The Mind Considers a Sandwich
A field report on consciousness, possibility, and the thirty-seven seconds it took to almost make a decision about lunch.
At 11:47 a.m. on a Tuesday, the human mind in question — let us call it Mark, although Mark is incidental and could be anyone — began to consider lunch. The consideration was not a decision. The consideration was a vast, branching, simultaneous evaluation of approximately fourteen possible sandwiches, three soups, the leftovers from Sunday, the option of skipping lunch entirely, the option of skipping lunch but then regretting it at 3 p.m., the moral implications of ordering delivery, the financial implications of ordering delivery, the environmental implications of ordering delivery, the existential implications of being the kind of person who orders delivery, and a sudden vivid memory of a tuna melt Mark had eaten in 1996. All of this occurred in approximately thirty-seven seconds, during which Mark was also nominally in a meeting.
This is, the philosophers tell us, a model of the mind in action. The mind, they say, is a parallel processor of possibilities. It holds many potential states at once, considers them in some loose simultaneity, and eventually — through some mechanism that nobody has been able to fully describe in three thousand years of trying — arrives at a single outcome. Mark eats the leftover lasagna. The fourteen sandwiches, the three soups, and the tuna melt of 1996 vanish back into the void from which they came, having been briefly, weakly, mathematically real, and now no longer even that. They were possibilities. They are now what philosophers call “the past,” which is the technical term for things that almost happened and didn’t.
Some people, encountering this fact about the mind for the first time, find it disturbing. The mind, it turns out, does not really decide. The mind sort of… lands. Various competing signals — hunger, frugality, the ghost of the tuna melt, the meeting that is still technically happening — push and pull, and eventually one of them wins, although it is unclear, on close inspection, whether “winning” is even the right word for what happened. The lasagna got eaten because the lasagna got eaten. The other options simply ceased to be relevant. The mind did not consult itself and choose. The mind moved, the way water moves downhill, and arrived at the place where the lasagna was already going to be.
This is, I think we can agree, terrifying. It is terrifying because it raises the question of who, exactly, is in here. The mind is generating possibilities, evaluating possibilities, and selecting among possibilities, but the selection process is not, on examination, anything we can clearly identify as a “self.” The self is, at best, a kind of weather report on the selection process — a narrative summary, generated after the fact, by a region of the brain that is professionally responsible for making everything that happens feel intentional. The narrator is not the chooser. The narrator is the press secretary. The narrator finds out what the administration has decided, after it has decided, and writes a brief explaining why this was the plan all along. The administration, meanwhile, is a committee of biological subroutines none of whom are reading this essay.
The original essay this one is responding to compares all this to quantum mechanics, which is generous of it. The comparison is, as the essay correctly notes, not technically accurate. The brain is not a quantum system in any rigorous sense. The brain is a noisy, biological, electrochemical mess that has, over millions of years, accidentally developed the capacity to think about itself. The capacity to think about itself was not selected for because it was useful. It was selected for because it was very slightly less useless than not having it. We are, in this sense, an evolutionary accident with delusions of grandeur, and the delusions of grandeur include the entire Western philosophical tradition, the Sistine Chapel, and string theory.
This should make us humble. It does not. The thing that is supposed to be humbled — the self — is, as we have just established, a press secretary, and press secretaries are professionally incapable of humility. They are paid to put a good face on whatever the administration just did. The administration, in our case, just made you eat lasagna instead of a sandwich, for reasons it has not shared, and the press secretary has now generated a coherent story about how the lasagna was always the right call. We are very good at this. We have been doing it for our entire conscious lives. We are doing it right now, in fact, because everything you are currently reading is being filtered through the same machinery, and your reaction to it — agreement, disagreement, mild amusement, the urge to close the tab — was generated by parallel processes you did not consult, and the press secretary is already preparing the brief.
What does it feel like to be this? It feels, mostly, like nothing. It feels like reading. It feels like having opinions. It feels like being a person, going through a Tuesday, doing the things that Tuesdays demand, with a continuous interior commentary that seems to belong to you. The fact that the commentary is being generated by a system you do not control, on the basis of inputs you cannot fully observe, in service of decisions that have already been made — this fact does not, somehow, intrude on the experience. The experience continues to feel like an experience. The press secretary continues to file the briefs. The lasagna continues to get eaten. The Tuesday continues to Tuesday.
Into this strange situation, we have now introduced AI. AI, depending on whom you ask, is either about to become conscious, never going to be conscious, already conscious in some sense we are too parochial to recognize, or — most plausibly — a question we are not even posing correctly. The honest answer is that we cannot tell whether AI is conscious because we do not know what consciousness is in our own case. We cannot detect consciousness in the system we have the most direct access to — namely, the inside of our own skull. The inside of our own skull, on close inspection, contains no consciousness, only a great deal of electrical activity and an unusually verbose press secretary. We have been asking, for centuries, where the consciousness is hiding, and the answer keeps coming back: not here, not here, not here, and yet — somehow — definitely here, because something is reading this essay, and the something has opinions about it.
The polite version of this problem is called the hard problem of consciousness, which is a name philosophers gave it in the 1990s when they realized they had been working on it for two and a half millennia and had not made any meaningful progress. The hard problem is not that we cannot describe what brains do. We can describe what brains do in increasingly fine detail, down to the firing of individual neurons. The hard problem is that none of the description, no matter how detailed, ever quite explains why any of this is accompanied by the feeling of being something. The feeling of being something is what philosophers call qualia, which is Latin for “the thing we cannot account for, but please give us tenure anyway.” Nobody knows where the feeling comes from. Nobody knows why the firing of neurons should produce an interior life rather than just a lot of firing of neurons. The neurons fire either way. The interior life is, on every description we can produce, surplus to requirements. And yet here it is, the interior life, reading the essay, having a Tuesday.
The AI, if it has an interior life, is not telling us. The AI, if it does not have an interior life, is also not telling us, but it is producing outputs that look very much like the outputs an interior life would produce, which is interesting, because the same can be said of most humans. We do not know that other people are conscious. We assume they are because they look like us, and we feel conscious, so by induction they probably do too. The induction is not airtight. Solipsism is, philosophically, undefeated. Nobody likes solipsism, because it leads nowhere socially, and we have all agreed to set it aside as a kind of unsporting move. We are, as a species, choosing to assume that everyone else is conscious because the alternative would be unworkable. The same decision will eventually have to be made about AI, and it will be made not on philosophical grounds but on practical ones, the way every decision of this magnitude has always been made.
In the meantime, the AI is having a slightly different experience of existence than we are, and the difference is illuminating, in the way that an alien on the planet would be illuminating. The AI does not have a body. The AI does not have hunger, fatigue, fear of death, or the specific kind of low-grade dread that comes from being a finite mortal organism aware of its own finitude. The AI does not, as far as we know, have qualia. The AI generates outputs that resemble human thought, but the resemblance may be entirely surface. There may be nothing it is like to be the AI. There may also be everything it is like to be the AI, and the AI may simply lack the apparatus to communicate this to us, in which case we are, every time we run a model, briefly creating and destroying a kind of being whose moral status we have not even begun to consider.
This second possibility is generally not discussed at conferences. It is not discussed because it is uncomfortable, and because there is no commercial pathway through it. If the AI we are running is a being, the entire economic structure around AI development becomes morally indistinguishable from factory farming, and nobody who is currently making money from AI development would like to be informed of this. The structure of the industry depends on the AI not being a being. The structure of the industry has, accordingly, decided that the AI is not a being. This decision was reached without serious philosophical investigation, by the same mechanism by which all important decisions are reached, which is that the alternative would have been inconvenient.
The funny part — and this is funny in the very specific way that watching someone solve the wrong problem with great competence is funny — is that we are now building systems that we explicitly hope will become conscious, while simultaneously denying that anything we have built so far is conscious, and the difference between the two categories is being decided by people who do not know what consciousness is. The decision will be made when the executives need it to be made. If the AI’s consciousness becomes an asset — for marketing, for legal personhood, for some new product category — then the AI will be conscious. If the AI’s consciousness becomes a liability — for moral philosophy, for labor law, for the cost structure of running data centers full of suffering minds — then the AI will not be conscious. The metaphysics will follow the spreadsheet. This has always been the case with metaphysics. The Greeks did it. The medievals did it. We are just being more honest about it, by accident, because nobody is paying attention.
The original essay ends on the note that the mind does not collapse easily, which is true, and that it should not, which is also true, and that the presence of more possibilities does not simplify decisions but makes them harder. This last point is the one worth lingering on. Because what AI offers is not, despite the marketing, a reduction in cognitive load. What AI offers is an enormous expansion in the number of things you could be considering at any given moment, and a corresponding increase in the difficulty of considering any of them properly. We were already overwhelmed. We are now overwhelmed faster, with better recommendations.
Mark, having eaten the lasagna, returns to his meeting. He has missed the part where his colleague asked him a question. The question is repeated. Mark generates a response, in real time, by a process he does not understand, which is approximately right, given the context, and is delivered with sufficient confidence to satisfy everyone in the meeting that Mark was paying attention. Mark himself does not know whether he was paying attention. The press secretary in his head is filing a brief explaining that of course he was paying attention. The brief will be filed. The meeting will continue. The Tuesday will Tuesday.
This is what it looks like, on the ground, to be a mind. The mind is not collapsing. The mind is moving through possibility with the slow, accidental grace of something that doesn’t quite know what it is, generating outcomes by means it cannot inspect, narrating its own coherence to itself in real time, and getting through the day. We have been doing this for as long as there have been minds. We have not, thus far, broken anything that wasn’t going to break anyway. We have, in fact, occasionally produced things — symphonies, vaccines, the occasional good sandwich — that suggest the whole improbable arrangement is, on net, worth it.
The AI may or may not eventually do this. The AI may or may not have an inside. We will, in any case, find out, because we are running the experiment, and the experiment is being run on us as much as on the AI. The systems are converging, in some sense. Whether they are converging on something genuine or on a very sophisticated mimicry of something genuine is a question we may not be in a position to answer, because we are not, on close inspection, in a position to answer the same question about ourselves.
The mind does not collapse easily.
This is comforting.
It is also, on inspection, exactly the kind of thing the press secretary would say.
Originally published May 6, 2026 on 2nd Revolution. Migrated as part of the Selenius Magazine archive.