ATLAS OF LANGUAGE

AT-134-037 · ATLAS 260

Defeasible Inference

An inference that holds under normal assumptions but can be retracted by additional information.

Part of the subject path Argumentation, Logic and Discourse

Why is this entry in the Atlas?

A · Analytical concept R1 · Distinct contribution to understanding

Defeasible inference adds a disciplined account of conclusions that can lose support when exceptions emerge. Revision conditions distinguish such reasoning from deduction and from guesses lacking a warranted default.

Colours and admission grounds

The rectangular label identifies the entry function: F – linguistic figure or procedure; A – analytical concept; G – foundational concept; E – selected concrete case. The rounded badge indicates the admission ground: R1 – distinctive contribution; R2 – necessary foundation; R3 – justified illustrative case. This editorial rationale explains the current selection rather than claiming an undocumented original selection history.

Explanation

A defeasible inference is a reasonable yet inherently revisable conclusion. It holds under normal or currently known conditions but loses its warrant when relevant exception information appears. Unlike a deductively valid inference, its form does not guarantee the conclusion; unlike a guess, it relies on rules, experience, or institutionally recognised assumptions. The crucial feature is organised retractability: new evidence does not merely supplement the judgement but can cancel the previous conclusion. The conceptual movement makes normality assumptions explicit, actively searches for defeating conditions, and records when and why revision becomes epistemically required.

Movement of thought

1. Observed facts and a normality rule support an initially reasonable provisional conclusion. 2. Tacit background assumptions are made explicit and tested for their proper domain. 3. Possible exceptions and counterevidence are actively sought rather than merely tolerated later. 4. New information may adjust confidence or fully retract the previous conclusion. 5. The revised justification records which assumption failed and which parts remain warranted.

Scientific use case

Argumentation theory, artificial intelligence, law, medicine, and everyday epistemology model defeasible inference when decisions must be made under incomplete information. Strong applications document exceptions, evidential priority, and revision rules.

Example

From the darkened office and ordinary working hours, someone concludes that nobody remains inside. A recorded night shift defeats this reasonable presumption without making the original generalisation worthless in every case.

Counterexample

An inference is not defeasible merely because someone changes an opinion arbitrarily. The initial conclusion must be supported by identifiable normality assumptions, and revision must respond to relevant new information or a recognised exception.

Distinction

Distinguish defeasible inference from deduction, statistical probability, and arbitrary conjecture. Defeasibility is the rule-governed cancellation of an initially warranted conclusion by relevant additional information.

Effect on thought and perception

Defeasible inference combines the capacity to act with the capacity to correct. It permits warranted judgement without false certainty and makes intellectual flexibility part of rational practice.

Resonance

A conclusion remains open to an answer. New evidence does not merely stand beside prior thought; it can retract the path and reopen an earlier fork from which reasoning begins again.

Cite this entry

Reich, Michael (Hrsg./ed.): „Defeasible Inference“. Atlas der Sprache, AT-134-037. ATLAS 260, data model 1.6.0. https://atlasoflanguage.com/concept/AT-134-037

https://atlasoflanguage.com/concept/AT-134-037

The address remains stable; editorial content may evolve. Editorial reading state: 2026-10-15. Include your access date for precise references.

Historical background

Pollock’s author-hosted manuscript A Recursive Semantics for Defeasible Reasoning treats the withdrawal of a conclusion without rejecting its original premises. Rebutting defeat attacks the conclusion; undercutting defeat attacks its support without establishing a contrary conclusion. This is evidence for a specific theoretical account, not the term’s coinage. The inspected opening sections replace the earlier reliance on article metadata alone. The manuscript’s final publication details were not established in this pass, and its content is not attributed to the unread 1987 or 1992 articles.

How conceptual history changes the analysis

When revising a belief, distinguish evidence against its conclusion from evidence against the way it was supported. Losing a justification does not by itself establish the opposite claim.

Additional example

Original example: A nod in a silent video suggests agreement. A complete recording containing explicit refusal supplies contrary evidence. Learning that the nod merely acknowledges receipt instead undermines the inference to agreement without establishing refusal. The reasons for revising the interpretation differ.