2

Is minimal logic equivalent to intuitionistic?

Obviously this is false if we interpret the symbol $\bot_M$ from minimal logic as meaning the same thing as the symbol $\bot_I$ of intuitionistic logic. But what if we don't make that identification?

More precisely, I'm asking the following two questions:

Suppose $\Gamma$ is a list of formulae and $A$ is a formula, all using just the logical symbols $\to, \vee, \wedge$. Suppose further that $\Gamma \vdash A$ is a theorem of intuitionistic logic. Is $\Gamma \vdash A$ a theorem of minimal logic?

and

Suppose we add a bottom element to minimal logic; i.e. we add a new constant symbol $\bot_I$ and the rule $\bot_I \vdash P$. Is this intuitionistic logic?

If I should be asking a different question, I'm open to hearing about that too.

Bernard
  • 175,478
  • 2
    You may find the discussion between Toby and Finn here interesting. Other paraconsistent systems around minimal logic may cast some light as well. – Derek Elkins left SE Jan 29 '18 at 00:38
  • 1
    The way things are normally treated, the language for minimal logic already contains the symbol $\bot$, and so rather than adjoining a symbol and a rule to obtain intuitionistic logic, we only need to adjoin an inference rule. In this way we can take any formula of classical logic and ask: is it provable in intuitionistic logic? Is it provable in minimal logic? – Carl Mummert Jan 29 '18 at 01:43

1 Answers1

3

The answer to both your questions is positive. But I totally disagree with your claim that minimal logic and intuitionistic logic are equivalent. To explain my viewpoint with an analogy, it is not because classical logic can be embedded in intuitionistic logic (via negative translation) that intuitionistic logic and classical logic are equivalent. Unless you redefine in a unnatural way the notion of "equivalence".

I answer your technical questions.

Suppose $\Gamma$ is a list of formulae and $A$ is a formula, all using just the logical symbols $\to,\lor,\land$. Suppose further that $\Gamma \vdash A$ is a theorem of intuitionistic logic. Is $\Gamma \vdash A$ a theorem of minimal logic?

Yes! If $\Gamma \vdash A$ is provable in intuitionistic logic, then there is a normal derivation $\pi$ of $\Gamma \vdash A$ in intuitionistic natural deduction, i.e. $\pi$ posses the subformula property. This entails that $\pi$ contains only inference rules for $\to, \lor, \land$, which are sound in minimal logic. Therefore, $\pi$ is a derivation in minimal natural deduction, i.e. $\Gamma \vdash A$ is provable in minimal logic.

So, all formulas provable in intuitionistic logic but not in minimal logic contain $\bot$ or $\lnot$ (remember that $\lnot A := A \to \bot$).

Suppose we add a bottom element to minimal logic; i.e. we add a new constant symbol $\bot_I$ and the rule $\bot_I⊢P$. Is this intuitionistic logic?

Yes! This is exactly the difference between intuitionistic logic and minimal logic. More precisely, they have the same language (with bottom $\bot$) and any derivation system for intuitionistic logic is defined as a derivation system for minimal logic plus the inference rule (that allows us to derive) $\bot \vdash P$ for every formula $P$ (the so called principe of explosion or ex falso quodlibet). Note that in minimal logic the bottom $\bot$ is nothing but a propositional variable, since in minimal logic there are no special inference rules for $\bot$ (there is no point in having a minimal bottom $\bot_M$ and an intuitionistic bottom $\bot_I$): $\bot$ in minimal logic does not have the same properties as in intuitionistic logic (as well as $\lor$ in intuitionistic logic does not have the same meaning as in classical logic).

Minimal logic can also be formulated in a language without $\bot$ but with negation $\lnot$ as primitive, adding (an inference rule that allows us to derive) the axiom $((A \to B) \land (A \to \lnot B)) \to \lnot A$ (see here, thanks to Peter Smith who advised me about this reference in his answer to a question of mine), but it is essentially equivalent to the formulation of minimal logic with $\bot$ and without the principle of explosion.