Non-Monotonic Logic

The core property that distinguishes non-monotonic from classical logic. A conclusion is defeasible if it is rationally warranted based on current premises but can be retracted when new, contradictory information is added.

• In legal reasoning, a prima facie obligation (e.g., 'contracts must be performed') is defeasible by a force majeure clause
• The inference engine must maintain justification structures to track which conclusions depend on which assumptions
• When a premise is invalidated, all downstream conclusions that relied on it are recursively withdrawn via truth maintenance

Non-monotonic systems encode default rules that apply in the typical case but admit explicit exceptions. This mirrors how legal statutes operate: a general rule establishes a norm, while specific provisions carve out exceptions.

• Default: 'A valid contract creates binding obligations'
• Exception: 'Unless the contract was signed under duress'
• The lex specialis principle is computationally implemented as an exception handler that overrides the default when the exception's applicability conditions are satisfied
• This avoids the combinatorial explosion of enumerating all possible exceptions within the general rule itself

When multiple applicable rules yield contradictory conclusions, non-monotonic systems resolve the conflict through explicit preference orderings rather than treating the knowledge base as inconsistent.

• Lex superior: Rules from higher authorities override lower ones
• Lex posterior: Later-enacted rules override earlier ones
• Lex specialis: Specific rules override general ones
• These precedence relations form a normative hierarchy graph—a directed acyclic structure that the inference engine traverses to determine which rule prevails in any given conflict

Non-monotonic systems incorporate formal belief revision operators that rationally update the knowledge base when new rules or facts are introduced. The goal is to achieve a maximal consistent subset of the rule base.

• The AGM postulates (Alchourrón, Gärdenfors, Makinson) define rationality criteria for belief change
• When a new statute is enacted, the system must: (1) incorporate the new rule, (2) remove or weaken conflicting rules, and (3) preserve as much of the existing knowledge as possible
• This is computationally modeled through normative repair operators that minimally modify the rule set to restore consistency

A distinctive capability of non-monotonic deontic systems is handling contrary-to-duty obligations—what an agent must do after violating a primary obligation. Classical deontic logic generates paradoxes here; non-monotonic extensions resolve them.

• Primary obligation: 'You must not damage property'
• Contrary-to-duty: 'If you damage property, you must compensate the owner'
• The system models this as a conditional obligation that activates only upon violation of the primary norm
• This is critical for modeling remedial clauses in contracts and penalty provisions in statutes

To ensure deterministic conflict resolution, non-monotonic legal systems organize rules into ordered strata based on priority. The inference engine consults higher strata first and only descends to lower strata when no applicable rule is found.

• Stratum 1: Constitutional provisions
• Stratum 2: Statutory law
• Stratum 3: Administrative regulations
• Stratum 4: Contractual terms
• This rule base stratification directly encodes the hierarchy of legal authority and ensures that a constitutional right cannot be overridden by a contractual clause

A principle of legal interpretation stating that a law governing a specific subject matter overrides a general law governing a broader category. This forms the basis for rule exception handling in normative systems.

• Specificity test: Algorithmically determining which of two conflicting rules has the narrower scope of application
• Exception carving: The specific rule does not invalidate the general rule; it merely creates a limited exception
• Computational implementation: Encoded as a priority function in rule-based reasoning engines

The algorithmic process of identifying contradictory obligations, permissions, or prohibitions within a normative corpus. A direct collision between a mandatory rule and a prohibitive rule triggers resolution protocols.

• Obligation-obligation conflict: Two mandatory but mutually exclusive actions
• Obligation-prohibition conflict: A single action is both required and forbidden
• Permissive-prohibitive conflict: An action is simultaneously permitted and prohibited
• Detection algorithms: Pattern matching over deontic modal operators in formalized rule representations

A declarative programming paradigm based on stable model semantics, particularly well-suited for modeling complex combinatorial problems like normative conflict resolution and default reasoning.

• Stable models: Each answer set represents one consistent, non-monotonic interpretation of a logic program
• Default negation: The operator not p means 'p is not believed,' enabling default assumptions that can be retracted
• Legal application: Used to compute all possible consistent outcomes from a set of conflicting legal rules

The process of rationally changing a set of legal rules or beliefs to incorporate a new rule while maintaining overall consistency. Often guided by formal postulates like the AGM theory.

• Expansion: Adding a new rule without checking consistency
• Contraction: Removing a rule to eliminate inconsistency
• Revision: Adding a new rule while removing conflicting rules to preserve consistency
• Entrenchment ordering: A priority ranking that determines which beliefs are sacrificed first during revision

A computational method for resolving normative conflicts by identifying the largest subset of non-contradictory rules from an inconsistent rule base. This enables conflict-free reasoning without arbitrary rule deletion.

• Subset maximality: No additional rule can be added without introducing inconsistency
• Multiple MCS problem: An inconsistent rule base may yield several distinct maximal consistent subsets
• Selection strategies: Domain-specific heuristics or preference orderings choose among competing MCS outputs