Simpson's Intuitionistic Modal Logics
- Simpson's intuitionistic modal logics are a family of modal systems based on intuitionistic logic and bi-relational semantics with primitive, non-dual modal operators.
- The framework extends to systems like constructive K, intuitionistic S4/S5, and tense logics, enriching the study of modal interactions and proof theory.
- A central contribution is the use of single-conclusion proofs and uniform cut-admissibility, advancing feasible proof search and model-theoretic characterization.
Searching arXiv for recent and foundational papers on Simpson's intuitionistic modal logics and closely related proof theory. Simpson’s intuitionistic modal logics are a family of modal systems built over an intuitionistic propositional base and interpreted by bi-relational Kripke semantics, typically with a preorder for intuitionistic information growth and an accessibility relation for modality. Their central mono-modal system is intuitionistic (), but the framework extends to systems such as constructive (), intuitionistic and , tense logics with converse modalities, grammar-logical generalizations, and explicit-term refinements. A defining feature is that and are primitive and, in general, are not dual under intuitionistic negation; this breaks a central classical simplification and drives much of the proof theory, semantics, and model theory of the area (Tabatabai et al., 2022, Lyon, 27 Nov 2025, Groot et al., 30 Jun 2026).
1. Historical position and conceptual profile
Simpson’s framework systematized a line of work going back to Fischer–Servi, Plotkin–Stirling, and Ewald, and it made bi-relational Kripke semantics the standard setting for intuitionistic modal and tense logics. In this setting, intuitionistic implication is governed by a preorder, while modality is governed by a separate accessibility relation, constrained by compatibility conditions linking the two. Later work treats 0 as the canonical intuitionistic counterpart of classical normal modal 1, and uses it as the reference point for extensions such as 2, 3, tense logics, grammar logics, and justification logics (Girlando et al., 2023, Groot et al., 30 Jun 2026).
A characteristic feature of the Simpson tradition is single-conclusion proof theory. This appears in labelled systems, sequent systems, nested sequents, and natural deduction, and it matches the non-classical behavior of implication and the failure of Boolean dualities. In later developments, the same tradition supports uniform cut-admissibility for large families of intuitionistic grammar logics, feasible admissibility of Visser’s rules for broad sequent-calculus families, decidability for 4, interpolation and Beth definability, and a van Benthem-style characterization of 5 via intuitionistic first-order logic (Lyon, 27 Nov 2025, Tabatabai et al., 2022, Girlando et al., 2023, Groot et al., 30 Jun 2026).
The framework also became a point of comparison for competing notions of “constructive” or “minimal” intuitionistic modality. Later papers show that 6 is neither the only plausible base system nor the weakest normal one in every sense; instead, the treatment of 7, the choice of semantic clauses, and the degree of interaction between 8 and 9 generate several distinct families (Balbiani et al., 26 Feb 2025, Groot et al., 2024, Balbiani et al., 2024).
2. Language and foundational systems
The standard mono-modal language is
0
and, as later work repeatedly emphasizes, 1 and 2 are generally not dual under intuitionistic negation (Tabatabai et al., 2022).
Two foundational Hilbert-style systems are 3 and Simpson’s 4: 5 where
6
and
7
with
8
9
0
This presentation makes explicit that 1 strengthens the constructive 2-base by adding interaction and distributivity principles for 3 (Tabatabai et al., 2022).
Several standard extensions are obtained by adding intuitionistic versions of 4, 5, 6, and 7: 8
9
0
1
For 2, one writes 3 and 4 for the corresponding extensions. Two prominent cases are constructive 5,
6
and intuitionistic 7,
8
also called 9 in one presentation (Tabatabai et al., 2022).
| System | Definition | Note |
|---|---|---|
| 0 | 1 | Constructive 2-base |
| 3 | 4 | Simpson’s mono-modal system |
| 5 | 6 | Constructive 7 |
| 8 | 9 | Also denoted 0 |
The same framework generalizes naturally to multi-modal and tense settings. In intuitionistic grammar logics one fixes a finite alphabet 1, split into forward and backward indices, and gives each 2 both a necessity 3 and a possibility 4. In the mono-modal case 5, one recovers 6; with 7, one recovers intuitionistic tense logic 8 (Lyon, 27 Nov 2025).
3. Bi-relational semantics
The standard semantic setting is Simpson-style bi-relational semantics. A frame is typically a tuple
9
or, in the multi-modal case,
0
where 1 is nonempty and 2 is a preorder. Valuations are monotone along 3, yielding persistence: if 4 and 5, then 6 (Lyon, 27 Nov 2025, Groot et al., 30 Jun 2026).
For mono-modal 7, the characteristic clauses are
8
9
0
with compatibility conditions between 1 and 2 ensuring monotonicity of the modal clauses (Groot et al., 30 Jun 2026). In the multi-modal setting, the possibility clause is commonly written
3
and the necessity clause quantifies over all such 4 pairs (Lyon, 27 Nov 2025).
The key compatibility conditions are the familiar monotonicity or confluence requirements linking 5 and 6. One formulation is: 7
8
These are exactly the conditions highlighted in the later model-theoretic characterization of 9, where they are treated as the semantic background for intuitionistic modal bisimulation and the intuitionistic first-order standard translation (Groot et al., 30 Jun 2026).
Standard modal frame properties reappear in intuitionistic form. In grammar-logical notation, reflexivity, transitivity, symmetry, Euclideanness, and seriality are encoded either by axiom pairs such as
0
1
or by intuitionistic path axioms. This provides a uniform semantics for 2, 3, and their 4, 5, 6, 7, 8 extensions (Lyon, 27 Nov 2025).
Later semantic work also broadens the setting. One line uses constructive or CK-style hereditary clauses for both modalities together with an exploding world or fallible worlds; another isolates alternative “minimal” or “local” readings of 9. These semantic variants do not merely reformulate 00; they generate genuinely different logics and explain why systems between 01 and 02 can disagree even on diamond-free fragments (Groot et al., 2024, Balbiani et al., 26 Feb 2025, Balbiani et al., 2024).
4. Proof theory and calculi
Simpson’s family has been studied through Hilbert systems, natural deduction, labelled calculi, ordinary sequents, nested sequents, and cyclic systems. A recurring constraint is single-conclusion proof theory, which matches intuitionistic implication and the failure of classical duality (Lyon, 27 Nov 2025, Das et al., 2023).
A basic sequent presentation starts from single-conclusion 03 and adds modal rules. For 04, representative rules are
05
yielding
06
Then
07
Within this framework, a large class of modal rules is isolated syntactically as “constructive” by classifying formulas as basic, almost positive, and constructive; this classification controls positive occurrences of 08, 09, implications, and nested boxes (Tabatabai et al., 2022).
A major later development is the nested-sequent treatment of intuitionistic grammar logics. The calculus 10 is single-conclusioned, works over nested trees of sequents, and introduces a structural shift rule that subsumes the structural behavior otherwise separately required for 11, 12, 13, 14, 15, and general intuitionistic path axioms. This yields a purely syntactic proof of cut-admissibility uniformly across all intuitionistic grammar logics in the class, and completeness follows as a corollary (Lyon, 27 Nov 2025). A related structural-refinement program derives cut-free labelled and nested systems from the semantics of intuitionistic grammar logics, proves conservativity over the mono-modal fragment, and identifies a decidable “simple” subclass (Lyon, 2022).
For specific classical-strength extensions inside Simpson’s family, proof theory can be highly concrete. The long-open decidability and finite model property for intuitionistic 16 were established by a fully labelled sequent calculus with two explicit relations, one for 17 and one for 18, together with a terminating proof search that outputs either a cut-free derivation or a finite countermodel (Girlando et al., 2023). There is also a cyclic labelled calculus for intuitionistic Gödel–Löb logic in Simpson’s style, where both modalities are retained and the characteristic semantic condition is converse well-foundedness of the composition 19 (Das et al., 2023).
The proof-theoretic landscape therefore contains both cut-friendly and cut-eliminating methodologies. One line deliberately avoids cut elimination and works by polynomial-time proof transformation; another establishes analyticity, hp-admissibility of structural rules, and syntactic cut-admissibility in modular nested calculi (Tabatabai et al., 2022, Lyon, 27 Nov 2025).
5. Meta-theory, admissibility, and model theory
A central recent theorem is that strong constructive calculi of the form 20, under a mild 21-free or 22-full condition, have the feasible Visser–Harrop property. Concretely, there is a polynomial-time algorithm that transforms a proof of
23
into a proof of one of
24
As a consequence, all Visser’s rules are feasibly admissible for 25, 26, their 27, 28, 29, 30 extensions, bounded width and bounded depth systems, and even for the 31-fragment systems 32 and 33 (Tabatabai et al., 2022).
The same work establishes a sharp negative boundary: if a strong enough 34-free or 35-full intuitionistic modal logic fails at least one Visser rule, then it has no constructive sequent calculus in the paper’s sense. In particular, no intermediate logic 36 has a constructive sequent calculus over the propositional language. This result places Simpson-style proof theory inside a broader “universal proof theory” program, where the shape of admissible rules becomes a logical invariant (Tabatabai et al., 2022).
Several classical metatheoretic properties have also been recovered in modern nested frameworks. For intuitionistic grammar logics, single-conclusion nested sequents yield uniform cut-admissibility, completeness, an interpolation algorithm on proofs, Lyndon interpolation, and Beth definability. These results cover all subsumed 37 and 38 systems with combinations of 39 (Lyon, 27 Nov 2025). For 40, the labelled proof-search procedure establishes both decidability and the finite model property, thereby solving a problem left open in Simpson’s thesis (Girlando et al., 2023).
On the semantic side, 41 now has an exact model-theoretic characterization: it is the 42-bisimulation-invariant fragment of intuitionistic first-order logic. The corresponding standard translation is
43
and the proof combines intuitionistic analogues of Łoś’s theorem, elementary embeddings, countable saturation, and a Hennessy–Milner theorem for modally saturated bi-relational models (Groot et al., 30 Jun 2026). This gives 44 the same kind of invariance-theoretic profile that van Benthem’s theorem gives classical modal logic, but now in an intuitionistic first-order environment.
6. Variants, boundaries, and later reinterpretations
Later work makes clear that Simpson’s 45 is central but not exhaustive. One major semantic study places logics between 46 and 47 on a common CK-based hereditary semantics with an exploding world, identifies exact or sufficient frame conditions for 48, 49, and 50, and proves a precise conservativity theorem: for any 51, the extension 52 is conservative over the box-only logic 53 iff not both 54 and 55 are in 56 (Groot et al., 2024). In particular, 57 is not conservative over the box-only fragment, and the combination of diamond normalization with the Fischer–Servi interaction is exactly what breaks conservativity.
Another proposal isolates a logic strictly contained in 58 as a candidate “minimal normal intuitionistic modal logic.” This logic, 59, keeps the common 60-clause but replaces the usual Fischer–Servi diamond clause by a weaker clause using 61. It is sound and complete for all bi-relational frames under that clause, has the finite frame property, and is decidable. The paper proves that 62 is strictly weaker than 63, and uses countermodels to separate the Fischer–Servi interaction axioms from the minimal normal core (Balbiani et al., 26 Feb 2025).
A further reinterpretation is LIK, where both modalities are given local clauses
64
while heredity is recovered via forward and downward confluence. LIK is stronger than constructive WK, incomparable with 65, and admits bi-nested calculi with terminating proof search and finite countermodel extraction (Balbiani et al., 2024). This suggests that the semantic choice between local and hereditary readings of modality is not merely presentational.
At the opposite end, some distinctions collapse rather than proliferate. The constructive and intuitionistic versions of 66 coincide: for every formula 67,
68
This contrasts sharply with the 69 case, where the constructive and intuitionistic variants diverge even on diamond-free formulas (Pacheco, 2024).
Simpson’s framework has also been extended upward and outward. Explicit-term semantics for intuitionistic justification logic with native diamonds yields a realization theorem from a justification system JIK to 70, refining 71 by proof terms and 72 by satisfier terms while preserving the underlying bi-relational behavior (Marin et al., 30 Jun 2026). In a different direction, second-order intuitionistic tense logic recovers both diamonds from boxes once forward and backward modalities are present, using impredicative propositional quantification; the resulting labelled calculus is sound, complete, and cut-admissible (Becker et al., 5 Feb 2026). These developments preserve the Simpsonian core—bi-relational semantics, persistence, and non-dual modalities—while expanding the framework into explicit evidence, tense, and higher-order definability.
Taken together, these results show that “Simpson’s intuitionistic modal logics” names both a specific family centered on 73 and a durable research program. The family is unified by bi-relational semantics, primitive 74 and 75, and single-conclusion proof theory; it is differentiated by the treatment of 76, by how strongly 77 and 78 are linked, and by whether one works in mono-modal, tense, grammar-logical, constructive, or explicit-term settings (Tabatabai et al., 2022, Lyon, 27 Nov 2025, Groot et al., 2024).