A version-one arXiv preprint dated 20 August 2026 presents a probability framework for LET +K, a logical system that can represent gaps, gluts and reliability. Its central result is formal: the paper defines several probability descriptions for the system and reports that semantic and syntactic forms of Jeffrey updating coincide in both the single-valued and six-valued cases.
The work is not a study of people, observed events or measured frequencies. It develops definitions, logical models, probability measures and assignment states, with a completeness construction based on formal assignments rather than an empirical sample. Its conclusions therefore concern what follows from the paper’s assumptions, not how often contradictions occur in the world or how people update beliefs.
A logic built for mixed information
LET +K is described as an expansion of FDE and classical propositional logic. Its formal semantics uses a six-valued logical matrix, M6, with T, t and b as designated values. This gives the system more categories than a simple true-or-false description, allowing the paper to distinguish different kinds of information within its model.
For each formula, the twist-model state space is divided into six disjoint regions: reliable belief, belief, conflict, disbelief, reliable disbelief and uncertainty. The labels separate positive and negative information from judgments about reliability, while keeping unresolved information distinct from outright conflict.
In plain terms, a gap is a case where the available information does not support a straightforward conclusion, while a glut is a case where conflicting information is present. The paper’s framework is built to represent both conditions instead of forcing every formula into a single, consistent category. Conflict and uncertainty are therefore treated as named regions of the model.
Putting probability on the regions
The probability construction follows the distinctions in that model. The paper defines one-, three- and six-dimensional probability functions through both axiomatic and twist-model semantic approaches. These are different formal ways of recording the information associated with a formula while preserving the structure of LET +K.
The representations known as P1, P3 and P6 are connected through inverse mappings. In practical terms, the paper says they can be translated back and forth within the stated framework, so the choice of representation does not create three unrelated probability systems.
The paper reports soundness and completeness for the axiomatically and semantically defined probability functions. In this setting, soundness means the formal rules stay within the intended semantics, while completeness means the semantic consequences can be captured by the formal system under the paper’s definitions. These are mathematical properties of the framework, not results from statistical estimation or hypothesis testing.
The completeness construction uses a finite set of assignments from propositional variables to the six truth values. That finite assignment space supplies the setting for the formal construction; it is not a population sample and does not estimate a real-world frequency.
Updating without flattening conflict
The paper then turns to Jeffrey updating, a way of revising probability weights when a target belief or region is changed without specifying every underlying detail. In the single-valued case, the semantic update is well-defined when the original positive extension has probability strictly greater than 0 and strictly less than 1. The update sets the positive mass to λ.
The authors report that the semantic single-valued Jeffrey update and its syntactic counterpart are equivalent. That result connects an update described through the model’s interpretation with one described through the rules of the formal language.
The six-valued version works region by region. An admissible target vector can expand or contract each partition region, while any region that began with zero mass remains at zero. The rule can therefore redistribute probability across reliable belief, belief, conflict, disbelief, reliable disbelief and uncertainty while respecting the framework’s constraints.
The paper also reports equivalence between the semantic and syntactic forms of the six-valued Jeffrey update. The correspondence holds under the stated admissibility conditions, making the two descriptions agree within the formal system.
The order of updates still matters. Six-valued Jeffrey updating is not commutative in general, meaning that applying one update and then another can produce a different result from reversing the order. The paper also treats six-valued Bayesian updating as an extremal special case of Jeffrey updating.
The boundary of the result
Taken together, the translation and equivalence results form a formal bridge between model-based descriptions and rule-based descriptions of probability and updating. The bridge is established through definitions, algebraic derivations, logical-matrix and twist-structure constructions, and proofs of soundness, completeness and inverse translations.
The conclusions are conditional on the LET +K axioms, the six-valued logical matrix, the twist structures and the probability-measure assumptions used in the paper. The completeness construction is explicitly finite, so the result does not by itself establish how the framework would behave in every possible model or setting.
The paper does not show that LET +K probabilities describe human beliefs or real-world event frequencies. It does not test the update rules against observed data, compare their predictive accuracy with classical probability, or establish that the reliability operator improves practical inference or decision-making. No causal evidence is reported.
For researchers in nonclassical logic, formal probability and belief revision, the contribution is a set of formal constructions and equivalence results rather than an empirical prediction. The preprint extends the paper’s logical and probabilistic machinery within its stated assumptions; it does not claim to validate that machinery outside them.
Paper data and sources
Original title: Probabilities beyond Belnap-Dunn logic: dealing with gaps, gluts and reliability
Authors: Verónica Borja Macias, Marcelo E. Coniglio, Alejandro Hernández-Tello
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text