Hossein Jorjani
First publication: 2026 – 08 – 06
A peculiar difficulty arises whenever a statement, rule, or theory is allowed to apply to itself. The liar paradox, Russell’s paradox, and Wittgenstein’s ladder all involve forms of such self-application and are therefore naturally compared. Yet they should not be regarded as versions of the same paradox. The liar produces a semantic contradiction involving truth and falsity; Russell’s paradox produces a formal contradiction involving set membership; Wittgenstein’s ladder presents a different problem, in which a philosophical account of meaningful language appears to exclude the meaningful formulation of that account itself. Their common interest lies not in producing identical logical results, but in showing the different difficulties that can arise when the conditions established by a system are turned back upon the system itself.
The Liar Paradox
The simplest form of the liar paradox is the sentence:
“This sentence is false.”
Let the sentence be L. If L is true, then what it says must be correct; but what it says is that L is false:
L is true → L is false.
If L is false, however, then its assertion that it is false appears to be correct:
L is false → L is true.
Thus:
L is true ↔ L is false.
The difficulty arises because the sentence does not merely make an assertion; it applies a truth-related predicate to itself.
Not every form of self-reference, however, produces the same contradiction. Imagine a board containing twenty statements, one of which says:
“All the statements on this board are false.”
Let this statement be B. If B is true, then every statement on the board must be false, including B itself. Hence:
B is true → B is false.
But if B is false, we can conclude only that at least one statement on the board is true. That statement could be one of the other nineteen. No contradiction necessarily follows. Only under additional conditions—for example, if all nineteen other statements are false—does the self-reference reproduce the structure of the pure liar paradox. Self-reference therefore creates the possibility of paradox, but does not by itself guarantee one.
One way of controlling such difficulties is to distinguish an object language from a metalanguage. Statements about objects belong to one level; statements assigning truth or falsity to those statements are made at another. If the statement evaluating the first language remains outside the collection it evaluates, self-application can be avoided. The problem returns when a language is permitted unrestrictedly to apply its own truth predicate to its own sentences.
Russell’s Paradox
Russell’s paradox arose from the foundations of set theory. If any definable condition were permitted to determine a set, one could apparently construct the set of all sets that are not members of themselves. Let:
R = {x : x ∉ x}.
We may then ask whether R is a member of itself.
If:
R ∈ R,
then, according to the defining condition of R:
R ∉ R.
Conversely, if:
R ∉ R,
then R satisfies the condition for membership in R, and therefore:
R ∈ R.
Thus:
R ∈ R ↔ R ∉ R.
The resemblance to the liar paradox is evident. In one case a sentence applies the predicate “is false” to itself; in the other, the membership condition “is not a member of itself” is applied to the set defined by that very condition. But the domains are different. The liar concerns semantic predicates such as truth and falsity, whereas Russell’s paradox exposes an inconsistency generated by unrestricted set formation.
Russell’s response involved restrictions upon such self-application, developed most systematically through the theory of logical types. Expressions belonging to different logical levels could not be applied indiscriminately to themselves or to expressions of inappropriate types. The purpose was to prevent the vicious circularity from which the contradiction arose.
Wittgenstein’s Ladder
Wittgenstein’s ladder presents a subtler problem. It is neither an ordinary semantic paradox nor a contradiction within set theory. It arises from the account of language developed in the Tractatus Logico-Philosophicus.
According to the early Wittgenstein, a meaningful proposition represents a possible state of affairs. Such representation is possible because proposition and reality share a logical form. Yet logical form cannot itself be represented as though it were another fact in the world. A proposition displays or shows its logical form through its structure; it cannot stand outside that structure and state the logical conditions that make representation possible.
The difficulty is that the Tractatus itself appears to formulate propositions about language, reality, representation, and logical form. The book therefore seems to use language to state the conditions governing meaningful language while simultaneously maintaining that those conditions cannot themselves be stated in the same manner.
Schematically, the problem is not:
T ↔ ¬T.
Rather, it is:
If T is correct, the propositions used to state T cannot themselves function as ordinary meaningful propositions of the kind described by T.
This is not the strict contradiction found in the liar or Russell’s paradox. It is a problem of theoretical self-exclusion: the theory appears to undermine the status of the discourse through which the theory itself is expressed.
Wittgenstein explicitly confronted this difficulty near the end of the Tractatus. In proposition 6.54, he compared his propositions to a ladder. The reader uses them to attain a certain understanding and then recognizes that they cannot be retained as ordinary meaningful propositions. The ladder, once climbed, is to be thrown away. Whatever interpretive difficulties this solution may create, Wittgenstein clearly recognized that his philosophical propositions occupied an unusual position relative to the account of meaningful language they were intended to elucidate.
Russell’s Hierarchy of Languages
In his introduction to the Tractatus, Russell considered whether the difficulty could be addressed through a hierarchy of languages. A language might be unable to speak about its own structure, while a higher-order language could speak about the structure of the first:
L₁ ← L₂ ← L₃ ← …
Here L₂ functions as a language in which something can be said about L₁, while L₃ may in turn speak about L₂. No individual language need therefore provide a complete account of its own logical structure.
Russell anticipated, however, that Wittgenstein would maintain that the same fundamental limitation applies to the hierarchy: introducing successively higher languages does not provide a standpoint outside the logical conditions governing representation. Russell, in turn, questioned whether the hierarchy itself should be regarded as a completed totality about which the same demand could be made.
The disagreement therefore reaches beyond the technical management of self-reference. Russell’s approach seeks an appropriate logical level from which statements about another level can legitimately be made. Wittgenstein’s concern is more fundamental: no hierarchy of representational languages provides language with a position altogether outside the logical conditions that make representation possible.
Three Forms of Self-Application
The three cases can now be compared more precisely. Each involves a system applying to itself a condition that it ordinarily applies to something within its domain:
Liar paradox: a sentence applies a truth-related predicate to itself.
Russell’s paradox: a membership condition is applied to the set defined by that condition.
Wittgenstein’s ladder: an account of the conditions of meaningful language is confronted with the status of the propositions through which that account is expressed.
The consequences, however, are importantly different. The liar produces a semantic contradiction:
L is true ↔ L is false.
Russell produces a formal contradiction:
R ∈ R ↔ R ∉ R.
Wittgenstein presents theoretical self-exclusion:
If the theory is correct, its own formulation cannot straightforwardly possess the status it assigns to ordinary meaningful propositions.
Self-reference and self-application should therefore not themselves be treated as paradoxes. Their significance lies in forcing a system to confront its own conditions. Sometimes the result is contradiction; sometimes restrictions on levels or types can prevent the difficulty; and sometimes, as in the Tractatus, the theory appears to place its own formulation at the boundary it has established.
The broader methodological lesson extends well beyond these three examples. Whenever a theory defines a domain and establishes conditions for legitimate statements, classifications, or operations within that domain, one further question should be asked: what happens when those conditions are applied to the theory itself? A theory need not satisfy its own conditions in precisely the same manner as the objects it describes, but the relationship must be made explicit. Otherwise, a principle that appears convincing when directed outward may become unstable when turned back upon the conceptual structure from which it originated.
Hossein Jorjani 