An examined document · read the margins
Masked-Man Fallacy
The masked-man fallacy is the invalid inference that two things must be distinct because someone knows or believes something about one (under one description) that they do not know or believe about the other (under another description), when in fact they may be identical.
It is a formal logical fallacy, not an experimentally measured cognitive bias: it arises from misapplying Leibniz's Law (substitutivity of identicals) inside intensional contexts such as "knows that," "believes that," or "does not recognize," where co-referring terms cannot be freely swapped. The classic form is "I know who my father is; I do not know who the masked man is; therefore the masked man is not my father" — invalid because the masked man could be one's father without one knowing it. The puzzle traces to Eubulides of Miletus (4th century BCE) as the "Hooded Man"/Electra paradox, and its modern logical diagnosis comes from Quine's analysis of referential opacity (1956) and Hintikka's epistemic logic (1962).
Leibniz's Law (the indiscernibility of identicals) licenses substituting co-referring terms 'salva veritate' (without change of truth value) in extensional/referentially transparent contexts. Inside intensional contexts created by propositional-attitude verbs ('knows,' 'believes,' 'recognizes,' 'doubts') or modal operators, terms no longer occur purely referentially — they are referentially opaque — so substitution is no longer truth-preserving. The fallacy is committed by treating an opaque context as if it were transparent: from 'S knows a' and 'S does not know b' one wrongly concludes 'a ≠ b.' Quine diagnosed this as the failure of substitutivity in opaque positions; Hintikka modeled it via possible-worlds semantics where the same object can be picked out by different descriptions across the agent's epistemic alternatives.
There is genuine philosophical dispute about the underlying logic rather than about an empirical effect: Frege (1892) handled opacity by positing that terms in such contexts refer to their sense (Sinn) rather than ordinary referent; Quine rejected senses/intensions as 'creatures of darkness' and concluded one cannot coherently quantify into opaque contexts; Hintikka and later possible-worlds and de re/de dicto theorists (e.g., the de re reading restores substitutivity) reject Quine's pessimism. So the 'mechanism' debate is about the correct semantics of intensional contexts, not about a psychological process.
Catch it in the act
Watch for an argument that infers two things are different (or the same) from what someone knows, believes, recognizes, or can name about them. The tell is a mental-state verb — "knows," "believes," "recognizes," "is aware that," "doubts" — sitting next to a substitution of one name/description for another. If swapping the terms changes what the person knows or believes, the context is opaque and the substitution is illegitimate (e.g., "Citizens know Batman saved Gotham, therefore they know Bruce Wayne saved Gotham").
Cross-references in the margin
- Intensional fallacysiblingDirect synonym/alias; emphasizes that the error treats an intensional (opaque) context as extensional.
- Epistemic fallacysiblingAnother name used for the same schema, stressing the role of 'knows'/'believes.'
- Illicit substitution of identicalssiblingNames the precise logical move: misapplying Leibniz's Law / substitutivity in an opaque context.
- Referential opacity (Quine)mechanistically-linkedThe semantic property of intensional contexts that explains WHY the substitution fails; the fallacy is the failure to respect it.
- Leibniz's Law / indiscernibility of identicalsparentThe valid principle being misapplied; the fallacy is its illegitimate extension to opaque contexts.
- De re / de dicto distinctioneasily-confusedA de re reading of a belief/knowledge ascription can legitimately license substitution, so the same sentence may be valid or fallacious depending on scope — a frequent source of confusion.
- EquivocationsiblingBoth involve a hidden shift in how a term is functioning, but equivocation shifts meaning of a term whereas the masked-man fallacy shifts between transparent and opaque occurrences.