The program of proof-theoretic semantics (PTS) seeks to characterize the meanings of logical operators in terms of characteristic rules of inference. We argue that the concept of a universal construction is the best way to formalize this method of specifying meaning. From this point of view, operators have a natural polarity: some familiar operators that Gentzen studied (⊃, ∧, ¬, and ∀) appear as right-universal constructions, while the others (∨, ∃) appear as left-universal constructions. The same idea can be used to specify the meanings of operators that Gentzen did not treat. The “multiplicative or” (unlike its “additive” counterpart) and the first-order identity predicate = both have right-universal constructions, while the unary predicate for being a natural number N has a left-universal construction. The roles of natural deduction’s introduction and elimination rules are sensitive to the operator’s polarity.
Most advocates of PTS have preferred a uniform, asymmetric theory of meaning, according to which it is either always an operator’s introduction rule or always an operator’s elimination rule that defines that operator. We will see that developing such an asymmetric account leads to cumbersome higher-order rules that seem to be more general than the lower-level rules arrived at on our symmetric account. But we will show that higher-order rules arrived at in this way uniformly collapse: They are interderivable with their lower-level counterparts. This observation underscores the idea that what really matter for PTS are the universal constructions of logical operators, not the particular forms of their presentation.