Non-Distributive Semantics for Epistemic Modals

19 concepts7 papers2 proposals

In classical logic, we take certain structural rules for granted. Among the most deeply entrenched is the distributive law, which states that conjunction distributes over disjunction:

$$A \land (B \lor C) \equiv (A \land B) \lor (A \land C)$$

For decades, this algebraic property has underpinned standard semantic frameworks. However, when we attempt to model how humans reason about uncertainty—specifically using epistemic modals like might or must—classical distributivity introduces severe empirical contradictions and paradoxes.

A growing body of research suggests that these linguistic paradoxes are not inherent flaws in human reasoning, but are rather artifacts of over-constrained classical rules. By shifting the mathematical foundation of modal logic from classical Boolean algebras to non-distributive algebraic structures, such as ortholattices, we can naturally capture the nuances of epistemic reasoning.

The Breakdown of Distributivity

Why does classical logic fail when reasoning about what might be true? The core issue is that epistemic modals inherently express states of ignorance or partial information, which do not behave like fixed, objective facts.

As Matthew Mandelkern argues in The Orthologic of Epistemic Modals, inference patterns like distributivity intuitively fail in modal contexts. When we force natural language statements involving uncertainty into a Boolean framework, the logic over-corrects, forcing a level of precision that destroys the underlying epistemic meaning.

In A fundamental non-classical logic, Wesley H. Holliday demonstrates that applying classical distributive laws to statements of possibility leads to unintended or contradictory conclusions. For example, knowing that a coin might be heads and might be tails represents a single, cohesive state of uncertainty. Classical logic, by enforcing distributivity, attempts to decompose this state into a rigid disjunction of absolute certainties, breaking the semantics of natural language.

From Boolean Algebras to Ortholattices

To fix this, we must change the underlying algebra of logic. Classical logic is built on Boolean algebra, where the lattice of truth values is strictly distributive. If we relax this constraint, we move into the realm of non-distributive lattices and ortholattices.

Classical Logic:   Boolean Algebra  ---> Enforces Distributivity ---> Fails for Modals
Epistemic Logic:   Ortholattices    ---> Relaxes Distributivity  ---> Captures Uncertainty

In Non-classical modal logic for natural language, Wesley H. Holliday proposes using ortholattices as an algebraic foundation instead of standard Boolean algebra. This generalization allows logical connectives like negation and disjunction to be modeled in ways that accommodate human uncertainty. Instead of treating linguistic quirks as errors to be patched with ad-hoc constraints, this approach treats them as structural features of a non-classical reasoning system.

The power of this shift is that it invalidates the problematic inference patterns specifically for modal contexts while preserving classical logic where it belongs. As developed by Matthew Mandelkern in The Orthologic of Epistemic Modals, an epistemic orthologic can isolate the failure of distributivity to sentences containing modals, ensuring that non-modal statements still evaluate classically.

The Computational Echo: Shannon Expansion Failure

The failure of distributivity is not just a philosophical or linguistic curiosity; it has profound computational consequences for AI and automated reasoning.

In classical knowledge compilation, complex logical formulas are translated into tractable formats like Ordered Binary Decision Diagrams (OBDDs) to enable polynomial-time queries. This process relies fundamentally on the Shannon expansion principle, which recursively decomposes a formula based on a variable $x$:

$$F \equiv (x \land F_{x}) \lor (\neg x \land F_{\neg x})$$

Notice that this expansion relies entirely on the validity of the distributive law to split and recombine the formula's internal structure.

In Knowledge Compilation in the Modal Logic S5, Meghyn Bienvenu et al. demonstrate that the Shannon expansion property does not hold within epistemic modal logic (S5). Because the semantics of agent knowledge do not preserve this recursive decomposition, standard propositional compilation techniques cannot be directly applied to epistemic knowledge bases. This limitation necessitates entirely new compilation strategies that achieve tractability without relying on variable-based branching.

Building a Unified Fundamental Logic

By systematically removing the distributive law and other over-constrained classical principles (such as double-negation elimination), researchers have mapped out a highly generalized "base" logic.

In Modal Logic, Fundamentally, Wesley H. Holliday constructs a semantic framework on non-distributive lattices, proving that necessity ($\Box$) and possibility ($\Diamond$) can still be effectively represented using a single accessibility relation, even when they are no longer strict classical duals. This mathematical generalization allows modal operators to be applied to fields ranging from natural language to quantum mechanics, where non-distributive structures are already standard.

Ultimately, this research direction provides a unified way to handle multiple distinct challenges in logic. As shown in Vagueness and the Connectives, dropping the law of distributivity alongside the law of the excluded middle allows a single framework to resolve both the behavior of epistemic modals and the Sorites Paradox (the problem of vagueness). By grounding logic in compatibility relations and lattice representations rather than rigid Boolean truths, as detailed in Compatibility and accessibility: Lattice representations for semantics of non-classical and modal logics, non-distributive semantics offers a more faithful, resilient mirror for the inherent flexibility of human thought.

Key Foundational Work

Non-classical modal logic for natural language (2022) — contributed 4 concept elements to this direction.
Knowledge Compilation in the Modal Logic S5 (2010) — contributed 4 concept elements to this direction.
A fundamental non-classical logic (2022) — contributed 2 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 7 papers
  2. Version 1 · 2026-10-04 · 9 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_non_distributive_semantics_7751501c,
  title = {non-distributive semantics},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/non-distributive-semantics/rdr_7751501c75954eeb983fd4fec7b70f0f},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 2, frozen text):

@misc{lacuna_non_distributive_semantics_7751501c_v2,
  title = {non-distributive semantics},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/non-distributive-semantics/rdr_7751501c75954eeb983fd4fec7b70f0f/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake