Epistemic Mode Of A Proposition¶
Core Idea¶
The epistemic mode of a proposition is the structural pattern by which one and the same proposition can be held in different ways — as an asserted fact, an assumption, an axiom, a hypothesis, a premise, a belief, or a conjecture — and the way it is held, independent of what it says, gates which operations are licensed on it. The decisive observation is that the mode is a dimension orthogonal to the content. The sentence "the bridge bears 40 tonnes" is one proposition; it is an entirely different object depending on whether it is asserted as a verified fact, assumed for the sake of a calculation, posed as a hypothesis to be tested against load data, taken as a premise in a safety argument, or held as a belief by an engineer. The proposition is identical in all five; the mode differs, and the mode is what determines the permissible moves: you may provisionally rely on an assumption while flagging that the conclusion is conditional on it; you must test a hypothesis rather than rely on it; you cannot refute an axiom from within the system that posits it (only change the system); you must eventually discharge a premise to convert a conditional conclusion into a categorical one; you should update a belief on evidence; you treat a datum as given but subject to measurement error; you hold a conjecture as a target for proof or disproof. The defining commitments are four. First, there is a proposition — a truth-apt content that could be true or false. Second, there is a mode attached to it — a label drawn from a small typed vocabulary (asserted-fact, assumption, axiom, hypothesis, premise, belief, conjecture, datum, stipulation) that records how the content is being held. Third, the mode is content-invariant: the same proposition can carry any mode, and changing the mode changes nothing about what the proposition says. Fourth, and load-bearing, the mode gates the licensed operations: each mode comes with a characteristic set of moves that are permitted, required, and forbidden on the proposition, and a reasoning error results from applying a move licensed by one mode to a proposition that is actually in another.
The structural signature is a typed handle that travels with a proposition and determines its operation set. It is what lets a reasoner answer, of any claim on the table, not "is this true?" but the prior question "in what mode is this being held, and therefore what am I allowed to do with it?" The single most consequential fact the prime names is that the mode, not the content, fixes the operations — which is why the same sentence can be untouchable in one mode (an axiom one cannot refute from inside the system) and mandatory-to-attack in another (a conjecture one is obliged to try to disprove), and why the most common and damaging reasoning errors are mode errors: treating an assumption as an established fact (relying on it without flagging the conditionality), treating a hypothesis as a belief (defending it instead of testing it), treating a stipulation as a discovery, or treating an axiom as an empirical claim open to falsification. From this single fact follow the prime's downstream properties. Modes are re-assignable: the central operations of inquiry are mode transitions — a hypothesis confirmed becomes an asserted fact, an assumption justified becomes a datum, a conjecture proved becomes a theorem, a premise discharged is converted into a conditional — and tracking these transitions is what disciplined reasoning is. Modes compose with the dependency structure: a conclusion inherits the weakest mode among the propositions it rests on, so a result resting on an untested assumption is itself only as secure as that assumption, no matter how rigorous the inference. And the mode discipline is substrate-general: the same typed-handle-gates-operations structure appears in formal logic, empirical science, law, and programming, each with its own mode vocabulary but the identical underlying move. What epistemic_mode provides as a prime is the recognition that "how a proposition is held" is a first-class structural feature, separable from the content, that silently governs every legitimate operation on the claim.
This is a candidate proposed at lower confidence, so it is worth stating directly why it clears the bar for a distinct prime rather than collapsing into its neighbors. The pattern earns independent standing on three grounds. First, it is not any single mode: assumption, axiom, and hypothesis are each one value of the mode dimension; this prime is the dimension itself — the recognition that mode is a content-orthogonal variable ranging over those values and gating operations accordingly — which no single-mode prime names, just as a prime for "color" is not subsumed by a prime for "red." Second, it is not the process of forming a belief: belief_formation names how an agent comes to hold a proposition true; this prime names the static typed status a proposition carries and the operation set that status licenses, which is a different object (a held belief is one mode among several, and the same proposition can be in a non-belief mode that no act of belief-formation produced). Third, the mode-gates-licensed-operations structure travels with the same shape across at least four independent disciplines — the axiom/theorem/conjecture typing of mathematics, the hypothesis/law/datum typing of science, the stipulated/alleged/adjudicated-fact typing of law, and the assert/assume/invariant typing of programming — and a structural pattern that recurs identically across logic, science, law, and code, gating what may be done to a claim in each, is exactly the kind of cross-substrate invariant the catalog exists to name. It is, candidly, the weakest of its drafting batch — its content overlaps heavily with the assumption family and its boundary against "the role a proposition plays" requires care — but the orthogonal-dimension-that-gates-operations reading is real, recurs widely, and is not captured by any existing prime.
How would you explain it like I'm…
How You Hold It
The Mode, Not The Message
Structural Signature¶
the truth-apt proposition (the content) — the mode label attached to it (the typed handle) — the content-invariance of the mode — the operation set the mode licenses, requires, and forbids — the re-assignability of modes through transitions of inquiry — the weakest-mode inheritance through the dependency structure
An epistemic mode is present when each of the following holds:
- A truth-apt proposition (the content). A claim with a content that could be true or false — the thing held, independent of how it is held.
- A mode label (the typed handle). A label drawn from a small typed vocabulary — asserted-fact, assumption, axiom, hypothesis, premise, belief, conjecture, datum, stipulation — recording how the content is currently being held.
- Content-invariance (the orthogonality invariant). The mode is independent of the content: the same proposition can carry any mode, and changing the mode changes nothing about what the proposition asserts. Mode is a dimension orthogonal to truth-value and meaning.
- A licensed operation set (the gating invariant). Each mode comes with a characteristic set of operations that are permitted, required, and forbidden — rely-on-provisionally for an assumption, test for a hypothesis, do-not-refute-from-within for an axiom, discharge for a premise, update-on-evidence for a belief. This is the load-bearing feature: the mode, not the content, fixes what may be done.
- Re-assignability (the transition invariant). Modes can be reassigned, and the core operations of inquiry are such transitions: hypothesis confirmed becomes fact, assumption justified becomes datum, conjecture proved becomes theorem, premise discharged becomes a conditional. Tracking the transitions is the discipline.
- Weakest-mode inheritance (the composition invariant). A conclusion inherits the weakest (least secure) mode among the propositions it rests on, so a result built on an untested assumption is itself only as secure as that assumption, regardless of inferential rigor.
The components compose into a single object — a content-orthogonal typed handle that travels with a proposition and gates the operations licensed on it — which is why the axiom/theorem/conjecture of mathematics, the hypothesis/law/datum of science, the alleged/adjudicated fact of law, and the assert/assume/invariant of code are recognized as the same structure: a mode dimension determining, in each, what may legitimately be done to a claim.
What It Is Not¶
- Not an assumption.
assumptionis one mode — a proposition held as if true, provisionally, for the purposes of some reasoning, licensing the operation "rely on while flagging conditionality." Epistemic mode is the dimension of which assumption is one value; it ranges over assumption, axiom, hypothesis, belief, conjecture, premise, datum, and stipulation, and its content is precisely that the choice among these gates the operations. Naming the dimension is not naming any of its values, just as naming "temperature" is not naming "hot." - Not an axiom.
axiomis another single mode — a proposition held as foundationally true within a formal system, licensing "use freely, do not refute from within." It is one value of the mode dimension, distinguished from assumption precisely by its mode (foundational-and-irrevisable versus provisional-and-revisable). The prime is what makes "axiom versus assumption" a difference in mode at all. - Not belief formation.
belief_formationis the process by which an agent comes to hold a proposition true and act accordingly. Epistemic mode is the static typed status a proposition carries and the operation set that status licenses. A held belief is one mode among several; the prime concerns the whole typing, including modes (an assumption one disbelieves, an axiom one merely stipulates) that no act of belief-formation produces. - Not the truth-value of the proposition. Truth-value is whether the content is in fact true or false; mode is how it is being held, which is orthogonal. A false proposition can be held as an assumption, a true one as a mere conjecture; the mode neither asserts nor settles the truth-value, it only fixes what operations are licensed pending the truth-value's determination.
- Not the proposition's content or its logical role in a single argument. The content is what the proposition says; mode is independent of it. And mode is not merely the local role a proposition plays within one inference (antecedent, conclusion) — it is a standing epistemic status the proposition carries across a body of reasoning, governing operations beyond any single argument's structure.
- Not a confidence level or probability. A credence (0.7 probable) is a graded quantity on a single mode (belief); mode is a categorical type that changes the kind of operation licensed, not its degree. A hypothesis at credence 0.9 still must be tested, not relied on, because its mode — not its credence — forbids reliance; raising the number does not change the mode or its operation set.
- Common misclassification. Treating a proposition's mode as fixed by its content or its source, and so applying a move licensed by the wrong mode. Catch it by asking, of any claim on the table, "in what mode is this being held — and is the operation I am about to perform one this mode licenses?": surprising reasoning failures concentrate in mode errors (relying on an assumption as a fact, defending a hypothesis as a belief, falsifying an axiom as if it were empirical), not in the content itself.
Broad Use¶
The epistemic-mode structure recurs wherever propositions are held and operated on under a discipline that distinguishes how they are held. In logic and mathematics, its origin substrate, the mode vocabulary is sharply typed and the gating is explicit: an axiom is used freely but cannot be refuted from within its system, a theorem is an asserted result that has been proved, a conjecture is a target held open for proof or disproof, a premise under conditional proof is discharged to convert a conditional into a categorical, and a lemma is a proved waypoint — and the whole architecture of proof is a discipline of mode transitions (conjecture → theorem, premise → discharged conditional). In empirical science, the modes are hypothesis (must be tested, never merely relied on), law or established theory (asserted, but provisionally and open to revision), datum (taken as given but carrying measurement error), and auxiliary assumption (relied on while flagged), and the central scientific operations — confirmation, falsification, replication — are mode transitions that move a proposition from hypothesis toward established fact or back to refuted. In law, the gating is institutionalized: a stipulated fact is agreed and not contested, an alleged fact must be proved to a standard, an adjudicated fact has been established by a tribunal, a presumption is held true until rebutted, and a legal fiction is treated as true for a purpose while known false — each licensing different operations (an alleged fact may be challenged; a stipulated one may not), and the trial is a machine for transitioning facts from alleged to adjudicated. In programming and formal methods, the modes are explicit in the language: an assert checks a proposition that should hold and halts if it does not, an assume (in verification) is relied on without checking, an invariant is a proposition required to hold at every iteration, a precondition is a premise the caller must discharge, and a postcondition is an asserted guarantee — and a verification tool's soundness depends on never confusing an assume (taken on trust) with an assert (checked). In everyday and professional reasoning, the modes are looser but real: a working assumption, a hunch, a settled fact, a tentative hypothesis, a stipulated ground rule, and good reasoning is the practice of keeping track of which is which. Across all of these the recurring fact is identical: a proposition carries a content-orthogonal mode that determines what operations are legitimate on it, and the recurring failure is the mode error — operating on a claim as if it were in a mode it is not.
Clarity¶
Naming the epistemic mode of a proposition separates two questions that reasoners routinely fuse: what does this claim say (and is it true)? and how is this claim being held (and therefore what may I do with it)? Ordinary attention runs straight to the content and its truth, treating "how it is held" as either obvious or irrelevant; the prime makes the mode a first-class, separable feature and insists it be read before any operation is performed. The clarifying force is to convert a vague unease — "something about how we're using this claim is off" — into the precise question "which mode is this in, and is the move I am making licensed by that mode?" This exposes a whole class of errors as mode errors rather than content errors. A team that relies on an untested assumption as though it were an established fact has not made a factual mistake — the content may be fine — they have made a mode error, operating in fact-mode on a proposition that is in assumption-mode, and the fix is not to re-examine the content but to either flag the conditionality or transition the assumption to a datum by justifying it. A scientist who defends a hypothesis instead of testing it, a lawyer who treats an alleged fact as adjudicated, a programmer who assumes what should have been asserted — each has applied an operation from the wrong mode's license. The prime also clarifies what disciplined inquiry is: not the accumulation of true contents but the correct management of modes and their transitions — knowing when a proposition has earned the right to move from hypothesis to fact, from conjecture to theorem, from alleged to adjudicated, and refusing to operate on it as though it had made that transition before it has. By relocating the discussion from the content to the mode, the prime makes visible the typed handle that was silently governing every legitimate move all along.
Manages Complexity¶
The epistemic-mode structure compresses the open-ended question "what is the standing of every claim in this body of reasoning, and what may I do with each?" into a small typed vocabulary and a fixed table of licensed operations. Rather than reasoning case-by-case about how to treat each proposition, the reasoner tags each with a mode and reads off the permitted, required, and forbidden operations from the mode alone — relying-but-flagging an assumption, testing a hypothesis, using-but-not-refuting an axiom, discharging a premise. This is a substantial reduction because it replaces a per-proposition judgment with a per-type rule: there are only a handful of modes, each with a stable operation set, so the management of arbitrarily many propositions reduces to assigning each a type and honoring the type's gating. The compression also makes the security of a conclusion computable from the modes of its inputs via the weakest-mode-inheritance rule: a conclusion is only as secure as the least-secure mode among the propositions it rests on, so an analyst can locate a conclusion's true epistemic standing by finding the weakest link in its support — a chain resting on one untested assumption is an untested-assumption-grade conclusion, however many proved theorems surround it. This converts the diffuse worry "how much should I trust this result?" into the bounded task "what is the weakest mode in its dependency set?" And it organizes the work of inquiry as a small set of mode transitions with known entry conditions: to strengthen a conclusion, find its weakest-moded support and do the work that transitions it (test the hypothesis, justify the assumption, prove the conjecture, discharge the premise). Because the modes are few, their operation sets fixed, and their composition rule simple, the prime keeps the otherwise-unbounded task of tracking the epistemic standing of a complex argument within reach — the complexity of "what may I do with each of these claims?" reduced to a typed handle and a lookup table.
Abstract Reasoning¶
The epistemic-mode pattern licenses several substrate-portable moves. Read the mode before operating: before performing any move on a proposition — relying on it, testing it, refuting it, building on it — the reasoner first asks what mode it is in, because the mode, not the content, determines whether the move is licensed; this single discipline prevents the entire family of mode errors. Detect the mode error: when a reasoning process goes wrong despite sound content, the reasoner suspects that an operation from one mode's license was applied to a proposition in another — an assumption relied on as a fact, a hypothesis defended as a belief, an axiom attacked as empirical — and locates the fault in the mismatch rather than re-auditing the content. Compute the conclusion's grade by weakest-mode inheritance: to assess how secure a result is, the reasoner finds the least-secure mode among its supporting propositions, because the conclusion can be no more secure than its weakest-moded input, no matter how rigorous the inference connecting them. Strengthen by transition, not by re-derivation: to make a conclusion more secure, the reasoner does not re-run the inference but identifies the weakest-moded support and performs the operation that transitions it upward — testing the hypothesis, justifying the assumption, proving the conjecture, discharging the premise — because the conclusion's grade tracks its inputs' modes, not the inference's length. And separate mode from credence: the reasoner distinguishes the categorical kind of holding (mode) from the graded degree of confidence (credence), recognizing that a high credence does not change a hypothesis's mode or license reliance on it — the operation set is fixed by the mode, and a 0.9-probable hypothesis must still be tested rather than relied on. Each move follows directly from the bare structure: because the mode is a content-orthogonal handle that gates operations, the legitimacy of any move is a function of the mode, the security of any conclusion is a function of its inputs' modes, and the work of inquiry is the work of effecting mode transitions.
Knowledge Transfer¶
Because the epistemic mode of a proposition is the bare structural relation of a content-orthogonal typed handle that gates operations, a discipline built around it in one field transfers to any other by re-identifying the field's mode vocabulary and the operation set each mode licenses — and the prime's reach is the reach of that one structure across logic, science, law, and code. The read-the-mode-before-operating discipline transfers verbatim: a mathematician who will not refute an axiom from within its system, a scientist who will not rely on an untested hypothesis, a lawyer who will not treat an alleged fact as adjudicated, and a verification engineer who will not confuse an assume with an assert are all running the identical move — check the mode, then perform only the operations it licenses — distinguished only by each field's mode names. The weakest-mode-inheritance rule transfers as a universal grading principle: a physicist locating the auxiliary assumption that makes an experimental conclusion only as secure as that assumption, a mathematician noting that a theorem resting on an unproved conjecture is itself only conjectural, and a software team recognizing that a verified module is only as sound as the assumes it was checked against are all computing a conclusion's grade from the weakest mode in its support. The strengthen-by-transition move transfers as the shared logic of inquiry: confirming a hypothesis (science), proving a conjecture (mathematics), adjudicating an alleged fact (law), and discharging a verification assumption by checking it (programming) are the same operation — moving a proposition's weakest-moded support up the mode hierarchy to raise the conclusion's grade. And the mode-error diagnostic transfers as a portable failure analysis: a surprising breakdown that survives a content audit is, across all four substrates, a candidate mode error — an operation applied from the wrong mode's license — and the repair is to correct the mode, not the content. In every transfer the practitioner runs the identical diagnosis: identify the proposition, read its mode from the field's vocabulary, confirm that the operation being performed is one the mode licenses, and trace a conclusion's grade to the weakest mode in its dependency set. The transfer is secure because none of these steps names a particular field's content — a logician, an experimentalist, a litigator, and a programmer are reasoning about the same content-orthogonal typed handle, distinguished only by the mode vocabulary their discipline uses and the operations it attaches to each mode. What does not transfer is any application outside an inferential practice: there is no epistemic mode where there is no reasoner or formal system to track the mode and honor its gating, because the pattern is constitutively about how a proposition is held within some reasoning, not about any mute physical process.
Examples¶
Formal/abstract¶
Conditional proof in formal logic is the prime in its purest worked form, because the mode and its gating are fully explicit and a mode transition is the whole point. To prove a conditional \(P \to Q\), the reasoner introduces \(P\) as a premise — and this is a precise mode assignment: \(P\) is held in premise-mode, which licenses the operations "use \(P\) freely within the subproof" and "must eventually discharge \(P\)," and forbids "treat \(P\) as categorically established outside the subproof." The same proposition \(P\) could elsewhere be an axiom (used freely, never discharged), a hypothesis (to be tested), or an asserted theorem (proved and freely usable) — the content-invariance is exact: \(P\) says the same thing in every mode, and only the mode differs. Within the subproof the reasoner derives \(Q\) from \(P\) (plus the system's axioms, themselves in axiom-mode: usable, non-refutable from within). Then comes the transition that is conditional proof's entire move: discharging the premise \(P\) converts the locally-derived \(Q\), which depended on \(P\), into the categorical conditional \(P \to Q\), which depends on \(P\) no longer. This is a mode operation, licensed specifically by premise-mode, that no other mode permits — one cannot "discharge" an axiom or a datum. The weakest-mode inheritance is visible too: inside the subproof, \(Q\) is only as secure as the assumed \(P\) (premise-grade, conditional); after discharge, \(P \to Q\) inherits the secure mode of a proved result. The gating is what makes the proof valid: were the reasoner to treat \(P\) as an asserted fact rather than a premise — using \(Q\) categorically without discharging \(P\) — the proof would be fallacious, a textbook mode error. The structural payoff is that the legitimacy of every step is fixed by the modes of the propositions, not their contents: discharge is licensed by premise-mode, free use by axiom-mode, and the conditional conclusion is exactly the record of a premise introduced and discharged.
Mapped back: Conditional proof instantiates every component — a truth-apt proposition \(P\), an explicit mode label (premise), the content-invariance of that label, the operation set premise-mode licenses (use-then-discharge), the transition (discharge converting conditional to categorical), and weakest-mode inheritance (\(Q\) conditional until discharge) — and shows the prime's core claim: the mode, not the content, gates which operations are valid.
Applied/industry¶
A formal-verification pipeline for safety-critical software runs the identical structure in a programming substrate, with no logic-textbook vocabulary, and the soundness of the whole effort depends on honoring the gating. The codebase carries propositions in sharply different modes, marked by distinct language constructs. An assert(x > 0) places the proposition "x > 0" in assert-mode: it must be checked, and the operation it licenses is "verify, and fail if false." An assume(input is sanitized) in the verification harness places its proposition in assume-mode: it is relied on without checking, licensing "use freely as a given, do not verify" — and crucially the same proposition "input is sanitized" could appear elsewhere as an assert (to be checked) or as a precondition (a premise the caller must discharge). The content-invariance is exact and consequential: "input is sanitized" says the same thing in every mode, but an assume is taken on trust while an assert is checked, so the verifier's soundness depends entirely on the mode, not the content. A loop invariant places its proposition in invariant-mode: it is required to hold at every iteration, licensing the inductive operations "show it holds initially" and "show each iteration preserves it." The weakest-mode inheritance is the pipeline's central safety fact: a module "proved correct" is only as sound as the assumes it was verified against — its guarantee inherits the assume-grade of its weakest premise — so a verified component resting on an unchecked assumption carries exactly that assumption's risk, however rigorous its proof. The catastrophic mode error is concrete: writing assume where one needed assert — relying on a proposition that should have been checked — produces a "verified" system that is sound only if the unchecked assumption happens to hold, the verification-engineering analogue of relying on an assumption as a fact. The prime's discipline is the cure: read each proposition's mode, confirm the operation matches (check the asserts, discharge the preconditions, induct on the invariants, flag the assumes), and trace the system's true guarantee to the weakest mode in its assumption set. The same typed-handle-gates-operations structure governs a scientific result (hypothesis-mode forbids reliance, datum-mode carries measurement error) and a legal case (alleged-mode invites challenge, adjudicated-mode does not).
Mapped back: The verification pipeline runs the prime end-to-end — propositions in assert-, assume-, precondition-, and invariant-modes; the content-invariance that makes the same claim checkable or trusted depending only on its mode; the operation sets each mode licenses; and the weakest-mode inheritance that makes a proof only as sound as its weakest assume — and demonstrates the transfer: a logician discharging a premise, an engineer honoring an assume/assert distinction, a scientist refusing to rely on a hypothesis, and a court weighing an alleged versus an adjudicated fact are operating the same content-orthogonal mode dimension.
Structural Tensions¶
T1 — Mode versus Content (the Orthogonality That Gets Collapsed). The prime's foundational tension is that mode is orthogonal to content, yet reasoners persistently read the permitted operations off the content (or the source) rather than the mode. The failure mode is the mode error: applying an operation licensed by one mode to a proposition that is in another — relying on an assumption as a fact, testing an axiom as a hypothesis, refuting a stipulation as a discovery — because attention went to what the proposition says, not how it is held. Diagnostic: before any operation, ask "in what mode is this proposition being held, and does this mode license this move?"; where the operation's legitimacy is being inferred from the content, the orthogonality has been collapsed and a mode error is imminent.
T2 — Stable Mode versus Re-assignment (Tracking the Transition). Modes are not fixed: the operations of inquiry are mode transitions, and a proposition's mode at one stage is not its mode at another. The tension is between treating a mode as a settled label and tracking its movement as inquiry proceeds. The failure mode is stale-mode reasoning: continuing to treat a proposition in its old mode after it should have transitioned (relying on a hypothesis as if it were now confirmed though the test never ran) or after it has transitioned (still testing a proposition that has earned fact-status). Diagnostic: ask whether the proposition's current mode reflects the operations actually performed on it; a hypothesis is not a fact until tested, an alleged fact not adjudicated until ruled, and operating on the basis of an unearned transition is the error.
T3 — The Weakest-Mode Inheritance versus Local Rigor (Conclusion Grade). A conclusion inherits the weakest mode among its supports, but rigorous local inference creates the illusion that the conclusion is as secure as its tightest step. The tension is between the security of the inference and the security of its inputs. The failure mode is rigor laundering: presenting a conclusion as well-established because the reasoning is impeccable, while it rests on an untested assumption that caps its true grade — the proof is perfect, the premise is a guess, and the conclusion is only a guess. Diagnostic: trace the conclusion's dependency set and find the weakest mode in it; the conclusion's grade is that weakest mode's grade, and no amount of inferential rigor raises it above the weakest-moded input.
T4 — Categorical Mode versus Graded Credence (Type versus Degree). Mode is a categorical type that fixes the kind of licensed operation; credence is a graded degree of confidence within a mode. The tension is the temptation to let a high credence substitute for a mode transition. The failure mode is credence-for-mode substitution: treating a high-probability hypothesis as if its probability licensed reliance, when its mode (hypothesis) still forbids reliance and requires testing — confidence is not confirmation, and a 0.95-probable hypothesis is still a hypothesis. Diagnostic: ask whether the move being licensed depends on the proposition's mode or its credence; reliance, refutation-immunity, and dischargeability are mode-gated and unaffected by credence, so raising the number changes the degree of belief but not the operation set.
T5 — Explicit Mode-Tracking versus Tacit Holding (Surfacing Cost). Disciplined reasoning tracks modes explicitly (the logician's premise markers, the law's evidentiary categories, the verifier's assert/assume distinction), but everyday reasoning holds propositions in tacit, unmarked modes, and surfacing every mode is costly. The tension is between the safety of explicit mode-tracking and the friction of marking every claim. The failure mode at one pole is tacit-mode drift — propositions silently changing mode (an assumption hardening into an unexamined fact) because no one tracked them — and at the other, mode-tracking overhead, annotating trivial claims until reasoning stalls. Diagnostic: make modes explicit for the load-bearing propositions (those whose mode gates a consequential operation), and accept tacit holding for the rest; the budget for surfacing modes should track which mode errors would actually be costly.
T6 — Structural Pattern versus Epistemic-Practice Binding (Scope of Transfer). Stated abstractly, the prime is a content-orthogonal typed handle that gates operations — a clean structural pattern — but it is inseparable from some inferential practice: there must be a reasoner or a formal system that tracks the mode and honors its gating. The tension is between the pattern's substrate-generality (it recurs identically across logic, science, law, and code) and its dependence on an epistemic substrate. The failure mode is over-transfer: applying the mode vocabulary to a mute physical process that holds no propositions and honors no gating, mistaking a metaphor for an instance. Diagnostic: ask whether there is an inferential agent or formal system actually tracking the proposition's mode and constraining operations by it; the pattern travels across every reasoning substrate precisely because all of them track modes, and not beyond that band.
Structural–Framed Character¶
Epistemic mode sits near the structural end of the structural–framed spectrum, graded mixed-structural with a frontmatter aggregate of 0.3 — a structural pattern with a faint epistemic-practice tint, and, honestly, the most structural of its drafting batch. Its placement closely tracks its child assumption (0.2, structural), and for the same reasons: the core is bare relational structure, three diagnostics read flatly zero, and only the inseparability from an inferential activity lifts it off the pole. The prime is graded a half-step more framed than assumption not because it carries more practice-binding — it carries the same — but because the dimension it names is one notch more abstract and the import-of-stance is marginally heavier, so the honest reading nudges the aggregate from 0.2 to 0.3.
Three diagnostics read flatly zero and set the structural baseline. The pattern carries no home vocabulary that must travel (vocab_travels 0.0): the same typed-handle-gates-operations structure appears as the axiom/theorem/conjecture of mathematics, the hypothesis/law/datum of science, the alleged/adjudicated fact of law, and the assert/assume/invariant of code, each named in its own field's words, with no single lexicon imported wholesale. It carries no evaluative weight (evaluative_weight 0.0): a proposition's mode is neither good nor bad — an assumption is not worse than a fact, only held differently and licensing different operations — and the prime is the bare typing, not a judgment on any mode. And its origin is formal (institutional_origin 0.0): the notion of how a proposition is held, and that the holding gates the licensed operations, is a piece of logic and epistemology, not the product of any human institution.
What lifts it just off the pole is the two half-points, both honest and both inherited from the structure's dependence on reasoning. The pattern is faintly human-practice-bound (human_practice_bound 0.5): a mode exists only relative to some reasoning or formal system that tracks it and honors its gating — there is no epistemic mode where there is no reasoner or system to hold the proposition and constrain operations by its type — yet that system need not be human, since any inferential substrate (a formal proof, a verification tool, a typed program) tracks modes and honors their gating, which is why the score is a half rather than a full point. And invoking it half-imports (import_vs_recognize 0.5): to call a proposition's standing its "epistemic mode" is partly to recognize a typed handle already governing the operations in some reasoning, but partly to take up the interpretive stance of tracking the epistemic status of a claim and auditing operations against it — a move tied to the practice of disciplined inference rather than read off a physical system. The 0.3 aggregate is the right reading: a structurally clean, vocabulary-free, evaluatively-neutral pattern whose only departure from the pole is its inseparability from the activity of reasoning — the same departure that places its child assumption at 0.2, registered a notch higher here because the dimension is the more abstract object and its invocation imports marginally more of the auditing stance.
Substrate Independence¶
Epistemic mode is a strongly substrate-independent prime — composite 4 / 5 on the substrate-independence scale, the same grade as its child assumption and for parallel reasons. Its breadth is broad (domain breadth 4): the content-orthogonal-typed-handle-that-gates-operations structure recurs with the same shape across logic and mathematics (axiom, theorem, conjecture, premise, each licensing distinct operations), empirical science (hypothesis to be tested, law held provisionally, datum carrying measurement error), law (stipulated, alleged, adjudicated facts, presumptions, legal fictions, each gating what may be done), and programming and formal methods (assert, assume, invariant, precondition, postcondition) — distinct inferential media in which the same mode dimension determines the licensed operations. Structural abstraction sits at 4: the structure — a mode label, orthogonal to content, that fixes the operation set, with conclusions inheriting the weakest mode of their supports — is medium-neutral and the vocabulary itself is bare, but the pattern is inseparable from some inferential activity, so it presupposes a reasoner or formal system that tracks the mode and honors its gating rather than running in a mute physical loop. Transfer evidence is concrete (4): the read-the-mode-before-operating discipline, the weakest-mode-inheritance grading rule, the strengthen-by-transition logic of inquiry, and the mode-error failure analysis port unchanged from a conditional-proof subproof to an experimental auxiliary assumption to an alleged-versus-adjudicated fact to an assert/assume distinction, and the catastrophic confusion of "assume where you needed assert" is recognizably the same defect as relying on an assumption as a fact. The composite of 4 records a pattern recognized across nearly every inferential domain, held just short of 5 only by its dependence on the activity of reasoning — the identical ceiling that holds its child assumption at 4.
- Composite substrate independence — 4 / 5
- Domain breadth — 4 / 5
- Structural abstraction — 4 / 5
- Transfer evidence — 4 / 5
Relationships to Other Primes¶
Foundational — no parent edges in the catalog.
Children (2) — more specific cases that build on this
-
Assumption is a kind of Epistemic Mode Of A Proposition
The file: assumption is ONE VALUE of the mode dimension (held-as-if-true provisionally). Dimension-to-value, like temperature-to-hot. Clean child; nearest neighbor (0.82).
-
Axiom is a kind of Epistemic Mode Of A Proposition
The file: axiom is another single mode (foundationally-true-within-a-system, do-not-refute-from-within). One value of the mode dimension. Clean child.
Neighborhood in Abstraction Space¶
Epistemic Mode Of A Proposition sits in a moderately populated region (49th percentile for distinctiveness): it has near-neighbors but no dense thicket of synonyms.
Family — Inference & Evidence (26 primes)
Nearest neighbors
- Epistemic Humility — 0.74
- Modal Reasoning — 0.71
- Inductive Reasoning — 0.71
- Indexicality — 0.71
- Axiom — 0.71
Computed from structural-signature embeddings · 2026-06-14
Not to Be Confused With¶
The most important confusion is with assumption, the chosen nearest neighbor (proposer similarity 0.82), because the two are tightly related and the boundary is exactly the one this prime exists to draw. The relation is dimension-to-value: an assumption is one mode — a proposition held as if true, provisionally, for the purposes of some reasoning, licensing the operation "rely on while flagging that the conclusion is conditional on it." The epistemic mode of a proposition is the dimension of which assumption is one value among several (axiom, hypothesis, premise, belief, conjecture, datum, stipulation), and its content is precisely that the choice among these modes — independent of the proposition's content — gates which operations are licensed. Naming the dimension is not naming any value of it: a prime for "the mode dimension and its gating" is no more subsumed by a prime for "assumption" than a prime for "temperature" is subsumed by a prime for "hot." The distinction is load-bearing because the two do different work. assumption answers "what is this particular held-as-if-true proposition, and what may I do with it?" — and its own entry already lists the neighboring modes (axiom, hypothesis, premise, belief, fact) and observes that "the role determines the appropriate intervention," which is exactly the epistemic-mode insight viewed from inside one of its values. This prime answers the prior, general question "modes are a content-orthogonal variable, and which mode a proposition is in fixes the operation set" — the recognition that "axiom versus assumption versus hypothesis" is a difference in mode at all, that the same proposition can move between them, and that conclusions inherit the weakest mode of their supports. A practitioner who collapses the dimension into the single value assumption loses the general machinery: the mode-transition account of inquiry, the weakest-mode grading rule across all modes (not just assumptions), and the recognition that an axiom, a hypothesis, and a stipulation are mode-siblings of the assumption, each with its own gating.
A second genuine confusion is with belief_formation. A belief is a proposition an agent holds as true, with a psychology of how it was acquired and updated; belief_formation names that process of coming to hold a proposition true and act accordingly. The epistemic mode of a proposition names the static typed status a proposition carries and the operation set that status licenses — a different object. The distinction is sharp on two points. First, belief is one mode among several: a proposition can be in assumption-mode (relied on while disbelieved), axiom-mode (stipulated, not believed-as-discovered), or conjecture-mode (held open, not yet believed), none of which is a belief, and none of which any act of belief-formation produced — so the mode dimension ranges over far more than the outputs of belief-formation. Second, mode concerns the licensed operations, not the acquisition: the prime is about what may be done with a proposition given how it is held (test it, rely on it, discharge it), whereas belief-formation is about how the agent came to hold it. The discriminating question is whether the concern is the process by which a proposition became held-true (belief_formation) or the typed status that gates the operations on it (epistemic mode); a modeler who assumes normality while actively disbelieving it shows the two apart at once — the proposition is in assumption-mode (licensing reliance-while-flagged) with no corresponding belief, so no belief-formation has occurred, yet the mode and its gating are fully operative.
A third confusion is with the proposition's truth-value and its credence. Truth-value is whether the content is true; credence is how probable the agent holds it; mode is how it is held and what that licenses — three orthogonal things. Mode is not truth-value: a false proposition can be soundly held as an assumption (for a reductio, or as a deliberately false legal fiction), and a true one can be a mere conjecture pending proof; the mode neither asserts nor settles the truth-value, it only fixes the operations licensed pending the truth-value's determination. Mode is not credence: a credence is a graded quantity attached to a single mode (belief), whereas mode is a categorical type that changes the kind of operation licensed — a hypothesis at credence 0.9 must still be tested, not relied on, because its mode forbids reliance regardless of how probable it is. The discriminating questions are whether the feature in play is whether the content is true (truth-value), how probable it is held (credence), or how it is held and what that permits (mode). Conflating mode with truth-value makes a reasoner think assigning a mode settles whether the claim is true (it does not); conflating mode with credence makes a reasoner think a high probability licenses a mode-forbidden operation (it does not).
For a practitioner these distinctions decide what is being claimed and how to act. Confusing the epistemic mode with assumption collapses the whole content-orthogonal dimension into one of its values, losing the mode-transition logic of inquiry and the weakest-mode grading rule that ranges over all modes. Confusing it with belief_formation mistakes the static typed status that gates operations for the process of coming to hold a proposition true, missing the modes (assumption, axiom, conjecture) that no belief-formation produces. Confusing it with truth-value or credence mistakes a categorical type that licenses operations for either the fact of the matter or the degree of confidence, and so applies an operation the mode forbids. The unifying discipline is the prime's mode check: of any claim on the table, read its mode from the field's vocabulary, confirm that the operation being performed is one that mode licenses, track the transitions as inquiry moves the proposition between modes, and grade any conclusion by the weakest mode in its support — because the content can be flawless and the inference rigorous, yet the reasoning fail entirely on a mode error.
Solution Archetypes¶
No catalogued solution archetypes reference this prime yet.