“How You Look from the Outside” Comes Later: How a Particular Record Acquires Generalized Status

Kirill Shebetov

Research

“How You Look from the Outside” Comes Later

Kirill Shebetov

The words “view from outside” appear in a caption. The caption alone does not establish how others see the person in the image.

On this page

How to examine a caption

Sepia photograph of a child sitting on a blanket, holding an object near their mouth. — This is how others see him

See the image with this caption

Cropped from a photograph supplied by the author as a childhood photo. The caption was added for this example.

The caption names ‘others’, but does not specify who they are, the conditions in which they look, or a rule for combining their views. The image and caption shown here do not supply those observations. Reading the caption as a shared view therefore adds something that this example has not established.

The same questions can now be applied to another caption:

Preserve the selected image and, separately, the exact caption. Specify their order of presentation.
Identify the claim: about the camera’s position, the person in the image, or how “others” see them. Who are these “others,” and under what conditions are they looking?
Specify the material supporting the claim. If several observations are involved, state how they are combined into a single result.
Consider whether the same starting data, under the stated conditions, allow different answers to the question being examined.

If they do, the supplied grounds are insufficient for a definite conclusion. This does not establish that the claim is false.

What remains a record

The caption does not become the view it attributes to others.

Additional data can support a specific claim about the image. Even a well-supported caption remains writing; it does not become what it names.

The examination concerns the selected material and the stated conditions. Insufficient grounds do not establish the absence of external positions, observation, or recognition of a person.

The explanation, criteria, and outcome are also presented as records. Labels such as “confirmed” and “beyond the record” do not replace the grounds for assigning them. The assessment itself is subject to the same questions.

The study’s title, its formulas, and this text receive no exemption. Letters assembled into words do not become what they name.

What this research is for

Captions such as “how others see you” and “what you really look like” invite readers to draw a general conclusion about a person from a single image. The study examines what supports that reading.

The camera’s position, recognition of a person, and agreement among observers each require their own supporting evidence. Establishing one does not establish the others.

The aim is to consider three things separately: the image, its caption, and the grounds for treating the image as a view shared by others. The same examination can be applied to a photograph, a screenshot from a video call, or a generated image. The way the record was produced also needs to be examined.

The analysis in the paper

Notation, conditions, and revision of conclusions

What the notation stands for

W₀: the selected visual record
A₁: a later written claim about it
C: the specific claim being tested
I: the conditions and rules declared before the test
K(W₀,I): the situations admissible under that record and those conditions
P: the checking procedure

Source order

In the W₀/A₁ pair analyzed here, a particular visual record W₀ is fixed first. A written attribution A₁ follows: “view from outside,” “how others see you,” or “what you really look like.” The order of presentation does not determine whether the attribution is true. Even a supported attribution A₁ does not become part of the earlier W₀ retroactively.

The study breaks the move from a particular frame to a general claim about an outside view into a sequence of claims. Each step is tested under an explicit contract I. Independent data and a declared procedure may establish the camera’s position relative to a selected boundary, the causal origin of the image, or agreement within a selected group. No such narrow result by itself specifies the observer class and aggregation rule needed for a single generalized “view from outside.”

The formal procedure

Level 0 — Source order: W₀ is bound to an exact file, frame set, hash, or identifier. A₁ is recorded verbatim and separately. The content of the later attribution is not read back into W₀.
Conditions for the test: Before any semantic status is assigned, the procedure checks that C is well formed, I is specified, K(W₀,I) is nonempty, and C has a defined value in every admissible situation.
Level 1 — Invariance: If admissible situations preserving the same W₀ and I give C different values, the claim is beyond the record. If C is invariant throughout a nonempty K(W₀,I), it receives a TRUE or FALSE status relative to the declared contract.
Checking the procedure: Procedure P is checked separately. It may confirm a value already shown to be invariant under the stated conditions, but it cannot turn a beyond-record claim into a record-determined one.

t(W₀) < t(A₁)

W₀ → “an image of a person” → “this is S” → “S viewed from outside” → “this is how others see S” → “this is what S really looks like”

Each arrow adds a distinct claim. Support for one step does not automatically support the next.

Cases examined

Multiple external positions: Even if several camera positions outside the selected boundary are established, their particular results do not produce a single generalized result without a declared selection or aggregation function.
One projection: One frame does not determine surfaces it does not show, another moment, motion, or the result from another position. Admissible situations can preserve every pixel of W₀ while assigning different values to the generalized claim in the caption.
Unspecified “others”: The formula “this is how others see you” requires a declared observer class, observation conditions, and a rule for aggregating their potentially divergent results.
Model agreement and provenance: Agreement with a stated optical model does not establish which scene actually produced W₀. Capture, rendering, and any transition between them must be included in the declared causal chain.

|V| > 1 ∧ s is unspecified ∧ g is unspecified ⇒ a single “view from outside” is not determined

Revising a particular classification

The methodological classification is wrong if a claim classified as BEYOND THE RECORD is invariant throughout the predeclared, nonempty K(W₀,I), or if an admissible situation gives the opposite value to a claim classified as RECORD-DETERMINED.

The condition for the paper’s general claim

The scoped claim is refuted by a stable, universal, procedure-independent representation of “S as seen from outside” whose value remains unchanged across all admissible external positions, conditions, and observers. Neither high recognizability nor a stable facial feature is enough: the falsifier must match the level of the generalized claim.

These conditions specify when to revise a classification or the stated general claim. The written criteria and conclusions remain material for the same examination.

Full text

English language edition · 13 September 2026. Original research files are listed separately below.

Download the English reading edition · ZIP

English language editionPDF · 11 pp.

Original editions

Original English editionPDF Russian versionPDF · 12 pp.

Citation

Shebetov, K. (2026). “How You Look from the Outside” Comes Later: How a Particular Record Acquires Generalized Status (English language edition, 13 September 2026).