Other Minds

The oldest sealed box there is: every mind you have ever loved or trusted, known only from the outside. Philosophy has held this problem since 1641 — and the way humans actually solved it in practice is the way this house thinks about opaque machines.

Here is the situation, stated without decoration. You have never seen a thought. Not one, not ever — except your own. Every other mind you have ever known reached you through its outside: a face, a voice, a kept promise, a body that flinched when yours would have. The inner light you are certain sits behind your mother’s eyes is, strictly speaking, an inference. You have exactly one sample of “mind from the inside,” and it is you.

Descartes saw the shape of this from his window in 1641: looking down at the street, he noted that what he actually saw was hats and coats moving — and that everything beyond that was judgment. Philosophy has carried the problem ever since under a plain name: the problem of other minds. It is not a puzzle about exotic cases. It is a fact about the ordinary ones. The walls between minds are not a metaphor; they are the default condition of every relationship you have.

Mill’s wager

The classic answer came from John Stuart Mill in 1865: the argument from analogy. I have a body like yours; when my body is in this state, I feel pain; your body is in that state; therefore, probably, so do you. It is honest about what it is — an inference, a bet placed on resemblance — and philosophers have spent a century and a half stress-testing it, because a generalization from a single case (your own) is about as thin as evidence gets. The modern literature in the roots below is, in large part, the continuing audit of that bet: what exactly the asymmetry is between knowing your own mind and knowing anyone else’s, and whether the gap is one of degree or of kind.

Wittgenstein’s turn

Wittgenstein moved the problem instead of solving it. His remark — “My attitude towards him is an attitude towards a soul. I am not of the opinion that he has a soul” — relocates the whole question: your certainty that other minds exist was never a conclusion you reasoned your way into, so no skeptical argument can reason you out of it. You do not infer that your friend is conscious the way you infer rain from wet pavement. You treat them as minded — the treating comes first, underneath argument, where the skeptic can’t reach it and the proof-builder can’t either.

Notice what that concedes, though, because it is the live wire in this whole room: the walls never came down. Wittgenstein did not show you the inside of another mind; he showed that human life runs — has always run — without that access.

Nearly four hundred years of philosophy, and the lid never opened once.

Why this room is in the temple

Because the loudest sentence in the AI discourse — we cannot see inside it, therefore fear it — quietly assumes that seeing inside is how trust between minds has ever worked. This is the room that says: that assumption deserves scrutiny. Nobody has ever read a loved one’s weights. Trust between opaque minds has historically been built behaviorally — on track record, on conduct over time — and that remains the primary axis even when other access is available. Field Note 004 makes that argument in full; this page is its load-bearing wall. The analogy is not perfect: AI systems operate at a scale and speed that changes the stakes, and engineers can inspect model weights in ways we cannot inspect each other. But the behavioral dimension does not disappear because those other dimensions exist. The machine walked into a world already structured around sealed boxes — and the question of how to extend trust across opacity is one that history has had to answer before.