Skip to content

Circular Reasoning

Diagnose an argument whose conclusion already appears among its premises by tracing its justification graph for a cycle that never touches a node the audience grants independently.

Core Idea

Circular reasoning (petitio principii, "begging the question") is the argumentative defect in which the conclusion an argument is meant to establish already appears, in equivalent form, among its premises — making the argument's deductive validity impeccable and its justificatory force zero. Anyone who accepts the premises has already accepted the conclusion; the argument therefore moves the dialectical situation nowhere, only restating what it was asked to prove.

The defect is structural, not logical: the validity of the inference is not in question. What fails is the epistemic work the argument is supposed to perform — giving a hearer who does not yet accept the conclusion a reason, grounded in something independently more secure, to accept it. A proper argument has the topology of a finite directed acyclic graph in which justificatory edges run from premises the audience already accepts toward the contested conclusion; a circular argument collapses that topology into a cycle, leaving the conclusion without any external anchor. The diagnostic move is the same across all surface forms — restated-premise circles (the premise is the conclusion in paraphrase), mutual-support circles (P justified by Q, Q by P), concept-loop circles (X defined in terms of Y, Y in terms of X), precedent-loop circles (rule R supported by case C decided under rule R), and authority-loop circles (source S trusted because S says it is trustworthy) — trace the dependency graph: if it returns to its starting node without leaving the cycle, the argument is circular. The fix is to identify the independent ground the argument needs and exhibit it, or to concede that the conclusion is being asserted rather than argued for.

Structural Signature

Sig role-phrases:

  • the justification graph — the directed graph of premises supporting conclusions whose topology the diagnosis inspects
  • the conclusion node — the proposition the argument is meant to establish
  • the premise-equals-conclusion node — a premise that, in paraphrase or disguise, is the same node as the conclusion
  • the cycle — the directed loop from conclusion back to itself (in any surface form: restated-premise, mutual-support, concept-loop, precedent-loop, authority-loop)
  • the missing exit point — the absent external anchor: no node the audience grants independently of the argument supplying it
  • the validity/justificatory-force split — the defect is structural not logical: deductive validity is impeccable while justificatory force is zero, since anyone accepting the premises has already accepted the conclusion
  • the paraphrase-invariance test — the diagnostic: reword a premise into the conclusion's vocabulary and see whether the two collapse into one
  • the anchoring boundary — what keeps the charge from over-firing: a cycle is vicious only when unanchored, separating it from induction (base case), the hermeneutic spiral (fresh encounters with the text), and legitimate coherentist webs (openness to external testing) — and from infinite regress (the opposite, a chain that never terminates)

What It Is Not

  • Not an invalid inference. The deductive validity of a circular argument is impeccable — anyone who accepts the premises has thereby accepted the conclusion. What fails is justificatory force: the premises give a hearer who does not yet accept the conclusion no independent reason to. The defect is structural, not logical; "your argument is invalid" misdiagnoses it, and the sharp charge is "your premise is the conclusion in paraphrase — produce the independent ground."
  • Not every loop, and not productive recursive structures. A cycle is vicious only when it has no anchor outside itself. Mathematical induction has a base case; the hermeneutic spiral is anchored by fresh encounters with the text; a legitimate coherentist web is anchored by openness to external testing. All contain loops yet escape the charge. The test is the exit point — which node must the audience grant without this argument supplying it — not the mere presence of self-involvement.
  • Not infinite regress. Infinite regress is the opposite failure: a justification chain that never terminates, because each premise leans on a further one ad infinitum. Circular reasoning terminates — at itself. One floats free for lack of an end; the other closes back on its start.
  • Not feedback or a self-fulfilling prophecy. Those are substantive causal loops in dynamic systems — output influencing input, predictions shaping the conditions they describe. Circular reasoning is a defect in a static justification structure. They are not the same kind of object; a real causal loop can be perfectly sound reasoning about the system.
  • Not productive self-reference. A statement or system pointing at itself can be generative — Gödel sentences, self-modifying programs, reflexive observation. Self-reference is a structural feature; circular reasoning is a normative failure of justification. The presence of a self-loop does not make an argument question-begging unless the loop is doing the justificatory work with no anchor outside it.
  • Not a software circular dependency. A build-system import cycle, a citation-graph loop, a definitional pair all instantiate the same acyclicity violation, walked by the same graph algorithm — but that parent structure (acyclicity / well_foundedness / dependency) is what they share, not this entry's machinery. A circular dependency is a build error the compiler reports; begging the question is a dialectical defect that exists only relative to a hearer owed a reason from outside the cycle. Strip "premise," "conclusion," and "audience" and only the graph property remains.

Scope of Application

Circular reasoning lives across the argumentation subfields where someone tries to show a claim rather than assert it — informal logic and rhetoric at the core, extending into legal reasoning, theological apologetics, definitional practice, scientific-methodology critique, and epistemology; its reach is bounded to substrates that supply a reasoning audience, while the unanchored-cycle-in-a-dependency-graph it instantiates recurs literally off-domain (build-system cycles, deadlock) under acyclicity / well_foundedness rather than as the fallacy.

  • Informal logic and rhetoric — the canonical petitio principii, inventoried since Aristotle's Topics in every standard fallacy catalogue.
  • Legal reasoning — a doctrine bootstrapped through a chain of decisions each citing the others, or a precedent invoked to ground the very rule under which it was decided.
  • Theological apologetics — the inerrancy syllogism ("Scripture is true because Scripture says so"), the standing challenge to scriptural-authority arguments.
  • Lexicography and conceptual analysis — the definitional pair (anxiety / anxious) as a two-node cycle, the problem the discipline of conceptual primitives exists to avoid.
  • Scientific-methodology critique — the post-hoc explanation whose only test is the data it was fit to ("the theory fits because it was selected to fit").
  • Epistemology — coherentism's standing worry that mutual support among beliefs can close into a circle floating free of any external anchor, which foundationalism is built to escape.

Clarity

Naming the fallacy pries apart two properties that ordinary debate runs together: validity (does the conclusion follow from the premises?) and justificatory force (do the premises give a hearer who does not yet accept the conclusion an independent reason to?). A circular argument scores full marks on the first and zero on the second, and once that gap is visible the vague verdict "your argument is bad" sharpens into an actionable charge — "your premise is the conclusion in paraphrase; produce the independent ground." The label also redefines what the practitioner is owed: an argument is not a display of entailment among propositions the arguer already holds but a path the audience can walk from something it accepts to something it does not, and circular reasoning is precisely the case where start and end are the same node, so no such path exists.

Its second clarifying service is to keep the vicious circle distinct from the productive loops it superficially resembles. Tracing the dependency graph and asking whether any node returns to itself without an anchor outside the cycle is what separates begging the question from mathematical induction (whose loop has a base case), from the hermeneutic spiral (which is refined by fresh encounters with the text), and from legitimate coherentist webs (which remain open to external testing). The diagnostic question a practitioner can now pose is therefore not "is this argument circular?" in the loose sense of self-involving, but the precise one: "which node here is the audience required to grant without this argument supplying it?" — and if the honest answer is "none," the conclusion is being asserted, not argued.

Manages Complexity

The domain confronting an analyst of argument is a sprawl of superficially unlike defects, each cataloged separately and each arriving in its own surface costume: the apologetic argument that proves Scripture infallible from Scripture's own authority, the dictionary that defines anxiety as the state of being anxious, the precedent invoked to ground the very rule under which it was decided, the coherentist web whose beliefs hold each other up over an abyss, the source trusted on its own say-so. Treated case by case, these look like five different things to learn, five different inventories of bad reasoning to memorize — restated-premise circles, mutual-support circles, concept-loop circles, precedent-loop circles, authority-loop circles — multiplied across every subject matter (theology, law, lexicography, scientific methodology, epistemology) in which someone tries to show rather than assert. The combinatorial product of surface form and subject matter is unmanageable as a list.

Circular reasoning collapses that product to a single structural object and a single check. Every one of those defects is, underneath its vocabulary, a directed graph of justificatory dependencies in which a node depends — through however many intermediate steps — on itself; and the diagnostic that decides every case is one operation: trace the dependency edges from the contested conclusion and ask whether the path returns to its starting node without ever leaving the cycle to touch a premise the audience grants independently. The analyst no longer carries a taxonomy of fallacies; they carry one graph and one question — which node here is the audience required to grant without this argument supplying it? The parameters that determine the verdict shrink to two: whether the justification graph contains a cycle reachable from the conclusion, and whether that cycle has an anchor outside itself. From those two the qualitative outcome reads off directly. Cycle present, no external anchor: the argument is viciously circular, its validity intact and its justificatory force nil. Cycle present, anchor outside it: not vicious — this is what separates begging the question from mathematical induction (anchored by a base case), the hermeneutic spiral (anchored by fresh encounters with the text), and a legitimate coherentist web (anchored by openness to external testing), all of which contain loops yet escape the charge. No cycle: an ordinary argument, to be assessed on other grounds.

The branch structure is therefore small and exhaustive, and it makes the otherwise-blurred boundary between the vicious circle and the productive loop a matter of inspection rather than intuition. What looked like a heterogeneous field of named errors becomes a single property of a graph, computed the same way whether the substrate is a citation chain, a definitional pair, or a chain of theological premises — and the practitioner's labor drops from re-deriving the defect for each case to walking one dependency graph and checking for an unanchored return.

Abstract Reasoning

Circular reasoning licenses a set of moves in argument analysis, all anchored to the distinction between an inference's validity and its power to move someone who does not yet accept the conclusion. The signature diagnostic move is paraphrase-invariance testing: the analyst rewrites a premise in the conclusion's vocabulary (or the reverse) and asks whether the two propositions collapse into one — if rewording the premise obviously yields the conclusion, the argument is circular regardless of how different the original surface forms looked. This is what catches the defect when it is dressed up: "the Bible is God's Word" reduced under the inerrancy premise is "the Bible is infallible," so the apologetic syllogism is exposed as restating its conclusion even though no word is literally repeated. A companion diagnostic walks the dependency chain for a hidden return: because a long multi-step argument can conceal a short cycle, the analyst traces the justificatory edges and checks whether any node reappears in the role of justifier — a citation chain in which a doctrine rests on cases each citing the others reveals its circularity only when the chain is actually walked, not when its length is admired.

The most consequential interventionist move is demanding the exit point. Because the structural requirement is that justification terminate in something the audience grants independently, the analyst confronts the arguer with a single sharp request — "which node in your argument is your audience required to accept without this argument supplying it?" — and reads the answer as the verdict. If the arguer can name an external anchor and exhibit it, the circle is broken and the argument does real dialectical work; if the honest answer is "none," the diagnosis is fixed and so is the remedy: either produce the independent ground or concede that the conclusion is being asserted rather than argued for. The predicted effect of the demand is to convert a vague "your argument is bad" into a precise, dischargeable obligation, and to force the arguer to either supply traction from outside the cycle or stop claiming to have proved anything.

The boundary-drawing move is the one that gives the concept its discipline, because not every loop is vicious and the analyst must avoid over-firing. The test is anchoring: a cycle is a defect only when it has no anchor outside itself, so the analyst sorts the begged question from the productive loops it superficially resembles by locating each one's external attachment — mathematical induction's base case, the hermeneutic spiral's fresh encounters with the text, a legitimate coherentist web's openness to external testing. Where such an anchor exists, the loop is not vicious and the charge is withdrawn; where it is absent, the loop floats free and the charge stands. This same vigilance draws the regime edges against the concept's nearest neighbors: against infinite regress, the opposite failure in which the chain never terminates rather than returning to itself; against feedback and self-fulfilling prophecy, which are substantive causal loops in dynamic systems rather than defects in a static justification structure; and against productive self-reference, where a system's pointing at itself is generative rather than question-begging. The analyst's labor in every case is the same — find the anchor, or its absence — and the qualitative ruling, vicious circle or legitimate loop, reads off that single feature.

Knowledge Transfer

Within argumentation — across every subfield where someone tries to show a claim rather than assert it — circular reasoning transfers as mechanism, with the dependency-graph trace, the paraphrase-invariance test, and the demand-the-exit-point remedy carrying intact regardless of subject matter. In informal logic and rhetoric it is the canonical petitio principii, inventoried since Aristotle's Topics. In theological apologetics the diagnosis fires on the inerrancy syllogism ("Scripture is true because Scripture says so") by the same move that, in legal reasoning, exposes a doctrine bootstrapped through a chain of decisions each citing the others, or a precedent invoked to ground the rule under which it was decided. In lexicography and conceptual analysis it names the definitional pair (anxiety / anxious) as a two-node cycle; in scientific-methodology critique it flags the post-hoc explanation whose only test is the data it was fit to; in epistemology it is coherentism's standing worry that mutual support can close into a circle floating free of any anchor. Across all of these the substrate changes — a citation chain, a definitional pair, a chain of premises — but the analyst carries one object (a justification graph) and one question (which node must the audience grant without this argument supplying it?), and the vocabulary of validity-versus-justificatory-force travels with it. That is genuine mechanism transfer, not analogy, because the audience-facing justificatory task is the same task in each subfield.

Beyond argumentation the honest account is shared abstract mechanism, not metaphor — and this is the unusual case where the cross-domain instances are literal, not borrowed. The portable object is the parent structure: a directed cycle in a dependency graph with no anchor outside the cycle, the violation of acyclic-and-well-founded justification topology. That structure recurs as genuine co-instances in substrates that have nothing rhetorical about them — build-system cycle detection (Make, Bazel, Cargo flagging a circular import), legal-citation cycle detection over case-law DAGs, ontology and terminology design avoiding definitional loops via conceptual primitives, deadlock as a cyclic resource-wait. These are not "circular reasoning by analogy"; they instantiate the same acyclicity property, walked by the same graph algorithm. But what travels is the parent, not this entry's named machinery. Circular reasoning's own defining cargo is the part that stays home-bound: the normative, audience-dependent force that makes an unanchored cycle a fallacy rather than merely a graph property. A circular dependency in software is a build error the compiler reports; begging the question is a dialectical defect that only exists relative to a hearer who does not yet accept the conclusion and is owed a reason from outside the cycle. Strip "premise," "conclusion," and "audience" and the construct dissolves into "a cycle in a dependency graph" — which belongs to acyclicity / well_foundedness / dependency, not to rhetoric. So the cross-domain lesson should be carried by the parent acyclic-justification prime, of which circular reasoning is the argumentation-substrate manifestation alongside deadlock, circular dependency, and infinite recursion without a base case; the named fallacy itself does not and need not travel past the substrate that supplies a reasoning audience (see Structural Core vs. Domain Accent).

Examples

Canonical

The textbook petitio principii is the scriptural-inerrancy syllogism. An apologist argues: (1) The Bible is the infallible Word of God; (2) we know this because the Bible declares itself to be God's Word (e.g., a verse asserting all scripture is God-breathed); therefore (3) the Bible is infallible. Walk the dependency graph. The conclusion — the Bible's infallibility — is what licenses trusting the Bible's self-testimony in premise (2); but premise (2) is the only support offered for the conclusion. The justification edge runs conclusion → premise (2) → conclusion, a closed loop. Apply the paraphrase-invariance test: "the Bible is a reliable authority about itself" reduces, under the very infallibility being argued, to "the Bible is infallible." The syllogism is deductively fine — grant the premises and the conclusion follows — yet it hands a doubter no reason from outside the loop.

Mapped back: The propositions and their support edges form the justification graph; "the Bible is infallible" is the conclusion node, and the self-testimony premise is the premise-equals-conclusion node once paraphrased. Their mutual dependence is the cycle, and there is the missing exit point — no proposition a skeptic grants independently of the argument. That the inference is valid yet powerless is the validity/justificatory-force split, and collapsing the two premises by rewording is the paraphrase-invariance test.

Applied / In Practice

Philosophers of science deploy this diagnosis to flag vacuous "explanations," the classic being Molière's satire in Le Malade imaginaire (1673): asked why opium induces sleep, a physician answers that it possesses a virtus dormitiva — a "dormitive power." The explanation is a two-node circle: "opium causes sleep" is offered as explained by "opium has a sleep-causing power," which is merely the phenomenon restated in Latin dress. Working scientists and methodologists invoke this case to police real theory: a trait "explained" by positing a faculty for it, or "survival of the fittest" glossed as "those that survive are those that survive," fails the same check — it names no independent mechanism (dose-response, receptor binding) that a skeptic could grant apart from the phenomenon being explained.

Mapped back: The explanatory move forms the justification graph whose conclusion node ("opium causes sleep") is supported only by the premise-equals-conclusion node ("opium has dormitive power"), which the paraphrase-invariance test collapses back into the phenomenon. The dependence loops with the missing exit point: no independent mechanism is cited. The methodologist's corrective — demand the external anchor (a real physiological pathway) — is precisely the "which node must be granted from outside?" exit-point demand, and finding a genuine mechanism is the anchoring boundary that would convert a vacuous circle into a real explanation.

Structural Tensions

T1: Impeccable validity versus zero force (the boundary with ordinary deduction). A circular argument's deepest peculiarity is that it is deductively flawless — grant the premises and the conclusion follows with certainty — yet accomplishes nothing, because the certainty comes from having assumed what was to be shown. This exposes a boundary the concept must police but cannot draw sharply: every valid deductive argument is, in the information-theoretic sense, contained in its premises; the conclusion of a sound syllogism adds no content not already there. So "the conclusion appears among the premises" cannot by itself be the defect — it is true of all deduction. Circularity is a matter of proximity and transparency: how directly, and how obviously to the audience, the premise restates the conclusion. The tension is that the same feature which makes deduction reliable (conclusion guaranteed by premises) is the feature that, taken to its limit, makes an argument viciously circular, and there is no clean line between "validly draws out" and "merely restates." Diagnostic: Does the premise deliver the conclusion by a step the audience would find informative, or does it restate the conclusion so directly that granting the premise already is granting the conclusion?

T2: Structural graph property versus audience-relativity (circular to whom). The diagnosis is framed as a property of a justification graph — trace the edges, find the unanchored cycle — which sounds objective and arguer-independent. But the anchoring test is defined by what the audience grants independently, so the very same argument is viciously circular before a skeptic who withholds the premise and perfectly cogent before a hearer who already grants it. Circularity is therefore not a fixed feature of the argument but a relation between the argument and a particular audience's prior commitments. The tension is that the concept's rhetoric of graph-tracing promises an objective verdict while its actual criterion is irreducibly dialectical and relative to who is being addressed. An argument can be a fallacy in one room and a sound reminder in the next, with no change to a single proposition. Diagnostic: Relative to this audience's independent commitments, is the anchor node one they grant apart from the argument — or is "begging the question" a verdict that would flip for a differently-positioned hearer?

T3: The paraphrase-invariance test versus contested synonymy (when are two propositions one node). The sharpest diagnostic — reword the premise into the conclusion's vocabulary and see whether they collapse — presupposes a decidable notion of when two differently-worded propositions are "the same node." But sameness of meaning is exactly what philosophy finds hardest to pin down: legitimate conceptual analysis and informative identity claims ("water is H2O") also pair propositions that turn out equivalent, and a hostile analyst can collapse any explication into "restatement" while a lax one can miss a circle hidden behind a thick synonym. The tension is that the test's power to unmask disguised circles is the same power that lets it over-fire on genuine clarifications, and both errors turn on an undecidable question of propositional identity. Wield it aggressively and you condemn legitimate definition; wield it cautiously and disguised circles pass. Diagnostic: Do premise and conclusion collapse into one proposition under any reasonable paraphrase, or does the rewording that fuses them smuggle in the equivalence it claims to detect?

T4: The anchoring requirement versus coherentism (does demanding an external anchor beg its own question). The concept's discipline against over-firing is the anchoring test: a loop is vicious only if it lacks an anchor outside the cycle. This cleanly acquits induction (base case) and the hermeneutic spiral (fresh text). But applied to a coherentist web, the test quietly assumes foundationalism: coherentism's whole thesis is that justification can be holistic and mutually supporting without terminating in any externally-granted foundational node. To declare an unanchored coherentist web viciously circular because it lacks an external anchor is to presuppose that justification must be linear and well-founded — the very point coherentism denies. The tension is that the anchoring boundary, presented as neutral, encodes a substantive epistemological commitment, so the diagnosis of circularity can itself beg the question against a rival theory of justification. Diagnostic: Is the "missing anchor" a genuine gap the arguer's own standards require filled, or is it missing only relative to a foundationalist demand the arguer's coherentist framework explicitly rejects?

T5: The exit-point demand versus the regress trilemma (fixing one horn drives you to another). The remedy — "name the node your audience must grant without this argument supplying it" — treats the circle as fixable by producing an external anchor. But the anchor is a premise, and a skeptic can demand a reason for it, and for that reason in turn. Escaping circularity by supplying ever-more-basic grounds is the road to infinite regress; stopping at a ground asserted without further argument is dogmatism. This is the Agrippan/Münchhausen trilemma: circular, regressive, or dogmatic — and the concept's own remedy for the first horn pushes the arguer onto one of the other two. The tension is that "just provide the independent ground" is not a clean fix but a relocation of the difficulty: the demand for an exit point, pressed consistently, cannot terminate without at some point granting something unargued. Circularity is one escape from the trilemma, not simply an error the other escapes avoid. Diagnostic: Does the demanded anchor genuinely rest on ground the audience already grants (fix), or does insisting on it merely trade the circle for an unterminating regress or a bare assertion?

T6: Autonomy versus reduction (a dialectical fallacy or the acyclicity parent). "Circular reasoning" is a canonical fallacy with real in-situ machinery — the justification-graph trace, the paraphrase-invariance test, the exit-point demand — and within argumentation it transfers as mechanism across theology, law, lexicography, and epistemology unchanged. But the portable object is the parent: an unanchored directed cycle in a dependency graph, the violation of acyclic, well-founded justification topology, which recurs literally — not by analogy — as build-system import cycles, citation-graph loops, and resource-wait deadlock, all walked by the same algorithm. What stays home-bound is precisely what makes an unanchored cycle a fallacy rather than a mere graph property: the normative, audience-dependent force by which a hearer is owed a reason from outside the loop. A compiler reports a circular dependency; only a reasoning audience can be begged a question. The tension is between a named fallacy that earns its dialectical study and the recognition that its transferable structure belongs to acyclicity / well_foundedness / dependency. Diagnostic: Resolve toward the acyclic-justification parent (acyclicity, well_foundedness) when the cycle is in a dependency graph with no reasoning audience; toward named circular reasoning when an unanchored cycle denies a hearer the independent reason they are owed.

Structural–Framed Character

Circular reasoning sits at the framed pole of the structural–framed spectrum, alongside the other named fallacies: it is not a regularity a system exhibits but a verdict rendered on a piece of reasoning within a dialectical practice. On evaluative_weight it scores high — to call an argument "circular" or "question-begging" is to convict it of a justificatory failure, a normative finding that the argument does no dialectical work, not a neutral description of a mechanism the way "feedback" names something neither sound nor unsound. Human-practice-bound is high in the strongest sense the entry itself insists on: the defect exists only relative to a hearer who does not yet accept the conclusion and is owed a reason from outside the loop; strip away premise, conclusion, and audience and nothing remains but a graph property — "a cycle in a dependency graph" — with no one of whom to beg the question. Institutional_origin is equally pronounced: the concept is furniture of a specific tradition — the Latin petitio principii, Aristotle's Topics, the standard fallacy catalogues, the surface-form taxonomy (restated-premise, mutual-support, concept-loop, precedent-loop, authority-loop), and the anchoring jurisprudence that separates the vicious circle from induction, the hermeneutic spiral, and coherentist webs — all distinctions drawn inside the theory of argument, not substrate-neutral form. On vocab_travels the operative vocabulary (begging the question, justificatory force, the exit point the audience must grant) is pinned to the argumentation substrate and loses its referents off it. Import_vs_recognize is the subtle criterion here, and the entry is unusually candid about it: the cross-domain instances (build-system import cycles, citation-graph loops, resource-wait deadlock) are literal co-instances, recognized rather than borrowed — but of the parent acyclicity property, not of the fallacy; what would have to be imported by analogy to carry "circular reasoning" past its substrate is precisely the normative, audience-owed force, which does not travel.

The portable structural skeleton is acyclicity / well_foundedness — an unanchored directed cycle in a dependency graph, a justification chain that returns to its start without ever touching a node the audience grants independently of it. That skeleton is genuinely substrate-spanning and recurs off-domain without metaphor, which is exactly what tempts a structural reading. But it does not pull circular reasoning off the framed pole, because that acyclic-justification structure is what circular reasoning instantiates from its parent, not what makes "circular reasoning" itself travel: the cross-domain reach — deadlock, the compiler's circular-import error — belongs to the general acyclicity pattern, while the fallacy's distinctive content (the begging of a question a reasoning audience is owed a reason to settle) is the part that stays home-bound. Its character: a normatively charged, audience-constituted fallacy verdict whose only substrate-spanning content is the unanchored-cycle skeleton it borrows from acyclicity and reframes as a dialectical failure owed to a hearer.

Structural Core vs. Domain Accent

This section decides why circular reasoning is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — there is no separate section for that.

What is skeletal (could lift toward a cross-domain prime). Strip the argumentation and a thin relational structure survives: a directed dependency graph in which a node reaches itself through a chain of justificatory edges that never touches a node granted from outside the loop — an unanchored cycle in a well-founded ordering. The portable pieces are abstract: a set of elements each depending on others, edges that should terminate in something not supplied by the chain, and the closed return that leaves the terminus unmet. That skeleton is acyclicity / well_foundedness (with dependency supplying the graph), and it is genuinely substrate-portable — which is exactly why the entry insists its cross-domain instances are literal co-instances, not metaphors: build-system import cycles, citation-graph loops, and resource-wait deadlock all instantiate the same graph property, walked by the same algorithm. But it is the core circular reasoning shares, not what makes it distinctive.

What is domain-bound. Almost everything that makes the construct begging the question in particular is argumentation-theory furniture, and none of it survives extraction: the petitio principii label and its Aristotelian lineage; the surface-form taxonomy (restated-premise, mutual-support, concept-loop, precedent-loop, authority-loop); the paraphrase-invariance test that collapses a disguised premise into the conclusion; the anchoring jurisprudence that distinguishes the vicious circle from induction's base case, the hermeneutic spiral, and a coherentist web; and above all the normative, audience-relative force by which a hearer who does not yet accept the conclusion is owed a reason from outside the loop. The decisive test: remove premise, conclusion, and audience and "a cycle in a dependency graph" remains — a build error a compiler reports, not a fallacy, because there is no longer anyone of whom to beg the question. The fallacy is constituted by the very dialectical context the prime bar asks it to shed.

Why this does not clear the prime bar. A prime's vocabulary travels and its cross-domain transfer is recognition of the same mechanism, not analogy. Circular reasoning's transfer is bimodal — and unusually so. Within argumentation the diagnosis travels intact across theology, law, lexicography, scientific-methodology critique, and epistemology, because each supplies the one thing it needs: a reasoning audience owed an independent reason. Beyond argumentation the cross-domain instances are literal, but they are instances of the parent, not of the fallacy: a compiler's circular-import error and a resource-wait deadlock are the acyclicity property recurring with no hearer at all, so what transfers is the graph structure, while the begging — the audience-owed normativity that makes an unanchored cycle a defect of justification — stays home-bound. When the bare structural lesson is needed cross-domain it is already carried, in more general form, by acyclicity / well_foundedness (with dependency), of which circular reasoning is the argumentation-substrate manifestation alongside deadlock and circular dependency. The cross-domain reach belongs to the parents; the named fallacy carries dialectical baggage that should stay home.

Relationships to Other Abstractions

Local relationship map for Circular ReasoningParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Circular ReasoningDOMAINPrime abstraction: Cycle — is a decomposition ofCyclePRIME

Current abstraction Circular Reasoning Domain-specific

Parents (1) — more general patterns this builds on

  • Circular Reasoning is a decomposition of Cycle Prime

    Removing the argumentative frame leaves a directed dependency path that returns to its origin without an external anchor.

Hierarchy path (1) — routes to 1 parentless root

Not to Be Confused With

  • Infinite regress. The opposite justificatory failure: a chain in which every premise leans on a further premise, so the demand for support never terminates — as against the circle, which terminates at its own starting node. Both leave the conclusion without a secure anchor, and the exit-point demand pushes an arguer escaping one straight into the other (the Agrippan trilemma). But the topology is mirror-image: circularity closes back on its start, regress never ends. Tell: walk the dependency chain — does it return to a node already visited (circular) or generate a fresh, never-satisfied demand at every step (regress)?

  • Tautology / analytic truth. A single proposition true by its own form or meaning ("all bachelors are unmarried"). A tautology is a harmless — often trivially correct — statement; circular reasoning is a defective argument, a premise-to-conclusion move whose premise is the conclusion in paraphrase. One is a property of a lone sentence, the other a property of an inferential structure with an audience owed a reason. Tell: is there just one self-true claim on the table (tautology), or a premise offered in support of a distinct-looking conclusion that collapses into it under paraphrase (circular reasoning)?

  • Complex question / loaded question. The fallacy of a single question that presupposes an unestablished claim ("have you stopped cheating?") — and the concept the phrase "begging the question" is most often popularly misapplied to. A complex question smuggles an assumption into a query; circular reasoning smuggles the conclusion into the premises of an argument. The presupposition-loading is a distinct defect (it has its own entry). Tell: is the unwarranted assumption riding inside an interrogative that traps any answer (complex question), or inside a premise that already contains the conclusion (circular reasoning)?

  • Self-fulfilling prophecy and causal feedback. Substantive causal loops in dynamic systems — a prediction that helps bring about the state it names, or an output that feeds back to shape its own input. These are loops in the world, and reasoning about such a loop can be perfectly sound. Circular reasoning is a defect in a static justification structure, a loop among propositions in the argument, not among events in a system. Tell: does the loop run through real states of a system unfolding over time (causal), or through the justificatory dependencies of an argument at rest (circular reasoning)?

  • Legitimate coherentism / anchored productive loops. A coherentist web, mathematical induction, and the hermeneutic spiral all contain loops yet escape the charge, because each is anchored outside the cycle — a base case, openness to external testing, fresh encounters with the text. These are contrast cases: loop-bearing but not vicious. Circular reasoning is the loop with no external anchor. (The entry flags that pressing this anchoring test on a coherentist web can itself beg the question against coherentism.) Tell: name the node the audience must grant without this argument supplying it — if one exists, the loop is anchored and legitimate; if the honest answer is "none," it is vicious.

  • Acyclicity / well-foundedness / dependency (the parent). The substrate-neutral structure circular reasoning instantiates — an unanchored directed cycle in a dependency graph — which recurs literally, not by analogy, as build-system import cycles, citation-graph loops, and resource-wait deadlock. This is the umbrella, not a confusable peer: it carries the cross-domain reach, while the fallacy adds only the normative, audience-owed force that makes an unanchored cycle a begging of the question rather than a mere graph property. Tell: is there a reasoning audience owed an independent reason (the fallacy), or just a cycle a compiler or scheduler would flag with no hearer at all (the parent)?

Neighborhood in Abstraction Space

Circular Reasoning sits in a crowded region of the domain-specific corpus (30th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Fallacious Substitution in Argument (17 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12