The Unfalsifiable Frame

What it is

This is the first member of the vault's structural cluster that lives in an argument rather than in an apparatus. The others are things you find at the back of a book. This one you find in the sentences, and no amount of citation quality touches it. The book that produced the category has one of the best apparatus on the shelf.

What is not being said matters as much as what is. An unfalsifiable frame is not a lie, not carelessness and not automatically worthless. Frameworks are how fields organise themselves before they know enough to test anything, and a good one earns its keep by making legible a set of phenomena that were previously a heap. The defect is silence about the status of what is being offered: a frame presented in the register of a finding, with an apparatus behind it that belongs to somebody else's findings.

The defining instance is The Elephant in the Brain, by Kevin Simler and Robin Hanson. The thesis is that people conceal their real motives from themselves, the better to conceal them from others, and that content is self-sealing by construction. If the mechanism is that a person cannot see the motive, then someone who reports not seeing it is displaying exactly what the theory predicts, and someone who does see it has confirmed it outright. The cleanest single instance is offered as a strength: having argued that laughter is a play signal, the authors write that the account also explains many of the occasions when nobody laughs, working through a clown's pratfall against a grandmother's fall, which becomes funny again once she signals she is unhurt. A theory that predicts laughter, then no laughter, then laughter again has an account of every possible observation. The milder version follows a page later, where our ignorance of what laughter is for is itself read as evidence that something is being hidden, so understanding and not understanding both confirm the thesis.

In effect

The symptom by which the defect is detected is two-sided confirmation, where a behaviour and its exact opposite are both counted as confirming. The vault considered giving that its own category and decided against it, because every instance of it is also an instance of this. Laughter and its absence. Impractical art and practical craft. Warmth and coldness between friends. Technical perfection and technical imperfection. When a theory covers a thing and its negation, it has stopped making predictions and started supplying vocabulary.

Three further instances show that this is not a complaint about one discipline. Derald Wing Sue's taxonomy of microaggressions contains a category, microinvalidation, for the denial or negation of a recipient's reported experience, and Adrian Furnham's reflexive observation is that criticism of the framework can itself be classified under the scheme's own categories. That is the same structure arrived at from an entirely different direction. Ernest Dichter's psychoanalytic readings of consumer behaviour in the 1950s, recorded under motivational research, produced an interpretation for every purchase and a test for none. And The Molecule of More supplies the cleanest single move in the collection: a dopamine-boosting drug injected into stump-tailed macaques produced submissive behaviour, the result cutting against the book's thesis that dopamine drives domination, and the book reports it honestly and then widens the theory rather than yielding, redefining domination as domination of the environment rather than of the people in it, so that submission now confirms the thesis too. The frame becomes unfalsifiable at the moment the thing to be explained is redescribed rather than the hypothesis revised.

Both sides are owed, and the defence half matters as much as the first. A framework is not automatically worthless and is not the same thing as a failed finding. Robert Trivers on self-deception is a framework, elegant and influential, faulted by reviewers for lacking direct empirical support for the central evolutionary claim rather than for having been tested and failed. Robert Kurzban's modular mind is a synthesis and an interpretive stance, to be judged on coherence. Geoffrey Miller's fitness-display theory of art is the same shape, with one small direct test behind it. Filing any of the three as failed findings would be a straightforward category error. No writer is obliged to design a test that could refute the theory in a trade book, and a theory that is hard to test today is not thereby false, continental drift having been unfalsifiable in practice for decades.

For the prosecution, the obligation is not to be falsifiable but to say which kind of thing you are offering, and to notice when your account has swallowed a phenomenon and its negation. The defining instance does notice in four places and should be credited for each: where the authors state that a suspicious reason is almost impossible to prove counterfeit, where they concede that positive side effects of signalling might be weaker and harder to detect, where the art motive is hedged to one motive among several, and where the conclusion concedes that the explanations are more compelling about the species than about a person. Where the book does not notice is where it matters most.

The detection test is the usable part. Ask what observation would have counted against the claim, and ask whether the author ever names one. Three cheap proxies follow. Look for the pair, by finding a behaviour the theory explains and then looking for its opposite in the same chapter. Look for the consciousness clause, the sentence saying the actor need not be aware of the motive, which is usually honest and is always the sentence that makes introspective evidence inadmissible. And ask what the notes are attached to, since in the defining instance the citations cluster on the material being reinterpreted rather than on the reinterpretation.

What it does not say

It does not say the theory is false. A vocabulary can be extremely useful as long as nobody mistakes it for a result, and the remedy is one sentence: say what would change your mind.

It does not impugn the authors' integrity. Simler and Hanson tell the reader that they have probably selected their evidence, that they wrote the book for prestige, and that their own explanations are often hollow applied to an individual. Very few authors in this library do any of that. The frame is a property of what they built.

It does not apply to every framework book. The Art of Clear Thinking is the contrast case: Hasard Lee states the conditions under which his own method did not work, repeatedly and in public, which is the opposite of an unfalsifiable frame, and the defect that book has instead is that its falsifiable half and its borrowed half are written in the same voice.

It is not reached by preregistration or by replication reform. Those govern claims that can be tested, and a frame never enters the pipeline.

---