On thinking about thinking, and the man who was supposed to be doing it.

I. A Privilege of the Gods

Aristotle allowed one being in the universe to think about thinking and about nothing else, and he was careful about who it was. The first mover of the twelfth book of the Metaphysics will not think of anything outside itself, because for a mind of that rank any other object would be a step down — “change would be for the worse,” and change of any kind is already a kind of motion, which the unmoved cannot afford. So it thinks the only thing worthy of it. “It is itself, therefore, that it understands, if indeed it is the most excellent thing, and the active understanding is active understanding of active understanding.” The rest of us were not forbidden the exercise; we were told where it takes place. Knowledge and perception and opinion and understanding, Aristotle notes in the same chapter, “have always something else as their object, and themselves only by the way.” A man perceives that he thinks as he perceives that he walks: in passing, on his way to something else. Thought with no object but itself was not, in Aristotle, a study skill. It was the definition of God, and it was reserved to him for a reason that flatters no one else: only the most excellent thing in the universe has nothing better to look at than itself.

Twenty-three centuries later the privilege was extended to schoolchildren. In the 1970s the developmental psychologist John Flavell gave the divine occupation a name, metacognition, and in 1979 a definition, “knowledge and cognition about cognitive phenomena”, and a four-part architecture of knowledge, experiences, goals and strategies by which a child might be taught to monitor its own memory and comprehension. The paper was modest and the discipline it announced was not. Within a generation the word had left the laboratory for the classroom and left the classroom for the shelf marked self-improvement, where it now sells, under the instruction to plan, monitor and evaluate, a faculty that Aristotle thought too expensive for anything short of the first cause. The instruction is always the same. Step back from your thinking. Watch it. Correct it. And the instruction rests on an assumption it never states, because to state it would be to expose it: that there is someone available to do the stepping back. The science, to be fair to it, posits no one; its models are of processes, not persons. The someone arrives with the instruction, which no process can follow, and which is followed every day by people who have taken it to be addressed to them.

II. The Second Man

The assumption has a shape, and the shape is a man. Behind the thinking there is supposed to stand a second one, quieter and better placed, who observes the first at work, notes his errors, and intervenes — an overseer, a supervisor of the mind, whose post is above the thoughts and whose reports are what the metacognitive practitioner calls self-knowledge. No one who buys the instruction believes he is buying a second man. He believes he is developing a skill. But the skill has a grammar, and the grammar requires a subject: someone must be doing the monitoring, and it cannot be the thing monitored, or the exercise is a dog chasing the part of itself that is chasing. Comte looked at the arrangement in 1830 and refused it in a sentence: “The thinker cannot divide himself into two, of whom one reasons whilst the other observes him reason.” He pronounced the method “radically null and void”. It went on to fill a shelf.

The tradition put that man in his post and then, one after another, went to see whether he was there. Descartes installed him as “a thing that thinks”, the residue that survives every doubt, and he was allowed to hold the position on the strength of the cogito, which is an excellent reference and not an inspection. Lichtenberg, in a notebook not meant for publication, wrote down what the cogito had actually earned: “It thinks, we should say, just as we say, it lightnings. To say cogito is already too much if we translate it as I think.” The lightning does not require a lightener. Nietzsche, who had read him, made it a rule of hygiene for logicians: “a thought comes when ‘it’ wishes, and not when ‘I’ wish”, so that the subject I is not the condition of the predicate think but an addition to it, a grammatical courtesy mistaken for a fact.

Hume had gone to the post in person, which is more than either of them did, and reported an absence with the exactness of a man who expected to find someone in. “For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never can catch myself at any time without a perception, and never can observe any thing but the perception.” He went in looking for the overseer and met only what the overseer was supposed to be overseeing. There was no second man in the room. There was the room.

Kant rebuilt the whole office around the vacancy. The “I think”, he granted, must be able to accompany all my representations — nothing is mine that it cannot accompany — but that accompaniment is a form, not a face; it is the condition of there being an experience at all, and not an item that could appear inside one. Whatever inner sense delivers, it delivers “only as we appear to ourselves, not as we are in ourselves”; and the conclusion, drawn with his usual scruple, is that consciousness of self is not knowledge of self. What survives of the overseer, in other words, is real, and cannot be inspected, and does not know itself; it is the signature at the bottom of every page and the subject of none of them. This is not a man who could be trained to monitor anything. It is the reason the page has a reader at all, and the reason the reader will never be one of the things read.

III. Telling More Than We Can Know

Suppose the post were staffed after all. Suppose the second man existed, and reported. The next question is what his reports are worth, and it has been asked with instruments rather than arguments.

In 1977 Nisbett and Wilson laid four pairs of identical stockings on a table in a shopping centre and asked passers-by to choose the best. The pair on the far right was chosen almost four times as often as the pair on the far left, a pure effect of position; and when the choosers were asked why, they offered the reasons one offers for stockings — the knit, the sheerness, the weave — and not one of them mentioned where the pair had been lying. They were not lying. They were doing what the overseer is supposed to do — reporting on the process by which a judgement had been reached — and the report was a confident and complete fabrication, delivered by people with no idea that they were fabricating. The paper’s title is the verdict: Telling More Than We Can Know. Half a century of work has narrowed the verdict without softening it. What a man was thinking he can often report, and when Ericsson and Simon checked such reports against the clock and the movements of the eye, they held; why he thought it is another matter, and why is the overseer’s whole business. In 2005 Johansson and his colleagues showed people two faces at a time, asked which was the more attractive, and by sleight of hand gave them the other one to explain. The switch was noticed less than a third of the time, and the explanations came anyway. One man had chosen the face before him, he said, “because I love blondes”. He had chosen a dark-haired woman.

Eric Schwitzgebel concludes that introspective judgements “arise from a shifting confluence of many processes, recruited opportunistically, none of which can be called introspection proper” — that is, the faculty the second man is supposed to exercise does not, on inspection, exist as a faculty, but as a committee of expedients that meets under one name. Ask undergraduates how well they understand a zipper and they will tell you; ask them to explain how a zipper works and the estimate falls at once, without anyone having told them anything about zippers. The name for this is the illusion of explanatory depth, and it is what the second man’s reports look like when someone checks them against the thing reported on.

The last of the instruments is the one that closes the shop. Kruger and Dunning argued in 1999 that “the skills that engender competence in a particular domain are often the very same skills necessary to evaluate competence in that domain” — that judging a performance takes the ability the performance takes, so that those least able in a field are, by the same defect, least able to know it. The result was named for them and is now cited, mostly, by people who believe it describes someone else. Its size has been disputed by statisticians ever since, which would matter more if the argument needed their numbers. Plato had made it without any. In the Charmides a knowledge of knowledge is proposed — a science of the sciences, which would let its possessor tell who knows and who only thinks he knows — and taken apart in a few pages: to tell the true physician from the pretender one must judge what he says of health and disease, and “no one at all, it would seem, except the physician can have this knowledge; and therefore not the wise man; he would have to be a physician as well as a wise man.” Turned inward, the difficulty does not ease. In doubting one’s own competence “one is using one’s judgment to make a negative assessment of one’s judgment”, and the instrument is the thing on the bench.

The consequence for the metacognitive trade deserves a moment’s respectful attention. The overseer’s post can only be filled by a man who already possesses the competence he is posted to check. Those who lack the competence cannot fill the post, and do not know they cannot; those who possess it were doing the checking before the word was coined, in the ordinary course of thinking well. The trade will reply that its classrooms show results, and they do: a review of three hundred and fifty-five studies credits pupils taught to plan, monitor and evaluate with some eight months of additional progress in a year. It also records where the progress is made. The strategies work when they are “applied to challenging tasks rooted in the usual curriculum content” — when the overseer is sent, that is, not to watch the thinker but to work on the problem. Kruger and Dunning had found the same cure for their own subjects. The least competent could be brought to see their incompetence, and the insight came, they reported, “by making them more competent”: they were taught some logic, and the overseer improved with the man he oversaw. The industry therefore addresses its product to a market of two kinds: those who cannot use it, and those who do not need it; and what it delivers to the first, when it delivers anything, is the competence it was supposed to supervise. It has done remarkably well on those terms, and one hesitates to disturb an arrangement so evidently satisfactory to all parties.

IV. Where the Reviewer Sits

Grant everything, then. Grant a second man, competent, honest, and sober about stockings. Post him above the thinking with instructions to review it. The difficulty that remains is not in his reports. It is in his chair.

Ryle gave the difficulty its exact form. A man can review what he has just thought; he can then review the reviewing; he can review that. But the commenting “is not, and cannot be, the step on which that commentary is being made” — the commentary is always about the performance before it and never about the performance it is. Self-commentary, he concluded, is “logically condemned to eternal penultimacy”. The reviewer can catch every thought but one, and the one he cannot catch is the catching: “He never succeeds in jumping on to the shadow of his own head, yet he is never more than one jump behind.” The distance does not close, because closing it would be one more jump. Ryle drew the moral without sentiment: “I” no more names an elusive Self than “now” names an elusive Present. The word does its work perfectly well. It points at the man who says it, and at no one standing behind him.

The philosophers of consciousness who rebuilt the second man in the language of representations came to the same chair by a different route. On the higher-order thought theory a state becomes conscious when a further thought is directed at it; and when the objection is raised that this further thought would need a further thought to be conscious, and so on without end, the standard reply, Rosenthal’s — offered in good faith and, as far as it goes, correct — is that the higher thought is not itself conscious. The overseer, that is, does his work in the dark. He is aware of the thought and not of his awareness; to become aware of that, a third man is needed, who is likewise unaware of himself. The post is filled, on this account, by a succession of officials each of whom sees the one below and none of whom has ever seen himself. The phenomenologists, coming from the other direction, reached the same place and called it by a gentler name: every experience carries an implicit awareness of oneself as its subject “without explicitly representing the self as an object of awareness”. Pre-reflective, they called it — the self as the one having the experience, never the one appearing in it.

All of which had been put, more briefly than by any of them, in a handful of lines that were written about something else. Wittgenstein, asking where in the world the subject is to be found, began by withdrawing the candidate: “There is no such thing as the subject that thinks or entertains ideas.” What remains is not a man. “The subject does not belong to the world: rather, it is a limit of the world” — and not, he added, the human soul “with which psychology deals”, which is where the instruction has sent everyone to look. To the objection that surely this is like the eye and what it sees — that the seeing has to be done from somewhere — he gave the answer the whole tradition had been circling. You do not see the eye. “And nothing in the visual field allows you to infer that it is seen by an eye.” Then he drew the picture every overseer keeps of himself — an outline like an egg, and at its narrow end a small eye, looking in — under a sentence that strikes it out: “For the form of the visual field is surely not like this.” The eye does not sit at the rim of what it sees; it is nowhere in it, rim included. Every object in sight can be examined; the one thing that cannot be examined is the condition of there being anything in sight. The metacognitive practitioner has been instructed to look for the overseer inside the field. He will find, as Hume found, the furniture; he will find, as Nisbett and Wilson found, an obliging story about the furniture; and he will find, as Ryle found, the shadow of his own head, moving each time he moves. What he will not find is the eye, for the same reason the eye has never been found: the search is being conducted with it. He is looking for the limit of everything he could look at, and no one has ever looked at a limit from the far side.

V. I, Your Glass

What, then, is thinking about thinking, once the second man has been relieved of his post? Not nothing; but not what was sold.

The soundest answer in the literature is also the least flattering to the trade, because it points the wrong way. Asked whether he believes it will rain, a man does not turn inward to inspect a belief. His eyes, as Evans put it, are “directed outward — upon the world”: he looks at the sky. To find out what one believes, one thinks about the thing; the belief is not an object to be discovered in a drawer but a stance to be taken on evidence, and the question “what do I believe?” is answered by the same act that answers “what is true?”. Understanding, which is what the curriculum actually monitors, submits to the same test: one finds out whether one understands a zipper by explaining the zipper, and an illusion that survived every inward glance does not survive the first attempt at the mechanism. The honest form of metacognition is therefore not a watch kept on the thinking but a watch kept on the thing thought about. The overseer’s only useful instruction is to stop watching the thinker and look where the thinker is looking.

The other answer is older, and the tradition has never been able to decide whether it is a consolation or a warning. “Tell me, good Brutus, can you see your face?” “No, Cassius; for the eye sees not itself, / But by reflection, by some other things.” Brutus has half of Wittgenstein’s point, three centuries early, and draws the only conclusion open to a man who has half of it: if he is to be seen, something else must do the showing. Cassius agrees, and offers his services. “And since you know you cannot see yourself / So well as by reflection, I, your glass, / Will modestly discover to yourself / That of yourself which you yet know not of.” It is the most reasonable offer in the play. It is also the first move in a conspiracy, and it trades on the half of the point that Brutus lacks: a glass returns a face, never the seeing, and a face can be worked on. Cassius says as much the moment he is alone — “Thy honourable metal may be wrought / From that it is disposed” — and the self his glass discovers to Brutus is the one Cassius needs him to have: a Brutus fit to kill Caesar, reflected back with such care that Brutus takes it for a discovery. The eye that cannot see itself must borrow another’s; and the other has reasons.

That is where the essence of the thing lies, and it is not where the curriculum put it. Thinking about thinking is not a vantage above thought. It is thought’s discovery that it has no such vantage — that there is no seat for the reviewer, in the room or at its rim, that his reports about the room are stories, that the competence to check is the competence being checked, and that the only mirrors available are the world, which does not flatter, and other people, who do. The god in the twelfth book could make his thinking its only object because nothing better was available to him and no one was there to deceive him. The rest of us know our thinking, as Aristotle said, only by the way, and we are thinking with the eye we are looking for; the honest end of the search is not to find it but to notice which way it is turned. The post was always unattended. It was simply never a post — and the man who believes he holds it is the one looking hardest into the field for the shape of his own head.

The Unattended Post

On thinking about thinking, and the man who was supposed to be doing it.

Achraf SOLTANI — September 28, 2026

Sources

Stanford Encyclopedia of Philosophy (principal source): Self-Knowledge (Gertler, rev. 2021) · Introspection (Schwitzgebel, rev. 2024) · Self-Consciousness (J. Smith, rev. 2024) · Epistemic Self-Doubt (Roush, 2017) · Higher-Order Theories of Consciousness (Carruthers and Gennaro, rev. 2026) · Kant’s View of the Mind and Consciousness of Self (Brook and Wuerth, rev. 2020) · Gilbert Ryle (Tanney, rev. 2026) · Aristotle’s Metaphysics (Cohen and Reeve, rev. 2025).

Primary texts: Aristotle, Metaphysics Λ.9, 1074b (Reeve; 1074b35–36 Ross) and Nicomachean Ethics IX.9 (Ross); Comte, Cours de philosophie positive (1830), in James’s translation, The Principles of Psychology (1890); Descartes, Meditations II (Cottingham); Lichtenberg, Sudelbücher K 76 (Tester); Nietzsche, Beyond Good and Evil §17 (Zimmern); Hume, A Treatise of Human Nature 1.4.6; Kant, Critique of Pure Reason B132, B153, B158 (Guyer and Wood; Kemp Smith); Plato, Charmides 167a–171c (Jowett); Ryle, The Concept of Mind (1949), ch. VI; Wittgenstein, Tractatus Logico-Philosophicus 5.631–5.641 (Pears and McGuinness); Shakespeare, Julius Caesar I.2.

Secondary: J. H. Flavell, “Metacognition and Cognitive Monitoring”, American Psychologist 34 (1979), 906–911; R. E. Nisbett and T. D. Wilson, “Telling More Than We Can Know”, Psychological Review 84 (1977), 231–259; K. A. Ericsson and H. A. Simon, Protocol Analysis (1984; rev. 1993); P. Johansson, L. Hall, S. Sikström and A. Olsson, “Failure to Detect Mismatches Between Intention and Outcome in a Simple Decision Task”, Science 310 (2005), 116–119, and P. Johansson et al., “How Something Can Be Said About Telling More Than We Can Know”, Consciousness and Cognition 15 (2006), 673–692; L. Rozenblit and F. Keil, “The Misunderstood Limits of Folk Science: An Illusion of Explanatory Depth”, Cognitive Science 26 (2002), 521–562; J. Kruger and D. Dunning, “Unskilled and Unaware of It”, Journal of Personality and Social Psychology 77 (1999), 1121–1134; Education Endowment Foundation, Teaching and Learning Toolkit, “Metacognition and Self-Regulation” (review updated May 2025); G. Evans, The Varieties of Reference (1982); R. Moran, Authority and Estrangement (2001); S. Tester, “G. C. Lichtenberg on Self-Consciousness and Personal Identity”, Archiv für Geschichte der Philosophie 95 (2013), 336–359.