Classic AI: search, logic & planning· 1980
A logic for default reasoning
Conclusions can be withdrawn.
On the reading lists
- #129 in the Top 500 (2026 edition)
- #92 of 100 papers to read in your lifetime (2026 edition)
- #5 in Classic AI: search, logic & planning (2026 edition)
Why it matters
Formalized default reasoning for drawing revisable conclusions from incomplete knowledge.
Condensed from the expert lists' notes.
Inside the paper
- The problem
- Ordinary deduction does not easily express everyday assumptions that hold unless an exception is known. For instance, knowing that an individual is a bird does not prove that it is not among the exceptions to flight.
- The idea
- Reiter gives default rules a formal interpretation in which a conclusion can depend on the consistency of an assumption with a set of beliefs. The resulting extensions represent possible sets of conclusions supported by the facts and defaults.
- How it works
- A default rule specifies a prerequisite, one or more assumptions that must remain consistent, and a conclusion.
- A default theory combines ordinary facts with a set of such rules, including rules that represent typical cases or closed-world assumptions.
- An extension is defined by a fixed-point construction: its conclusions are generated using defaults whose justifications are consistent with that same extension.
- New information can change which defaults apply, and different extensions can represent alternative sets of conclusions.
- What it showed
- The paper develops the semantics and proof theory of default logic and studies properties of classes of default theories. Its examples include ordinary exceptions, database assumptions and the persistence of facts through changes in the world.
- Read it for
- Start with the examples in Section 1.1, especially the bird example, before reading the definition of an extension in Section 2. The later theorems and proof procedures reward a second pass; follow the examples first if the notation is unfamiliar.
Written in our own words from the paper itself. A draft the editors have not yet reviewed.
Reading it
Difficulty: Advanced
Read this first: first-order logic.
The record
- First released
- 1980 (from OpenAlex)
- Citations
- 4,584 citations, as counted by
Semantic Scholar (non-commercial licence) - Named by
- both expert lists
- Score
- 0.627 · computed rank 129 · consensus 0.7, must-read 1, citations percentile 0.905 (1980s · classic-ai), awards 0, existing lists 0. Method
Editor's notes
- Readability check: pass: Reiter's default logic, the classic of non-monotonic reasoning. (AI draft (Claude), not yet reviewed, 2026-10-29)
Data sources
Catalogue data from OpenAlex (CC0), Crossref (CC0 metadata; no abstracts are shown), arXiv, and Papers With Code (CC BY-SA 4.0), and
Semantic Scholar (used under its non-commercial API licence). Each count names its source. See the method.