But set the count aside, because here is what both sides do agree on: looking and reasoning are two real ways of knowing. No one at the table disputes those two. And they are exactly the two you reached for this week. When you wanted to check whether the parts are really the bottom, what did you do? You looked harder — that is perception. Or you reasoned it out — that is inference. So those are the two we will put to work tonight. Not because they are the only two anyone has named — that is the very thing the two sides argue about — but because they are the two you trust, the two you actually used, the two that seem as though they could not let you down. If looking cannot settle the self question, and reasoning cannot settle it either, then the plan — settle it by looking harder and thinking longer — is in trouble before it starts.
So tonight is simple to state and hard to accept. We will send each of the two methods out to do the job, and watch what comes back.
But first I have to show you the difficulty they run into — because it is a difficulty you already know, from a setting that has nothing to do with the self. Let me build it on plain ground.
Consider two neighbours and a dented car. One backed out of a driveway; the other was parked on the street; now there is a long crease down the parked car's door, and they disagree, completely and in good faith, about whose fault it is. So they do the sensible thing. They agree not simply to argue, but to let the evidence decide.
And here is where it goes wrong, quietly. One of them says: there is a doorbell camera across the street, it caught the whole thing; let us watch the footage and settle it. The other says: cameras mislead — the angle deceives, you cannot judge distance through that lens — what matters is the paint, the height of the crease, where the metal actually transferred; let us have someone read the marks. Look at what has happened. They both agreed to let the evidence decide. They both meant it. But they do not agree on what counts as evidence. One will trust only the footage. The other will trust only the marks. And the footage says one thing; the marks say another.
So which neighbour is right? This is the part that matters, because everything tonight rests on it. You cannot settle it by looking at more evidence — because what counts as evidence is the very thing they disagree about. Show the footage-truster more footage, and he is convinced. Show the marks-truster more marks, and she is convinced. Each method delivers a result, cleanly. The two results do not match. And there is no third thing — no further evidence, no neutral standard both can appeal to — that stands above both methods and rules on which was the right one to use. To rule on it, that neutral standard would need a method of its own. And then you could ask the same question again about that.
This is not the neighbours being stupid. They are both being careful. It is something about the situation itself. When two people disagree not about the facts but about what counts as establishing the facts, there is no neutral place left to stand. The impartial judge they both want would have to come from outside every method — and there is no outside.
That is the shape of it. Two careful parties. The same world in front of them. Opposite results — not because one is careless, but because the method each trusts already favoured a different answer before either began. And no neutral standard above them.
Now take that same shape back to the question you actually care about. Is there a self? Are the parts the bottom? You were going to settle it by looking, or by reasoning. So let us send each method out, and see whether it comes back with a clean result — or whether it comes back as the neighbours did.
Start with looking, because it seems the one that cannot fail. Reasoning, you might grant, can go wrong — you can drop a step. But looking? Looking simply shows you what is there. If you want to know whether there is a coin under the cushion, you lift the cushion and look. What could be more neutral than that?
So look for the self. You have done this — we did it together on the first night. Sit still, the instruction goes, and watch your own mind. Let the thoughts come and go, and find the one who is doing the watching. The seer behind the seeing. Look straight at it.
And you could not. You went looking for the watcher and found another thought, and then another — things watched, never the watcher. You looked, honestly, and the looking turned up no self. For someone in this tradition, that is what perception reports: there is no separate seer inside. Only the seeing, one moment after another, with no one standing behind it.
Now bring back last week's realist — the one who would not be argued out of existing. Tell him you looked and found no self. He will laugh, and not unkindly. "You looked," he says, "and found no self? I look, and I find one constantly. Right now. There is plainly someone here doing the seeing. The seeing is mine; it is unified; it holds together; one person is having every part of it at once. You say you find only thoughts. I ask: who is finding them? The finding is the self, caught in the act. I do not have to hunt for the seer. The seer is the one doing the hunting. I find it every time I look."
Stop on that, because something odd has happened, and it is easy to pass over. You both looked. You looked at the same place — your own experience, right now — with the same honesty. And you came back with opposite reports. You: only thoughts, no watcher. He: plainly a watcher, in every act of watching. Two careful people, the same world, opposite reports. You have seen this shape before, a minute ago, with the dented car.
So how do you settle it? You want to say: look again, look more carefully. But that is the difficulty. The disagreement was never about how hard you looked. It is about what looking even delivers. And here the two sides divide, cleanly, and each has an account it has worked out and defended for centuries.
The Buddhist logicians say: real perception is the bare flash — the raw hit of colour and sound and sensation, before the mind has done anything to it. That is all that is actually given. The unified self having all of it is not in the flash. It is something the mind adds a half-second later, so quickly that you never catch it being added. So when the realist says "I see the self," they answer: no. You see the flashes, you add the self, and you have mistaken your own addition for something you found.
And the realist has an answer just as worked out. He says: that begs the question. You have quietly decided in advance that only the bare flash counts as really seeing, and that everything else is "adding" — but that is not a discovery, it is a definition, and you chose it because it hands you the answer you already wanted. When I see my friend across the room, I do not see a patch of colour and then work out that it is my friend. I see my friend. The seeing arrives already shaped, already of a thing, of someone. That shaping is part of the perceiving, not something added afterward. You do not get to strip perception down to the bare flash simply because the stripped-down version conveniently lacks the self you had already decided was not there.
Hear what they are actually disagreeing about. Not what is in front of them. What counts as seeing it. One admits only the bare flash as evidence; the other admits the shaped, of-a-thing seeing as evidence. The footage and the marks again. And the bare flash shows no self, while the shaped seeing shows one in every glance. Each method delivers its result cleanly. The results are opposite. And you cannot settle which method is the right one by looking — because looking is the very thing in dispute.
So the first method comes back with nothing. Not because there is nothing to see. Because the two parties cannot agree on what seeing shows, and there is no third kind of looking — no neutral view from nowhere — to settle it over their heads.
All right. Looking is deadlocked. But you have a second method, and it is probably the one you trust more anyway. Reasoning. Inference. You do not have to see the self directly. Perhaps you can prove it, or disprove it — the way you can establish that there is fire on a hillside you cannot see into, by noticing the smoke above it.
That example is not mine. It is the example, the one the Indian logicians used for a thousand years, because it shows the whole shape of inference in a single picture. You are standing in a valley. You cannot see the far hillside — too distant, too hazy. But you see smoke rising from it. And you know, with as much confidence as you know anything, that there is fire up there. You did not see the fire. You inferred it, from the smoke.
Now Dharmakīrti and the tradition he worked in did something worth slowing down for. They asked: when is an inference like that actually sound? When does the smoke really earn you the fire — and when are you fooling yourself? And they wrote down a test. Three marks. Three things that must all hold at once, or the inference is no good. I will give all three in plain words, because they are not hard, and they are worth having.
One. The clue must actually be present, in the case in front of you. There must really be smoke on this hill — not a smudge on your glasses, not a low cloud you mistook for it. The sign is genuinely there.
Two. Everywhere you have independently found the thing you are inferring, you have found the clue with it. Every kitchen, every campfire, every furnace — where there has been fire, there has been smoke. The clue and the target travel together, reliably, across every case you can check.
Three. And everywhere the target is absent, the clue is absent too. A cold lake. A block of ice. No fire — and, sure enough, no smoke. The clue never appears where the target is missing.
Get all three, and the inference holds. Smoke here; smoke-with-fire everywhere you have checked; no smoke wherever there is no fire. Then smoke on the far hill really does deliver you the fire. That is not mysticism. It is careful, honest reasoning — set down a very long time ago, by people most of us were never taught to read.
So let us use it. Let us settle the self by inference. We want to prove — or disprove — that there is a self. What is our smoke? What is the clue that travels with selves and stays away wherever there is none?
And the test cannot get started. It fails at marks two and three, and it is worth seeing exactly where, because this is the whole point of the evening arriving.
Mark two asked: everywhere you have independently found a self, what clue came with it? But to answer that, both sides have to agree on some clear cases of "here is a self" — cases they can point to together and study, the way they can both point at a kitchen and agree there is fire in it. And they have none. The realist points at a person reasoning and says, "there — a self, plainly." The Buddhist looks at the same person and says, "there — a bundle running smoothly, no self." There is no agreed kitchen. There is no shared example of a self for the test to learn from — because whether any given case is a case of a self is exactly what is in dispute.
Mark three is worse. It asked: where the self is absent, the clue is absent too. So point to a clear case of no-self — something both sides agree has no self in it — so we can check what goes missing. The Buddhist says every case is one. The realist says that of any conscious being, none is. No agreement again — and not out at the edges, where you would expect it, but right at the centre.
So the reasoning — and it is rigorous; it works flawlessly on the smoke and the fire — cannot even begin on the self. Not because it is too weak. Because it runs on agreed examples, and on this question there are none. Feed it the self and it asks, quite reasonably, "show me the kitchens, show me the cold lakes," and the two parties hand it different lists. And the lists are the argument.
There is your second method, and it produces as little as the first. Reasoning did not fail because someone made a bad step. It could not take a step at all — because the ground it needs, shared examples, a case both call a self and a case both call selfless, was never there to stand on.
So take stock. You came in tonight holding the one move you were sure could not fail — when in doubt, look harder, or think it through. And both came back from the self question with nothing. Not defeated. Simply with nothing to show. The looking deadlocked, because the two sides do not agree on what looking shows. The reasoning stalled, because it needs shared examples and there are none. And why that matters is not that you are stuck on one stubborn question. It is what it tells you about the methods themselves.
You had been treating looking and reasoning as neutral — as impartial judges, standing outside the dispute between "there is a self" and "there is not," so you could send them in to settle it fairly. They do not stand outside. Each one already favours an answer. The bare-flash way of looking favours no-self before it has seen a thing; the shaped way of looking favours a self. The inference test cannot run without examples, and which examples you feed it is the answer, already decided and slipped in as a premise. Neither method was impartial. And there is no impartial method above them, because whatever a higher method used to reach its ruling would be one more way of knowing, as open to dispute as the ones below it. There is no method that stands free of all method. It too would favour an answer.
What tonight comes down to is sharper than it first sounds. Let me say exactly what it is — and exactly what it is not.
It is not — and this is the exit I said we were not taking — it is not "so nothing can be known, it is all opinion, believe whatever you like." Look again at the neighbours. Within the footage method there are right answers and wrong ones; the footage really does show what it shows. Within the marks method, the same. The results are real. Some arguments genuinely win. What you cannot do is rise to a standpoint above both methods and rule on which one was the right method to use — not when the disagreement runs this deep, down to what even counts as evidence. That is a long way from "anything goes." It is narrower, and much less comfortable: the method you would use to settle the deepest question is not neutral between the answers. It has already taken a side. "Look harder" was never the fair test you took it for.
And it is not mysticism. There was not one mystical sentence tonight. There was a doorbell camera, a dented door, smoke on a hill, and a test with three plain parts. This is the opposite of being told to stop thinking. It is thinking carried far enough to reach its own edge.
Now I have to be honest about one more thing, because if I left it out you would notice it yourself — and you should. This cuts at everyone. It cuts at the tradition we have been following too. The whole no-self picture you have been drawn to these last weeks — the bundle, the parts, the self as a useful name — was reached by looking and by reasoning. By the bare-flash way of looking, and by inference. Which means it came out of one of these methods too. It favours an answer as well. I am not going to do anything with that tonight. I am not going to tell you it sinks the no-self case, and I am not going to tell you it survives it. I am going to leave it standing here, unanswered, because it does not get answered for two more weeks, and pretending I could answer it now would be its own kind of evasion. Just notice that it is there. The method that carried you this far was not neutral either.
So here is what you have now that you did not have an hour ago. You came in with two methods you trusted without a second thought, and a plan to settle everything by using them harder. You leave knowing that the two you trusted most — looking, reasoning — are not the impartial judges you took them for. Neither stands outside the question; each favours an answer before it starts; and there is no third place, no neutral standpoint, to appeal to over their heads. That is not a small thing to carry. Most people go a whole life without once examining the method they are using.
This is the fifth of eight, and it is the narrowest, hardest stretch of the series — one question followed all the way down. Next week the argument moves on. The senior voice of this whole tradition steps forward — the figure these four weeks have been building toward — and he does not offer you a better way to look, or a more neutral standard. He does something you will not expect. He takes the very move you made at the start — the chariot taken to pieces, the self dissolved into its parts — and he turns it around and runs it one level lower. On the parts. On the bottom you have just laid down.
So carry this out with you. This week a question will not sit still, and you will reach for the reflex — "I will just look closer." "I will just reason it through." When you do, stop, and ask what we asked tonight. Not "am I looking hard enough?" The other question. Which method is this, and what does it decide before I have even begun? And could you carry on — not knowing whether there is any neutral place to stand and settle it from at all?
That is where we begin next time. Thanks for listening. I'll see you next time.