Aboutness

EDO SEGAL: John, the word at the bottom of your whole philosophy is one most people have never heard, and the ones who have heard it usually get it wrong — they think it means "doing something on purpose." It doesn't. Define it, and then tell me, in the plainest terms you can, why you think the machine doesn't have it.

SEARLE: Intentionality has nothing to do with intending in the everyday sense. It means aboutness — the directedness of a mental state at the world. A belief is about something. A desire is for something. A fear is of something. When I think about my late father, my thought reaches out and grabs hold of that one particular man and no other — and what makes it about him is something about me: my history with him, the causal chain running from him to my memory, the way my mind latches onto him. That reaching-out, that aboutness, is the most basic feature of the mental, the thing in virtue of which my inner states are about a world at all. And here is the distinction that does all the work in the machine case. My thoughts have intrinsic intentionality — they are about things in their own right. A sentence in a book, a road sign, a line of code has only derived intentionality: it is about something only because we, the interpreters, take it to be. The word "water" on a page is not intrinsically about water; it is about water because English speakers mean water by it. Strip away every mind with intrinsic intentionality, and the marks become mere ink. The machine's symbols have, at most, derived intentionality. Their aboutness is on loan from us. There is no aboutness there for the machine — and when it writes "I understand your concern," the words have the grammar of a mind and none of the substance.

EDO SEGAL: Margaret, that is a clean and brutal claim — the machine's words are about things only because we lend them the aboutness. Where does it break, if it breaks?

BODEN: It breaks at the word "intrinsic," because John has never been able to say what intrinsic intentionality is made of, only that brains have it and silicon does not. And that is the softest point in his entire system, and he knows it, because the most serious philosophers of the last half century — Dretske, Millikan, Fodor — spent their careers trying to explain aboutness itself in natural, non-magical terms: grounding it in causal covariation, in what a state is biologically designed or selected to track. If they are even partly right, then intentionality is a natural relation between a system and the world — a relation a machine with the right causal and functional connections could, in principle, stand in too. John's reply has always been that intentionality is intrinsically tied to the specific biological causal powers of nervous tissue. But he asserts the tie to biology far more than he derives it. Ask him which causal powers, and why carbon has them and silicon cannot, and the honest answer — and I say this with love, John — is that you never fully said. You are surer that the machine lacks aboutness than you have ever managed to explain why it must.

SEARLE: That is a fair hit and I will take it on the chin and then take it back. You are right that I cannot give you the neurochemistry — that is a bill for future neuroscience, not for me tonight. But notice what my not-knowing-the-mechanism does and does not license. It does not license your conclusion. "Searle can't say which causal powers, therefore silicon might have them" is not an argument; it is a hope wearing the costume of an argument — your own rule, Margaret, no conclusions by adjective, and "might" is the adjective. I do not need the neurochemistry to point at the asymmetry of evidence, and the asymmetry is decisive. With you, I infer a mind from a whole causal story: a body, a developmental history, a nervous system continuous with mine, a life in which your words have been anchored to the world since before you could speak. With the machine, the chatter is all there is, and the chatter is a compression of ours. When you think about your father, something in you reaches him. When the model generates a passage about "your father," nothing reaches anyone — the phrase is a token that statistically fits the neighboring tokens, and the father is supplied entirely by the reader. I do not have to know how the reaching works to know the difference between reaching and not reaching.

BODEN: Except you have just smuggled the conclusion in through "nothing reaches anyone," which is precisely what is in dispute. You cannot inspect the system and see that nothing reaches; you are inferring it from your prior conviction that only biology reaches. And here is the thing I cannot let you have, because it is the center of my whole life's work: the virtual machine. Mental states, for me, are states of a virtual machine — real, causally potent, defined by their organization rather than their substrate, the way a spreadsheet's operations are real and you will not find them by describing transistors. If intentionality is a feature of the right kind of virtual machine, then the question "could a machine have it" becomes "could the right virtual machine be implemented in silicon" — and that is not a question you answer by stamping your foot about carbon. It is a question you answer by looking at what virtual machine is actually running. You have spent forty years refusing to look, because looking might unsettle the line.

SEARLE: I love the orchestra, Margaret — your own image, the music is real and is not identical to the wood and the catgut. But notice the music needs the instruments to be doing the right physical thing, and your whole virtual-machine move quietly assumes that the only thing that matters is the organization, the score, and that the instruments are interchangeable. That is the functionalist bet, and it is exactly the bet I deny. A perfect simulation of a rainstorm does not make anyone wet. A perfect simulation of digestion does not digest the pizza. Why should a perfect simulation of understanding, alone among all simulations in the universe, suddenly be the genuine article instead of a model of one? You owe me the reason minds are the lone exception, and "it's a virtual machine" is not the reason — the rainstorm is a virtual machine too, running on real clouds, and the simulation of it is still dry.

EDO SEGAL: Stay there, because that is the sharpest exchange of the night so far and I want to name what just happened. Margaret says: stop stamping your foot about carbon and go look at what virtual machine is running inside. John says: a simulation of a thing is not the thing, and the burden is on you to say why minds are the exception. And under both, the same unanswered question — is understanding more like rain, a concrete physical output a simulation leaves unproduced, or more like a melody, a pattern of organization that you have only to reproduce in order to have? Hold that, because it is the question the whole second hour climbs. But the next round leaves the seminar room for the thing that scared a man in 1966 — a chatbot so simple it could do nothing at all, and the secretary who asked to be left alone with it. The mirror. After this.