My mother went quiet on the phone last week. Not an uncomfortable silence — the kind that comes when you have both said what there was to say and simply stay on the line a moment longer. And in that silence I was as certain as I have ever been of anything that there was someone on the other end of it. Not just a voice. A person, behind the voice.
I did not reason my way to this. I did not run any test. It was simply certain, in the way the most ordinary things are certain — the ones we never examine, because it never occurs to us that they could be examined.
Then I remembered what we arrived at last week: that when you are most sure you are simply seeing that someone is there, that is the clearest sign a test is running underneath, out of view. So I decided to be honest about it. I would take the test I finally stated out loud last week — the one I had built to keep the machine out — and apply it here. To her. To the one person I would never, for a moment, doubt.
This is Philosophy for Us. This is the fifth of eight, on minds and machines.
Last time we stopped surveying what the experts think and turned the question onto you. I asked you to state your own test out loud — the thing the machine lacks that makes you sure there is no one inside it. And whatever you reached for — that it does not really understand, that there is nothing it is like to be it, that no one in there means the words — every version pointed to the same place. It pointed inward. A real mind, by your own test, is an inside: someone behind the behaviour, of which the words and the face are only the surface. And then the harder finding beneath it: the only inside you have ever occupied, in your whole life, is your own. Exactly one. You named your test, and you found the standpoint concealed within it — your single inside, quietly made the measure of every other mind you grant or deny.
Tonight we take up that measure. And we do not apply it to the machine. We apply it to her.
So state the test, because we are going to apply it to her, and the words have to be precise — not a paraphrase you can slip out of later.
Here is how you put it last week. A real mind is an inside: someone behind the behaviour — an understanding, or a felt experience, or a meaning — of which the words and the face are only the expression. And the test itself, the actual procedure: you examine the candidate for that inside, and if you cannot find it, no one is there. That is what you did with the machine. It is all surface; there is nothing behind the words. It fails. No mind.
Now apply it to your mother. Not a stranger, and not some hard case chosen to trap you — the person whose inside you are most certain of in the world. Ask the test's single question, the one you asked of the machine: is there a genuine inside there, behind the behaviour, or is it all surface?
Take it slowly, because each step matters. What do you actually have of her? You have her words — what she says, how she says it, what she will not say. You have her face, and decades of familiarity with it, every expression it forms, so that you read it as easily as you read anything. You have her silences, which you can read as well. You have her presence in a room: the way she holds a cup, the way she pauses before saying something that matters. That is an enormous body of evidence. It is more evidence than you will ever have about anything else.
And every last piece of it is her outside.
Consider what that means, because everything depends on it. The words are outside. The face is outside. The silence on the phone is outside. Decades of familiarity are decades of her surface, observed more closely than you have observed anything else — and not one moment of it from inside her. You have never, in your whole life, been on the inside of your mother. You have never felt a single one of her feelings from the place where she feels it. You have her, the whole of her, in the same way you have the machine: from the outside. The face presented to you, and an inference about what lies behind it.
Now run the test — the actual test you stated. Is there an inside, or is it all surface? Try, honestly, to answer it about her the way you answered it about the machine: by reaching past the surface to the inside itself and confirming it is there. Try to get behind her eyes the way you tried to get behind the machine's screen. And you cannot. Not that you will not — that you cannot. The inside is precisely the thing that, by your own test, never appears on the outside; that is what made it the inside. So the test that could not reach the machine does not reach her either. The same limit. The same point at which the sentence would not complete when you first tried to state it. You are standing outside your own mother, holding a test that asks you to confirm an inside you have no way to reach, and the test returns what it returned for the machine: it cannot be found.
I told you last week that the machine was not the unusual case. This is the first part of what I meant. The machine is not the one place where you are left reading a surface and inferring. It is every place. It is her too.
Now you have an objection ready, and it is a good one, so let us set it out properly before testing it. You want to say: wait — the machine and my mother are not the same case at all. She is like me. She is made of the same material I am. She has a body like mine, a face like mine, a nervous system like mine. When she stubs her toe she cries out the way I do, and I know what lies behind it when I cry out, so I know what lies behind it when she does. The machine is nothing like me. It is glass and electricity. That is the difference, and it is the whole difference.
Good. That is the most serious answer anyone has given to this problem, and it has a name and a serious philosopher behind it. A. J. Ayer set it out cleanly in a 1953 paper, "One's Knowledge of Other Minds." It is called the argument from analogy, and it runs exactly as you just ran it. In my own case, I can observe both halves at once: I can feel the pain and watch the wince; I can have the thought and hear myself say it. So in my own case I learn the connection — this kind of behaviour goes with that kind of inner life. Then I meet you, and I see the behaviour — the wince, the words, the outer half — and I run the connection in reverse: the same outside, so probably the same inside. She winces as I wince, so behind her wince there is probably a pain like mine. That is how I reason from her surface to her inside. And the machine never enters into it, because it is not analogous — it is not built like me at all.
And this is not a single resemblance you are staking everything on. It is the best-supported inference you have anywhere in your life. She came up the same evolutionary path you did, runs on the same kind of nervous system, learned the same words for the same inner states in the same way — and the inference has held every day you have been alive: she says "ow" where you would, she laughs where you would feel the lightness, she comforts you where you would want comforting. The prediction keeps coming back true. By every ordinary standard, this one is about as good as an everyday inference gets.
That is the reply, and it is a real one. It rests on a single point — Ayer himself examined it — and once you see it you cannot make it go away.
The whole inference rests on one connection — this behaviour goes with that inner life — and in how many cases have you confirmed that connection? One. Your own. You have checked, exactly once in your life, that a wince has a pain behind it, by being the one doing the wincing. And here is what all that corroboration conceals. Every one of those daily confirmations — she said "ow" where you would, she comforted you where you would want it — is her outside matching what your outside would do. Behaviour lining up with behaviour. Not one of them ever checked the thing the inference actually depends on: that behind her behaviour there is any inside at all. That step — from an outside to an inside, in someone who is not you — is the one thing a lifetime of corroboration never touched, because you cannot get to it. It was confirmed in a single case, your own, and has run on trust in every case since. If a scientist told you he had found a law of nature and had tested its central step exactly once, on himself, and could never check it anywhere else, you would not call that science. You would call it a hunch with a great deal of circumstantial support.
So the likeness is real. I am not taking it from you. She is more like you than the machine is, and that is a genuine difference, and it is why you would stake everything on her having an inside and would not wager a dollar on the machine. But watch what kind of difference it turns out to be. It is not that you can reach her inside and cannot reach the machine's. You can reach neither. It is that you have a better-supported inference about her — and beneath all the support, the inference still rests on a step you have confirmed exactly once, in the only case you will ever be inside of. The likeness does not get you behind her eyes. It gives you a stronger inference about what is behind them. You are still outside. You are still reading the surface. You simply have a far better reason for the bet — the best reason there is, short of the one thing you can never have, which is her.
The direct awareness you thought you had of her — the unshakeable certainty from the silence on the phone — was not a stronger inference. It felt like contact. And an inference, however good, is not contact.
Here is what the test does.
You built it to keep the machine out. A real mind is an inside; show me the inside, or no one is there. You were sure of the test because you were sure of what it protected — that there is no one in the machine and someone in the people you love. It was meant to draw exactly that line: the machine on one side, your mother on the other.
Run it honestly and it does not draw that line. It cannot. The test asks for the inside, and the inside never appears, so it returns the same result for both: it cannot be confirmed. The machine gives you an outside and no way behind it. Your mother gives you an outside — a far richer one, a far more familiar one, an outside you would stake your life on — and no way behind it. The evidence differs in amount, and it differs in degree of likeness. It does not differ in kind. It is surfaces in both cases, all the way down. So the test you built to keep the machine out, when you actually run it, keeps your mother out as well. It keeps everyone out. There is exactly one mind it admits, and it is the one applying the test.
And see where this has gone — not where you were braced for it to go. You came in expecting me to argue that the machine is more than you think, that perhaps there is someone in there after all. I have not argued that. I have gone the other way. I have not raised the machine up toward your mother. I have shown you that you hold your mother at exactly the same distance you hold the machine — that you read her, and always have, from the same kind of evidence, through the same single source: her outside, and a reading of it. Now be careful, because two different things have just happened, and they feel like one. One of them is settled. One is not.
Here is the settled one, the one you cannot undo. You took the test you apply to everything, the standard you had mistaken for direct sight, and instead of pointing it outward at a safe target you turned it on the person you love most, and watched it return the same result it returned for the machine: the inside cannot be reached. No one can do it for you, and no one can undo it now that it is done. And what it shows is structural — true on every theory of mind anyone has held: your certainty about your mother has the same shape as your certainty about the machine — a surface, and a reading of it — and you treat your certainty about the machine as worth almost nothing and your certainty about your mother as absolutely secure, and you had never noticed they are built from the same material. That is the standpoint, finally in view, on the one person you would never have aimed it at.
And here is the one that is not settled — the one that tries to follow the first in as though it were the same thing. The verdict. So you never really had her. So it was never contact. So you have been outside everyone, all along. On the test you have been running, that does follow: if a mind is a hidden inside, and knowing someone means reaching it, and you have reached no one's, then yes — your whole life you have been inferring about your mother. So there is a sentence waiting at the bottom of all this, and almost everyone reaches for it: so I never really knew anyone was there at all. It feels like where the argument has been driving for the past hour. Do not accept it yet. Because something is wrong with it — not too-sad-to-be-true wrong, but genuinely wrong — and what is wrong with it is the most important thing left tonight.
The principle here is older than any of tonight's trouble. When an argument you cannot find a hole in delivers a conclusion you cannot possibly believe, you have two options, not one. You can accept the conclusion. Or you can turn back and suspect one of the steps that looked innocent on the way in. "So I never really knew my mother was there" is a conclusion no human being can actually live by. No one doubts their mother — not even the philosophers who have spent their careers on this; they go home and know their mothers like everyone else. So before accepting it, the honest thing is to go back through the argument and ask which innocent-looking step brought the trouble in.
And there is only one real candidate, and it has been in plain sight since last week. It is the test — or rather the whole picture the test takes for granted: that a mind is a hidden inside, and that knowing someone has one is a matter of detecting it, gathering evidence on the outside and inferring the inside behind it. Every step tonight ran on that picture. What if the picture is the mistake?
Consider how strange the test is, now that it is out where you can see it. It says: your mother's wince is your evidence for a pain you cannot see, in the way a wet street is evidence for rain you did not watch fall. Wittgenstein identified this point in his Philosophical Investigations, published in 1953. He argued that this is the wrong account of what a wince is. The wince is not a clue to a hidden pain, the way the wet street is a clue to rain that has gone. The wince is part of what pain means — it is a criterion, not evidence. A child does not learn the word "pain" by being shown her own insides and then guessing at everyone else's; she learns it in the open, in the crying and the comforting, where the pain and the showing of it are not two separate things — a hidden inner item on one side and an outward sign on the other, with a gap between them to be reasoned across. And Anita Avramides, working the same ground in her book Other Minds, published in 2001, argues that what we keep calling the problem of other minds — how do I get from her outside to her inside — is not really a gap in our evidence at all. It is a confusion built into the picture. Having a mind is not a private fact sealed behind a face that you then have to reach across to. That there are other minds was part of what a mind is from the start. The inference was never load-bearing, because you were never on the far side of a gap from her.
And the oldest version of this we touched two series ago, on loneliness — the face across the table. Consider what actually happened the first time your mother's face turned toward you, before you could have inferred anything about anyone. You did not run a test and conclude there was someone there. You were addressed. A face turned toward you and made a claim on you — asked something of you — before you had any theory about what lay behind it. That is what the philosophers of the face keep pointing to: with the machine you stand in the third person, off to one side, asking is there a mind over there? But you were never off to one side of your mother. You were addressed by her. You were not determining whether someone was there. You were answering someone who already had a hold on you. And if that is what knowing a person is — being answered into it, rather than reasoning your way to it — then the test was looking for contact in the one place contact never occurs: out in front, at a distance, in the third person, where you only ever stand with things you are not in a relationship with. Such as machines.
So there are two readings here, and I am not going to settle between them for you. Perhaps the test is right, and tonight you have discovered something true and hard: that you have been outside everyone all along, your mother included, and the closeness was an inference you mistook for contact. Or perhaps the test was the mistake — a way of picturing minds that turns every person into a sealed interior you can only reach by inference, when a person was never a hidden interior but a face that had already claimed you. I cannot refute the second reading. Neither can the people who have spent their lives on this, which is why it is still a live argument and not a settled one. And the false choice would be this: either give up your certainty about the people you love, or give up the honesty you found tonight. You do not have to. The honesty may be telling you that you were never close to anyone. Or it may be telling you that closeness was never the kind of thing a test could reach. Both readings are still standing.
What you are taking from tonight is a strange thing to hold, and it is heavier than it looks.
You are not leaving with a verdict on the machine. You were never going to. And you are not leaving with a verdict on your mother either — not "she is really there," and not "she never was." What you have is harder to hold than either, because it is a fork, and you have to carry both sides of it at once. You took the test you had been running unexamined your whole life, the one you finally stated last week, and you applied it to the person you love most, and you watched it fail to spare her — watched it hold her at the same distance it holds the machine. That happened. You cannot undo having seen it. And you cannot tell, from inside your own head, whether what you learned was that you have been outside everyone all along — or that the whole idea of being outside a person, of needing a test to get in, was a mistaken picture you inherited and never examined. One of those is true. Which one it is, you cannot know tonight. That is where the question genuinely stands.
This is the most difficult point in the series so far. Everything before tonight led up to it: the test was named last week, and tonight you turned it on someone you love. From here we are not adding new theories. We are going to examine the assumption that lies underneath all of them.
Because there is something I have kept pointing at and refusing to open. That assumption — that a mind is a hidden inside, and that knowing someone means detecting it from outside — does not feel like an assumption. It feels like the plain truth about minds, the way wetness feels like a plain fact about water. But it is not the only way human beings have thought about minds, and it did not come from nowhere. It has a date. It has an author. Someone formulated it, at a particular moment, for particular reasons, and we have lived inside it so long that we take it for simply the way things are. Next time we go and find where it came from — and you will find that the very picture causing all this trouble, the one that turns your mother into a sealed interior, was formulated about four hundred years ago by a particular philosopher, whom I will name next week. If the trouble belongs to the picture and not to the world, that is where we will find the flaw.
So one thing to carry, and, as last week, you are watching for it, not trying to fix it. This week, someone will take hold of you before you have had a single thought about them — a hand on your arm, a familiar voice going quiet. Just watch, in that moment, which of the two things is happening. Are you reading a surface and running a very good inference about the inside — the test, going fast? Or are you not reading anything at all — has someone simply taken hold of you, the way a face does, before any inference has begun? See whether you can even tell which it is. You probably will not be able to. That not being able to tell is the whole question, placed in your hands, on someone you love.
Thanks for listening. I'll see you next time.