The Ideal/Casuistic Principle: A Common Diagnosis for Lewis’s Explosion and the Two-Barbaras Problem

Record of a unification found in this conversation: two problems in classical logic — the explosion of a contradiction (C.I. Lewis) and the unexplained asymmetry of Aristotle’s mixed modal syllogism (“the two Barbaras”) — turn out to share the same underlying diagnosis.

1. The general principle

Every universal premise (“All A is B”) admits two readings that must not be conflated:

• Ideal reading: the premise is held as true, without reservation, among those reasoning together — “we affirm there is no case of A that is not B.” It need not be called “necessary” in the strong modal sense; it is enough that it be accepted without exception, dialogically, for it to function in an inference.
• Casuistic (or real) reading: the premise describes how things happen to stand on this occasion, without closing off the possibility that it could be otherwise — “it happens to be the case that all A is B, this time.” Explicit forms of this reading: “all A can be B,” “all A is B, for now.”

The confusion that generates the problems below consists in treating a casuistically-read premise as though it were ideally read — that is, letting an un-eliminated possibility keep circulating through the inference while acting as though it had already been ruled out.

This principle is an application, to two concrete cases in formal logic, of the distinction already recognized in philosophy of science between nomological generalizations (lawlike, support counterfactuals) and accidental generalizations (a coincidence of the occasion, do not support them) — see Hempel, and Goodman’s “grue” paradox.

2. First application: Lewis’s explosion and the Boole/Jevons dispute

C. I. Lewis’s proof that a contradiction implies any proposition uses three steps:

1. ∧-elimination: from A∧¬A, extract A and ¬A.
2. ∨-introduction (addition): from A, infer A∨B, for any B.
3. Disjunctive syllogism: from A∨B and ¬A, infer B.

Step 2 is where the confusion enters. AND and XOR are purely formal operations (Boole’s): they combine truth values without committing to anything about what is real. The inclusive OR, in the form Jevons popularized and that Lewis uses in addition, does something different: it injects a new possibility (that of B, entirely unrelated to what the original premises held) under the guise of a purely formal step. That possibility of B was never “held as true” by anyone — it is added unilaterally by the inference rule, and yet the rest of the proof treats it as though it carried the same dialogical force as the original premises.

3. Second application: the problem of the two Barbaras

3.1 The classical problem

Aristotle (Prior Analytics I.9–22) accepts the mixed modal syllogism Barbara LXL (necessary major premise, assertoric minor → necessary conclusion) but rejects Barbara XLL (assertoric major, necessary minor → necessary conclusion) — same figure, premises swapped, asymmetric verdict with no structural reason agreed upon in nearly 2400 years of commentary (Theophrastus: neither is valid; Łukasiewicz: both are valid; modern discussion — Malink 2013, Johnston 2016, Botting 2023 — still unresolved).

3.2 Formal verification: position doesn’t matter, the reading does

Finite models were built (2 worlds, a domain of 2 individuals) to check this directly. Result:

Casuistic major, ideal minor:    176 of 576 models fail
Ideal major, casuistic minor:    176 of 576 models fail   <- identical
Both major and minor ideal:        0 of 256 models fail   <- always valid
Both major and minor casuistic:  448 of 1024 models fail

Position (major/minor) has no effect on the outcome whatsoever — it is an artifact of which case one chooses to emphasize when describing it, not a variable the logic itself respects. The only thing that determines whether the syllogism preserves universality in the conclusion is whether any premise receives the casuistic reading. With both premises read ideally, the conclusion is always universal, without a single exception across 256 models tested.

3.3 The “problematized” form isn’t a failing Barbara — it’s a different form in disguise

Every case of A can (or cannot) be a case of B
C is a case of A
∴ C can (or cannot) be a case of B

This major premise is not “All A is B” with less force — it is an explicit modal disjunction (◇B ∧ ◇¬B) that never had the categorical form the syllogism requires. “The possibility is not eliminated, therefore there is no necessity” — a direct corollary: there was never, in that premise, a possibility to eliminate, because the premise never closed that door to begin with.

4. Unified diagnosis

In both cases, the error lies in no single inference rule taken in isolation (∨-introduction is valid; Barbara is valid). The error lies in treating a premise that retains an open possibility as though that possibility had already been ruled out by dialogical agreement:

The syllogism — and formal inference in general — is a dialogical tool: its validity depends on the premises being held as true, without reservation, among those reasoning together — not on their being demonstrated as true of reality. When a premise retains an explicit possibility (assertoric in the casuistic sense, or an inflated disjunction like Jevons’s), it no longer meets the entry condition the inference requires, and using it anyway — treating it as though it were ideally read — is what generates, in both historical cases, results that look paradoxical but are in fact errors of reading, not failures of the logic itself.

5. Status

Diagnosis closed and formally verified for both cases (§2 by direct argument, §3 by finite models). The general principle (§1, §4) is recorded as a result of this conversation, with credit for the original diagnosis — the ideal/casuistic distinction, and its connection to Boole/Jevons — belonging to Diego Tentor.