Knowledge and Corrigibility

T2 is the second independent theory document in the T1–T9 programme. It develops the epistemology of DET SOM ER and asks what is required for a person, group, or institution rightly to assign knowledge status to a claim.

T1 established that knowledge must concern something that is. T2 examines access, presence, recognition, justification, uncertainty, and correction. It does not by itself decide whether a particular empirical, historical, legal, or scientific claim is true; it establishes requirements against which such claims must be examined.

T2 does not identify knowledge with direct perception, inner certainty, belief, authority, consensus, documentation, or instrumental usefulness. Knowledge is understood as recognition of what is through a sufficiently clarified, traceable, and reviewable relation to actuality.

Knowledge cannot be the ground of its own actuality. If something is known, there must be a relation, event, structure, or state of affairs that the knowledge concerns. What is known does not become true because it is called known, enacted, or repeated.

Error and uncertainty display the same dependence. A representation can be wrong only because something differs from what the representation says, and it can be uncertain only because access, material, or inference does not yet provide sufficient clarification.

Ontology therefore grounds epistemology but does not replace inquiry. That something is does not automatically provide knowledge of what it is, how it operates, or what follows from it.

Presence in T2 means that something is available to recognition by virtue of actuality. Presence does not guarantee truth; it makes truth, error, and testing possible. Access may take different forms and possess different strengths.

Direct presence concerns what is experienced without the documentary or inferential mediation at issue. Mediated presence concerns what becomes available through traces, memory, language, documents, images, instruments, archives, testimony, or records. Derived presence concerns what is established through valid inference from present or documented conditions.

The common requirement is not a single form of expression but traceable grounding: the referent must be identifiable, mediations explainable, limitations visible, and the conclusion no stronger than its ground.

Experience is an encounter with something that is. Registration is the trace or data form left by the encounter. Recognition is the clarified understanding of what is encountered or registered. Representation is the linguistic, visual, numerical, or institutional expression of that recognition.

The formulation “experience precedes interpretation” does not mean that human beings always possess pure and unmediated access. It means that interpretation cannot be the ground of its own actuality and must answer to something encountered, registered, or derived.

Knowledge does not arise from registration alone. Data must be related to provenance, context, category, and valid inference. Understanding must also remain open to the possibility that the registration was incomplete, miscalibrated, misclassified, or placed within an unsuitable frame.

T2 distinguishes knowledge, belief, opinion, and conception. Knowledge is recognition of what is with a sufficient relation and traceability to actuality. Belief is trust, assent, or credit given to a representation, source, or authority. Opinion is an assessment or interpretation that does not by itself possess knowledge status. Conception is possible, hypothetical, or imagined content.

Belief may correspond with knowledge, follow knowledge, or motivate inquiry. It is nevertheless not what makes a representation true or known. Strong conviction may be false, and an actually recognised relation does not become more true because an agent expresses stronger assent.

The absence of a constitutive belief requirement does not turn accidentally correct behaviour, guessing, or mechanical repetition into knowledge. There must still be a real epistemic relation: the agent must have access to and be capable of standing responsible for what is attributed as known.

Knowledge is factive: if p is known, p must be true. It does not follow that every claim called knowledge is knowledge. Language must therefore distinguish p is known, an agent believes p is known, p was treated as known, and p was well supported at a particular time.

When later inquiry shows p to have been false, it is more precise to say that the attribution of knowledge was mistaken, the justification insufficient, or the material extended beyond what it could bear. The false claim was not true or knowledge during the period in which it was accepted.

An earlier true and bounded recognition may nevertheless be refined or extended. Correction then concerns scope, explanation, or context, not a conversion of the true state of affairs.

Traceability is more than a link to a source. A knowledge claim must be capable of being followed through an explainable chain from what it concerns, through material and registration, to interpretation, inference, and formulated conclusion.

The chain includes at least what the claim concerns, which material supports it, where the material originates, how it was registered, which mediations are present, which interpretation or inference was made, which alternatives were considered, and what may be tested again.

A source may be authentic without proving the claim. A document may be genuine yet incomplete, misinterpreted, outside its valid scope, or itself dependent on unverified information. Traceability therefore requires both provenance control and examination of the epistemic transition.

T2 distinguishes material, datum, trace, evidence, justification, proof, confirmation, and counter-material. These terms may have field-specific meanings but should not be used as though they automatically carried the same force.

Material becomes evidence for a claim when it bears a relevant and justified relation to that claim. Evidence becomes proof only within an explicit standard stating what must be established and to what degree. Confirmation is not merely the absence of contradiction; it requires positive ground.

A formal document, statistical association, expert opinion, or technical measurement is not automatically proof of every claim attached to it. Material cannot possess greater epistemic force than its provenance, method, valid scope, and alternative interpretations can support.

Uncertainty is not a degree of actuality but a feature of access, material, interpretation, or inference. What is the case is not more or less actual according to an agent’s confidence. Epistemic status should state how far the representation can be defended.

T2 distinguishes known, strongly supported, provisional, possible, unresolved, insufficient ground, contradicted, and refuted. Status must be revisable when new material appears, sources weaken, or scope is narrowed.

Institutions should mark status expressly. A provisional assessment should not migrate into registers, media, or decisions as though final. Unresolved does not mean all explanations are equally good; it means that the material does not yet support a responsible final attribution.

• Known — a sufficiently established and traceable state of affairs.
• Strongly supported — extensive ground with visible residual uncertainty or limitation.
• Provisional — reasonable ground requiring further examination.
• Possible — not excluded but not sufficiently established.
• Unresolved — the material does not support a responsible conclusion.
• Insufficient ground — the claim exceeds the material.
• Contradicted — significant material speaks against the claim.
• Refuted — the claim does not hold against sufficient ground.

Corrigibility means that human recognition, representation, justification, scope, and status can be corrected in relation to actuality. It does not mean that truth itself is fluid or that a true representation can be made false by decision.

Factual correction replaces a false statement. Scope correction narrows an overbroad claim. Clarifying correction makes a broadly true representation more precise. Source revision changes the ground where provenance, authenticity, or relevance was misjudged. Status correction moves a claim from known to provisional or unresolved. Withdrawal removes a claim that can no longer be defended.

Corrigibility does not require that everything remain uncertain. It requires that no human formulation may shield itself from relevant counter-material by appeal to its own authority, identity, or institutional position.

Testimony and expertise are necessary sources of much human knowledge. An agent need not personally perform every measurement, witness every event, or master every field in order to know through others.

The transfer holds when the source possesses relevant access or competence, speaks within that competence, can be identified, makes the ground reviewable, and discloses conflicts, limitations, and counter-material.

Authority is epistemically relevant as an indicator of access, training, and responsibility, but not ontologically decisive. “A competent authority says so” may be a good practical reason for trust; it is not identical with a full explanation of why the claim is true.

Institutions can assemble, preserve, and operationalise knowledge that no individual carries alone. Archives, registers, research communities, court records, medical records, and technical logs may stabilise mediated knowledge over time.

Collective storage does not remove responsibility for provenance, source control, role limitation, access control, contradiction, revision, and correction. Information may be copied institutionally without becoming more true.

An institution may establish a decision, status, or legal effect within its mandate. It does not thereby automatically establish the truth of every premise on which the decision rests. T2 therefore distinguishes institutional validity from epistemic correspondence.

Human beings recognise actuality from their position within it. Body, time, language, role, instruments, and social location affect what is available and how it is represented. Perspective may therefore limit and organise attention.

It does not follow that standpoint produces truth or that conflicting claims are automatically equally valid. Perspectives may be complementary, differently informed, misplaced, or directly contradictory. They must be examined against the matter and against one another.

Situation explains limitation and the importance of contradiction. It does not abolish correspondence; it makes requirements of scope, openness, and correction more precise.

Local doubt may be legitimate and methodologically necessary. An agent may doubt a source, measurement, memory, interpretation, or inference without denying all knowledge. Such doubt can strengthen examination.

Total denial of all access is less stable. It itself employs language, reference, and a claimed access to the conditions of human recognition. This shows that absolute denial of access cannot receive unlimited status while using what it denies.

The performative point does not grant infallible access. It secures only that access cannot be denied absolutely while the denial is demanded as valid. What is actually known must still be examined in each case.

The epistemology must meet objections capable of limiting or changing it. The replies below are working replies with visible residual problems, not insulation from further criticism.

T2 enters discussion with classical factive accounts of knowledge, analyses of knowledge as justified true belief, the Gettier problem, fallibilism, foundationalism and coherentism, internalism and externalism, reliabilism, virtue epistemology, and social epistemology.

The position shares the factivity requirement of classical epistemology but denies that a separate act of belief must constitute knowledge. It shares with fallibilism that human access and attribution can be corrected while maintaining that what is genuinely known must be true.

Traceability has affinities with externalist and reliabilist concerns, while T2 also requires visible justification and responsibility where claims have public or institutional effects. Full comparison and literature criticism belong to T9.

T2 requires later documents to distinguish material, observation, interpretation, support, finding, and decision. They must also mark uncertainty and permit correction of both claim and status.

If T2 is substantively revised, T3–T5 and R1–R7 in particular must be reviewed because they operationalise access, evidence, findings, and correction.

• T3: Observation must receive a bounded epistemic status.
• T4: Linguistic claims must remain traceable to what they concern.
• T5: Model support must be distinguished from ontological truth.
• T6: Self-representation and identity must be examinable without erasing the person.
• T7: Authority and institution may carry knowledge but may not make themselves incorrigible.
• T8: Action requires a clarified knowledge ground and marked uncertainty.
• T9: The epistemology must be compared, bounded, and criticised.
• R1: The method requires criteria for evidence, findings, uncertainty, and correction.
• R2–R7: Operative documents must preserve the distinction among material, interpretation, finding, and decision.