💻 computer science

A taxonomy for controlling (in)consistency

This paper introduces the hierarchy of Logics of Controlled Consistency (L $_n^k$ ), a two-dimensional taxonomy of Logics of Formal Inconsistency that models varying degrees of paraconsistent commitment from skepticism to dogmatism, and provides sound and complete semantic accounts for this family and its specific extensions using swap structures, twist structures, and RNmatrices.

Original authors: Marcelo E. Coniglio, Rafael Ongaratto

Published 2026-04-22

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Marcelo E. Coniglio, Rafael Ongaratto

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are a detective trying to solve a mystery, but the world you are investigating is a bit messy. Sometimes, witnesses lie. Sometimes, two witnesses give contradictory stories. Sometimes, a witness is so unreliable that you can't even trust their claim about whether they are lying.

This paper is about building a toolbox of logic to handle these messy situations. The authors, Marcelo Coniglio and Rafael Ongaratto, introduce a new family of logical systems called Logics of Controlled Consistency (LCC).

Here is the breakdown of their big idea, using simple analogies.

1. The Problem: The "Infinite Regress" of Doubt

In standard logic (like math), if you say "A is true" and "A is false," the whole system crashes. It's like a computer program hitting a fatal error. But in real life, contradictions happen all the time without the world ending.

Paraconsistent logic allows us to handle contradictions without crashing. However, a new question arises: How much can we doubt?

The Skeptic's View: "I don't trust this witness. And I don't trust the method we used to check the witness. And I don't trust the method used to check the method..." This goes on forever. This is like a dog chasing its own tail.
The Dogmatist's View: "I trust my witness completely. I trust my method completely. I don't even question if my method is reliable." This is rigid and ignores reality.
The Pragmatist's View: "I will trust the witness, but I admit the method might be shaky. However, I won't question the method of the method because that's too much work for this specific case."

The authors ask: Can we build a logic that lets us choose exactly how deep our doubt goes?

2. The Solution: The "Depth of Doubt" Ladder

The authors created a hierarchy of logics (a ladder) called $L^k_n$ . Think of this as a dial you can turn to control how many layers of "reliability checks" you allow.

The "n" Dial (Depth of Consistency): This controls how many times you can question the reliability of a statement before you stop.
- Level 0 (The Dogmatist): You check if a statement is true. If it's contradictory, you stop. You never ask, "Is the check itself reliable?"
- Level 1 (The Skeptic): You check the statement. You also check if the check is reliable. But you stop there. You don't ask, "Is the check of the check reliable?"
- Level 2, 3, etc.: You can go deeper. You can question the reliability of the reliability of the reliability... up to a specific limit you set.
The "k" Dial (Strength of Negation): This controls how strictly you handle "Not."
- In some logics, "Not Not A" might not equal "A" (double negation is fuzzy).
- In stronger logics, "Not Not A" definitely equals "A."
- As you go up the "k" ladder, your logic becomes more like standard, classical logic, but with the safety net of paraconsistency.

3. The "Swap Structure" Analogy: The Truth Dashboard

To make these logics work, the authors use something called Swap Structures. Imagine a dashboard in a car with three lights for every statement:

Is it True? (Green)
Is it False? (Red)
Is the reliability of this statement trustworthy? (Yellow)

In the authors' new systems, these lights can be on or off in complex combinations.

Scenario A: The light is Green (True), but the Yellow light (Reliability) is blinking. This means "It's true, but we aren't sure if our source is good."
Scenario B: The light is Green, but the Yellow light is Red. This means "It's true, but we know our source is lying."

The authors show that by changing the rules of how these lights interact (the "Swap"), they can create logics that represent Skepticism (where the Yellow light is always blinking) or Dogmatism (where the Yellow light is always solid Green).

4. The New Discovery: LFI3 (The 5-Valued Logic)

The paper also introduces a specific new logic called LFI3.

Most simple paraconsistent logics have 3 values: True, False, and "Both/Contradictory."
The authors realized that sometimes 3 isn't enough. They built a 5-valued logic.
Think of it like a color palette. Instead of just Red, Blue, and Purple, you have Red, Blue, Purple, Pink, and Orange.
This allows for much more nuance. For example, you can distinguish between "A contradiction that is reliable" and "A contradiction that is unreliable."

They proved that this new 5-valued logic is actually a "super-set" of an older, famous logic (LFI1). It's like upgrading from a black-and-white TV to a high-definition one; you can see everything the old one could, plus much more detail.

5. Why Does This Matter? (The Real-World Application)

The authors argue that different situations require different levels of doubt.

A Courtroom: A lawyer might say, "The witness is unreliable." The prosecutor might say, "The lawyer's claim about the witness is also unreliable." But the judge might say, "Stop! We can't question the judge's ability to judge the lawyer's claim about the witness." The judge needs a logic that stops the questioning at Level 2.
Scientific Research: A scientist might doubt a result, doubt the experiment, but eventually, they must stop doubting and accept the data to move forward.
Everyday Life: If your friend says, "I saw a ghost," you might doubt them. If they say, "I know I'm crazy, but I saw a ghost," you might doubt their sanity. But you probably won't doubt the logic of their doubt.

Summary

This paper is a manual for building custom logic systems.

It gives us a hierarchy (a ladder) to control how deep our skepticism goes.
It provides mathematical tools (Swap Structures and RNmatrices) to prove these systems work.
It introduces a new, more detailed logic (LFI3) that handles complex contradictions better than older systems.

In short, the authors are saying: "Don't just use one size fits all logic. Depending on whether you are a skeptic, a dogmatist, or a pragmatist, you can now choose the exact logical tool that fits your situation."

1. Problem Statement

The paper addresses a gap in the hierarchy of Logics of Formal Inconsistency (LFIs). While LFIs introduce a consistency operator ( $\circ$ ) to recover classical reasoning in the presence of contradictions, existing systems often present a binary choice regarding the "depth" of consistency:

Extreme Skepticism (e.g., $mbCciw$): Consistency can be inconsistent at any level of iteration (meta-consistency is never guaranteed).
Extreme Dogmatism (e.g., $mbCci$): Consistency is guaranteed to be consistent at all levels (any judgment of consistency is itself consistent).

The authors argue that real-world reasoning often requires controlled consistency, where the reliability of criteria can be questioned up to a specific depth ( $n$ ) but must stabilize thereafter. Furthermore, existing hierarchies (like da Costa's $C_n$ ) rely on definable consistency operators, whereas LFIs require a primitive operator. The paper seeks to construct a two-dimensional hierarchy of LFIs that allows for:

Controlled Iteration ( $n$ ): Limiting the depth at which consistency can be contradictory.
Strengthened Negation ( $k$ ): Increasing the determinism of the negation operator.

2. Methodology

The authors employ a multi-faceted methodological approach combining algebraic semantics, non-deterministic matrices, and axiomatic extensions:

Hierarchy Construction ( $L^k_n$ ): They define a family of logics $L^k_n$ $L_{n}^{k}$ where:
- $n$ controls the iteration of the consistency operator (via axiom $\circ^{n+2}\alpha$ ).
- $k$ controls the strength of negation (via axioms like $\neg\neg\alpha \leftrightarrow \alpha$ and $\neg\circ^j\neg\alpha \leftrightarrow \neg\circ^j\alpha$ ).
Semantics via Swap Structures: For the base cases, they utilize swap structures (non-deterministic matrices where elements are tuples representing truth values of a formula and its related operators). This allows for a fine-grained representation of "snapshots" of truth, consistency, and inconsistency.
Twist Structures: For the 5-valued logic $LFI3$, they transition from non-deterministic swap structures to deterministic twist structures to define a standard logical matrix.
RNmatrices (Restricted Nmatrices): To handle the general hierarchy ( $L^k_n$ ) without an explosion in the number of truth values, they use RNmatrices. This restricts the set of valid valuations on a base Nmatrix to satisfy specific conditions (e.g., if $\circ^n\alpha$ is true/false, then $\circ^{n+1}\alpha$ must be true).
Axiomatization: They provide Hilbert-style calculi for specific members of the hierarchy, proving soundness and completeness against their respective semantic models.

3. Key Contributions

A. The $L^k_n$ Hierarchy

The paper introduces a unified taxonomy of Logics of Controlled Consistency (LCC):

Dimension $n$ (Control Level): Determines the maximum depth of inconsistency allowed for the consistency operator.
- $L^k_0$ (e.g., $mbCci$): Consistency is always consistent ( $\circ\circ\alpha$ ).
- $L^k_1$ : Consistency can be inconsistent, but consistency of consistency is guaranteed ( $\circ\circ\circ\alpha$ ).
- $L^k_n$ : Consistency can be inconsistent up to $n$ iterations, but $\circ^{n+2}\alpha$ is valid.
Dimension $k$ (Negation Strength):
- $L^0_n$ : Basic negation.
- $L^1_n$ : Adds double negation elimination ( $\neg\neg\alpha \leftrightarrow \alpha$ ).
- $L^2_n$ : Adds interaction between negation and consistency ( $\neg\circ\neg\alpha \leftrightarrow \neg\circ\alpha$ ).
- Higher $k$ : Further strengthens the interaction between negation and iterated consistency.

B. The Logic $LFI3$

The authors introduce $LFI3$, a new 5-valued LFI that serves as a sublogic of $LFI1$ (the 3-valued logic $J3$ ).

Significance: It resolves the limitation of 3-valued logics by distinguishing more cases of consistency and inconsistency.
Semantics: Defined via a twist structure over a 5-element domain $\{F, f, b, t, T\}$ (False, False-Contradictory, Contradictory, True-Contradictory, True).
Recovery Operators: $LFI3$ defines new consistency operators ( $\circ^*$ and $\circ^\#$ ) that allow the recovery of Classical Logic and $LFI1$ respectively within the system.
Maximality: It is proven that $LFI3$ is maximal with respect to $LFI1$; adding any theorem from $LFI1$ that is not in $LFI3$ collapses the system into $LFI1$.

C. Philosophical Taxonomy

The hierarchy maps to philosophical stances:

Skepticism: Represented by $mbCciw$ (no iteration limit on inconsistency).
Dogmatism: Represented by $L^0_0$ ($mbCci$), where criteria are never questioned beyond the first level.
Pragmatism: The intermediate levels ( $L^k_n$ ) allow for context-dependent reasoning (e.g., legal testimony or detective games) where the reliability of a criterion can be questioned up to a fixed depth $n$ , but not infinitely.

4. Key Results

Soundness and Completeness: The authors prove that the axiomatic systems $L^k_n$ are sound and complete with respect to their swap structure semantics and induced RNmatrices.
Fixed Point Theorem: They prove that for $n \ge 0$ and $m \ge n+2$ , the logic $L^{n+1}_n$ is a fixed point, meaning $L^{n+1}_n = L^m_n$ . Adding further axioms for higher iterations of negation does not change the logic once the consistency depth is fixed.
Sublogic Relations:
- $LFI3$ is a sublogic of $LFI1$.
- $LFI3$ is a sublogic of Classical Logic (CPL).
- $LFI1$ can be axiomatized as $LFI3 + (\circ\circ\alpha)$ .
Recovery of Classicality: The paper demonstrates how specific consistency operators ( $\circ^*$ and $\circ^\#$ ) can recover CPL and $LFI1$ reasoning within $LFI3$, respectively.

5. Significance

Unification: The paper unifies various known LFIs (such as $mbCci$, $Cbr$, $Cie$, and $LFI1$) into a single, coherent two-dimensional hierarchy.
Beyond 3-Valuedness: It provides a robust framework for constructing many-valued LFIs with more than three values (demonstrated by the 5-valued $LFI3$), addressing a historical challenge in paraconsistency.
Pragmatic Application: By formalizing "degrees of paraconsistent commitment," the hierarchy offers a logical tool for modeling real-world scenarios where the depth of skepticism or dogmatism varies by context (e.g., legal reasoning, scientific inquiry).
Semantic Innovation: The use of RNmatrices to handle the general hierarchy avoids the complexity of defining massive truth tables for high $n$ , offering a scalable semantic approach for future research in LFIs.

In summary, the paper successfully constructs a flexible, mathematically rigorous taxonomy for paraconsistent logics that allows for the precise control of inconsistency depth and negation strength, bridging the gap between abstract logical theory and pragmatic philosophical application.