Kripke-Style Intensional Models

Updated 11 February 2026

Kripke-style intensional models are frameworks that generalize possible-worlds semantics by encoding context-dependent intensions alongside extensional truth conditions.
They organize worlds via structured orders, trees, or algebraic systems, enabling precise analysis in modal, provability, and substructural logics.
The models support advanced semantic analyses through categorical and vector-space embeddings, facilitating modular enrichment and compositional evaluation.

Kripke-style intensional models generalize traditional possible-worlds semantics by providing a structured framework for evaluating modalities and intensional constructs across a wide range of logical systems, including provability logics, natural language semantics, ultrafinitism, and set theory in residuated and substructural logics. These models explicate not just the extensional truth conditions of formulas but encode index-parameterized variation in meaning, resource-sensitivity, and context dependency, aligning semantic theory with both proof-theoretic and algebraic perspectives.

1. Fundamental Framework

A Kripke-style intensional model comprises a structure of "worlds" (or contexts, indices, states), typically organized via a preorder, partial order, or more complex frame such as a tree, monoid, or sequence space. For each "sort" of intensional parameter—worlds (modal), times (temporal logic), locations, or resources—a corresponding relational or algebraic structure is specified. The extension of a syntactic formula is computed relative to such a world, but the meaning (intension) is understood as a world-dependent function or algebraic object, thus distinguishing between sense and reference in the Frege-Montague tradition (Majkic, 2011).

Formally, models are specified as tuples combining:

Domains for extensional (base) types;
Families of Kripke frames for each intensional parameter;
An interpretation function assigning (possibly world-dependent) extensions or intensions to constants, function symbols, and predicates;
Modal (or intensional) evaluation clauses, typically via accessibility relations or general algebraic connectives.

Kripke-style intensional semantics are central in provability logics such as GL, GLP(Λ), and interpretability logics (ILM), including their transfinite polymodal generalizations (Fernández-Duque et al., 2012). Consider the closed fragment GLP^0_Λ, where formulas lack propositional variables and are built from falsity and a tower of modalities ξ. The frame I(Θ,Λ) consists of "ℓ-sequences": functions f : Λ → Θ constrained by iterated logarithmic and hyperexponential supports, encoding how much "descendant strength" remains for each modality. Accessibility relations are defined via componentwise agreement along initial segments and strict increase at a specific index. Formula satisfaction is recursively specified: e.g., f ⊨ ⟨ξ⟩φ iff some g <_ξ f and g ⊨ φ.

These constructions validate characteristic axioms:

K-schema: ξ → ([ξ]φ → [ξ]ψ)
Löb: ξ → [ξ]φ
Monotonicity and frame-specific interactions (e.g., A4: ⟨ξ⟩φ → [ζ]⟨ξ⟩φ for ξ<ζ<Λ)

The completeness of GLP^0_Λ with respect to I(Θ,Λ) is sharply characterized by the hyperexponential bound Θ ≥ e^Λ(1). Worlds encode not just extensional reachable sets but intensional information about modal "depths" and provability resources, generalizing Ignatiev’s models to arbitrary modality towers (Fernández-Duque et al., 2012).

3. Category-Theoretic and Algebraic Embeddings

Categorical frameworks formalize intensional models as objects in a category ModInt, whose morphisms are bounded morphisms between Kripke frames and identity-on-extensional components. Objects are specified by:

A family of extensional domains (for base types);
A family of Kripke frames (for each intensional parameter type);
Interpretation functions assigning to each symbol an intension (i.e., a context-dependent mapping).

This structure allows a fully faithful embedding of extensional models (where all intensional frames are trivial singletons) and a canonical "trivialization" reducing an intensional model to its extensional shadow (Quigley, 2024). Moreover, each semantic construction is functorial: composition and substitution respect the categorical structure, and addition of new intensional sorts or dimensions is modular. This provides a foundation for integrating multiple context parameters (e.g., worlds × times × locations) and supports future enrichment (e.g., higher categorical structure, Day convolution).

4. Two-Step and Compositional Semantics

Kripke-style intensional semantics instantiate a two-step meaning assignment:

Syntax is mapped to intensional algebraic entities (concepts, properties, functions, etc., via an interpretation I).
Each world (or index) selects an extensionalization h, mapping the intensional object to its extension in that world (a set of tuples, truth-value, etc.).

The satisfaction relation and algebraic structure interact as follows (Majkic, 2011):

The intensional algebra A_int captures term-level composition, conjunction, quantification, negation as algebraic operations (e.g., concept conjunction, negation, existential abstraction).
The extensional algebra A_ℛ realizes these as operations on extensions, e.g., joins and projections on relations.
There exists a homomorphic diagram: syntactic algebra → intensional algebra → extensional algebra, such that extensions respect syntactic-compositional structure in every world.

This two-stage process explicitly separates the semantic evaluation into a layer for "sense" or intension, and a layer for "reference" or extension, clarifying semantic distinctions not present in pure extensional models (as in Tarski semantics).

5. Advanced Model Variants: Ultrafinitist, Substructural, and Vector Embeddings

Substantial model-theoretic variants arise:

Ultrafinitist/Esenin-Volpin semantics: Worlds are nodes in a finite-depth tree, with domains expanding only via feasible term formation. Evaluation enforces resource sensitivity—truth at bounded depth corresponds to feasible derivability. Term equality depends on explicit generability along the construction path (Mannucci, 2023).
Modal Residuated Logics: Kripke-style models are constructed over complete residuated monoids (lattices with a monoidal operation and adjoint), supporting substructural connectives—tensor, implication, with generalized forcing clauses. Modalities (e.g., possibility ◇) are indexed by conucleus operators, and the set hierarchy and constructible universe are mirrored inside residuated models, supporting translation to Heyting-valued models (Moncayo et al., 2024).
Vector-Space Embeddings: Each component of a Kripke-style intensional model (entities, indices, functions, etc.) is embedded injectively into an appropriate Hilbert space. Modal operators are represented as adjacency matrices or integral kernels, and their evaluation reduces to linear algebraic operations and thresholding. This homomorphism translates intensional semantics into the framework of high-dimensional vector space semantics, while ensuring full compositionality and injectivity: distinct intensions map to distinct vectors or operators (Quigley, 3 Feb 2026).

6. Interpretations, Limitations, and Connections

Kripke-style intensional models rigorously encode Montague’s intuition that meaning is context-dependent. However, not all modal logics support nontrivial intensionality: e.g., FOL_K (standard Kripke reading of quantifiers) remains extensionally equivalent to classical FOL unless the universe is enriched with genuine intensional sorts or context-parameters (Majkic, 2011). The precise algebraic structure (lattice, monoid, topological space, etc.) and the choice of morphisms (monotone, bounded, measure-preserving, etc.) impact both the expressiveness and the nature of intensionality represented.

These models unify and generalize extensional, intensional, substructural, and categorical semantics, providing a toolbox for constructing semantic theories tailored to the needs of various fields—provability, formal linguistics, ultrafinitism, and nonclassical logics. Future directions include the full categorification of the intensional model framework, integration of indexical contexts, and the analysis of algebraic invariants and dualities across intensional semantic landscapes (Quigley, 2024, Quigley, 3 Feb 2026).