Logical Varieties in Normative Reasoning

Although conventional logical systems based on logical calculi have been successfully used in mathematics and beyond, they have definite limitations that restrict their application in many cases. For instance, the principal condition for any logical calculus is its consistency. At the same time, knowledge about large object domains (in science or in practice) is essentially inconsistent. Logical prevarieties and varieties were introduced to eliminate these limitations in a logically correct way. In this paper, the Logic of Reasonable Inferences is described. This logic has been applied successfully to model legal reasoning with inconsistent knowledge. It is demonstrated that this logic is a logical variety and properties of logical varieties related to legal reasoning are developed.

💡 Research Summary

The paper begins by pointing out a fundamental limitation of traditional logical calculi: they require consistency, yet many real‑world domains—especially legal knowledge—are riddled with contradictions. To overcome this, the authors introduce the notions of logical prevarieties and logical varieties. A prevariety is a collection of logical calculi whose axioms and theorems are mapped into a common language via partial functions; a variety strengthens this by demanding that the intersections of these mapped components themselves form a logical calculus. The authors focus on syntactic (or deductive) prevarieties/varieties, defining them in terms of a logical language L, a set of inference rules R, and a class of partial mappings F. Each calculus C_i = (A_i, H_i, T_i) contributes its axioms A_i and theorems T_i, which are projected into L by f_i and g_i. When the images intersect in a way that can be represented again by a calculus (possibly after bijective renaming), the structure qualifies as a logical variety of depth k; a “full” variety satisfies this condition for any depth.

The authors then apply this framework to normative reasoning, arguing that legal reasoning inherently involves both an “object level” (deriving normative conclusions from facts and statutes) and a “meta level” (selecting a common perspective among competing conclusions). Because legal systems must often operate with contradictory statutes, precedents, or factual claims, a logic that tolerates inconsistency while still delivering “reasonable” inferences is required.

Enter the Logic of Reasonable Inferences (LRI). LRI is built on classical first‑order predicate calculus but augments it with a meta‑rule that selects only those inferences that satisfy two universal legal constraints: (1) justification – every derived normative opinion must be backed by a complete argument, and (2) rationality – the argument must be non‑contradictory. LRI therefore allows multiple, potentially conflicting conclusions to coexist, yet it can still produce a single “reasonable” decision when needed. The paper demonstrates that LRI itself forms a logical variety: its components (different legal sub‑systems, temporal versions of statutes, or jurisdictional fragments) can be treated as separate calculi, and the intersections of their projected axioms and theorems are again representable as an LRI calculus. Consequently, LRI is a full syntactic logical variety of any depth.

Beyond law, the authors discuss how logical varieties support large knowledge bases in artificial intelligence. By partitioning a massive KB into tightly coupled sub‑theories (each a calculus) and loosely coupling them via shared language, one obtains a graph‑like variety that enables efficient local inference while controlling combinatorial explosion. This approach has been used in robot motion planning, many‑worlds quantum interpretations, and pluralistic quantum field theories.

In summary, the paper provides a rigorous formalism—logical varieties—that generalizes traditional logical calculi to handle inconsistent, dynamic, and multi‑perspective knowledge. It validates the approach by showing that the Logic of Reasonable Inferences, a system designed for legal reasoning, satisfies the full variety conditions. The work thus bridges abstract logical theory with practical normative reasoning, offering a promising foundation for future legal AI, multi‑modal reasoning systems, and any domain where contradictory information must be managed without sacrificing logical rigor.