Being Looked At, and What It Costs to Run

Every other page in this chamber describes the same obstacle from a different angle, and describes it from the inside: the part of attention that never quite arrives, the monitoring that will not switch off, the sense of being slightly outside an experience while having it. This page is about the one place where that thing has been taken into a laboratory and measured — and the measurement is more useful than any amount of encouragement.

John Berger put it in Ways of Seeing in 1972, and the formulation has outlived most of what was written around it. Men act, he argued, and women appear. Men look at women; women watch themselves being looked at. The consequence he drew is the part that still stings: a woman comes to contain both roles at once. The surveyor and the surveyed live in one person, and the surveying does not stop when the room empties.

For fifty years that was an argument from observation — persuasive, literary, and easy to dismiss as interpretation. Then somebody tested it.

In 1998 Barbara Fredrickson and colleagues ran a study that has never stopped being cited. Participants were asked to try on a garment and, while wearing it, complete a maths problem. The garment was either a sweater or a swimsuit. Nobody watched them. There was no audience, no photograph, no evaluation of appearance at all.

Women in the swimsuit condition performed worse on the maths. Men, tested the same way, showed no comparable effect.

The finding is not that women are fragile or bad at arithmetic. It is that self-objectification — attention turned onto one’s own body as an object to be evaluated — consumes cognitive resources, and that those resources are then not available for anything else. It is a measurable, quantifiable load, produced without a single other person in the room. Berger’s surveyor, it turns out, costs something to run.

Line the finding up with what the other pages have been describing and a single mechanism appears underneath all of them.

Arousal requires attention in the body. So does pleasure. So does interoception, which is the capacity to register what is happening inside at all. And self-surveillance draws on the same finite pool. That is not a metaphor about being distracted; it is competition for one resource. Which means being looked at, including by the copy of an observer you carry around, is not an unfortunate accompaniment to the experience. It is in direct competition with it.

It also explains why the standard advice fails so reliably. Relax. Stop thinking about it. Just be present. Every one of those is a new instruction to monitor compliance with, which draws on the very pool it was meant to free. You cannot spend attention on not spending attention. Only the conditions can change, not the effort.

The 1998 result is usually repeated as a fact about women, and the way it is repeated is wrong in a way that matters. In 2004 Hebl, King and Lin ran the paradigm again with around four hundred undergraduates and made one change: the men, instead of swim trunks, were given something as revealing as what the women wore. The effect appeared in the men too.

That single modification moves the finding from one category to another. It is not a claim about a female mind, and a page that let it stand as one would be doing something this library objects to everywhere else. It is a claim about a STATE: a person whose appearance has been made salient and evaluable pays a cognitive cost, and it appears to be a human effect rather than a sexed one. What differs between men and women is not the machinery. It is how many hours of an ordinary week are spent in that state.

And the limits, because a single famous experiment is exactly the kind of evidence this library has a whole page about treating carefully. The original study was small. The paradigm is one manipulation among several used in this literature, and reviews of the experimental work on state self-objectification report results that are consistent on some outcomes and mixed on others. What is best supported is not the maths result specifically but the broader correlational finding underneath it: habitual self-monitoring is associated with reduced awareness of internal bodily states. That is the claim this page actually needs, and it is the sturdier one.

Calling it a load is accurate and not yet useful, because you cannot put down a thing you cannot see the parts of. In practice self-surveillance seems to run as three separate processes at once, and they are distinguishable from the inside once somebody has named them.

First, checking: an intermittent sampling of how you currently look, which is why a mirror, a dark window or a phone screen catches attention involuntarily. Second, simulating: running a model of what the other person is seeing and concluding. This is the expensive one, because it is a whole second mind being maintained at your own cost, and it runs even when nobody is there. Third, correcting: the small continuous adjustments of posture, expression, angle and voice that follow from the first two.

The reason the decomposition earns its place is that the three do not respond to the same things. Checking is reduced by removing surfaces. Correcting fades on its own once the other two quieten. But simulating — the expensive one — is not reduced by anything a person does with her own effort, because effort is what it is made of. It is reduced only by information: by actually learning what the other person thinks, which ends the simulation by answering it.

Which is why an explicit sentence from somebody else does more in three seconds than an hour of trying. Not reassurance, which is generic and gets discounted, but information: this is fine, I am not looking at that, nothing is expected here. A person who has been told specifically what is true has nothing left to model.

Berger wrote in 1972, when a person saw her own face a handful of times a day, in fixed places, on purpose: a bathroom mirror, a shop window, a photograph developed a week later. The surveyor he described had to be maintained largely from memory and imagination.

The front-facing camera removed that constraint. A great many people now see their own face during conversations, continuously, in real time, while also trying to listen — a condition that had no equivalent in any previous period and that an entire generation has never been without. Whatever else video calls did, they took the most expensive of the three processes above and gave it a live feed.

This page will not make a claim about what that has done at the population level, because the honest position is that nobody knows yet and the research is younger than the phenomenon. What is worth saying is narrower and checkable by anybody in about a minute: hide your own image in the next call you take, and notice whether the call is easier. If it is, that is the mechanism on this page, demonstrated on yourself, for free, in the one setting where the surface can simply be switched off.

Everything so far has treated being looked at as a cost, and a reader could finish it with the impression that the library thinks attention is a problem. It does not, and the complication is worth the space, because the thing a great many people most want is to be looked at by somebody, properly, for a while.

The awkward part is that the mechanism does not distinguish. Wanted attention recruits the same monitoring as unwanted attention, because the simulating process does not care whether the verdict it is modelling is a kind one — it still has to be run. Which explains a report this house hears often and which the person making it usually offers as a confession: he was looking at me, and I wanted him to be, and I felt almost nothing.

That is not a failure of desire and it is not evidence about the person or the evening. It is arithmetic. And the resolution is not to want less attention but to change what the attention is doing: being watched runs the simulator, and being ATTENDED TO with the verdict already settled does not. The difference between the two is whether an assessment is still pending, which is a property of the arrangement rather than of either person’s feelings.

Japan supplies unusually explicit versions of this, which has one advantage: an explicit rule can be argued with, while an atmosphere cannot. 身だしなみ is an ordinary workplace expectation with real content rather than a vague preference for neatness, and appearance standards here are frequently written down.

The clearest recent demonstration is #KuToo, begun in 2019 by 石川優実, over workplace requirements that women wear high heels. What made it significant was not the shoes. It was that a great many people had assumed such expectations were a matter of taste, and the campaign established that they functioned as requirements with consequences attached — which is a different category entirely, and one that can be contested.

Set that beside 空気を読む — the continuous inference of what a room expects — and the load described above is running more or less permanently for a great many women here. Not as anxiety. As a background process, competent, unremarkable, and expensive.

Berger said the surveyor is internalised, which is correct and leaves out the part a reader most needs: when. Almost nobody describes having had it as a small child. What people describe, over and over, is a period — usually somewhere between about ten and fourteen — in which a body that had simply been theirs became a thing that other people remarked on. Relatives commenting on shape. A school uniform that fitted differently than it had. A first piece of public appraisal from a stranger, which is remembered with a precision that almost nothing else from that year is.

That the thing was installed at a datable point, by identifiable inputs, is not a small correction. A person who believes this is how she is has a permanent property and two responses available, which is the box this library keeps finding. A person who knows it was trained in has something that was learned — and things that were learned respond differently, not because they can be simply un-learned, but because they can be examined, attributed to the people who did the training, and noticed as they fire rather than obeyed.

And it does not run at a constant volume across a life, which is the part almost never said out loud. A great many women report that somewhere around fifty the load drops substantially, and describe it in a phrase that carries both halves at once: I became invisible. It is a loss, and it is named as one, and the same people frequently say in the next sentence that they can eat in a restaurant alone now, or walk somewhere at dusk, or stop thinking about what they look like from behind — and that this is the most attention they have had available for anything since they were twelve.

Both of those are true at once and neither cancels the other. The page has no advice attached to this and is not going to manufacture any. It is here because a woman in the middle of the loud years is generally told, if anything, that the volume is permanent, and it is not; and because a woman in the quiet years is generally offered only the loss half, with nobody having mentioned that the thing she got back is the one this entire page has been describing.

Since the load is produced by conditions rather than by attitude, the interventions that work are unromantic and mostly environmental. Lighting that does not present the body for inspection. Not being positioned to see yourself. Being warm, because cold recruits attention to the surface of the body, which is exactly where the surveyor lives. Not being observed during transitions — undressing, moving, arranging — which is when most people report the spike.

One of them is not environmental at all, and it follows directly from the decomposition above: the simulating process ends when it is answered. So the intervention nobody suggests, because it sounds too small to be worth doing, is to ask. Not "do I look all right", which invites exactly the verdict the simulator is modelling and confirms that assessment is live. Something flatter, which asks for information rather than for reassurance: is this all right with you, or, are you comfortable. A specific answer closes a process that an hour of effort cannot close, and it costs one sentence.

And the largest one, which is structural rather than physical: being in a room where nothing about you is being assessed. Not a room where you are being assessed favourably — that is still assessment, and praise reports the same surveillance that criticism does. A room where the question is not live at all.

Read against this page, most of Moonlight’s arrangements stop looking like atmosphere and start looking like load reduction. The scope written in advance means nothing has to be negotiated in the room. The absence of any appraisal — spoken or implied, admiring or otherwise — means the surveyor has nothing to report. The unhurried pace means no clock is recruiting attention. The longer window afterwards means the transition out is not another performance.

None of that is a claim about what will be felt. It is a claim about what will not be spent. That is a smaller promise and it is the only one anybody can actually keep, since what a person does with returned attention is entirely her own.

Berger said a woman watches herself being looked at. What nobody could show in 1972, and can be shown now, is the price: the watching runs on the same supply as feeling anything, and it runs whether or not anyone is there. Which means the problem was never that you were thinking too much. It was that two things were drawing on one account, and only one of them had been told it was optional.