---
title: Blok's Dichotomy Theorem
url: https://www.emergentmind.com/topics/blok-s-dichotomy-theorem
type: topic
---

# Blok's Dichotomy Theorem

Searching arXiv for the cited paper and closely related context.
Search query: title:"Degree of Kripke-incompleteness of Tense Logics"
Blok's Dichotomy Theorem is a result about the semantic indistinguishability of logics under Kripke semantics. In its classical form, for every normal modal logic \(L\in\mathsf{NExt}(\mathsf{K})\), the number of logics with exactly the same validating Kripke frames as \(L\) is either \(1\) or \(2^{\aleph_0}\). The theorem thus partitions normal modal logics into those whose Kripke semantics determines them uniquely and those that belong to a continuum-sized equivalence class under frame semantics. In the tense setting, the theorem has been generalized to the lattices \(\mathsf{NExt}(\mathsf{K}_t)\), \(\mathsf{NExt}(\mathsf{K4}_t)\), and \(\mathsf{NExt}(\mathsf{S4}_t)\), where the same dichotomy holds and where iterated splittings characterize the degree-\(1\) cases [2507.04533].

## 1. Formal setting and semantic degree

Blok's dichotomy concerns the relation between a logic and the class of Kripke frames validating it. In the tense framework considered in "Degree of Kripke-incompleteness of Tense Logics" [2507.04533], a frame is a pair \(F=(X,R)\) where \(X\) is a nonempty set and \(R\subseteq X\times X\), while a general frame is a triple \(F=(X,R,A)\) with \(A\subseteq\mathcal{P}(X)\) closed under \(\cap\), complementation, \(R[\cdot]\), and \(\breve{R}[\cdot]\). The language is bimodal, with future and past tense operators, and validity is defined in the usual way via valuations into admissible sets [2507.04533].

For a tense logic \(L\), Kripke completeness means \(L=\mathsf{Log}(\mathsf{Fr}(L))\), and the finite model property means \(L=\mathsf{Log}(\mathsf{Fin}(L))\). General-frame completeness holds uniformly: every tense logic is complete with respect to its class of general frames, and also with respect to its rooted general frames [2507.04533]. This separates the ubiquitous completeness of tense logics for general frames from the much more delicate issue of completeness for ordinary Kripke frames.

The central quantitative invariant is the degree of Kripke-incompleteness. For a lattice \(\mathcal{L}\) of logics and \(L\in\mathcal{L}\), the degree is defined by
\[
\mathsf{deg}_{\mathcal{L}}(L)=\bigl|\{L'\in\mathcal{L}: \mathsf{Fr}(L')=\mathsf{Fr}(L)\}\bigr|.
\]
Equivalently, it is the cardinality of the equivalence class of \(L\) under the relation \(L_1\equiv_{\mathsf{Fr}}L_2\iff \mathsf{Fr}(L_1)=\mathsf{Fr}(L_2)\) [2507.04533]. A logic is strictly Kripke-complete in \(\mathcal{L}\) exactly when this degree is \(1\). The paper also defines the analogous degree for the finite model property,
\[
\mathsf{df}_{\mathcal{L}}(L)=\bigl|\{L'\in\mathcal{L}: \mathsf{Fin}(L')=\mathsf{Fin}(L)\}\bigr|,
\]
and shows that in the tense lattices studied, \(\mathsf{deg}(L)=\mathsf{df}(L)\) [2507.04533].

The conceptual content of the dichotomy is therefore stark. Either frame semantics determines a logic uniquely, or frame semantics leaves room for continuum many distinct logics with exactly the same validating frames.

## 2. Classical theorem in modal logic

The classical theorem recalled in the tense generalization states that every modal logic \(L\in\mathsf{NExt}(\mathsf{K})\) has degree of Kripke-incompleteness either \(1\) or \(2^{\aleph_0}\) [2507.04533]. Thus, within the lattice of normal extensions of \(\mathsf{K}\), there are no intermediate cardinalities for frame-semantic equivalence classes.

In this formulation, degree \(1\) means that the logic is strictly Kripke-complete: among all normal modal logics, exactly one logic has its frame class. Degree \(2^{\aleph_0}\) means that there is a continuum of pairwise distinct normal modal logics sharing that same frame class [2507.04533]. A plausible implication is that Kripke semantics is maximally discriminating on one part of the lattice and maximally non-discriminating on the remainder.

The paper attributes the original proof to Blok (1978), who worked algebraically in \(\mathsf{NExt}(\mathsf{K})\). In that setting, union-splittings are exactly the consistent strictly Kripke-complete logics, while every other consistent logic has degree \(2^{\aleph_0}\) [2507.04533]. Another proof, based on relational semantics, is due to Chagrov–Zakharyaschev (1997, §10.5), also as reported there. The result is often regarded as a canonical theorem on Kripke-incompleteness because it connects a semantic phenomenon to a precise lattice-theoretic structure.

The tense generalization preserves the dichotomous cardinal structure but modifies the structural criterion for the degree-\(1\) side. In particular, for \(\mathsf{NExt}(\mathsf{S4}_t)\), the classical identification of degree \(1\) with union-splitting does not persist unchanged [2507.04533].

## 3. Splittings, union-splittings, and iterated splittings

The structural notions governing the theorem are splitting, union-splitting, and iterated splitting. For a base logic \(L_0\) and \(L_1,L_2\in\mathsf{NExt}(L_0)\), the pair \(\{L_1,L_2\}\) is a splitting pair in \(\mathsf{NExt}(L_0)\) if for every \(L\in\mathsf{NExt}(L_0)\), exactly one of \(L\subseteq L_1\) and \(L\supseteq L_2\) holds [2507.04533]. In that case \(L_1\) splits the lattice and \(L_2\) is the corresponding splitting logic \(L_0/L_1\).

A union-splitting in \(\mathsf{NExt}(L_0)\) is a logic expressible as a join of splittings. An iterated splitting is obtained by successive applications of the splitting operation:
\[
L=L_0/L_1/\cdots/L_n,
\]
where each \(L_i\) splits the appropriate residual lattice; the definition also counts \(L_0\) itself as an iterated splitting [2507.04533]. These notions are not merely syntactic. The underlying lattice theory states that an element splits the lattice iff it is completely \(\wedge\)-prime, and an element is a splitting iff it is completely \(\vee\)-prime [2507.04533].

In the classical modal case, union-splittings characterize the degree-\(1\) logics. In the tense setting, the picture is more nuanced. For \(\mathsf{NExt}(\mathsf{K}_t)\) and \(\mathsf{NExt}(\mathsf{K4}_t)\), iterated splittings coincide with union-splittings and with the strictly Kripke-complete logics. For \(\mathsf{NExt}(\mathsf{S4}_t)\), by contrast, iterated splittings still characterize strict Kripke-completeness, but union-splittings do not capture all degree-\(1\) cases [2507.04533].

This shift is one of the principal conceptual refinements in the tense generalization. It suggests that iterated splitting is the structurally stable notion behind strict Kripke-completeness once one moves from unimodal to bimodal tense lattices.

## 4. Tense-logical generalization

The main theorem of "Degree of Kripke-incompleteness of Tense Logics" is that Blok's dichotomy extends from \(\mathsf{NExt}(\mathsf{K})\) to three lattices of tense logics: \(\mathsf{NExt}(\mathsf{K}_t)\), \(\mathsf{NExt}(\mathsf{K4}_t)\), and \(\mathsf{NExt}(\mathsf{S4}_t)\) [2507.04533]. In each case, every logic has degree either \(1\) or \(2^{\aleph_0}\), and in each case the degree of Kripke-incompleteness coincides with the corresponding finite-model degree.

The main characterizations can be organized as follows.

| Lattice | Degree-\(1\) logics | Dichotomy |
|---|---|---|
| \(\mathsf{NExt}(\mathsf{K}_t)\) | exactly the union-splittings, equivalently exactly the iterated splittings | \(\mathsf{deg}(L)=\mathsf{df}(L)\in\{1,2^{\aleph_0}\}\) |
| \(\mathsf{NExt}(\mathsf{K4}_t)\) | exactly the union-splittings, equivalently exactly the iterated splittings | \(\mathsf{deg}(L)=\mathsf{df}(L)\in\{1,2^{\aleph_0}\}\) |
| \(\mathsf{NExt}(\mathsf{S4}_t)\) | exactly the iterated splittings | \(\mathsf{deg}(L)=\mathsf{df}(L)\in\{1,2^{\aleph_0}\}\) |

In \(\mathsf{NExt}(\mathsf{K}_t)\), Kracht's result yields a unique splitting pair: a logic \(L\) splits the lattice iff \(L=\mathsf{Log}(\bullet)\), where \(\bullet\) is the one-point frame [2507.04533]. Writing
\[
L^*:=\mathsf{K}_t/\mathsf{Log}(\bullet),
\]
the paper proves \(L^*=\mathsf{K}_t\oplus(\top\vee\blacksquare\top)\), and shows that the only iterated splittings are \(\mathsf{K}_t\) and \(L^*\) [2507.04533]. Both have degree \(1\), and every other logic in the lattice has degree \(2^{\aleph_0}\).

The same pattern holds in \(\mathsf{NExt}(\mathsf{K4}_t)\). Again there is a unique splitting pair, again one obtains a corresponding \(L^*=\mathsf{K4}_t/\bullet\), and again the only iterated splittings are \(\mathsf{K4}_t\) and \(L^*\) [2507.04533]. These are exactly the strictly Kripke-complete logics in the lattice.

The \(\mathsf{S4}_t\) case is more intricate. Kracht's theorem gives exactly two splitting pairs in \(\mathsf{NExt}(\mathsf{S4}_t)\): \(\{\mathsf{Log}(Ch_2),\mathsf{S5}_t\}\) and \(\{\mathsf{Log}(Ch_1),L\}\), where \(L\) is the least reflexive-transitive tense logic whose rooted frames have no nontrivial chains [2507.04533]. Here the iterated splittings are precisely
\[
\mathsf{NExt}(\mathsf{S5}_t)\cup\{\mathsf{S4}_t\},
\]
and these are exactly the degree-\(1\) logics. The paper emphasizes that only \(\mathsf{S4}_t\), \(\mathsf{S5}_t\), and \(L\) are union-splittings, so degree \(1\) is strictly broader than union-splitting in this lattice [2507.04533].

## 5. Proof architecture and frame constructions

The proofs combine splitting arguments for the degree-\(1\) side with continuum-size constructions for the degree-\(2^{\aleph_0}\) side. The general strategy is modeled on the relational proof of Blok's theorem by Chagrov–Zakharyaschev, but it is adapted to the bimodal tense setting and to transitive and reflexive-transitive frame classes [2507.04533].

A central technical device is reflective unfolding. Given a frame \(F\) and points \(w,u\), the paper defines reflective unfoldings \(F^n_{w,u}\) by repeated combination of disjoint copies of \(F\). There is also a transitive variant using a transitive combination operation. The crucial property is the existence of a natural \(t\)-morphism \(\pi_n:F^n_{w,u}\twoheadrightarrow F\), and in the transitive case the corresponding map from \((F^n_{w,u})^t\) onto \(F\) [2507.04533]. This permits the transfer of semantic information from the original finite rooted frame to arbitrarily large finite rooted unfoldings.

The resulting corollaries provide finite rooted frames of arbitrarily large reachability degree that preserve satisfaction of a chosen counterexample condition. If a formula is satisfiable in some finite rooted frame other than \(\bullet\), then for each \(n\in\omega\) there is a finite rooted frame with reachability degree at least \(n\) satisfying it. In the transitive setting, a corresponding statement holds for rooted non-symmetric \(\mathsf{K4}_t\)- or \(\mathsf{S4}_t\)-frames [2507.04533]. These large-depth witnesses are then used to encode continuum many different logics without altering the Kripke frame class.

For a logic \(L\) that is not an iterated splitting, the construction proceeds by choosing a formula outside the maximal degree-\(1\) region, selecting a sufficiently deep finite rooted countermodel, and then combining it with specially designed general frames \(F'_I\) indexed by subsets \(I\subseteq\mathbb{Z}^+\) [2507.04533]. The combined general frames \(F_I\) yield logics
\[
L_I:=L\cap\mathsf{Log}(F_I),
\]
with two key properties: \(\mathsf{Fr}(L_I)=\mathsf{Fr}(L)\) for all \(I\), yet \(L_I\neq L_J\) whenever \(I\neq J\). In the \(\mathsf{K}_t\) and \(\mathsf{K4}_t\) cases, the coding uses formulas such as \(\gamma_n\) and \(\gamma_m^*\); in the \(\mathsf{S4}_t\) case, the construction is more elaborate and uses frames generalizing the Rieger–Nishimura ladder together with defining formulas \(\varphi_{x_0},\varphi_{a_n},\varphi_{c_n}\) [2507.04533].

Since there are continuum many subsets \(I\subseteq\mathbb{Z}^+\), this produces continuum many pairwise distinct logics with the same frame class, giving \(\mathsf{deg}(L)\ge 2^{\aleph_0}\). As the ambient lattices have cardinality at most \(2^{\aleph_0}\), equality follows [2507.04533].

## 6. Structural consequences and comparison with the modal case

Across the three tense lattices, the general outcome is that iterated splittings are exactly the strictly Kripke-complete logics, while all other logics have degree \(2^{\aleph_0}\) [2507.04533]. This yields a unified Blok-type characterization for tense logics:
for \(L_0\in\{\mathsf{K}_t,\mathsf{K4}_t,\mathsf{S4}_t\}\) and \(L\in\mathsf{NExt}(L_0)\), \(L\) is strictly Kripke-complete iff \(L\) is an iterated splitting in \(\mathsf{NExt}(L_0)\); otherwise \(\mathsf{deg}_{L_0}(L)=2^{\aleph_0}\) [2507.04533].

The relation to the original modal theorem is twofold. First, the cardinal dichotomy itself is preserved unchanged. Second, the structural criterion for degree \(1\) is preserved exactly in the \(\mathsf{K}_t\) and \(\mathsf{K4}_t\) lattices, where union-splittings and iterated splittings coincide. The main divergence appears in \(\mathsf{NExt}(\mathsf{S4}_t)\), where there are infinitely many iterated splittings inside \(\mathsf{NExt}(\mathsf{S5}_t)\), but only \(\mathsf{S4}_t\), \(\mathsf{S5}_t\), and \(L\) are union-splittings [2507.04533]. Thus the naive transfer of the modal slogan “degree \(1\) iff union-splitting” fails in the reflexive-transitive tense setting.

The paper identifies several sources of additional difficulty in tense logic. The presence of future and past modalities creates more complex frame configurations; transitivity and reflexivity constrain the unfolding constructions; and in the \(\mathsf{S4}_t\) case, the logic is not finitely transitive, which complicates the use of master-modality methods and necessitates finer control over width, depth, and related frame parameters [2507.04533]. The proof accordingly relies on transitivity-preserving combinations, large reachability degree, and formulas such as \(\mathsf{bz}_n,\mathsf{bw}^\pm_n,\mathsf{alt}^\pm_n,\mathsf{bd}_n\) for bound control [2507.04533].

A plausible implication is that the tense generalization does not merely reproduce the modal theorem in a richer syntax; it isolates which lattice-theoretic notions remain invariant under the passage from one modality to two and which do not.

## 7. Significance, limitations, and open questions

The theorem has three principal consequences in the tense setting. First, it extends Blok's dichotomy to major lattices of tense logics, namely \(\mathsf{NExt}(\mathsf{K}_t)\), \(\mathsf{NExt}(\mathsf{K4}_t)\), and \(\mathsf{NExt}(\mathsf{S4}_t)\) [2507.04533]. Second, it establishes that in these lattices the degree of Kripke-incompleteness coincides with the degree determined by finite frames: \(\mathsf{deg}(L)=\mathsf{df}(L)\) [2507.04533]. Third, it identifies iterated splitting as the robust structural marker of strict Kripke-completeness.

The result also clarifies a common misunderstanding. Kripke completeness and strict Kripke-completeness are not the same notion. A Kripke-complete logic may still fail to be uniquely determined by its validating frames, whereas strict Kripke-completeness requires semantic uniqueness within the lattice. Blok's dichotomy concerns the latter notion, measured via the size of the frame-semantic equivalence class [2507.04533].

Several questions remain open. The paper points to possible extensions to other finitely transitive tense logics such as \(\mathsf{S4.2}_t\) and \(\mathsf{S4.3}_t\), where Kracht showed that there are infinitely many splittings [2507.04533]. It also raises the possibility of anti-dichotomy phenomena analogous to those studied elsewhere for degrees of the finite model property in intuitionistic and transitive modal logics. More generally, it asks for a broader account of the relation among union-splittings, iterated splittings, and strictly Kripke-complete tense logics across wider classes of tense lattices [2507.04533].

Within the scope established so far, the theorem yields a precise and uniform picture. In the principal lattices of tense logic treated, semantic equivalence under Kripke frames is either trivial or continuum-sized, and the exact frontier between those two regimes is given by iterated splitting.

Source: https://www.emergentmind.com/topics/blok-s-dichotomy-theorem