Skip to content

Quasi-conjunction of trivalent conditionals

Quasi-conjunction combines trivalent conditional objects by conjoining their material implications under the union of their conditioning events.

Version
v1 · 2026-10-31 · History
Domain-specific #
14561
Domain group
Formal Sciences
Origin domain
Mathematics
Subdomains
Mathematical Logic, Probability Theory → Mathematics

Core Idea

A basic conditional a|b is true where both a and b hold, false where b holds without a, and inapplicable where b does not hold. Quasi-conjunction combines two such objects over a Boolean event algebra by

(a|b) ∧Q (c|d) = ((b→a) ∧ (d→c)) | (b∨d).

Here → is material implication in the underlying event algebra. The combined conditional is assessed on the union of the two conditioning events. The formula defines an operation on conditional objects without assigning probabilities. Its use in probability-coherent entailment needs additional assumptions.[1][2]

How would you explain it like I'm…

Two Promises Together

Think of a promise like 'If it's sunny, we go to the park.' If it's sunny and we go, the promise is kept. If it's sunny and we don't go, it's broken. If it isn't sunny, the promise just doesn't count today. Quasi-conjunction puts two promises together: the pair doesn't count only when neither 'if' happens, and it's broken if any promise that counts gets broken.

Joining If-Then Rules

An 'if' statement like 'if b, then a' can be true, false, or not apply at all: it is true when b and a both happen, false when b happens without a, and it doesn't count when b doesn't happen. Quasi-conjunction joins two of these 'if' statements into one. The joined statement counts whenever at least one of the two 'if' parts happens. When it counts, it is true only if neither statement is broken, meaning no applicable 'if' part happened without its matching result. It is just a rule for combining statements, not a way of computing chances.

Combining Three-Valued Conditionals

A conditional a|b ('a given b') has three possible values: true when a and b both hold, false when b holds but a does not, and inapplicable when b fails. Quasi-conjunction combines two such conditionals into one: (a|b) ∧Q (c|d) = ((b→a) ∧ (d→c)) | (b∨d), where → is ordinary 'if-then' (material implication). So the combined conditional applies whenever b or d holds. When it applies, it is false exactly when one of the original conditionals is false; a conditional that does not apply in that case counts as harmless because its material implication is true. This defines an operation on conditionals; it does not assign probabilities, and using it to reason about probability needs extra assumptions.

 

In trivalent (three-valued) treatments of conditionals, a conditional event a|b over a Boolean algebra of events takes the value true on a∧b, false on ¬a∧b, and void (inapplicable) on ¬b. Quasi-conjunction is the binary operation (a|b) ∧Q (c|d) = ((b→a) ∧ (d→c)) | (b∨d), where → is material implication in the underlying event algebra. The result's conditioning event is the union b∨d, so it is applicable whenever at least one component is. Within that region, the consequent (b→a)∧(d→c) is false exactly when some applicable component is false, because material implication is vacuously true where its antecedent fails. This defines an algebraic operation on conditional objects only; it assigns no probabilities, and using it in probabilistic or coherence-based entailment requires further assumptions.

Structural Signature

Sig role-phrases:

  • Event base: A Boolean algebra supplies events and its complement, conjunction and disjunction.
  • Conditional inputs: a|b and c|d each have true, false and inapplicable regions; the consequent outside its conditioning event does not affect the conditional object.
  • Contextual combination: Quasi-conjunction intersects b→a and d→c and conditions the result on b∨d.
  • Conditional output: The result is again a basic trivalent conditional object, even where its displayed numerator has events outside the new context; only its restriction to that context determines the object.
  • Separate inference layer: A p-entailment conclusion requires a p-consistent family and coherence-based reasoning, rather than following from the Q formula alone.[1][2]

What It Is Not

This is not the Boolean conjunction of a constructed Boolean algebra of conditionals. Flaminio, Godo and Hosni distinguish quasi-conjunction ∧Q from an interval-based conjunction ∧I on the same trivalent conditional family and report that neither corresponding algebra is Boolean. These two operations are competing rules, not two instances of Q. Nor is Q a probability ratio or an observed expert-system implementation.[1][3]

Scope of Application

The two examples below are formal realizations over different event bases. The first is a transparent four-outcome calculation derived here from the published formula. The second is a source-reported illustrative knowledge base with four logically independent party propositions and a separate coherence-based p-entailment argument. Neither is empirical evidence of deployed software or human reasoning.[1][2]

Clarity

For each input, name the antecedent and consequent before combining them. The output's context is b∨d, not b∧d; an event in the raw implication numerator outside b∨d has no effect on the conditional. State separately if a later inference also invokes p-consistency and inclusion.

Manages Complexity

The Q rule keeps a combination of two conditional assertions within the same basic conditional-object form. This is useful for tracing which conditional follows from selected premises in a formal rule base. It does not determine a probability assignment or validate a whole rule base on its own.[1][2]

Abstract Reasoning

The formula is invariant under changing a consequent outside its antecedent: a|b and (a∧b)|b denote the same trivalent conditional. In the output, the meaningful numerator is therefore ((b→a)∧(d→c))∧(b∨d). This explains why an apparently larger raw implication set can name a smaller normalized output without changing the conditional.[1]

Knowledge Transfer

The portable question is whether another formalism combines conditional claims by preserving each claim's applicability region and specifying a new joint context. Carrying Q itself into another setting requires its Boolean event operations and trivalent conditional semantics; shared “if–then” wording is insufficient. A probability-based inference must also carry its own coherence hypotheses.[1][2]

Examples

Canonical: derived four-outcome calculation

In the event algebra P({1,2,3,4}), take ({1}|{1,2}) and ({2}|{2,3}). Their material implications are {1,3,4} and {1,2,4}. Their intersection is {1,4}, while the conditioning union is {1,2,3}. The result is ({1,4}|{1,2,3}), equivalently ({1}|{1,2,3}) after restricting the consequent to its context. This specific calculation is derived for this entry from Flaminio and colleagues' Q formula; it is not a worked example printed by them. No probability is assigned.[1]

Mapped back: The four outcomes supply the event base; the two displayed conditionals are the inputs; material implication and intersection supply the raw numerator; the union supplies the output context. The output remains one trivalent conditional.

Applied/practice: published party-rule illustration

Gilio and Sanfilippo's Example 2 introduces logically independent propositions L (Linda goes to the party), S (Steve goes), G (the party is great) and N (the party is noisy). In their stipulated p-consistent family F, select S|L and ¬N|(L∧S). Their quasi-conjunction is (S∧¬N)|L. In their coherence-based framework, its inclusion in ¬N|L, together with p-consistency and their Quasi And rule, supports the stated p-entailment of ¬N|L from F. The paper presents an illustrative formal knowledge base, not a deployed party-prediction system.[2]

Mapped back: The four independent propositions generate a Boolean event base distinct from the four-outcome base; the selected rules are the Q inputs; Q produces a combined conditional on context L. The p-entailment conclusion belongs to the paper's additional coherence and inclusion argument, not merely to the Q computation.

Structural Tensions

T1 — Choice of conditional conjunction. On the same trivalent conditional family, ∧Q and ∧I can give different results. For the derived four-outcome inputs above, the published ∧I formula gives (∅|{2,3}), whereas Q gives ({1}|{1,2,3}). Choosing Q retains the specific union-context composition used in the cited nonmonotonic-rule analysis; interval conjunction follows a different operation rule. The contrast establishes a definitional choice, not an empirical ranking.[1]

T2 — Composition and entailment. Q can be computed without a probability. The published party example's p-entailment needs a p-consistent family, coherent assessments and inclusion reasoning. Dropping those assumptions leaves the Q output intact but removes the stated basis for that p-entailment.[2]

Diagnostic: Is the task only to combine conditional objects, or also to justify a probabilistic consequence under explicit coherence assumptions?

Structural–Framed Character

The operation has a portable structural skeleton—conditional inputs, a context-sensitive combination and a conditional output—but its identity is mathematically framed by Boolean event operations and trivalent applicability. In the cited research, the Q rule is specified as a formal choice among conditional conjunctions and then used for a coherence-based inference analysis. Its merits are judged by such mathematical tasks as representing conditional combinations and tracing qualified entailment; these are task-dependent criteria, not observations of deployed performance. Ordinary conditional language may prompt recognition of the pattern, but importing Q requires the stated algebra and semantics. Its character: a formally specified composition rule whose inference use is conditional on a separately supplied probabilistic framework.[1][2]

Structural Core vs. Domain Accent

The structural core is contextual composition: each assertion contributes its material implication, and the output applies wherever either antecedent applies. Boolean operations, trivalent inapplicability and the exact Q formula are constitutive mathematical commitments here. Other fields may also combine context-bound assertions, but that resemblance alone does not transfer this operation or its inference rules.

Boolean Conditional-Event Algebra Constructions is a neighboring formalism with a Boolean constructed carrier and qualified probability extension. It does not supply Q as its Boolean conjunction. Default Logic also studies nonmonotonic inference, but its Reiter-style extension semantics is distinct from contextual combination of trivalent conditionals.

Neighborhood in Abstraction Space

Quasi-conjunction of trivalent conditionals sits in a moderately populated region (49th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Formal Models & Logical Foundations (37 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-31

Not to Be Confused With

Do not treat trivalent conditional values as a claim that all three-valued logics use Q. Do not infer a probability ratio, universal default conclusion, or empirical rule-system behavior from the Q formula. The historical associations with Schay and Calabrese follow Flaminio and colleagues' comparison; their original works and the Goodman–Nguyen work were not directly examined for this account.[1][3]

References

[1] Tommaso Flaminio, Lluís Godo and Hykel Hosni, “Boolean algebras of conditionals, probability and logic,” Artificial Intelligence 286 (2020), 103347, §8.1. https://doi.org/10.1016/j.artint.2020.103347; author full text. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k

[2] Angelo Gilio and Giuseppe Sanfilippo, “Quasi Conjunction, Quasi Disjunction, T-norms and T-conorms: Probabilistic Aspects,” Information Sciences 245 (2013), 146–167. https://doi.org/10.1016/j.ins.2013.03.019; Definitions 3–5, Corollary 1 and Example 2. Author PDF. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h

[3] Didier Dubois and Henri Prade, “Measure-free conditioning, probability and non-monotonic reasoning,” Proceedings of IJCAI 1989, §§2, 4. Conference PDF. registry ↩a ↩b