Lecture 15: Validity and Predicate Logic 1 Goals Today ✤ Learn the definition of valid and invalid arguments in terms of the semantics of predicate logic, and look at several examples. ✤ Learn how to get equivalents in predicate logic, and look at how equivalents can help us show things about validities. 2 Recall: Propositional Validity ✤ ✤ Here’s the definition we learned in Lecture 12 for formulas of propositional logic. We say that ϕ1, …, ϕn ⊨ ψ if any truth-table which contains columns for each of ϕ1, …, ϕn, ψ has this feature: whenever a row contains a T in each of the ϕ1, …, ϕn columns, this row also has a T in the ψ column. 3 Defn: Validity in Predicate Logic ✤ Suppose ϕ1, …, ϕn , ψ, are formulas of predicate logic. Then we say ϕ1, …, ϕn ⊨ ψ if any model M which interprets the language of ϕ1, …, ϕn , ψ has this feature: if VM(ϕ1)=T, . . . , VM(ϕn)=T then VM(ψ)=T. ✤ Again, we call ϕ1, …, ϕn the premises and we call ψ the conclusion. And the argument with premises ϕ1, …, ϕn and conclusion ψ is said to be valid if ϕ1, …, ϕn ⊨ ψ. We may also continue to use the “follows from” and “has a consequence” locutions. ✤ Again, the way to see the relationship between propositional logic and predicate logic is as follows: individual models correspond to individual rows of the truth-table. 4 Example 1 ✤ Let’s show that ∀x Fx ⊨ Fa. ✤ So we need to show that any model M in the language with an individual constant a and predicate letter F is such that if VM(∀x Fx)=T then VM(Fa)=T ✤ Well, suppose that M is such a model, and suppose that it’s the case that VM(∀x Fx)=T. We have to show that VM(Fa)=T. 5 ✤ But, by the clause for ∀ in the semantics, VM(∀x Fx)=T implies that for all individual constants c, one has that VM(Fc)=T ✤ In particular, for the individual constant a, we have that VM(Fa)=T which is what we wanted to show. Example 2 ✤ ✤ ✤ Let’s show that (∀x Ax) ∨ (∀x Bx) ⊨ ∀x (Ax∨Bx) ✤ Suppose first that VM(∀x Ax)=T. Then the clause for ∀ says that VM(Ac)=T for each c. Then the clause for ∨ implies that VM(Ac ∨ Bc)=T for each c. But then the clause for ∀ says that VM(∀x (Ax∨Bx)) = T. ✤ We argue similarly in the case where VM(∀x Bx)=T. Hence, we conclude that in either case we have VM(∀x (Ax∨Bx)) = T, which is what we wanted to show. So suppose M is a model with VM((∀x Ax) ∨ (∀x Bx))=T We must show we also have VM(∀x (Ax∨Bx)) = T Since VM((∀x Ax) ∨ (∀x Bx))=T, the clause for ∨ says that VM(∀x Ax)=T or VM(∀x Bx)=T 6 Definition of Invalidity ✤ Suppose ϕ1, …, ϕn , ψ, are formulas of predicate logic. ✤ Then we say ϕ1, …, ϕn ⊭ ψ if at least one model M interpreting the language of ϕ1, …, ϕn , ψ has this feature: VM(ϕ1)=T, . . . , VM(ϕn)=T and VM(ψ)=F. ✤ Again, we call ϕ1, …, ϕn the premises and we call ψ the conclusion. And the argument with premises ϕ1, …, ϕn and conclusion ψ is said to be invalid if ϕ1, …, ϕn ⊭ ψ. ✤ There’s an obvious sense in which showing invalidity is easier than showing validity: to show invalidity, you just need to find one model. 7 Example 3 ✤ ✤ Let’s show that ∀x (Ax∨Bx) ⊭ (∀x Ax) ∨ (∀x Bx) ✤ So we just need to find one model M with the property that VM(∀x (Ax∨Bx))=T and VM((∀x Ax) ∨ (∀x Bx))=F ✤ ✤ Well, consider our old friend M with domain natural numbers and with A interpreted as the evens and B interpreted as the odds. 8 Let’s check VM(∀x (Ax∨Bx))=T. By the clause for ∀, we have that it suffices to show that VM(Ac∨Bc)=T for all c from M. But every number is even or odd, and so we know this! Let’s check that we indeed have VM((∀x Ax) ∨ (∀x Bx))=F Well, VM(∀x Ax)=F, since 3 is not even, and VM(∀x Bx)=F, since 2 is not odd. So by clause for ∨, VM((∀x Ax) ∨ (∀x Bx))=F. Example 3 (Venn Diagrams) ✤ Let’s show that ∀x (Ax∨Bx) ⊭ (∀x Ax) ∨ (∀x Bx) ✤ So we just need to find one model M with the property that VM(∀x (Ax∨Bx))=T and VM((∀x Ax) ∨ (∀x Bx))=F ✤ Well, consider our old friend M with domain natural numbers and with A interpreted as the evens and B interpreted as the odds. ✤ There’s a traditional way of illustrating these models with pictures called Venn diagrams: A 9 M B More Venn-Diagram Examples ✤ Most simple relationships between two predicates A,B can be easily visualized in terms of Venn Diagram. ✤ Here’s an example of a diagram of model M where the following is true: ∀ x (Ax→ Bx) ✤ B So in terms of this picture, this sentence represents the containment of A in B. A 10 M More Venn-Diagram Examples ✤ Most simple relationships between two predicates A,B can be easily visualized in terms of Venn Diagram. ✤ Here’s an example of a diagram of model M where the following is true: ∃ x (Ax ∧ Bx) ✤ B A M So in terms of this picture, this sentence represents that A and B overlap. 11 More Venn-Diagram Examples ✤ ✤ ✤ So what would a diagram of ∀ x Ax look like? Well, it just corresponds to the claim that everything in the domain is in A. A M It’s easy to visualize with the following animation. 12 More Venn-Diagram Examples ✤ ✤ So what would a diagram of ∃ x Ax look like? Well, it just corresponds to the claim that something is in A. So it corresponds to A just being any shape within M. M A 13 Example 4 ✤ Let’s show that we have ∀x∃y Rxy ⊭ ∃y ∀x Rxy ✤ Consider the model M, where Rab iff there’s an arrow from a to b in the picture: 1 2 3 4 14 ✤ Let’s check VM(∀x∃y Rxy)=T. We just need to check that for each number, there’s an arrow from that number to another number. But we see it’s so! ✤ Let’s check VM(∃y ∀x Rxy)=F. Suppose it was true, that is, suppose VM(∃y ∀x Rxy)=T. By the clause for ∃, we would have VM(∀x Rxc)=T for some c. But c can’t be 1 (since . . .), and it can’t be 2 (since . . ), etc. Ω 15
© Copyright 2026 Paperzz