For two weeks now I have been handing you other people's arguments and asking you to build them at full strength.
Two weeks ago it was the case that the machine cannot think — Searle, the locked room, fluent Chinese produced by a person who understands none of it. We built it until the question seemed closed. Then last week the opposite case — that a mind is a set of functions a thing performs, not the material it is made of, so a machine performing those functions could have one. We built that just as carefully. And that is where the matter was left: two finished arguments pointing in opposite directions, with no honest way to settle on either one. You can no longer say obviously not. You cannot say obviously it could either. Both arguments stand intact, and you are left unable to choose between them.
The natural thing to want now is a third argument — a better theory than Searle's, a better one than the functionalist's, something that finally settles the matter. No such argument is coming. Not because the theories run out, but because the deadlock was never going to be broken out there, in the machine. The reason for the deadlock has been on your own side of the question the whole time, and we have not once set it down and looked at it.
This is Philosophy for Us. This is the fourth of eight, on minds and machines.
Last time we built the case that the machine could think after all — not a weak version of it, the difficult one, the argument that is roughly sixty years old. A mind, on this view, is not a substance but a set of functions: whatever performs the function is the mind, neurons or silicon alike, in the way that what makes something a carburettor is that it mixes fuel and air, not the metal it is cast from. "It is only silicon" turned out to be no better an argument than "it is only meat." And then we reached the difficulty even that case cannot meet: you could run the entire function, every input and every output, on a crowd of a billion people and their telephones, and still ask whether anyone, anywhere in that crowd, was actually experiencing a thing. So we declared no winner. Two cases, each as strong as anyone has made it, and a deadlock.
Tonight we stop. We stop surveying what other people think a mind is — not because the experts have nothing more to offer, but because the thing we most need to examine now is the one thing no survey of experts could show you: the test you brought to this question yourself, before any of them said a word. And we turn the examination around. For the rest of the hour, what is in front of us is not the machine, and not Searle, and not the functionalist. It is you — and one particular thing you have been doing, silently and fluently, since before this series began, and have never put into words. Tonight you are going to state it.
Let me show you why a third theory would not help, so that you can be satisfied turning inward is not an evasion.
Look back at the argument you have just watched. Searle says the machine cannot have a mind because it only ever manipulates symbols by their shape and never understands them. The functionalist says it could, because if it performs the whole function a mind performs, then understanding simply is that function being performed. Notice what they are actually disputing. They are not disputing what the machine does. They agree entirely about that — it takes in words, runs its process, puts out words. They are disputing whether that is enough: whether doing all of it amounts to the real thing, or falls short of it.
And to dispute whether something is enough, both sides need an account of what would be enough — a standard the thing has to meet. Searle has a standard: real understanding, the genuine grasp of meaning, and he says the machine falls short of it. The functionalist has a standard: the full functional role, and he says the machine meets it. They disagree about where the standard lies and what it consists in. But notice that both of them are applying a standard. Both hold that the way you settle whether there is a mind here is to specify what a mind must have, and then check whether this candidate has it. A test. A requirement. Pass or fail.
That is the form of every move in the whole debate. It is the form of your own first reaction, too — obviously not is a failing grade, who's to say is an incomplete. Every position in the dispute, expert and amateur alike, runs the same procedure: name the standard, apply it, read off the verdict. The arguments are all about which standard. No one is asking whether applying a standard is the right thing to be doing in the first place.
So the real reason a third theory would not release us tonight is not that the experts have nothing left to say. Suppose a better account did arrive, sharper than Searle's or the functionalist's, good enough to win the argument. You would have a new verdict on the machine. And you would be standing in the very place you are standing now, in the one respect that turns out to matter: you would have a standard, you would have applied it, you would have read off an answer — and you would still never have turned to examine the standard you brought to this question yourself, the one that has been running under your "obviously not" since before any philosopher said a word. A winning theory would settle the machine. It would not show you the thing you have never examined. No one can hand you that. You have to turn and look at it yourself.
There is exactly one thing in this entire question that has never been examined, and it is the least likely candidate, because it is the most familiar. It is not Searle's standard and not the functionalist's. It is yours — the one you arrived with, two weeks ago, before you had heard a single argument, the standard that made obviously not feel like something you could simply see. We have spent three episodes examining the experts' tests in great detail. The test you brought with you, we have left unexamined. Tonight, before anything else, you are going to take it out and look at it. And the first surprise will be that you have one at all.
So take it out. I will set it up as I set up the sentence on the first night — but this time it leads somewhere new.
Here is the standard, and this time it is not the sentence I am after — it is the requirement itself. Finish this: the one thing the machine lacks — the thing that, if it had it, would stop me being sure there is nobody there — is ______. Not the cleverest answer. The one you would actually stake the obviously not on. State it, aloud if you can, under your breath if you cannot.
I will tell you where most people arrive, and yours is probably one of these, almost word for word — and that sameness matters.
The first: it does not really understand; it moves words around without grasping what any of them mean. Good — that is a genuine standard, and the commonest one. The second: there is nothing it is like to be it; behind all the talk, there is no experience going on, no one inside having one. Also genuine, also common. The third: it does not actually mean anything by what it says — there is no inner source the words come from, no one who meant them. Three standards. Perhaps yours was slightly different — no genuine feeling, no real wanting, no soul, if you are being old-fashioned about it — but I would wager it belongs to the same family, and I will show you shortly why they are all one family.
Now look at what you have just done. Not at the answer you gave — at the act. You held the machine up against a requirement. You had something in mind — understanding, or inner experience, or genuine meaning — and you checked the machine for it, and the machine came up short, and you read off the verdict: nobody there. That is not what seeing is. When you see that the wall is white, you do not hold the wall up against a standard for whiteness and check whether it qualifies. You simply see white. But holding a thing up against a standard is just what you did with the machine. You ran it through a test. There was a step in it — does it have the thing? no? then no mind — a step you perform so quickly and so often that it had disappeared entirely into the feeling of sight.
That is the thing you have never stated. Not the answer to the test — you have given that a hundred times, it does not really understand, easily. What you have never stated, never even noticed you were doing, is that there is a test. That obviously not is the result of a measurement you are running, with a standard you chose, on evidence you are reading — and not something you are simply perceiving. Three episodes ago you would have told me, in complete honesty, that you could just see there was no one in the machine. You cannot see that. No one can see that. What you can do — what you just did — is check a thing you have against a thing it lacks, and call the result a perception. We caught the edge of this on the first night, when the sentence would not finish. Tonight you have the whole of it: not a missing word, but an entire procedure you have been running without noticing. You have a test. It was always a test. Now we can finally ask what it consists of.
So let us set the three standards side by side and ask what they have in common — because they have something in common so obvious it goes unseen.
Real understanding. Where would that be, if the machine had it? Not in the words on the screen — those it produces perfectly well. Somewhere behind the words: an inner grasp, an understanding, that the words come from. Inner experience, something it is like to be it. The same — not the behaviour, not the talk, but a felt state underneath the behaviour, in some interior the behaviour is the outward sign of. Genuine meaning, someone who meant it. Again: not the sentence, but a someone behind the sentence, doing the meaning. Every standard you might have reached for points to the same place. They all point inward. They all say: the mind is not the part you can see and hear and read off from outside — the mind is something behind that, an inside, of which all the visible activity is only the outward expression.
That is your test, stated all the way down. Set aside whether you said "understanding" or "experience" or "meaning"; underneath, every version asks one question — is there an inside in there? Is there a someone, a felt inner life, a private source the behaviour comes from — or is it all outward, all output, with nothing behind it? That is the standard. Not any particular inner item. The inside as such.
Now you can see why the sentence would not finish on the first night. Remember how it slipped away, every time, the moment you tried to say how you would check? That was not slowness on your part. That was the test showing you what it is made of. You built a standard — has an inside — out of the one thing that, by its own definition, never appears on the outside. The inside is precisely the part that is not the outward behaviour; that is what makes it the inside. So the test you are running asks you to detect, from the outside, something whose whole nature is that it does not appear on the outside. Of course no means of checking arrived. You were trying to find a direct view of the one thing your own test defines as the thing no direct view can reach.
And before you conclude that this is a problem with the machine — that the machine is the very case of a convincing outside and a hollow inside — hold one thing here, because next week begins at this point. That whole picture your test takes for granted: a mind as a private inner room, the behaviour as a face on the front of it, the real thing always one layer back. You did not invent that picture. You inherited it, fully formed, and it feels like the plain truth about minds. It is not the only picture anyone has ever held, and it came from somewhere in particular, and a few episodes from now we are going to find out exactly where — because it turns out to be causing much of this trouble. Tonight I only need you to see that it is a picture you are holding — not the world itself, a picture — because of what happens the moment you ask the one question the picture has kept you from asking. Which is this. The mind is an inside — how many insides have you ever been inside of?
So: the mind is an inside — that is your test, the one you have run for years without knowing it. A real mind is a someone behind the behaviour, and the behaviour is only the surface they present. Keep that in view. Now the question the picture has kept you from, the one the whole night turns on.
How many minds have you ever known from the inside?
Count honestly. Not how many minds you are sure are out there — billions, plainly. How many you have ever been inside of: known from within rather than observed from without. The answer, when you actually look, is one. Yours. Exactly one mind in the whole history of the universe has ever shown you its inside, and it is the one doing the counting. Every other mind you have ever met — everyone you have loved, every stranger, the dog at your feet, every human being who has ever lived — you have met from the outside. The words, the face, the voice, the behaviour. The surface. You have never once been on the inside of anyone else.
Now set that beside your test, and the whole thing falls into place. Your test says a real mind is an inside. And the only inside you have ever had access to is your own. So when you run the test — does this thing have a genuine inside, or is it all surface? — what are you actually doing? You are taking the one inside you happen to know, from the single standpoint you happen to occupy, and using it as the master example — the one sample of what an inside is like, against which everything else is measured. That is the standard you brought with you. Not "does it have a mind," measured against minds in general — you have never met minds in general. It is "is there, in there, something like the one thing here behind my own eyes" — measured against the only example you were ever given.
That is the thing you would never have put into words. Not "it does not understand." Underneath that: I am the hidden standard. The one mind I have from the inside has been quietly appointed judge of which other things count as minds at all — and I never noticed the appointment, because from in here it does not feel like a standard. It feels like simply knowing what a mind is. Of course I know what a mind is. I am in one.
Look at where that leaves the machine — and then at how it does not leave only the machine there. When you said the machine has no inside, it is all surface, the evidence you had was its outside: its words, its behaviour, the surface it shows you. And I have just made you grant that the outside is the only kind of evidence you have ever had about anything that is not you. The machine is not a special hard case, the one place you are stuck with the surface. It is the ordinary case. It is every case but one. The machine gives you an outside and asks you to rule on the inside — and so does every person who has ever come towards you. The same kind of evidence — the surface, the words, the behaviour — for the machine and for the people in your life alike; and you have been reading every one of them from the single standpoint you have. What that does to the confident yes you have always given the people — that is next week's question.
I am going to stop there, just before the next step. Because tonight you have already done the hard thing: you caught the test. You named it, you saw what it consists of, and you found the standpoint hidden inside it — your own single inside, quietly doing the judging for all the rest. That is the catch. What it costs you comes next.
What you hold in your hands tonight is heavier than the first night's, and heavy in a different way.
The first night gave you only a glimpse — that there might be a test at work where you had assumed there was only perception. That was the gentle version. Tonight the test is fully in the open: named, and taken apart. You know what it says: a real mind is an inside, a someone behind the surface — and beneath even that, the standpoint you did not know you occupied: the only inside you have ever known is your own, and you have been using it, quietly, as the standard for every other mind you grant or deny. You hold the whole of it now, and you read it from the single standpoint you occupy. You can set all of that down for the week — but not back to running unnoticed. A test you can see is a test you can question, and you can see this one now, all the way down.
This is where the episodes begin to demand more. The first three weeks were preparation — meet the puzzle, build both sides. From here it stops being about the machine.
Now, next time. We have your standard now: a real mind is an inside, and I read insides from the single standpoint I have. Next time we do not apply it to the machine. We apply it to a person — not a stranger, not a hard case, but someone you would never for a moment doubt has a mind behind the face. And we run your own test on them, the exact one you have just named, and we watch, together, what it does. I am not going to tell you tonight how that goes. I will only say that the people who have thought hardest about this have not found it a comfortable afternoon, and that it is the reason none of this was ever really about machines.
So one thing to carry out the door — a thing to watch for, not to solve. This week you will have a moment — in a doorway, on a telephone call, lying beside someone — when you are simply, wordlessly certain that there is someone there, behind the face in front of you. It will not feel like a conclusion. It will feel like the most obvious fact in the world, the way nobody's home in the machine felt obvious three weeks ago. And just this once, instead of going along with the certainty, catch the test running underneath it. Notice that what you actually have, in that moment, is an outside — a face, a voice, a warmth — and that somewhere too quick to catch, you turned it into an inside and called the whole thing knowing. Do not doubt the person; I am not asking you to. Only catch that you ran a test, and that you ran it from the only standpoint you have.
Thanks for listening. I'll see you next time.